From aaf8f79c679ea058c58a1114d380d99e1ff34156 Mon Sep 17 00:00:00 2001 From: Philippe Blain Date: Fri, 14 Feb 2025 17:36:17 +0000 Subject: [PATCH 001/259] config/remote.txt: reunite 'severOption' description paragraphs When 'remote..followRemoteHEAD' was added in b7f7d16562 (fetch: add configuration for set_head behaviour, 2024-11-29), its description was added to remote.txt in between the two paragraphs describing 'remote..serverOption'. Reunite these two paragraphs. Signed-off-by: Philippe Blain Signed-off-by: Junio C Hamano --- Documentation/config/remote.txt | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/Documentation/config/remote.txt b/Documentation/config/remote.txt index 4118c219c13ac6..1b9814e8aa4237 100644 --- a/Documentation/config/remote.txt +++ b/Documentation/config/remote.txt @@ -101,6 +101,11 @@ remote..serverOption:: The default set of server options used when fetching from this remote. These server options can be overridden by the `--server-option=` command line arguments. ++ +This is a multi-valued variable, and an empty value can be used in a higher +priority configuration file (e.g. `.git/config` in a repository) to clear +the values inherited from a lower priority configuration files (e.g. +`$HOME/.gitconfig`). remote..followRemoteHEAD:: How linkgit:git-fetch[1] should handle updates to `remotes//HEAD`. @@ -114,8 +119,3 @@ remote..followRemoteHEAD:: Setting to "always" will silently update it to the value on the remote. Finally, setting it to "never" will never change or create the local reference. -+ -This is a multi-valued variable, and an empty value can be used in a higher -priority configuration file (e.g. `.git/config` in a repository) to clear -the values inherited from a lower priority configuration files (e.g. -`$HOME/.gitconfig`). From 0d03fda6a5db0b56df7e629ce856d2d38c89a4de Mon Sep 17 00:00:00 2001 From: Philippe Blain Date: Fri, 14 Feb 2025 17:36:18 +0000 Subject: [PATCH 002/259] config/remote.txt: improve wording for 'remote..followRemoteHEAD' Signed-off-by: Philippe Blain Signed-off-by: Junio C Hamano --- Documentation/config/remote.txt | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/Documentation/config/remote.txt b/Documentation/config/remote.txt index 1b9814e8aa4237..25fe219d103cc4 100644 --- a/Documentation/config/remote.txt +++ b/Documentation/config/remote.txt @@ -110,12 +110,12 @@ the values inherited from a lower priority configuration files (e.g. remote..followRemoteHEAD:: How linkgit:git-fetch[1] should handle updates to `remotes//HEAD`. The default value is "create", which will create `remotes//HEAD` - if it exists on the remote, but not locally, but will not touch an - already existing local reference. Setting to "warn" will print - a message if the remote has a different value, than the local one and + if it exists on the remote, but not locally; this will not touch an + already existing local reference. Setting it to "warn" will print + a message if the remote has a different value than the local one; in case there is no local reference, it behaves like "create". A variant on "warn" is "warn-if-not-$branch", which behaves like "warn", but if `HEAD` on the remote is `$branch` it will be silent. - Setting to "always" will silently update it to the value on the remote. - Finally, setting it to "never" will never change or create the local - reference. + Setting it to "always" will silently update `remotes//HEAD` to + the value on the remote. Finally, setting it to "never" will never + change or create the local reference. From a97ca95160b510dfc8e609638fb9df1b863b947f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?V=C5=A9=20Ti=E1=BA=BFn=20H=C6=B0ng?= Date: Thu, 6 Mar 2025 12:41:39 +0700 Subject: [PATCH 003/259] l10n: Updated translation for vi-2.49 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Vũ Tiến Hưng --- po/vi.po | 533 +++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 358 insertions(+), 175 deletions(-) diff --git a/po/vi.po b/po/vi.po index 80a5f191a85c7a..803a68e1b6025d 100644 --- a/po/vi.po +++ b/po/vi.po @@ -11,7 +11,7 @@ # Vũ Tiến Hưng , 2024-2025. # --- # BẢNG THUẬT NGỮ / TERMINOLOGY -# Updated: 2024-07-26, git 2.46 +# Updated: 2025-03-06, git 2.49 # # Ghi chú: # - Bảng thuật ngữ này chưa hoàn thiện. @@ -59,15 +59,17 @@ # | (v.) rebase | cải tổ | # | (v.) squash | squash | # | (v.) amend | tu bổ | +# | (n.) revision | cải biên | +# | (n.) repo/repository | kho chứa | # | | | # | ... TODO ... | | # +------------------------------------------------------------------+ msgid "" msgstr "" -"Project-Id-Version: git 2.48\n" +"Project-Id-Version: git 2.49\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-23 18:57+0000\n" -"PO-Revision-Date: 2025-01-05 01:20+0700\n" +"POT-Creation-Date: 2025-03-05 22:57+0000\n" +"PO-Revision-Date: 2025-03-06 08:20+0700\n" "Last-Translator: Vũ Tiến Hưng \n" "Language-Team: Vietnamese \n" "Language: vi\n" @@ -1601,7 +1603,7 @@ msgid "" "git bisect cannot work properly in this case.\n" "Maybe you mistook %s and %s revs?\n" msgstr "" -"Một số điểm xét duyệt %s không phải tổ tiên của điểm xét duyệt %s.\n" +"Một số lần cải biên %s không phải tổ tiên của lần cải biên %s.\n" "git bisect không thể làm việc đúng trong trường hợp này.\n" "Liệu có phải bạn nhầm lẫn các điểm %s và %s không?\n" @@ -1621,7 +1623,7 @@ msgstr "Đang bisect: gốc hòa trộn cần phải được kiểm tra\n" #, c-format msgid "a %s revision is needed" -msgstr "cần một điểm xét duyệt %s" +msgstr "cần lần cải biên %s" #, c-format msgid "could not create file '%s'" @@ -1661,7 +1663,7 @@ msgstr[0] "(cần khoảng chừng %d bước)" #, c-format msgid "Bisecting: %d revision left to test after this %s\n" msgid_plural "Bisecting: %d revisions left to test after this %s\n" -msgstr[0] "Bisecting: còn %d điểm xét duyệt để kiểm tra %s\n" +msgstr[0] "Bisecting: còn %d lần cải biên để kiểm tra %s\n" msgid "--contents and --reverse do not blend well." msgstr "tùy chọn --contents và --reverse không nên đi với nhau." @@ -1671,7 +1673,7 @@ msgstr "" "cùng sử dụng --reverse và --first-parent cần chỉ định lần chuyển giao cuối" msgid "revision walk setup failed" -msgstr "cài đặt việc di chuyển qua các điểm xét duyệt gặp lỗi" +msgstr "cài đặt việc duyệt qua các lần cải biên gặp lỗi" msgid "" "--reverse --first-parent together require range along first-parent chain" @@ -2371,6 +2373,18 @@ msgstr "git archive: lỗi giao thức" msgid "git archive: expected a flush" msgstr "git archive: cần flush dữ liệu" +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +msgid "problem loading sparse-checkout" +msgstr "không thể tải checkout thưa" + +msgid "Minimum number of objects to request at a time" +msgstr "Số đối tượng tối thiểu mỗi lần yêu cầu" + +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "Chỉ lấy về đối tượng còn thiếu trong checkout thưa hiện tại" + msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" "checkout] [--first-parent] [ [...]] [--] [...]" @@ -2441,7 +2455,7 @@ msgstr "Đối số bisect_write sai: %s" #, c-format msgid "couldn't get the oid of the rev '%s'" -msgstr "không thể lấy oid của điểm xét duyệt '%s'" +msgstr "không thể lấy oid của lần cải biên '%s'" #, c-format msgid "couldn't open the file '%s'" @@ -2466,7 +2480,7 @@ msgid "" "You can use \"git bisect %s\" and \"git bisect %s\" for that." msgstr "" "Bạn cần bắt đầu bằng lệnh \"git bisect start\".\n" -"Bạn sau đó cần phải chỉ cho tôi ít nhất một điểm xét duyệt %s và một %s.\n" +"Bạn sau đó cần phải chỉ cho tôi ít nhất một lần cải biên %s và một %s.\n" "Bạn có thể sử dụng \"git bisect %s\" và \"git bisect %s\" cho chúng." #, c-format @@ -2526,7 +2540,7 @@ msgstr "không nhận ra tuỳ chọn: '%s'" #, c-format msgid "'%s' does not appear to be a valid revision" -msgstr "'%s' không có vẻ như là một điểm xét duyệt hợp lệ" +msgstr "'%s' không có vẻ như là một lần cải biên hợp lệ" msgid "bad HEAD - I need a HEAD" msgstr "sai HEAD - Tôi cần một HEAD" @@ -2587,11 +2601,11 @@ msgstr "bisect chạy gặp lỗi: không đưa ra lệnh." #, c-format msgid "unable to verify %s on good revision" -msgstr "không thể xác nhận '%s' trên điểm xét duyệt tốt" +msgstr "không thể xác nhận '%s' trên lần cải biên tốt" #, c-format msgid "bogus exit code %d for good revision" -msgstr "mã trả về %d bất thường cho điểm xét duyệt tốt" +msgstr "mã trả về %d bất thường cho lần cải biên tốt" #, c-format msgid "bisect run failed: exit code %d from %s is < 0 or >= 128" @@ -2658,7 +2672,7 @@ msgstr "phải kết thúc bằng một màu" #, c-format msgid "cannot find revision %s to ignore" -msgstr "không thể tìm thấy điểm xét duyệt %s để bỏ qua" +msgstr "không thể tìm thấy lần cải biên %s để bỏ qua" msgid "show blame entries as we find them, incrementally" msgstr "hiển thị các mục 'blame' theo thời gian, tăng dần" @@ -2717,7 +2731,7 @@ msgid "ignore when blaming" msgstr "bỏ qua khi blame" msgid "ignore revisions from " -msgstr "bỏ qua các điểm xét duyệt từ " +msgstr "bỏ qua các lần cải biên từ " msgid "color redundant metadata from previous line differently" msgstr "tô màu khác cho siêu dữ liệu dư thừa từ dòng trước" @@ -2730,7 +2744,7 @@ msgstr "tiêu thụ thêm tài nguyên để tìm kiếm tốt hơn nữa" msgid "use revisions from instead of calling git-rev-list" msgstr "" -"sử dụng các điểm xét duyệt (revision) từ thay vì gọi git-rev-list" +"sử dụng các lần cải biên (revision) từ thay vì gọi git-rev-list" msgid "use 's contents as the final image" msgstr "sử dụng nội dung của như là ảnh cuối cùng" @@ -3114,10 +3128,6 @@ msgstr "" msgid "git version:\n" msgstr "phiên bản git:\n" -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() gặp lỗi '%s' (%d)\n" - msgid "compiler info: " msgstr "thông tin trình biên dịch: " @@ -3710,7 +3720,7 @@ msgstr[0] "" "\n" msgid "internal error in revision walk" -msgstr "lỗi nội bộ trong khi di chuyển qua các điểm xét duyệt" +msgstr "lỗi nội bộ trong khi di chuyển qua các lần cải biên" msgid "Previous HEAD position was" msgstr "Vị trí trước kia của HEAD là" @@ -4116,8 +4126,91 @@ msgstr "" "clean.requireForce được đặt thành true và không có tuỳ chọn -f; từ chối dọn " "dẹp" -msgid "git clone [] [--] []" -msgstr "git clone [] [--] []" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "thông tin: không thể thêm thay thế cho '%s': %s\n" + +#, c-format +msgid "failed to stat '%s'" +msgstr "gặp lỗi khi stat '%s'" + +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s có tồn tại nhưng lại không phải là một thư mục" + +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "'%s' là liên kết mềm, từ chối sao chép với --local" + +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "gặp lỗi khi bắt đầu lặp qua '%s'" + +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "liên kết mềm '%s' đã tồn tại, từ chối sao chép với --local" + +#, c-format +msgid "failed to unlink '%s'" +msgstr "gặp lỗi khi unlink '%s'" + +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "không thể kiểm tra liên kết cứng '%s'" + +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "liên kết cứng '%s' khác với nguồn" + +#, c-format +msgid "failed to create link '%s'" +msgstr "gặp lỗi khi tạo liên kết mềm %s" + +#, c-format +msgid "failed to copy file to '%s'" +msgstr "gặp lỗi khi sao chép tập tin tới '%s'" + +#, c-format +msgid "failed to iterate over '%s'" +msgstr "gặp lỗi khi lặp qua '%s'" + +#, c-format +msgid "done.\n" +msgstr "hoàn tất.\n" + +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Việc nhân bản thành công, nhưng checkout gặp lỗi.\n" +"Kiểm tra xem cái gì đã được checkout bằng lệnh 'git status'\n" +"và thử lại với lệnh 'git restore --source=HEAD :/'\n" + +msgid "remote did not send all necessary objects" +msgstr "máy chủ đã không gửi tất cả các đối tượng cần thiết" + +#, c-format +msgid "unable to update %s" +msgstr "không thể cập nhật %s" + +msgid "failed to initialize sparse-checkout" +msgstr "gặp lỗi khi khởi tạo sparse-checkout" + +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "HEAD ở máy chủ chỉ đến ref không tồn tại, không thể checkout" + +msgid "unable to checkout working tree" +msgstr "không thể checkout cây làm việc" + +msgid "unable to write parameters to config file" +msgstr "không thể ghi các tham số vào tập tin cấu hình" + +msgid "cannot repack to clean up" +msgstr "không thể đóng gói để dọn dẹp" + +msgid "cannot unlink temporary alternates file" +msgstr "không thể bỏ liên kết tập tin thay thế tạm thời" msgid "don't clone shallow repository" msgstr "đừng nhân bản từ kho nông" @@ -4170,6 +4263,9 @@ msgstr "dùng thay cho 'origin' để theo dõi thượng nguồn" msgid "checkout instead of the remote's HEAD" msgstr "checkout thay cho HEAD của máy chủ" +msgid "clone single revision and check out" +msgstr "nhân bản chỉ lần cải biên and check out" + msgid "path to git-upload-pack on the remote" msgstr "đường dẫn đến git-upload-pack trên máy chủ" @@ -4191,10 +4287,8 @@ msgstr "làm sâu hơn lịch sử của bản sao shallow, loại trừ tham ch msgid "clone only one branch, HEAD or --branch" msgstr "chỉ nhân bản một nhánh, HEAD hoặc --branch" -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "" -"đứng có nhân bản bất kỳ nhánh nào, và làm cho những lần lấy về sau không " -"theo chúng nữa" +msgid "clone tags, and make later fetches not to follow them" +msgstr "nhân bản thẻ, và làm cho những lần lấy về sau không theo chúng nữa" msgid "any cloned submodules will be shallow" msgstr "mọi mô-đun-con nhân bản sẽ là shallow (nông)" @@ -4235,95 +4329,8 @@ msgstr "uri" msgid "a URI for downloading bundles before fetching from origin remote" msgstr "URI để tải xuống bundle trước khi lấy về từ máy chủ origin" -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "thông tin: không thể thêm thay thế cho '%s': %s\n" - -#, c-format -msgid "failed to stat '%s'" -msgstr "gặp lỗi khi stat '%s'" - -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s có tồn tại nhưng lại không phải là một thư mục" - -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "'%s' là liên kết mềm, từ chối sao chép với --local" - -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "gặp lỗi khi bắt đầu lặp qua '%s'" - -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "liên kết mềm '%s' đã tồn tại, từ chối sao chép với --local" - -#, c-format -msgid "failed to unlink '%s'" -msgstr "gặp lỗi khi unlink '%s'" - -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "không thể kiểm tra liên kết cứng '%s'" - -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "liên kết cứng '%s' khác với nguồn" - -#, c-format -msgid "failed to create link '%s'" -msgstr "gặp lỗi khi tạo liên kết mềm %s" - -#, c-format -msgid "failed to copy file to '%s'" -msgstr "gặp lỗi khi sao chép tập tin tới '%s'" - -#, c-format -msgid "failed to iterate over '%s'" -msgstr "gặp lỗi khi lặp qua '%s'" - -#, c-format -msgid "done.\n" -msgstr "hoàn tất.\n" - -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Việc nhân bản thành công, nhưng checkout gặp lỗi.\n" -"Kiểm tra xem cái gì đã được checkout bằng lệnh 'git status'\n" -"và thử lại với lệnh 'git restore --source=HEAD :/'\n" - -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "Không tìm thấy nhánh máy chủ %s để nhân bản (clone)." - -msgid "remote did not send all necessary objects" -msgstr "máy chủ đã không gửi tất cả các đối tượng cần thiết" - -#, c-format -msgid "unable to update %s" -msgstr "không thể cập nhật %s" - -msgid "failed to initialize sparse-checkout" -msgstr "gặp lỗi khi khởi tạo sparse-checkout" - -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "HEAD ở máy chủ chỉ đến ref không tồn tại, không thể checkout" - -msgid "unable to checkout working tree" -msgstr "không thể checkout cây làm việc" - -msgid "unable to write parameters to config file" -msgstr "không thể ghi các tham số vào tập tin cấu hình" - -msgid "cannot repack to clean up" -msgstr "không thể đóng gói để dọn dẹp" - -msgid "cannot unlink temporary alternates file" -msgstr "không thể bỏ liên kết tập tin thay thế tạm thời" +msgid "git clone [] [--] []" +msgstr "git clone [] [--] []" msgid "Too many arguments." msgstr "Quá nhiều đối số." @@ -4430,6 +4437,10 @@ msgstr "trình vận chuyển đã báo lỗi" msgid "Remote branch %s not found in upstream %s" msgstr "Nhánh máy chủ %s không tìm thấy trong thượng nguồn %s" +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "Lần cải biên %s không tìm thấy trong thượng nguồn %s" + msgid "You appear to have cloned an empty repository." msgstr "Bạn hình như là đã nhân bản một kho trống rỗng." @@ -4605,7 +4616,7 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree: gặp lỗi khi đọc" msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -5576,7 +5587,7 @@ msgstr "" #, c-format msgid "traversed %lu commits\n" -msgstr "đã xuyên %lu qua lần chuyển giao\n" +msgstr "đã chạy qua %lu lần chuyển giao\n" #, c-format msgid "found %i tags; gave up search at %s\n" @@ -6021,8 +6032,8 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Chạy 'git remote set-head %s %s' để làm theo thay đổi, hoặc đặt tuỳ chọn\n" "'remote.%s.followRemoteHEAD' sang giá trị khác nếu bạn không muốn thấy\n" @@ -6073,7 +6084,7 @@ msgid "" "remote name from which new revisions should be fetched" msgstr "" "chưa chỉ ra kho chứa máy chủ; xin hãy chỉ định hoặc là URL hoặc\n" -"tên máy chủ từ cái mà những điểm xét duyệt mới có thể được fetch (lấy về)" +"tên máy chủ từ cái mà những lần cải biên mới có thể được fetch (lấy về)" msgid "you need to specify a tag name" msgstr "bạn cần chỉ định một tên thẻ" @@ -6167,7 +6178,7 @@ msgid "specify fetch refmap" msgstr "chỉ ra refmap cần lấy về" msgid "revision" -msgstr "điểm xét duyệt" +msgstr "lần cải biên" msgid "report that we have only objects reachable from this object" msgstr "báo rằng ta chỉ có các đối tượng tiếp cận được từ đối tượng này" @@ -6705,6 +6716,9 @@ msgstr "buộc gc chạy ngay cả khi có tiến trình gc khác đang chạy" msgid "repack all other packs except the largest pack" msgstr "đóng gói lại tất cả các gói khác ngoại trừ gói lớn nhất" +msgid "pack prefix to store a pack containing pruned objects" +msgstr "tiền tố của gói để lưu gói gồm những đối tượng đã loại bỏ" + #, c-format msgid "failed to parse gc.logExpiry value %s" msgstr "gặp lỗi khi đọc giá trị gc.logExpiry %s" @@ -7113,7 +7127,7 @@ msgstr "--no-index hay --untracked không được sử dụng cùng với revs" #, c-format msgid "unable to resolve revision: %s" -msgstr "không thể phân giải điểm xét duyệt: %s" +msgstr "không thể phân giải lần cải biên: %s" msgid "--untracked not supported with --recurse-submodules" msgstr "tùy chọn --untracked không được hỗ trợ với --recurse-submodules" @@ -7490,6 +7504,10 @@ msgstr "không thể hoàn tất pack-objects để đóng gói các link cục msgid "Cannot come back to cwd" msgstr "Không thể quay lại thư mục hiện hành" +#, c-format +msgid "bad --pack_header: %s" +msgstr "--pack_header sai: %s" + #, c-format msgid "bad %s" msgstr "%s sai" @@ -7758,10 +7776,10 @@ msgid "failed to find exact merge base" msgstr "gặp lỗi khi tìm gốc hòa trộn chính xác" msgid "base commit should be the ancestor of revision list" -msgstr "lần chuyển giao nền không là tổ tiên của danh sách điểm xét duyệt" +msgstr "lần chuyển giao nền không là tổ tiên của danh sách lần cải biên" msgid "base commit shouldn't be in revision list" -msgstr "lần chuyển giao nền không được trong danh sách điểm xét duyệt" +msgstr "lần chuyển giao nền không được trong danh sách lần cải biên" msgid "cannot get patch id" msgstr "không thể lấy mã bản vá" @@ -8356,10 +8374,6 @@ msgstr "không hiểu chiến lược: -X%s" msgid "malformed input line: '%s'." msgstr "dòng đầu vào sai quy cách: '%s'." -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "không thể tiếp tục hoà trộn; kết quả không hoàn toàn %d" - msgid "git merge [] [...]" msgstr "git merge [] [...]" @@ -9186,6 +9200,13 @@ msgstr "" "git pack-objects [] [< | < ]" +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "tùy chọn --name-hash-version không hợp lệ: %d" + +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "--write-bitmap-index hiện cần --name-hash-version=1" + #, c-format msgid "" "write_reuse_object: could not locate %s, expected at offset % in " @@ -9363,7 +9384,7 @@ msgstr "không phải một rev '%s'" #, c-format msgid "bad revision '%s'" -msgstr "điểm xem xét sai '%s'" +msgstr "lần cải biên sai '%s'" msgid "unable to add recent objects" msgstr "không thể thêm các đối tượng mới dùng" @@ -9509,6 +9530,9 @@ msgstr "giao thức" msgid "exclude any configured uploadpack.blobpackfileuri with this protocol" msgstr "loại trừ bất kỳ cấu hình uploadpack.blobpackfileuri với giao thức này" +msgid "use the specified name-hash function to group similar objects" +msgstr "dùng hàm băm này để nhóm các đối tượng giống nhau" + #, c-format msgid "delta chain depth %d is too deep, forcing %d" msgstr "mức delta chain %d là quá sâu, buộc dùng %d" @@ -10261,7 +10285,7 @@ msgid "" msgstr "" "\n" "git gặp phải một lỗi trong khi đang chuẩn bị các bản vá để diễn lại\n" -"những điểm xét duyệt này:\n" +"những lần cải biên này:\n" "\n" " %s\n" "\n" @@ -10750,8 +10774,8 @@ msgstr "chưa chỉ ra reflog để xóa" msgid "invalid ref format: %s" msgstr "định dạng tham chiếu không hợp lệ: %s" -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format=<định dạng> [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format=<định dạng> [--no-reflog] [--dry-run]" msgid "git refs verify [--strict] [--verbose]" msgstr "git refs verify [--strict] [--verbose]" @@ -10762,6 +10786,9 @@ msgstr "chỉ định định dạng tham chiếu để chuyển đổi sang" msgid "perform a non-destructive dry-run" msgstr "chạy thử mà không thay đổi gì" +msgid "drop reflogs entirely during the migration" +msgstr "bỏ reflog khi chuyển đổi" + msgid "missing --ref-format=" msgstr "thiếu --ref-format=<định dạng>" @@ -11216,8 +11243,14 @@ msgstr "Sẽ không xóa những địa chỉ URL không-push" msgid "be verbose; must be placed before a subcommand" msgstr "chi tiết; phải được đặt trước một lệnh-con" -msgid "git repack []" -msgstr "git repack []" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" msgid "" "Incremental repacks are incompatible with bitmap indexes. Use\n" @@ -11293,6 +11326,10 @@ msgstr "chuyển --no-reuse-delta cho git-pack-objects" msgid "pass --no-reuse-object to git-pack-objects" msgstr "chuyển --no-reuse-object cho git-pack-objects" +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "phiên bản hàm băm để nhóm đối tượng giống nhau theo đường dẫn" + msgid "do not run git-update-server-info" msgstr "không chạy git-update-server-info" @@ -11341,9 +11378,6 @@ msgstr "tìm một tiến trình hình học với hệ số " msgid "write a multi-pack index of the resulting packs" msgstr "ghi chỉ mục 'multi-pack' của các gói kết quả" -msgid "pack prefix to store a pack containing pruned objects" -msgstr "tiền tố của gói để lưu gói gồm những đối tượng đã loại bỏ" - msgid "pack prefix to store a pack containing filtered out objects" msgstr "tiền tố của gói để lưu gói gồm những đối tượng đã lọc bỏ" @@ -11560,11 +11594,8 @@ msgstr "chỉ được chỉ định một mẫu cùng với tùy chọn -l" msgid "need some commits to replay" msgstr "cần các chuyển giao để phát lại" -msgid "--onto and --advance are incompatible" -msgstr "--onto và --advance xung khắc" - msgid "all positive revisions given must be references" -msgstr "mọi điểm xét duyệt cộng thêm phải là tên tham chiếu" +msgstr "mọi lần cải biên cộng thêm phải là tên tham chiếu" msgid "argument to --advance must be a reference" msgstr "tham số cho --advance phải là tham chiếu" @@ -11610,11 +11641,11 @@ msgid "" "some rev walking options will be overridden as '%s' bit in 'struct rev_info' " "will be forced" msgstr "" -"một số tuỳ chọn duyệt qua điểm xét duyệt sẽ bị bỏ qua do bit '%s' trong " +"một số tuỳ chọn duyệt qua lần cải biên sẽ bị bỏ qua do bit '%s' trong " "'struct rev_info' bị ép bật/tắt" msgid "error preparing revisions" -msgstr "gặp lỗi khi chuẩn bị các điểm xét duyệt" +msgstr "gặp lỗi khi chuẩn bị các lần cải biên" msgid "replaying down to root commit is not supported yet!" msgstr "chưa hỗ trợ phát lại đến lần chuyển giao gốc!" @@ -11709,7 +11740,7 @@ msgstr "chỉ ghi lại những đường dẫn thực sự sẽ được thêm #, c-format msgid "Failed to resolve '%s' as a valid revision." -msgstr "Gặp lỗi khi phân giải '%s' thành điểm xét duyệt hợp lệ." +msgstr "Gặp lỗi khi phân giải '%s' thành lần cải biên hợp lệ." #, c-format msgid "Failed to resolve '%s' as a valid tree." @@ -11742,7 +11773,7 @@ msgstr "" #, c-format msgid "Could not reset index file to revision '%s'." -msgstr "Không thể đặt lại (reset) chỉ mục thành điểm xét duyệt '%s'." +msgstr "Không thể đặt lại (reset) chỉ mục thành lần cải biên '%s'." msgid "Could not write new index file." msgstr "Không thể ghi tập tin chỉ mục mới." @@ -11784,7 +11815,7 @@ msgid "missing opt-spec before option flags" msgstr "thiếu opt-spec trước các tuỳ chọn" msgid "Needed a single revision" -msgstr "Cần một điểm xét duyệt đơn" +msgstr "Cần một lần cải biên đơn" msgid "" "git rev-parse --parseopt [] -- [...]\n" @@ -12160,7 +12191,7 @@ msgstr "không có tham chiếu nào như thế %s" #, c-format msgid "cannot handle more than %d rev." msgid_plural "cannot handle more than %d revs." -msgstr[0] "không thể xử lý nhiều hơn %d điểm xét duyệt." +msgstr[0] "không thể xử lý nhiều hơn %d lần cải biên." #, c-format msgid "'%s' is not a valid ref." @@ -12438,7 +12469,7 @@ msgstr "'%s' không phải là lần chuyển giao kiểu-stash" #, c-format msgid "Too many revisions specified:%s" -msgstr "Chỉ ra quá nhiều điểm xét duyệt: %s" +msgstr "Chỉ ra quá nhiều lần cải biên: %s" msgid "No stash entries found." msgstr "Không tìm thấy các mục tạm cất (stash) nào." @@ -12742,7 +12773,7 @@ msgstr "" "dẫn>]" msgid "could not fetch a revision for HEAD" -msgstr "không thể lấy về một điểm xem xét cho HEAD" +msgstr "không thể lấy về một lần cải biên cho HEAD" #, c-format msgid "Synchronizing submodule url for '%s'\n" @@ -12977,8 +13008,7 @@ msgstr "" #, c-format msgid "Unable to find current revision in submodule path '%s'" -msgstr "" -"Không tìm thấy điểm xét duyệt hiện hành trong đường dẫn mô-đun-con '%s'" +msgstr "Không tìm thấy lần cải biên hiện hành trong đường dẫn mô-đun-con '%s'" #, c-format msgid "Unable to fetch in submodule path '%s'" @@ -12986,7 +13016,7 @@ msgstr "Không thể lấy về trong đường dẫn mô-đun-con '%s'" #, c-format msgid "Unable to find %s revision in submodule path '%s'" -msgstr "Không tìm thấy điểm xét duyệt %s trong đường dẫn mô-đun-con '%s'" +msgstr "Không tìm thấy lần cải biên %s trong đường dẫn mô-đun-con '%s'" #, c-format msgid "Failed to recurse into submodule path '%s'" @@ -13017,8 +13047,7 @@ msgid "use the 'rebase' update strategy" msgstr "dùng chiến lược hòa trộn 'rebase'" msgid "create a shallow clone truncated to the specified number of revisions" -msgstr "" -"tạo một bản sao nông được cắt ngắn thành số lượng điểm xét duyệt đã cho" +msgstr "tạo một bản sao nông được cắt ngắn thành số lượng lần cải biên đã cho" msgid "parallel jobs" msgstr "công việc đồng thời" @@ -14190,6 +14219,9 @@ msgstr "Nhập một kho GNU Arch vào một kho Git" msgid "Create an archive of files from a named tree" msgstr "Tạo một kho nén các tập tin từ cây làm việc có tên" +msgid "Download missing objects in a partial clone" +msgstr "Tải về các đối tượng còn thiếu khi nhân bản một phần" + msgid "Use binary search to find the commit that introduced a bug" msgstr "Tìm kiếm dạng nhị phân để tìm ra lần chuyển giao nào đưa ra lỗi" @@ -14727,7 +14759,7 @@ msgid "Git Repository Layout" msgstr "Bố cục kho Git" msgid "Specifying revisions and ranges for Git" -msgstr "Chỉ định điểm xét duyệt và vùng cho Git" +msgstr "Chỉ định lần cải biên và vùng cho Git" msgid "Mounting one repository inside another" msgstr "Gắn một kho chứa vào trong một cái khác" @@ -15482,7 +15514,7 @@ msgstr "tham chiếu '%s' không chỉ đến một blob nào cả" #, c-format msgid "unable to resolve config blob '%s'" -msgstr "không thể phân giải điểm xét duyệt '%s'" +msgstr "không thể phân giải lần cải biên '%s'" msgid "unable to parse command-line config" msgstr "không thể đọc cấu hình dòng lệnh" @@ -16106,6 +16138,12 @@ msgstr "tham số cho %s không hợp lệ" msgid "invalid regex given to -I: '%s'" msgstr "đưa cho -I biểu thức chính quy không hợp lệ: '%s'" +msgid "-G requires a non-empty argument" +msgstr "-G cần một tham số" + +msgid "-S requires a non-empty argument" +msgstr "-S cần một tham số" + #, c-format msgid "failed to parse --submodule option parameter: '%s'" msgstr "gặp lỗi khi đọc đối số tùy chọn --submodule: '%s'" @@ -18274,6 +18312,10 @@ msgstr "tập tin '%s' và '%s' có nội dung khác nhau" msgid "unable to write file %s" msgstr "không thể ghi tập tin %s" +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "không thể ghi vào tập tin biến mất liên tục %s" + #, c-format msgid "unable to set permission to '%s'" msgstr "không thể đặt quyền thành '%s'" @@ -18735,7 +18777,7 @@ msgstr "checksum không hợp lệ" #, c-format msgid "invalid rev-index position at %: % != %" -msgstr "vị trí mục xét duyệt không hợp lệ %: % != %" +msgstr "vị trí mục cải biên không hợp lệ %: % != %" msgid "multi-pack-index reverse-index chunk is the wrong size" msgstr "chunk chỉ mục ngược của chỉ mục đa gói có kích thước sai" @@ -18833,6 +18875,52 @@ msgstr "không hiểu tùy chọn '%c'" msgid "unknown non-ascii option in string: `%s'" msgstr "không hiểu tùy chọn non-ascii trong chuỗi: '%s'" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid " <%s>" +msgstr " <%s>" + msgid "..." msgstr "..." @@ -18919,6 +19007,21 @@ msgstr "sai giá trị kiểu boolean của cấu hình '%s' cho '%s'" msgid "failed to parse %s" msgstr "gặp lỗi khi đọc cú pháp %s" +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "gặp lỗi khi duyệt nhánh của cây %s: không tìm thấy" + +#, c-format +msgid "failed to find object %s" +msgstr "gặp lỗi khi tìm đối tượng '%s'." + +#, c-format +msgid "failed to find tag %s" +msgstr "gặp lỗi khi tìm thẻ %s" + +msgid "failed to setup revision walk" +msgstr "gặp lỗi khi thiết lập duyệt lần cải biên" + #, c-format msgid "Could not make %s writable by group" msgstr "Không thể làm %s được ghi bởi nhóm" @@ -19063,6 +19166,22 @@ msgstr "tên máy chủ promisor không thể bắt đầu bằng '/': %s" msgid "could not fetch %s from promisor remote" msgstr "không thể tải %s từ máy chủ promisor" +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "có máy chủ '%s' nhưng với url '%s' thay vì '%s'" + +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "không hiểu giá trị '%s' cho cho cấu hình '%s'" + +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "không hiểu phần '%s' từ thông tin máy chủ" + +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "không tìm thấy accepted promisor remote '%s'" + msgid "object-info: expected flush after arguments" msgstr "object-info: cần đẩy dữ liệu lên đĩa sau các tham số" @@ -19750,7 +19869,7 @@ msgid "refusing to update ref with bad name '%s'" msgstr "từ chối cập nhật tham chiếu với tên sai '%s'" msgid "refusing to force and skip creation of reflog" -msgstr "từ chối bỏ qua việc tạo log tham chiếu" +msgstr "từ chối bỏ qua việc tạo reflog" #, c-format msgid "update_ref failed for ref '%s': %s" @@ -19883,6 +20002,14 @@ msgstr "tên tham chiếu %s là tham chiếu biểu trưng, không hỗ trợ s msgid "invalid refspec '%s'" msgstr "refspec không hợp lệ '%s'" +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "giá trị '%s' không có '*'" + +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "thay thế '%s' không có '*'" + #, c-format msgid "invalid quoting in push-option value: '%s'" msgstr "sai trích dẫn trong giá trị push-option :'%s'" @@ -20002,6 +20129,27 @@ msgstr "remote-curl: đã cố gắng fetch mà không có kho nội bộ" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl: không hiểu lệnh '%s' từ git" +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"đang đọc máy chủ từ \"%s/%s\", đã được đề cử để loại bỏ.\n" +"Nếu bạn vẫn còn sử dụng thư mục \"remotes\", chúng tôi khuyên bạn\n" +"hãy chuyển đổi sang sử dụng máy chủ qua tập tin cấu hình:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"Nếu bạn không thể đổi, hãy cho chúng tôi biết tại sao bạn cần nó\n" +"bằng cách gửi e-mail đến ." + #, c-format msgid "config remote shorthand cannot begin with '/': %s" msgstr "cấu hình viết tắt máy chủ không thể bắt đầu bằng '/': %s" @@ -20036,14 +20184,6 @@ msgstr "%s thường theo dõi %s, không phải %s" msgid "%s tracks both %s and %s" msgstr "%s theo dõi cả %s và %s" -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "khóa '%s' của mẫu k có '*'" - -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "giá trị '%s' của mẫu k có '*'" - #, c-format msgid "src refspec %s does not match any" msgstr "refspec (đặc tả tham chiếu) nguồn %s không khớp bất kỳ cái gì" @@ -20317,7 +20457,7 @@ msgid "update the index with reused conflict resolution if possible" msgstr "cập nhật chỉ mục với phân giải xung đột dùng lại nếu được" msgid "could not determine HEAD revision" -msgstr "không thể dò tìm điểm xét duyệt HEAD" +msgstr "không thể dò tìm lần cải biên HEAD" #, c-format msgid "failed to find tree of %s" @@ -21456,10 +21596,10 @@ msgid "" "Use '--' to separate paths from revisions, like this:\n" "'git [...] -- [...]'" msgstr "" -"tham số chưa rõ ràng '%s': chưa biết điểm xét duyệt hay đường dẫn không " -"trong cây làm việc.\n" -"Dùng '--' để ngăn cách các đường dẫn khỏi điểm xét duyệt, như thế này:\n" -"'git [<điểm xét duyệt>...] -- [...]'" +"tham số chưa rõ ràng '%s': chưa biết lần cải biên hay đường dẫn không trong " +"cây làm việc.\n" +"Dùng '--' để ngăn cách các đường dẫn khỏi lần cải biên, như thế này:\n" +"'git [...] -- [...]'" #, c-format msgid "option '%s' must come before non-option arguments" @@ -21471,9 +21611,9 @@ msgid "" "Use '--' to separate paths from revisions, like this:\n" "'git [...] -- [...]'" msgstr "" -"tham số chưa rõ ràng '%s': cả điểm xem xét và tên tập tin.\n" -"Dùng '--' để ngăn cách các đường dẫn khỏi điểm xem xét, như thế này:\n" -"'git [<điểm xem xét>...] -- [...]'" +"tham số chưa rõ ràng '%s': cả lần cải biên và tên tập tin.\n" +"Dùng '--' để ngăn cách các đường dẫn khỏi lần cải biên, như thế này:\n" +"'git [...] -- [...]'" msgid "unable to set up work tree using invalid config" msgstr "không thể thiết lập thư mục làm việc với cấu hình không hợp lệ" @@ -21935,6 +22075,27 @@ msgstr "dọn cây nhớ tạm trước mỗi chu kỳ" msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "số mục cần huỷ trong câu nhớ tạm (mặc định 0)" +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk -- " + +msgid "toggle inclusion of blob objects" +msgstr "bao gồm các đối tượng blob" + +msgid "toggle inclusion of commit objects" +msgstr "bao gồm các đối tượng chuyển giao" + +msgid "toggle inclusion of tag objects" +msgstr "bao gồm các đối tượng thẻ" + +msgid "toggle inclusion of tree objects" +msgstr "bao gồm các đối tượng cây" + +msgid "toggle pruning of uninteresting paths" +msgstr "lược bỏ những đường dẫn ít ý nghĩa" + +msgid "read a pattern list over stdin" +msgstr "đọc các mẫu từ stdin" + #, c-format msgid "commit %s is not marked reachable" msgstr "lần chuyển giao %s chưa được đánh dấu là tiếp cận được" @@ -22579,6 +22740,10 @@ msgstr "lỗi: " msgid "warning: " msgstr "cảnh báo: " +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() gặp lỗi '%s' (%d)\n" + msgid "Fetching objects" msgstr "Đang lấy về các đối tượng" @@ -23536,3 +23701,21 @@ msgstr "Bỏ qua %s với hậu tố sao lưu '%s'.\n" #, perl-format msgid "Do you really want to send %s? [y|N]: " msgstr "Bạn có thực sự muốn gửi %s? [y|N](có/KHÔNG): " + +#, c-format +#~ msgid "Could not find remote branch %s to clone." +#~ msgstr "Không tìm thấy nhánh máy chủ %s để nhân bản (clone)." + +#, c-format +#~ msgid "merging cannot continue; got unclean result of %d" +#~ msgstr "không thể tiếp tục hoà trộn; kết quả không hoàn toàn %d" + +#~ msgid "git repack []" +#~ msgstr "git repack []" + +#~ msgid "--onto and --advance are incompatible" +#~ msgstr "--onto và --advance xung khắc" + +#, c-format +#~ msgid "key '%s' of pattern had no '*'" +#~ msgstr "khóa '%s' của mẫu k có '*'" From cc2eb7ece2c50f49ff06cfbee2b578023c1205da Mon Sep 17 00:00:00 2001 From: Alexander Shopov Date: Sat, 1 Mar 2025 22:28:22 +0100 Subject: [PATCH 004/259] l10n: bg.po: Updated Bulgarian translation (5836t) Signed-off-by: Alexander Shopov --- po/bg.po | 446 ++++++++++++++++++++++++++++++++++++++----------------- 1 file changed, 308 insertions(+), 138 deletions(-) diff --git a/po/bg.po b/po/bg.po index 2fc6c2b60ca2e9..37d127faac7169 100644 --- a/po/bg.po +++ b/po/bg.po @@ -1,7 +1,7 @@ # Bulgarian translation of git po-file. -# Copyright (C) 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024 Alexander Shopov . +# Copyright (C) 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024, 2025 Alexander Shopov . # This file is distributed under the same license as the git package. -# Alexander Shopov , 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024. +# Alexander Shopov , 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024, 2025. # ======================== # DICTIONARY TO MERGE IN GIT GUI # ------------------------ @@ -233,6 +233,7 @@ # cache-tree кеша на обектите-дървета # acquire lock придобивам ключалка # detached отделѐн, несвързан +# revision walk обхождане на версиите # # ------------------------ # „$var“ - може да не сработва за shell има gettext и eval_gettext - проверка - намират се лесно по „$ @@ -261,8 +262,8 @@ msgid "" msgstr "" "Project-Id-Version: git 2.48\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-27 22:37+0100\n" -"PO-Revision-Date: 2024-12-27 22:40+0100\n" +"POT-Creation-Date: 2025-03-05 22:57+0000\n" +"PO-Revision-Date: 2025-03-06 09:15+0100\n" "Last-Translator: Alexander Shopov \n" "Language-Team: Bulgarian \n" "Language: bg\n" @@ -2655,6 +2656,18 @@ msgstr "git archive: протоколна грешка" msgid "git archive: expected a flush" msgstr "git archive: очакваше се изчистване на буферите чрез „flush“" +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=БРОЙ] [--[no-]sparse]" + +msgid "problem loading sparse-checkout" +msgstr "проблем при зареждане на частично хранилище" + +msgid "Minimum number of objects to request at a time" +msgstr "Минимален БРОЙ обекти заявявани наведнъж" + +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "Ограничаване на липсващите обекти до текущото частично хранилище" + msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" "checkout] [--first-parent] [ [...]] [--] [...]" @@ -3434,10 +3447,6 @@ msgstr "" msgid "git version:\n" msgstr "версия на git:\n" -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "грешка при изпълнението на „uname()“ — „%s“ (%d)\n" - msgid "compiler info: " msgstr "компилатор: " @@ -4466,8 +4475,98 @@ msgstr "" "Настройката „clean.requireForce“ е зададена, което изисква опцията „-f“. " "Няма да се извърши изчистване" -msgid "git clone [] [--] []" -msgstr "git clone [ОПЦИЯ…] [--] ХРАНИЛИЩЕ [ДИРЕКТОРИЯ]" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "" +"ПРЕДУПРЕЖДЕНИЕ: не може да се добави алтернативен източник на „%s“: %s\n" + +#, c-format +msgid "failed to stat '%s'" +msgstr "не може да бъде получена информация чрез „stat“ за „%s“" + +#, c-format +msgid "%s exists and is not a directory" +msgstr "„%s“ съществува и не е директория" + +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "„%s“ е символна връзка, не може да се клонира с опцията „--local“" + +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "неуспешно итериране по „%s“" + +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "" +"символната връзка „%s“ съществува, не може да се клонира с опцията „--local“" + +#, c-format +msgid "failed to unlink '%s'" +msgstr "неуспешно изтриване на „%s“" + +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "твърдата връзка не може да се провери при „%s“" + +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "твърдата връзка е различна от източника „%s“" + +#, c-format +msgid "failed to create link '%s'" +msgstr "връзката „%s“ не може да бъде създадена" + +#, c-format +msgid "failed to copy file to '%s'" +msgstr "файлът не може да бъде копиран като „%s“" + +#, c-format +msgid "failed to iterate over '%s'" +msgstr "неуспешно итериране по „%s“" + +#, c-format +msgid "done.\n" +msgstr "действието завърши.\n" + +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Клонирането бе успешно за разлика от подготовката на работното дърво\n" +"за определен клон. Все пак може да проверите кои файлове и от кой\n" +"клон в момента са изтеглени с командата „git status“. Може да\n" +"завършите изтеглянето на клона с командата:\n" +"\n" +" git restore --source=HEAD :/\n" + +msgid "remote did not send all necessary objects" +msgstr "отдалеченото хранилище не изпрати всички необходими обекти." + +#, c-format +msgid "unable to update %s" +msgstr "обектът „%s“ не може да бъде обновен" + +msgid "failed to initialize sparse-checkout" +msgstr "частичното изтегляне не може да се инициализира" + +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "" +"указателят „HEAD“ от отдалеченото хранилище сочи към нещо, което не " +"съществува. Изтегляне не може да се извърши" + +msgid "unable to checkout working tree" +msgstr "работното дърво не може да бъде подготвено" + +msgid "unable to write parameters to config file" +msgstr "настройките не може да бъдат записани в конфигурационния файл" + +msgid "cannot repack to clean up" +msgstr "не може да се извърши пакетиране за изчистване на файловете" + +msgid "cannot unlink temporary alternates file" +msgstr "временният файл за алтернативни обекти не може да бъде изтрит" msgid "don't clone shallow repository" msgstr "без клониране на плитко хранилище" @@ -4521,6 +4620,9 @@ msgstr "използване на това ИМЕ вместо „origin“ пр msgid "checkout instead of the remote's HEAD" msgstr "изтегляне на този КЛОН, а не соченият от отдалечения указател „HEAD“" +msgid "clone single revision and check out" +msgstr "клониране на единствена ВЕРСИЯ и изтегляне в работната директория" + msgid "path to git-upload-pack on the remote" msgstr "път към командата „git-upload-pack“ на отдалеченото хранилище" @@ -4544,9 +4646,8 @@ msgstr "" "клониране само на един клон — или сочения от отдалечения „HEAD“, или изрично " "зададения с „--branch“" -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "" -"без клониране на етикети, като последващите доставяния няма да ги следят" +msgid "clone tags, and make later fetches not to follow them" +msgstr "клониране на етикети, като последващите доставяния няма да ги следят" msgid "any cloned submodules will be shallow" msgstr "всички клонирани подмодули ще са плитки" @@ -4590,104 +4691,8 @@ msgid "a URI for downloading bundles before fetching from origin remote" msgstr "" "АДРЕС за доставяне на пратки на git преди доставяне от отдалеченото хранилище" -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "" -"ПРЕДУПРЕЖДЕНИЕ: не може да се добави алтернативен източник на „%s“: %s\n" - -#, c-format -msgid "failed to stat '%s'" -msgstr "не може да бъде получена информация чрез „stat“ за „%s“" - -#, c-format -msgid "%s exists and is not a directory" -msgstr "„%s“ съществува и не е директория" - -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "„%s“ е символна връзка, не може да се клонира с опцията „--local“" - -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "неуспешно итериране по „%s“" - -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "" -"символната връзка „%s“ съществува, не може да се клонира с опцията „--local“" - -#, c-format -msgid "failed to unlink '%s'" -msgstr "неуспешно изтриване на „%s“" - -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "твърдата връзка не може да се провери при „%s“" - -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "твърдата връзка е различна от източника „%s“" - -#, c-format -msgid "failed to create link '%s'" -msgstr "връзката „%s“ не може да бъде създадена" - -#, c-format -msgid "failed to copy file to '%s'" -msgstr "файлът не може да бъде копиран като „%s“" - -#, c-format -msgid "failed to iterate over '%s'" -msgstr "неуспешно итериране по „%s“" - -#, c-format -msgid "done.\n" -msgstr "действието завърши.\n" - -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Клонирането бе успешно за разлика от подготовката на работното дърво\n" -"за определен клон. Все пак може да проверите кои файлове и от кой\n" -"клон в момента са изтеглени с командата „git status“. Може да\n" -"завършите изтеглянето на клона с командата:\n" -"\n" -" git restore --source=HEAD :/\n" - -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "" -"Клонът „%s“ от отдалеченото хранилище, което клонирате,\n" -"и който следва да бъде изтеглен, не съществува." - -msgid "remote did not send all necessary objects" -msgstr "отдалеченото хранилище не изпрати всички необходими обекти." - -#, c-format -msgid "unable to update %s" -msgstr "обектът „%s“ не може да бъде обновен" - -msgid "failed to initialize sparse-checkout" -msgstr "частичното изтегляне не може да се инициализира" - -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "" -"указателят „HEAD“ от отдалеченото хранилище сочи към нещо, което не " -"съществува. Изтегляне не може да се извърши" - -msgid "unable to checkout working tree" -msgstr "работното дърво не може да бъде подготвено" - -msgid "unable to write parameters to config file" -msgstr "настройките не може да бъдат записани в конфигурационния файл" - -msgid "cannot repack to clean up" -msgstr "не може да се извърши пакетиране за изчистване на файловете" - -msgid "cannot unlink temporary alternates file" -msgstr "временният файл за алтернативни обекти не може да бъде изтрит" +msgid "git clone [] [--] []" +msgstr "git clone [ОПЦИЯ…] [--] ХРАНИЛИЩЕ [ДИРЕКТОРИЯ]" msgid "Too many arguments." msgstr "Прекалено много аргументи." @@ -4796,6 +4801,10 @@ msgstr "отдалеченият транспорт върна грешка" msgid "Remote branch %s not found in upstream %s" msgstr "Отдалеченият клон „%s“ липсва в клонираното хранилище „%s“" +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "Отдалечената версия „%s“ липсва в клонираното хранилище „%s“" + msgid "You appear to have cloned an empty repository." msgstr "Изглежда клонирахте празно хранилище." @@ -4975,7 +4984,7 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree: не може да се прочете" msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -4985,7 +4994,7 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a|--interactive|--patch] [-s] [-v] [-u РЕЖИМ] [--amend]\n" +"git commit [-a|--interactive|--patch] [-s] [-v] [-u[РЕЖИМ]] [--amend]\n" " [--dry-run] [(-c|-C|--squash) ПОДАВАНЕ |--fixup [(amend|" "reword):]ПОДАВАНЕ]\n" " [-F ФАЙЛ|-m СЪОБЩЕНИЕ] [--reset-author] [--allow-empty]\n" @@ -6424,8 +6433,8 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Изпълнете\n" "\n" @@ -6435,7 +6444,7 @@ msgstr "" "„remote.%s.followRemoteHEAD, ако не искате тези съобщения.\n" "Изпълнението на\n" "\n" -" git config set remote.%s.followRemoteHEAD %s\n" +" git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s\n" "\n" "ще изключи предупреждението, докато отдалеченият указател HEAD не\n" "започне да сочи нещо друго." @@ -7124,6 +7133,9 @@ msgstr "" msgid "repack all other packs except the largest pack" msgstr "препакетиране на всичко без най-големия пакет" +msgid "pack prefix to store a pack containing pruned objects" +msgstr "префикс на имената на пакетите за окастрени обекти" + #, c-format msgid "failed to parse gc.logExpiry value %s" msgstr "неразпозната стойност на „gc.logExpiry“ %s" @@ -7943,6 +7955,10 @@ msgstr "" msgid "Cannot come back to cwd" msgstr "Процесът не може да се върне към предишната работна директория" +#, c-format +msgid "bad --pack_header: %s" +msgstr "неправилна стойност за „--pack_header“: „%s“" + #, c-format msgid "bad %s" msgstr "неправилна стойност „%s“" @@ -8835,10 +8851,6 @@ msgstr "непозната опция за стратегия: -X%s" msgid "malformed input line: '%s'." msgstr "входен ред с неправилен формат: „%s“." -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "сливането не може да продължи — %d-тото завърши с грешка" - msgid "git merge [] [...]" msgstr "git merge [ОПЦИЯ…] [ПОДАВАНЕ…]" @@ -9684,6 +9696,13 @@ msgstr "" "git pack-objects [ОПЦИЯ…] ПРЕФИКС_НА_ИМЕТО [< СПИСЪК_С_УКАЗАТЕЛИ|< " "СПИСЪК_С_ОБЕКТИ]" +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "неправилна опция „--name-hash-version“: %d" + +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "текущо опцията „--write-bitmap-index“ изисква „--name-hash-version=1“" + #, c-format msgid "" "write_reuse_object: could not locate %s, expected at offset % in " @@ -10025,6 +10044,11 @@ msgstr "протокол" msgid "exclude any configured uploadpack.blobpackfileuri with this protocol" msgstr "без ползване на настройки „uploadpack.blobpackfileuri“ с този протокол" +msgid "use the specified name-hash function to group similar objects" +msgstr "" +"използване на указаната функция за контролни суми за групиране на подобните " +"обекти" + #, c-format msgid "delta chain depth %d is too deep, forcing %d" msgstr "веригата с разлики е прекалено дълбока — %d, ще се ползва %d" @@ -11304,8 +11328,8 @@ msgstr "не е указан журнал с подавания за изтри msgid "invalid ref format: %s" msgstr "неправилен формат на указател: %s" -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format=ФОРМАТ [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format=ФОРМАТ [--no-reflog] [--dry-run]" msgid "git refs verify [--strict] [--verbose]" msgstr "git refs verify [--strict] [--verbose]" @@ -11316,6 +11340,9 @@ msgstr "указване на форма̀та за указател, към к msgid "perform a non-destructive dry-run" msgstr "пробно изпълнение — без промяна на данни" +msgid "drop reflogs entirely during the migration" +msgstr "журналът на указателите да се изтрие изцяло по време на миграцията" + msgid "missing --ref-format=" msgstr "липсва опцията --ref-format=ФОРМАТ" @@ -11786,8 +11813,14 @@ msgstr "Никой от адресите, които не са за изтлас msgid "be verbose; must be placed before a subcommand" msgstr "повече подробности. Поставя се пред подкоманда" -msgid "git repack []" -msgstr "git repack [ОПЦИЯ…]" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=БРОЙ] [--depth=БРОЙ] [--threads=БРОЙ] [--keep-pack=ИМЕ_НА_ПАКЕТ]\n" +"[--write-midx] [--name-hash-version=БРОЙ]" msgid "" "Incremental repacks are incompatible with bitmap indexes. Use\n" @@ -11875,6 +11908,11 @@ msgid "pass --no-reuse-object to git-pack-objects" msgstr "" "подаване на опцията „--no-reuse-object“ на командата „git-pack-objects“" +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "" +"укажете функция за контролни суми за групиране на подобните обекти по път" + msgid "do not run git-update-server-info" msgstr "без изпълнение на командата „git-update-server-info“" @@ -11927,9 +11965,6 @@ msgstr "откриване на геометрична прогресия с ч msgid "write a multi-pack index of the resulting packs" msgstr "запазване на многопакетен индекс за създадените пакети" -msgid "pack prefix to store a pack containing pruned objects" -msgstr "префикс на имената на пакетите за окастрени обекти" - msgid "pack prefix to store a pack containing filtered out objects" msgstr "префикс на имената на пакетите за филтрирани обекти" @@ -12146,9 +12181,6 @@ msgstr "опцията „-l“ приема точно един шаблон" msgid "need some commits to replay" msgstr "необходимо е да има подавания за прилагане отново" -msgid "--onto and --advance are incompatible" -msgstr "опциите „--onto“ и „--advance“ са несъвместими" - msgid "all positive revisions given must be references" msgstr "всички зададени положителни версии трябва да са указатели" @@ -14852,6 +14884,9 @@ msgstr "Внасяне на хранилище на GNU Arch в Git" msgid "Create an archive of files from a named tree" msgstr "Създаване на архив с файловете от именовано дърво" +msgid "Download missing objects in a partial clone" +msgstr "Изтегляне на липсващите обекти в частично хранилище" + msgid "Use binary search to find the commit that introduced a bug" msgstr "Двоично търсене на промяната, която е причинила грешка" @@ -16852,6 +16887,12 @@ msgstr "неправилен аргумент към „%s“" msgid "invalid regex given to -I: '%s'" msgstr "неправилен регулярен израз подаден към „-I“: „%s“" +msgid "-G requires a non-empty argument" +msgstr "опцията „-G“ изисква аргумент" + +msgid "-S requires a non-empty argument" +msgstr "опцията „-S“ изисква аргумент" + #, c-format msgid "failed to parse --submodule option parameter: '%s'" msgstr "неразпознат параметър към опцията „--submodule“: „%s“" @@ -19122,6 +19163,10 @@ msgstr "съдържанието на файловете „%s“ и „%s“ е msgid "unable to write file %s" msgstr "файлът „%s“ не може да бъде записан" +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "смаляващият се файл „%s“ не може да бъде записван" + #, c-format msgid "unable to set permission to '%s'" msgstr "права̀та за достъп до „%s“ не може да бъдат зададени" @@ -19722,6 +19767,52 @@ msgstr "непознат флаг „%c“" msgid "unknown non-ascii option in string: `%s'" msgstr "непозната стойност извън „ascii“ в низа: „%s“" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[=<%s>]" +msgstr "[=%s]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[<%s>]" +msgstr "[%s]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid " <%s>" +msgstr " %s" + msgid "..." msgstr "…" @@ -19808,6 +19899,21 @@ msgstr "неправилна булева стойност „%s“ за „%s msgid "failed to parse %s" msgstr "„%s“ не може да бъде анализиран" +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "неуспешно обхождане на дъщерните елементи на дървото „%s“: то липсва" + +#, c-format +msgid "failed to find object %s" +msgstr "обектът „%s“ липсва" + +#, c-format +msgid "failed to find tag %s" +msgstr "етикетът „%s“ липсва" + +msgid "failed to setup revision walk" +msgstr "неуспешно настройване на обхождането на версиите" + #, c-format msgid "Could not make %s writable by group" msgstr "Не може да се дадат права̀ за запис в директорията „%s“ на групата" @@ -19957,6 +20063,22 @@ msgstr "" msgid "could not fetch %s from promisor remote" msgstr "„%s“ не може да се достави от гарантиращото хранилище" +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "има хранилище с име „%s“, но адресът му сочи към „%s“, а не към „%s“" + +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "непозната стойност „%s“ за настройката „%s“" + +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "непознат елемент „%s“ в информацията за отдалеченото хранилище" + +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "липсва приетото вече хранилище-гарант „%s“" + msgid "object-info: expected flush after arguments" msgstr "object-info: след аргументите се очаква изчистване на буферите" @@ -20802,6 +20924,14 @@ msgstr "името на указател „%s“ е символен указа msgid "invalid refspec '%s'" msgstr "неправилен указател: „%s“" +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "шаблонът „%s“ не съдържа „*“" + +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "заместителят „%s“ не съдържа „*“" + #, c-format msgid "invalid quoting in push-option value: '%s'" msgstr "" @@ -20929,6 +21059,28 @@ msgstr "remote-curl: опит за доставяне без локално хр msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl: непозната команда „%s“ от git" +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"изчитането на отдалеченото хранилище от „%s/%s“ предстои да бъде " +"премахнато.\n" +"Ако все още ползвате директорията „remotes/“, препоръчваме да я мигрирате\n" +"към следящи директории на база настройки чрез командата:\n" +"\n" +" git remote rename %s %s\n" +"\n" +"Ако не може поради някаква причина, молим да ни известите за нея с\n" +"е-писмо до пощенския списък: ." + #, c-format msgid "config remote shorthand cannot begin with '/': %s" msgstr "" @@ -20964,14 +21116,6 @@ msgstr "„%s“ обикновено следи „%s“, а не „%s“" msgid "%s tracks both %s and %s" msgstr "„%s“ следи както „%s“, така и „%s“" -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "ключ „%s“ на шаблона не съдържа „*“" - -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "стойност „%s“ на шаблона не съдържа „*“" - #, c-format msgid "src refspec %s does not match any" msgstr "указателят на версия-източник „%s“ не съвпада с никой обект" @@ -22945,6 +23089,28 @@ msgstr "" "какъв брой записи в кеша на обектите-дървета да се отбележат като невалидни " "(стандартно е 0)" +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk ОПЦИЯ… -- ОПЦИЯ_ЗА_ВЕРСИИ…" + +msgid "toggle inclusion of blob objects" +msgstr "превключване на поместването на обекти-BLOB" + +msgid "toggle inclusion of commit objects" +msgstr "превключване на поместването на обекти-подавания" + +msgid "toggle inclusion of tag objects" +msgstr "превключване на поместването на обекти-етикети" + +msgid "toggle inclusion of tree objects" +msgstr "превключване на поместването на обекти-дърво" + +msgid "toggle pruning of uninteresting paths" +msgstr "" +"превключване на окастрянето на пътищата, които не представляват интерес" + +msgid "read a pattern list over stdin" +msgstr "изчитане на списък с шаблони от стандартния вход" + #, c-format msgid "commit %s is not marked reachable" msgstr "подаването „%s“ не е отбелязано като достижимо" @@ -23581,6 +23747,10 @@ msgstr "грешка: " msgid "warning: " msgstr "предупреждение: " +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "грешка при изпълнението на „uname()“ — „%s“ (%d)\n" + msgid "Fetching objects" msgstr "Доставяне на обектите" From 60b9a254b67342955208cb514520fd1b59cb7ed4 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jean-No=C3=ABl=20Avila?= Date: Sat, 1 Mar 2025 15:02:22 +0100 Subject: [PATCH 005/259] l10n: fr: 2.49 round 2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Jean-Noël Avila --- po/fr.po | 452 +++++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 321 insertions(+), 131 deletions(-) diff --git a/po/fr.po b/po/fr.po index 64a75aecf5859c..8e3339030761a1 100644 --- a/po/fr.po +++ b/po/fr.po @@ -87,8 +87,8 @@ msgid "" msgstr "" "Project-Id-Version: git\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-23 18:57+0000\n" -"PO-Revision-Date: 2024-12-29 18:26+0100\n" +"POT-Creation-Date: 2025-03-05 22:57+0000\n" +"PO-Revision-Date: 2025-03-06 16:46+0100\n" "Last-Translator: Cédric Malard \n" "Language-Team: Jean-Noël Avila \n" "Language: fr\n" @@ -2452,6 +2452,18 @@ msgstr "git archive : erreur de protocole" msgid "git archive: expected a flush" msgstr "git archive : vidage attendu" +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +msgid "problem loading sparse-checkout" +msgstr "problème lors du chargement de l'extraction clairsemée" + +msgid "Minimum number of objects to request at a time" +msgstr "Nombre minimum d'objets à demander à chaque fois" + +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "Restreindre les objets manquants à l'extraction clairsemée actuelle" + msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" "checkout] [--first-parent] [ [...]] [--] [...]" @@ -3217,10 +3229,6 @@ msgstr "" msgid "git version:\n" msgstr "version git ::\n" -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "échec de uname() avec l'erreur '%s' (%d)\n" - msgid "compiler info: " msgstr "info compilateur : " @@ -4243,8 +4251,93 @@ msgid "clean.requireForce is true and -f not given: refusing to clean" msgstr "" "clean.requireForce positionné est true et -f non fourni ; refus de nettoyer" -msgid "git clone [] [--] []" -msgstr "git clone [] [--] []" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "info : impossible d'ajouter une alternative pour '%s' : %s\n" + +#, c-format +msgid "failed to stat '%s'" +msgstr "échec du stat de '%s'" + +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s existe et n'est pas un répertoire" + +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "'%s' est un lien symbolique, refus de cloner avec --local" + +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "échec du démarrage un itérateur sur '%s'" + +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "le lien symbolique '%s' existe, refus de cloner avec --local" + +#, c-format +msgid "failed to unlink '%s'" +msgstr "échec pour délier '%s'" + +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "le lien dur ne peut pas être vérifié à '%s'" + +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "le lien dur est différent de la source à '%s'" + +#, c-format +msgid "failed to create link '%s'" +msgstr "échec de la création du lien '%s'" + +#, c-format +msgid "failed to copy file to '%s'" +msgstr "échec de la copie vers '%s'" + +#, c-format +msgid "failed to iterate over '%s'" +msgstr "échec de l'itération sur '%s'" + +#, c-format +msgid "done.\n" +msgstr "fait.\n" + +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Le clone a réussi, mais l'extraction a échoué.\n" +"Vous pouvez inspecter ce qui a été extrait avec 'git status'\n" +"et réessayer avec 'git restore --source=HEAD :/'\n" + +msgid "remote did not send all necessary objects" +msgstr "le serveur distant n'a pas envoyé tous les objets nécessaires" + +#, c-format +msgid "unable to update %s" +msgstr "impossible de mettre à jour %s" + +msgid "failed to initialize sparse-checkout" +msgstr "échec lors de l'initialisation l'extraction clairsemée" + +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "" +"la HEAD distante réfère à une référence non existante, impossible de " +"l'extraire" + +msgid "unable to checkout working tree" +msgstr "impossible d'extraire la copie de travail" + +msgid "unable to write parameters to config file" +msgstr "impossible d'écrire les paramètres dans le fichier de configuration" + +msgid "cannot repack to clean up" +msgstr "impossible de remballer pour nettoyer" + +msgid "cannot unlink temporary alternates file" +msgstr "impossible de délier le fichier temporaire alternates" msgid "don't clone shallow repository" msgstr "ne pas cloner un dépôt superficiel" @@ -4297,6 +4390,9 @@ msgstr "utiliser au lieu de 'origin' pour suivre la branche amont" msgid "checkout instead of the remote's HEAD" msgstr "extraire au lieu de la HEAD du répertoire distant" +msgid "clone single revision and check out" +msgstr "cloner la révision unique et l'extraire" + msgid "path to git-upload-pack on the remote" msgstr "chemin vers git-upload-pack sur le serveur distant" @@ -4318,10 +4414,9 @@ msgstr "approfondit l'historique d'un clone superficiel en excluant une ref" msgid "clone only one branch, HEAD or --branch" msgstr "cloner seulement une branche, HEAD ou --branch" -msgid "don't clone any tags, and make later fetches not to follow them" +msgid "clone tags, and make later fetches not to follow them" msgstr "" -"ne pas cloner les tags et indiquer aux récupérations futures de ne pas le " -"faire" +"cloner les tags et indiquer aux récupérations futures de ne pas le faire" msgid "any cloned submodules will be shallow" msgstr "tous les sous-modules clonés seront superficiels" @@ -4367,97 +4462,8 @@ msgstr "" "un URI pour télécharger des paquets avant de récupérer depuis le distant " "d'origine" -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "info : impossible d'ajouter une alternative pour '%s' : %s\n" - -#, c-format -msgid "failed to stat '%s'" -msgstr "échec du stat de '%s'" - -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s existe et n'est pas un répertoire" - -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "'%s' est un lien symbolique, refus de cloner avec --local" - -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "échec du démarrage un itérateur sur '%s'" - -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "le lien symbolique '%s' existe, refus de cloner avec --local" - -#, c-format -msgid "failed to unlink '%s'" -msgstr "échec pour délier '%s'" - -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "le lien dur ne peut pas être vérifié à '%s'" - -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "le lien dur est différent de la source à '%s'" - -#, c-format -msgid "failed to create link '%s'" -msgstr "échec de la création du lien '%s'" - -#, c-format -msgid "failed to copy file to '%s'" -msgstr "échec de la copie vers '%s'" - -#, c-format -msgid "failed to iterate over '%s'" -msgstr "échec de l'itération sur '%s'" - -#, c-format -msgid "done.\n" -msgstr "fait.\n" - -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Le clone a réussi, mais l'extraction a échoué.\n" -"Vous pouvez inspecter ce qui a été extrait avec 'git status'\n" -"et réessayer avec 'git restore --source=HEAD :/'\n" - -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "Impossible de trouver la branche distante '%s' à cloner." - -msgid "remote did not send all necessary objects" -msgstr "le serveur distant n'a pas envoyé tous les objets nécessaires" - -#, c-format -msgid "unable to update %s" -msgstr "impossible de mettre à jour %s" - -msgid "failed to initialize sparse-checkout" -msgstr "échec lors de l'initialisation l'extraction clairsemée" - -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "" -"la HEAD distante réfère à une référence non existante, impossible de " -"l'extraire" - -msgid "unable to checkout working tree" -msgstr "impossible d'extraire la copie de travail" - -msgid "unable to write parameters to config file" -msgstr "impossible d'écrire les paramètres dans le fichier de configuration" - -msgid "cannot repack to clean up" -msgstr "impossible de remballer pour nettoyer" - -msgid "cannot unlink temporary alternates file" -msgstr "impossible de délier le fichier temporaire alternates" +msgid "git clone [] [--] []" +msgstr "git clone [] [--] []" msgid "Too many arguments." msgstr "Trop d'arguments." @@ -4563,6 +4569,10 @@ msgstr "le transport distant a retourné une erreur" msgid "Remote branch %s not found in upstream %s" msgstr "La branche distante %s n'a pas été trouvée dans le dépôt amont %s" +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "La révision distante %s n'a pas été trouvée dans le dépôt amont %s" + msgid "You appear to have cloned an empty repository." msgstr "Vous semblez avoir cloné un dépôt vide." @@ -4737,7 +4747,7 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree : échec de la lecture" msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -4747,14 +4757,14 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" -"reword):])]\n" +"reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" " [--allow-empty-message] [--no-verify] [-e] [--author=]\n" " [--date=] [--cleanup=] [--[no-]status]\n" " [-i | -o] [--pathspec-from-file= [--pathspec-file-nul]]\n" -" [(--trailer [(=|:)])...] [-S[]]\n" +" [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgid "git status [] [--] [...]" @@ -6168,13 +6178,14 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Lancez 'git remote set-head %s %s' pour suivre la modification, ou\n" "réglez l'option de configuration 'remote.%s.followRemoteHEAD' à une\n" "valeur différente si vous ne souhaitez pas voir ce message. Lancer\n" -"spécifiquement 'git config set remote.%s.followRemoteHEAD %s'\n" +"spécifiquement 'git config set remote.%s.followRemoteHEAD warn-if-not-branch-" +"%s'\n" "va désactiver l'alerte jusqu'à ce que le distant change HEAD." msgid "multiple branches detected, incompatible with --set-upstream" @@ -6860,6 +6871,9 @@ msgstr "" msgid "repack all other packs except the largest pack" msgstr "recompacter tous les autres paquets excepté le plus gros paquet" +msgid "pack prefix to store a pack containing pruned objects" +msgstr "préfixe de paquet pour stocker un paquet contenant les objets élagués" + #, c-format msgid "failed to parse gc.logExpiry value %s" msgstr "impossible d'analyser gc.logExpiry %s" @@ -7673,6 +7687,10 @@ msgstr "" msgid "Cannot come back to cwd" msgstr "Impossible de revenir au répertoire de travail courant" +#, c-format +msgid "bad --pack_header: %s" +msgstr "mauvais --pack_header : %s" + #, c-format msgid "bad %s" msgstr "mauvais %s" @@ -8561,10 +8579,6 @@ msgstr "option de stratégie inconnue : -X%s" msgid "malformed input line: '%s'." msgstr "ligne en entrée malformée : '%s'." -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "la fusion ne peut pas continuer ; résultat non propre retourné %d" - msgid "git merge [] [...]" msgstr "git merge [] [...]" @@ -9397,6 +9411,13 @@ msgstr "" "git pack-objects [] [< | < ]" +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "option --name-hash-version invalide : %d" + +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "actuellement, --write-bitmap-index nécessite --name-hash-version=1" + #, c-format msgid "" "write_reuse_object: could not locate %s, expected at offset % in " @@ -9730,6 +9751,11 @@ msgstr "protocole" msgid "exclude any configured uploadpack.blobpackfileuri with this protocol" msgstr "exclure tout uploadpack.blobpackfileuri configuré avec ce protocole" +msgid "use the specified name-hash function to group similar objects" +msgstr "" +"utiliser la fonction d'empreinte de nom spécifiée pour grouper les objets " +"similaires" + #, c-format msgid "delta chain depth %d is too deep, forcing %d" msgstr "la profondeur %d de chaîne de delta est trop grande, forcée à %d" @@ -10992,8 +11018,8 @@ msgstr "pas de journal de références à supprimer spécifié" msgid "invalid ref format: %s" msgstr "format de référence invalide : %s" -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format= [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format= [--no-reflog] [--dry-run]" msgid "git refs verify [--strict] [--verbose]" msgstr "git refs verify [--strict] [--verbose]" @@ -11004,6 +11030,9 @@ msgstr "spécifier le format de réference vers lequel convertir" msgid "perform a non-destructive dry-run" msgstr "faire l'action en mode simulé non destructif" +msgid "drop reflogs entirely during the migration" +msgstr "abandonner complètement le reflog pendant la migration" + msgid "missing --ref-format=" msgstr "--ref-format= manquant" @@ -11475,8 +11504,14 @@ msgstr "Pas de suppression de toutes les URLs non-push" msgid "be verbose; must be placed before a subcommand" msgstr "être verbeux : doit être placé avant une sous-commande" -msgid "git repack []" -msgstr "git repack []" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" msgid "" "Incremental repacks are incompatible with bitmap indexes. Use\n" @@ -11557,6 +11592,12 @@ msgstr "passer --no-reuse-delta à git-pack-objects" msgid "pass --no-reuse-object to git-pack-objects" msgstr "passer --no-reuse-object à git-pack-objects" +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "" +"spécifier la verison d'empreinte de nom à utiliser pour grouper les objets " +"similaires par chemin" + msgid "do not run git-update-server-info" msgstr "ne pas lancer git-update-server-info" @@ -11606,9 +11647,6 @@ msgstr "trouver une progression géométrique avec un facteur " msgid "write a multi-pack index of the resulting packs" msgstr "écrire un index de multi-paquet des paquets résultants" -msgid "pack prefix to store a pack containing pruned objects" -msgstr "préfixe de paquet pour stocker un paquet contenant les objets élagués" - msgid "pack prefix to store a pack containing filtered out objects" msgstr "préfixe de paquet pour stocker un paquet contenant les objets filtrés" @@ -11825,9 +11863,6 @@ msgstr "-l n'accepte qu'un motifs" msgid "need some commits to replay" msgstr "commits requis pour pouvoir rejouer" -msgid "--onto and --advance are incompatible" -msgstr "--onto et --advance sont incompatibles" - msgid "all positive revisions given must be references" msgstr "toutes les révisions positives fournies doivent être des références" @@ -14539,6 +14574,9 @@ msgstr "Importer dans Git un dépôt GNU Arch" msgid "Create an archive of files from a named tree" msgstr "Créer une archive des fichiers depuis un arbre nommé" +msgid "Download missing objects in a partial clone" +msgstr "Télécharger les objets manquants dans un clone partiel" + msgid "Use binary search to find the commit that introduced a bug" msgstr "Trouver par recherche binaire la modification qui a introduit un bogue" @@ -16531,6 +16569,12 @@ msgstr "argument invalide pour %s" msgid "invalid regex given to -I: '%s'" msgstr "regex invalide fournie à -I : '%s'" +msgid "-G requires a non-empty argument" +msgstr "-G exige un argument non vide" + +msgid "-S requires a non-empty argument" +msgstr "-S exige un argument non vide" + #, c-format msgid "failed to parse --submodule option parameter: '%s'" msgstr "échec de l'analyse du paramètre de l'option --submodule : '%s'" @@ -18766,6 +18810,10 @@ msgstr "les fichiers '%s' et '%s' diffèrent par leur contenu" msgid "unable to write file %s" msgstr "impossible d'écrire le fichier %s" +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "impossible d'écrire le fichier %s qui disparaît répétitivement" + #, c-format msgid "unable to set permission to '%s'" msgstr "impossible de régler les droits de '%s'" @@ -19338,6 +19386,52 @@ msgstr "bascule inconnue « %c »" msgid "unknown non-ascii option in string: `%s'" msgstr "option non-ascii inconnue dans la chaîne : '%s'" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid " <%s>" +msgstr " <%s>" + msgid "..." msgstr "..." @@ -19424,6 +19518,21 @@ msgstr "valeur booléenne d'environnement invalide '%s' pour '%s'" msgid "failed to parse %s" msgstr "échec de l'analyse de %s" +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "échec de parcours des enfants de l'arbre %s : non trouvé" + +#, c-format +msgid "failed to find object %s" +msgstr "échec de la recherche de l'objet %s" + +#, c-format +msgid "failed to find tag %s" +msgstr "impossible de trouver l'étiquette %s" + +msgid "failed to setup revision walk" +msgstr "impossible définir un parcours de révisions" + #, c-format msgid "Could not make %s writable by group" msgstr "Impossible de rendre %s inscriptible pour le groupe" @@ -19581,6 +19690,22 @@ msgstr "un nom de prometteur distant ne peut pas commencer par '/' : %s" msgid "could not fetch %s from promisor remote" msgstr "impossible de récupérer %s depuis le distant de prometteur" +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "distant connu nommé '%s' mais avec l'url '%s' au lieu de '%s'" + +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "valeur inconnue '%s' pour l'option de config '%s'" + +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "élément inconnu '%s' pour l'info de distant" + +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "distant accpté de prometteur '%s' non trouvé" + msgid "object-info: expected flush after arguments" msgstr "object-info : vidage attendu après les arguments" @@ -20417,6 +20542,14 @@ msgstr "" msgid "invalid refspec '%s'" msgstr "spécificateur de réference invalide : '%s'" +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "la valeur '%s' du motif n'a pas de '*'" + +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "le remplacement '%s' n'a pas de '*'" + #, c-format msgid "invalid quoting in push-option value: '%s'" msgstr "citation invalide dans la valeur push-option : '%s'" @@ -20538,6 +20671,28 @@ msgstr "remote-curl : récupération tentée sans dépôt local" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl : commande inconnue '%s' depuis git" +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"lecture depuis \"%s/%s\", dont la suppression est programmée.\n" +"\n" +"Si vous utilisez encore le répertoire \"remotes\", il est recommandé de\n" +"migrer vers une gestion à base de configuration :\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"S'il vous est impossible de migrer, veuillez nous avertir par\n" +"un courriel à ." + #, c-format msgid "config remote shorthand cannot begin with '/': %s" msgstr "" @@ -20573,14 +20728,6 @@ msgstr "%s suit habituellement %s, pas %s" msgid "%s tracks both %s and %s" msgstr "%s suit à la fois %s et %s" -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "la clé '%s' du modèle n'a pas de '*'" - -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "la valeur '%s' du modèle n'a pas de '*'" - #, c-format msgid "src refspec %s does not match any" msgstr "" @@ -22505,6 +22652,27 @@ msgstr "effacer l'arbre de cache avant chaque itération" msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "nombre d'entrées dans l'arbre de cache à invalider (par défaut, 0)" +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk -- " + +msgid "toggle inclusion of blob objects" +msgstr "activer l'inclusion des objets blob" + +msgid "toggle inclusion of commit objects" +msgstr "activer l'inclusion des objets commit" + +msgid "toggle inclusion of tag objects" +msgstr "activer l'inclusion des objets étiquette" + +msgid "toggle inclusion of tree objects" +msgstr "activer l'inclusion des objets arbre" + +msgid "toggle pruning of uninteresting paths" +msgstr "activer l'élagage des chemins inintéressants" + +msgid "read a pattern list over stdin" +msgstr "lire les motifs depuis stdin" + #, c-format msgid "commit %s is not marked reachable" msgstr "le commit %s n'est pas marqué joignable" @@ -23151,6 +23319,10 @@ msgstr "erreur : " msgid "warning: " msgstr "avertissement : " +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "échec de uname() avec l'erreur '%s' (%d)\n" + msgid "Fetching objects" msgstr "Récupération des objets" @@ -24127,6 +24299,24 @@ msgstr "%s sauté avec un suffix de sauvegarde '%s'.\n" msgid "Do you really want to send %s? [y|N]: " msgstr "Souhaitez-vous réellement envoyer %s ?[y|N] : " +#, c-format +#~ msgid "Could not find remote branch %s to clone." +#~ msgstr "Impossible de trouver la branche distante '%s' à cloner." + +#, c-format +#~ msgid "merging cannot continue; got unclean result of %d" +#~ msgstr "la fusion ne peut pas continuer ; résultat non propre retourné %d" + +#~ msgid "git repack []" +#~ msgstr "git repack []" + +#~ msgid "--onto and --advance are incompatible" +#~ msgstr "--onto et --advance sont incompatibles" + +#, c-format +#~ msgid "key '%s' of pattern had no '*'" +#~ msgstr "la clé '%s' du modèle n'a pas de '*'" + #, c-format #~ msgid "preferred pack (%s) is invalid" #~ msgstr "le paquet préféré (%s) est invalide" From 49d9cd8dea520a318412792b50ab4b91ce71b9fe Mon Sep 17 00:00:00 2001 From: Toon Claes Date: Fri, 7 Mar 2025 15:18:07 +0100 Subject: [PATCH 006/259] help: include git-zlib.h to print zlib version In 41f1a8435a (git-compat-util: move include of "compat/zlib.h" into "git-zlib.h", 2025-01-28) some code was refactored to enable easier linking against zlib-ng. This removed `zlib.h` being indirectly included in `help.c`. As this file uses `ZLIB_VERSION` to print the version number of zlib when running git-version(1) with `--build-options`, this resulted in a regression. Include `git-zlib.h` directly into `help.c` to print zlib version information. This brings back the zlib version in the output of `git version --build-options`. Signed-off-by: Toon Claes Reviewed-by: Patrick Steinhardt Signed-off-by: Junio C Hamano --- help.c | 1 + 1 file changed, 1 insertion(+) diff --git a/help.c b/help.c index 5483ea8fd293fc..d3d3adcc5c16c1 100644 --- a/help.c +++ b/help.c @@ -2,6 +2,7 @@ #define DISABLE_SIGN_COMPARE_WARNINGS #include "git-compat-util.h" +#include "git-zlib.h" #include "config.h" #include "builtin.h" #include "exec-cmd.h" From 2b1e0f8cd5058c7e0b85760e54706b9e74cb4766 Mon Sep 17 00:00:00 2001 From: Toon Claes Date: Fri, 7 Mar 2025 15:18:08 +0100 Subject: [PATCH 007/259] help: print zlib-ng version number When building against zlib-ng, the header file `zlib.h` is not included, but `zlib-ng.h` is included instead. It's `zlib.h` that defines `ZLIB_VERSION` and that macro is used to print out zlib version in `git-version(1)` with `--build-options`. But when it's not defined, no version is printed. `zlib-ng.h` defines another macro: `ZLIBNG_VERSION`. Use that macro to print the zlib-ng version in `git version --build-options` when it's set. Otherwise fallback to `ZLIB_VERSION`. Signed-off-by: Toon Claes Helped-by: Patrick Steinhardt Reviewed-by: Patrick Steinhardt Signed-off-by: Junio C Hamano --- help.c | 4 +++- t/t0091-bugreport.sh | 3 ++- 2 files changed, 5 insertions(+), 2 deletions(-) diff --git a/help.c b/help.c index d3d3adcc5c16c1..a08c645fd97c37 100644 --- a/help.c +++ b/help.c @@ -776,7 +776,9 @@ void get_version_info(struct strbuf *buf, int show_build_options) #if defined OPENSSL_VERSION_TEXT strbuf_addf(buf, "OpenSSL: %s\n", OPENSSL_VERSION_TEXT); #endif -#if defined ZLIB_VERSION +#if defined ZLIBNG_VERSION + strbuf_addf(buf, "zlib-ng: %s\n", ZLIBNG_VERSION); +#elif defined ZLIB_VERSION strbuf_addf(buf, "zlib: %s\n", ZLIB_VERSION); #endif } diff --git a/t/t0091-bugreport.sh b/t/t0091-bugreport.sh index e11d819b62a150..e38ca7a9018751 100755 --- a/t/t0091-bugreport.sh +++ b/t/t0091-bugreport.sh @@ -47,7 +47,8 @@ test_expect_success 'sanity check "System Info" section' ' # This is bound to differ from environment to environment, # so we just do some rather high-level checks. grep "uname: ." system && - grep "compiler info: ." system + grep "compiler info: ." system && + grep "zlib." system ' test_expect_success 'dies if file with same name as report already exists' ' From 6167370b87895530ede03f56d28c96cd5f8668e8 Mon Sep 17 00:00:00 2001 From: Peter Krefting Date: Mon, 10 Mar 2025 17:48:34 +0100 Subject: [PATCH 008/259] l10n: sv.po: Update Swedish translation - Update for 2.49.0. - Fix numerous typos found by spelling checker. - Fix more straight quotes. - Harmonize translation of "blob" (to "blob", not "blobb"). - Harmonize translation of "reflog" (to "referenslogg"). Signed-off-by: Peter Krefting --- po/sv.po | 531 ++++++++++++++++++++++++++++++++++++------------------- 1 file changed, 352 insertions(+), 179 deletions(-) diff --git a/po/sv.po b/po/sv.po index 5377b3ece7d6e2..32da53fee0ccab 100644 --- a/po/sv.po +++ b/po/sv.po @@ -1,14 +1,14 @@ # Swedish translations for Git. -# Copyright (C) 2010-2024 Peter Krefting +# Copyright (C) 2010-2025 Peter Krefting # This file is distributed under the same license as the Git package. -# Peter Krefting , 2010-2024. +# Peter Krefting , 2010-2025. # msgid "" msgstr "" -"Project-Id-Version: git 2.48.0\n" +"Project-Id-Version: git 2.49.0\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-30 11:57+0100\n" -"PO-Revision-Date: 2024-12-30 12:03+0100\n" +"POT-Creation-Date: 2025-03-10 17:45+0100\n" +"PO-Revision-Date: 2025-03-10 17:48+0100\n" "Last-Translator: Peter Krefting \n" "Language-Team: Svenska \n" "Language: sv\n" @@ -1477,7 +1477,7 @@ msgid "" "Use '\\!' for literal leading exclamation." msgstr "" "Negativa mönster ignoreras i git-attribut\n" -"Använd '\\!' för att inleda med ett utropstecken." +"Använd ”\\!” för att inleda med ett utropstecken." #, c-format msgid "cannot fstat gitattributes file '%s'" @@ -2060,7 +2060,7 @@ msgid "" "It does not apply to blobs recorded in its index." msgstr "" "Har du handredigerat din patch?\n" -"Den kan inte tillämpas på blobbar som antecknats i dess index." +"Den kan inte tillämpas på blob:ar som antecknats i dess index." msgid "Falling back to patching base and 3-way merge..." msgstr "" @@ -2316,6 +2316,18 @@ msgstr "git archive: protokollfel" msgid "git archive: expected a flush" msgstr "git archive: förväntade en tömning (flush)" +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +msgid "problem loading sparse-checkout" +msgstr "problem med att läsa filen sparse-checkout" + +msgid "Minimum number of objects to request at a time" +msgstr "Minsta antal objekt att be om varje gång" + +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "Begränsa saknade objekt till befintlig ”sparse-checkout”" + msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" "checkout] [--first-parent] [ [...]] [--] [...]" @@ -2921,7 +2933,7 @@ msgid "delete branch (even if not merged)" msgstr "ta bort gren (även om inte helt sammanslagen)" msgid "move/rename a branch and its reflog" -msgstr "flytta/ta bort en gren och dess reflogg" +msgstr "flytta/ta bort en gren och dess referenslogg" msgid "move/rename a branch, even if target exists" msgstr "flytta/ta bort en gren, även om målet finns" @@ -2930,7 +2942,7 @@ msgid "do not output a newline after empty formatted refs" msgstr "skriv inte ut ett nyradstecken efter tomma formaterade referenser" msgid "copy a branch and its reflog" -msgstr "kopiera en gren och dess reflogg" +msgstr "kopiera en gren och dess referenslogg" msgid "copy a branch, even if target exists" msgstr "kopiera en gren, även om målet finns" @@ -2942,7 +2954,7 @@ msgid "show current branch name" msgstr "visa namn på aktuell gren" msgid "create the branch's reflog" -msgstr "skapa grenens reflogg" +msgstr "skapa grenens referenslogg" msgid "edit the description for the branch" msgstr "redigera beskrivning för grenen" @@ -3055,10 +3067,6 @@ msgstr "" msgid "git version:\n" msgstr "git version:\n" -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() misslyckades med felet ”%s” (%d)\n" - msgid "compiler info: " msgstr "kompilatorinfo:" @@ -3332,11 +3340,11 @@ msgid "blob|tree" msgstr "blob|träd" msgid "use a for (--textconv | --filters); Not with 'batch'" -msgstr "använd en för (--textconv | --filters): Inte med 'batch'" +msgstr "använd en för (--textconv | --filters): Inte med ”batch”" #, c-format msgid "'%s=<%s>' needs '%s' or '%s'" -msgstr "'%s=<%s>' behöver '%s' eller '%s'" +msgstr "”%s=<%s>” behöver ”%s” eller ”%s”" msgid "path|tree-ish" msgstr "sökväg|träd-igt" @@ -3887,7 +3895,7 @@ msgid "create/reset and checkout a branch" msgstr "skapa/nollställ och checka ut en gren" msgid "create reflog for new branch" -msgstr "skapa reflogg för ny gren" +msgstr "skapa referenslogg för ny gren" msgid "second guess 'git checkout ' (default)" msgstr "förutspå ”git checkout ” (förval)" @@ -4062,8 +4070,91 @@ msgstr "ta endast bort ignorerade filer" msgid "clean.requireForce is true and -f not given: refusing to clean" msgstr "clean.requireForce är true och -f angavs inte: vägrar städa" -msgid "git clone [] [--] []" -msgstr "git clone [] [--] []" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "info: Kan inte skapa suppleant för ”%s”: %s\n" + +#, c-format +msgid "failed to stat '%s'" +msgstr "misslyckades ta status på ”%s”" + +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s finns och är ingen katalog" + +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "”%s” är en symbolisk länk, vägrar klona med --local" + +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "misslyckades starta iterator över ”%s”" + +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "symbolisk länk ”%s” finns redan, vägrar klona med --local" + +#, c-format +msgid "failed to unlink '%s'" +msgstr "misslyckades ta bort länken ”%s”" + +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "hård länk kan inte kontrolleras vid ”%s”" + +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "hård länk skiljer sig från källan vid ”%s”" + +#, c-format +msgid "failed to create link '%s'" +msgstr "misslyckades skapa länken ”%s”" + +#, c-format +msgid "failed to copy file to '%s'" +msgstr "misslyckades kopiera filen till ”%s”" + +#, c-format +msgid "failed to iterate over '%s'" +msgstr "misslyckades iterera över ”%s”" + +#, c-format +msgid "done.\n" +msgstr "klart.\n" + +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Klonen lyckades, men utcheckningen misslyckades.\n" +"Du kan inspektera det som checkades ut med ”git status”\n" +"och försöka med ”git restore --source=HEAD :/”\n" + +msgid "remote did not send all necessary objects" +msgstr "fjärren sände inte alla nödvändiga objekt" + +#, c-format +msgid "unable to update %s" +msgstr "kan inte uppdatera %s" + +msgid "failed to initialize sparse-checkout" +msgstr "misslyckades initiera sparse-checkout" + +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "HEAD hos fjärren pekar på en obefintlig referens, kan inte checka ut" + +msgid "unable to checkout working tree" +msgstr "kan inte checka ut arbetskatalogen" + +msgid "unable to write parameters to config file" +msgstr "kan inte skriva parametrar till konfigurationsfilen" + +msgid "cannot repack to clean up" +msgstr "kan inte packa om för att städa upp" + +msgid "cannot unlink temporary alternates file" +msgstr "kunde inte ta bort temporär ”alternates”-fil" msgid "don't clone shallow repository" msgstr "klona inte grunt arkiv" @@ -4116,6 +4207,9 @@ msgstr "använd istället för ”origin” för att spåra uppströms" msgid "checkout instead of the remote's HEAD" msgstr "checka ut istället för fjärrens HEAD" +msgid "clone single revision and check out" +msgstr "klona ensam revision och checka ut" + msgid "path to git-upload-pack on the remote" msgstr "sökväg till git-upload-pack på fjärren" @@ -4137,8 +4231,8 @@ msgstr "fördjupa historik för grund klon, exkludera ref" msgid "clone only one branch, HEAD or --branch" msgstr "klona endast en gren, HEAD eller --branch" -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "klona inga taggar och gör att senare hämtningar inte följer dem" +msgid "clone tags, and make later fetches not to follow them" +msgstr "klona taggar och gör att senare hämtningar inte följer dem" msgid "any cloned submodules will be shallow" msgstr "klonade undermoduler kommer vara grunda" @@ -4179,95 +4273,8 @@ msgstr "uri" msgid "a URI for downloading bundles before fetching from origin remote" msgstr "en URI för att hämta buntar innan de hämtas från ursprungsfjärr" -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "info: Kan inte skapa suppleant för ”%s”: %s\n" - -#, c-format -msgid "failed to stat '%s'" -msgstr "misslyckades ta status på ”%s”" - -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s finns och är ingen katalog" - -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "”%s” är en symbolisk länk, vägrar klona med --local" - -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "misslyckades starta iterator över ”%s”" - -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "symbolisk länk ”%s” finns redan, vägrar klona med --local" - -#, c-format -msgid "failed to unlink '%s'" -msgstr "misslyckades ta bort länken ”%s”" - -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "hård länk kan inte kontrolleras vid ”%s”" - -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "hård länk skiljer sig från källan vid ”%s”" - -#, c-format -msgid "failed to create link '%s'" -msgstr "misslyckades skapa länken ”%s”" - -#, c-format -msgid "failed to copy file to '%s'" -msgstr "misslyckades kopiera filen till ”%s”" - -#, c-format -msgid "failed to iterate over '%s'" -msgstr "misslyckades iterera över ”%s”" - -#, c-format -msgid "done.\n" -msgstr "klart.\n" - -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Klonen lyckades, men utcheckningen misslyckades.\n" -"Du kan inspektera det som checkades ut med ”git status”\n" -"och försöka med ”git restore --source=HEAD :/”\n" - -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "Kunde inte hitta fjärrgrenen %s för att klona." - -msgid "remote did not send all necessary objects" -msgstr "fjärren sände inte alla nödvändiga objekt" - -#, c-format -msgid "unable to update %s" -msgstr "kan inte uppdatera %s" - -msgid "failed to initialize sparse-checkout" -msgstr "misslyckades initiera sparse-checkout" - -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "HEAD hos fjärren pekar på en obefintlig referens, kan inte checka ut" - -msgid "unable to checkout working tree" -msgstr "kan inte checka ut arbetskatalogen" - -msgid "unable to write parameters to config file" -msgstr "kan inte skriva parametrar till konfigurationsfilen" - -msgid "cannot repack to clean up" -msgstr "kan inte packa om för att städa upp" - -msgid "cannot unlink temporary alternates file" -msgstr "kunde inte ta bort temporär ”alternates”-fil" +msgid "git clone [] [--] []" +msgstr "git clone [] [--] []" msgid "Too many arguments." msgstr "För många argument." @@ -4367,6 +4374,10 @@ msgstr "fjärrtransport rapporterade fel" msgid "Remote branch %s not found in upstream %s" msgstr "Fjärrgrenen %s hittades inte i uppströmsarkivet %s" +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "Fjärr-revisionen %s hittades inte i uppströmsarkivet %s" + msgid "You appear to have cloned an empty repository." msgstr "Du verkar ha klonat ett tomt arkiv." @@ -4540,7 +4551,7 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree: misslyckades läsa" msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -4550,7 +4561,7 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -4791,7 +4802,7 @@ msgstr "Ange meddelandet en av flaggorna -m eller -F.\n" #, c-format msgid "--author '%s' is not 'Name ' and matches no existing author" msgstr "" -"--author '%s' är inte 'Namn ' och motsvarar ingen befintlig författare" +"--author ”%s” är inte ”Namn ” och motsvarar ingen befintlig författare" #, c-format msgid "Invalid ignored mode '%s'" @@ -5202,7 +5213,7 @@ msgid "writing to stdin is not supported" msgstr "skriva till standard in stöds inte" msgid "writing config blobs is not supported" -msgstr "skriva konfigurations-blobbar stöds inte" +msgstr "skriva konfigurations-blob:ar stöds inte" #, c-format msgid "" @@ -5239,7 +5250,7 @@ msgid "" "section in \"git help worktree\" for details" msgstr "" "--worktree kan inte användas med flera arbetskataloger om inte\n" -"konfigurationsutöknignen worktreeConfig har aktiverats. Läsa stycket\n" +"konfigurationsutökningen worktreeConfig har aktiverats. Läsa stycket\n" "”KONFIGURATIONSFIL” i ”git help worktree” för detaljer" msgid "Other" @@ -5325,7 +5336,7 @@ msgid "editing stdin is not supported" msgstr "redigering av standard in stöds ej" msgid "editing blobs is not supported" -msgstr "redigering av blobbar stöds ej" +msgstr "redigering av blob:ar stöds ej" #, c-format msgid "cannot create configuration file %s" @@ -5601,7 +5612,7 @@ msgstr "objektet ”%s” som angavs är felaktigt." #, c-format msgid "more than two blobs given: '%s'" -msgstr "mer än två blobbar angavs: ”%s”" +msgstr "mer än två blob:ar angavs: ”%s”" #, c-format msgid "unhandled object '%s' given." @@ -5760,7 +5771,7 @@ msgid "reference parents which are not in fast-export stream by object id" msgstr "referera föräldrar som inte finns i fast-export-ström med objekt-id" msgid "show original object ids of blobs/commits" -msgstr "visa ursprungliga objekt-id för blobbar/incheckningar" +msgstr "visa ursprungliga objekt-id för blob:ar/incheckningar" msgid "label tags with mark ids" msgstr "märk taggar med märke-id" @@ -5931,14 +5942,14 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Kör ”git remote set-head %s %s” för att följa ändringen, eller sätt\n" "konfigurationsflaggan ”remote %s.followRemoteHEAD” till ett annat värde\n" "om du inte vill se det här meddelandet. Du kan specifikt inaktivera\n" "varningen till fjärren ändrar HEAD till något annat genom att köra\n" -"”git config set remote %s.followRemoteHEAD %s”." +"”git config set remote %s.followRemoteHEAD warn-if-not-branch-%s”." msgid "multiple branches detected, incompatible with --set-upstream" msgstr "flera grenar upptäcktes, inkompatibelt med --set-upstream" @@ -6328,11 +6339,11 @@ msgstr "%s: objekt trasigt eller saknas" #, c-format msgid "%s: invalid reflog entry %s" -msgstr "%s: ogiltig reflog-post %s" +msgstr "%s: ogiltig referensloggpost %s" #, c-format msgid "Checking reflog %s->%s" -msgstr "Kontrollerar reflog %s→%s" +msgstr "Kontrollerar referenslogg %s→%s" #, c-format msgid "%s: invalid sha1 pointer %s" @@ -6441,7 +6452,7 @@ msgid "make index objects head nodes" msgstr "gör indexobjekt till huvudnoder" msgid "make reflogs head nodes (default)" -msgstr "gör refloggar till huvudnoder (standard)" +msgstr "gör referensloggar till huvudnoder (standard)" msgid "also consider packs and alternate objects" msgstr "ta även hänsyn till paket och supplerande objekt" @@ -6502,11 +6513,11 @@ msgstr "kunde inte skapa fsmonitor-kaka ”%s”" #, c-format msgid "fsmonitor: cookie_result '%d' != SEEN" -msgstr "fsmonitor: cookie_result '%d' != SEEN" +msgstr "fsmonitor: cookie_result ”%d” != SEEN" #, c-format msgid "could not start IPC thread pool on '%s'" -msgstr "kunde inte starta IPC-trådpol på ”%s”" +msgstr "kunde inte starta IPC-trådpool på ”%s”" msgid "could not start fsmonitor listener thread" msgstr "kunde inte starta fsmonitor-lyssnartråd" @@ -6618,6 +6629,9 @@ msgstr "tvinga gc-körning även om en annan gc kanske körs" msgid "repack all other packs except the largest pack" msgstr "packa om alla paket förutom det största paketet" +msgid "pack prefix to store a pack containing pruned objects" +msgstr "paketprefix att lagra ett paket som innehåller bortrensade objekt" + #, c-format msgid "failed to parse gc.logExpiry value %s" msgstr "misslyckades tolka värdet %s för gc.logExpiry" @@ -7176,7 +7190,7 @@ msgstr "flaggan ”%s” tar inte några argument som inte är flaggor" msgid "" "the '--no-[external-commands|aliases]' options can only be used with '--all'" msgstr "" -"flaggorna '--no-[external-commands|aliases]' kan endast användas med ”--all”" +"flaggorna ”--no-[external-commands|aliases]” kan endast användas med ”--all”" #, c-format msgid "usage: %s%s" @@ -7406,6 +7420,10 @@ msgstr "kunde inte avsluta pack-objects för att packa om lokala länkar" msgid "Cannot come back to cwd" msgstr "Kan inte gå tillbaka till arbetskatalogen (cwd)" +#, c-format +msgid "bad --pack_header: %s" +msgstr "felaktig --pack_header: %s" + #, c-format msgid "bad %s" msgstr "felaktig %s" @@ -7977,7 +7995,7 @@ msgid "suppress duplicate entries" msgstr "undertryck dublettposter" msgid "show sparse directories in the presence of a sparse index" -msgstr "visa glesa kataloger när et glest index existerar" +msgstr "visa glesa kataloger när ett glest index existerar" msgid "" "--format cannot be used with -s, -o, -k, -t, --resolve-undo, --deduplicate, " @@ -8261,10 +8279,6 @@ msgstr "okänd strategiflagga: -X%s" msgid "malformed input line: '%s'." msgstr "felaktig indatarad: ”%s”." -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "sammanslagning kan inte fortsätta; fick inte rent resultat från %d" - msgid "git merge [] [...]" msgstr "git merge [] [...]" @@ -8969,7 +8983,7 @@ msgid "failed to remove 'git notes merge' worktree" msgstr "misslyckades ta bort arbetskatalogen för ”git notes merge”" msgid "failed to read ref NOTES_MERGE_PARTIAL" -msgstr "misslyckades läsa references NOTES_MERGE_PARTIAL" +msgstr "misslyckades läsa referensen NOTES_MERGE_PARTIAL" msgid "could not find commit from NOTES_MERGE_PARTIAL." msgstr "kunde inte hitta incheckning från NOTES_MERGE_PARTIAL." @@ -9080,6 +9094,13 @@ msgid "" msgstr "" "git pack-objects [] [< | < ]" +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "ogiltig flagga för --name-hash-version: %d" + +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "--write-bitmap-index kräver för närvarande, --name-hash-version=1" + #, c-format msgid "" "write_reuse_object: could not locate %s, expected at offset % in " @@ -9190,14 +9211,14 @@ msgid "" "value of uploadpack.blobpackfileuri must be of the form ' ' (got '%s')" msgstr "" -"värdet på uploadpack.blobpackfileuri måste vara på formen ' " -" ' (fick '%s')" +"värdet på uploadpack.blobpackfileuri måste vara på formen ” " +" ” (fick ”%s”)" #, c-format msgid "" "object already configured in another uploadpack.blobpackfileuri (got '%s')" msgstr "" -"objektet redan konfigurerat i et annat uploadpack.blobpackfileuri (fick '%s)" +"objektet redan konfigurerat i ett annat uploadpack.blobpackfileuri (fick ”%s”" #, c-format msgid "could not get type of object %s in pack %s" @@ -9401,6 +9422,10 @@ msgid "exclude any configured uploadpack.blobpackfileuri with this protocol" msgstr "" "uteslut redan konfigurerade uploadpack.blobpackfileuri med detta protokoll" +msgid "use the specified name-hash function to group similar objects" +msgstr "" +"använd den angivna namn-hash-funktionen för att gruppera liknande objekt" + #, c-format msgid "delta chain depth %d is too deep, forcing %d" msgstr "deltakedjedjupet %d är för djupt, påtvingar %d" @@ -10189,7 +10214,7 @@ msgstr "" "Ange vilken gren du vill ombasera mot.\n" "Se git-rebase(1) för detaljer.\n" "\n" -" git rebase ''\n" +" git rebase ””\n" "\n" #, c-format @@ -10603,7 +10628,8 @@ msgid "process the reflogs of all references" msgstr "hantera referensloggar för alla referenser" msgid "limits processing to reflogs from the current worktree only" -msgstr "begränsar hantering av referensloggar till endast aktuell arbetskatalog" +msgstr "" +"begränsar hantering av referensloggar till endast aktuell arbetskatalog" #, c-format msgid "Marking reachable objects..." @@ -10620,8 +10646,8 @@ msgstr "ingen referenslogg att ta bort angavs" msgid "invalid ref format: %s" msgstr "felaktigt referensformat: %s" -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format= [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format= [--no-reflog] [--dry-run]" msgid "git refs verify [--strict] [--verbose]" msgstr "git refs verify [--strict] [--verbose]" @@ -10632,6 +10658,9 @@ msgstr "ange referensformatet att konvertera till" msgid "perform a non-destructive dry-run" msgstr "utför ett icke-destruktiv testkörning" +msgid "drop reflogs entirely during the migration" +msgstr "kasta referensloggar helt under migreringen" + msgid "missing --ref-format=" msgstr "saknad --ref-format=" @@ -11091,8 +11120,14 @@ msgstr "Kommer inte ta bort alla icke-sänd-URL:er" msgid "be verbose; must be placed before a subcommand" msgstr "var pratsam; måste skrivas före ett underkommando" -msgid "git repack []" -msgstr "git repack []" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" msgid "" "Incremental repacks are incompatible with bitmap indexes. Use\n" @@ -11166,6 +11201,12 @@ msgstr "sänd --no-reuse-delta till git-pack-objects" msgid "pass --no-reuse-object to git-pack-objects" msgstr "sänd --no-reuse-object till git-pack-objects" +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "" +"ange den namnhash-version som ska användas för att gruppera liknande objekt " +"efter sökväg" + msgid "do not run git-update-server-info" msgstr "kör inte git-update-server-info" @@ -11214,9 +11255,6 @@ msgstr "hitta ett geometrisk förlopp med faktor " msgid "write a multi-pack index of the resulting packs" msgstr "skriv ett flerpaketsindex för de skapade paketen" -msgid "pack prefix to store a pack containing pruned objects" -msgstr "paketprefix att lagra ett paket som innehåller bortrensade objekt" - msgid "pack prefix to store a pack containing filtered out objects" msgstr "paketprefix att lagra ett paket som innehåller utfiltrerade objekt" @@ -11433,9 +11471,6 @@ msgstr "endast ett mönster kan anges med -l" msgid "need some commits to replay" msgstr "behöver några incheckningar för omspelning" -msgid "--onto and --advance are incompatible" -msgstr "--onto och --advance kan inte kombineras" - msgid "all positive revisions given must be references" msgstr "alla positiva revisioner som anges måste vara referenser" @@ -12020,7 +12055,7 @@ msgid "[,]" msgstr "[,]" msgid "show most recent ref-log entries starting at base" -msgstr "visa nyaste refloggposter med början på bas" +msgstr "visa nyaste referensloggposter med början på bas" msgid "no branches given, and HEAD is not valid" msgstr "inga grenar angavs, och HEAD är inte giltigt" @@ -12134,7 +12169,7 @@ msgid "" "directory '%s' contains untracked files, but is not in the sparse-checkout " "cone" msgstr "" -"katalogen ”%s” innehåller ospårade filer, men är inte i området som ages i " +"katalogen ”%s” innehåller ospårade filer, men är inte i området som anges i " "”sparse-checkout”" #, c-format @@ -12726,7 +12761,7 @@ msgstr "misslyckades klona ”%s” till undermodulsökvägen ”%s”" #, c-format msgid "could not get submodule directory for '%s'" -msgstr "kunde inte få tag i undermodulkatalog för ”%s”" +msgstr "kunde inte få tag i undermodulskatalog för ”%s”" msgid "alternative anchor for relative paths" msgstr "alternativa ankare för relativa sökvägar" @@ -12830,8 +12865,8 @@ msgid "" "Fetched in submodule path '%s', but it did not contain %s. Direct fetching " "of that commit failed." msgstr "" -"Hämtade i undermodulssökvägen ”%s”, men den innehöll inte %s. Direkt " -"hämtning av incheckningen misslyckades." +"Hämtade i undermodulsökvägen ”%s”, men den innehöll inte %s. Direkt hämtning " +"av incheckningen misslyckades." #, c-format msgid "could not initialize submodule at path '%s'" @@ -13199,7 +13234,7 @@ msgid "replace the tag if exists" msgstr "ersätt taggen om den finns" msgid "create a reflog" -msgstr "skapa en reflog" +msgstr "skapa en referenslogg" msgid "Tag listing options" msgstr "Alternativ för listning av taggar" @@ -13924,7 +13959,7 @@ msgstr "okänd kapabilitet ”%s”" #, c-format msgid "'%s' does not look like a v2 or v3 bundle file" -msgstr "'%s' ser inte ut som en v2- eller v3-bunt-fil" +msgstr "”%s” ser inte ut som en v2- eller v3-bunt-fil" #, c-format msgid "unrecognized header: %s%s (%d)" @@ -14058,6 +14093,9 @@ msgstr "Importera ett GNU Arch-arkiv till Git" msgid "Create an archive of files from a named tree" msgstr "Skapa ett arkiv över filer från ett namngivet träd" +msgid "Download missing objects in a partial clone" +msgstr "Hämta saknade objekt i en delvis kloning" + msgid "Use binary search to find the commit that introduced a bug" msgstr "Använd binärsökning för att hitta ändringen som introducerade ett fel" @@ -14164,7 +14202,7 @@ msgid "Compare a tree to the working tree or index" msgstr "Jämför en träd med arbetskatalogen eller indexet" msgid "Compares the content and mode of blobs found via two tree objects" -msgstr "Visar innehåll och läge för blobbar som hittats via två trädobjekt" +msgstr "Visar innehåll och läge för blob:ar som hittats via två trädobjekt" msgid "Show changes using common diff tools" msgstr "Visa ändringar med vanliga diff-verktyg" @@ -14996,19 +15034,19 @@ msgstr "kunde inte läsa katalogändringar [GLE %ld]" #, c-format msgid "opendir('%s') failed" -msgstr "opendir('%s') misslyckades" +msgstr "opendir(”%s”) misslyckades" #, c-format msgid "lstat('%s') failed" -msgstr "lstat('%s') misslyckades" +msgstr "lstat(”%s”) misslyckades" #, c-format msgid "strbuf_readlink('%s') failed" -msgstr "strbuf_readlink('%s') misslyckades" +msgstr "strbuf_readlink(”%s”) misslyckades" #, c-format msgid "closedir('%s') failed" -msgstr "closedir('%s') misslyckades" +msgstr "closedir(”%s”) misslyckades" #, c-format msgid "[GLE %ld] unable to open for read '%ls'" @@ -15335,7 +15373,7 @@ msgstr "referensen ”%s” pekar inte på en blob" #, c-format msgid "unable to resolve config blob '%s'" -msgstr "kan inte slå upp konfigurerings-blobben ”%s”" +msgstr "kan inte slå upp konfigurerings-blob:en ”%s”" msgid "unable to parse command-line config" msgstr "kan inte tolka kommandoradskonfiguration" @@ -15968,6 +16006,12 @@ msgstr "ogiltigt argument för %s" msgid "invalid regex given to -I: '%s'" msgstr "ogiltigt reguljärt uttryck angavs för -I: ”%s”" +msgid "-G requires a non-empty argument" +msgstr "-G kräver ett icke-tomt argument" + +msgid "-S requires a non-empty argument" +msgstr "-S kräver ett icke-tomt argument" + #, c-format msgid "failed to parse --submodule option parameter: '%s'" msgstr "misslyckades tolka argument till flaggan --submodule: ”%s”" @@ -16080,7 +16124,7 @@ msgid "" "do not munge pathnames and use NULs as output field terminators in --raw or " "--numstat" msgstr "" -"skriv inte om sökvägsnamn och använd NUL-tecken som fältseparerare i --raw " +"skriv inte om sökvägsnamn och använd NUL-tecken som fältavdelare i --raw " "eller --numstat" msgid "" @@ -16689,7 +16733,7 @@ msgstr "" "”git help -a” och ”git help -g” visar tillgängliga underkommandon och\n" "några konceptvägledningar. Se ”git help ” eller ”git help\n" "” för att läsa mer om specifika underkommandon och koncept.\n" -"See ”git help git” för en översikt över systemet." +"Se ”git help git” för en översikt över systemet." #, c-format msgid "unsupported command listing type '%s'" @@ -17020,7 +17064,7 @@ msgstr "inte ett git-arkiv" #, c-format msgid "argument to --packfile must be a valid hash (got '%s')" msgstr "" -"argumentet till --packfile måste vara ett giltigt hashvärde (fick '%s')" +"argumentet till --packfile måste vara ett giltigt hashvärde (fick ”%s”)" #, c-format msgid "negative value for http.postBuffer; defaulting to %d" @@ -17056,7 +17100,7 @@ msgid "" " asked for: %s\n" " redirect: %s" msgstr "" -"kan inte uppdatera urlbas från omdirigerin:\n" +"kan inte uppdatera urlbas från omdirigering:\n" " bad om: %s\n" " omdirigering: %s" @@ -17914,7 +17958,7 @@ msgid "multi-pack-index stores a 64-bit offset, but off_t is too small" msgstr "multi-pack-index innehåller 64-bitars offset, men off_t är för liten" msgid "multi-pack-index large offset out of bounds" -msgstr "stort offset för mult-pack-index utanför gränsen" +msgstr "stort offset för multi-pack-index utanför gränsen" msgid "multi-pack-index file exists, but failed to parse" msgstr "multi-pack-indexfilen finns, men kunde inte tolkas" @@ -17985,7 +18029,7 @@ msgstr "Kan inte checka in oinitierat/orefererat anteckningsträd" #, c-format msgid "Bad notes.rewriteMode value: '%s'" -msgstr "Felaktigt värde för notes.rewriteMode: '%s'" +msgstr "Felaktigt värde för notes.rewriteMode: ”%s”" #, c-format msgid "Refusing to rewrite notes in %s (outside of refs/notes/)" @@ -18134,6 +18178,10 @@ msgstr "filerna ”%s” och ”%s” har olika innehåll" msgid "unable to write file %s" msgstr "kan inte skriva filen %s" +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "kan inte skriva till filen %s som hela tiden försvinner" + #, c-format msgid "unable to set permission to '%s'" msgstr "kan inte sätta behörigheten till ”%s”" @@ -18538,7 +18586,8 @@ msgstr "bitkarteresultat stämmer inte överens" #, c-format msgid "pseudo-merge index out of range (% >= %)" -msgstr "pseudosammanslagningsindex utanför intervallet (% ≥ %)" +msgstr "" +"pseudosammanslagningsindex utanför intervallet (% ≥ %)" #, c-format msgid "could not find '%s' in pack '%s' at offset %" @@ -18694,6 +18743,52 @@ msgstr "okänd flagga ”%c”" msgid "unknown non-ascii option in string: `%s'" msgstr "okänd icke-ascii-flagga i strängen: ”%s”" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid " <%s>" +msgstr " <%s>" + msgid "..." msgstr "..." @@ -18778,6 +18873,21 @@ msgstr "felaktigt booleskt miljövariabelvärde ”%s” för ”%s”" msgid "failed to parse %s" msgstr "misslyckades tolka %s" +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "misslyckades traversera löven i trädet %s: hittades inte" + +#, c-format +msgid "failed to find object %s" +msgstr "misslyckades hitta objektet %s" + +#, c-format +msgid "failed to find tag %s" +msgstr "misslyckades hitta taggen %s" + +msgid "failed to setup revision walk" +msgstr "misslyckades starta revisionstraversering" + #, c-format msgid "Could not make %s writable by group" msgstr "Kunde inte göra %s skrivbar för gruppen" @@ -18923,6 +19033,22 @@ msgstr "kontraktsfjärr kan inte börja med ”/”: %s" msgid "could not fetch %s from promisor remote" msgstr "kunde inte hämta %s från kontraktsfjärr" +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "känd fjärr som heter ”%s” med med url:en ”%s” istället för ”%s”" + +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "okänt värde ”%s” för inställningen ”%s”" + +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "okänt värde ”%s” från fjärrinformation" + +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "godkänd kontraktsfjärr ”%s” hittades inte" + msgid "object-info: expected flush after arguments" msgstr "object-info: förväntade ”flush” efter argument" @@ -19248,7 +19374,7 @@ msgstr "" " enrads, om inget incheckningsmeddelande angavs); använd\n" " -c för att skriva om meddelandet.\n" "u, update-ref = spåra en platshållare för att uppdatera\n" -" till denna position bland nya inchecknngar.\n" +" till denna position bland nya incheckningar.\n" " uppdateras i slutet av ombaseringen.\n" "\n" "Du kan byta ordning på raderna; de utförs uppifrån och ned.\n" @@ -19616,7 +19742,7 @@ msgid "refusing to update ref with bad name '%s'" msgstr "vägrar uppdatera referens med trasigt namn ”%s”" msgid "refusing to force and skip creation of reflog" -msgstr "vägrar att tvinga och hoppa över skapande av reflogg" +msgstr "vägrar att tvinga och hoppa över skapande av referenslogg" #, c-format msgid "update_ref failed for ref '%s': %s" @@ -19748,6 +19874,14 @@ msgstr "referensnamnet %s är en symbolisk referens, kopiering stöds inte" msgid "invalid refspec '%s'" msgstr "felaktig referensspecifikation: ”%s”" +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "mönstret ”%s” innehåller ingen ”*”" + +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "ersättningen ”%s” innehåller ingen ”*”" + #, c-format msgid "invalid quoting in push-option value: '%s'" msgstr "felaktig citering på värde för push-option: ”%s”" @@ -19866,6 +20000,28 @@ msgstr "remote-curl: försökte ta emot utan lokalt arkiv" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl: okänt kommando ”%s” från git" +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"läser fjärren från ”%s/%s”, som har nominerats för borttagning.\n" +"\n" +"Om du fortfarande använder ”remotes/”-katalogen rekommenderas du\n" +"migrera till konfigurationsbaserade fjärrar:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"Om du inte kan det, berätta för oss varför du fortfarande behöver\n" +"använda det på e-post till ." + #, c-format msgid "config remote shorthand cannot begin with '/': %s" msgstr "konfigurerad kortform för fjärr kan inte börja med ”/”: %s" @@ -19900,14 +20056,6 @@ msgstr "%s spårar vanligtvis %s, inte %s" msgid "%s tracks both %s and %s" msgstr "%s spårar både %s och %s" -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "nyckeln ”%s” i mönstret innehåller ingen ”*”" - -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "värdet ”%s” i mönstret innehåller ingen ”*”" - #, c-format msgid "src refspec %s does not match any" msgstr "käll-referensspecifikationen %s motsvarar ingen" @@ -21798,6 +21946,27 @@ msgstr "töm cacheträdet före varje iteration" msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "antal poster i cacheträdet att ogiltigförklara (förval är 0)" +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk -- " + +msgid "toggle inclusion of blob objects" +msgstr "växla om blob-objekt ska vara med eller inte" + +msgid "toggle inclusion of commit objects" +msgstr "växla om incheckningsobjekt ska vara med eller inte" + +msgid "toggle inclusion of tag objects" +msgstr "växla om taggobjekt ska vara med eller inte" + +msgid "toggle inclusion of tree objects" +msgstr "växla om trädobjekt ska vara med eller inte" + +msgid "toggle pruning of uninteresting paths" +msgstr "växla bortrensning av ointressanta sökvägar" + +msgid "read a pattern list over stdin" +msgstr "läs en mönsterlista från standard in" + #, c-format msgid "commit %s is not marked reachable" msgstr "incheckning %s är inte märkt nåbar" @@ -22431,6 +22600,10 @@ msgstr "fel: " msgid "warning: " msgstr "varning: " +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() misslyckades med felet ”%s” (%d)\n" + msgid "Fetching objects" msgstr "Hämtar objekt" @@ -23265,7 +23438,7 @@ msgstr "Nödvändig SMTP-server har inte angivits korrekt." #, perl-format msgid "Server does not support STARTTLS! %s" -msgstr "Servern stöder inte SMARTTLS! %s" +msgstr "Servern stöder inte STARTTLS! %s" #, perl-format msgid "STARTTLS failed! %s" From 7d5a9e6b9984ad5fa699ef1245dbd27aab07d3fa Mon Sep 17 00:00:00 2001 From: Tuomas Ahola Date: Sat, 8 Mar 2025 22:45:10 +0200 Subject: [PATCH 009/259] l10n: sv.po: Fix Swedish typos Signed-off-by: Tuomas Ahola --- po/sv.po | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/po/sv.po b/po/sv.po index 32da53fee0ccab..42c357e54f093e 100644 --- a/po/sv.po +++ b/po/sv.po @@ -3103,7 +3103,7 @@ msgid "" "You can delete any lines you don't wish to share.\n" msgstr "" "Tack för att du skriver en buggraport för Git!\n" -"Om du svarar på följande frågor är det lättare för oss att första " +"Om du svarar på följande frågor är det lättare för oss att förstå " "problemet.\n" "Skriv gärna på engelska\n" "\n" @@ -3376,13 +3376,13 @@ msgid "" "git check-attr [--source ] [-a | --all | ...] [--] " "..." msgstr "" -"git check-attr [--source ] [-a | --all | ...] [--] " +"git check-attr [--source ] [-a | --all | ...] [--] " "..." msgid "" "git check-attr --stdin [-z] [--source ] [-a | --all | ...]" msgstr "" -"git check-attr --stdin [-z] [--source ] [-a | --all | ...]" +"git check-attr --stdin [-z] [--source ] [-a | --all | ...]" msgid "report all attributes set on file" msgstr "visa alla attribut som satts på filen" @@ -16539,7 +16539,7 @@ msgid "already have %s (%s)" msgstr "har redan %s (%s)" msgid "fetch-pack: unable to fork off sideband demultiplexer" -msgstr "fetch-patch: kan inte grena (fork) av sidbandsmultiplexare" +msgstr "fetch-patch: kan inte grena (fork) av sidbands-avmultiplexare" msgid "protocol error: bad pack header" msgstr "protokollfel: felaktigt packhuvud" From 83b278ef74b7eed02a4c7046c830eea19bbd10f9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Martin=20=C3=85gren?= Date: Mon, 10 Mar 2025 12:07:56 +0100 Subject: [PATCH 010/259] git-clone doc: fix indentation MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Commit bc26f7690a (clone: make it possible to specify --tags, 2025-02-06) added a new paragraph in the middle of this list item. By adding an empty line rather than using a list continuation, we broke the list continuation, with the new paragraph ending up funnily indented. Restore the chain of list continuations. Signed-off-by: Martin Ågren Signed-off-by: Junio C Hamano --- Documentation/git-clone.txt | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/Documentation/git-clone.txt b/Documentation/git-clone.txt index 8d0476f6dcaf6f..ebfb60888519d9 100644 --- a/Documentation/git-clone.txt +++ b/Documentation/git-clone.txt @@ -279,9 +279,9 @@ corresponding `--mirror` and `--no-tags` options instead. `remote..tagOpt=--no-tags` configuration. This ensures that future `git pull` and `git fetch` won't follow any tags. Subsequent explicit tag fetches will still work (see linkgit:git-fetch[1]). - - By default, tags are cloned and passing `--tags` is thus typically a - no-op, unless it cancels out a previous `--no-tags`. ++ +By default, tags are cloned and passing `--tags` is thus typically a +no-op, unless it cancels out a previous `--no-tags`. + Can be used in conjunction with `--single-branch` to clone and maintain a branch with no references other than a single cloned From ee01097f2840deb2f042b670f4b5b6a0cb45fb5d Mon Sep 17 00:00:00 2001 From: Ruggero Turra Date: Sat, 22 Feb 2025 23:18:23 +0100 Subject: [PATCH 011/259] l10n: it: fix spelling of "sorgente" (Italian for "source") Signed-off-by: Ruggero Turra --- po/it.po | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/po/it.po b/po/it.po index c31560834d8f2b..20fd8bb28d8589 100644 --- a/po/it.po +++ b/po/it.po @@ -17602,7 +17602,7 @@ msgstr "Controllo la ridenominazione di '%s' in '%s'\n" #: builtin/mv.c:185 msgid "bad source" -msgstr "sourgente errata" +msgstr "sorgente errata" #: builtin/mv.c:188 msgid "can not move directory into itself" From ec507acbfd5e039cdc6690f56fbc576b17f3b930 Mon Sep 17 00:00:00 2001 From: seoyeon-kwon Date: Thu, 23 Jan 2025 15:06:37 +0900 Subject: [PATCH 012/259] l10n: ko: fix minor typo in Korean translation Signed-off-by: seoyeon-kwon --- po/ko.po | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/po/ko.po b/po/ko.po index 5d190e52380864..7a6847f0236711 100644 --- a/po/ko.po +++ b/po/ko.po @@ -7,6 +7,7 @@ # Changwoo Ryu , 2015-2018. # Sihyeon Jang , 2018. # Gwan-gyeong Mun , 2018. +# Seoyeon Kwon , 2025. # # - 작업자는 위 Contributors 목록에 추가해 주세요. # - 번역하면서 80컬럼을 넘어가지 않도록 해 주세요. @@ -7666,7 +7667,7 @@ msgid "" " git commit --allow-empty\n" "\n" msgstr "" -"이전 커맷 빼오기가 비어 있습니다. 아마도 충돌 해결 과정에서 그렇게 됐을\n" +"이전 커밋 빼오기가 비어 있습니다. 아마도 충돌 해결 과정에서 그렇게 됐을\n" "것입니다. 그래도 커밋하려면 다음과 같이 하십시오:\n" "\n" " git commit --allow-empty\n" From f17f45f3874caeec4c8794b680a2fb88d1da1e65 Mon Sep 17 00:00:00 2001 From: Emir SARI Date: Thu, 27 Feb 2025 13:22:50 +0300 Subject: [PATCH 013/259] l10n: tr: Update Turkish translations for 2.49.0 Signed-off-by: Emir SARI --- po/tr.po | 436 ++++++++++++++++++++++++++++++++++++++----------------- 1 file changed, 304 insertions(+), 132 deletions(-) diff --git a/po/tr.po b/po/tr.po index 1bc43453aa7e21..4ed91cbd975c89 100644 --- a/po/tr.po +++ b/po/tr.po @@ -96,8 +96,8 @@ msgid "" msgstr "" "Project-Id-Version: Git Turkish Localization Project\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-23 18:57+0000\n" -"PO-Revision-Date: 2025-01-01 15:00+0300\n" +"POT-Creation-Date: 2025-03-11 15:01+0300\n" +"PO-Revision-Date: 2025-03-11 15:00+0300\n" "Last-Translator: Emir SARI \n" "Language-Team: Turkish (https://github.com/bitigchi/git-po/)\n" "Language: tr\n" @@ -2408,6 +2408,18 @@ msgstr "git archive: Protokol hatası" msgid "git archive: expected a flush" msgstr "git archive: Floş bekleniyordu" +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +msgid "problem loading sparse-checkout" +msgstr "aralıklı çıkış yüklenirken sorun" + +msgid "Minimum number of objects to request at a time" +msgstr "Bir kerede istenecek en çok nesne sayısı" + +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "Eksik nesneleri geçerli aralıklı çıkışa sınırla" + msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" "checkout] [--first-parent] [ [...]] [--] [...]" @@ -3149,10 +3161,6 @@ msgstr "" msgid "git version:\n" msgstr "git sürümü:\n" -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() '%s' hatasını verip çıktı (%d)\n" - msgid "compiler info: " msgstr "derleyici bilgisi: " @@ -4158,8 +4166,91 @@ msgstr "" "clean.requireForce 'true' olarak ayarlı ve -f verilmedi; temizlik " "reddediliyor" -msgid "git clone [] [--] []" -msgstr "git clone [] [--] []" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "info: '%s' için alternatif eklenemedi: %s\n" + +#, c-format +msgid "failed to stat '%s'" +msgstr "'%s' dosyasının bilgileri alınamadı" + +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s var ve bir dizin değil" + +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "'%s' bir sembolik bağ; --local ile klonlama reddediliyor" + +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "yineleyici '%s' üzerinden çalıştırılamadı" + +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "'%s' sembolik bağı var, --local ile klonlama reddediliyor" + +#, c-format +msgid "failed to unlink '%s'" +msgstr "'%s' bağlantısı kesilemedi" + +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "sabit bağlantı, '%s' konumunda denetlenemiyor" + +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "sabit bağlantı, '%s' konumundaki kaynaktan farklı" + +#, c-format +msgid "failed to create link '%s'" +msgstr "'%s' bağı oluşturulamadı" + +#, c-format +msgid "failed to copy file to '%s'" +msgstr "dosya şuraya kopyalanamadı: '%s'" + +#, c-format +msgid "failed to iterate over '%s'" +msgstr "'%s' üzerinde yinelenemedi" + +#, c-format +msgid "done.\n" +msgstr "bitti.\n" + +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Klonlama başarılı oldu; ancak çıkış yapılamadı.\n" +"Neyin çıkış yapılıp yapılmadığını 'git status' ile inceleyebilir\n" +"ve 'git restore --source=HEAD' ile yeniden deneyebilirsiniz.\n" + +msgid "remote did not send all necessary objects" +msgstr "uzak konum gereken tüm nesneleri göndermedi" + +#, c-format +msgid "unable to update %s" +msgstr "%s güncellenemiyor" + +msgid "failed to initialize sparse-checkout" +msgstr "sparse-checkout ilklendirilemedi" + +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "uzak konum HEAD'i, var olmayan başvuruya başvuruyor; çıkış yapılamıyor" + +msgid "unable to checkout working tree" +msgstr "çalışma ağacı çıkış yapılamıyor" + +msgid "unable to write parameters to config file" +msgstr "parametreler yapılandırma dosyasına yazılamıyor" + +msgid "cannot repack to clean up" +msgstr "temizlik için yeniden paketlenemiyor" + +msgid "cannot unlink temporary alternates file" +msgstr "geçici alternatifler dosyasının bağlantısı kesilemiyor" msgid "don't clone shallow repository" msgstr "sığ depoyu klonlama" @@ -4212,6 +4303,9 @@ msgstr "üstkaynağı izlemek için 'origin' yerine kullan" msgid "checkout instead of the remote's HEAD" msgstr "uzak konumun HEAD'i yerine 'ı çıkış yap" +msgid "clone single revision and check out" +msgstr "tek revizyonlu 'i klonla ve çıkış yap" + msgid "path to git-upload-pack on the remote" msgstr "uzak konumdaki git-upload-pack'e olan yol" @@ -4233,8 +4327,8 @@ msgstr "başvuru hariç tutarak sığ klonun geçmişini derinleştir" msgid "clone only one branch, HEAD or --branch" msgstr "yalnızca bir dal klonla, HEAD veya --branch" -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "etiket klonlama ve sonraki getirmeler de onları izlemesin" +msgid "clone tags, and make later fetches not to follow them" +msgstr "etiketleri klonla ve sonraki getirmelerin onları izlememesini sağla" msgid "any cloned submodules will be shallow" msgstr "klonlanan altmodüller sığ olacak" @@ -4277,95 +4371,8 @@ msgstr "uri" msgid "a URI for downloading bundles before fetching from origin remote" msgstr "uzak konum kökeninden getirmeden önce demetleri indirmek için bir URI" -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "info: '%s' için alternatif eklenemedi: %s\n" - -#, c-format -msgid "failed to stat '%s'" -msgstr "'%s' dosyasının bilgileri alınamadı" - -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s var ve bir dizin değil" - -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "'%s' bir sembolik bağ; --local ile klonlama reddediliyor" - -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "yineleyici '%s' üzerinden çalıştırılamadı" - -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "'%s' sembolik bağı var, --local ile klonlama reddediliyor" - -#, c-format -msgid "failed to unlink '%s'" -msgstr "'%s' bağlantısı kesilemedi" - -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "sabit bağlantı, '%s' konumunda denetlenemiyor" - -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "sabit bağlantı, '%s' konumundaki kaynaktan farklı" - -#, c-format -msgid "failed to create link '%s'" -msgstr "'%s' bağı oluşturulamadı" - -#, c-format -msgid "failed to copy file to '%s'" -msgstr "dosya şuraya kopyalanamadı: '%s'" - -#, c-format -msgid "failed to iterate over '%s'" -msgstr "'%s' üzerinde yinelenemedi" - -#, c-format -msgid "done.\n" -msgstr "bitti.\n" - -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Klonlama başarılı oldu; ancak çıkış yapılamadı.\n" -"Neyin çıkış yapılıp yapılmadığını 'git status' ile inceleyebilir\n" -"ve 'git restore --source=HEAD' ile yeniden deneyebilirsiniz.\n" - -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "Klonlanacak %s uzak dal bulunamadı." - -msgid "remote did not send all necessary objects" -msgstr "uzak konum gereken tüm nesneleri göndermedi" - -#, c-format -msgid "unable to update %s" -msgstr "%s güncellenemiyor" - -msgid "failed to initialize sparse-checkout" -msgstr "sparse-checkout ilklendirilemedi" - -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "uzak konum HEAD'i, var olmayan başvuruya başvuruyor; çıkış yapılamıyor" - -msgid "unable to checkout working tree" -msgstr "çalışma ağacı çıkış yapılamıyor" - -msgid "unable to write parameters to config file" -msgstr "parametreler yapılandırma dosyasına yazılamıyor" - -msgid "cannot repack to clean up" -msgstr "temizlik için yeniden paketlenemiyor" - -msgid "cannot unlink temporary alternates file" -msgstr "geçici alternatifler dosyasının bağlantısı kesilemiyor" +msgid "git clone [] [--] []" +msgstr "git clone [] [--] []" msgid "Too many arguments." msgstr "Çok fazla argüman." @@ -4465,6 +4472,10 @@ msgstr "uzak konum taşıması hata bildirdi" msgid "Remote branch %s not found in upstream %s" msgstr "%s uzak dalı %s üstkaynağında bulunamadı" +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "%s uzak revizyonu, %s üstkaynağında bulunamadı" + msgid "You appear to have cloned an empty repository." msgstr "Boş bir depoyu klonlamış görünüyorsunuz." @@ -4641,7 +4652,7 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree: okunamadı" msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -4651,11 +4662,11 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" -" [--dry-run] [(-c | -C | --squash) | --fixup\n" -" [(amend|reword):]] [-F | -m ] [--" -"reset-author] [--allow-empty]\n" -" [--allow-empty-message] [--no-verify] [-e] [--author=]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" +" [--dry-run] [(-c | -C | --squash) |\n" +" --fixup [(amend|reword):]]\n" +" [-F | -m ] [--reset-author] [--allow-empty]\n" +" [--allow-empty-message] [--no-verify] [-e] [--author=]\n" " [--date=] [--cleanup=] [--[no-]status]\n" " [-i | -o] [--pathspec-from-file= [--pathspec-file-nul]]\n" " [(--trailer [(=|:)])...] [-S[]]\n" @@ -6044,15 +6055,15 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Değişikliği izlemek için 'git remote set-head %s %s' yapın veya\n" "'remote.%s.followRemoteHEAD' yapılandırma seçeneğini başka bir\n" "değere ayarlayın (bu iletiyi görmek istemiyorsanız). Özellikle\n" -"'git config set remote.%s.followRemoteHEAD %s' komutunu çalıştırmak\n" -"uyarıyı HEAD'e veya başka bir şeye uzaktan değişiklik olana dek\n" -"devre dışı bırakır." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"komutunu çalıştırmak, uyarıyı HEAD'e veya başka bir şeye uzaktan\n" +"değişiklik olana dek devre dışı bırakır." msgid "multiple branches detected, incompatible with --set-upstream" msgstr "birden çok dal algılandı, --set-upstream ile uyumsuz" @@ -6729,6 +6740,9 @@ msgstr "başka bir gc çalışıyor olsa bile zorla gc çalıştır" msgid "repack all other packs except the largest pack" msgstr "en büyük paket dışındaki diğer tüm paketleri yeniden paketle" +msgid "pack prefix to store a pack containing pruned objects" +msgstr "budanan nesneler içeren paketi depolamak için paket öneki" + #, c-format msgid "failed to parse gc.logExpiry value %s" msgstr "gc.logExpiry değeri %s ayrıştırılamadı" @@ -7525,6 +7539,10 @@ msgstr "yerel bağlantıları yeniden paketleme için pack-objects bitirilemedi" msgid "Cannot come back to cwd" msgstr "Geçerli çalışma dizinine geri gelinemiyor" +#, c-format +msgid "bad --pack_header: %s" +msgstr "hatalı --pack_header: %s" + #, c-format msgid "bad %s" msgstr "hatalı %s" @@ -8384,10 +8402,6 @@ msgstr "bilinmeyen strateji seçeneği: -X%s" msgid "malformed input line: '%s'." msgstr "hatalı oluşturulmuş girdi satırı: '%s'." -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "birleştirme sürdürülemiyor; %d için temiz olmayan sonuçlar alındı" - msgid "git merge [] [...]" msgstr "git merge [] [...]" @@ -9201,6 +9215,13 @@ msgstr "" "git pack-objects [] [< | < ]" +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "geçersiz --name-hash-version seçeneği: %d" + +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "şu anda --write-bitmap-index, --name-hash-version=1 gerektiriyor" + #, c-format msgid "" "write_reuse_object: could not locate %s, expected at offset % in " @@ -9526,6 +9547,9 @@ msgstr "" "bu protokol ile herhangi bir yapılandırılmış uploadpack.blobpackfileuri " "ögesini hariç tut" +msgid "use the specified name-hash function to group similar objects" +msgstr "benzer nesneleri gruplamak için belirtilen name-hash işlevini kullan" + #, c-format msgid "delta chain depth %d is too deep, forcing %d" msgstr "delta zincir derinliği %d çok derin, %d zorlanıyor" @@ -10756,8 +10780,8 @@ msgstr "silmek için bir başvuru günlüğü belirtilmedi" msgid "invalid ref format: %s" msgstr "geçersiz başvuru biçimi: %s" -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format= [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format= [--no-reflog] [--dry-run]" msgid "git refs verify [--strict] [--verbose]" msgstr "git refs verify [--strict] [--verbose]" @@ -10768,6 +10792,9 @@ msgstr "dönüştürülecek başvuru biçimini belirt" msgid "perform a non-destructive dry-run" msgstr "yıkıcı olmayan bir deneme gerçekleştir" +msgid "drop reflogs entirely during the migration" +msgstr "göç sırasında başvuru günlüklerini tümüyle bırak" + msgid "missing --ref-format=" msgstr "--ref-format= eksik" @@ -11234,8 +11261,14 @@ msgstr "Tüm itme olmayan URL'ler silinmeyecek" msgid "be verbose; must be placed before a subcommand" msgstr "ayrıntılı anlat; bir altkomuttan önce yerleştirilmelidir" -msgid "git repack []" -msgstr "git repack []" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" msgid "" "Incremental repacks are incompatible with bitmap indexes. Use\n" @@ -11309,6 +11342,11 @@ msgstr "'git-pack-objects'e --no-reuse-delta geçir" msgid "pass --no-reuse-object to git-pack-objects" msgstr "'git-pack-objects'e --no-reuse-object geçir" +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "" +"benzer nesneleri yola göre gruplamada kullanılacak name-hash sürümünü belirt" + msgid "do not run git-update-server-info" msgstr "'git-update-server-info' çalıştırma" @@ -11357,9 +11395,6 @@ msgstr " faktörlü bir geometrik ilerleme bul" msgid "write a multi-pack index of the resulting packs" msgstr "ortaya çıkan paketlerin bir çoklu paket indeksini yaz" -msgid "pack prefix to store a pack containing pruned objects" -msgstr "budanan nesneler içeren paketi depolamak için paket öneki" - msgid "pack prefix to store a pack containing filtered out objects" msgstr "süzülen nesneler içeren paketi depolamak için paket öneki" @@ -11576,9 +11611,6 @@ msgstr "-l ile yalnızca bir dizgi verilebilir" msgid "need some commits to replay" msgstr "yeniden oynatmak için birkaç işleme gerekli" -msgid "--onto and --advance are incompatible" -msgstr "--onto ve --advance birbiriyle uyumsuz" - msgid "all positive revisions given must be references" msgstr "verilen tüm pozitif revizyonlar, başvuru olmalı" @@ -14200,6 +14232,9 @@ msgstr "Git'e bir GNU Arch deposu içe aktar" msgid "Create an archive of files from a named tree" msgstr "Ad verilmiş ağaçtan bir dosyalar arşivi oluştur" +msgid "Download missing objects in a partial clone" +msgstr "Eksik nesneleri kısımsal bir klonda indir" + msgid "Use binary search to find the commit that introduced a bug" msgstr "Hatalara neden olan işlemeyi bulmada ikili arama kullan" @@ -16111,6 +16146,12 @@ msgstr "%s için geçersiz argüman" msgid "invalid regex given to -I: '%s'" msgstr "-I'ya geçersiz düzenli ifade verildi: '%s'" +msgid "-G requires a non-empty argument" +msgstr "-G, boş olmayan bir argüman gerektiriyor" + +msgid "-S requires a non-empty argument" +msgstr "-S, boş olmayan bir argüman gerektiriyor" + #, c-format msgid "failed to parse --submodule option parameter: '%s'" msgstr "--submodule seçenek parametresi ayrıştırılamadı: '%s'" @@ -18289,6 +18330,10 @@ msgstr "'%s' ve '%s' dosyalarının içeriği farklı" msgid "unable to write file %s" msgstr "%s dosyası yazılamıyor" +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "sürekli olarak kaybolan %s dosyası yazılamıyor" + #, c-format msgid "unable to set permission to '%s'" msgstr "'%s' ögesine izin ayarlanamıyor" @@ -18851,6 +18896,52 @@ msgstr "bilinmeyen anahtar '%c'" msgid "unknown non-ascii option in string: `%s'" msgstr "dizi içinde bilinmeyen ascii dışı seçenek: '%s'" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid " <%s>" +msgstr " <%s>" + msgid "..." msgstr "..." @@ -18936,6 +19027,21 @@ msgstr "hatalı Boole çevre değeri '%s', '%s' için" msgid "failed to parse %s" msgstr "%s ayrıştırılamadı" +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "%s ağacının alt ögeleri yürütülemedi: bulunamadı" + +#, c-format +msgid "failed to find object %s" +msgstr "%s nesnesi bulunamadı" + +#, c-format +msgid "failed to find tag %s" +msgstr "%s etiketi bulunamadı" + +msgid "failed to setup revision walk" +msgstr "revizyon yürüyüşü ayarlanamadı" + #, c-format msgid "Could not make %s writable by group" msgstr "%s grup ile yazılabilir yapılamadı" @@ -19079,6 +19185,22 @@ msgstr "vaatçi uzak konum adı '/' ile başlayamaz: %s" msgid "could not fetch %s from promisor remote" msgstr "vaatçi uzak konumundan %s getirilemedi" +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "bilinen uzak konum adı '%s'; ancak url'si '%s'; '%s' olmalı" + +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "'%s' yapılandırma seçeneği için bilinmeyen değer: '%s'" + +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "uzak bilgiden bilinmeyen öge '%s'" + +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "kabul edilmiş vaatçi uzak konum '%s' bulunamadı" + msgid "object-info: expected flush after arguments" msgstr "object-info: argümanlardan sonra floş bekleniyordu" @@ -19901,6 +20023,14 @@ msgstr "başvuru adı %s bir sembolik bağ, onu kopyalamak desteklenmiyor" msgid "invalid refspec '%s'" msgstr "geçersiz başvuru belirteci '%s'" +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "'%s' dizgisinde '*' yok" + +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "'%s' yedeğinde '*' yok" + #, c-format msgid "invalid quoting in push-option value: '%s'" msgstr "push-option değerinde geçersiz tırnak içine alım: '%s'" @@ -20020,6 +20150,31 @@ msgstr "remote-curl: yerel bir depo olmadan getirme yapılmaya çalışıldı" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl: git'ten bilinmeyen komut '%s'" +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"Kaldırma için aday gösterilmiş \"%s/%s\" konumundan\n" +"uzak konum okunuyor.\n" +"\n" +"Eğer hâlâ \"remotes\" dizinini kullanıyorsanız\n" +"yapılandırma tabanlı uzak konumlara geçiş yapmanız\n" +"önerilir:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"Eğer geçiş yapamıyorsanız bunu neden kullandığınıza\n" +"dair bir iletiyi adresine\n" +"gönderin." + #, c-format msgid "config remote shorthand cannot begin with '/': %s" msgstr "uzak konum yapılandırma stenografisi '/' ile başlayamaz: %s" @@ -20054,14 +20209,6 @@ msgstr "%s genelde %s ögesini izler, %s değil" msgid "%s tracks both %s and %s" msgstr "%s hem %s hem %s ögelerini izler" -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "dizginin '%s' anahtarında '*' yoktu" - -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "dizginin '%s' değerinde '*' yok" - #, c-format msgid "src refspec %s does not match any" msgstr "kaynak başvuru belirteci %s başka hiçbir şeyle eşleşmiyor" @@ -21948,6 +22095,27 @@ msgstr "her bir yinelemeden önce önbellek ağacını temizle" msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "önbellek ağacındaki geçersizleştirilecek girdi sayısı (öntanımlı 0)" +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk -- " + +msgid "toggle inclusion of blob objects" +msgstr "ikili nesnelerin içerilmesini aç/kapat" + +msgid "toggle inclusion of commit objects" +msgstr "işleme nesnelerinin içerilmesini aç/kapat" + +msgid "toggle inclusion of tag objects" +msgstr "etiket nesnelerinin içerilmesini aç/kapat" + +msgid "toggle inclusion of tree objects" +msgstr "ağaç nesnelerinin içerilmesini aç/kapat" + +msgid "toggle pruning of uninteresting paths" +msgstr "ilgisiz yolların budanmasını aç/kapat" + +msgid "read a pattern list over stdin" +msgstr "stdin'den bir dizgi listesi oku" + #, c-format msgid "commit %s is not marked reachable" msgstr "%s işlemesi ulaşılabilir olarak imlenmedi" @@ -22575,6 +22743,10 @@ msgstr "hata: " msgid "warning: " msgstr "uyarı: " +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() '%s' hatasını verip çıktı (%d)\n" + msgid "Fetching objects" msgstr "Nesneler getiriliyor" From 5b75ad9ee87d9944a7210f14c430aed99d5241fb Mon Sep 17 00:00:00 2001 From: Arkadii Yakovets Date: Tue, 11 Mar 2025 19:39:45 -0700 Subject: [PATCH 014/259] l10n: uk: add 2.49 translation Co-authored-by: Kate Golovanova Co-authored-by: Mikhail T. Co-authored-by: Tamara Lazerka Signed-off-by: Arkadii Yakovets Signed-off-by: Kate Golovanova Signed-off-by: Mikhail T. Signed-off-by: Tamara Lazerka --- po/uk.po | 469 +++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 319 insertions(+), 150 deletions(-) diff --git a/po/uk.po b/po/uk.po index 5f68f36c7db024..39c22289fa62d5 100644 --- a/po/uk.po +++ b/po/uk.po @@ -8,8 +8,8 @@ msgid "" msgstr "" "Project-Id-Version: Git v2.46\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2025-01-04 19:26-0800\n" -"PO-Revision-Date: 2025-01-03 14:17-0800\n" +"POT-Creation-Date: 2025-03-09 15:40-0700\n" +"PO-Revision-Date: 2025-03-09 16:53-0700\n" "Last-Translator: Kateryna Golovanova \n" "Language-Team: Ukrainian \n" "Language: uk\n" @@ -1433,7 +1433,7 @@ msgid "write the archive to this file" msgstr "записати архів до цього файлу" msgid "read .gitattributes in working directory" -msgstr "прочитати .gitattributes робочої директорії" +msgstr "читати .gitattributes робочої директорії" msgid "report archived files on stderr" msgstr "звітувати про заархівовані файли в stderr" @@ -2357,6 +2357,18 @@ msgstr "git archive: помилка протоколу" msgid "git archive: expected a flush" msgstr "git archive: очікувалось flush" +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=<н>] [--[no-]sparse]" + +msgid "problem loading sparse-checkout" +msgstr "проблема із завантаженням розрідженого переходу" + +msgid "Minimum number of objects to request at a time" +msgstr "Мінімальна кількість обʼєктів для запиту за один раз" + +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "Обмежити відсутні обʼєкти поточним розрідженим переходом" + msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" "checkout] [--first-parent] [ [...]] [--] [...]" @@ -3117,15 +3129,11 @@ msgstr "" msgid "git version:\n" msgstr "версія git:\n" -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() завершився невдало з помилкою \"%s\" (%d)\n" - msgid "compiler info: " msgstr "інформація щодо компілятора: " msgid "libc info: " -msgstr "информація щодо libc: " +msgstr "інформація щодо libc: " msgid "not run from a git repository - no hooks to show\n" msgstr "запущено не з git сховища - немає гачків для показу\n" @@ -3448,7 +3456,7 @@ msgid "use .gitattributes only from the index" msgstr "використовувати .gitattributes тільки з індексу" msgid "read file names from stdin" -msgstr "зчитувати назви файлів з stdin" +msgstr "читати назви файлів з stdin" msgid "terminate input and output records by a NUL character" msgstr "завершувати вхідні та вихідні записи символом NUL" @@ -3493,13 +3501,13 @@ msgid "also read contacts from stdin" msgstr "також читати контакти з stdin" msgid "read additional mailmap entries from file" -msgstr "зчитувати додаткові записи mailmap з файлу" +msgstr "читати додаткові записи mailmap з файлу" msgid "blob" msgstr "blob" msgid "read additional mailmap entries from blob" -msgstr "зчитувати додаткові записи mailmap з blob" +msgstr "читати додаткові записи mailmap з blob" msgid "no contacts specified" msgstr "контакти не вказані" @@ -3538,10 +3546,10 @@ msgid "update stat information in the index file" msgstr "оновити статистичну інформацію в індексному файлі" msgid "read list of paths from the standard input" -msgstr "зчитати список шляхів зі стандартного вводу" +msgstr "читати список шляхів зі стандартного вводу" msgid "write the content to temporary files" -msgstr "записати вміст у тимчасові файли" +msgstr "записати вміст до тимчасових файлів" msgid "copy out the files from named stage" msgstr "скопіювати файли з іменованої стадії" @@ -4146,11 +4154,94 @@ msgid "clean.requireForce is true and -f not given: refusing to clean" msgstr "" "clean.requireForce встановлено у true і -f не задано: відмовлено в прибиранні" -msgid "git clone [] [--] []" -msgstr "git clone [<опції>] [--] <сховище> [<директорія>]" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "інфо: Не вдалося додати запозичений обʼєкт для \"%s\": %s\n" + +#, c-format +msgid "failed to stat '%s'" +msgstr "не вдалося виконати stat \"%s\"" + +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s існує і не є директорією" + +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "\"%s\" є символьним посиланням, відмовлено в клонуванні з --local" + +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "не вдалося запустити ітератор для \"%s\"" + +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "символьне посилання \"%s\" існує, відмовлено в клонуванні з --local" + +#, c-format +msgid "failed to unlink '%s'" +msgstr "не вдалося видалити \"%s\"" + +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "неможливо перевірити жорстке посилання \"%s\"" + +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "жорстке посилання відрізняється від джерела в \"%s\"" + +#, c-format +msgid "failed to create link '%s'" +msgstr "не вдалося створити посилання \"%s\"" + +#, c-format +msgid "failed to copy file to '%s'" +msgstr "не вдалося скопіювати файл у \"%s\"" + +#, c-format +msgid "failed to iterate over '%s'" +msgstr "не вдалося здійснити перебір для \"%s\"" + +#, c-format +msgid "done.\n" +msgstr "готово.\n" + +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Клонування пройшло успішно, але не вдалося перейти на гілку.\n" +"Ви можете перевірити, що було додано за допомогою 'git status'\n" +"і повторити спробу за допомогою \"git restore --source=HEAD :/\"\n" + +msgid "remote did not send all necessary objects" +msgstr "віддалене сховище не надіслало всі необхідні обʼєкти" + +#, c-format +msgid "unable to update %s" +msgstr "не вдалося оновити %s" + +msgid "failed to initialize sparse-checkout" +msgstr "не вдалося ініціалізувати розріджений перехід" + +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "віддалений HEAD вказує на неіснуюче посилання, неможливо перейти" + +msgid "unable to checkout working tree" +msgstr "не вдалося завантажити стан робочої директорії" + +msgid "unable to write parameters to config file" +msgstr "не вдалося записати параметри до конфігураційного файлу" + +msgid "cannot repack to clean up" +msgstr "неможливо перепакувати для очищення" + +msgid "cannot unlink temporary alternates file" +msgstr "неможливо видалити тимчасовий файл запозичених обʼєктів" msgid "don't clone shallow repository" -msgstr "не клонувати неглибоке сховище" +msgstr "не клонувати поверхневе сховище" msgid "don't create a checkout" msgstr "не переходити на гілку" @@ -4202,6 +4293,9 @@ msgstr "" msgid "checkout instead of the remote's HEAD" msgstr "перейти до <гілки> замість HEAD віддаленого сховища" +msgid "clone single revision and check out" +msgstr "клонувати одну ревізію і перейти на неї" + msgid "path to git-upload-pack on the remote" msgstr "шлях до git-upload-pack на віддаленому сервері" @@ -4223,9 +4317,8 @@ msgstr "поглибити історію неглибокого клону, з msgid "clone only one branch, HEAD or --branch" msgstr "клонувати лише одну гілку, HEAD або --branch" -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "" -"не клонувати жодних тегів і не слідувати за ними під час отримувань пізніше" +msgid "clone tags, and make later fetches not to follow them" +msgstr "клонувати теги і більше не слідкувати за ними" msgid "any cloned submodules will be shallow" msgstr "будь-які клоновані підмодулі будуть неглибокими" @@ -4269,95 +4362,8 @@ msgstr "uri" msgid "a URI for downloading bundles before fetching from origin remote" msgstr "URI для завантаження пакунків перед отриманням з віддаленого джерела" -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "инфо: Не вдалося додати запозичений обʼєкт для \"%s\": %s\n" - -#, c-format -msgid "failed to stat '%s'" -msgstr "не вдалося виконати stat \"%s\"" - -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s існує і не є директорією" - -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "\"%s\" є символьним посиланням, відмовлено в клонуванні з --local" - -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "не вдалося запустити перебір для \"%s\"" - -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "символьне посилання \"%s\" існує, не можу клонувати з --local" - -#, c-format -msgid "failed to unlink '%s'" -msgstr "не вдалося видалити \"%s\"" - -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "неможливо перевірити жорстке посилання \"%s\"" - -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "жорстке посилання відрізняється від джерела в \"%s\"" - -#, c-format -msgid "failed to create link '%s'" -msgstr "не вдалося створити посилання \"%s\"" - -#, c-format -msgid "failed to copy file to '%s'" -msgstr "не вдалося скопіювати файл у \"%s\"" - -#, c-format -msgid "failed to iterate over '%s'" -msgstr "не вдалося перебрати \"%s\"" - -#, c-format -msgid "done.\n" -msgstr "готово.\n" - -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Клонування пройшло успішно, але не вдалося перейти на гілку.\n" -"Ви можете перевірити, що було додано за допомогою 'git status'\n" -"і повторити спробу за допомогою 'git restore --source=HEAD :/'\n" - -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "Не вдалося знайти віддалену гілку %s для клонування." - -msgid "remote did not send all necessary objects" -msgstr "віддалене сховище не надіслало всі необхідні обʼєкти" - -#, c-format -msgid "unable to update %s" -msgstr "не вдалося оновити %s" - -msgid "failed to initialize sparse-checkout" -msgstr "не вдалося ініціалізувати розріджений перехід" - -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "віддалений HEAD посилається на неіснуючого рефа, неможливо перейти" - -msgid "unable to checkout working tree" -msgstr "не вдалося завантажити стан робочої директорії" - -msgid "unable to write parameters to config file" -msgstr "не вдалося записати параметри до конфігураційного файлу" - -msgid "cannot repack to clean up" -msgstr "неможливо перепакувати, щоб очистити" - -msgid "cannot unlink temporary alternates file" -msgstr "неможливо видалити тимчасовий файл запозичених обʼєктів" +msgid "git clone [] [--] []" +msgstr "git clone [<опції>] [--] <сховище> [<директорія>]" msgid "Too many arguments." msgstr "Забагато аргументів." @@ -4463,6 +4469,10 @@ msgstr "операція віддаленого отримання повідо msgid "Remote branch %s not found in upstream %s" msgstr "Віддалену гілку %s не знайдено у першоджерельному сховищі %s" +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "Віддалена ревізія %s не знайдена у першоджерельному сховищі %s" + msgid "You appear to have cloned an empty repository." msgstr "Здається, ви клонували порожнє сховище." @@ -4639,7 +4649,7 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree: не вдалося прочитати" msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -4649,7 +4659,7 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u<режим>] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[<режим>]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) <коміт> | --fixup [(amend|" "reword):]<коміт>]\n" " [-F <файл> | -m <допис>] [--reset-author] [--allow-empty]\n" @@ -5214,7 +5224,7 @@ msgid "blob-id" msgstr "blob-id" msgid "read config from given blob object" -msgstr "прочитати конфігурацію з наданого blob-обʼєкту" +msgstr "читати конфігурацію з наданого blob-обʼєкту" msgid "Type" msgstr "Тип" @@ -6050,15 +6060,15 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Запустіть \"git remote set-head %s %s\", щоб відстежити зміни, або " "встановіть\n" "\"remote.%s.followRemoteHEAD\" параметр конфігурації на інше значення\n" -"якщо ви не хочете бачити це повідомлення. Зокрема, виконання команди\n" -"\"git config set remote.%s.followRemoteHEAD %s\" вимкне попередження\n" -"доки віддалений сервер не змінить HEAD на щось інше." +"якщо ви не хочете бачити це повідомлення. Зокрема, виконання\n" +"\"git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s\"\n" +"вимкне попередження, доки віддалене призначення не змінить HEAD на щось інше." msgid "multiple branches detected, incompatible with --set-upstream" msgstr "виявлено кілька гілок, несумісних з --set-upstream" @@ -6740,6 +6750,9 @@ msgstr "примусово запускати збирач сміття, нав msgid "repack all other packs except the largest pack" msgstr "перепакувати всі пакунки, крім найбільшого" +msgid "pack prefix to store a pack containing pruned objects" +msgstr "префікс для зберігання пакунка з обрізаними обʼєктами" + #, c-format msgid "failed to parse gc.logExpiry value %s" msgstr "не вдалося розібрати gc.logExpiry значення %s" @@ -7120,7 +7133,7 @@ msgid "show the surrounding function" msgstr "показати навколишню функцію" msgid "read patterns from file" -msgstr "зчитувати шаблони з файлу" +msgstr "читати шаблони з файлу" msgid "match " msgstr "зіставляти <шаблон>" @@ -7196,7 +7209,7 @@ msgid "write the object into the object database" msgstr "записати об’єкт до бази даних об’єктів" msgid "read the object from stdin" -msgstr "прочитати об’єкт з stdin" +msgstr "читати об’єкт з stdin" msgid "store file as is without filters" msgstr "зберегти файл як є без фільтрів" @@ -7547,6 +7560,10 @@ msgstr "не вдалося завершити pack-objects для перепа msgid "Cannot come back to cwd" msgstr "Неможливо повернутися до поточної робочої директорії" +#, c-format +msgid "bad --pack_header: %s" +msgstr "невірний --pack_header: %s" + #, c-format msgid "bad %s" msgstr "невірний %s" @@ -8420,10 +8437,6 @@ msgstr "невідомий варіант стратегії: -X%s" msgid "malformed input line: '%s'." msgstr "невірно сформований рядок вводу: \"%s\"." -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "неможливо продовжити злиття; отримано брудний результат для %d" - msgid "git merge [] [...]" msgstr "git merge [<опції>] [<коміт>...]" @@ -9094,7 +9107,7 @@ msgid "Removing note for object %s\n" msgstr "Видалення нотатки для обʼєкта %s\n" msgid "read objects from stdin" -msgstr "зчитати обʼєкти з stdin" +msgstr "читати обʼєкти з stdin" msgid "load rewriting config for (implies --stdin)" msgstr "" @@ -9216,7 +9229,7 @@ msgid "attempt to remove non-existent note is not an error" msgstr "спроба видалити неіснуючу нотатку не є помилкою" msgid "read object names from the standard input" -msgstr "зчитати імена обʼєктів зі стандартного вводу" +msgstr "читати імена обʼєктів зі стандартного вводу" msgid "do not remove, show only" msgstr "не видаляти, тільки показувати" @@ -9245,6 +9258,13 @@ msgstr "" "git pack-objects [<опції>] <базова-назва> [< <список-посилань> | < <список-" "обʼєктів>]" +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "неприпустима --name-hash-version опція: %d" + +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "наразі --write-bitmap-index потребує --name-hash-version=1" + #, c-format msgid "" "write_reuse_object: could not locate %s, expected at offset % in " @@ -9481,7 +9501,7 @@ msgid "do not create an empty pack output" msgstr "не створювати вивід порожнього пакунка" msgid "read revision arguments from standard input" -msgstr "зчитувати аргументи ревізії зі стандартного вводу" +msgstr "читати аргументи ревізії зі стандартного вводу" msgid "limit the objects to those that are not yet packed" msgstr "обмежувати обʼєкти тільки тими, які ще не запаковані" @@ -9570,6 +9590,10 @@ msgstr "протокол" msgid "exclude any configured uploadpack.blobpackfileuri with this protocol" msgstr "вилучити всі налаштовані uploadpack.blobpackfileuri з цим протоколом" +msgid "use the specified name-hash function to group similar objects" +msgstr "" +"використовувати вказану name-hash функцію для групування схожих обʼєктів" + #, c-format msgid "delta chain depth %d is too deep, forcing %d" msgstr "глибина дельта ланцюжка %d занадто глибока, примусове %d" @@ -10826,8 +10850,8 @@ msgstr "не вказано журнал посилань для видален msgid "invalid ref format: %s" msgstr "неприпустимий формат посилання: %s" -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format=<формат> [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format=<формат> [--no-reflog] [--dry-run]" msgid "git refs verify [--strict] [--verbose]" msgstr "git refs verify [--strict] [--verbose]" @@ -10838,6 +10862,9 @@ msgstr "вкажіть формат посилання, в який потріб msgid "perform a non-destructive dry-run" msgstr "виконати неруйнівний пробний запуск" +msgid "drop reflogs entirely during the migration" +msgstr "повністю видалити журнал посилань під час міграції" + msgid "missing --ref-format=" msgstr "відсутній --ref-format=<формат>" @@ -11312,8 +11339,14 @@ msgstr "Не видалятиме всі URL-адреси, що не є приз msgid "be verbose; must be placed before a subcommand" msgstr "розгорнутий вивід; має стояти перед підкомандою" -msgid "git repack []" -msgstr "git repack [<опції>]" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=<н>] [--depth=<н>] [--threads=<н>] [--keep-pack=<назва-пакунка>]\n" +"[--write-midx] [--name-hash-version=<н>]" msgid "" "Incremental repacks are incompatible with bitmap indexes. Use\n" @@ -11388,6 +11421,10 @@ msgstr "передати --no-reuse-delta до git-pack-objects" msgid "pass --no-reuse-object to git-pack-objects" msgstr "передати --no-reuse-object до git-pack-objects" +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "вказати версію назви хешу для групування схожих обʼєктів за шляхом" + msgid "do not run git-update-server-info" msgstr "не запускати git-update-server-info" @@ -11437,9 +11474,6 @@ msgstr "знайти геометричну прогресію з факторо msgid "write a multi-pack index of the resulting packs" msgstr "записати multi-pack-index результуючих пакунків" -msgid "pack prefix to store a pack containing pruned objects" -msgstr "префікс для зберігання пакунка з обрізаними обʼєктами" - msgid "pack prefix to store a pack containing filtered out objects" msgstr "префікс для зберігання пакунка з відфільтрованими обʼєктами" @@ -11656,9 +11690,6 @@ msgstr "тільки один шаблон може бути заданий з - msgid "need some commits to replay" msgstr "потрібні деякі коміти для відтворення" -msgid "--onto and --advance are incompatible" -msgstr "--onto та --advance несумісні" - msgid "all positive revisions given must be references" msgstr "всі надані позитивні ревізії мають бути посиланнями" @@ -12135,7 +12166,7 @@ msgid "use stateless RPC protocol" msgstr "використовувати протокол RPC без збереження стану" msgid "read refs from stdin" -msgstr "прочитати посилання з stdin" +msgstr "читати посилання з stdin" msgid "print status from remote helper" msgstr "вивести статус з віддаленого помічника" @@ -13623,7 +13654,7 @@ msgid "with --stdin: input lines are terminated by null bytes" msgstr "з --stdin: вхідні рядки завершуються нульовими байтами" msgid "read list of paths to be updated from standard input" -msgstr "прочитати список шляхів для оновлення зі стандартного вводу" +msgstr "читати список шляхів для оновлення зі стандартного вводу" msgid "add entries from standard input to the index" msgstr "додати записи зі стандартного вводу до індексу" @@ -14335,8 +14366,11 @@ msgstr "Імпортувати GNU Arch сховище до Git" msgid "Create an archive of files from a named tree" msgstr "Створити архів файлів з названого дерева" +msgid "Download missing objects in a partial clone" +msgstr "Завантажити відсутні обʼєкти у частковому клонуванні" + msgid "Use binary search to find the commit that introduced a bug" -msgstr "Використати бінарний пошук, щоб знайти коміт, який вніс помилку" +msgstr "Використати бінарний пошук, щоб знайти коміт, який ввів помилку" msgid "Show what revision and author last modified each line of a file" msgstr "Показати, яка ревізія та автор востаннє змінювали кожен рядок файлу" @@ -15728,7 +15762,7 @@ msgstr "помилка протоколу: неочікувані здібнос #, c-format msgid "protocol error: expected shallow sha-1, got '%s'" -msgstr "помилка протоколу: очікувалось неглибоке sha-1, отримано \"%s\"" +msgstr "помилка протоколу: очікувалось поверхневе sha-1, отримано \"%s\"" msgid "repository on the other end cannot be shallow" msgstr "сховище на іншому кінці не може бути неглибоким" @@ -16256,6 +16290,12 @@ msgstr "неприпустимий аргумент до %s" msgid "invalid regex given to -I: '%s'" msgstr "неприпустимий regex, переданий до -I: \"%s\"" +msgid "-G requires a non-empty argument" +msgstr "-G потребує непорожнього аргументу" + +msgid "-S requires a non-empty argument" +msgstr "-S потребує непорожнього аргументу" + #, c-format msgid "failed to parse --submodule option parameter: '%s'" msgstr "не вдалося розібрати параметр опції --submodule: \"%s\"" @@ -17247,7 +17287,7 @@ msgstr "" #, c-format msgid "git: '%s' is not a git command. See 'git --help'." -msgstr "git: \"%s\" не є командою git. Дивітья git --help." +msgstr "git: \"%s\" не є командою git. Дивіться git --help." msgid "Uh oh. Your system reports no Git commands at all." msgstr "Ой-ой. Ваша система повідомляє про повну відсутність Git команд." @@ -18447,6 +18487,10 @@ msgstr "файли \"%s\" та \"%s\" відрізняються за вміст msgid "unable to write file %s" msgstr "не вдалося записати файл %s" +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "не вдалося записати файл %s, який постійно зникає" + #, c-format msgid "unable to set permission to '%s'" msgstr "не вдалося встановити дозволи для \"%s\"" @@ -19020,6 +19064,52 @@ msgstr "невідомий перемикач \"%c\"" msgid "unknown non-ascii option in string: `%s'" msgstr "невідомий non-ascii параметр у рядку: \"%s\"" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid " <%s>" +msgstr " <%s>" + msgid "..." msgstr "..." @@ -19106,6 +19196,21 @@ msgstr "невірне булеве значення оточення \"%s\" д msgid "failed to parse %s" msgstr "не вдалося розібрати %s" +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "не вдалося пройти дочірні елементи дерева %s: не знайдено" + +#, c-format +msgid "failed to find object %s" +msgstr "не вдалося знайти обʼєкт %s" + +#, c-format +msgid "failed to find tag %s" +msgstr "не вдалося знайти тег %s" + +msgid "failed to setup revision walk" +msgstr "не вдалося налаштувати проходження по ревізіям" + #, c-format msgid "Could not make %s writable by group" msgstr "Не вдалося зробити %s доступним для запису групою" @@ -19255,6 +19360,22 @@ msgstr "назва віддаленого promisor не може починат msgid "could not fetch %s from promisor remote" msgstr "не вдалося отримати %s з віддаленого promisor" +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "відоме віддалене сховище з імʼям \"%s\" має URL \"%s\" замість \"%s\"" + +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "невідоме значення \"%s\" для параметра конфігурації \"%s\"" + +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "невідомий елемент \"%s\" з віддаленої інформації" + +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "прийнятий віддалений promisor \"%s\" не знайдено" + msgid "object-info: expected flush after arguments" msgstr "object-info: очікувався flush після аргументів" @@ -20083,6 +20204,14 @@ msgstr "" msgid "invalid refspec '%s'" msgstr "неприпустимий визначник посилання \"%s\"" +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "шаблон \"%s\" не має \"*\"" + +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "заміна \"%s\" не має \"*\"" + #, c-format msgid "invalid quoting in push-option value: '%s'" msgstr "неприпустимі лапки у значенні push-опції: \"%s\"" @@ -20132,7 +20261,7 @@ msgstr "віддалений сервер надіслав неочікуван msgid "unable to rewind rpc post data - try increasing http.postBuffer" msgstr "" -"не вдалося перемотати вперед rpc post дані - спробуйте збільшити " +"не вдалося перемотати вперед rpc post дані - спробуйте збільшити " "http.postBuffer" #, c-format @@ -20206,9 +20335,32 @@ msgstr "remote-curl: спроба отримання без локального msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl: невідома команда \"%s\" з git" +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"читання віддаленого призначення з \"%s/%s\", який номіновано на вилучення.\n" +"\n" +"Якщо ви все ще використовуєте директорію \"remotes/\", рекомендується\n" +"перейти віддалені призначення на основі конфігурації:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"Якщо ви не можете цього зробити, будь ласка, дайте нам знати, чому ви все ще " +"використовуєте її\n" +"надіславши листа на адресу ." + #, c-format msgid "config remote shorthand cannot begin with '/': %s" -msgstr "скорочення віддаленої конфігураціі не може починатися з \"/\": %s" +msgstr "скорочене ім'я віддаленої конфігураціі не може починатися з \"/\": %s" msgid "more than one receivepack given, using the first" msgstr "надано більше одного пакунка для отримання, використано перший" @@ -20240,14 +20392,6 @@ msgstr "%s зазвичай відстежує %s, а не %s" msgid "%s tracks both %s and %s" msgstr "%s відстежує як %s, так і %s" -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "ключ \"%s\" шаблону не містив '*'" - -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "значення \"%s\" шаблону не містить '*'" - #, c-format msgid "src refspec %s does not match any" msgstr "визначник посилання джерела %s не збігається з жодним" @@ -22179,6 +22323,27 @@ msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "" "кількість записів у дереві кешу, які потрібно анулювати (за замовчуванням 0)" +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk <опції> -- <опції-ревізії>." + +msgid "toggle inclusion of blob objects" +msgstr "перемикач включення обʼєктів blob" + +msgid "toggle inclusion of commit objects" +msgstr "перемикач включення обʼєктів коміту" + +msgid "toggle inclusion of tag objects" +msgstr "перемикач включення обʼєктів тегів" + +msgid "toggle inclusion of tree objects" +msgstr "перемикач включення обʼєктів дерева" + +msgid "toggle pruning of uninteresting paths" +msgstr "перемикач обрізання нецікавих шляхів" + +msgid "read a pattern list over stdin" +msgstr "читати список шаблонів через stdin" + #, c-format msgid "commit %s is not marked reachable" msgstr "коміт %s не позначений як досяжний" @@ -22819,6 +22984,10 @@ msgstr "помилка: " msgid "warning: " msgstr "попередження: " +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() завершився невдало з помилкою \"%s\" (%d)\n" + msgid "Fetching objects" msgstr "Отримання обʼєктів" From ed99a5d9b8d3d7ac0074ffcffc36b955edd17954 Mon Sep 17 00:00:00 2001 From: Teng Long Date: Sun, 9 Mar 2025 20:32:13 +0800 Subject: [PATCH 015/259] l10n: zh_CN: updated translation for 2.49 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Helped-by: 依云 Helped-by: Jiang Xin Signed-off-by: Teng Long --- po/zh_CN.po | 540 +++++++++++++++++++++++++++++++++++----------------- 1 file changed, 370 insertions(+), 170 deletions(-) diff --git a/po/zh_CN.po b/po/zh_CN.po index 12a0fb510b10cc..5cde4011e750ce 100644 --- a/po/zh_CN.po +++ b/po/zh_CN.po @@ -93,6 +93,7 @@ # pack index | 包索引 # packfile | 包文件 # parent | 父提交 +# partial clone | 部分克隆 # patch | 补丁 # pathspec | 路径规格 # pattern | 模式 @@ -154,8 +155,8 @@ msgid "" msgstr "" "Project-Id-Version: Git\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2025-01-02 20:43+0800\n" -"PO-Revision-Date: 2025-01-05 19:01+0800\n" +"POT-Creation-Date: 2025-03-09 20:34+0800\n" +"PO-Revision-Date: 2025-03-12 14:47+0800\n" "Last-Translator: Teng Long \n" "Language-Team: GitHub \n" "Language: zh_CN\n" @@ -2962,6 +2963,22 @@ msgstr "git archive：协议错误" msgid "git archive: expected a flush" msgstr "git archive：应有一个 flush 包" +#: builtin/backfill.c +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +#: builtin/backfill.c +msgid "problem loading sparse-checkout" +msgstr "加载稀疏检出时出现问题" + +#: builtin/backfill.c +msgid "Minimum number of objects to request at a time" +msgstr "单次请求的最少对象数量" + +#: builtin/backfill.c +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "将缺少的对象限制在当前的稀疏检出中" + #: builtin/bisect.c msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" @@ -3381,7 +3398,7 @@ msgstr "显示作者的邮箱而不是名字（默认：关闭）" msgid "ignore whitespace differences" msgstr "忽略空白差异" -#: builtin/blame.c builtin/log.c +#: builtin/blame.c builtin/clone.c builtin/log.c msgid "rev" msgstr "版本" @@ -3803,8 +3820,8 @@ msgstr "HEAD 没有位于 /refs/heads 之下！" #: builtin/branch.c msgid "" -"branch with --recurse-submodules can only be used if submodule." -"propagateBranches is enabled" +"branch with --recurse-submodules can only be used if " +"submodule.propagateBranches is enabled" msgstr "" "带有 --recurse-submodules 的分支只能在 submodule.propagateBranches 启用时使用" @@ -3891,11 +3908,6 @@ msgstr "不再支持选项 '--set-upstream'。请使用 '--track' 或 '--set-ups msgid "git version:\n" msgstr "git 版本：\n" -#: builtin/bugreport.c -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() 失败，错误为 '%s'（%d）\n" - #: builtin/bugreport.c msgid "compiler info: " msgstr "编译器信息：" @@ -5134,8 +5146,112 @@ msgid "clean.requireForce is true and -f not given: refusing to clean" msgstr "clean.requireForce 设置为 true 且未提供 -f 选项：拒绝执行清理动作" #: builtin/clone.c -msgid "git clone [] [--] []" -msgstr "git clone [<选项>] [--] <仓库> [<路径>]" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "info: 不能为 '%s' 添加一个备用：%s\n" + +#: builtin/clone.c builtin/diff.c builtin/rm.c grep.c setup.c +#, c-format +msgid "failed to stat '%s'" +msgstr "无法对 '%s' 调用 stat" + +#: builtin/clone.c +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s 存在且不是一个目录" + +#: builtin/clone.c +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "'%s' 为符号链接，拒绝用 --local 克隆" + +#: builtin/clone.c +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "无法在 '%s' 上启动迭代器" + +#: builtin/clone.c +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "符号链接 '%s' 存在，拒绝用 --local 克隆" + +#: builtin/clone.c compat/precompose_utf8.c +#, c-format +msgid "failed to unlink '%s'" +msgstr "无法删除 '%s'" + +#: builtin/clone.c +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "无法检查 '%s' 处的硬链接" + +#: builtin/clone.c +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "硬链接与 '%s' 处的来源不同" + +#: builtin/clone.c +#, c-format +msgid "failed to create link '%s'" +msgstr "无法创建链接 '%s'" + +#: builtin/clone.c +#, c-format +msgid "failed to copy file to '%s'" +msgstr "无法拷贝文件至 '%s'" + +#: builtin/clone.c refs/files-backend.c +#, c-format +msgid "failed to iterate over '%s'" +msgstr "无法在 '%s' 上迭代" + +#: builtin/clone.c +#, c-format +msgid "done.\n" +msgstr "完成。\n" + +#: builtin/clone.c +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"克隆成功，但是检出失败。\n" +"您可以通过 'git status' 检查哪些已被检出，然后使用命令\n" +"'git restore --source=HEAD :/' 重试\n" + +#: builtin/clone.c fetch-pack.c +msgid "remote did not send all necessary objects" +msgstr "远程没有发送所有必需的对象" + +#: builtin/clone.c +#, c-format +msgid "unable to update %s" +msgstr "不能更新 %s" + +#: builtin/clone.c +msgid "failed to initialize sparse-checkout" +msgstr "无法初始化稀疏检出" + +#: builtin/clone.c +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "远程 HEAD 指向一个不存在的引用，无法检出" + +#: builtin/clone.c +msgid "unable to checkout working tree" +msgstr "不能检出工作区" + +#: builtin/clone.c +msgid "unable to write parameters to config file" +msgstr "无法将参数写入配置文件" + +#: builtin/clone.c +msgid "cannot repack to clean up" +msgstr "无法执行 repack 来清理" + +#: builtin/clone.c +msgid "cannot unlink temporary alternates file" +msgstr "无法删除临时的备用文件" #: builtin/clone.c msgid "don't clone shallow repository" @@ -5207,6 +5323,10 @@ msgstr "使用 <名称> 而不是 'origin' 去跟踪上游" msgid "checkout instead of the remote's HEAD" msgstr "检出 <分支> 而不是远程 HEAD" +#: builtin/clone.c +msgid "clone single revision and check out" +msgstr "克隆单个版本 <版本> 并检出" + #: builtin/clone.c msgid "path to git-upload-pack on the remote" msgstr "远程 git-upload-pack 路径" @@ -5236,8 +5356,8 @@ msgid "clone only one branch, HEAD or --branch" msgstr "只克隆一个分支、HEAD 或 --branch" #: builtin/clone.c -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "不要克隆任何标签，并且后续获取操作也不下载它们" +msgid "clone tags, and make later fetches not to follow them" +msgstr "克隆标签，并在后续获取时不跟随它们" #: builtin/clone.c msgid "any cloned submodules will be shallow" @@ -5294,117 +5414,8 @@ msgid "a URI for downloading bundles before fetching from origin remote" msgstr "用于在从 origin 远程获取之前下载归档包的 URI" #: builtin/clone.c -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "info: 不能为 '%s' 添加一个备用：%s\n" - -#: builtin/clone.c builtin/diff.c builtin/rm.c grep.c setup.c -#, c-format -msgid "failed to stat '%s'" -msgstr "无法对 '%s' 调用 stat" - -#: builtin/clone.c -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s 存在且不是一个目录" - -#: builtin/clone.c -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "'%s' 为符号链接，拒绝用 --local 克隆" - -#: builtin/clone.c -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "无法在 '%s' 上启动迭代器" - -#: builtin/clone.c -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "符号链接 '%s' 存在，拒绝用 --local 克隆" - -#: builtin/clone.c compat/precompose_utf8.c -#, c-format -msgid "failed to unlink '%s'" -msgstr "无法删除 '%s'" - -#: builtin/clone.c -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "无法检查 '%s' 处的硬链接" - -#: builtin/clone.c -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "硬链接与 '%s' 处的源不同" - -#: builtin/clone.c -#, c-format -msgid "failed to create link '%s'" -msgstr "无法创建链接 '%s'" - -#: builtin/clone.c -#, c-format -msgid "failed to copy file to '%s'" -msgstr "无法拷贝文件至 '%s'" - -#: builtin/clone.c refs/files-backend.c -#, c-format -msgid "failed to iterate over '%s'" -msgstr "无法在 '%s' 上迭代" - -#: builtin/clone.c -#, c-format -msgid "done.\n" -msgstr "完成。\n" - -#: builtin/clone.c -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"克隆成功，但是检出失败。\n" -"您可以通过 'git status' 检查哪些已被检出，然后使用命令\n" -"'git restore --source=HEAD :/' 重试\n" - -#: builtin/clone.c -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "不能发现要克隆的远程分支 %s。" - -#: builtin/clone.c fetch-pack.c -msgid "remote did not send all necessary objects" -msgstr "远程没有发送所有必需的对象" - -#: builtin/clone.c -#, c-format -msgid "unable to update %s" -msgstr "不能更新 %s" - -#: builtin/clone.c -msgid "failed to initialize sparse-checkout" -msgstr "无法初始化稀疏检出" - -#: builtin/clone.c -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "远程 HEAD 指向一个不存在的引用，无法检出" - -#: builtin/clone.c -msgid "unable to checkout working tree" -msgstr "不能检出工作区" - -#: builtin/clone.c -msgid "unable to write parameters to config file" -msgstr "无法将参数写入配置文件" - -#: builtin/clone.c -msgid "cannot repack to clean up" -msgstr "无法执行 repack 来清理" - -#: builtin/clone.c -msgid "cannot unlink temporary alternates file" -msgstr "无法删除临时的 alternates 文件" +msgid "git clone [] [--] []" +msgstr "git clone [<选项>] [--] <仓库> [<目录>]" #: builtin/clone.c msgid "Too many arguments." @@ -5530,6 +5541,11 @@ msgstr "远程传输报告错误" msgid "Remote branch %s not found in upstream %s" msgstr "远程分支 %s 在上游 %s 未发现" +#: builtin/clone.c +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "在上游 %2$s 未发现远程版本 %1$s" + #: builtin/clone.c msgid "You appear to have cloned an empty repository." msgstr "您似乎克隆了一个空仓库。" @@ -5593,7 +5609,8 @@ msgstr "" "[no-]progress]\n" " <切分选项>" -#: builtin/commit-graph.c builtin/fetch.c builtin/log.c builtin/repack.c +#: builtin/commit-graph.c builtin/fetch.c builtin/gc.c builtin/log.c +#: builtin/repack.c msgid "dir" msgstr "目录" @@ -5751,7 +5768,7 @@ msgstr "git commit-tree：无法读取" #: builtin/commit.c msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -5761,7 +5778,7 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u<模式>] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[<模式>]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) <提交> | --fixup [(amend|" "reword):]<提交>]\n" " [-F <文件> | -m <消息>] [--reset-author] [--allow-empty]\n" @@ -7473,14 +7490,13 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" -"运行 'git remote set-head %s %s' 以跟随更改，或者\n" -"如果您不想看到此消息，则将'remote.%s.followRemoteHEAD' 配置选项设置为不同的" -"值。\n" -"特别地，运行 'git config set remote.%s.followRemoteHEAD %s' 将禁用警告，直到" -"远程将 HEAD 更改为其他内容。\"" +"运行 'git remote set-head %s %s' 同步变更，或通过配置选项\n" +"'remote.%s.followRemoteHEAD' 设置为其他值以屏蔽此提示。具体可通过\n" +"运行命令 'git config remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"以禁用该警告，直到远程将 HEAD 更改为其他内容。" #: builtin/fetch.c msgid "multiple branches detected, incompatible with --set-upstream" @@ -7721,8 +7737,8 @@ msgstr "协议不支持 --negotiate-only，退出" #: builtin/fetch.c msgid "" -"--filter can only be used with the remote configured in extensions." -"partialclone" +"--filter can only be used with the remote configured in " +"extensions.partialclone" msgstr "只可以将 --filter 用于在 extensions.partialclone 中配置的远程仓库" #: builtin/fetch.c @@ -8328,6 +8344,10 @@ msgstr "强制执行 gc 即使另外一个 gc 正在执行" msgid "repack all other packs except the largest pack" msgstr "除了最大的包之外，对所有其它包文件重新打包" +#: builtin/gc.c builtin/repack.c +msgid "pack prefix to store a pack containing pruned objects" +msgstr "用于存储修剪对象的包前缀" + #: builtin/gc.c #, c-format msgid "failed to parse gc.logExpiry value %s" @@ -9327,6 +9347,11 @@ msgstr "无法完成 pack-objects 来重新打包本地链接" msgid "Cannot come back to cwd" msgstr "无法返回当前工作目录" +#: builtin/index-pack.c builtin/unpack-objects.c +#, c-format +msgid "bad --pack_header: %s" +msgstr "错误的 --pack_header：%s" + #: builtin/index-pack.c #, c-format msgid "bad %s" @@ -10432,11 +10457,6 @@ msgstr "未知的策略选项：-X%s" msgid "malformed input line: '%s'." msgstr "格式错误的输入行：'%s'。" -#: builtin/merge-tree.c -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "合并无法继续；得到不干净的结果 %d" - #: builtin/merge.c msgid "git merge [] [...]" msgstr "git merge [<选项>] [<提交>...]" @@ -11452,6 +11472,15 @@ msgid "" "git pack-objects [] [< | < ]" msgstr "git pack-objects [<选项>] <前缀名称> [< <引用列表> | < <对象列表>]" +#: builtin/pack-objects.c +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "无效的 --name-hash-version 选项：%d" + +#: builtin/pack-objects.c +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "当前，--write-bitmap-index 要求 --name-hash-version=1" + #: builtin/pack-objects.c #, c-format msgid "" @@ -11857,7 +11886,11 @@ msgstr "协议" #: builtin/pack-objects.c msgid "exclude any configured uploadpack.blobpackfileuri with this protocol" -msgstr "使用此协议排除任何已配置的 uploadpack.blobpackfileuri" +msgstr "排除掉采用该协议的 uploadpack.blobpackfileuri 配置项" + +#: builtin/pack-objects.c +msgid "use the specified name-hash function to group similar objects" +msgstr "使用指定的名称哈希函数对相似的对象进行分组" #: builtin/pack-objects.c #, c-format @@ -12284,8 +12317,8 @@ msgid "" "upstream, see 'push.autoSetupRemote' in 'git help config'.\n" msgstr "" "\n" -"为了让没有追踪上游的分支自动配置，参见 'git help config' 中的 push." -"autoSetupRemote。\n" +"为了让没有追踪上游的分支自动配置，参见 'git help config' 中的 " +"push.autoSetupRemote。\n" #: builtin/push.c #, c-format @@ -13294,8 +13327,8 @@ msgid "invalid ref format: %s" msgstr "无效的引用格式：%s" #: builtin/refs.c -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format=<格式> [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format=<格式> [--no-reflog] [--dry-run]" #: builtin/refs.c msgid "git refs verify [--strict] [--verbose]" @@ -13309,6 +13342,10 @@ msgstr "指定要转换的引用格式" msgid "perform a non-destructive dry-run" msgstr "进行非破坏性的试运行（dry-run）" +#: builtin/refs.c +msgid "drop reflogs entirely during the migration" +msgstr "在迁移期间丢弃引用日志" + #: builtin/refs.c msgid "missing --ref-format=" msgstr "缺少 --ref-format=<格式>" @@ -13881,8 +13918,14 @@ msgid "be verbose; must be placed before a subcommand" msgstr "冗长输出；必须置于子命令之前" #: builtin/repack.c -msgid "git repack []" -msgstr "git repack [<选项>]" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=<包名>]\n" +"[--write-midx] [--name-hash-version=]" #: builtin/repack.c msgid "" @@ -13974,6 +14017,11 @@ msgstr "向 git-pack-objects 传递参数 --no-reuse-delta" msgid "pass --no-reuse-object to git-pack-objects" msgstr "向 git-pack-objects 传递参数 --no-reuse-object" +#: builtin/repack.c +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "指定要使用的名称哈希算法，以实现按照路径对相似对象分组" + #: builtin/repack.c msgid "do not run git-update-server-info" msgstr "不运行 git-update-server-info" @@ -14038,10 +14086,6 @@ msgstr "使用因子 查找几何级数" msgid "write a multi-pack index of the resulting packs" msgstr "写入结果包的多包索引" -#: builtin/repack.c -msgid "pack prefix to store a pack containing pruned objects" -msgstr "储存被清除的对象的包的前缀" - #: builtin/repack.c msgid "pack prefix to store a pack containing filtered out objects" msgstr "储存被过滤的对象的包的前缀" @@ -14314,10 +14358,6 @@ msgstr "只能为 -l 提供一个模式" msgid "need some commits to replay" msgstr "需要一些提交来重放" -#: builtin/replay.c -msgid "--onto and --advance are incompatible" -msgstr "--onto 和 --advance 不兼容" - #: builtin/replay.c msgid "all positive revisions given must be references" msgstr "提供的所有正向版本必须为引用" @@ -17545,6 +17585,10 @@ msgstr "将一个 GNU Arch 仓库导入到 Git" msgid "Create an archive of files from a named tree" msgstr "基于一个指定的树创建文件存档" +#: command-list.h +msgid "Download missing objects in a partial clone" +msgstr "下载部分克隆中缺失的对象" + #: command-list.h msgid "Use binary search to find the commit that introduced a bug" msgstr "通过二分查找定位引入 bug 的提交" @@ -18927,8 +18971,8 @@ msgid "" "remote URLs cannot be configured in file directly or indirectly included by " "includeIf.hasconfig:remote.*.url" msgstr "" -"远程 URL 不能在文件中配置，不管直接地还是通过 includeIf.hasconfig:remote.*." -"url 间接地包含。" +"远程 URL 不能在文件中配置，不管直接地还是通过 " +"includeIf.hasconfig:remote.*.url 间接地包含。" #: config.c #, c-format @@ -19925,6 +19969,14 @@ msgstr "%s 的参数无效" msgid "invalid regex given to -I: '%s'" msgstr "选项 -I 的正则表达式无效：'%s'" +#: diff.c +msgid "-G requires a non-empty argument" +msgstr "-G 需要一个非空参数" + +#: diff.c +msgid "-S requires a non-empty argument" +msgstr "-S 需要一个非空参数" + #: diff.c #, c-format msgid "failed to parse --submodule option parameter: '%s'" @@ -22430,7 +22482,7 @@ msgstr "无法读取替代文件" #: object-file.c msgid "unable to move new alternates file into place" -msgstr "无法将新的替代文件移动到位" +msgstr "无法将新的备用文件移动到位" #: object-file.c #, c-format @@ -22551,6 +22603,11 @@ msgstr "文件 '%s' 和 '%s' 的内容不同" msgid "unable to write file %s" msgstr "无法写文件 %s" +#: object-file.c +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "无法写入反复消失的文件 %s" + #: object-file.c #, c-format msgid "unable to set permission to '%s'" @@ -23233,6 +23290,55 @@ msgstr "未知开关 `%c'" msgid "unknown non-ascii option in string: `%s'" msgstr "字符串中未知的非 ascii 字符选项：`%s'" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid " <%s>" +msgstr " <%s>" + #: parse-options.c msgid "..." msgstr "..." @@ -23336,6 +23442,25 @@ msgstr "对于 '%2$s' 的错误的布尔环境取值 '%1$s'" msgid "failed to parse %s" msgstr "无法解析 %s" +#: path-walk.c +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "无法遍历树 %s 的子节点：未找到" + +#: path-walk.c +#, c-format +msgid "failed to find object %s" +msgstr "无法找到对象 %s" + +#: path-walk.c +#, c-format +msgid "failed to find tag %s" +msgstr "无法找到标签 %s" + +#: path-walk.c +msgid "failed to setup revision walk" +msgstr "无法设置版本遍历" + #: path.c #, c-format msgid "Could not make %s writable by group" @@ -23517,6 +23642,26 @@ msgstr "promisor 远程名称不能以 '/' 开始：%s" msgid "could not fetch %s from promisor remote" msgstr "无法从承诺者远程获取 %s" +#: promisor-remote.c +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "已知远程名称为 '%s'，但 url 为 '%s' 而不是 '%s'" + +#: promisor-remote.c +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "配置项 '%2$s' 为未知的取值 '%1$s'" + +#: promisor-remote.c +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "远程信息中的未知元素 '%s'" + +#: promisor-remote.c +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "未找到已接受的承诺者远程 '%s'" + #: protocol-caps.c msgid "object-info: expected flush after arguments" msgstr "object-info：在参数之后应有一个 flush" @@ -24484,6 +24629,16 @@ msgstr "引用名 %s 是一个符号引用，不支持复制" msgid "invalid refspec '%s'" msgstr "无效的引用规格：'%s'" +#: refspec.c +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "模式 '%s' 没有 '*'" + +#: refspec.c +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "替换 '%s' 没有 '*'" + #: remote-curl.c #, c-format msgid "invalid quoting in push-option value: '%s'" @@ -24635,6 +24790,28 @@ msgstr "remote-curl：尝试没有本地仓库下获取" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl：未知的来自 git 的命令 '%s'" +#: remote.c +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"正在从 \"%s/%s\" 读取远程内容，该内容被提名删除。\n" +"\n" +"如果你仍然在使用 \"remotes/\" 目录，建议迁移为基于配置远程的方式：\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"如果你无法迁移，请发送电子邮件至 告知我们你\n" +"仍然需要使用它的原因。" + #: remote.c #, c-format msgid "config remote shorthand cannot begin with '/': %s" @@ -24678,16 +24855,6 @@ msgstr "%s 通常跟踪 %s，而非 %s" msgid "%s tracks both %s and %s" msgstr "%s 同时跟踪 %s 和 %s" -#: remote.c -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "模式的键 '%s' 没有 '*'" - -#: remote.c -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "模式的值 '%s' 没有 '*'" - #: remote.c #, c-format msgid "src refspec %s does not match any" @@ -26956,6 +27123,34 @@ msgstr "在每次迭代前清除缓存树" msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "缓存树中无效化的条目数量（默认 0）" +#: t/helper/test-path-walk.c +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk <选项> -- <版本选项>" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of blob objects" +msgstr "切换是否包含数据对象" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of commit objects" +msgstr "切换是否包含提交对象" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of tag objects" +msgstr "切换是否包含标签对象" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of tree objects" +msgstr "切换是否包含树对象" + +#: t/helper/test-path-walk.c +msgid "toggle pruning of uninteresting paths" +msgstr "切换对无趣路径的修剪" + +#: t/helper/test-path-walk.c +msgid "read a pattern list over stdin" +msgstr "从标准输入读取模式列表" + #: t/helper/test-reach.c #, c-format msgid "commit %s is not marked reachable" @@ -27703,6 +27898,11 @@ msgstr "错误：" msgid "warning: " msgstr "警告：" +#: version.c +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() 失败，错误为 '%s'（%d）\n" + #: walker.c msgid "Fetching objects" msgstr "正在获取对象" From 4b68faf6b93311254efad80e554780e372deb42f Mon Sep 17 00:00:00 2001 From: Junio C Hamano Date: Wed, 12 Mar 2025 12:06:30 -0700 Subject: [PATCH 016/259] A bit more updates after -rc2 Signed-off-by: Junio C Hamano --- Documentation/RelNotes/2.49.0.adoc | 1 + 1 file changed, 1 insertion(+) diff --git a/Documentation/RelNotes/2.49.0.adoc b/Documentation/RelNotes/2.49.0.adoc index a173eb193075c2..494c83096fae8d 100644 --- a/Documentation/RelNotes/2.49.0.adoc +++ b/Documentation/RelNotes/2.49.0.adoc @@ -285,3 +285,4 @@ Fixes since v2.48 (merge 832f56f06a jc/doc-boolean-synonyms later to maint). (merge 3eeed876a9 ac/doc-http-ssl-type-config later to maint). (merge c268e3285d jc/breaking-changes-early-adopter-option later to maint). + (merge 0d03fda6a5 pb/doc-follow-remote-head later to maint). From ab0072438931c9b4f7ee2d94ca9465ceebb2af43 Mon Sep 17 00:00:00 2001 From: Bagas Sanjaya Date: Tue, 4 Mar 2025 15:26:27 +0700 Subject: [PATCH 017/259] l10n: po-id for 2.49 Update following components: * builtin/clone.c * builtin/commit.c * builtin/fetch.c * builtin/index-pack.c * builtin/pack-objects.c * builtin/refs.c * builtin/repack.c * builtin/unpack-objects.c * command-list.h * diff.c * object-file.c * parse-options.c * promisor-remote.c * refspec.c * remote.c Translate following new components: * path-walk.c * builtin/backfill.c * t/helper/test-path-walk.c Signed-off-by: Bagas Sanjaya --- po/id.po | 520 ++++++++++++++++++++++++++++++++++++++----------------- 1 file changed, 362 insertions(+), 158 deletions(-) diff --git a/po/id.po b/po/id.po index 3965c9e941d9c6..7d32c51ac77f04 100644 --- a/po/id.po +++ b/po/id.po @@ -7,8 +7,8 @@ msgid "" msgstr "" "Project-Id-Version: Git\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-23 18:57+0000\n" -"PO-Revision-Date: 2025-01-06 15:50+0700\n" +"POT-Creation-Date: 2025-03-05 22:57+0000\n" +"PO-Revision-Date: 2025-03-09 17:44+0700\n" "Last-Translator: Bagas Sanjaya \n" "Language-Team: Indonesian\n" "Language: id\n" @@ -2890,6 +2890,22 @@ msgstr "git archive: kesalahan protokol" msgid "git archive: expected a flush" msgstr "git archive: sebuah bilasan diharapkan" +#: builtin/backfill.c +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +#: builtin/backfill.c +msgid "problem loading sparse-checkout" +msgstr "galat memuat sparse-checkout" + +#: builtin/backfill.c +msgid "Minimum number of objects to request at a time" +msgstr "Jumlah objek minum yang diminta pada suatu waktu" + +#: builtin/backfill.c +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "Batasi objek yang hilang ke sparse-checkout saat ini" + #: builtin/bisect.c msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" @@ -3312,7 +3328,7 @@ msgstr "perlihatkan email pengarang daripada nama (asali: off)" msgid "ignore whitespace differences" msgstr "abaikan perbedaan spasi putih" -#: builtin/blame.c builtin/log.c +#: builtin/blame.c builtin/clone.c builtin/log.c msgid "rev" msgstr "revisi" @@ -3828,11 +3844,6 @@ msgstr "" msgid "git version:\n" msgstr "versi git:\n" -#: builtin/bugreport.c -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() gagal dengan kesalahan '%s' (%d)\n" - #: builtin/bugreport.c msgid "compiler info: " msgstr "info pengompilasi: " @@ -5090,8 +5101,112 @@ msgid "clean.requireForce is true and -f not given: refusing to clean" msgstr "clean.requireForce 'true' dan -f tidak diberikan: menolak membersihkan" #: builtin/clone.c -msgid "git clone [] [--] []" -msgstr "git clone [] [--] []" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "info: Tidak dapat menambahkan alternatif untuk '%s': %s\n" + +#: builtin/clone.c builtin/diff.c builtin/rm.c grep.c setup.c +#, c-format +msgid "failed to stat '%s'" +msgstr "gagal men-stat '%s'" + +#: builtin/clone.c +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s ada dan bukan direktori" + +#: builtin/clone.c +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "'%s' tautan simbolik, menolak mengkloning dengan --local" + +#: builtin/clone.c +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "gagal memulai iterator pada '%s'" + +#: builtin/clone.c +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "tautan simbolik '%s' ada, menolak mengkloning dengan --local" + +#: builtin/clone.c compat/precompose_utf8.c +#, c-format +msgid "failed to unlink '%s'" +msgstr "gagal menghapus tautan '%s'" + +#: builtin/clone.c +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "tautan keras tidak dapat diperiksa pada '%s'" + +#: builtin/clone.c +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "tautan keras berbeda dari sumber pada '%s'" + +#: builtin/clone.c +#, c-format +msgid "failed to create link '%s'" +msgstr "gagal membuat tautan '%s'" + +#: builtin/clone.c +#, c-format +msgid "failed to copy file to '%s'" +msgstr "gagal menyalin berkas ke '%s'" + +#: builtin/clone.c refs/files-backend.c +#, c-format +msgid "failed to iterate over '%s'" +msgstr "gagal iterasi pada '%s'" + +#: builtin/clone.c +#, c-format +msgid "done.\n" +msgstr "selesai.\n" + +#: builtin/clone.c +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Klon sukses, tapi checkout gagal.\n" +"Anda dapat periksa apa yang dicheckout dengan 'git status'\n" +"dan coba lagi dengan 'git restore --source=HEAD :/'\n" + +#: builtin/clone.c fetch-pack.c +msgid "remote did not send all necessary objects" +msgstr "remote tidak mengirim semua objek yang dibutuhkan" + +#: builtin/clone.c +#, c-format +msgid "unable to update %s" +msgstr "tidak dapat memperbarui %s" + +#: builtin/clone.c +msgid "failed to initialize sparse-checkout" +msgstr "gagal menginisalisasi checkout tipis" + +#: builtin/clone.c +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "HEAD remote merujuk pada ref yang tidak ada, tidak dapat men-checkout" + +#: builtin/clone.c +msgid "unable to checkout working tree" +msgstr "tidak dapat men-checkout pohon kerja" + +#: builtin/clone.c +msgid "unable to write parameters to config file" +msgstr "tidak dapat menulis parameter ke berkas konfigurasi" + +#: builtin/clone.c +msgid "cannot repack to clean up" +msgstr "tidak dapat memaket ulang untuk pembersihan" + +#: builtin/clone.c +msgid "cannot unlink temporary alternates file" +msgstr "tidak dapat batal-taut berkas alternatif sementara" #: builtin/clone.c msgid "don't clone shallow repository" @@ -5163,6 +5278,10 @@ msgstr "gunakan daripada 'origin' untuk lacak hulu" msgid "checkout instead of the remote's HEAD" msgstr "checkout daripada HEAD remote" +#: builtin/clone.c +msgid "clone single revision and check out" +msgstr "klon satu revisi dan check out" + #: builtin/clone.c msgid "path to git-upload-pack on the remote" msgstr "jalur ke git-upload-pack pada remote" @@ -5192,8 +5311,8 @@ msgid "clone only one branch, HEAD or --branch" msgstr "klon hanya satu cabang, HEAD atau --branch" #: builtin/clone.c -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "jangan klon tag apapun, dan buat pengambilan nanti tidak mengikutinya" +msgid "clone tags, and make later fetches not to follow them" +msgstr "klon tag dan jangan ikuti mereka pada pengambilan berikutnya" #: builtin/clone.c msgid "any cloned submodules will be shallow" @@ -5251,117 +5370,8 @@ msgid "a URI for downloading bundles before fetching from origin remote" msgstr "sebuah URI untuk mengunduh bundel sebelum mengambil dari remote asal" #: builtin/clone.c -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "info: Tidak dapat menambahkan alternatif untuk '%s': %s\n" - -#: builtin/clone.c builtin/diff.c builtin/rm.c grep.c setup.c -#, c-format -msgid "failed to stat '%s'" -msgstr "gagal men-stat '%s'" - -#: builtin/clone.c -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s ada dan bukan direktori" - -#: builtin/clone.c -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "'%s' tautan simbolik, menolak mengkloning dengan --local" - -#: builtin/clone.c -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "gagal memulai iterator pada '%s'" - -#: builtin/clone.c -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "tautan simbolik '%s' ada, menolak mengkloning dengan --local" - -#: builtin/clone.c compat/precompose_utf8.c -#, c-format -msgid "failed to unlink '%s'" -msgstr "gagal menghapus tautan '%s'" - -#: builtin/clone.c -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "tautan keras tidak dapat diperiksa pada '%s'" - -#: builtin/clone.c -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "tautan keras berbeda dari sumber pada '%s'" - -#: builtin/clone.c -#, c-format -msgid "failed to create link '%s'" -msgstr "gagal membuat tautan '%s'" - -#: builtin/clone.c -#, c-format -msgid "failed to copy file to '%s'" -msgstr "gagal menyalin berkas ke '%s'" - -#: builtin/clone.c refs/files-backend.c -#, c-format -msgid "failed to iterate over '%s'" -msgstr "gagal iterasi pada '%s'" - -#: builtin/clone.c -#, c-format -msgid "done.\n" -msgstr "selesai.\n" - -#: builtin/clone.c -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Klon sukses, tapi checkout gagal.\n" -"Anda dapat periksa apa yang dicheckout dengan 'git status'\n" -"dan coba lagi dengan 'git restore --source=HEAD :/'\n" - -#: builtin/clone.c -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "Tidak dapat menemukan cabang remote %s untuk diklon." - -#: builtin/clone.c fetch-pack.c -msgid "remote did not send all necessary objects" -msgstr "remote tidak mengirim semua objek yang dibutuhkan" - -#: builtin/clone.c -#, c-format -msgid "unable to update %s" -msgstr "tidak dapat memperbarui %s" - -#: builtin/clone.c -msgid "failed to initialize sparse-checkout" -msgstr "gagal menginisalisasi checkout tipis" - -#: builtin/clone.c -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "HEAD remote merujuk pada ref yang tidak ada, tidak dapat men-checkout" - -#: builtin/clone.c -msgid "unable to checkout working tree" -msgstr "tidak dapat men-checkout pohon kerja" - -#: builtin/clone.c -msgid "unable to write parameters to config file" -msgstr "tidak dapat menulis parameter ke berkas konfigurasi" - -#: builtin/clone.c -msgid "cannot repack to clean up" -msgstr "tidak dapat memaket ulang untuk pembersihan" - -#: builtin/clone.c -msgid "cannot unlink temporary alternates file" -msgstr "tidak dapat batal-taut berkas alternatif sementara" +msgid "git clone [] [--] []" +msgstr "git clone [] [--] []" #: builtin/clone.c msgid "Too many arguments." @@ -5491,6 +5501,11 @@ msgstr "transportasi remote melaporkan kesalahan" msgid "Remote branch %s not found in upstream %s" msgstr "Cabang remote %s tidak ditemukan di hulu %s" +#: builtin/clone.c +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "Revisi remote %s tidak ditemukan di hulu %s" + #: builtin/clone.c msgid "You appear to have cloned an empty repository." msgstr "Anda tampaknya mengklon repositori kosong." @@ -5555,7 +5570,8 @@ msgstr "" "[no-]progress]\n" " " -#: builtin/commit-graph.c builtin/fetch.c builtin/log.c builtin/repack.c +#: builtin/commit-graph.c builtin/fetch.c builtin/gc.c builtin/log.c +#: builtin/repack.c msgid "dir" msgstr "direktori" @@ -5715,7 +5731,7 @@ msgstr "git commit-tree: gagal membaca" #: builtin/commit.c msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -5725,7 +5741,7 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -7472,14 +7488,14 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Jalankan 'git remote set-head %s %s' untuk mengikuti perubahan, atau setel\n" "opsi konfigurasi 'remote.%s.followRemoteHEAD' ke nilai yang berbeda jika\n" -"Anda tidak ingin melihat pesan ini lagi. Secara rinci menjalakan\n" -"'git config set remote.%s followRemoteHEAD %s' akan mematikan peringatan\n" -"ini sampai remote mengubah HEAD ke yang lain." +"Anda tidak ingin melihat pesan ini lagi. Secara rinci menjalankan\n" +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s' akan\n" +"mematikan peringatan ini sampai remote mengubah HEAD ke yang lain." #: builtin/fetch.c msgid "multiple branches detected, incompatible with --set-upstream" @@ -8341,6 +8357,10 @@ msgstr "paksa jalankan gc bahkan jika mungkin ada gc lain yang berjalan" msgid "repack all other packs except the largest pack" msgstr "pak ulang semua pak yang lain kecuali pak terbesar" +#: builtin/gc.c builtin/repack.c +msgid "pack prefix to store a pack containing pruned objects" +msgstr "awalan pak untuk menyimpan pak berisi objek terpangkas" + #: builtin/gc.c #, c-format msgid "failed to parse gc.logExpiry value %s" @@ -9353,6 +9373,11 @@ msgstr "tidak dapat menyelesaikan pack-objects untuk mempak ulang tautan lokal" msgid "Cannot come back to cwd" msgstr "tidak dapat kembali ke direktori kerja saat ini" +#: builtin/index-pack.c builtin/unpack-objects.c +#, c-format +msgid "bad --pack_header: %s" +msgstr "--pack_header jelek: %s" + #: builtin/index-pack.c #, c-format msgid "bad %s" @@ -10476,11 +10501,6 @@ msgstr "opsi strategi tidak dikenal: -X%s" msgid "malformed input line: '%s'." msgstr "baris masukan jelek: '%s'." -#: builtin/merge-tree.c -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "penggabungan tidak dapat berlanjut; dapat hasil kotor dari %d" - #: builtin/merge.c msgid "git merge [] [...]" msgstr "git merge [] [...]" @@ -11522,6 +11542,15 @@ msgstr "" "git pack-objects [...] [< | < ]" +#: builtin/pack-objects.c +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "opsi --name-hash-version tidak valid: %d" + +#: builtin/pack-objects.c +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "saat ini, --write-bitmap-index memerlukan --name-hash-version=1" + #: builtin/pack-objects.c #, c-format msgid "" @@ -11934,6 +11963,12 @@ msgstr "" "abaikan uploadpack.blobpackfileuri apapun yang dikonfigurasikan dengan " "protokol ini" +#: builtin/pack-objects.c +msgid "use the specified name-hash function to group similar objects" +msgstr "" +"gunakan fungsi nama-hash yang dirincikan untuk mengelompokan objek yang " +"serupa" + #: builtin/pack-objects.c #, c-format msgid "delta chain depth %d is too deep, forcing %d" @@ -13428,8 +13463,8 @@ msgid "invalid ref format: %s" msgstr "format referensi tidak valid: %s" #: builtin/refs.c -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format= [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format= [--no-reflog] [--dry-run]" #: builtin/refs.c msgid "git refs verify [--strict] [--verbose]" @@ -13443,6 +13478,10 @@ msgstr "sebutkan format referensi untuk dikonversi" msgid "perform a non-destructive dry-run" msgstr "lakukan uji coba non desktruktif" +#: builtin/refs.c +msgid "drop reflogs entirely during the migration" +msgstr "buang keseluruhan log referensi selama migrasi" + #: builtin/refs.c msgid "missing --ref-format=" msgstr "--ref-format= hilang" @@ -14024,8 +14063,14 @@ msgid "be verbose; must be placed before a subcommand" msgstr "jadi lebih bertele-tele; harus ditempatkan sebelum subperintah" #: builtin/repack.c -msgid "git repack []" -msgstr "git repack []" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" #: builtin/repack.c msgid "" @@ -14118,6 +14163,13 @@ msgstr "lewatkan --no-reuse-delta ke git-pack-objects" msgid "pass --no-reuse-object to git-pack-objects" msgstr "lewatkan --no-reuse-object ke git-pack-objects" +#: builtin/repack.c +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "" +"tentukan versi nama hash yang digunakan untuk mengelompokan objek yang " +"serupa berdasarkan jalur" + #: builtin/repack.c msgid "do not run git-update-server-info" msgstr "jangan jalankan git-update-server-info" @@ -14184,10 +14236,6 @@ msgstr "temukan deret geometri dengan faktor " msgid "write a multi-pack index of the resulting packs" msgstr "tulis indeks multipak dari pak yang dihasilkan" -#: builtin/repack.c -msgid "pack prefix to store a pack containing pruned objects" -msgstr "awalan pak untuk menyimpan pak berisi objek terpangkas" - #: builtin/repack.c msgid "pack prefix to store a pack containing filtered out objects" msgstr "awalan pak untuk menyimpan pak berisi objek tersaring" @@ -14462,10 +14510,6 @@ msgstr "hanya satu pola yang dapat diberikan dengan -l" msgid "need some commits to replay" msgstr "butuh beberapa komit untuk dimainkan ulang" -#: builtin/replay.c -msgid "--onto and --advance are incompatible" -msgstr "--onto dan --advance tidak kompatibel" - #: builtin/replay.c msgid "all positive revisions given must be references" msgstr "semua revisi positif yang diberikan haruslah referensi" @@ -17767,6 +17811,10 @@ msgstr "Impor repositori GNU Arch ke dalam Git" msgid "Create an archive of files from a named tree" msgstr "Buat arsip berkas dari pohon bernama" +#: command-list.h +msgid "Download missing objects in a partial clone" +msgstr "Unduh objek yang hilang dalam klon parsial" + #: command-list.h msgid "Use binary search to find the commit that introduced a bug" msgstr "Gunakan pencarian biner untuk mencari komit yang memasukkan bug" @@ -20190,6 +20238,14 @@ msgstr "argumen tidak valid ke %s" msgid "invalid regex given to -I: '%s'" msgstr "regex tidak valid diberikan ke -I: '%s'" +#: diff.c +msgid "-G requires a non-empty argument" +msgstr "-G butuh sebuah argumen bukan kosong" + +#: diff.c +msgid "-S requires a non-empty argument" +msgstr "-S butuh sebuah argumen bukan kosong" + #: diff.c #, c-format msgid "failed to parse --submodule option parameter: '%s'" @@ -22886,6 +22942,11 @@ msgstr "berkas '%s' dan '%s' berbeda konteks" msgid "unable to write file %s" msgstr "tidak dapat menulis berkas %s" +#: object-file.c +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "tidak dapat menulis berkas yang menghilang %s terus-menerus" + #: object-file.c #, c-format msgid "unable to set permission to '%s'" @@ -23579,6 +23640,55 @@ msgstr "sakelar tidak dikenal `%c'" msgid "unknown non-ascii option in string: `%s'" msgstr "opsi non-ascii di dalam untai tidak dikenal: `%s'" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid " <%s>" +msgstr " <%s>" + #: parse-options.c msgid "..." msgstr "..." @@ -23682,6 +23792,25 @@ msgstr "nilai lingkungan boolean '%s' jelek untuk '%s'" msgid "failed to parse %s" msgstr "gagal menguraikan %s" +#: path-walk.c +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "gagal berjalan anak pohon %s: tidak ditemukan" + +#: path-walk.c +#, c-format +msgid "failed to find object %s" +msgstr "gagal menemukan objek %s" + +#: path-walk.c +#, c-format +msgid "failed to find tag %s" +msgstr "gagal menemukan tag %s" + +#: path-walk.c +msgid "failed to setup revision walk" +msgstr "gagal men-setup jalan revisi" + #: path.c #, c-format msgid "Could not make %s writable by group" @@ -23868,6 +23997,26 @@ msgstr "nama remote penjanji tidak dapat diawali dengan '/': %s" msgid "could not fetch %s from promisor remote" msgstr "tidak dapat mengambil %s dari remote penjanji" +#: promisor-remote.c +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "remote yang terkenal bernama '%s' tapi dengan url '%s' daripada '%s'" + +#: promisor-remote.c +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "nilai '%s' tidak dikenal untuk opsi konfigurasi '%s'" + +#: promisor-remote.c +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "elemen '%s' dari info remote tidak dikenal" + +#: promisor-remote.c +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "remote penjanji yang diterima '%s' tidak ditemukan" + #: protocol-caps.c msgid "object-info: expected flush after arguments" msgstr "object-info: bilasan diharapkan setelah argumen" @@ -24870,6 +25019,16 @@ msgstr "nama referensi %s simbolik, menyalinnya tidak didukung" msgid "invalid refspec '%s'" msgstr "spek referensi tidak valid '%s'" +#: refspec.c +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "pola '%s' tidak mempunyai '*'" + +#: refspec.c +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "pengganti '%s' tidak mempunyai '*'" + #: remote-curl.c #, c-format msgid "invalid quoting in push-option value: '%s'" @@ -25023,6 +25182,28 @@ msgstr "remote-curl: pengambilan dicoba tanpa repositori lokal" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl: perintah tidak dikenal '%s' dari git" +#: remote.c +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"membaca remote dari \"%s/%s\", yang dinominasikan untuk dihapus.\n" +"\n" +"Jika Anda masih menggunakan direktori \"remotes/\" disarankan untuk\n" +"migrasi ke remote berdasarkan konfigurasi:\n" +"\n" +"\tgit remote rename %s %s\n" +"Jika tidak, mohon beri tahu kami mengapa Anda masih menggunakannya dengan\n" +"mengirimkan surel ke ." + #: remote.c #, c-format msgid "config remote shorthand cannot begin with '/': %s" @@ -25066,16 +25247,6 @@ msgstr "%s biasanya melacak %s, bukan %s" msgid "%s tracks both %s and %s" msgstr "%s melacak baik %s dan %s" -#: remote.c -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "kunci '%s' dari pola tidak ada '*'" - -#: remote.c -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "nilai '%s' dari pola tidak ada '*'" - #: remote.c #, c-format msgid "src refspec %s does not match any" @@ -27389,6 +27560,34 @@ msgstr "bersihkan pohon tembolok sebelum setiap iterasi" msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "jumlah entri di dalam pohon tembolok untuk dinirvalidasi (asali 0)" +#: t/helper/test-path-walk.c +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk -- " + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of blob objects" +msgstr "sertakan objek blob" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of commit objects" +msgstr "sertakan objek komit" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of tag objects" +msgstr "sertakan objek tag" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of tree objects" +msgstr "sertakan objek pohon" + +#: t/helper/test-path-walk.c +msgid "toggle pruning of uninteresting paths" +msgstr "pangkas jalur yang tidak menarik" + +#: t/helper/test-path-walk.c +msgid "read a pattern list over stdin" +msgstr "baca daftar pola dari masukan standar" + #: t/helper/test-reach.c #, c-format msgid "commit %s is not marked reachable" @@ -28148,6 +28347,11 @@ msgstr "kesalahan: " msgid "warning: " msgstr "peringatan: " +#: version.c +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() gagal dengan kesalahan '%s' (%d)\n" + #: walker.c msgid "Fetching objects" msgstr "Mengambil objek" From 9db5ab6f6ca8e675db8f3f517b3b27a95a02b1fd Mon Sep 17 00:00:00 2001 From: Ralf Thielow Date: Thu, 13 Mar 2025 07:03:42 +0100 Subject: [PATCH 018/259] l10n: update German translation Signed-off-by: Ralf Thielow --- po/de.po | 447 ++++++++++++++++++++++++++++++++++++++----------------- 1 file changed, 309 insertions(+), 138 deletions(-) diff --git a/po/de.po b/po/de.po index addd5919bdb015..4d9c46fb19de01 100644 --- a/po/de.po +++ b/po/de.po @@ -8,8 +8,8 @@ msgid "" msgstr "" "Project-Id-Version: Git\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-20 17:44+0100\n" -"PO-Revision-Date: 2024-12-27 16:43+0100\n" +"POT-Creation-Date: 2025-03-06 18:29+0100\n" +"PO-Revision-Date: 2025-03-07 17:28+0100\n" "Last-Translator: Ralf Thielow \n" "Language-Team: German\n" "Language: de\n" @@ -2400,6 +2400,18 @@ msgstr "git archive: Protokollfehler" msgid "git archive: expected a flush" msgstr "git archive: erwartete eine Spülung (flush)" +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +msgid "problem loading sparse-checkout" +msgstr "Problem beim Laden von sparse-checkout" + +msgid "Minimum number of objects to request at a time" +msgstr "Mindestanzahl der gleichzeitig anzufordernden Objekte" + +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "fehlenden Objekte im aktuellen sparse-checkout beschränken" + msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" "checkout] [--first-parent] [ [...]] [--] [...]" @@ -3167,10 +3179,6 @@ msgstr "" msgid "git version:\n" msgstr "git Version:\n" -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() ist fehlgeschlagen mit Fehler '%s' (%d)\n" - msgid "compiler info: " msgstr "Compiler Info: " @@ -4201,8 +4209,95 @@ msgstr "" "clean.requireForce ist 'true' und -f ist nicht angegeben: clean wird " "verweigert" -msgid "git clone [] [--] []" -msgstr "git clone [] [--] []" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "info: Konnte Alternative für '%s' nicht hinzufügen: %s\n" + +#, c-format +msgid "failed to stat '%s'" +msgstr "Konnte '%s' nicht lesen" + +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s existiert und ist kein Verzeichnis" + +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "" +"'%s' ist eine symbolische Verknüpfung, verweigere das Klonen mit --local" + +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "Fehler beim Starten der Iteration über '%s'" + +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "" +"symbolische Verknüpfung '%s' existiert, verweigere das Klonen mit --local" + +#, c-format +msgid "failed to unlink '%s'" +msgstr "Konnte '%s' nicht entfernen." + +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "Hardlink bei '%s' kann nicht geprüft werden" + +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "Hardlink unterscheidet sich von der Quelle bei '%s'" + +#, c-format +msgid "failed to create link '%s'" +msgstr "Konnte Verweis '%s' nicht erstellen" + +#, c-format +msgid "failed to copy file to '%s'" +msgstr "Konnte Datei nicht nach '%s' kopieren" + +#, c-format +msgid "failed to iterate over '%s'" +msgstr "Fehler beim Iterieren über '%s'" + +#, c-format +msgid "done.\n" +msgstr "Fertig.\n" + +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"Klonen erfolgreich, Auschecken ist aber fehlgeschlagen.\n" +"Sie können mit 'git status' prüfen, was ausgecheckt worden ist\n" +"und das Auschecken mit 'git restore --source=HEAD :/' erneut versuchen.\n" + +msgid "remote did not send all necessary objects" +msgstr "Remote-Repository hat nicht alle erforderlichen Objekte gesendet" + +#, c-format +msgid "unable to update %s" +msgstr "kann %s nicht aktualisieren" + +msgid "failed to initialize sparse-checkout" +msgstr "Fehler beim Initialisieren vom partiellen Checkout." + +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "" +"HEAD des Remote-Repositories verweist auf nicht existierende Referenz, kann " +"nicht ausgecheckt werden" + +msgid "unable to checkout working tree" +msgstr "Arbeitsverzeichnis konnte nicht ausgecheckt werden" + +msgid "unable to write parameters to config file" +msgstr "konnte Parameter nicht in Konfigurationsdatei schreiben" + +msgid "cannot repack to clean up" +msgstr "Kann \"repack\" zum Aufräumen nicht aufrufen" + +msgid "cannot unlink temporary alternates file" +msgstr "Kann temporäre \"alternates\"-Datei nicht entfernen" msgid "don't clone shallow repository" msgstr "Repository mit unvollständiger Historie nicht klonen" @@ -4255,6 +4350,9 @@ msgstr " statt 'origin' für Upstream-Repository verwenden" msgid "checkout instead of the remote's HEAD" msgstr " auschecken, anstatt HEAD des Remote-Repositories" +msgid "clone single revision and check out" +msgstr "einzelnen Commit klonen und auschecken" + msgid "path to git-upload-pack on the remote" msgstr "Pfad zu \"git-upload-pack\" auf der Gegenseite" @@ -4280,8 +4378,8 @@ msgstr "Historie eines flachen Klons vertiefen, Referenz exkludiert" msgid "clone only one branch, HEAD or --branch" msgstr "nur einen Branch klonen, HEAD oder --branch" -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "keine Tags klonen, und auch bei späteren Abrufen nicht beachten" +msgid "clone tags, and make later fetches not to follow them" +msgstr "Tags klonen und dafür sorgen, dass spätere Abrufe ihnen nicht folgen" msgid "any cloned submodules will be shallow" msgstr "jedes geklonte Submodul mit unvollständiger Historie (shallow)" @@ -4326,99 +4424,8 @@ msgstr "" "eine URI für das Herunterladen von Bundles vor dem Abruf\n" "aus dem ursprünglichen Remote-Repository" -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "info: Konnte Alternative für '%s' nicht hinzufügen: %s\n" - -#, c-format -msgid "failed to stat '%s'" -msgstr "Konnte '%s' nicht lesen" - -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s existiert und ist kein Verzeichnis" - -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "" -"'%s' ist eine symbolische Verknüpfung, verweigere das Klonen mit --local" - -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "Fehler beim Starten der Iteration über '%s'" - -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "" -"symbolische Verknüpfung '%s' existiert, verweigere das Klonen mit --local" - -#, c-format -msgid "failed to unlink '%s'" -msgstr "Konnte '%s' nicht entfernen." - -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "Hardlink bei '%s' kann nicht geprüft werden" - -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "Hardlink unterscheidet sich von der Quelle bei '%s'" - -#, c-format -msgid "failed to create link '%s'" -msgstr "Konnte Verweis '%s' nicht erstellen" - -#, c-format -msgid "failed to copy file to '%s'" -msgstr "Konnte Datei nicht nach '%s' kopieren" - -#, c-format -msgid "failed to iterate over '%s'" -msgstr "Fehler beim Iterieren über '%s'" - -#, c-format -msgid "done.\n" -msgstr "Fertig.\n" - -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"Klonen erfolgreich, Auschecken ist aber fehlgeschlagen.\n" -"Sie können mit 'git status' prüfen, was ausgecheckt worden ist\n" -"und das Auschecken mit 'git restore --source=HEAD :/' erneut versuchen.\n" - -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "Konnte zu klonenden Remote-Branch %s nicht finden." - -msgid "remote did not send all necessary objects" -msgstr "Remote-Repository hat nicht alle erforderlichen Objekte gesendet" - -#, c-format -msgid "unable to update %s" -msgstr "kann %s nicht aktualisieren" - -msgid "failed to initialize sparse-checkout" -msgstr "Fehler beim Initialisieren vom partiellen Checkout." - -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "" -"HEAD des Remote-Repositories verweist auf nicht existierende Referenz, kann " -"nicht ausgecheckt werden" - -msgid "unable to checkout working tree" -msgstr "Arbeitsverzeichnis konnte nicht ausgecheckt werden" - -msgid "unable to write parameters to config file" -msgstr "konnte Parameter nicht in Konfigurationsdatei schreiben" - -msgid "cannot repack to clean up" -msgstr "Kann \"repack\" zum Aufräumen nicht aufrufen" - -msgid "cannot unlink temporary alternates file" -msgstr "Kann temporäre \"alternates\"-Datei nicht entfernen" +msgid "git clone [] [--] []" +msgstr "git clone [] [--] []" msgid "Too many arguments." msgstr "Zu viele Argumente." @@ -4531,6 +4538,10 @@ msgstr "Remoteübertragung meldete Fehler" msgid "Remote branch %s not found in upstream %s" msgstr "Remote-Branch %s nicht im Upstream-Repository %s gefunden" +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "Remote-Commit %s nicht im Upstream-Repository %s gefunden" + msgid "You appear to have cloned an empty repository." msgstr "Sie scheinen ein leeres Repository geklont zu haben." @@ -4710,7 +4721,7 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree: Fehler beim Lesen" msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -4720,14 +4731,14 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" " [--allow-empty-message] [--no-verify] [-e] [--author=]\n" " [--date=] [--cleanup=] [--[no-]status]\n" " [-i | -o] [--pathspec-from-file= [--pathspec-file-nul]]\n" -" [(--trailer [(=|:)])...] [-S[]]\n" +" [(--trailer [(=|:)])...] [-S[>]]\n" " [--] [...]" msgid "git status [] [--] [...]" @@ -6139,14 +6150,15 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" "Führen Sie 'git remote set-head %s %s' aus, um der Änderung zu folgen,\n" "oder setzen Sie die Konfiguration 'remote.%s.followRemoteHEAD' auf einen\n" -"anderen Wert, wenn Sie diese Meldung nicht sehen wollen. Konkret wird diese\n" -"Warnung mit 'git config set remote.%s.followRemoteHEAD %s' deaktiviert\n" -"bis die Gegenstelle HEAD in etwas anderes ändert." +"anderen Wert wenn Sie diese Meldung nicht sehen wollen. Speziell der Befehl\n" +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"wird die Warnung deaktivieren, bis das Remote-Repository HEAD in etwas\n" +"anderes ändert." msgid "multiple branches detected, incompatible with --set-upstream" msgstr "mehrere Branches erkannt, inkompatibel mit --set-upstream" @@ -6842,6 +6854,9 @@ msgstr "" msgid "repack all other packs except the largest pack" msgstr "alle anderen Pakete, außer das größte Paket, neu packen" +msgid "pack prefix to store a pack containing pruned objects" +msgstr "pack-Präfix zum Speichern eines Pakets mit gelöschten Objekten" + #, c-format msgid "failed to parse gc.logExpiry value %s" msgstr "Fehler beim Parsen des Wertes '%s' von gc.logExpiry" @@ -7658,6 +7673,10 @@ msgstr "" msgid "Cannot come back to cwd" msgstr "Kann nicht zurück zum Arbeitsverzeichnis wechseln" +#, c-format +msgid "bad --pack_header: %s" +msgstr "ungültiger --pack_header: %s" + #, c-format msgid "bad %s" msgstr "%s ist ungültig" @@ -8537,11 +8556,6 @@ msgstr "unbekannte Strategie-Option: -X%s" msgid "malformed input line: '%s'." msgstr "Fehlerhafte Eingabezeile: '%s'." -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "" -"Merge kann nicht fortgesetzt werden; unsauberes Ergebnis von %d erhalten" - msgid "git merge [] [...]" msgstr "git merge [] [...]" @@ -9381,6 +9395,13 @@ msgstr "" "git pack-objects [] [< | < " "]" +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "ungültige Option für --name-hash-version: %d" + +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "derzeit erfordert --write-bitmap-index --name-hash-version=1" + #, c-format msgid "" "write_reuse_object: could not locate %s, expected at offset % in " @@ -9718,6 +9739,11 @@ msgstr "" "jegliche konfigurierte uploadpack.blobpackfileuri für dieses Protkoll " "ausschließen" +msgid "use the specified name-hash function to group similar objects" +msgstr "" +"die angegebene Name-Hash-Funktion verwenden, um ähnliche Objekte zu " +"gruppieren" + #, c-format msgid "delta chain depth %d is too deep, forcing %d" msgstr "Tiefe für Verkettung von Unterschieden %d ist zu tief, erzwinge %d" @@ -11006,8 +11032,8 @@ msgstr "Kein Reflog zum Löschen angegeben." msgid "invalid ref format: %s" msgstr "Ungültiges Format für Referenzen: %s" -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format= [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format= [--no-reflog] [--dry-run]" msgid "git refs verify [--strict] [--verbose]" msgstr "git refs verify [--strict] [--verbose]" @@ -11018,6 +11044,9 @@ msgstr "das Referenzformat angeben, in das konvertiert werden soll" msgid "perform a non-destructive dry-run" msgstr "einen zerstörungsfreien Trockenlauf durchführen" +msgid "drop reflogs entirely during the migration" +msgstr "Reflogs während der Migration vollständig zu löschen" + msgid "missing --ref-format=" msgstr "fehlendes --ref-format=" @@ -11488,8 +11517,14 @@ msgstr "Werde keine URLs entfernen, die nicht für \"push\" bestimmt sind" msgid "be verbose; must be placed before a subcommand" msgstr "erweiterte Ausgaben; muss vor einem Unterbefehl angegeben werden" -msgid "git repack []" -msgstr "git repack []" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" msgid "" "Incremental repacks are incompatible with bitmap indexes. Use\n" @@ -11568,6 +11603,12 @@ msgstr "--no-reuse-delta an git-pack-objects übergeben" msgid "pass --no-reuse-object to git-pack-objects" msgstr "--no-reuse-object an git-pack-objects übergeben" +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "" +"die Version des Hash-Namens angeben, die für die Gruppierung ähnlicher " +"Objekte nach Pfad verwendet werden soll" + msgid "do not run git-update-server-info" msgstr "git-update-server-info nicht ausführen" @@ -11619,9 +11660,6 @@ msgstr "eine geometrische Progression mit Faktor finden" msgid "write a multi-pack index of the resulting packs" msgstr "ein Multi-Pack-Index des resultierenden Pakets schreiben" -msgid "pack prefix to store a pack containing pruned objects" -msgstr "pack-Präfix zum Speichern eines Pakets mit gelöschten Objekten" - msgid "pack prefix to store a pack containing filtered out objects" msgstr "Paketpräfix, um ein Paket mit herausgefilterten Objekten zu speichern" @@ -11839,9 +11877,6 @@ msgstr "Mit -l kann nur ein Muster angegeben werden" msgid "need some commits to replay" msgstr "zum erneuten Abspielen werden Commits benötigt" -msgid "--onto and --advance are incompatible" -msgstr "--onto und --advance sind inkompatibel" - msgid "all positive revisions given must be references" msgstr "alle angegebenen positiven Commits müssen Referenzen sein" @@ -14571,6 +14606,9 @@ msgstr "ein GNU Arch Repository in Git importieren" msgid "Create an archive of files from a named tree" msgstr "Dateiarchiv von angegebenem Verzeichnis erstellen" +msgid "Download missing objects in a partial clone" +msgstr "fehlender Objekte in einem partiellen Klon herunterladen" + msgid "Use binary search to find the commit that introduced a bug" msgstr "" "Binärsuche verwenden, um den Commit zu finden, der einen Fehler verursacht " @@ -16543,6 +16581,12 @@ msgstr "ungültiges Argument für %s" msgid "invalid regex given to -I: '%s'" msgstr "ungültiger regulärer Ausdruck für -I gegeben: '%s'" +msgid "-G requires a non-empty argument" +msgstr "-G erfordert ein nicht leeres Argument" + +msgid "-S requires a non-empty argument" +msgstr "-S erfordert ein nicht leeres Argument" + #, c-format msgid "failed to parse --submodule option parameter: '%s'" msgstr "Fehler beim Parsen des --submodule Optionsparameters: '%s'" @@ -18781,6 +18825,10 @@ msgstr "die Dateien '%s' und '%s' unterscheiden sich im Inhalt" msgid "unable to write file %s" msgstr "Konnte Datei %s nicht schreiben." +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "konnte nicht in eine wiederholt verschwindende Datei %s schreiben" + #, c-format msgid "unable to set permission to '%s'" msgstr "Konnte Zugriffsberechtigung auf '%s' nicht setzen." @@ -19355,6 +19403,52 @@ msgstr "Unbekannter Schalter `%c'" msgid "unknown non-ascii option in string: `%s'" msgstr "Unbekannte nicht-Ascii Option in String: `%s'" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#, c-format +msgid " <%s>" +msgstr " <%s>" + msgid "..." msgstr "..." @@ -19441,6 +19535,21 @@ msgstr "falscher boolescher Wert von Umgebungsvariable '%s' für '%s'" msgid "failed to parse %s" msgstr "Fehler beim Parsen von %s." +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "Fehlen beim Durchlaufen der Kinder von Baum %s: nicht gefunden" + +#, c-format +msgid "failed to find object %s" +msgstr "Objekt nicht gefunden: %s" + +#, c-format +msgid "failed to find tag %s" +msgstr "Tag nicht gefunden: %s" + +msgid "failed to setup revision walk" +msgstr "Einrichtung des Revisionsgangs fehlgeschlagen" + #, c-format msgid "Could not make %s writable by group" msgstr "Konnte Gruppenschreibrecht für %s nicht setzen." @@ -19591,6 +19700,25 @@ msgstr "Promisor-Remote-Name kann nicht mit '/' beginnen: %s" msgid "could not fetch %s from promisor remote" msgstr "konnte %s nicht von Promisor-Remote abrufen" +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "" +"bekanntes Remote-Repository mit dem Namen '%s', aber mit der URL '%s' statt " +"'%s'" + +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "unbekannter '%s'-Wert für '%s'-Konfigurationsoption" + +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "unbekanntes Element '%s' aus Information aus Remote-Repository" + +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "" +"akzeptiertes partiell geklontes \"Remote-Repository '%s' nicht gefunden" + msgid "object-info: expected flush after arguments" msgstr "object-info: erwartete Flush nach Argumenten" @@ -20427,6 +20555,14 @@ msgstr "" msgid "invalid refspec '%s'" msgstr "ungültige Refspec '%s'" +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "Muster '%s' hat keinen '*'" + +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "Ersetzung '%s' hat keinen '*'" + #, c-format msgid "invalid quoting in push-option value: '%s'" msgstr "Ungültiges Quoting beim \"push-option\"-Wert: '%s'" @@ -20549,6 +20685,28 @@ msgstr "remote-curl: \"fetch\" ohne lokales Repository versucht" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl: Unbekannter Befehl '%s' von Git" +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"Lese von Remote-Repository \"%s/%s\", das zum Entfernen vorgeschlagen wird.\n" +"\n" +"Wenn Sie noch das Verzeichnis \"remotes/\" verwenden, wird empfohlen\n" +"auf konfigurationsbasierte Remote-Repositories umzusteigen:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"Wenn dies nicht möglich ist, teilen Sie uns bitte mit, warum Sie das noch\n" +"verwenden müssen, indem Sie eine E-Mail an senden." + #, c-format msgid "config remote shorthand cannot begin with '/': %s" msgstr "" @@ -20585,14 +20743,6 @@ msgstr "%s folgt üblicherweise %s, nicht %s" msgid "%s tracks both %s and %s" msgstr "%s folgt sowohl %s als auch %s" -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "Schlüssel '%s' des Musters hatte kein '*'." - -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "Wert '%s' des Musters hat kein '*'." - #, c-format msgid "src refspec %s does not match any" msgstr "Src-Refspec %s entspricht keiner Referenz." @@ -22519,6 +22669,27 @@ msgstr "" "Anzahl der Einträge im Cache-Verzeichnis, die ungültig gemacht werden sollen " "(Standardwert 0)" +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk -- " + +msgid "toggle inclusion of blob objects" +msgstr "Einbeziehung von Blob-Objekten umschalten" + +msgid "toggle inclusion of commit objects" +msgstr "Einbeziehung von Commit-Objekten umschalten" + +msgid "toggle inclusion of tag objects" +msgstr "Einbeziehung von Tag-Objekten umschalten" + +msgid "toggle inclusion of tree objects" +msgstr "Einbeziehung von Tree-Objekten umschalten" + +msgid "toggle pruning of uninteresting paths" +msgstr "Auslassen von uninteressanten Pfaden umschalten" + +msgid "read a pattern list over stdin" +msgstr "Liste von Mustern von der Standard-Eingabe lesen" + #, c-format msgid "commit %s is not marked reachable" msgstr "Commit %s ist nicht als erreichbar gekennzeichnet." @@ -23168,6 +23339,10 @@ msgstr "Fehler: " msgid "warning: " msgstr "Warnung: " +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() ist fehlgeschlagen mit Fehler '%s' (%d)\n" + msgid "Fetching objects" msgstr "Anfordern der Objekte" @@ -24190,7 +24365,3 @@ msgstr "Lasse %s mit Backup-Suffix '%s' aus.\n" #, perl-format msgid "Do you really want to send %s? [y|N]: " msgstr "Wollen Sie %s wirklich versenden? [y|N]: " - -#, c-format -#~ msgid "preferred pack (%s) is invalid" -#~ msgstr "bevorzugtes Paket (%s) ist ungültig" From 7bc205bec2ead74030a9290aade4df0e3337b9b6 Mon Sep 17 00:00:00 2001 From: Yi-Jyun Pan Date: Sun, 9 Mar 2025 10:54:02 +0800 Subject: [PATCH 019/259] l10n: zh_TW: Git 2.49.0 round 1 Co-authored-by: Lumynous Signed-off-by: Yi-Jyun Pan --- po/zh_TW.po | 574 ++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 402 insertions(+), 172 deletions(-) diff --git a/po/zh_TW.po b/po/zh_TW.po index a61f544304d34d..aa74d6537accf4 100644 --- a/po/zh_TW.po +++ b/po/zh_TW.po @@ -30,8 +30,8 @@ msgid "" msgstr "" "Project-Id-Version: Git\n" "Report-Msgid-Bugs-To: Git Mailing List \n" -"POT-Creation-Date: 2024-12-28 13:16+0800\n" -"PO-Revision-Date: 2024-12-28 13:23+0800\n" +"POT-Creation-Date: 2025-03-09 10:39+0800\n" +"PO-Revision-Date: 2025-03-09 10:52+0800\n" "Last-Translator: Yi-Jyun Pan \n" "Language-Team: Chinese (Traditional) \n" @@ -2841,6 +2841,22 @@ msgstr "git archive：通訊協定錯誤" msgid "git archive: expected a flush" msgstr "git archive：預期收到 flush 封包" +#: builtin/backfill.c +msgid "git backfill [--min-batch-size=] [--[no-]sparse]" +msgstr "git backfill [--min-batch-size=] [--[no-]sparse]" + +#: builtin/backfill.c +msgid "problem loading sparse-checkout" +msgstr "載入稀疏簽出時發生問題" + +#: builtin/backfill.c +msgid "Minimum number of objects to request at a time" +msgstr "一次請求的最小物件數量" + +#: builtin/backfill.c +msgid "Restrict the missing objects to the current sparse-checkout" +msgstr "將缺少的物件限制於目前的稀疏簽出" + #: builtin/bisect.c msgid "" "git bisect start [--term-(new|bad)= --term-(old|good)=] [--no-" @@ -2872,22 +2888,22 @@ msgstr "git bisect run [...]" #: builtin/bisect.c #, c-format msgid "cannot open file '%s' in mode '%s'" -msgstr "無法以「%2$s」模式開啟「%1$s」檔案" +msgstr "無法以「%2$s」模式開啟檔案「%1$s」" #: builtin/bisect.c #, c-format msgid "could not write to file '%s'" -msgstr "無法寫入「%s」檔案" +msgstr "無法寫入檔案「%s」" #: builtin/bisect.c #, c-format msgid "cannot open file '%s' for reading" -msgstr "無法開啟「%s」檔案進行讀取" +msgstr "無法開啟檔案「%s」來讀取" #: builtin/bisect.c #, c-format msgid "'%s' is not a valid term" -msgstr "「%s」不是有效術語" +msgstr "「%s」不是有效的術語" #: builtin/bisect.c #, c-format @@ -2917,7 +2933,7 @@ msgstr "「%s」不是有效的提交" #, c-format msgid "" "could not check out original HEAD '%s'. Try 'git bisect reset '." -msgstr "無法簽出原始 HEAD「%s」。請嘗試「git bisect reset 」。" +msgstr "無法簽出原本的 HEAD「%s」。請嘗試「git bisect reset 」。" #: builtin/bisect.c #, c-format @@ -2937,7 +2953,7 @@ msgstr "無法開啟檔案「%s」" #: builtin/bisect.c #, c-format msgid "Invalid command: you're currently in a %s/%s bisect" -msgstr "命令無效：您目前正處於二分搜尋 %s/%s 的狀態" +msgstr "命令無效：您正處於二分搜尋 %s/%s 的狀態" #: builtin/bisect.c #, c-format @@ -3263,7 +3279,7 @@ msgstr "顯示作者信箱而非名稱（預設值：off）" msgid "ignore whitespace differences" msgstr "忽略空白差異" -#: builtin/blame.c builtin/log.c +#: builtin/blame.c builtin/clone.c builtin/log.c msgid "rev" msgstr "rev" @@ -3775,11 +3791,6 @@ msgstr "" msgid "git version:\n" msgstr "git 版本：\n" -#: builtin/bugreport.c -#, c-format -msgid "uname() failed with error '%s' (%d)\n" -msgstr "uname() 失敗，錯誤：「%s」(%d)\n" - #: builtin/bugreport.c msgid "compiler info: " msgstr "編譯器資訊： " @@ -5010,8 +5021,112 @@ msgid "clean.requireForce is true and -f not given: refusing to clean" msgstr "clean.requireForce 是 true 且未給定 -f 選項：拒絕清理" #: builtin/clone.c -msgid "git clone [] [--] []" -msgstr "git clone [] [--] []" +#, c-format +msgid "info: Could not add alternate for '%s': %s\n" +msgstr "info: 不能為「%s」新增一個備用：%s\n" + +#: builtin/clone.c builtin/diff.c builtin/rm.c grep.c setup.c +#, c-format +msgid "failed to stat '%s'" +msgstr "對「%s」呼叫 stat 失敗" + +#: builtin/clone.c +#, c-format +msgid "%s exists and is not a directory" +msgstr "%s 存在且不是目錄" + +#: builtin/clone.c +#, c-format +msgid "'%s' is a symlink, refusing to clone with --local" +msgstr "「%s」是符號連結，故不能使用 --local 複製" + +#: builtin/clone.c +#, c-format +msgid "failed to start iterator over '%s'" +msgstr "無法在「%s」上啟動迭代器" + +#: builtin/clone.c +#, c-format +msgid "symlink '%s' exists, refusing to clone with --local" +msgstr "符號連結「%s」已存在，拒絕使用 --local 複製" + +#: builtin/clone.c compat/precompose_utf8.c +#, c-format +msgid "failed to unlink '%s'" +msgstr "無法刪除「%s」" + +#: builtin/clone.c +#, c-format +msgid "hardlink cannot be checked at '%s'" +msgstr "無法檢查位於「%s」的硬連結" + +#: builtin/clone.c +#, c-format +msgid "hardlink different from source at '%s'" +msgstr "硬連結與位於「%s」的來源不同" + +#: builtin/clone.c +#, c-format +msgid "failed to create link '%s'" +msgstr "建立連結「%s」失敗" + +#: builtin/clone.c +#, c-format +msgid "failed to copy file to '%s'" +msgstr "複製檔案至「%s」失敗" + +#: builtin/clone.c refs/files-backend.c +#, c-format +msgid "failed to iterate over '%s'" +msgstr "無法在「%s」上迭代" + +#: builtin/clone.c +#, c-format +msgid "done.\n" +msgstr "完成。\n" + +#: builtin/clone.c +msgid "" +"Clone succeeded, but checkout failed.\n" +"You can inspect what was checked out with 'git status'\n" +"and retry with 'git restore --source=HEAD :/'\n" +msgstr "" +"複製成功，但是簽出失敗。\n" +"您可以透過「git status」檢查哪些已被簽出，然後使用指令\n" +"「git restore --source=HEAD :/」重試\n" + +#: builtin/clone.c fetch-pack.c +msgid "remote did not send all necessary objects" +msgstr "遠端沒有傳送所有必需的物件" + +#: builtin/clone.c +#, c-format +msgid "unable to update %s" +msgstr "不能更新 %s" + +#: builtin/clone.c +msgid "failed to initialize sparse-checkout" +msgstr "無法初始化稀疏簽出" + +#: builtin/clone.c +msgid "remote HEAD refers to nonexistent ref, unable to checkout" +msgstr "遠端 HEAD 指向不存在的引用，無法簽出" + +#: builtin/clone.c +msgid "unable to checkout working tree" +msgstr "不能簽出工作區" + +#: builtin/clone.c +msgid "unable to write parameters to config file" +msgstr "無法將參數寫入組態檔案" + +#: builtin/clone.c +msgid "cannot repack to clean up" +msgstr "無法執行 repack 來清理" + +#: builtin/clone.c +msgid "cannot unlink temporary alternates file" +msgstr "無法刪除暫存 alternates 檔案" #: builtin/clone.c msgid "don't clone shallow repository" @@ -5083,6 +5198,10 @@ msgstr "使用 而不是「origin」追蹤上游" msgid "checkout instead of the remote's HEAD" msgstr "簽出 而不是遠端 HEAD" +#: builtin/clone.c +msgid "clone single revision and check out" +msgstr "複製單個修訂版 並簽出" + #: builtin/clone.c msgid "path to git-upload-pack on the remote" msgstr "遠端 git-upload-pack 路徑" @@ -5112,8 +5231,8 @@ msgid "clone only one branch, HEAD or --branch" msgstr "只複製一個分支、HEAD 或 --branch" #: builtin/clone.c -msgid "don't clone any tags, and make later fetches not to follow them" -msgstr "不要複製任何標籤，之後取得也不要追蹤這些標籤" +msgid "clone tags, and make later fetches not to follow them" +msgstr "複製標籤，並使後續抓取不要追蹤這些標籤" #: builtin/clone.c msgid "any cloned submodules will be shallow" @@ -5170,117 +5289,8 @@ msgid "a URI for downloading bundles before fetching from origin remote" msgstr "在從 origin 遠端抓取前，用來下載套件包的 URI" #: builtin/clone.c -#, c-format -msgid "info: Could not add alternate for '%s': %s\n" -msgstr "info: 不能為 '%s' 新增一個備用：%s\n" - -#: builtin/clone.c builtin/diff.c builtin/rm.c grep.c setup.c -#, c-format -msgid "failed to stat '%s'" -msgstr "對 '%s' 呼叫 stat 失敗" - -#: builtin/clone.c -#, c-format -msgid "%s exists and is not a directory" -msgstr "%s 存在且不是一個目錄" - -#: builtin/clone.c -#, c-format -msgid "'%s' is a symlink, refusing to clone with --local" -msgstr "「%s」是個符號連結，故不能使用 --local 複製" - -#: builtin/clone.c -#, c-format -msgid "failed to start iterator over '%s'" -msgstr "無法在 '%s' 上啟動疊代器" - -#: builtin/clone.c -#, c-format -msgid "symlink '%s' exists, refusing to clone with --local" -msgstr "「%s」符號連結已存在，拒絕使用 --local 複製" - -#: builtin/clone.c compat/precompose_utf8.c -#, c-format -msgid "failed to unlink '%s'" -msgstr "無法刪除「%s」" - -#: builtin/clone.c -#, c-format -msgid "hardlink cannot be checked at '%s'" -msgstr "無法檢查位於「%s」的硬連結" - -#: builtin/clone.c -#, c-format -msgid "hardlink different from source at '%s'" -msgstr "硬連結與位於 '%s' 的來源不同" - -#: builtin/clone.c -#, c-format -msgid "failed to create link '%s'" -msgstr "建立連結 '%s' 失敗" - -#: builtin/clone.c -#, c-format -msgid "failed to copy file to '%s'" -msgstr "複製檔案至 '%s' 失敗" - -#: builtin/clone.c refs/files-backend.c -#, c-format -msgid "failed to iterate over '%s'" -msgstr "無法在 '%s' 上疊代" - -#: builtin/clone.c -#, c-format -msgid "done.\n" -msgstr "完成。\n" - -#: builtin/clone.c -msgid "" -"Clone succeeded, but checkout failed.\n" -"You can inspect what was checked out with 'git status'\n" -"and retry with 'git restore --source=HEAD :/'\n" -msgstr "" -"複製成功，但是簽出失敗。\n" -"您可以透過 'git status' 檢查哪些已被簽出，然後使用指令\n" -"'git restore --source=HEAD :/' 重試\n" - -#: builtin/clone.c -#, c-format -msgid "Could not find remote branch %s to clone." -msgstr "找不到要複製的遠端分支 %s。" - -#: builtin/clone.c fetch-pack.c -msgid "remote did not send all necessary objects" -msgstr "遠端沒有傳送所有必需的物件" - -#: builtin/clone.c -#, c-format -msgid "unable to update %s" -msgstr "不能更新 %s" - -#: builtin/clone.c -msgid "failed to initialize sparse-checkout" -msgstr "無法初始化稀疏簽出" - -#: builtin/clone.c -msgid "remote HEAD refers to nonexistent ref, unable to checkout" -msgstr "遠端 HEAD 指向一個不存在的引用，無法簽出" - -#: builtin/clone.c -msgid "unable to checkout working tree" -msgstr "不能簽出工作區" - -#: builtin/clone.c -msgid "unable to write parameters to config file" -msgstr "無法將參數寫入設定檔案" - -#: builtin/clone.c -msgid "cannot repack to clean up" -msgstr "無法執行 repack 來清理" - -#: builtin/clone.c -msgid "cannot unlink temporary alternates file" -msgstr "無法刪除暫存 alternates 檔案" +msgid "git clone [] [--] []" +msgstr "git clone [] [--] []" #: builtin/clone.c msgid "Too many arguments." @@ -5404,7 +5414,12 @@ msgstr "遠端傳輸回報錯誤" #: builtin/clone.c #, c-format msgid "Remote branch %s not found in upstream %s" -msgstr "遠端分支 %s 在上游 %s 未發現" +msgstr "上游 %2$s 上找不到遠端分支 %1$s" + +#: builtin/clone.c +#, c-format +msgid "Remote revision %s not found in upstream %s" +msgstr "上游 %2$s 上找不到遠端修訂版 %1$s" #: builtin/clone.c msgid "You appear to have cloned an empty repository." @@ -5469,7 +5484,8 @@ msgstr "" "[no-]progress]\n" " " -#: builtin/commit-graph.c builtin/fetch.c builtin/log.c builtin/repack.c +#: builtin/commit-graph.c builtin/fetch.c builtin/gc.c builtin/log.c +#: builtin/repack.c msgid "dir" msgstr "目錄" @@ -5626,8 +5642,19 @@ msgid "git commit-tree: failed to read" msgstr "git commit-tree：讀取失敗" #: builtin/commit.c -msgid "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +#| msgid "" +#| "git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +#| " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" +#| "reword):]]\n" +#| " [-F | -m ] [--reset-author] [--allow-empty]\n" +#| " [--allow-empty-message] [--no-verify] [-e] [--" +#| "author=]\n" +#| " [--date=] [--cleanup=] [--[no-]status]\n" +#| " [-i | -o] [--pathspec-from-file= [--pathspec-file-nul]]\n" +#| " [(--trailer [(=|:)])...] [-S[]]\n" +#| " [--] [...]" +msgid "" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -5637,7 +5664,7 @@ msgid "" " [(--trailer [(=|:)])...] [-S[]]\n" " [--] [...]" msgstr "" -"git commit [-a | --interactive | --patch] [-s] [-v] [-u] [--amend]\n" +"git commit [-a | --interactive | --patch] [-s] [-v] [-u[]] [--amend]\n" " [--dry-run] [(-c | -C | --squash) | --fixup [(amend|" "reword):]]\n" " [-F | -m ] [--reset-author] [--allow-empty]\n" @@ -7358,14 +7385,14 @@ msgid "" "Run 'git remote set-head %s %s' to follow the change, or set\n" "'remote.%s.followRemoteHEAD' configuration option to a different value\n" "if you do not want to see this message. Specifically running\n" -"'git config set remote.%s.followRemoteHEAD %s' will disable the warning\n" -"until the remote changes HEAD to something else." +"'git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s'\n" +"will disable the warning until the remote changes HEAD to something else." msgstr "" -"執行「git remote set-head %s %s」以追蹤這個變更，或者\n" -"如果您不想看到這則訊息，請將「remote.%s.followRemoteHEAD」\n" -"組態選項設定成不同的值。更具體些來說，執行\n" -"「git config set remote.%s.followRemoteHEAD %s」會停用這個警告，\n" -"直到遠端將 HEAD 變更為其他內容。" +"執行「git remote set-head %s %s」來跟進此差異，或者\n" +"如果您不想看到這則訊息，請將「remote.%s.followRemoteHEAD」組態選項\n" +"設定成別的值。更具體些來說，執行\n" +"「git config set remote.%s.followRemoteHEAD warn-if-not-branch-%s」\n" +"會停用這個警告，直到遠端將 HEAD 改為指向其他東西。" #: builtin/fetch.c msgid "multiple branches detected, incompatible with --set-upstream" @@ -8216,6 +8243,10 @@ msgstr "強制執行 gc 即使另外一個 gc 正在執行" msgid "repack all other packs except the largest pack" msgstr "除了最大的包之外，對所有其它包重新打包" +#: builtin/gc.c builtin/repack.c +msgid "pack prefix to store a pack containing pruned objects" +msgstr "將前綴打包並儲存為包含已剪除物件的包" + #: builtin/gc.c #, c-format msgid "failed to parse gc.logExpiry value %s" @@ -9209,6 +9240,11 @@ msgstr "無法結束 pack-objects 來重新封包" msgid "Cannot come back to cwd" msgstr "無法返回目前工作目錄" +#: builtin/index-pack.c builtin/unpack-objects.c +#, c-format +msgid "bad --pack_header: %s" +msgstr "無效的 --pack_header：%s" + #: builtin/index-pack.c #, c-format msgid "bad %s" @@ -10311,11 +10347,6 @@ msgstr "未知的策略選項：-X%s" msgid "malformed input line: '%s'." msgstr "格式錯誤的輸入行：'%s'。" -#: builtin/merge-tree.c -#, c-format -msgid "merging cannot continue; got unclean result of %d" -msgstr "無法繼續合併：從 %d 收到的結果不乾淨" - #: builtin/merge.c msgid "git merge [] [...]" msgstr "git merge [<選項>] [<提交>...]" @@ -11332,6 +11363,15 @@ msgid "" "git pack-objects [] [< | < ]" msgstr "git pack-objects [<選項>] <前綴名稱> [< <引用列表> | < <物件列表>]" +#: builtin/pack-objects.c +#, c-format +msgid "invalid --name-hash-version option: %d" +msgstr "無效的 --name-hash-version 選項：%d" + +#: builtin/pack-objects.c +msgid "currently, --write-bitmap-index requires --name-hash-version=1" +msgstr "目前 --write-bitmap-index 需要指定 --name-hash-version=1" + #: builtin/pack-objects.c #, c-format msgid "" @@ -11468,8 +11508,8 @@ msgid "" "value of uploadpack.blobpackfileuri must be of the form ' ' (got '%s')" msgstr "" -"uploadpack.blobpackfileuri 的值格式必須為 ' " -"' (收到 '%s')" +"uploadpack.blobpackfileuri 的值格式必須為「 」" +"（收到「%s」）" #: builtin/pack-objects.c #, c-format @@ -11739,6 +11779,10 @@ msgstr "通訊協定" msgid "exclude any configured uploadpack.blobpackfileuri with this protocol" msgstr "排除任何設定過，使用此通訊協定的 uploadpack.blobpackfileuri" +#: builtin/pack-objects.c +msgid "use the specified name-hash function to group similar objects" +msgstr "使用指定的名稱雜湊函式來為相似物件分組" + #: builtin/pack-objects.c #, c-format msgid "delta chain depth %d is too deep, forcing %d" @@ -13178,8 +13222,9 @@ msgid "invalid ref format: %s" msgstr "無效的引用格式：%s" #: builtin/refs.c -msgid "git refs migrate --ref-format= [--dry-run]" -msgstr "git refs migrate --ref-format=<格式> [--dry-run]" +#| msgid "git refs migrate --ref-format= [--dry-run]" +msgid "git refs migrate --ref-format= [--no-reflog] [--dry-run]" +msgstr "git refs migrate --ref-format= [--no-reflog] [--dry-run]" #: builtin/refs.c msgid "git refs verify [--strict] [--verbose]" @@ -13193,6 +13238,10 @@ msgstr "指定要轉換成的引用格式" msgid "perform a non-destructive dry-run" msgstr "進行非破壞性的試執行" +#: builtin/refs.c +msgid "drop reflogs entirely during the migration" +msgstr "在遷移過程中完全拋棄 reflog" + #: builtin/refs.c msgid "missing --ref-format=" msgstr "缺少 --ref-format=<格式>" @@ -13764,8 +13813,14 @@ msgid "be verbose; must be placed before a subcommand" msgstr "詳細輸出；必須置於子指令之前" #: builtin/repack.c -msgid "git repack []" -msgstr "git repack [<選項>]" +msgid "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" +msgstr "" +"git repack [-a] [-A] [-d] [-f] [-F] [-l] [-n] [-q] [-b] [-m]\n" +"[--window=] [--depth=] [--threads=] [--keep-pack=]\n" +"[--write-midx] [--name-hash-version=]" #: builtin/repack.c msgid "" @@ -13857,6 +13912,11 @@ msgstr "向 git-pack-objects 傳遞參數 --no-reuse-delta" msgid "pass --no-reuse-object to git-pack-objects" msgstr "向 git-pack-objects 傳遞參數 --no-reuse-object" +#: builtin/repack.c +msgid "" +"specify the name hash version to use for grouping similar objects by path" +msgstr "指定要用來以路徑為相似物件分組的名稱雜湊版本" + #: builtin/repack.c msgid "do not run git-update-server-info" msgstr "不執行 git-update-server-info" @@ -13921,10 +13981,6 @@ msgstr "尋找因數是 的等比數列" msgid "write a multi-pack index of the resulting packs" msgstr "寫入結果包的多包索引" -#: builtin/repack.c -msgid "pack prefix to store a pack containing pruned objects" -msgstr "封裝前綴，儲存為包含過時物件的套件包" - #: builtin/repack.c msgid "pack prefix to store a pack containing filtered out objects" msgstr "將前綴進行包裝，儲存為包含已過濾物件的封裝" @@ -14197,10 +14253,6 @@ msgstr "只能為 -l 提供一個模式" msgid "need some commits to replay" msgstr "需要一些提交才能重放" -#: builtin/replay.c -msgid "--onto and --advance are incompatible" -msgstr "--onto 和 --advance 不相容" - #: builtin/replay.c msgid "all positive revisions given must be references" msgstr "提供的所有正向修訂集必須為引用" @@ -17423,6 +17475,10 @@ msgstr "將一個 GNU Arch 版本庫匯入到 Git" msgid "Create an archive of files from a named tree" msgstr "基於命名過的樹建立檔案封存" +#: command-list.h +msgid "Download missing objects in a partial clone" +msgstr "在部分複製中下載缺少的物件" + #: command-list.h msgid "Use binary search to find the commit that introduced a bug" msgstr "透過二分搜尋定位引入 bug 的提交" @@ -18378,8 +18434,8 @@ msgstr "嘗試寫入提交圖，但「core.commitGraph」已停用" #: commit-graph.c #, c-format msgid "" -"attempting to write a commit-graph, but 'commitGraph." -"changedPathsVersion' (%d) is not supported" +"attempting to write a commit-graph, but 'commitGraph.changedPathsVersion' " +"(%d) is not supported" msgstr "嘗試寫入提交圖，但不支援「commitGraph.changedPathsVersion」（%d）" #: commit-graph.c @@ -19790,6 +19846,14 @@ msgstr "%s 的參數無效" msgid "invalid regex given to -I: '%s'" msgstr "傳入 -I 的常規表示式無效：「%s」" +#: diff.c +msgid "-G requires a non-empty argument" +msgstr "-G 需要非空白引數" + +#: diff.c +msgid "-S requires a non-empty argument" +msgstr "-S 需要非空白引數" + #: diff.c #, c-format msgid "failed to parse --submodule option parameter: '%s'" @@ -22417,6 +22481,11 @@ msgstr "「%s」和「%s」檔案內容不同" msgid "unable to write file %s" msgstr "無法寫檔案 %s" +#: object-file.c +#, c-format +msgid "unable to write repeatedly vanishing file %s" +msgstr "無法寫入重複消失的檔案 %s" + #: object-file.c #, c-format msgid "unable to set permission to '%s'" @@ -23100,6 +23169,55 @@ msgstr "未知開關 `%c'" msgid "unknown non-ascii option in string: `%s'" msgstr "字串中未知的非 ascii 字元選項：`%s'" +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the long form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid "[=<%s>]" +msgstr "[=<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string +#. stands for an optional value given to a command +#. line option in the short form, and "<>" is there +#. as a convention to signal that it is a +#. placeholder (i.e. the user should substitute it +#. with the real value). If your language uses a +#. different convention, you can change "<%s>" part +#. to match yours, e.g. it might use "|%s|" instead, +#. or if the alphabet is different enough it may use +#. "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid "[<%s>]" +msgstr "[<%s>]" + +#. TRANSLATORS: The "<%s>" part of this string stands for a +#. value given to a command line option, and "<>" is there +#. as a convention to signal that it is a placeholder +#. (i.e. the user should substitute it with the real value). +#. If your language uses a different convention, you can +#. change "<%s>" part to match yours, e.g. it might use +#. "|%s|" instead, or if the alphabet is different enough it +#. may use "%s" without any placeholder signal. Most +#. translations leave this message as is. +#. +#: parse-options.c +#, c-format +msgid " <%s>" +msgstr " <%s>" + #: parse-options.c msgid "..." msgstr "..." @@ -23203,6 +23321,25 @@ msgstr "「%2$s」的「%1$s」布林環境值無效" msgid "failed to parse %s" msgstr "解析 %s 失敗" +#: path-walk.c +#, c-format +msgid "failed to walk children of tree %s: not found" +msgstr "無法走訪樹 %s 的子代：找不到" + +#: path-walk.c +#, c-format +msgid "failed to find object %s" +msgstr "找不到物件 %s" + +#: path-walk.c +#, c-format +msgid "failed to find tag %s" +msgstr "找不到標籤 %s" + +#: path-walk.c +msgid "failed to setup revision walk" +msgstr "無法設置修訂版走訪" + #: path.c #, c-format msgid "Could not make %s writable by group" @@ -23384,6 +23521,26 @@ msgstr "promisor 遠端名稱不能以 '/' 開始：%s" msgid "could not fetch %s from promisor remote" msgstr "無法從承諾者遠端抓取 %s" +#: promisor-remote.c +#, c-format +msgid "known remote named '%s' but with url '%s' instead of '%s'" +msgstr "已知有遠端名為「%s」，但其 URL 為「%s」而非「%s」" + +#: promisor-remote.c +#, c-format +msgid "unknown '%s' value for '%s' config option" +msgstr "「%2$s」組態選項的值「%1$s」未知" + +#: promisor-remote.c +#, c-format +msgid "unknown element '%s' from remote info" +msgstr "遠端資訊的元素「%s」未知" + +#: promisor-remote.c +#, c-format +msgid "accepted promisor remote '%s' not found" +msgstr "找不到接受的承諾者遠端「%s」" + #: protocol-caps.c msgid "object-info: expected flush after arguments" msgstr "object-info：引數後預期要有 flush" @@ -24351,6 +24508,16 @@ msgstr "引用名稱 %s 是符號引用，不支援複製" msgid "invalid refspec '%s'" msgstr "無效的引用規格：「%s」" +#: refspec.c +#, c-format +msgid "pattern '%s' has no '*'" +msgstr "符合模式「%s」中沒有「*」" + +#: refspec.c +#, c-format +msgid "replacement '%s' has no '*'" +msgstr "取代文字「%s」中沒有「*」" + #: remote-curl.c #, c-format msgid "invalid quoting in push-option value: '%s'" @@ -24502,6 +24669,28 @@ msgstr "remote-curl：嘗試沒有本機版本庫下取得" msgid "remote-curl: unknown command '%s' from git" msgstr "remote-curl：未知的來自 git 的指令 '%s'" +#: remote.c +#, c-format +msgid "" +"reading remote from \"%s/%s\", which is nominated for removal.\n" +"\n" +"If you still use the \"remotes/\" directory it is recommended to\n" +"migrate to config-based remotes:\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"If you cannot, please let us know why you still need to use it by\n" +"sending an e-mail to ." +msgstr "" +"正在自「%s/%s」讀取遠端，該路徑已被標記為即將移除。\n" +"\n" +"如果您仍在使用「remotes/」目錄，建議您遷移至基於組態的遠端：\n" +"\n" +"\tgit remote rename %s %s\n" +"\n" +"如果您無法遷移，請告訴我們您為何仍需使用此功能，\n" +"請寄信至 。" + #: remote.c #, c-format msgid "config remote shorthand cannot begin with '/': %s" @@ -24545,16 +24734,6 @@ msgstr "%s 通常追蹤 %s，而非 %s" msgid "%s tracks both %s and %s" msgstr "%s 同時追蹤 %s 和 %s" -#: remote.c -#, c-format -msgid "key '%s' of pattern had no '*'" -msgstr "模式的鍵 '%s' 沒有 '*'" - -#: remote.c -#, c-format -msgid "value '%s' of pattern has no '*'" -msgstr "模式的值 '%s' 沒有 '*'" - #: remote.c #, c-format msgid "src refspec %s does not match any" @@ -26824,6 +27003,34 @@ msgstr "每次迭代前清除快取樹狀物件" msgid "number of entries in the cache tree to invalidate (default 0)" msgstr "在快取樹狀物件中，要使失效的項目數量（預設值為 0）" +#: t/helper/test-path-walk.c +msgid "test-tool path-walk -- " +msgstr "test-tool path-walk -- " + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of blob objects" +msgstr "切換是否包含資料物件" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of commit objects" +msgstr "切換是否包含提交物件" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of tag objects" +msgstr "切換是否包含標籤物件" + +#: t/helper/test-path-walk.c +msgid "toggle inclusion of tree objects" +msgstr "切換是否包含樹狀物件" + +#: t/helper/test-path-walk.c +msgid "toggle pruning of uninteresting paths" +msgstr "切換是否剪除不重要路徑" + +#: t/helper/test-path-walk.c +msgid "read a pattern list over stdin" +msgstr "從標準輸入讀取模式清單" + #: t/helper/test-reach.c #, c-format msgid "commit %s is not marked reachable" @@ -27571,6 +27778,11 @@ msgstr "錯誤: " msgid "warning: " msgstr "警告: " +#: version.c +#, c-format +msgid "uname() failed with error '%s' (%d)\n" +msgstr "uname() 失敗，錯誤：「%s」(%d)\n" + #: walker.c msgid "Fetching objects" msgstr "正在抓取物件" @@ -28757,6 +28969,24 @@ msgstr "略過 %s 含備份後綴 '%s'。\n" msgid "Do you really want to send %s? [y|N]: " msgstr "您真的要傳送 %s？[y|N]： " +#, c-format +#~ msgid "Could not find remote branch %s to clone." +#~ msgstr "找不到要複製的遠端分支 %s。" + +#, c-format +#~ msgid "merging cannot continue; got unclean result of %d" +#~ msgstr "無法繼續合併：從 %d 收到的結果不乾淨" + +#~ msgid "git repack []" +#~ msgstr "git repack [<選項>]" + +#~ msgid "--onto and --advance are incompatible" +#~ msgstr "--onto 和 --advance 不相容" + +#, c-format +#~ msgid "key '%s' of pattern had no '*'" +#~ msgstr "模式的鍵 '%s' 沒有 '*'" + #, c-format #~ msgid "preferred pack (%s) is invalid" #~ msgstr "偏好的封包 (%s) 無效" From 683c54c999c301c2cd6f715c411407c413b1d84e Mon Sep 17 00:00:00 2001 From: Junio C Hamano Date: Fri, 14 Mar 2025 09:19:41 -0700 Subject: [PATCH 020/259] Git 2.49 Signed-off-by: Junio C Hamano --- GIT-VERSION-GEN | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/GIT-VERSION-GEN b/GIT-VERSION-GEN index 8979fb2803285a..3abfe7d3d7e79d 100755 --- a/GIT-VERSION-GEN +++ b/GIT-VERSION-GEN @@ -1,6 +1,6 @@ #!/bin/sh -DEF_VER=v2.49.0-rc2 +DEF_VER=v2.49.0 LF=' ' From 328ff864183cdd0a4b779b5b88a3271b39a1b1a2 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 6 Nov 2024 20:34:50 +0100 Subject: [PATCH 021/259] sideband: mask control characters The output of `git clone` is a vital component for understanding what has happened when things go wrong. However, these logs are partially under the control of the remote server (via the "sideband", which typically contains what the remote `git pack-objects` process sends to `stderr`), and is currently not sanitized by Git. This makes Git susceptible to ANSI escape sequence injection (see CWE-150, https://cwe.mitre.org/data/definitions/150.html), which allows attackers to corrupt terminal state, to hide information, and even to insert characters into the input buffer (i.e. as if the user had typed those characters). To plug this vulnerability, disallow any control character in the sideband, replacing them instead with the common `^` (e.g. `^[` for `\x1b`, `^A` for `\x01`). There is likely a need for more fine-grained controls instead of using a "heavy hammer" like this, which will be introduced subsequently. Signed-off-by: Johannes Schindelin --- sideband.c | 17 +++++++++++++++-- t/t5409-colorize-remote-messages.sh | 12 ++++++++++++ 2 files changed, 27 insertions(+), 2 deletions(-) diff --git a/sideband.c b/sideband.c index 251e9615ed0239..81b1ff0805b3f6 100644 --- a/sideband.c +++ b/sideband.c @@ -66,6 +66,19 @@ void list_config_color_sideband_slots(struct string_list *list, const char *pref list_config_item(list, prefix, keywords[i].keyword); } +static void strbuf_add_sanitized(struct strbuf *dest, const char *src, int n) +{ + strbuf_grow(dest, n); + for (; n && *src; src++, n--) { + if (!iscntrl(*src) || *src == '\t' || *src == '\n') + strbuf_addch(dest, *src); + else { + strbuf_addch(dest, '^'); + strbuf_addch(dest, 0x40 + *src); + } + } +} + /* * Optionally highlight one keyword in remote output if it appears at the start * of the line. This should be called for a single line only, which is @@ -81,7 +94,7 @@ static void maybe_colorize_sideband(struct strbuf *dest, const char *src, int n) int i; if (!want_color_stderr(use_sideband_colors())) { - strbuf_add(dest, src, n); + strbuf_add_sanitized(dest, src, n); return; } @@ -114,7 +127,7 @@ static void maybe_colorize_sideband(struct strbuf *dest, const char *src, int n) } } - strbuf_add(dest, src, n); + strbuf_add_sanitized(dest, src, n); } diff --git a/t/t5409-colorize-remote-messages.sh b/t/t5409-colorize-remote-messages.sh index fa5de4500a4f50..d0745c391b2625 100755 --- a/t/t5409-colorize-remote-messages.sh +++ b/t/t5409-colorize-remote-messages.sh @@ -98,4 +98,16 @@ test_expect_success 'fallback to color.ui' ' grep "error: error" decoded ' +test_expect_success 'disallow (color) control sequences in sideband' ' + write_script .git/color-me-surprised <<-\EOF && + printf "error: Have you \\033[31mread\\033[m this?\\n" >&2 + exec "$@" + EOF + test_config_global uploadPack.packObjectshook ./color-me-surprised && + test_commit need-at-least-one-commit && + git clone --no-local . throw-away 2>stderr && + test_decode_color decoded && + test_grep ! RED decoded +' + test_done From ab2eb6c0043c643935ea0fbdaed68e15bc831b11 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 6 Nov 2024 21:07:51 +0100 Subject: [PATCH 022/259] sideband: introduce an "escape hatch" to allow control characters The preceding commit fixed the vulnerability whereas sideband messages (that are under the control of the remote server) could contain ANSI escape sequences that would be sent to the terminal verbatim. However, this fix may not be desirable under all circumstances, e.g. when remote servers deliberately add coloring to their messages to increase their urgency. To help with those use cases, give users a way to opt-out of the protections: `sideband.allowControlCharacters`. Signed-off-by: Johannes Schindelin --- Documentation/config.adoc | 2 ++ Documentation/config/sideband.adoc | 5 +++++ sideband.c | 10 ++++++++++ t/t5409-colorize-remote-messages.sh | 8 +++++++- 4 files changed, 24 insertions(+), 1 deletion(-) create mode 100644 Documentation/config/sideband.adoc diff --git a/Documentation/config.adoc b/Documentation/config.adoc index cc769251be339b..a8b04c4e5183d2 100644 --- a/Documentation/config.adoc +++ b/Documentation/config.adoc @@ -522,6 +522,8 @@ include::config/sequencer.adoc[] include::config/showbranch.adoc[] +include::config/sideband.adoc[] + include::config/sparse.adoc[] include::config/splitindex.adoc[] diff --git a/Documentation/config/sideband.adoc b/Documentation/config/sideband.adoc new file mode 100644 index 00000000000000..3fb5045cd79581 --- /dev/null +++ b/Documentation/config/sideband.adoc @@ -0,0 +1,5 @@ +sideband.allowControlCharacters:: + By default, control characters that are delivered via the sideband + are masked, to prevent potentially unwanted ANSI escape sequences + from being sent to the terminal. Use this config setting to override + this behavior. diff --git a/sideband.c b/sideband.c index 81b1ff0805b3f6..d1c326fa191068 100644 --- a/sideband.c +++ b/sideband.c @@ -26,6 +26,8 @@ static struct keyword_entry keywords[] = { { "error", GIT_COLOR_BOLD_RED }, }; +static int allow_control_characters; + /* Returns a color setting (GIT_COLOR_NEVER, etc). */ static int use_sideband_colors(void) { @@ -39,6 +41,9 @@ static int use_sideband_colors(void) if (use_sideband_colors_cached >= 0) return use_sideband_colors_cached; + git_config_get_bool("sideband.allowcontrolcharacters", + &allow_control_characters); + if (!git_config_get_string_tmp(key, &value)) use_sideband_colors_cached = git_config_colorbool(key, value); else if (!git_config_get_string_tmp("color.ui", &value)) @@ -68,6 +73,11 @@ void list_config_color_sideband_slots(struct string_list *list, const char *pref static void strbuf_add_sanitized(struct strbuf *dest, const char *src, int n) { + if (allow_control_characters) { + strbuf_add(dest, src, n); + return; + } + strbuf_grow(dest, n); for (; n && *src; src++, n--) { if (!iscntrl(*src) || *src == '\t' || *src == '\n') diff --git a/t/t5409-colorize-remote-messages.sh b/t/t5409-colorize-remote-messages.sh index d0745c391b2625..fb31e8525418a1 100755 --- a/t/t5409-colorize-remote-messages.sh +++ b/t/t5409-colorize-remote-messages.sh @@ -105,9 +105,15 @@ test_expect_success 'disallow (color) control sequences in sideband' ' EOF test_config_global uploadPack.packObjectshook ./color-me-surprised && test_commit need-at-least-one-commit && + git clone --no-local . throw-away 2>stderr && test_decode_color decoded && - test_grep ! RED decoded + test_grep ! RED decoded && + + rm -rf throw-away && + git -c sideband.allowControlCharacters clone --no-local . throw-away 2>stderr && + test_decode_color decoded && + test_grep RED decoded ' test_done From a369672c2e6974590ad0561854318a4f255e6893 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 18 Nov 2024 21:42:57 +0100 Subject: [PATCH 023/259] sideband: do allow ANSI color sequences by default The preceding two commits introduced special handling of the sideband channel to neutralize ANSI escape sequences before sending the payload to the terminal, and `sideband.allowControlCharacters` to override that behavior. However, some `pre-receive` hooks that are actively used in practice want to color their messages and therefore rely on the fact that Git passes them through to the terminal. In contrast to other ANSI escape sequences, it is highly unlikely that coloring sequences can be essential tools in attack vectors that mislead Git users e.g. by hiding crucial information. Therefore we can have both: Continue to allow ANSI coloring sequences to be passed to the terminal, and neutralize all other ANSI escape sequences. Signed-off-by: Johannes Schindelin --- Documentation/config/sideband.adoc | 17 ++++++-- sideband.c | 61 ++++++++++++++++++++++++++--- t/t5409-colorize-remote-messages.sh | 16 +++++++- 3 files changed, 84 insertions(+), 10 deletions(-) diff --git a/Documentation/config/sideband.adoc b/Documentation/config/sideband.adoc index 3fb5045cd79581..f347fd6b33004a 100644 --- a/Documentation/config/sideband.adoc +++ b/Documentation/config/sideband.adoc @@ -1,5 +1,16 @@ sideband.allowControlCharacters:: By default, control characters that are delivered via the sideband - are masked, to prevent potentially unwanted ANSI escape sequences - from being sent to the terminal. Use this config setting to override - this behavior. + are masked, except ANSI color sequences. This prevents potentially + unwanted ANSI escape sequences from being sent to the terminal. Use + this config setting to override this behavior: ++ +-- + color:: + Allow ANSI color sequences, line feeds and horizontal tabs, + but mask all other control characters. This is the default. + false:: + Mask all control characters other than line feeds and + horizontal tabs. + true:: + Allow all control characters to be sent to the terminal. +-- diff --git a/sideband.c b/sideband.c index d1c326fa191068..9084ca234dcecc 100644 --- a/sideband.c +++ b/sideband.c @@ -26,7 +26,11 @@ static struct keyword_entry keywords[] = { { "error", GIT_COLOR_BOLD_RED }, }; -static int allow_control_characters; +static enum { + ALLOW_NO_CONTROL_CHARACTERS = 0, + ALLOW_ALL_CONTROL_CHARACTERS = 1, + ALLOW_ANSI_COLOR_SEQUENCES = 2 +} allow_control_characters = ALLOW_ANSI_COLOR_SEQUENCES; /* Returns a color setting (GIT_COLOR_NEVER, etc). */ static int use_sideband_colors(void) @@ -41,8 +45,24 @@ static int use_sideband_colors(void) if (use_sideband_colors_cached >= 0) return use_sideband_colors_cached; - git_config_get_bool("sideband.allowcontrolcharacters", - &allow_control_characters); + switch (git_config_get_maybe_bool("sideband.allowcontrolcharacters", &i)) { + case 0: /* Boolean value */ + allow_control_characters = i ? ALLOW_ALL_CONTROL_CHARACTERS : + ALLOW_NO_CONTROL_CHARACTERS; + break; + case -1: /* non-Boolean value */ + if (git_config_get_string_tmp("sideband.allowcontrolcharacters", + &value)) + ; /* huh? `get_maybe_bool()` returned -1 */ + else if (!strcmp(value, "color")) + allow_control_characters = ALLOW_ANSI_COLOR_SEQUENCES; + else + warning(_("unrecognized value for `sideband." + "allowControlCharacters`: '%s'"), value); + break; + default: + break; /* not configured */ + } if (!git_config_get_string_tmp(key, &value)) use_sideband_colors_cached = git_config_colorbool(key, value); @@ -71,9 +91,37 @@ void list_config_color_sideband_slots(struct string_list *list, const char *pref list_config_item(list, prefix, keywords[i].keyword); } +static int handle_ansi_color_sequence(struct strbuf *dest, const char *src, int n) +{ + int i; + + /* + * Valid ANSI color sequences are of the form + * + * ESC [ [ [; ]*] m + */ + + if (allow_control_characters != ALLOW_ANSI_COLOR_SEQUENCES || + n < 3 || src[0] != '\x1b' || src[1] != '[') + return 0; + + for (i = 2; i < n; i++) { + if (src[i] == 'm') { + strbuf_add(dest, src, i + 1); + return i; + } + if (!isdigit(src[i]) && src[i] != ';') + break; + } + + return 0; +} + static void strbuf_add_sanitized(struct strbuf *dest, const char *src, int n) { - if (allow_control_characters) { + int i; + + if (allow_control_characters == ALLOW_ALL_CONTROL_CHARACTERS) { strbuf_add(dest, src, n); return; } @@ -82,7 +130,10 @@ static void strbuf_add_sanitized(struct strbuf *dest, const char *src, int n) for (; n && *src; src++, n--) { if (!iscntrl(*src) || *src == '\t' || *src == '\n') strbuf_addch(dest, *src); - else { + else if ((i = handle_ansi_color_sequence(dest, src, n))) { + src += i; + n -= i; + } else { strbuf_addch(dest, '^'); strbuf_addch(dest, 0x40 + *src); } diff --git a/t/t5409-colorize-remote-messages.sh b/t/t5409-colorize-remote-messages.sh index fb31e8525418a1..a755c49a74e634 100755 --- a/t/t5409-colorize-remote-messages.sh +++ b/t/t5409-colorize-remote-messages.sh @@ -100,7 +100,7 @@ test_expect_success 'fallback to color.ui' ' test_expect_success 'disallow (color) control sequences in sideband' ' write_script .git/color-me-surprised <<-\EOF && - printf "error: Have you \\033[31mread\\033[m this?\\n" >&2 + printf "error: Have you \\033[31mread\\033[m this?\\a\\n" >&2 exec "$@" EOF test_config_global uploadPack.packObjectshook ./color-me-surprised && @@ -108,12 +108,24 @@ test_expect_success 'disallow (color) control sequences in sideband' ' git clone --no-local . throw-away 2>stderr && test_decode_color decoded && + test_grep RED decoded && + test_grep "\\^G" stderr && + tr -dc "\\007" actual && + test_must_be_empty actual && + + rm -rf throw-away && + git -c sideband.allowControlCharacters=false \ + clone --no-local . throw-away 2>stderr && + test_decode_color decoded && test_grep ! RED decoded && + test_grep "\\^G" stderr && rm -rf throw-away && git -c sideband.allowControlCharacters clone --no-local . throw-away 2>stderr && test_decode_color decoded && - test_grep RED decoded + test_grep RED decoded && + tr -dc "\\007" actual && + test_file_not_empty actual ' test_done From 26f910108d89b420f979839691175c855f2832bd Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 30 Oct 2024 19:48:46 +0100 Subject: [PATCH 024/259] unix-socket: avoid leak when initialization fails When a Unix socket is initialized, the current directory's path is stored so that the cleanup code can `chdir()` back to where it was before exit. If the path that needs to be stored exceeds the default size of the `sun_path` attribute of `struct sockaddr_un` (which is defined as a 108-sized byte array on Linux), a larger buffer needs to be allocated so that it can hold the path, and it is the responsibility of the `unix_sockaddr_cleanup()` function to release that allocated memory. In Git's CI, this stack allocation is not necessary because the code is checked out to `/home/runner/work/git/git`. Concatenate the path `t/trash directory.t0301-credential-cache/.cache/git/credential/socket` and a terminating NUL, and you end up with 96 bytes, 12 shy of the default `sun_path` size. However, I use worktrees with slightly longer paths: `/home/me/projects/git/yes/i/nest/worktrees/to/organize/them/` is more in line with what I have. When I recently tried to locally reproduce a failure of the `linux-leaks` CI job, this t0301 test failed (where it had not failed in CI). The reason: When `credential-cache` tries to reach its daemon initially by calling `unix_sockaddr_init()`, it is expected that the daemon cannot be reached (the idea is to spin up the daemon in that case and try again). However, when this first call to `unix_sockaddr_init()` fails, the code returns early from the `unix_stream_connect()` function _without_ giving the cleanup code a chance to run, skipping the deallocation of above-mentioned path. The fix is easy: do not return early but instead go directly to the cleanup code. Signed-off-by: Johannes Schindelin --- unix-socket.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/unix-socket.c b/unix-socket.c index 8860203c3f46dc..1fa0cf6c15c721 100644 --- a/unix-socket.c +++ b/unix-socket.c @@ -84,7 +84,7 @@ int unix_stream_connect(const char *path, int disallow_chdir) struct unix_sockaddr_context ctx; if (unix_sockaddr_init(&sa, path, &ctx, disallow_chdir) < 0) - return -1; + goto fail; fd = socket(AF_UNIX, SOCK_STREAM, 0); if (fd < 0) goto fail; From a119a574e73bafe38818cb4d8d4b5d16134e8376 Mon Sep 17 00:00:00 2001 From: Sverre Rabbelier Date: Sun, 24 Jul 2011 15:54:04 +0200 Subject: [PATCH 025/259] t9350: point out that refs are not updated correctly This happens only when the corresponding commits are not exported in the current fast-export run. This can happen either when the relevant commit is already marked, or when the commit is explicitly marked as UNINTERESTING with a negative ref by another argument. This breaks fast-export basec remote helpers. Signed-off-by: Sverre Rabbelier --- t/t9350-fast-export.sh | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/t/t9350-fast-export.sh b/t/t9350-fast-export.sh index 40427883ec6dbf..abffd267e259c4 100755 --- a/t/t9350-fast-export.sh +++ b/t/t9350-fast-export.sh @@ -801,4 +801,15 @@ test_expect_success 'fast-export handles --end-of-options' ' test_cmp expect actual ' +cat > expected << EOF +reset refs/heads/master +from $(git rev-parse master) + +EOF + +test_expect_failure 'refs are updated even if no commits need to be exported' ' + git fast-export master..master > actual && + test_cmp expected actual +' + test_done From dfef2388316853d58f31928e266dd2dbd023026f Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sun, 10 Jul 2022 01:15:08 +0200 Subject: [PATCH 026/259] ci: stop linking the `prove` cache It is not useful because we do not have any persisted directory anymore, not since dropping our Travis CI support. Signed-off-by: Johannes Schindelin --- ci/run-build-and-tests.sh | 5 ----- ci/run-test-slice.sh | 5 ----- 2 files changed, 10 deletions(-) diff --git a/ci/run-build-and-tests.sh b/ci/run-build-and-tests.sh index 1c698467235943..066d4c4c8547e7 100755 --- a/ci/run-build-and-tests.sh +++ b/ci/run-build-and-tests.sh @@ -5,11 +5,6 @@ . ${0%/*}/lib.sh -case "$CI_OS_NAME" in -windows*) cmd //c mklink //j t\\.prove "$(cygpath -aw "$cache_dir/.prove")";; -*) ln -s "$cache_dir/.prove" t/.prove;; -esac - run_tests=t case "$jobname" in diff --git a/ci/run-test-slice.sh b/ci/run-test-slice.sh index e167e646f79e3d..0444c79c023c82 100755 --- a/ci/run-test-slice.sh +++ b/ci/run-test-slice.sh @@ -5,11 +5,6 @@ . ${0%/*}/lib.sh -case "$CI_OS_NAME" in -windows*) cmd //c mklink //j t\\.prove "$(cygpath -aw "$cache_dir/.prove")";; -*) ln -s "$cache_dir/.prove" t/.prove;; -esac - group "Run tests" make --quiet -C t T="$(cd t && ./helper/test-tool path-utils slice-tests "$1" "$2" t[0-9]*.sh | tr '\n' ' ')" || From ca0d213731357f5177286023209b6569d59bbbf5 Mon Sep 17 00:00:00 2001 From: Sverre Rabbelier Date: Sat, 28 Aug 2010 20:49:01 -0500 Subject: [PATCH 027/259] transport-helper: add trailing -- [PT: ensure we add an additional element to the argv array] Signed-off-by: Sverre Rabbelier Signed-off-by: Johannes Schindelin --- transport-helper.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/transport-helper.c b/transport-helper.c index d457b425501a74..fed5d51a2b3d34 100644 --- a/transport-helper.c +++ b/transport-helper.c @@ -499,6 +499,8 @@ static int get_exporter(struct transport *transport, for (size_t i = 0; i < revlist_args->nr; i++) strvec_push(&fastexport->args, revlist_args->items[i].string); + strvec_push(&fastexport->args, "--"); + fastexport->git_cmd = 1; return start_command(fastexport); } From 1c7fffeefc230206d0ebc121405a59c40566c6d8 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Thu, 30 Jan 2025 12:15:41 +0100 Subject: [PATCH 028/259] contrib/buildsystems: drop support for building .vcproj/.vcxproj files Before we had CMake support, the only way to build Git in Visual Studio was via this hacky `generate` script. For a while I tried to fix whenever things got broken, in particular to allow building confidence in embargoed releases by running the CI builds in Azure Pipelines in a private Azure DevOps project. I even carried the patches in Git for Windows with the intention of upstreaming them, eventually. However, it is a lot of work with too little benefit. CMake is much better supported by Visual Studio. So let's drop this hacky script (plus support code). Signed-off-by: Johannes Schindelin --- contrib/buildsystems/Generators.pm | 42 -- contrib/buildsystems/Generators/QMake.pm | 189 ------- contrib/buildsystems/Generators/Vcproj.pm | 579 --------------------- contrib/buildsystems/Generators/Vcxproj.pm | 402 -------------- contrib/buildsystems/engine.pl | 395 -------------- contrib/buildsystems/generate | 29 -- contrib/buildsystems/parse.pl | 228 -------- 7 files changed, 1864 deletions(-) delete mode 100644 contrib/buildsystems/Generators.pm delete mode 100644 contrib/buildsystems/Generators/QMake.pm delete mode 100644 contrib/buildsystems/Generators/Vcproj.pm delete mode 100644 contrib/buildsystems/Generators/Vcxproj.pm delete mode 100755 contrib/buildsystems/engine.pl delete mode 100755 contrib/buildsystems/generate delete mode 100755 contrib/buildsystems/parse.pl diff --git a/contrib/buildsystems/Generators.pm b/contrib/buildsystems/Generators.pm deleted file mode 100644 index aa4cbaa2adacb1..00000000000000 --- a/contrib/buildsystems/Generators.pm +++ /dev/null @@ -1,42 +0,0 @@ -package Generators; -require Exporter; - -use strict; -use File::Basename; -no strict 'refs'; -use vars qw($VERSION @AVAILABLE); - -our $VERSION = '1.00'; -our(@ISA, @EXPORT, @EXPORT_OK, @AVAILABLE); -@ISA = qw(Exporter); - -BEGIN { - local(*D); - my $me = $INC{"Generators.pm"}; - die "Couldn't find myself in \@INC, which is required to load the generators!" if ("$me" eq ""); - $me = dirname($me); - if (opendir(D,"$me/Generators")) { - foreach my $gen (readdir(D)) { - next unless ($gen =~ /\.pm$/); - require "${me}/Generators/$gen"; - $gen =~ s,\.pm,,; - push(@AVAILABLE, $gen); - } - closedir(D); - my $gens = join(', ', @AVAILABLE); - } - - push @EXPORT_OK, qw(available); -} - -sub available { - return @AVAILABLE; -} - -sub generate { - my ($gen, $git_dir, $out_dir, $rel_dir, %build_structure) = @_; - return eval("Generators::${gen}::generate(\$git_dir, \$out_dir, \$rel_dir, \%build_structure)") if grep(/^$gen$/, @AVAILABLE); - die "Generator \"${gen}\" is not available!\nAvailable generators are: @AVAILABLE\n"; -} - -1; diff --git a/contrib/buildsystems/Generators/QMake.pm b/contrib/buildsystems/Generators/QMake.pm deleted file mode 100644 index ff3b657e610575..00000000000000 --- a/contrib/buildsystems/Generators/QMake.pm +++ /dev/null @@ -1,189 +0,0 @@ -package Generators::QMake; -require Exporter; - -use strict; -use vars qw($VERSION); - -our $VERSION = '1.00'; -our(@ISA, @EXPORT, @EXPORT_OK, @AVAILABLE); -@ISA = qw(Exporter); - -BEGIN { - push @EXPORT_OK, qw(generate); -} - -sub generate { - my ($git_dir, $out_dir, $rel_dir, %build_structure) = @_; - - my @libs = @{$build_structure{"LIBS"}}; - foreach (@libs) { - createLibProject($_, $git_dir, $out_dir, $rel_dir, %build_structure); - } - - my @apps = @{$build_structure{"APPS"}}; - foreach (@apps) { - createAppProject($_, $git_dir, $out_dir, $rel_dir, %build_structure); - } - - createGlueProject($git_dir, $out_dir, $rel_dir, %build_structure); - return 0; -} - -sub createLibProject { - my ($libname, $git_dir, $out_dir, $rel_dir, %build_structure) = @_; - print "Generate $libname lib project\n"; - $rel_dir = "../$rel_dir"; - - my $sources = join(" \\\n\t", sort(map("$rel_dir/$_", @{$build_structure{"LIBS_${libname}_SOURCES"}}))); - my $defines = join(" \\\n\t", sort(@{$build_structure{"LIBS_${libname}_DEFINES"}})); - my $includes= join(" \\\n\t", sort(map("$rel_dir/$_", @{$build_structure{"LIBS_${libname}_INCLUDES"}}))); - my $cflags = join(" ", sort(@{$build_structure{"LIBS_${libname}_CFLAGS"}})); - - my $cflags_debug = $cflags; - $cflags_debug =~ s/-MT/-MTd/; - $cflags_debug =~ s/-O.//; - - my $cflags_release = $cflags; - $cflags_release =~ s/-MTd/-MT/; - - my @tmp = @{$build_structure{"LIBS_${libname}_LFLAGS"}}; - my @tmp2 = (); - foreach (@tmp) { - if (/^-LTCG/) { - } elsif (/^-L/) { - $_ =~ s/^-L/-LIBPATH:$rel_dir\//; - } - push(@tmp2, $_); - } - my $lflags = join(" ", sort(@tmp)); - - my $target = $libname; - $target =~ s/\//_/g; - $defines =~ s/-D//g; - $defines =~ s/"/\\\\"/g; - $includes =~ s/-I//g; - mkdir "$target" || die "Could not create the directory $target for lib project!\n"; - open F, ">$target/$target.pro" || die "Could not open $target/$target.pro for writing!\n"; - print F << "EOM"; -TEMPLATE = lib -TARGET = $target -DESTDIR = $rel_dir - -CONFIG -= qt -CONFIG += static - -QMAKE_CFLAGS = -QMAKE_CFLAGS_RELEASE = $cflags_release -QMAKE_CFLAGS_DEBUG = $cflags_debug -QMAKE_LIBFLAGS = $lflags - -DEFINES += \\ - $defines - -INCLUDEPATH += \\ - $includes - -SOURCES += \\ - $sources -EOM - close F; -} - -sub createAppProject { - my ($appname, $git_dir, $out_dir, $rel_dir, %build_structure) = @_; - print "Generate $appname app project\n"; - $rel_dir = "../$rel_dir"; - - my $sources = join(" \\\n\t", sort(map("$rel_dir/$_", @{$build_structure{"APPS_${appname}_SOURCES"}}))); - my $defines = join(" \\\n\t", sort(@{$build_structure{"APPS_${appname}_DEFINES"}})); - my $includes= join(" \\\n\t", sort(map("$rel_dir/$_", @{$build_structure{"APPS_${appname}_INCLUDES"}}))); - my $cflags = join(" ", sort(@{$build_structure{"APPS_${appname}_CFLAGS"}})); - - my $cflags_debug = $cflags; - $cflags_debug =~ s/-MT/-MTd/; - $cflags_debug =~ s/-O.//; - - my $cflags_release = $cflags; - $cflags_release =~ s/-MTd/-MT/; - - my $libs; - foreach (sort(@{$build_structure{"APPS_${appname}_LIBS"}})) { - $_ =~ s/\//_/g; - $libs .= " $_"; - } - my @tmp = @{$build_structure{"APPS_${appname}_LFLAGS"}}; - my @tmp2 = (); - foreach (@tmp) { - # next if ($_ eq "-NODEFAULTLIB:MSVCRT.lib"); - if (/^-LTCG/) { - } elsif (/^-L/) { - $_ =~ s/^-L/-LIBPATH:$rel_dir\//; - } - push(@tmp2, $_); - } - my $lflags = join(" ", sort(@tmp)); - - my $target = $appname; - $target =~ s/\.exe//; - $target =~ s/\//_/g; - $defines =~ s/-D//g; - $defines =~ s/"/\\\\"/g; - $includes =~ s/-I//g; - mkdir "$target" || die "Could not create the directory $target for app project!\n"; - open F, ">$target/$target.pro" || die "Could not open $target/$target.pro for writing!\n"; - print F << "EOM"; -TEMPLATE = app -TARGET = $target -DESTDIR = $rel_dir - -CONFIG -= qt embed_manifest_exe -CONFIG += console - -QMAKE_CFLAGS = -QMAKE_CFLAGS_RELEASE = $cflags_release -QMAKE_CFLAGS_DEBUG = $cflags_debug -QMAKE_LFLAGS = $lflags -LIBS = $libs - -DEFINES += \\ - $defines - -INCLUDEPATH += \\ - $includes - -win32:QMAKE_LFLAGS += -LIBPATH:$rel_dir -else: QMAKE_LFLAGS += -L$rel_dir - -SOURCES += \\ - $sources -EOM - close F; -} - -sub createGlueProject { - my ($git_dir, $out_dir, $rel_dir, %build_structure) = @_; - my $libs = join(" \\ \n", map("\t$_|$_.pro", @{$build_structure{"LIBS"}})); - my $apps = join(" \\ \n", map("\t$_|$_.pro", @{$build_structure{"APPS"}})); - $libs =~ s/\.a//g; - $libs =~ s/\//_/g; - $libs =~ s/\|/\//g; - $apps =~ s/\.exe//g; - $apps =~ s/\//_/g; - $apps =~ s/\|/\//g; - - my $filename = $out_dir; - $filename =~ s/.*\/([^\/]+)$/$1/; - $filename =~ s/\/$//; - print "Generate glue project $filename.pro\n"; - open F, ">$filename.pro" || die "Could not open $filename.pro for writing!\n"; - print F << "EOM"; -TEMPLATE = subdirs -CONFIG += ordered -SUBDIRS += \\ -$libs \\ -$apps -EOM - close F; -} - -1; diff --git a/contrib/buildsystems/Generators/Vcproj.pm b/contrib/buildsystems/Generators/Vcproj.pm deleted file mode 100644 index 737647e76afd42..00000000000000 --- a/contrib/buildsystems/Generators/Vcproj.pm +++ /dev/null @@ -1,579 +0,0 @@ -package Generators::Vcproj; -require Exporter; - -use strict; -use vars qw($VERSION); -use Digest::SHA qw(sha256_hex); - -our $VERSION = '1.00'; -our(@ISA, @EXPORT, @EXPORT_OK, @AVAILABLE); -@ISA = qw(Exporter); - -BEGIN { - push @EXPORT_OK, qw(generate); -} - -sub generate_guid ($) { - my $hex = sha256_hex($_[0]); - $hex =~ s/^(.{8})(.{4})(.{4})(.{4})(.{12}).*/{$1-$2-$3-$4-$5}/; - $hex =~ tr/a-z/A-Z/; - return $hex; -} - -sub generate { - my ($git_dir, $out_dir, $rel_dir, %build_structure) = @_; - my @libs = @{$build_structure{"LIBS"}}; - foreach (@libs) { - createLibProject($_, $git_dir, $out_dir, $rel_dir, \%build_structure); - } - - my @apps = @{$build_structure{"APPS"}}; - foreach (@apps) { - createAppProject($_, $git_dir, $out_dir, $rel_dir, \%build_structure); - } - - createGlueProject($git_dir, $out_dir, $rel_dir, %build_structure); - return 0; -} - -sub createLibProject { - my ($libname, $git_dir, $out_dir, $rel_dir, $build_structure) = @_; - print "Generate $libname vcproj lib project\n"; - $rel_dir = "..\\$rel_dir"; - $rel_dir =~ s/\//\\/g; - - my $target = $libname; - $target =~ s/\//_/g; - $target =~ s/\.a//; - - my $uuid = generate_guid($libname); - $$build_structure{"LIBS_${target}_GUID"} = $uuid; - - my @srcs = sort(map("$rel_dir\\$_", @{$$build_structure{"LIBS_${libname}_SOURCES"}})); - my @sources; - foreach (@srcs) { - $_ =~ s/\//\\/g; - push(@sources, $_); - } - my $defines = join(",", sort(@{$$build_structure{"LIBS_${libname}_DEFINES"}})); - my $includes= join(";", sort(map(""$rel_dir\\$_"", @{$$build_structure{"LIBS_${libname}_INCLUDES"}}))); - my $cflags = join(" ", sort(@{$$build_structure{"LIBS_${libname}_CFLAGS"}})); - $cflags =~ s/\"/"/g; - $cflags =~ s//>/g; - - my $cflags_debug = $cflags; - $cflags_debug =~ s/-MT/-MTd/; - $cflags_debug =~ s/-O.//; - - my $cflags_release = $cflags; - $cflags_release =~ s/-MTd/-MT/; - - my @tmp = @{$$build_structure{"LIBS_${libname}_LFLAGS"}}; - my @tmp2 = (); - foreach (@tmp) { - if (/^-LTCG/) { - } elsif (/^-L/) { - $_ =~ s/^-L/-LIBPATH:$rel_dir\//; - } - push(@tmp2, $_); - } - my $lflags = join(" ", sort(@tmp)); - - $defines =~ s/-D//g; - $defines =~ s/\"/\\"/g; - $defines =~ s//>/g; - $defines =~ s/\'//g; - $includes =~ s/-I//g; - mkdir "$target" || die "Could not create the directory $target for lib project!\n"; - open F, ">$target/$target.vcproj" || die "Could not open $target/$target.pro for writing!\n"; - binmode F, ":crlf"; - print F << "EOM"; - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -EOM - foreach(@sources) { - print F << "EOM"; - -EOM - } - print F << "EOM"; - - - - - -EOM - close F; -} - -sub createAppProject { - my ($appname, $git_dir, $out_dir, $rel_dir, $build_structure) = @_; - print "Generate $appname vcproj app project\n"; - $rel_dir = "..\\$rel_dir"; - $rel_dir =~ s/\//\\/g; - - my $target = $appname; - $target =~ s/\//_/g; - $target =~ s/\.exe//; - - my $uuid = generate_guid($appname); - $$build_structure{"APPS_${target}_GUID"} = $uuid; - - my @srcs = sort(map("$rel_dir\\$_", @{$$build_structure{"APPS_${appname}_SOURCES"}})); - my @sources; - foreach (@srcs) { - $_ =~ s/\//\\/g; - push(@sources, $_); - } - my $defines = join(",", sort(@{$$build_structure{"APPS_${appname}_DEFINES"}})); - my $includes= join(";", sort(map(""$rel_dir\\$_"", @{$$build_structure{"APPS_${appname}_INCLUDES"}}))); - my $cflags = join(" ", sort(@{$$build_structure{"APPS_${appname}_CFLAGS"}})); - $cflags =~ s/\"/"/g; - $cflags =~ s//>/g; - - my $cflags_debug = $cflags; - $cflags_debug =~ s/-MT/-MTd/; - $cflags_debug =~ s/-O.//; - - my $cflags_release = $cflags; - $cflags_release =~ s/-MTd/-MT/; - - my $libs; - foreach (sort(@{$$build_structure{"APPS_${appname}_LIBS"}})) { - $_ =~ s/\//_/g; - $libs .= " $_"; - } - my @tmp = @{$$build_structure{"APPS_${appname}_LFLAGS"}}; - my @tmp2 = (); - foreach (@tmp) { - if (/^-LTCG/) { - } elsif (/^-L/) { - $_ =~ s/^-L/-LIBPATH:$rel_dir\//; - } - push(@tmp2, $_); - } - my $lflags = join(" ", sort(@tmp)) . " -LIBPATH:$rel_dir"; - - $defines =~ s/-D//g; - $defines =~ s/\"/\\"/g; - $defines =~ s//>/g; - $defines =~ s/\'//g; - $defines =~ s/\\\\/\\/g; - $includes =~ s/-I//g; - mkdir "$target" || die "Could not create the directory $target for lib project!\n"; - open F, ">$target/$target.vcproj" || die "Could not open $target/$target.pro for writing!\n"; - binmode F, ":crlf"; - print F << "EOM"; - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -EOM - foreach(@sources) { - print F << "EOM"; - -EOM - } - print F << "EOM"; - - - - - -EOM - close F; -} - -sub createGlueProject { - my ($git_dir, $out_dir, $rel_dir, %build_structure) = @_; - print "Generate solutions file\n"; - $rel_dir = "..\\$rel_dir"; - $rel_dir =~ s/\//\\/g; - my $SLN_HEAD = "Microsoft Visual Studio Solution File, Format Version 10.00\n# Visual Studio 2008\n"; - my $SLN_PRE = "Project(\"{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}\") = "; - my $SLN_POST = "\nEndProject\n"; - - my @libs = @{$build_structure{"LIBS"}}; - my @tmp; - foreach (@libs) { - $_ =~ s/\//_/g; - $_ =~ s/\.a//; - push(@tmp, $_); - } - @libs = @tmp; - - my @apps = @{$build_structure{"APPS"}}; - @tmp = (); - foreach (@apps) { - $_ =~ s/\//_/g; - $_ =~ s/\.exe//; - if ($_ eq "git" ) { - unshift(@tmp, $_); - } else { - push(@tmp, $_); - } - } - @apps = @tmp; - - open F, ">git.sln" || die "Could not open git.sln for writing!\n"; - binmode F, ":crlf"; - print F "$SLN_HEAD"; - - my $uuid_libgit = $build_structure{"LIBS_libgit_GUID"}; - my $uuid_xdiff_lib = $build_structure{"LIBS_xdiff_lib_GUID"}; - foreach (@apps) { - my $appname = $_; - my $uuid = $build_structure{"APPS_${appname}_GUID"}; - print F "$SLN_PRE"; - print F "\"${appname}\", \"${appname}\\${appname}.vcproj\", \"${uuid}\"\n"; - print F " ProjectSection(ProjectDependencies) = postProject\n"; - print F " ${uuid_libgit} = ${uuid_libgit}\n"; - print F " ${uuid_xdiff_lib} = ${uuid_xdiff_lib}\n"; - print F " EndProjectSection"; - print F "$SLN_POST"; - } - foreach (@libs) { - my $libname = $_; - my $uuid = $build_structure{"LIBS_${libname}_GUID"}; - print F "$SLN_PRE"; - print F "\"${libname}\", \"${libname}\\${libname}.vcproj\", \"${uuid}\""; - print F "$SLN_POST"; - } - - print F << "EOM"; -Global - GlobalSection(SolutionConfigurationPlatforms) = preSolution - Debug|Win32 = Debug|Win32 - Release|Win32 = Release|Win32 - EndGlobalSection -EOM - print F << "EOM"; - GlobalSection(ProjectConfigurationPlatforms) = postSolution -EOM - foreach (@apps) { - my $appname = $_; - my $uuid = $build_structure{"APPS_${appname}_GUID"}; - print F "\t\t${uuid}.Debug|Win32.ActiveCfg = Debug|Win32\n"; - print F "\t\t${uuid}.Debug|Win32.Build.0 = Debug|Win32\n"; - print F "\t\t${uuid}.Release|Win32.ActiveCfg = Release|Win32\n"; - print F "\t\t${uuid}.Release|Win32.Build.0 = Release|Win32\n"; - } - foreach (@libs) { - my $libname = $_; - my $uuid = $build_structure{"LIBS_${libname}_GUID"}; - print F "\t\t${uuid}.Debug|Win32.ActiveCfg = Debug|Win32\n"; - print F "\t\t${uuid}.Debug|Win32.Build.0 = Debug|Win32\n"; - print F "\t\t${uuid}.Release|Win32.ActiveCfg = Release|Win32\n"; - print F "\t\t${uuid}.Release|Win32.Build.0 = Release|Win32\n"; - } - - print F << "EOM"; - EndGlobalSection -EndGlobal -EOM - close F; -} - -1; diff --git a/contrib/buildsystems/Generators/Vcxproj.pm b/contrib/buildsystems/Generators/Vcxproj.pm deleted file mode 100644 index b2e68a16715e39..00000000000000 --- a/contrib/buildsystems/Generators/Vcxproj.pm +++ /dev/null @@ -1,402 +0,0 @@ -package Generators::Vcxproj; -require Exporter; - -use strict; -use vars qw($VERSION); -use Digest::SHA qw(sha256_hex); - -our $VERSION = '1.00'; -our(@ISA, @EXPORT, @EXPORT_OK, @AVAILABLE); -@ISA = qw(Exporter); - -BEGIN { - push @EXPORT_OK, qw(generate); -} - -sub generate_guid ($) { - my $hex = sha256_hex($_[0]); - $hex =~ s/^(.{8})(.{4})(.{4})(.{4})(.{12}).*/{$1-$2-$3-$4-$5}/; - $hex =~ tr/a-z/A-Z/; - return $hex; -} - -sub generate { - my ($git_dir, $out_dir, $rel_dir, %build_structure) = @_; - my @libs = @{$build_structure{"LIBS"}}; - foreach (@libs) { - createProject($_, $git_dir, $out_dir, $rel_dir, \%build_structure, 1); - } - - my @apps = @{$build_structure{"APPS"}}; - foreach (@apps) { - createProject($_, $git_dir, $out_dir, $rel_dir, \%build_structure, 0); - } - - createGlueProject($git_dir, $out_dir, $rel_dir, %build_structure); - return 0; -} - -sub createProject { - my ($name, $git_dir, $out_dir, $rel_dir, $build_structure, $static_library) = @_; - my $label = $static_library ? "lib" : "app"; - my $prefix = $static_library ? "LIBS_" : "APPS_"; - my $config_type = $static_library ? "StaticLibrary" : "Application"; - print "Generate $name vcxproj $label project\n"; - my $cdup = $name; - $cdup =~ s/[^\/]+/../g; - $cdup =~ s/\//\\/g; - $rel_dir = $rel_dir eq "." ? $cdup : "$cdup\\$rel_dir"; - $rel_dir =~ s/\//\\/g; - - my $target = $name; - if ($static_library) { - $target =~ s/\.a//; - } else { - $target =~ s/\.exe//; - } - - my $uuid = generate_guid($name); - $$build_structure{"$prefix${target}_GUID"} = $uuid; - my $vcxproj = $target; - $vcxproj =~ s/(.*\/)?(.*)/$&\/$2.vcxproj/; - $vcxproj =~ s/([^\/]*)(\/lib)\/(lib.vcxproj)/$1$2\/$1_$3/; - $$build_structure{"$prefix${target}_VCXPROJ"} = $vcxproj; - - my @srcs = sort(map("$rel_dir\\$_", @{$$build_structure{"$prefix${name}_SOURCES"}})); - my @sources; - foreach (@srcs) { - $_ =~ s/\//\\/g; - push(@sources, $_); - } - my $defines = join(";", sort(@{$$build_structure{"$prefix${name}_DEFINES"}})); - my $includes= join(";", sort(map { s/^-I//; s/\//\\/g; File::Spec->file_name_is_absolute($_) ? $_ : "$rel_dir\\$_" } @{$$build_structure{"$prefix${name}_INCLUDES"}})); - my $cflags = join(" ", sort(map { s/^-[GLMOWZ].*//; s/.* .*/"$&"/; $_; } @{$$build_structure{"$prefix${name}_CFLAGS"}})); - $cflags =~ s//>/g; - - my $libs_release = "\n "; - my $libs_debug = "\n "; - if (!$static_library && $name ne 'headless-git') { - $libs_release = join(";", sort(grep /^(?!libgit\.lib|xdiff\/lib\.lib|vcs-svn\/lib\.lib|reftable\/libreftable\.lib)/, @{$$build_structure{"$prefix${name}_LIBS"}})); - $libs_debug = $libs_release; - $libs_debug =~ s/zlib\.lib/zlibd\.lib/g; - $libs_debug =~ s/libexpat\.lib/libexpatd\.lib/g; - $libs_debug =~ s/libcurl\.lib/libcurl-d\.lib/g; - } - - $defines =~ s/-D//g; - $defines =~ s//>/g; - $defines =~ s/\'//g; - - die "Could not create the directory $target for $label project!\n" unless (-d "$target" || mkdir "$target"); - - open F, ">$vcxproj" or die "Could not open $vcxproj for writing!\n"; - binmode F, ":crlf :utf8"; - print F chr(0xFEFF); - print F << "EOM"; - - - - - Debug - Win32 - - - Release - Win32 - - - Debug - x64 - - - Release - x64 - - - - $uuid - Win32Proj - x86-windows - x64-windows - $cdup\\compat\\vcbuild\\vcpkg\\installed\\\$(VCPKGArch) - \$(VCPKGArchDirectory)\\debug\\bin - \$(VCPKGArchDirectory)\\debug\\lib - \$(VCPKGArchDirectory)\\bin - \$(VCPKGArchDirectory)\\lib - \$(VCPKGArchDirectory)\\include - $libs_debug - $libs_release - - - - true - true - - - false - true - - - $config_type - v140 - - ..\\ - - - - - - - - - - - - - false - true - - - - $cflags %(AdditionalOptions) - $cdup;$cdup\\compat;$cdup\\compat\\regex;$cdup\\compat\\win32;$cdup\\compat\\poll;$cdup\\compat\\vcbuild\\include;\$(VCPKGIncludeDirectory);%(AdditionalIncludeDirectories) - - OnlyExplicitInline - - ProgramDatabase - - - true - - - \$(VCPKGLibDirectory);%(AdditionalLibraryDirectories) - \$(VCPKGLibs);\$(AdditionalDependencies) - invalidcontinue.obj %(AdditionalOptions) - wmainCRTStartup - $cdup\\compat\\win32\\git.manifest - Console - -EOM - if ($target eq 'libgit') { - print F << "EOM"; - - Initialize VCPKG - del "$cdup\\compat\\vcbuild\\vcpkg" - call "$cdup\\compat\\vcbuild\\vcpkg_install.bat" - -EOM - } - print F << "EOM"; - - - - MachineX86 - - - - - Disabled - WIN32;_DEBUG;$defines;%(PreprocessorDefinitions) - MultiThreadedDebugDLL - - - true - - - - - MaxSpeed - true - WIN32;NDEBUG;$defines;%(PreprocessorDefinitions) - MultiThreadedDLL - true - Speed - - - true - true - true - - - -EOM - foreach(@sources) { - print F << "EOM"; - -EOM - } - print F << "EOM"; - -EOM - if ((!$static_library || $target =~ 'vcs-svn' || $target =~ 'xdiff') && !($name =~ /headless-git/)) { - my $uuid_libgit = $$build_structure{"LIBS_libgit_GUID"}; - my $uuid_libreftable = $$build_structure{"LIBS_reftable/libreftable_GUID"}; - my $uuid_xdiff_lib = $$build_structure{"LIBS_xdiff/lib_GUID"}; - - print F << "EOM"; - - - $uuid_libgit - false - -EOM - if (!($name =~ /xdiff|libreftable/)) { - print F << "EOM"; - - $uuid_libreftable - false - -EOM - } - if (!($name =~ 'xdiff')) { - print F << "EOM"; - - $uuid_xdiff_lib - false - -EOM - } - if ($name =~ /(test-(line-buffer|svn-fe)|^git-remote-testsvn)\.exe$/) { - my $uuid_vcs_svn_lib = $$build_structure{"LIBS_vcs-svn/lib_GUID"}; - print F << "EOM"; - - $uuid_vcs_svn_lib - false - -EOM - } - print F << "EOM"; - -EOM - } - print F << "EOM"; - -EOM - if (!$static_library) { - print F << "EOM"; - - - - - - - -EOM - } - if ($target eq 'git') { - print F " \n"; - } - if ($target eq 'git-remote-http') { - print F " \n"; - } - print F << "EOM"; - -EOM - close F; -} - -sub createGlueProject { - my ($git_dir, $out_dir, $rel_dir, %build_structure) = @_; - print "Generate solutions file\n"; - $rel_dir = "..\\$rel_dir"; - $rel_dir =~ s/\//\\/g; - my $SLN_HEAD = "Microsoft Visual Studio Solution File, Format Version 11.00\n# Visual Studio 2010\n"; - my $SLN_PRE = "Project(\"{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}\") = "; - my $SLN_POST = "\nEndProject\n"; - - my @libs = @{$build_structure{"LIBS"}}; - my @tmp; - foreach (@libs) { - $_ =~ s/\.a//; - push(@tmp, $_); - } - @libs = @tmp; - - my @apps = @{$build_structure{"APPS"}}; - @tmp = (); - foreach (@apps) { - $_ =~ s/\.exe//; - if ($_ eq "git" ) { - unshift(@tmp, $_); - } else { - push(@tmp, $_); - } - } - @apps = @tmp; - - open F, ">git.sln" || die "Could not open git.sln for writing!\n"; - binmode F, ":crlf :utf8"; - print F chr(0xFEFF); - print F "$SLN_HEAD"; - - foreach (@apps) { - my $appname = $_; - my $uuid = $build_structure{"APPS_${appname}_GUID"}; - print F "$SLN_PRE"; - my $vcxproj = $build_structure{"APPS_${appname}_VCXPROJ"}; - $vcxproj =~ s/\//\\/g; - $appname =~ s/.*\///; - print F "\"${appname}\", \"${vcxproj}\", \"${uuid}\""; - print F "$SLN_POST"; - } - foreach (@libs) { - my $libname = $_; - my $uuid = $build_structure{"LIBS_${libname}_GUID"}; - print F "$SLN_PRE"; - my $vcxproj = $build_structure{"LIBS_${libname}_VCXPROJ"}; - $vcxproj =~ s/\//\\/g; - $libname =~ s/\//_/g; - print F "\"${libname}\", \"${vcxproj}\", \"${uuid}\""; - print F "$SLN_POST"; - } - - print F << "EOM"; -Global - GlobalSection(SolutionConfigurationPlatforms) = preSolution - Debug|x64 = Debug|x64 - Debug|x86 = Debug|x86 - Release|x64 = Release|x64 - Release|x86 = Release|x86 - EndGlobalSection -EOM - print F << "EOM"; - GlobalSection(ProjectConfigurationPlatforms) = postSolution -EOM - foreach (@apps) { - my $appname = $_; - my $uuid = $build_structure{"APPS_${appname}_GUID"}; - print F "\t\t${uuid}.Debug|x64.ActiveCfg = Debug|x64\n"; - print F "\t\t${uuid}.Debug|x64.Build.0 = Debug|x64\n"; - print F "\t\t${uuid}.Debug|x86.ActiveCfg = Debug|Win32\n"; - print F "\t\t${uuid}.Debug|x86.Build.0 = Debug|Win32\n"; - print F "\t\t${uuid}.Release|x64.ActiveCfg = Release|x64\n"; - print F "\t\t${uuid}.Release|x64.Build.0 = Release|x64\n"; - print F "\t\t${uuid}.Release|x86.ActiveCfg = Release|Win32\n"; - print F "\t\t${uuid}.Release|x86.Build.0 = Release|Win32\n"; - } - foreach (@libs) { - my $libname = $_; - my $uuid = $build_structure{"LIBS_${libname}_GUID"}; - print F "\t\t${uuid}.Debug|x64.ActiveCfg = Debug|x64\n"; - print F "\t\t${uuid}.Debug|x64.Build.0 = Debug|x64\n"; - print F "\t\t${uuid}.Debug|x86.ActiveCfg = Debug|Win32\n"; - print F "\t\t${uuid}.Debug|x86.Build.0 = Debug|Win32\n"; - print F "\t\t${uuid}.Release|x64.ActiveCfg = Release|x64\n"; - print F "\t\t${uuid}.Release|x64.Build.0 = Release|x64\n"; - print F "\t\t${uuid}.Release|x86.ActiveCfg = Release|Win32\n"; - print F "\t\t${uuid}.Release|x86.Build.0 = Release|Win32\n"; - } - - print F << "EOM"; - EndGlobalSection - GlobalSection(SolutionProperties) = preSolution - HideSolutionNode = FALSE - EndGlobalSection -EndGlobal -EOM - close F; -} - -1; diff --git a/contrib/buildsystems/engine.pl b/contrib/buildsystems/engine.pl deleted file mode 100755 index 069be7e4befcd7..00000000000000 --- a/contrib/buildsystems/engine.pl +++ /dev/null @@ -1,395 +0,0 @@ -#!/usr/bin/perl -w -###################################################################### -# Do not call this script directly! -# -# The generate script ensures that @INC is correct before the engine -# is executed. -# -# Copyright (C) 2009 Marius Storm-Olsen -###################################################################### -use strict; -use File::Basename; -use File::Spec; -use Cwd; -use Generators; -use Text::ParseWords; - -my (%build_structure, %compile_options, @makedry); -my $out_dir = getcwd(); -my $git_dir = $out_dir; -$git_dir =~ s=\\=/=g; -$git_dir = dirname($git_dir) while (!-e "$git_dir/git.c" && "$git_dir" ne ""); -die "Couldn't find Git repo" if ("$git_dir" eq ""); - -my @gens = Generators::available(); -my $gen = "Vcproj"; - -sub showUsage -{ - my $genlist = join(', ', @gens); - print << "EOM"; -generate usage: - -g --gen Specify the buildsystem generator (default: $gen) - Available: $genlist - -o --out Specify output directory generation (default: .) - --make-out Write the output of GNU Make into a file - -i --in Specify input file, instead of running GNU Make - -h,-? --help This help -EOM - exit 0; -} - -# Parse command-line options -my $make_out; -while (@ARGV) { - my $arg = shift @ARGV; - if ("$arg" eq "-h" || "$arg" eq "--help" || "$arg" eq "-?") { - showUsage(); - exit(0); - } elsif("$arg" eq "--out" || "$arg" eq "-o") { - $out_dir = shift @ARGV; - } elsif("$arg" eq "--make-out") { - $make_out = shift @ARGV; - } elsif("$arg" eq "--gen" || "$arg" eq "-g") { - $gen = shift @ARGV; - } elsif("$arg" eq "--in" || "$arg" eq "-i") { - my $infile = shift @ARGV; - open(F, "<$infile") || die "Couldn't open file $infile"; - @makedry = ; - close(F); - } else { - die "Unknown option: " . $arg; - } -} - -# NOT using File::Spec->rel2abs($path, $base) here, as -# it fails badly for me in the msysgit environment -$git_dir = File::Spec->rel2abs($git_dir); -$out_dir = File::Spec->rel2abs($out_dir); -my $rel_dir = makeOutRel2Git($git_dir, $out_dir); - -# Print some information so the user feels informed -print << "EOM"; ------ -Generator: $gen -Git dir: $git_dir -Out dir: $out_dir ------ -Running GNU Make to figure out build structure... -EOM - -# Pipe a make --dry-run into a variable, if not already loaded from file -# Capture the make dry stderr to file for review (will be empty for a release build). - -my $ErrsFile = "msvc-build-makedryerrors.txt"; -@makedry = `make -C $git_dir -n MSVC=1 SKIP_VCPKG=1 V=1 2>$ErrsFile` -if !@makedry; -# test for an empty Errors file and remove it -unlink $ErrsFile if -f -z $ErrsFile; - -if (defined $make_out) { - open OUT, ">" . $make_out; - print OUT @makedry; - close OUT; -} - -# Parse the make output into usable info -parseMakeOutput(); - -# Finally, ask the generator to start generating.. -Generators::generate($gen, $git_dir, $out_dir, $rel_dir, %build_structure); - -# main flow ends here -# ------------------------------------------------------------------------------------------------- - - -# 1) path: /foo/bar/baz 2) path: /foo/bar/baz 3) path: /foo/bar/baz -# base: /foo/bar/baz/temp base: /foo/bar base: /tmp -# rel: .. rel: baz rel: ../foo/bar/baz -sub makeOutRel2Git -{ - my ($path, $base) = @_; - my $rel; - if ("$path" eq "$base") { - return "."; - } elsif ($base =~ /^$path/) { - # case 1 - my $tmp = $base; - $tmp =~ s/^$path//; - foreach (split('/', $tmp)) { - $rel .= "../" if ("$_" ne ""); - } - } elsif ($path =~ /^$base/) { - # case 2 - $rel = $path; - $rel =~ s/^$base//; - $rel = "./$rel"; - } else { - my $tmp = $base; - foreach (split('/', $tmp)) { - $rel .= "../" if ("$_" ne ""); - } - $rel .= $path; - } - $rel =~ s/\/\//\//g; # simplify - $rel =~ s/\/$//; # don't end with / - return $rel; -} - -sub parseMakeOutput -{ - print "Parsing GNU Make output to figure out build structure...\n"; - my $line = 0; - while (my $text = shift @makedry) { - my $ate_next; - do { - $ate_next = 0; - $line++; - chomp $text; - chop $text if ($text =~ /\r$/); - if ($text =~ /\\$/) { - $text =~ s/\\$//; - $text .= shift @makedry; - $ate_next = 1; - } - } while($ate_next); - - if ($text =~ /^test /) { - # options to test (eg -o) may be mistaken for linker options - next; - } - - if ($text =~ /^(mkdir|msgfmt) /) { - # options to the Portable Object translations - # the line "mkdir ... && msgfmt ..." contains no linker options - next; - } - - if($text =~ / -c /) { - # compilation - handleCompileLine($text, $line); - - } elsif ($text =~ / -o /) { - # linking executable - handleLinkLine($text, $line); - - } elsif ($text =~ /\.o / && $text =~ /\.a /) { - # libifying - handleLibLine($text, $line); -# -# } elsif ($text =~ /^cp /) { -# # copy file around -# -# } elsif ($text =~ /^rm -f /) { -# # shell command -# -# } elsif ($text =~ /^make[ \[]/) { -# # make output -# -# } elsif ($text =~ /^echo /) { -# # echo to file -# -# } elsif ($text =~ /^if /) { -# # shell conditional -# -# } elsif ($text =~ /^tclsh /) { -# # translation stuff -# -# } elsif ($text =~ /^umask /) { -# # handling boilerplates -# -# } elsif ($text =~ /\$\(\:\)/) { -# # ignore -# -# } elsif ($text =~ /^FLAGS=/) { -# # flags check for dependencies -# -# } elsif ($text =~ /^'\/usr\/bin\/perl' -MError -e/) { -# # perl commands for copying files -# -# } elsif ($text =~ /generate-cmdlist\.sh/) { -# # command for generating list of commands -# -# } elsif ($text =~ /new locations or Tcl/) { -# # command for detecting Tcl/Tk changes -# -# } elsif ($text =~ /mkdir -p/) { -# # command creating path -# -# } elsif ($text =~ /: no custom templates yet/) { -# # whatever -# -# } else { -# print "Unhandled (line: $line): $text\n"; - } - } - -# use Data::Dumper; -# print "Parsed build structure:\n"; -# print Dumper(%build_structure); -} - -# variables for the compilation part of each step -my (@defines, @incpaths, @cflags, @sources); - -sub clearCompileStep -{ - @defines = (); - @incpaths = (); - @cflags = (); - @sources = (); -} - -sub removeDuplicates -{ - my (%dupHash, $entry); - %dupHash = map { $_, 1 } @defines; - @defines = keys %dupHash; - - %dupHash = map { $_, 1 } @incpaths; - @incpaths = keys %dupHash; - - %dupHash = map { $_, 1 } @cflags; - @cflags = keys %dupHash; -} - -sub handleCompileLine -{ - my ($line, $lineno) = @_; - my @parts = shellwords($line); - my $sourcefile; - shift(@parts); # ignore cmd - while (my $part = shift @parts) { - if ("$part" eq "-o") { - # ignore object file - shift @parts; - } elsif ("$part" eq "-c") { - # ignore compile flag - } elsif ("$part" eq "-c") { - } elsif ($part =~ /^.?-I/) { - push(@incpaths, $part); - } elsif ($part =~ /^.?-D/) { - push(@defines, $part); - } elsif ($part =~ /^-/) { - push(@cflags, $part); - } elsif ($part =~ /\.(c|cc|cpp)$/) { - $sourcefile = $part; - } else { - die "Unhandled compiler option @ line $lineno: $part"; - } - } - @{$compile_options{"${sourcefile}_CFLAGS"}} = @cflags; - @{$compile_options{"${sourcefile}_DEFINES"}} = @defines; - @{$compile_options{"${sourcefile}_INCPATHS"}} = @incpaths; - clearCompileStep(); -} - -sub handleLibLine -{ - my ($line, $lineno) = @_; - my (@objfiles, @lflags, $libout, $part); - # kill cmd and rm 'prefix' - $line =~ s/^rm -f .* && .* rcs //; - my @parts = shellwords($line); - while ($part = shift @parts) { - if ($part =~ /^-/) { - push(@lflags, $part); - } elsif ($part =~ /\.(o|obj)$/) { - push(@objfiles, $part); - } elsif ($part =~ /\.(a|lib)$/) { - $libout = $part; - $libout =~ s/\.a$//; - } else { - die "Unhandled lib option @ line $lineno: $part"; - } - } -# print "LibOut: '$libout'\nLFlags: @lflags\nOfiles: @objfiles\n"; -# exit(1); - foreach (@objfiles) { - my $sourcefile = $_; - $sourcefile =~ s/\.o$/.c/; - push(@sources, $sourcefile); - push(@cflags, @{$compile_options{"${sourcefile}_CFLAGS"}}); - push(@defines, @{$compile_options{"${sourcefile}_DEFINES"}}); - push(@incpaths, @{$compile_options{"${sourcefile}_INCPATHS"}}); - } - removeDuplicates(); - - push(@{$build_structure{"LIBS"}}, $libout); - @{$build_structure{"LIBS_${libout}"}} = ("_DEFINES", "_INCLUDES", "_CFLAGS", "_SOURCES", - "_OBJECTS"); - @{$build_structure{"LIBS_${libout}_DEFINES"}} = @defines; - @{$build_structure{"LIBS_${libout}_INCLUDES"}} = @incpaths; - @{$build_structure{"LIBS_${libout}_CFLAGS"}} = @cflags; - @{$build_structure{"LIBS_${libout}_LFLAGS"}} = @lflags; - @{$build_structure{"LIBS_${libout}_SOURCES"}} = @sources; - @{$build_structure{"LIBS_${libout}_OBJECTS"}} = @objfiles; - clearCompileStep(); -} - -sub handleLinkLine -{ - my ($line, $lineno) = @_; - my (@objfiles, @lflags, @libs, $appout, $part); - my @parts = shellwords($line); - shift(@parts); # ignore cmd - while ($part = shift @parts) { - if ($part =~ /^-IGNORE/) { - push(@lflags, $part); - } elsif ($part =~ /^-[GRIMDO]/) { - # eat compiler flags - } elsif ("$part" eq "-o") { - $appout = shift @parts; - } elsif ("$part" eq "-lz") { - push(@libs, "zlib.lib"); - } elsif ("$part" eq "-lcrypto") { - push(@libs, "libcrypto.lib"); - } elsif ("$part" eq "-lssl") { - push(@libs, "libssl.lib"); - } elsif ("$part" eq "-lcurl") { - push(@libs, "libcurl.lib"); - } elsif ("$part" eq "-lexpat") { - push(@libs, "libexpat.lib"); - } elsif ("$part" eq "-liconv") { - push(@libs, "iconv.lib"); - } elsif ($part =~ /^[-\/]/) { - push(@lflags, $part); - } elsif ($part =~ /\.(a|lib)$/) { - $part =~ s/\.a$/.lib/; - push(@libs, $part); - } elsif ($part eq 'invalidcontinue.obj') { - # ignore - known to MSVC - } elsif ($part =~ /\.o$/) { - push(@objfiles, $part); - } elsif ($part =~ /\.obj$/) { - # do nothing, 'make' should not be producing .obj, only .o files - } else { - die "Unhandled link option @ line $lineno: $part"; - } - } -# print "AppOut: '$appout'\nLFlags: @lflags\nLibs : @libs\nOfiles: @objfiles\n"; -# exit(1); - foreach (@objfiles) { - my $sourcefile = $_; - $sourcefile =~ s/^headless-git\.o$/compat\/win32\/headless.c/; - $sourcefile =~ s/\.o$/.c/; - push(@sources, $sourcefile); - push(@cflags, @{$compile_options{"${sourcefile}_CFLAGS"}}); - push(@defines, @{$compile_options{"${sourcefile}_DEFINES"}}); - push(@incpaths, @{$compile_options{"${sourcefile}_INCPATHS"}}); - } - removeDuplicates(); - - removeDuplicates(); - push(@{$build_structure{"APPS"}}, $appout); - @{$build_structure{"APPS_${appout}"}} = ("_DEFINES", "_INCLUDES", "_CFLAGS", "_LFLAGS", - "_SOURCES", "_OBJECTS", "_LIBS"); - @{$build_structure{"APPS_${appout}_DEFINES"}} = @defines; - @{$build_structure{"APPS_${appout}_INCLUDES"}} = @incpaths; - @{$build_structure{"APPS_${appout}_CFLAGS"}} = @cflags; - @{$build_structure{"APPS_${appout}_LFLAGS"}} = @lflags; - @{$build_structure{"APPS_${appout}_SOURCES"}} = @sources; - @{$build_structure{"APPS_${appout}_OBJECTS"}} = @objfiles; - @{$build_structure{"APPS_${appout}_LIBS"}} = @libs; - clearCompileStep(); -} diff --git a/contrib/buildsystems/generate b/contrib/buildsystems/generate deleted file mode 100755 index bc10f25ff24a74..00000000000000 --- a/contrib/buildsystems/generate +++ /dev/null @@ -1,29 +0,0 @@ -#!/usr/bin/perl -w -###################################################################### -# Generate buildsystem files -# -# This script generate buildsystem files based on the output of a -# GNU Make --dry-run, enabling Windows users to develop Git with their -# trusted IDE with native projects. -# -# Note: -# It is not meant as *the* way of building Git with MSVC, but merely a -# convenience. The correct way of building Git with MSVC is to use the -# GNU Make tool to build with the maintained Makefile in the root of -# the project. If you have the msysgit environment installed and -# available in your current console, together with the Visual Studio -# environment you wish to build for, all you have to do is run the -# command: -# make MSVC=1 -# -# Copyright (C) 2009 Marius Storm-Olsen -###################################################################### -use strict; -use File::Basename; -use Cwd; - -my $git_dir = getcwd(); -$git_dir =~ s=\\=/=g; -$git_dir = dirname($git_dir) while (!-e "$git_dir/git.c" && "$git_dir" ne ""); -die "Couldn't find Git repo" if ("$git_dir" eq ""); -exec join(" ", ("PERL5LIB=${git_dir}/contrib/buildsystems ${git_dir}/contrib/buildsystems/engine.pl", @ARGV)); diff --git a/contrib/buildsystems/parse.pl b/contrib/buildsystems/parse.pl deleted file mode 100755 index c9656ece9931c4..00000000000000 --- a/contrib/buildsystems/parse.pl +++ /dev/null @@ -1,228 +0,0 @@ -#!/usr/bin/perl -w -###################################################################### -# Do not call this script directly! -# -# The generate script ensures that @INC is correct before the engine -# is executed. -# -# Copyright (C) 2009 Marius Storm-Olsen -###################################################################### -use strict; -use File::Basename; -use Cwd; - -my $file = $ARGV[0]; -die "No file provided!" if !defined $file; - -my ($cflags, $target, $type, $line); - -open(F, "<$file") || die "Couldn't open file $file"; -my @data = ; -close(F); - -while (my $text = shift @data) { - my $ate_next; - do { - $ate_next = 0; - $line++; - chomp $text; - chop $text if ($text =~ /\r$/); - if ($text =~ /\\$/) { - $text =~ s/\\$//; - $text .= shift @data; - $ate_next = 1; - } - } while($ate_next); - - if($text =~ / -c /) { - # compilation - handleCompileLine($text, $line); - - } elsif ($text =~ / -o /) { - # linking executable - handleLinkLine($text, $line); - - } elsif ($text =~ /\.o / && $text =~ /\.a /) { - # libifying - handleLibLine($text, $line); - -# } elsif ($text =~ /^cp /) { -# # copy file around -# -# } elsif ($text =~ /^rm -f /) { -# # shell command -# -# } elsif ($text =~ /^make[ \[]/) { -# # make output -# -# } elsif ($text =~ /^echo /) { -# # echo to file -# -# } elsif ($text =~ /^if /) { -# # shell conditional -# -# } elsif ($text =~ /^tclsh /) { -# # translation stuff -# -# } elsif ($text =~ /^umask /) { -# # handling boilerplates -# -# } elsif ($text =~ /\$\(\:\)/) { -# # ignore -# -# } elsif ($text =~ /^FLAGS=/) { -# # flags check for dependencies -# -# } elsif ($text =~ /^'\/usr\/bin\/perl' -MError -e/) { -# # perl commands for copying files -# -# } elsif ($text =~ /generate-cmdlist\.sh/) { -# # command for generating list of commands -# -# } elsif ($text =~ /^test / && $text =~ /|| rm -f /) { -# # commands removing executables, if they exist -# -# } elsif ($text =~ /new locations or Tcl/) { -# # command for detecting Tcl/Tk changes -# -# } elsif ($text =~ /mkdir -p/) { -# # command creating path -# -# } elsif ($text =~ /: no custom templates yet/) { -# # whatever - - } else { -# print "Unhandled (line: $line): $text\n"; - } -} -close(F); - -# use Data::Dumper; -# print "Parsed build structure:\n"; -# print Dumper(%build_structure); - -# ------------------------------------------------------------------- -# Functions under here -# ------------------------------------------------------------------- -my (%build_structure, @defines, @incpaths, @cflags, @sources); - -sub clearCompileStep -{ - @defines = (); - @incpaths = (); - @cflags = (); - @sources = (); -} - -sub removeDuplicates -{ - my (%dupHash, $entry); - %dupHash = map { $_, 1 } @defines; - @defines = keys %dupHash; - - %dupHash = map { $_, 1 } @incpaths; - @incpaths = keys %dupHash; - - %dupHash = map { $_, 1 } @cflags; - @cflags = keys %dupHash; - - %dupHash = map { $_, 1 } @sources; - @sources = keys %dupHash; -} - -sub handleCompileLine -{ - my ($line, $lineno) = @_; - my @parts = split(' ', $line); - shift(@parts); # ignore cmd - while (my $part = shift @parts) { - if ("$part" eq "-o") { - # ignore object file - shift @parts; - } elsif ("$part" eq "-c") { - # ignore compile flag - } elsif ("$part" eq "-c") { - } elsif ($part =~ /^.?-I/) { - push(@incpaths, $part); - } elsif ($part =~ /^.?-D/) { - push(@defines, $part); - } elsif ($part =~ /^-/) { - push(@cflags, $part); - } elsif ($part =~ /\.(c|cc|cpp)$/) { - push(@sources, $part); - } else { - die "Unhandled compiler option @ line $lineno: $part"; - } - } - #print "Sources: @sources\nCFlags: @cflags\nDefine: @defines\nIncpat: @incpaths\n"; - #exit(1); -} - -sub handleLibLine -{ - my ($line, $lineno) = @_; - my (@objfiles, @lflags, $libout, $part); - # kill cmd and rm 'prefix' - $line =~ s/^rm -f .* && .* rcs //; - my @parts = split(' ', $line); - while ($part = shift @parts) { - if ($part =~ /^-/) { - push(@lflags, $part); - } elsif ($part =~ /\.(o|obj)$/) { - push(@objfiles, $part); - } elsif ($part =~ /\.(a|lib)$/) { - $libout = $part; - } else { - die "Unhandled lib option @ line $lineno: $part"; - } - } - #print "LibOut: '$libout'\nLFlags: @lflags\nOfiles: @objfiles\n"; - #exit(1); - removeDuplicates(); - push(@{$build_structure{"LIBS"}}, $libout); - @{$build_structure{"LIBS_${libout}"}} = ("_DEFINES", "_INCLUDES", "_CFLAGS", "_SOURCES", - "_OBJECTS"); - @{$build_structure{"LIBS_${libout}_DEFINES"}} = @defines; - @{$build_structure{"LIBS_${libout}_INCLUDES"}} = @incpaths; - @{$build_structure{"LIBS_${libout}_CFLAGS"}} = @cflags; - @{$build_structure{"LIBS_${libout}_SOURCES"}} = @sources; - @{$build_structure{"LIBS_${libout}_OBJECTS"}} = @objfiles; - clearCompileStep(); -} - -sub handleLinkLine -{ - my ($line, $lineno) = @_; - my (@objfiles, @lflags, @libs, $appout, $part); - my @parts = split(' ', $line); - shift(@parts); # ignore cmd - while ($part = shift @parts) { - if ($part =~ /^-[GRIDO]/) { - # eat compiler flags - } elsif ("$part" eq "-o") { - $appout = shift @parts; - } elsif ($part =~ /^-/) { - push(@lflags, $part); - } elsif ($part =~ /\.(a|lib)$/) { - push(@libs, $part); - } elsif ($part =~ /\.(o|obj)$/) { - push(@objfiles, $part); - } else { - die "Unhandled lib option @ line $lineno: $part"; - } - } - #print "AppOut: '$appout'\nLFlags: @lflags\nLibs : @libs\nOfiles: @objfiles\n"; - #exit(1); - removeDuplicates(); - push(@{$build_structure{"APPS"}}, $appout); - @{$build_structure{"APPS_${appout}"}} = ("_DEFINES", "_INCLUDES", "_CFLAGS", "_LFLAGS", - "_SOURCES", "_OBJECTS", "_LIBS"); - @{$build_structure{"APPS_${appout}_DEFINES"}} = @defines; - @{$build_structure{"APPS_${appout}_INCLUDES"}} = @incpaths; - @{$build_structure{"APPS_${appout}_CFLAGS"}} = @cflags; - @{$build_structure{"APPS_${appout}_LFLAGS"}} = @lflags; - @{$build_structure{"APPS_${appout}_SOURCES"}} = @sources; - @{$build_structure{"APPS_${appout}_OBJECTS"}} = @objfiles; - @{$build_structure{"APPS_${appout}_LIBS"}} = @libs; - clearCompileStep(); -} From 965d249c8b9db3e5367e0c2e560e4e212a9db065 Mon Sep 17 00:00:00 2001 From: Jeff King Date: Mon, 13 Jan 2025 01:26:01 -0500 Subject: [PATCH 029/259] grep: prevent `^$` false match at end of file In some implementations, `regexec_buf()` assumes that it is fed lines; Without `REG_NOTEOL` it thinks the end of the buffer is the end of a line. Which makes sense, but trips up this case because we are not feeding lines, but rather a whole buffer. So the final newline is not the start of an empty line, but the true end of the buffer. This causes an interesting bug: $ echo content >file.txt $ git grep --no-index -n '^$' file.txt file.txt:2: This bug is fixed by making the end of the buffer consistently the end of the final line. The patch was applied from https://lore.kernel.org/git/20250113062601.GD767856@coredump.intra.peff.net/ Reported-by: Olly Betts Signed-off-by: Johannes Schindelin --- grep.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/grep.c b/grep.c index 4e155ee9e66367..c4bb9f10814024 100644 --- a/grep.c +++ b/grep.c @@ -1646,6 +1646,8 @@ static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int colle bol = gs->buf; left = gs->size; + if (left && gs->buf[left-1] == '\n') + left--; while (left) { const char *eol; int hit; From 5c933d687a2265dbca1596d133f397dbae340059 Mon Sep 17 00:00:00 2001 From: Sverre Rabbelier Date: Sun, 24 Jul 2011 00:06:00 +0200 Subject: [PATCH 030/259] remote-helper: check helper status after import/export Signed-off-by: Johannes Schindelin Signed-off-by: Sverre Rabbelier --- t/t5801-remote-helpers.sh | 2 +- transport-helper.c | 15 +++++++++++++++ 2 files changed, 16 insertions(+), 1 deletion(-) diff --git a/t/t5801-remote-helpers.sh b/t/t5801-remote-helpers.sh index d21877150ed82e..3917da47276825 100755 --- a/t/t5801-remote-helpers.sh +++ b/t/t5801-remote-helpers.sh @@ -262,7 +262,7 @@ test_expect_success 'push update refs failure' ' echo "update fail" >>file && git commit -a -m "update fail" && git rev-parse --verify testgit/origin/heads/update >expect && - test_expect_code 1 env GIT_REMOTE_TESTGIT_FAILURE="non-fast forward" \ + test_must_fail env GIT_REMOTE_TESTGIT_FAILURE="non-fast forward" \ git push origin update && git rev-parse --verify testgit/origin/heads/update >actual && test_cmp expect actual diff --git a/transport-helper.c b/transport-helper.c index fed5d51a2b3d34..72588c4d14e3ff 100644 --- a/transport-helper.c +++ b/transport-helper.c @@ -505,6 +505,19 @@ static int get_exporter(struct transport *transport, return start_command(fastexport); } +static void check_helper_status(struct helper_data *data) +{ + int pid, status; + + pid = waitpid(data->helper->pid, &status, WNOHANG); + if (pid < 0) + die("Could not retrieve status of remote helper '%s'", + data->name); + if (pid > 0 && WIFEXITED(status)) + die("Remote helper '%s' died with %d", + data->name, WEXITSTATUS(status)); +} + static int fetch_with_import(struct transport *transport, int nr_heads, struct ref **to_fetch) { @@ -541,6 +554,7 @@ static int fetch_with_import(struct transport *transport, if (finish_command(&fastimport)) die(_("error while running fast-import")); + check_helper_status(data); /* * The fast-import stream of a remote helper that advertises @@ -1160,6 +1174,7 @@ static int push_refs_with_export(struct transport *transport, if (finish_command(&exporter)) die(_("error while running fast-export")); + check_helper_status(data); if (push_update_refs_status(data, remote_refs, flags)) return 1; From 09eff7ed3f309899cace4a1dce76128bb429ee83 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sat, 9 May 2020 14:08:36 +0200 Subject: [PATCH 031/259] config.mak.uname: drop the `vcxproj` target Now that we dropped `contrib/buildsystems/generate` to generate Visual Studio Solution files, it is time to also drop the `vcxproj` Makefile target that depended on that script. Signed-off-by: Johannes Schindelin --- config.mak.uname | 76 ------------------------------------------------ 1 file changed, 76 deletions(-) diff --git a/config.mak.uname b/config.mak.uname index b12d4e168ae119..c7a9fa1b775981 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -759,79 +759,3 @@ ifeq ($(uname_S),QNX) NO_STRCASESTR = YesPlease NO_STRLCPY = YesPlease endif - -vcxproj: - # Require clean work tree - git update-index -q --refresh && \ - git diff-files --quiet && \ - git diff-index --cached --quiet HEAD -- - - # Make .vcxproj files and add them - perl contrib/buildsystems/generate -g Vcxproj - git add -f git.sln {*,*/lib,t/helper/*}/*.vcxproj - - # Generate the LinkOrCopyBuiltins.targets and LinkOrCopyRemoteHttp.targets file - (echo '' && \ - echo ' ' && \ - for name in $(BUILT_INS);\ - do \ - echo ' '; \ - done && \ - echo ' ' && \ - echo '') >git/LinkOrCopyBuiltins.targets - (echo '' && \ - echo ' ' && \ - for name in $(REMOTE_CURL_ALIASES); \ - do \ - echo ' '; \ - done && \ - echo ' ' && \ - echo '') >git-remote-http/LinkOrCopyRemoteHttp.targets - git add -f git/LinkOrCopyBuiltins.targets git-remote-http/LinkOrCopyRemoteHttp.targets - - # Add generated headers - $(MAKE) MSVC=1 SKIP_VCPKG=1 prefix=/mingw64 $(GENERATED_H) - git add -f $(GENERATED_H) - - # Add scripts - rm -f perl/perl.mak - $(MAKE) MSVC=1 SKIP_VCPKG=1 prefix=/mingw64 $(SCRIPT_LIB) $(SCRIPTS) - # Strip out the sane tool path, needed only for building - sed -i '/^git_broken_path_fix ".*/d' git-sh-setup - git add -f $(SCRIPT_LIB) $(SCRIPTS) - - # Add Perl module - $(MAKE) $(LIB_PERL_GEN) - git add -f perl/build - - # Add bin-wrappers, for testing - rm -rf bin-wrappers/ - $(MAKE) MSVC=1 SKIP_VCPKG=1 prefix=/mingw64 $(test_bindir_programs) - # Ensure that the GIT_EXEC_PATH is a Unix-y one, and that the absolute - # path of the repository is not hard-coded (GIT_EXEC_PATH will be set - # by test-lib.sh according to the current setup) - sed -i -e 's/^\(GIT_EXEC_PATH\)=.*/test -n "$${\1##*:*}" ||\ - \1="$$(cygpath -u "$$\1")"/' \ - -e "s|'$$(pwd)|\"\$$GIT_EXEC_PATH\"'|g" bin-wrappers/* - # Ensure that test-* helpers find the .dll files copied to top-level - sed -i 's|^PATH=.*|&:"$$GIT_EXEC_PATH"|' bin-wrappers/test-* - # We do not want to force hard-linking builtins - sed -i 's|\(git\)-\([-a-z]*\)\.exe"|\1.exe" \2|g' \ - bin-wrappers/git-{receive-pack,upload-archive} - git add -f $(test_bindir_programs) - - # Add templates - $(MAKE) -C templates - git add -f templates/boilerplates.made templates/blt/ - - # Add the translated messages - make MSVC=1 SKIP_VCPKG=1 prefix=/mingw64 $(MOFILES) - git add -f $(MOFILES) - - # Add build options - $(MAKE) MSVC=1 SKIP_VCPKG=1 prefix=/mingw64 GIT-BUILD-OPTIONS - git add -f GIT-BUILD-OPTIONS - - # Commit the whole shebang - git commit -m "Generate Visual Studio solution" \ - -m "Auto-generated by \`$(MAKE)$(MAKEFLAGS) $@\`" From 7b7d20c524988fdd073b7d456f3b0fb10064b97b Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 9 Apr 2012 13:04:35 -0500 Subject: [PATCH 032/259] Always auto-gc after calling a fast-import transport After importing anything with fast-import, we should always let the garbage collector do its job, since the objects are written to disk inefficiently. This brings down an initial import of http://selenic.com/hg from about 230 megabytes to about 14. In the future, we may want to make this configurable on a per-remote basis, or maybe teach fast-import about it in the first place. Signed-off-by: Johannes Schindelin --- transport-helper.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/transport-helper.c b/transport-helper.c index 72588c4d14e3ff..7513fd7eea05e4 100644 --- a/transport-helper.c +++ b/transport-helper.c @@ -22,6 +22,8 @@ #include "packfile.h" static int debug; +/* TODO: put somewhere sensible, e.g. git_transport_options? */ +static int auto_gc = 1; struct helper_data { char *name; @@ -588,6 +590,13 @@ static int fetch_with_import(struct transport *transport, } } strbuf_release(&buf); + if (auto_gc) { + struct child_process cmd = CHILD_PROCESS_INIT; + + cmd.git_cmd = 1; + strvec_pushl(&cmd.args, "gc", "--auto", "--quiet", NULL); + run_command(&cmd); + } return 0; } From c75c3ecfb64ffb2248f058b0275aefc1826dd1b2 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sun, 20 Oct 2019 22:08:58 +0200 Subject: [PATCH 033/259] win32/pthread: avoid name clashes with winpthread The mingw-w64 GCC seems to link implicitly to libwinpthread, which does implement a pthread emulation (that is more complete than Git's). Let's keep preferring Git's. To avoid linker errors where it thinks that the `pthread_self` and the `pthread_create` symbols are defined twice, let's give our version a `win32_` prefix, just like we already do for `pthread_join()`. Signed-off-by: Johannes Schindelin --- compat/win32/pthread.c | 6 +++--- compat/win32/pthread.h | 8 +++++--- 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/compat/win32/pthread.c b/compat/win32/pthread.c index 58980a529c3eb9..4cc6356f757d80 100644 --- a/compat/win32/pthread.c +++ b/compat/win32/pthread.c @@ -21,8 +21,8 @@ static unsigned __stdcall win32_start_routine(void *arg) return 0; } -int pthread_create(pthread_t *thread, const void *attr UNUSED, - void *(*start_routine)(void *), void *arg) +int win32_pthread_create(pthread_t *thread, const void *attr UNUSED, + void *(*start_routine)(void *), void *arg) { thread->arg = arg; thread->start_routine = start_routine; @@ -53,7 +53,7 @@ int win32_pthread_join(pthread_t *thread, void **value_ptr) } } -pthread_t pthread_self(void) +pthread_t win32_pthread_self(void) { pthread_t t = { NULL }; t.tid = GetCurrentThreadId(); diff --git a/compat/win32/pthread.h b/compat/win32/pthread.h index e2b5c4f64c9b91..476e78b911b857 100644 --- a/compat/win32/pthread.h +++ b/compat/win32/pthread.h @@ -50,8 +50,9 @@ typedef struct { DWORD tid; } pthread_t; -int pthread_create(pthread_t *thread, const void *unused, - void *(*start_routine)(void*), void *arg); +int win32_pthread_create(pthread_t *thread, const void *unused, + void *(*start_routine)(void*), void *arg); +#define pthread_create win32_pthread_create /* * To avoid the need of copying a struct, we use small macro wrapper to pass @@ -62,7 +63,8 @@ int pthread_create(pthread_t *thread, const void *unused, int win32_pthread_join(pthread_t *thread, void **value_ptr); #define pthread_equal(t1, t2) ((t1).tid == (t2).tid) -pthread_t pthread_self(void); +pthread_t win32_pthread_self(void); +#define pthread_self win32_pthread_self static inline void NORETURN pthread_exit(void *ret) { From a1f4e5f83ee33768048a5078abdae89cf17a2ced Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 12 Aug 2022 12:44:15 +0200 Subject: [PATCH 034/259] git-compat-util: avoid redeclaring _DEFAULT_SOURCE We are about to vendor in `mimalloc`'s source code which we will want to include `git-compat-util.h` after defining that constant. Signed-off-by: Johannes Schindelin --- git-compat-util.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/git-compat-util.h b/git-compat-util.h index e123288e8f1393..6aee7e1b27a26b 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -201,7 +201,9 @@ DISABLE_WARNING(-Wsign-compare) #define _ALL_SOURCE 1 #define _GNU_SOURCE 1 #define _BSD_SOURCE 1 +#ifndef _DEFAULT_SOURCE #define _DEFAULT_SOURCE 1 +#endif #define _NETBSD_SOURCE 1 #define _SGI_SOURCE 1 From b30d3b28d568b77681195b4b2c417b383805019e Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 24 Jun 2019 21:31:30 +0200 Subject: [PATCH 035/259] Import the source code of mimalloc v2.1.2 This commit imports mimalloc's source code as per v2.1.2, fetched from the tag at https://github.com/microsoft/mimalloc. The .c files are from the src/ subdirectory, and the .h files from the include/ and include/mimalloc/ subdirectories. We will subsequently modify the source code to accommodate building within Git's context. Since we plan on using the `mi_*()` family of functions, we skip the C++-specific source code, some POSIX compliant functions to interact with mimalloc, and the code that wants to support auto-magic overriding of the `malloc()` function (mimalloc-new-delete.h, alloc-posix.c, mimalloc-override.h, alloc-override.c, alloc-override-osx.c, alloc-override-win.c and static.c). To appease the `check-whitespace` job of Git's Continuous Integration, this commit was washed one time via `git rebase --whitespace=fix`. Signed-off-by: Johannes Schindelin --- Makefile | 1 + compat/mimalloc/LICENSE | 21 + compat/mimalloc/alloc-aligned.c | 298 +++++ compat/mimalloc/alloc.c | 1060 ++++++++++++++++++ compat/mimalloc/arena.c | 935 ++++++++++++++++ compat/mimalloc/bitmap.c | 432 +++++++ compat/mimalloc/bitmap.h | 115 ++ compat/mimalloc/heap.c | 626 +++++++++++ compat/mimalloc/init.c | 709 ++++++++++++ compat/mimalloc/mimalloc.h | 565 ++++++++++ compat/mimalloc/mimalloc/atomic.h | 385 +++++++ compat/mimalloc/mimalloc/internal.h | 979 ++++++++++++++++ compat/mimalloc/mimalloc/prim.h | 323 ++++++ compat/mimalloc/mimalloc/track.h | 147 +++ compat/mimalloc/mimalloc/types.h | 670 +++++++++++ compat/mimalloc/options.c | 571 ++++++++++ compat/mimalloc/os.c | 689 ++++++++++++ compat/mimalloc/page-queue.c | 332 ++++++ compat/mimalloc/page.c | 939 ++++++++++++++++ compat/mimalloc/prim/windows/prim.c | 622 +++++++++++ compat/mimalloc/random.c | 254 +++++ compat/mimalloc/segment-cache.c | 0 compat/mimalloc/segment-map.c | 153 +++ compat/mimalloc/segment.c | 1617 +++++++++++++++++++++++++++ compat/mimalloc/stats.c | 467 ++++++++ 25 files changed, 12910 insertions(+) create mode 100644 compat/mimalloc/LICENSE create mode 100644 compat/mimalloc/alloc-aligned.c create mode 100644 compat/mimalloc/alloc.c create mode 100644 compat/mimalloc/arena.c create mode 100644 compat/mimalloc/bitmap.c create mode 100644 compat/mimalloc/bitmap.h create mode 100644 compat/mimalloc/heap.c create mode 100644 compat/mimalloc/init.c create mode 100644 compat/mimalloc/mimalloc.h create mode 100644 compat/mimalloc/mimalloc/atomic.h create mode 100644 compat/mimalloc/mimalloc/internal.h create mode 100644 compat/mimalloc/mimalloc/prim.h create mode 100644 compat/mimalloc/mimalloc/track.h create mode 100644 compat/mimalloc/mimalloc/types.h create mode 100644 compat/mimalloc/options.c create mode 100644 compat/mimalloc/os.c create mode 100644 compat/mimalloc/page-queue.c create mode 100644 compat/mimalloc/page.c create mode 100644 compat/mimalloc/prim/windows/prim.c create mode 100644 compat/mimalloc/random.c create mode 100644 compat/mimalloc/segment-cache.c create mode 100644 compat/mimalloc/segment-map.c create mode 100644 compat/mimalloc/segment.c create mode 100644 compat/mimalloc/stats.c diff --git a/Makefile b/Makefile index 7315507381e146..fbb9227e7d7909 100644 --- a/Makefile +++ b/Makefile @@ -1342,6 +1342,7 @@ BUILTIN_OBJS += builtin/write-tree.o # upstream unnecessarily (making merging in future changes easier). THIRD_PARTY_SOURCES += compat/inet_ntop.c THIRD_PARTY_SOURCES += compat/inet_pton.c +THIRD_PARTY_SOURCES += compat/mimalloc/% THIRD_PARTY_SOURCES += compat/nedmalloc/% THIRD_PARTY_SOURCES += compat/obstack.% THIRD_PARTY_SOURCES += compat/poll/% diff --git a/compat/mimalloc/LICENSE b/compat/mimalloc/LICENSE new file mode 100644 index 00000000000000..670b668a0c928e --- /dev/null +++ b/compat/mimalloc/LICENSE @@ -0,0 +1,21 @@ +MIT License + +Copyright (c) 2018-2021 Microsoft Corporation, Daan Leijen + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. diff --git a/compat/mimalloc/alloc-aligned.c b/compat/mimalloc/alloc-aligned.c new file mode 100644 index 00000000000000..e975af5f7c2ad4 --- /dev/null +++ b/compat/mimalloc/alloc-aligned.c @@ -0,0 +1,298 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2021, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/prim.h" // mi_prim_get_default_heap + +#include // memset + +// ------------------------------------------------------ +// Aligned Allocation +// ------------------------------------------------------ + +// Fallback primitive aligned allocation -- split out for better codegen +static mi_decl_noinline void* mi_heap_malloc_zero_aligned_at_fallback(mi_heap_t* const heap, const size_t size, const size_t alignment, const size_t offset, const bool zero) mi_attr_noexcept +{ + mi_assert_internal(size <= PTRDIFF_MAX); + mi_assert_internal(alignment != 0 && _mi_is_power_of_two(alignment)); + + const uintptr_t align_mask = alignment - 1; // for any x, `(x & align_mask) == (x % alignment)` + const size_t padsize = size + MI_PADDING_SIZE; + + // use regular allocation if it is guaranteed to fit the alignment constraints + if (offset==0 && alignment<=padsize && padsize<=MI_MAX_ALIGN_GUARANTEE && (padsize&align_mask)==0) { + void* p = _mi_heap_malloc_zero(heap, size, zero); + mi_assert_internal(p == NULL || ((uintptr_t)p % alignment) == 0); + return p; + } + + void* p; + size_t oversize; + if mi_unlikely(alignment > MI_ALIGNMENT_MAX) { + // use OS allocation for very large alignment and allocate inside a huge page (dedicated segment with 1 page) + // This can support alignments >= MI_SEGMENT_SIZE by ensuring the object can be aligned at a point in the + // first (and single) page such that the segment info is `MI_SEGMENT_SIZE` bytes before it (so it can be found by aligning the pointer down) + if mi_unlikely(offset != 0) { + // todo: cannot support offset alignment for very large alignments yet + #if MI_DEBUG > 0 + _mi_error_message(EOVERFLOW, "aligned allocation with a very large alignment cannot be used with an alignment offset (size %zu, alignment %zu, offset %zu)\n", size, alignment, offset); + #endif + return NULL; + } + oversize = (size <= MI_SMALL_SIZE_MAX ? MI_SMALL_SIZE_MAX + 1 /* ensure we use generic malloc path */ : size); + p = _mi_heap_malloc_zero_ex(heap, oversize, false, alignment); // the page block size should be large enough to align in the single huge page block + // zero afterwards as only the area from the aligned_p may be committed! + if (p == NULL) return NULL; + } + else { + // otherwise over-allocate + oversize = size + alignment - 1; + p = _mi_heap_malloc_zero(heap, oversize, zero); + if (p == NULL) return NULL; + } + + // .. and align within the allocation + const uintptr_t poffset = ((uintptr_t)p + offset) & align_mask; + const uintptr_t adjust = (poffset == 0 ? 0 : alignment - poffset); + mi_assert_internal(adjust < alignment); + void* aligned_p = (void*)((uintptr_t)p + adjust); + if (aligned_p != p) { + mi_page_t* page = _mi_ptr_page(p); + mi_page_set_has_aligned(page, true); + _mi_padding_shrink(page, (mi_block_t*)p, adjust + size); + } + // todo: expand padding if overallocated ? + + mi_assert_internal(mi_page_usable_block_size(_mi_ptr_page(p)) >= adjust + size); + mi_assert_internal(p == _mi_page_ptr_unalign(_mi_ptr_segment(aligned_p), _mi_ptr_page(aligned_p), aligned_p)); + mi_assert_internal(((uintptr_t)aligned_p + offset) % alignment == 0); + mi_assert_internal(mi_usable_size(aligned_p)>=size); + mi_assert_internal(mi_usable_size(p) == mi_usable_size(aligned_p)+adjust); + + // now zero the block if needed + if (alignment > MI_ALIGNMENT_MAX) { + // for the tracker, on huge aligned allocations only from the start of the large block is defined + mi_track_mem_undefined(aligned_p, size); + if (zero) { + _mi_memzero_aligned(aligned_p, mi_usable_size(aligned_p)); + } + } + + if (p != aligned_p) { + mi_track_align(p,aligned_p,adjust,mi_usable_size(aligned_p)); + } + return aligned_p; +} + +// Primitive aligned allocation +static void* mi_heap_malloc_zero_aligned_at(mi_heap_t* const heap, const size_t size, const size_t alignment, const size_t offset, const bool zero) mi_attr_noexcept +{ + // note: we don't require `size > offset`, we just guarantee that the address at offset is aligned regardless of the allocated size. + if mi_unlikely(alignment == 0 || !_mi_is_power_of_two(alignment)) { // require power-of-two (see ) + #if MI_DEBUG > 0 + _mi_error_message(EOVERFLOW, "aligned allocation requires the alignment to be a power-of-two (size %zu, alignment %zu)\n", size, alignment); + #endif + return NULL; + } + + if mi_unlikely(size > PTRDIFF_MAX) { // we don't allocate more than PTRDIFF_MAX (see ) + #if MI_DEBUG > 0 + _mi_error_message(EOVERFLOW, "aligned allocation request is too large (size %zu, alignment %zu)\n", size, alignment); + #endif + return NULL; + } + const uintptr_t align_mask = alignment-1; // for any x, `(x & align_mask) == (x % alignment)` + const size_t padsize = size + MI_PADDING_SIZE; // note: cannot overflow due to earlier size > PTRDIFF_MAX check + + // try first if there happens to be a small block available with just the right alignment + if mi_likely(padsize <= MI_SMALL_SIZE_MAX && alignment <= padsize) { + mi_page_t* page = _mi_heap_get_free_small_page(heap, padsize); + const bool is_aligned = (((uintptr_t)page->free+offset) & align_mask)==0; + if mi_likely(page->free != NULL && is_aligned) + { + #if MI_STAT>1 + mi_heap_stat_increase(heap, malloc, size); + #endif + void* p = _mi_page_malloc(heap, page, padsize, zero); // TODO: inline _mi_page_malloc + mi_assert_internal(p != NULL); + mi_assert_internal(((uintptr_t)p + offset) % alignment == 0); + mi_track_malloc(p,size,zero); + return p; + } + } + // fallback + return mi_heap_malloc_zero_aligned_at_fallback(heap, size, alignment, offset, zero); +} + + +// ------------------------------------------------------ +// Optimized mi_heap_malloc_aligned / mi_malloc_aligned +// ------------------------------------------------------ + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_malloc_aligned_at(mi_heap_t* heap, size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_malloc_zero_aligned_at(heap, size, alignment, offset, false); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_malloc_aligned(mi_heap_t* heap, size_t size, size_t alignment) mi_attr_noexcept { + if mi_unlikely(alignment == 0 || !_mi_is_power_of_two(alignment)) return NULL; + #if !MI_PADDING + // without padding, any small sized allocation is naturally aligned (see also `_mi_segment_page_start`) + if mi_likely(_mi_is_power_of_two(size) && size >= alignment && size <= MI_SMALL_SIZE_MAX) + #else + // with padding, we can only guarantee this for fixed alignments + if mi_likely((alignment == sizeof(void*) || (alignment == MI_MAX_ALIGN_SIZE && size > (MI_MAX_ALIGN_SIZE/2))) + && size <= MI_SMALL_SIZE_MAX) + #endif + { + // fast path for common alignment and size + return mi_heap_malloc_small(heap, size); + } + else { + return mi_heap_malloc_aligned_at(heap, size, alignment, 0); + } +} + +// ensure a definition is emitted +#if defined(__cplusplus) +static void* _mi_heap_malloc_aligned = (void*)&mi_heap_malloc_aligned; +#endif + +// ------------------------------------------------------ +// Aligned Allocation +// ------------------------------------------------------ + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_zalloc_aligned_at(mi_heap_t* heap, size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_malloc_zero_aligned_at(heap, size, alignment, offset, true); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_zalloc_aligned(mi_heap_t* heap, size_t size, size_t alignment) mi_attr_noexcept { + return mi_heap_zalloc_aligned_at(heap, size, alignment, 0); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_calloc_aligned_at(mi_heap_t* heap, size_t count, size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + size_t total; + if (mi_count_size_overflow(count, size, &total)) return NULL; + return mi_heap_zalloc_aligned_at(heap, total, alignment, offset); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_calloc_aligned(mi_heap_t* heap, size_t count, size_t size, size_t alignment) mi_attr_noexcept { + return mi_heap_calloc_aligned_at(heap,count,size,alignment,0); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_malloc_aligned_at(size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_malloc_aligned_at(mi_prim_get_default_heap(), size, alignment, offset); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_malloc_aligned(size_t size, size_t alignment) mi_attr_noexcept { + return mi_heap_malloc_aligned(mi_prim_get_default_heap(), size, alignment); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_zalloc_aligned_at(size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_zalloc_aligned_at(mi_prim_get_default_heap(), size, alignment, offset); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_zalloc_aligned(size_t size, size_t alignment) mi_attr_noexcept { + return mi_heap_zalloc_aligned(mi_prim_get_default_heap(), size, alignment); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_calloc_aligned_at(size_t count, size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_calloc_aligned_at(mi_prim_get_default_heap(), count, size, alignment, offset); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_calloc_aligned(size_t count, size_t size, size_t alignment) mi_attr_noexcept { + return mi_heap_calloc_aligned(mi_prim_get_default_heap(), count, size, alignment); +} + + +// ------------------------------------------------------ +// Aligned re-allocation +// ------------------------------------------------------ + +static void* mi_heap_realloc_zero_aligned_at(mi_heap_t* heap, void* p, size_t newsize, size_t alignment, size_t offset, bool zero) mi_attr_noexcept { + mi_assert(alignment > 0); + if (alignment <= sizeof(uintptr_t)) return _mi_heap_realloc_zero(heap,p,newsize,zero); + if (p == NULL) return mi_heap_malloc_zero_aligned_at(heap,newsize,alignment,offset,zero); + size_t size = mi_usable_size(p); + if (newsize <= size && newsize >= (size - (size / 2)) + && (((uintptr_t)p + offset) % alignment) == 0) { + return p; // reallocation still fits, is aligned and not more than 50% waste + } + else { + // note: we don't zero allocate upfront so we only zero initialize the expanded part + void* newp = mi_heap_malloc_aligned_at(heap,newsize,alignment,offset); + if (newp != NULL) { + if (zero && newsize > size) { + // also set last word in the previous allocation to zero to ensure any padding is zero-initialized + size_t start = (size >= sizeof(intptr_t) ? size - sizeof(intptr_t) : 0); + _mi_memzero((uint8_t*)newp + start, newsize - start); + } + _mi_memcpy_aligned(newp, p, (newsize > size ? size : newsize)); + mi_free(p); // only free if successful + } + return newp; + } +} + +static void* mi_heap_realloc_zero_aligned(mi_heap_t* heap, void* p, size_t newsize, size_t alignment, bool zero) mi_attr_noexcept { + mi_assert(alignment > 0); + if (alignment <= sizeof(uintptr_t)) return _mi_heap_realloc_zero(heap,p,newsize,zero); + size_t offset = ((uintptr_t)p % alignment); // use offset of previous allocation (p can be NULL) + return mi_heap_realloc_zero_aligned_at(heap,p,newsize,alignment,offset,zero); +} + +mi_decl_nodiscard void* mi_heap_realloc_aligned_at(mi_heap_t* heap, void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_realloc_zero_aligned_at(heap,p,newsize,alignment,offset,false); +} + +mi_decl_nodiscard void* mi_heap_realloc_aligned(mi_heap_t* heap, void* p, size_t newsize, size_t alignment) mi_attr_noexcept { + return mi_heap_realloc_zero_aligned(heap,p,newsize,alignment,false); +} + +mi_decl_nodiscard void* mi_heap_rezalloc_aligned_at(mi_heap_t* heap, void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_realloc_zero_aligned_at(heap, p, newsize, alignment, offset, true); +} + +mi_decl_nodiscard void* mi_heap_rezalloc_aligned(mi_heap_t* heap, void* p, size_t newsize, size_t alignment) mi_attr_noexcept { + return mi_heap_realloc_zero_aligned(heap, p, newsize, alignment, true); +} + +mi_decl_nodiscard void* mi_heap_recalloc_aligned_at(mi_heap_t* heap, void* p, size_t newcount, size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + size_t total; + if (mi_count_size_overflow(newcount, size, &total)) return NULL; + return mi_heap_rezalloc_aligned_at(heap, p, total, alignment, offset); +} + +mi_decl_nodiscard void* mi_heap_recalloc_aligned(mi_heap_t* heap, void* p, size_t newcount, size_t size, size_t alignment) mi_attr_noexcept { + size_t total; + if (mi_count_size_overflow(newcount, size, &total)) return NULL; + return mi_heap_rezalloc_aligned(heap, p, total, alignment); +} + +mi_decl_nodiscard void* mi_realloc_aligned_at(void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_realloc_aligned_at(mi_prim_get_default_heap(), p, newsize, alignment, offset); +} + +mi_decl_nodiscard void* mi_realloc_aligned(void* p, size_t newsize, size_t alignment) mi_attr_noexcept { + return mi_heap_realloc_aligned(mi_prim_get_default_heap(), p, newsize, alignment); +} + +mi_decl_nodiscard void* mi_rezalloc_aligned_at(void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_rezalloc_aligned_at(mi_prim_get_default_heap(), p, newsize, alignment, offset); +} + +mi_decl_nodiscard void* mi_rezalloc_aligned(void* p, size_t newsize, size_t alignment) mi_attr_noexcept { + return mi_heap_rezalloc_aligned(mi_prim_get_default_heap(), p, newsize, alignment); +} + +mi_decl_nodiscard void* mi_recalloc_aligned_at(void* p, size_t newcount, size_t size, size_t alignment, size_t offset) mi_attr_noexcept { + return mi_heap_recalloc_aligned_at(mi_prim_get_default_heap(), p, newcount, size, alignment, offset); +} + +mi_decl_nodiscard void* mi_recalloc_aligned(void* p, size_t newcount, size_t size, size_t alignment) mi_attr_noexcept { + return mi_heap_recalloc_aligned(mi_prim_get_default_heap(), p, newcount, size, alignment); +} diff --git a/compat/mimalloc/alloc.c b/compat/mimalloc/alloc.c new file mode 100644 index 00000000000000..961f6d53d0f2c7 --- /dev/null +++ b/compat/mimalloc/alloc.c @@ -0,0 +1,1060 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2022, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#ifndef _DEFAULT_SOURCE +#define _DEFAULT_SOURCE // for realpath() on Linux +#endif + +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" +#include "mimalloc/prim.h" // _mi_prim_thread_id() + +#include // memset, strlen (for mi_strdup) +#include // malloc, abort + +#define MI_IN_ALLOC_C +#include "alloc-override.c" +#undef MI_IN_ALLOC_C + +// ------------------------------------------------------ +// Allocation +// ------------------------------------------------------ + +// Fast allocation in a page: just pop from the free list. +// Fall back to generic allocation only if the list is empty. +extern inline void* _mi_page_malloc(mi_heap_t* heap, mi_page_t* page, size_t size, bool zero) mi_attr_noexcept { + mi_assert_internal(page->xblock_size==0||mi_page_block_size(page) >= size); + mi_block_t* const block = page->free; + if mi_unlikely(block == NULL) { + return _mi_malloc_generic(heap, size, zero, 0); + } + mi_assert_internal(block != NULL && _mi_ptr_page(block) == page); + // pop from the free list + page->used++; + page->free = mi_block_next(page, block); + mi_assert_internal(page->free == NULL || _mi_ptr_page(page->free) == page); + #if MI_DEBUG>3 + if (page->free_is_zero) { + mi_assert_expensive(mi_mem_is_zero(block+1,size - sizeof(*block))); + } + #endif + + // allow use of the block internally + // note: when tracking we need to avoid ever touching the MI_PADDING since + // that is tracked by valgrind etc. as non-accessible (through the red-zone, see `mimalloc/track.h`) + mi_track_mem_undefined(block, mi_page_usable_block_size(page)); + + // zero the block? note: we need to zero the full block size (issue #63) + if mi_unlikely(zero) { + mi_assert_internal(page->xblock_size != 0); // do not call with zero'ing for huge blocks (see _mi_malloc_generic) + mi_assert_internal(page->xblock_size >= MI_PADDING_SIZE); + if (page->free_is_zero) { + block->next = 0; + mi_track_mem_defined(block, page->xblock_size - MI_PADDING_SIZE); + } + else { + _mi_memzero_aligned(block, page->xblock_size - MI_PADDING_SIZE); + } + } + +#if (MI_DEBUG>0) && !MI_TRACK_ENABLED && !MI_TSAN + if (!zero && !mi_page_is_huge(page)) { + memset(block, MI_DEBUG_UNINIT, mi_page_usable_block_size(page)); + } +#elif (MI_SECURE!=0) + if (!zero) { block->next = 0; } // don't leak internal data +#endif + +#if (MI_STAT>0) + const size_t bsize = mi_page_usable_block_size(page); + if (bsize <= MI_MEDIUM_OBJ_SIZE_MAX) { + mi_heap_stat_increase(heap, normal, bsize); + mi_heap_stat_counter_increase(heap, normal_count, 1); +#if (MI_STAT>1) + const size_t bin = _mi_bin(bsize); + mi_heap_stat_increase(heap, normal_bins[bin], 1); +#endif + } +#endif + +#if MI_PADDING // && !MI_TRACK_ENABLED + mi_padding_t* const padding = (mi_padding_t*)((uint8_t*)block + mi_page_usable_block_size(page)); + ptrdiff_t delta = ((uint8_t*)padding - (uint8_t*)block - (size - MI_PADDING_SIZE)); + #if (MI_DEBUG>=2) + mi_assert_internal(delta >= 0 && mi_page_usable_block_size(page) >= (size - MI_PADDING_SIZE + delta)); + #endif + mi_track_mem_defined(padding,sizeof(mi_padding_t)); // note: re-enable since mi_page_usable_block_size may set noaccess + padding->canary = (uint32_t)(mi_ptr_encode(page,block,page->keys)); + padding->delta = (uint32_t)(delta); + #if MI_PADDING_CHECK + if (!mi_page_is_huge(page)) { + uint8_t* fill = (uint8_t*)padding - delta; + const size_t maxpad = (delta > MI_MAX_ALIGN_SIZE ? MI_MAX_ALIGN_SIZE : delta); // set at most N initial padding bytes + for (size_t i = 0; i < maxpad; i++) { fill[i] = MI_DEBUG_PADDING; } + } + #endif +#endif + + return block; +} + +static inline mi_decl_restrict void* mi_heap_malloc_small_zero(mi_heap_t* heap, size_t size, bool zero) mi_attr_noexcept { + mi_assert(heap != NULL); + #if MI_DEBUG + const uintptr_t tid = _mi_thread_id(); + mi_assert(heap->thread_id == 0 || heap->thread_id == tid); // heaps are thread local + #endif + mi_assert(size <= MI_SMALL_SIZE_MAX); + #if (MI_PADDING) + if (size == 0) { size = sizeof(void*); } + #endif + mi_page_t* page = _mi_heap_get_free_small_page(heap, size + MI_PADDING_SIZE); + void* const p = _mi_page_malloc(heap, page, size + MI_PADDING_SIZE, zero); + mi_track_malloc(p,size,zero); + #if MI_STAT>1 + if (p != NULL) { + if (!mi_heap_is_initialized(heap)) { heap = mi_prim_get_default_heap(); } + mi_heap_stat_increase(heap, malloc, mi_usable_size(p)); + } + #endif + #if MI_DEBUG>3 + if (p != NULL && zero) { + mi_assert_expensive(mi_mem_is_zero(p, size)); + } + #endif + return p; +} + +// allocate a small block +mi_decl_nodiscard extern inline mi_decl_restrict void* mi_heap_malloc_small(mi_heap_t* heap, size_t size) mi_attr_noexcept { + return mi_heap_malloc_small_zero(heap, size, false); +} + +mi_decl_nodiscard extern inline mi_decl_restrict void* mi_malloc_small(size_t size) mi_attr_noexcept { + return mi_heap_malloc_small(mi_prim_get_default_heap(), size); +} + +// The main allocation function +extern inline void* _mi_heap_malloc_zero_ex(mi_heap_t* heap, size_t size, bool zero, size_t huge_alignment) mi_attr_noexcept { + if mi_likely(size <= MI_SMALL_SIZE_MAX) { + mi_assert_internal(huge_alignment == 0); + return mi_heap_malloc_small_zero(heap, size, zero); + } + else { + mi_assert(heap!=NULL); + mi_assert(heap->thread_id == 0 || heap->thread_id == _mi_thread_id()); // heaps are thread local + void* const p = _mi_malloc_generic(heap, size + MI_PADDING_SIZE, zero, huge_alignment); // note: size can overflow but it is detected in malloc_generic + mi_track_malloc(p,size,zero); + #if MI_STAT>1 + if (p != NULL) { + if (!mi_heap_is_initialized(heap)) { heap = mi_prim_get_default_heap(); } + mi_heap_stat_increase(heap, malloc, mi_usable_size(p)); + } + #endif + #if MI_DEBUG>3 + if (p != NULL && zero) { + mi_assert_expensive(mi_mem_is_zero(p, size)); + } + #endif + return p; + } +} + +extern inline void* _mi_heap_malloc_zero(mi_heap_t* heap, size_t size, bool zero) mi_attr_noexcept { + return _mi_heap_malloc_zero_ex(heap, size, zero, 0); +} + +mi_decl_nodiscard extern inline mi_decl_restrict void* mi_heap_malloc(mi_heap_t* heap, size_t size) mi_attr_noexcept { + return _mi_heap_malloc_zero(heap, size, false); +} + +mi_decl_nodiscard extern inline mi_decl_restrict void* mi_malloc(size_t size) mi_attr_noexcept { + return mi_heap_malloc(mi_prim_get_default_heap(), size); +} + +// zero initialized small block +mi_decl_nodiscard mi_decl_restrict void* mi_zalloc_small(size_t size) mi_attr_noexcept { + return mi_heap_malloc_small_zero(mi_prim_get_default_heap(), size, true); +} + +mi_decl_nodiscard extern inline mi_decl_restrict void* mi_heap_zalloc(mi_heap_t* heap, size_t size) mi_attr_noexcept { + return _mi_heap_malloc_zero(heap, size, true); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_zalloc(size_t size) mi_attr_noexcept { + return mi_heap_zalloc(mi_prim_get_default_heap(),size); +} + + +// ------------------------------------------------------ +// Check for double free in secure and debug mode +// This is somewhat expensive so only enabled for secure mode 4 +// ------------------------------------------------------ + +#if (MI_ENCODE_FREELIST && (MI_SECURE>=4 || MI_DEBUG!=0)) +// linear check if the free list contains a specific element +static bool mi_list_contains(const mi_page_t* page, const mi_block_t* list, const mi_block_t* elem) { + while (list != NULL) { + if (elem==list) return true; + list = mi_block_next(page, list); + } + return false; +} + +static mi_decl_noinline bool mi_check_is_double_freex(const mi_page_t* page, const mi_block_t* block) { + // The decoded value is in the same page (or NULL). + // Walk the free lists to verify positively if it is already freed + if (mi_list_contains(page, page->free, block) || + mi_list_contains(page, page->local_free, block) || + mi_list_contains(page, mi_page_thread_free(page), block)) + { + _mi_error_message(EAGAIN, "double free detected of block %p with size %zu\n", block, mi_page_block_size(page)); + return true; + } + return false; +} + +#define mi_track_page(page,access) { size_t psize; void* pstart = _mi_page_start(_mi_page_segment(page),page,&psize); mi_track_mem_##access( pstart, psize); } + +static inline bool mi_check_is_double_free(const mi_page_t* page, const mi_block_t* block) { + bool is_double_free = false; + mi_block_t* n = mi_block_nextx(page, block, page->keys); // pretend it is freed, and get the decoded first field + if (((uintptr_t)n & (MI_INTPTR_SIZE-1))==0 && // quick check: aligned pointer? + (n==NULL || mi_is_in_same_page(block, n))) // quick check: in same page or NULL? + { + // Suspicous: decoded value a in block is in the same page (or NULL) -- maybe a double free? + // (continue in separate function to improve code generation) + is_double_free = mi_check_is_double_freex(page, block); + } + return is_double_free; +} +#else +static inline bool mi_check_is_double_free(const mi_page_t* page, const mi_block_t* block) { + MI_UNUSED(page); + MI_UNUSED(block); + return false; +} +#endif + +// --------------------------------------------------------------------------- +// Check for heap block overflow by setting up padding at the end of the block +// --------------------------------------------------------------------------- + +#if MI_PADDING // && !MI_TRACK_ENABLED +static bool mi_page_decode_padding(const mi_page_t* page, const mi_block_t* block, size_t* delta, size_t* bsize) { + *bsize = mi_page_usable_block_size(page); + const mi_padding_t* const padding = (mi_padding_t*)((uint8_t*)block + *bsize); + mi_track_mem_defined(padding,sizeof(mi_padding_t)); + *delta = padding->delta; + uint32_t canary = padding->canary; + uintptr_t keys[2]; + keys[0] = page->keys[0]; + keys[1] = page->keys[1]; + bool ok = ((uint32_t)mi_ptr_encode(page,block,keys) == canary && *delta <= *bsize); + mi_track_mem_noaccess(padding,sizeof(mi_padding_t)); + return ok; +} + +// Return the exact usable size of a block. +static size_t mi_page_usable_size_of(const mi_page_t* page, const mi_block_t* block) { + size_t bsize; + size_t delta; + bool ok = mi_page_decode_padding(page, block, &delta, &bsize); + mi_assert_internal(ok); mi_assert_internal(delta <= bsize); + return (ok ? bsize - delta : 0); +} + +// When a non-thread-local block is freed, it becomes part of the thread delayed free +// list that is freed later by the owning heap. If the exact usable size is too small to +// contain the pointer for the delayed list, then shrink the padding (by decreasing delta) +// so it will later not trigger an overflow error in `mi_free_block`. +void _mi_padding_shrink(const mi_page_t* page, const mi_block_t* block, const size_t min_size) { + size_t bsize; + size_t delta; + bool ok = mi_page_decode_padding(page, block, &delta, &bsize); + mi_assert_internal(ok); + if (!ok || (bsize - delta) >= min_size) return; // usually already enough space + mi_assert_internal(bsize >= min_size); + if (bsize < min_size) return; // should never happen + size_t new_delta = (bsize - min_size); + mi_assert_internal(new_delta < bsize); + mi_padding_t* padding = (mi_padding_t*)((uint8_t*)block + bsize); + mi_track_mem_defined(padding,sizeof(mi_padding_t)); + padding->delta = (uint32_t)new_delta; + mi_track_mem_noaccess(padding,sizeof(mi_padding_t)); +} +#else +static size_t mi_page_usable_size_of(const mi_page_t* page, const mi_block_t* block) { + MI_UNUSED(block); + return mi_page_usable_block_size(page); +} + +void _mi_padding_shrink(const mi_page_t* page, const mi_block_t* block, const size_t min_size) { + MI_UNUSED(page); + MI_UNUSED(block); + MI_UNUSED(min_size); +} +#endif + +#if MI_PADDING && MI_PADDING_CHECK + +static bool mi_verify_padding(const mi_page_t* page, const mi_block_t* block, size_t* size, size_t* wrong) { + size_t bsize; + size_t delta; + bool ok = mi_page_decode_padding(page, block, &delta, &bsize); + *size = *wrong = bsize; + if (!ok) return false; + mi_assert_internal(bsize >= delta); + *size = bsize - delta; + if (!mi_page_is_huge(page)) { + uint8_t* fill = (uint8_t*)block + bsize - delta; + const size_t maxpad = (delta > MI_MAX_ALIGN_SIZE ? MI_MAX_ALIGN_SIZE : delta); // check at most the first N padding bytes + mi_track_mem_defined(fill, maxpad); + for (size_t i = 0; i < maxpad; i++) { + if (fill[i] != MI_DEBUG_PADDING) { + *wrong = bsize - delta + i; + ok = false; + break; + } + } + mi_track_mem_noaccess(fill, maxpad); + } + return ok; +} + +static void mi_check_padding(const mi_page_t* page, const mi_block_t* block) { + size_t size; + size_t wrong; + if (!mi_verify_padding(page,block,&size,&wrong)) { + _mi_error_message(EFAULT, "buffer overflow in heap block %p of size %zu: write after %zu bytes\n", block, size, wrong ); + } +} + +#else + +static void mi_check_padding(const mi_page_t* page, const mi_block_t* block) { + MI_UNUSED(page); + MI_UNUSED(block); +} + +#endif + +// only maintain stats for smaller objects if requested +#if (MI_STAT>0) +static void mi_stat_free(const mi_page_t* page, const mi_block_t* block) { + #if (MI_STAT < 2) + MI_UNUSED(block); + #endif + mi_heap_t* const heap = mi_heap_get_default(); + const size_t bsize = mi_page_usable_block_size(page); + #if (MI_STAT>1) + const size_t usize = mi_page_usable_size_of(page, block); + mi_heap_stat_decrease(heap, malloc, usize); + #endif + if (bsize <= MI_MEDIUM_OBJ_SIZE_MAX) { + mi_heap_stat_decrease(heap, normal, bsize); + #if (MI_STAT > 1) + mi_heap_stat_decrease(heap, normal_bins[_mi_bin(bsize)], 1); + #endif + } + else if (bsize <= MI_LARGE_OBJ_SIZE_MAX) { + mi_heap_stat_decrease(heap, large, bsize); + } + else { + mi_heap_stat_decrease(heap, huge, bsize); + } +} +#else +static void mi_stat_free(const mi_page_t* page, const mi_block_t* block) { + MI_UNUSED(page); MI_UNUSED(block); +} +#endif + +#if MI_HUGE_PAGE_ABANDON +#if (MI_STAT>0) +// maintain stats for huge objects +static void mi_stat_huge_free(const mi_page_t* page) { + mi_heap_t* const heap = mi_heap_get_default(); + const size_t bsize = mi_page_block_size(page); // to match stats in `page.c:mi_page_huge_alloc` + if (bsize <= MI_LARGE_OBJ_SIZE_MAX) { + mi_heap_stat_decrease(heap, large, bsize); + } + else { + mi_heap_stat_decrease(heap, huge, bsize); + } +} +#else +static void mi_stat_huge_free(const mi_page_t* page) { + MI_UNUSED(page); +} +#endif +#endif + +// ------------------------------------------------------ +// Free +// ------------------------------------------------------ + +// multi-threaded free (or free in huge block if compiled with MI_HUGE_PAGE_ABANDON) +static mi_decl_noinline void _mi_free_block_mt(mi_page_t* page, mi_block_t* block) +{ + // The padding check may access the non-thread-owned page for the key values. + // that is safe as these are constant and the page won't be freed (as the block is not freed yet). + mi_check_padding(page, block); + _mi_padding_shrink(page, block, sizeof(mi_block_t)); // for small size, ensure we can fit the delayed thread pointers without triggering overflow detection + + // huge page segments are always abandoned and can be freed immediately + mi_segment_t* segment = _mi_page_segment(page); + if (segment->kind == MI_SEGMENT_HUGE) { + #if MI_HUGE_PAGE_ABANDON + // huge page segments are always abandoned and can be freed immediately + mi_stat_huge_free(page); + _mi_segment_huge_page_free(segment, page, block); + return; + #else + // huge pages are special as they occupy the entire segment + // as these are large we reset the memory occupied by the page so it is available to other threads + // (as the owning thread needs to actually free the memory later). + _mi_segment_huge_page_reset(segment, page, block); + #endif + } + + #if (MI_DEBUG>0) && !MI_TRACK_ENABLED && !MI_TSAN // note: when tracking, cannot use mi_usable_size with multi-threading + if (segment->kind != MI_SEGMENT_HUGE) { // not for huge segments as we just reset the content + memset(block, MI_DEBUG_FREED, mi_usable_size(block)); + } + #endif + + // Try to put the block on either the page-local thread free list, or the heap delayed free list. + mi_thread_free_t tfreex; + bool use_delayed; + mi_thread_free_t tfree = mi_atomic_load_relaxed(&page->xthread_free); + do { + use_delayed = (mi_tf_delayed(tfree) == MI_USE_DELAYED_FREE); + if mi_unlikely(use_delayed) { + // unlikely: this only happens on the first concurrent free in a page that is in the full list + tfreex = mi_tf_set_delayed(tfree,MI_DELAYED_FREEING); + } + else { + // usual: directly add to page thread_free list + mi_block_set_next(page, block, mi_tf_block(tfree)); + tfreex = mi_tf_set_block(tfree,block); + } + } while (!mi_atomic_cas_weak_release(&page->xthread_free, &tfree, tfreex)); + + if mi_unlikely(use_delayed) { + // racy read on `heap`, but ok because MI_DELAYED_FREEING is set (see `mi_heap_delete` and `mi_heap_collect_abandon`) + mi_heap_t* const heap = (mi_heap_t*)(mi_atomic_load_acquire(&page->xheap)); //mi_page_heap(page); + mi_assert_internal(heap != NULL); + if (heap != NULL) { + // add to the delayed free list of this heap. (do this atomically as the lock only protects heap memory validity) + mi_block_t* dfree = mi_atomic_load_ptr_relaxed(mi_block_t, &heap->thread_delayed_free); + do { + mi_block_set_nextx(heap,block,dfree, heap->keys); + } while (!mi_atomic_cas_ptr_weak_release(mi_block_t,&heap->thread_delayed_free, &dfree, block)); + } + + // and reset the MI_DELAYED_FREEING flag + tfree = mi_atomic_load_relaxed(&page->xthread_free); + do { + tfreex = tfree; + mi_assert_internal(mi_tf_delayed(tfree) == MI_DELAYED_FREEING); + tfreex = mi_tf_set_delayed(tfree,MI_NO_DELAYED_FREE); + } while (!mi_atomic_cas_weak_release(&page->xthread_free, &tfree, tfreex)); + } +} + +// regular free +static inline void _mi_free_block(mi_page_t* page, bool local, mi_block_t* block) +{ + // and push it on the free list + //const size_t bsize = mi_page_block_size(page); + if mi_likely(local) { + // owning thread can free a block directly + if mi_unlikely(mi_check_is_double_free(page, block)) return; + mi_check_padding(page, block); + #if (MI_DEBUG>0) && !MI_TRACK_ENABLED && !MI_TSAN + if (!mi_page_is_huge(page)) { // huge page content may be already decommitted + memset(block, MI_DEBUG_FREED, mi_page_block_size(page)); + } + #endif + mi_block_set_next(page, block, page->local_free); + page->local_free = block; + page->used--; + if mi_unlikely(mi_page_all_free(page)) { + _mi_page_retire(page); + } + else if mi_unlikely(mi_page_is_in_full(page)) { + _mi_page_unfull(page); + } + } + else { + _mi_free_block_mt(page,block); + } +} + + +// Adjust a block that was allocated aligned, to the actual start of the block in the page. +mi_block_t* _mi_page_ptr_unalign(const mi_segment_t* segment, const mi_page_t* page, const void* p) { + mi_assert_internal(page!=NULL && p!=NULL); + const size_t diff = (uint8_t*)p - _mi_page_start(segment, page, NULL); + const size_t adjust = (diff % mi_page_block_size(page)); + return (mi_block_t*)((uintptr_t)p - adjust); +} + + +void mi_decl_noinline _mi_free_generic(const mi_segment_t* segment, mi_page_t* page, bool is_local, void* p) mi_attr_noexcept { + mi_block_t* const block = (mi_page_has_aligned(page) ? _mi_page_ptr_unalign(segment, page, p) : (mi_block_t*)p); + mi_stat_free(page, block); // stat_free may access the padding + mi_track_free_size(block, mi_page_usable_size_of(page,block)); + _mi_free_block(page, is_local, block); +} + +// Get the segment data belonging to a pointer +// This is just a single `and` in assembly but does further checks in debug mode +// (and secure mode) if this was a valid pointer. +static inline mi_segment_t* mi_checked_ptr_segment(const void* p, const char* msg) +{ + MI_UNUSED(msg); + mi_assert(p != NULL); + +#if (MI_DEBUG>0) + if mi_unlikely(((uintptr_t)p & (MI_INTPTR_SIZE - 1)) != 0) { + _mi_error_message(EINVAL, "%s: invalid (unaligned) pointer: %p\n", msg, p); + return NULL; + } +#endif + + mi_segment_t* const segment = _mi_ptr_segment(p); + mi_assert_internal(segment != NULL); + +#if (MI_DEBUG>0) + if mi_unlikely(!mi_is_in_heap_region(p)) { + #if (MI_INTPTR_SIZE == 8 && defined(__linux__)) + if (((uintptr_t)p >> 40) != 0x7F) { // linux tends to align large blocks above 0x7F000000000 (issue #640) + #else + { + #endif + _mi_warning_message("%s: pointer might not point to a valid heap region: %p\n" + "(this may still be a valid very large allocation (over 64MiB))\n", msg, p); + if mi_likely(_mi_ptr_cookie(segment) == segment->cookie) { + _mi_warning_message("(yes, the previous pointer %p was valid after all)\n", p); + } + } + } +#endif +#if (MI_DEBUG>0 || MI_SECURE>=4) + if mi_unlikely(_mi_ptr_cookie(segment) != segment->cookie) { + _mi_error_message(EINVAL, "%s: pointer does not point to a valid heap space: %p\n", msg, p); + return NULL; + } +#endif + + return segment; +} + +// Free a block +// fast path written carefully to prevent spilling on the stack +void mi_free(void* p) mi_attr_noexcept +{ + if mi_unlikely(p == NULL) return; + mi_segment_t* const segment = mi_checked_ptr_segment(p,"mi_free"); + const bool is_local= (_mi_prim_thread_id() == mi_atomic_load_relaxed(&segment->thread_id)); + mi_page_t* const page = _mi_segment_page_of(segment, p); + + if mi_likely(is_local) { // thread-local free? + if mi_likely(page->flags.full_aligned == 0) // and it is not a full page (full pages need to move from the full bin), nor has aligned blocks (aligned blocks need to be unaligned) + { + mi_block_t* const block = (mi_block_t*)p; + if mi_unlikely(mi_check_is_double_free(page, block)) return; + mi_check_padding(page, block); + mi_stat_free(page, block); + #if (MI_DEBUG>0) && !MI_TRACK_ENABLED && !MI_TSAN + memset(block, MI_DEBUG_FREED, mi_page_block_size(page)); + #endif + mi_track_free_size(p, mi_page_usable_size_of(page,block)); // faster then mi_usable_size as we already know the page and that p is unaligned + mi_block_set_next(page, block, page->local_free); + page->local_free = block; + if mi_unlikely(--page->used == 0) { // using this expression generates better code than: page->used--; if (mi_page_all_free(page)) + _mi_page_retire(page); + } + } + else { + // page is full or contains (inner) aligned blocks; use generic path + _mi_free_generic(segment, page, true, p); + } + } + else { + // not thread-local; use generic path + _mi_free_generic(segment, page, false, p); + } +} + +// return true if successful +bool _mi_free_delayed_block(mi_block_t* block) { + // get segment and page + const mi_segment_t* const segment = _mi_ptr_segment(block); + mi_assert_internal(_mi_ptr_cookie(segment) == segment->cookie); + mi_assert_internal(_mi_thread_id() == segment->thread_id); + mi_page_t* const page = _mi_segment_page_of(segment, block); + + // Clear the no-delayed flag so delayed freeing is used again for this page. + // This must be done before collecting the free lists on this page -- otherwise + // some blocks may end up in the page `thread_free` list with no blocks in the + // heap `thread_delayed_free` list which may cause the page to be never freed! + // (it would only be freed if we happen to scan it in `mi_page_queue_find_free_ex`) + if (!_mi_page_try_use_delayed_free(page, MI_USE_DELAYED_FREE, false /* dont overwrite never delayed */)) { + return false; + } + + // collect all other non-local frees to ensure up-to-date `used` count + _mi_page_free_collect(page, false); + + // and free the block (possibly freeing the page as well since used is updated) + _mi_free_block(page, true, block); + return true; +} + +// Bytes available in a block +mi_decl_noinline static size_t mi_page_usable_aligned_size_of(const mi_segment_t* segment, const mi_page_t* page, const void* p) mi_attr_noexcept { + const mi_block_t* block = _mi_page_ptr_unalign(segment, page, p); + const size_t size = mi_page_usable_size_of(page, block); + const ptrdiff_t adjust = (uint8_t*)p - (uint8_t*)block; + mi_assert_internal(adjust >= 0 && (size_t)adjust <= size); + return (size - adjust); +} + +static inline size_t _mi_usable_size(const void* p, const char* msg) mi_attr_noexcept { + if (p == NULL) return 0; + const mi_segment_t* const segment = mi_checked_ptr_segment(p, msg); + const mi_page_t* const page = _mi_segment_page_of(segment, p); + if mi_likely(!mi_page_has_aligned(page)) { + const mi_block_t* block = (const mi_block_t*)p; + return mi_page_usable_size_of(page, block); + } + else { + // split out to separate routine for improved code generation + return mi_page_usable_aligned_size_of(segment, page, p); + } +} + +mi_decl_nodiscard size_t mi_usable_size(const void* p) mi_attr_noexcept { + return _mi_usable_size(p, "mi_usable_size"); +} + + +// ------------------------------------------------------ +// Allocation extensions +// ------------------------------------------------------ + +void mi_free_size(void* p, size_t size) mi_attr_noexcept { + MI_UNUSED_RELEASE(size); + mi_assert(p == NULL || size <= _mi_usable_size(p,"mi_free_size")); + mi_free(p); +} + +void mi_free_size_aligned(void* p, size_t size, size_t alignment) mi_attr_noexcept { + MI_UNUSED_RELEASE(alignment); + mi_assert(((uintptr_t)p % alignment) == 0); + mi_free_size(p,size); +} + +void mi_free_aligned(void* p, size_t alignment) mi_attr_noexcept { + MI_UNUSED_RELEASE(alignment); + mi_assert(((uintptr_t)p % alignment) == 0); + mi_free(p); +} + +mi_decl_nodiscard extern inline mi_decl_restrict void* mi_heap_calloc(mi_heap_t* heap, size_t count, size_t size) mi_attr_noexcept { + size_t total; + if (mi_count_size_overflow(count,size,&total)) return NULL; + return mi_heap_zalloc(heap,total); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_calloc(size_t count, size_t size) mi_attr_noexcept { + return mi_heap_calloc(mi_prim_get_default_heap(),count,size); +} + +// Uninitialized `calloc` +mi_decl_nodiscard extern mi_decl_restrict void* mi_heap_mallocn(mi_heap_t* heap, size_t count, size_t size) mi_attr_noexcept { + size_t total; + if (mi_count_size_overflow(count, size, &total)) return NULL; + return mi_heap_malloc(heap, total); +} + +mi_decl_nodiscard mi_decl_restrict void* mi_mallocn(size_t count, size_t size) mi_attr_noexcept { + return mi_heap_mallocn(mi_prim_get_default_heap(),count,size); +} + +// Expand (or shrink) in place (or fail) +void* mi_expand(void* p, size_t newsize) mi_attr_noexcept { + #if MI_PADDING + // we do not shrink/expand with padding enabled + MI_UNUSED(p); MI_UNUSED(newsize); + return NULL; + #else + if (p == NULL) return NULL; + const size_t size = _mi_usable_size(p,"mi_expand"); + if (newsize > size) return NULL; + return p; // it fits + #endif +} + +void* _mi_heap_realloc_zero(mi_heap_t* heap, void* p, size_t newsize, bool zero) mi_attr_noexcept { + // if p == NULL then behave as malloc. + // else if size == 0 then reallocate to a zero-sized block (and don't return NULL, just as mi_malloc(0)). + // (this means that returning NULL always indicates an error, and `p` will not have been freed in that case.) + const size_t size = _mi_usable_size(p,"mi_realloc"); // also works if p == NULL (with size 0) + if mi_unlikely(newsize <= size && newsize >= (size / 2) && newsize > 0) { // note: newsize must be > 0 or otherwise we return NULL for realloc(NULL,0) + mi_assert_internal(p!=NULL); + // todo: do not track as the usable size is still the same in the free; adjust potential padding? + // mi_track_resize(p,size,newsize) + // if (newsize < size) { mi_track_mem_noaccess((uint8_t*)p + newsize, size - newsize); } + return p; // reallocation still fits and not more than 50% waste + } + void* newp = mi_heap_malloc(heap,newsize); + if mi_likely(newp != NULL) { + if (zero && newsize > size) { + // also set last word in the previous allocation to zero to ensure any padding is zero-initialized + const size_t start = (size >= sizeof(intptr_t) ? size - sizeof(intptr_t) : 0); + _mi_memzero((uint8_t*)newp + start, newsize - start); + } + else if (newsize == 0) { + ((uint8_t*)newp)[0] = 0; // work around for applications that expect zero-reallocation to be zero initialized (issue #725) + } + if mi_likely(p != NULL) { + const size_t copysize = (newsize > size ? size : newsize); + mi_track_mem_defined(p,copysize); // _mi_useable_size may be too large for byte precise memory tracking.. + _mi_memcpy(newp, p, copysize); + mi_free(p); // only free the original pointer if successful + } + } + return newp; +} + +mi_decl_nodiscard void* mi_heap_realloc(mi_heap_t* heap, void* p, size_t newsize) mi_attr_noexcept { + return _mi_heap_realloc_zero(heap, p, newsize, false); +} + +mi_decl_nodiscard void* mi_heap_reallocn(mi_heap_t* heap, void* p, size_t count, size_t size) mi_attr_noexcept { + size_t total; + if (mi_count_size_overflow(count, size, &total)) return NULL; + return mi_heap_realloc(heap, p, total); +} + + +// Reallocate but free `p` on errors +mi_decl_nodiscard void* mi_heap_reallocf(mi_heap_t* heap, void* p, size_t newsize) mi_attr_noexcept { + void* newp = mi_heap_realloc(heap, p, newsize); + if (newp==NULL && p!=NULL) mi_free(p); + return newp; +} + +mi_decl_nodiscard void* mi_heap_rezalloc(mi_heap_t* heap, void* p, size_t newsize) mi_attr_noexcept { + return _mi_heap_realloc_zero(heap, p, newsize, true); +} + +mi_decl_nodiscard void* mi_heap_recalloc(mi_heap_t* heap, void* p, size_t count, size_t size) mi_attr_noexcept { + size_t total; + if (mi_count_size_overflow(count, size, &total)) return NULL; + return mi_heap_rezalloc(heap, p, total); +} + + +mi_decl_nodiscard void* mi_realloc(void* p, size_t newsize) mi_attr_noexcept { + return mi_heap_realloc(mi_prim_get_default_heap(),p,newsize); +} + +mi_decl_nodiscard void* mi_reallocn(void* p, size_t count, size_t size) mi_attr_noexcept { + return mi_heap_reallocn(mi_prim_get_default_heap(),p,count,size); +} + +// Reallocate but free `p` on errors +mi_decl_nodiscard void* mi_reallocf(void* p, size_t newsize) mi_attr_noexcept { + return mi_heap_reallocf(mi_prim_get_default_heap(),p,newsize); +} + +mi_decl_nodiscard void* mi_rezalloc(void* p, size_t newsize) mi_attr_noexcept { + return mi_heap_rezalloc(mi_prim_get_default_heap(), p, newsize); +} + +mi_decl_nodiscard void* mi_recalloc(void* p, size_t count, size_t size) mi_attr_noexcept { + return mi_heap_recalloc(mi_prim_get_default_heap(), p, count, size); +} + + + +// ------------------------------------------------------ +// strdup, strndup, and realpath +// ------------------------------------------------------ + +// `strdup` using mi_malloc +mi_decl_nodiscard mi_decl_restrict char* mi_heap_strdup(mi_heap_t* heap, const char* s) mi_attr_noexcept { + if (s == NULL) return NULL; + size_t n = strlen(s); + char* t = (char*)mi_heap_malloc(heap,n+1); + if (t == NULL) return NULL; + _mi_memcpy(t, s, n); + t[n] = 0; + return t; +} + +mi_decl_nodiscard mi_decl_restrict char* mi_strdup(const char* s) mi_attr_noexcept { + return mi_heap_strdup(mi_prim_get_default_heap(), s); +} + +// `strndup` using mi_malloc +mi_decl_nodiscard mi_decl_restrict char* mi_heap_strndup(mi_heap_t* heap, const char* s, size_t n) mi_attr_noexcept { + if (s == NULL) return NULL; + const char* end = (const char*)memchr(s, 0, n); // find end of string in the first `n` characters (returns NULL if not found) + const size_t m = (end != NULL ? (size_t)(end - s) : n); // `m` is the minimum of `n` or the end-of-string + mi_assert_internal(m <= n); + char* t = (char*)mi_heap_malloc(heap, m+1); + if (t == NULL) return NULL; + _mi_memcpy(t, s, m); + t[m] = 0; + return t; +} + +mi_decl_nodiscard mi_decl_restrict char* mi_strndup(const char* s, size_t n) mi_attr_noexcept { + return mi_heap_strndup(mi_prim_get_default_heap(),s,n); +} + +#ifndef __wasi__ +// `realpath` using mi_malloc +#ifdef _WIN32 +#ifndef PATH_MAX +#define PATH_MAX MAX_PATH +#endif +#include +mi_decl_nodiscard mi_decl_restrict char* mi_heap_realpath(mi_heap_t* heap, const char* fname, char* resolved_name) mi_attr_noexcept { + // todo: use GetFullPathNameW to allow longer file names + char buf[PATH_MAX]; + DWORD res = GetFullPathNameA(fname, PATH_MAX, (resolved_name == NULL ? buf : resolved_name), NULL); + if (res == 0) { + errno = GetLastError(); return NULL; + } + else if (res > PATH_MAX) { + errno = EINVAL; return NULL; + } + else if (resolved_name != NULL) { + return resolved_name; + } + else { + return mi_heap_strndup(heap, buf, PATH_MAX); + } +} +#else +/* +#include // pathconf +static size_t mi_path_max(void) { + static size_t path_max = 0; + if (path_max <= 0) { + long m = pathconf("/",_PC_PATH_MAX); + if (m <= 0) path_max = 4096; // guess + else if (m < 256) path_max = 256; // at least 256 + else path_max = m; + } + return path_max; +} +*/ +char* mi_heap_realpath(mi_heap_t* heap, const char* fname, char* resolved_name) mi_attr_noexcept { + if (resolved_name != NULL) { + return realpath(fname,resolved_name); + } + else { + char* rname = realpath(fname, NULL); + if (rname == NULL) return NULL; + char* result = mi_heap_strdup(heap, rname); + free(rname); // use regular free! (which may be redirected to our free but that's ok) + return result; + } + /* + const size_t n = mi_path_max(); + char* buf = (char*)mi_malloc(n+1); + if (buf == NULL) { + errno = ENOMEM; + return NULL; + } + char* rname = realpath(fname,buf); + char* result = mi_heap_strndup(heap,rname,n); // ok if `rname==NULL` + mi_free(buf); + return result; + } + */ +} +#endif + +mi_decl_nodiscard mi_decl_restrict char* mi_realpath(const char* fname, char* resolved_name) mi_attr_noexcept { + return mi_heap_realpath(mi_prim_get_default_heap(),fname,resolved_name); +} +#endif + +/*------------------------------------------------------- +C++ new and new_aligned +The standard requires calling into `get_new_handler` and +throwing the bad_alloc exception on failure. If we compile +with a C++ compiler we can implement this precisely. If we +use a C compiler we cannot throw a `bad_alloc` exception +but we call `exit` instead (i.e. not returning). +-------------------------------------------------------*/ + +#ifdef __cplusplus +#include +static bool mi_try_new_handler(bool nothrow) { + #if defined(_MSC_VER) || (__cplusplus >= 201103L) + std::new_handler h = std::get_new_handler(); + #else + std::new_handler h = std::set_new_handler(); + std::set_new_handler(h); + #endif + if (h==NULL) { + _mi_error_message(ENOMEM, "out of memory in 'new'"); + if (!nothrow) { + throw std::bad_alloc(); + } + return false; + } + else { + h(); + return true; + } +} +#else +typedef void (*std_new_handler_t)(void); + +#if (defined(__GNUC__) || (defined(__clang__) && !defined(_MSC_VER))) // exclude clang-cl, see issue #631 +std_new_handler_t __attribute__((weak)) _ZSt15get_new_handlerv(void) { + return NULL; +} +static std_new_handler_t mi_get_new_handler(void) { + return _ZSt15get_new_handlerv(); +} +#else +// note: on windows we could dynamically link to `?get_new_handler@std@@YAP6AXXZXZ`. +static std_new_handler_t mi_get_new_handler() { + return NULL; +} +#endif + +static bool mi_try_new_handler(bool nothrow) { + std_new_handler_t h = mi_get_new_handler(); + if (h==NULL) { + _mi_error_message(ENOMEM, "out of memory in 'new'"); + if (!nothrow) { + abort(); // cannot throw in plain C, use abort + } + return false; + } + else { + h(); + return true; + } +} +#endif + +mi_decl_export mi_decl_noinline void* mi_heap_try_new(mi_heap_t* heap, size_t size, bool nothrow ) { + void* p = NULL; + while(p == NULL && mi_try_new_handler(nothrow)) { + p = mi_heap_malloc(heap,size); + } + return p; +} + +static mi_decl_noinline void* mi_try_new(size_t size, bool nothrow) { + return mi_heap_try_new(mi_prim_get_default_heap(), size, nothrow); +} + + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_alloc_new(mi_heap_t* heap, size_t size) { + void* p = mi_heap_malloc(heap,size); + if mi_unlikely(p == NULL) return mi_heap_try_new(heap, size, false); + return p; +} + +mi_decl_nodiscard mi_decl_restrict void* mi_new(size_t size) { + return mi_heap_alloc_new(mi_prim_get_default_heap(), size); +} + + +mi_decl_nodiscard mi_decl_restrict void* mi_heap_alloc_new_n(mi_heap_t* heap, size_t count, size_t size) { + size_t total; + if mi_unlikely(mi_count_size_overflow(count, size, &total)) { + mi_try_new_handler(false); // on overflow we invoke the try_new_handler once to potentially throw std::bad_alloc + return NULL; + } + else { + return mi_heap_alloc_new(heap,total); + } +} + +mi_decl_nodiscard mi_decl_restrict void* mi_new_n(size_t count, size_t size) { + return mi_heap_alloc_new_n(mi_prim_get_default_heap(), size, count); +} + + +mi_decl_nodiscard mi_decl_restrict void* mi_new_nothrow(size_t size) mi_attr_noexcept { + void* p = mi_malloc(size); + if mi_unlikely(p == NULL) return mi_try_new(size, true); + return p; +} + +mi_decl_nodiscard mi_decl_restrict void* mi_new_aligned(size_t size, size_t alignment) { + void* p; + do { + p = mi_malloc_aligned(size, alignment); + } + while(p == NULL && mi_try_new_handler(false)); + return p; +} + +mi_decl_nodiscard mi_decl_restrict void* mi_new_aligned_nothrow(size_t size, size_t alignment) mi_attr_noexcept { + void* p; + do { + p = mi_malloc_aligned(size, alignment); + } + while(p == NULL && mi_try_new_handler(true)); + return p; +} + +mi_decl_nodiscard void* mi_new_realloc(void* p, size_t newsize) { + void* q; + do { + q = mi_realloc(p, newsize); + } while (q == NULL && mi_try_new_handler(false)); + return q; +} + +mi_decl_nodiscard void* mi_new_reallocn(void* p, size_t newcount, size_t size) { + size_t total; + if mi_unlikely(mi_count_size_overflow(newcount, size, &total)) { + mi_try_new_handler(false); // on overflow we invoke the try_new_handler once to potentially throw std::bad_alloc + return NULL; + } + else { + return mi_new_realloc(p, total); + } +} + +// ------------------------------------------------------ +// ensure explicit external inline definitions are emitted! +// ------------------------------------------------------ + +#ifdef __cplusplus +void* _mi_externs[] = { + (void*)&_mi_page_malloc, + (void*)&_mi_heap_malloc_zero, + (void*)&_mi_heap_malloc_zero_ex, + (void*)&mi_malloc, + (void*)&mi_malloc_small, + (void*)&mi_zalloc_small, + (void*)&mi_heap_malloc, + (void*)&mi_heap_zalloc, + (void*)&mi_heap_malloc_small, + // (void*)&mi_heap_alloc_new, + // (void*)&mi_heap_alloc_new_n +}; +#endif diff --git a/compat/mimalloc/arena.c b/compat/mimalloc/arena.c new file mode 100644 index 00000000000000..879ee9e7e773d4 --- /dev/null +++ b/compat/mimalloc/arena.c @@ -0,0 +1,935 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2019-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +/* ---------------------------------------------------------------------------- +"Arenas" are fixed area's of OS memory from which we can allocate +large blocks (>= MI_ARENA_MIN_BLOCK_SIZE, 4MiB). +In contrast to the rest of mimalloc, the arenas are shared between +threads and need to be accessed using atomic operations. + +Arenas are used to for huge OS page (1GiB) reservations or for reserving +OS memory upfront which can be improve performance or is sometimes needed +on embedded devices. We can also employ this with WASI or `sbrk` systems +to reserve large arenas upfront and be able to reuse the memory more effectively. + +The arena allocation needs to be thread safe and we use an atomic bitmap to allocate. +-----------------------------------------------------------------------------*/ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" + +#include // memset +#include // ENOMEM + +#include "bitmap.h" // atomic bitmap + +/* ----------------------------------------------------------- + Arena allocation +----------------------------------------------------------- */ + +// Block info: bit 0 contains the `in_use` bit, the upper bits the +// size in count of arena blocks. +typedef uintptr_t mi_block_info_t; +#define MI_ARENA_BLOCK_SIZE (MI_SEGMENT_SIZE) // 64MiB (must be at least MI_SEGMENT_ALIGN) +#define MI_ARENA_MIN_OBJ_SIZE (MI_ARENA_BLOCK_SIZE/2) // 32MiB +#define MI_MAX_ARENAS (112) // not more than 126 (since we use 7 bits in the memid and an arena index + 1) + +// A memory arena descriptor +typedef struct mi_arena_s { + mi_arena_id_t id; // arena id; 0 for non-specific + mi_memid_t memid; // memid of the memory area + _Atomic(uint8_t*) start; // the start of the memory area + size_t block_count; // size of the area in arena blocks (of `MI_ARENA_BLOCK_SIZE`) + size_t field_count; // number of bitmap fields (where `field_count * MI_BITMAP_FIELD_BITS >= block_count`) + size_t meta_size; // size of the arena structure itself (including its bitmaps) + mi_memid_t meta_memid; // memid of the arena structure itself (OS or static allocation) + int numa_node; // associated NUMA node + bool exclusive; // only allow allocations if specifically for this arena + bool is_large; // memory area consists of large- or huge OS pages (always committed) + _Atomic(size_t) search_idx; // optimization to start the search for free blocks + _Atomic(mi_msecs_t) purge_expire; // expiration time when blocks should be decommitted from `blocks_decommit`. + mi_bitmap_field_t* blocks_dirty; // are the blocks potentially non-zero? + mi_bitmap_field_t* blocks_committed; // are the blocks committed? (can be NULL for memory that cannot be decommitted) + mi_bitmap_field_t* blocks_purge; // blocks that can be (reset) decommitted. (can be NULL for memory that cannot be (reset) decommitted) + mi_bitmap_field_t blocks_inuse[1]; // in-place bitmap of in-use blocks (of size `field_count`) +} mi_arena_t; + + +// The available arenas +static mi_decl_cache_align _Atomic(mi_arena_t*) mi_arenas[MI_MAX_ARENAS]; +static mi_decl_cache_align _Atomic(size_t) mi_arena_count; // = 0 + + +//static bool mi_manage_os_memory_ex2(void* start, size_t size, bool is_large, int numa_node, bool exclusive, mi_memid_t memid, mi_arena_id_t* arena_id) mi_attr_noexcept; + +/* ----------------------------------------------------------- + Arena id's + id = arena_index + 1 +----------------------------------------------------------- */ + +static size_t mi_arena_id_index(mi_arena_id_t id) { + return (size_t)(id <= 0 ? MI_MAX_ARENAS : id - 1); +} + +static mi_arena_id_t mi_arena_id_create(size_t arena_index) { + mi_assert_internal(arena_index < MI_MAX_ARENAS); + return (int)arena_index + 1; +} + +mi_arena_id_t _mi_arena_id_none(void) { + return 0; +} + +static bool mi_arena_id_is_suitable(mi_arena_id_t arena_id, bool arena_is_exclusive, mi_arena_id_t req_arena_id) { + return ((!arena_is_exclusive && req_arena_id == _mi_arena_id_none()) || + (arena_id == req_arena_id)); +} + +bool _mi_arena_memid_is_suitable(mi_memid_t memid, mi_arena_id_t request_arena_id) { + if (memid.memkind == MI_MEM_ARENA) { + return mi_arena_id_is_suitable(memid.mem.arena.id, memid.mem.arena.is_exclusive, request_arena_id); + } + else { + return mi_arena_id_is_suitable(0, false, request_arena_id); + } +} + +bool _mi_arena_memid_is_os_allocated(mi_memid_t memid) { + return (memid.memkind == MI_MEM_OS); +} + +/* ----------------------------------------------------------- + Arena allocations get a (currently) 16-bit memory id where the + lower 8 bits are the arena id, and the upper bits the block index. +----------------------------------------------------------- */ + +static size_t mi_block_count_of_size(size_t size) { + return _mi_divide_up(size, MI_ARENA_BLOCK_SIZE); +} + +static size_t mi_arena_block_size(size_t bcount) { + return (bcount * MI_ARENA_BLOCK_SIZE); +} + +static size_t mi_arena_size(mi_arena_t* arena) { + return mi_arena_block_size(arena->block_count); +} + +static mi_memid_t mi_memid_create_arena(mi_arena_id_t id, bool is_exclusive, mi_bitmap_index_t bitmap_index) { + mi_memid_t memid = _mi_memid_create(MI_MEM_ARENA); + memid.mem.arena.id = id; + memid.mem.arena.block_index = bitmap_index; + memid.mem.arena.is_exclusive = is_exclusive; + return memid; +} + +static bool mi_arena_memid_indices(mi_memid_t memid, size_t* arena_index, mi_bitmap_index_t* bitmap_index) { + mi_assert_internal(memid.memkind == MI_MEM_ARENA); + *arena_index = mi_arena_id_index(memid.mem.arena.id); + *bitmap_index = memid.mem.arena.block_index; + return memid.mem.arena.is_exclusive; +} + + + +/* ----------------------------------------------------------- + Special static area for mimalloc internal structures + to avoid OS calls (for example, for the arena metadata) +----------------------------------------------------------- */ + +#define MI_ARENA_STATIC_MAX (MI_INTPTR_SIZE*MI_KiB) // 8 KiB on 64-bit + +static uint8_t mi_arena_static[MI_ARENA_STATIC_MAX]; +static _Atomic(size_t) mi_arena_static_top; + +static void* mi_arena_static_zalloc(size_t size, size_t alignment, mi_memid_t* memid) { + *memid = _mi_memid_none(); + if (size == 0 || size > MI_ARENA_STATIC_MAX) return NULL; + if ((mi_atomic_load_relaxed(&mi_arena_static_top) + size) > MI_ARENA_STATIC_MAX) return NULL; + + // try to claim space + if (alignment == 0) { alignment = 1; } + const size_t oversize = size + alignment - 1; + if (oversize > MI_ARENA_STATIC_MAX) return NULL; + const size_t oldtop = mi_atomic_add_acq_rel(&mi_arena_static_top, oversize); + size_t top = oldtop + oversize; + if (top > MI_ARENA_STATIC_MAX) { + // try to roll back, ok if this fails + mi_atomic_cas_strong_acq_rel(&mi_arena_static_top, &top, oldtop); + return NULL; + } + + // success + *memid = _mi_memid_create(MI_MEM_STATIC); + const size_t start = _mi_align_up(oldtop, alignment); + uint8_t* const p = &mi_arena_static[start]; + _mi_memzero(p, size); + return p; +} + +static void* mi_arena_meta_zalloc(size_t size, mi_memid_t* memid, mi_stats_t* stats) { + *memid = _mi_memid_none(); + + // try static + void* p = mi_arena_static_zalloc(size, MI_ALIGNMENT_MAX, memid); + if (p != NULL) return p; + + // or fall back to the OS + return _mi_os_alloc(size, memid, stats); +} + +static void mi_arena_meta_free(void* p, mi_memid_t memid, size_t size, mi_stats_t* stats) { + if (mi_memkind_is_os(memid.memkind)) { + _mi_os_free(p, size, memid, stats); + } + else { + mi_assert(memid.memkind == MI_MEM_STATIC); + } +} + +static void* mi_arena_block_start(mi_arena_t* arena, mi_bitmap_index_t bindex) { + return (arena->start + mi_arena_block_size(mi_bitmap_index_bit(bindex))); +} + + +/* ----------------------------------------------------------- + Thread safe allocation in an arena +----------------------------------------------------------- */ + +// claim the `blocks_inuse` bits +static bool mi_arena_try_claim(mi_arena_t* arena, size_t blocks, mi_bitmap_index_t* bitmap_idx) +{ + size_t idx = 0; // mi_atomic_load_relaxed(&arena->search_idx); // start from last search; ok to be relaxed as the exact start does not matter + if (_mi_bitmap_try_find_from_claim_across(arena->blocks_inuse, arena->field_count, idx, blocks, bitmap_idx)) { + mi_atomic_store_relaxed(&arena->search_idx, mi_bitmap_index_field(*bitmap_idx)); // start search from found location next time around + return true; + }; + return false; +} + + +/* ----------------------------------------------------------- + Arena Allocation +----------------------------------------------------------- */ + +static mi_decl_noinline void* mi_arena_try_alloc_at(mi_arena_t* arena, size_t arena_index, size_t needed_bcount, + bool commit, mi_memid_t* memid, mi_os_tld_t* tld) +{ + MI_UNUSED(arena_index); + mi_assert_internal(mi_arena_id_index(arena->id) == arena_index); + + mi_bitmap_index_t bitmap_index; + if (!mi_arena_try_claim(arena, needed_bcount, &bitmap_index)) return NULL; + + // claimed it! + void* p = mi_arena_block_start(arena, bitmap_index); + *memid = mi_memid_create_arena(arena->id, arena->exclusive, bitmap_index); + memid->is_pinned = arena->memid.is_pinned; + + // none of the claimed blocks should be scheduled for a decommit + if (arena->blocks_purge != NULL) { + // this is thread safe as a potential purge only decommits parts that are not yet claimed as used (in `blocks_inuse`). + _mi_bitmap_unclaim_across(arena->blocks_purge, arena->field_count, needed_bcount, bitmap_index); + } + + // set the dirty bits (todo: no need for an atomic op here?) + if (arena->memid.initially_zero && arena->blocks_dirty != NULL) { + memid->initially_zero = _mi_bitmap_claim_across(arena->blocks_dirty, arena->field_count, needed_bcount, bitmap_index, NULL); + } + + // set commit state + if (arena->blocks_committed == NULL) { + // always committed + memid->initially_committed = true; + } + else if (commit) { + // commit requested, but the range may not be committed as a whole: ensure it is committed now + memid->initially_committed = true; + bool any_uncommitted; + _mi_bitmap_claim_across(arena->blocks_committed, arena->field_count, needed_bcount, bitmap_index, &any_uncommitted); + if (any_uncommitted) { + bool commit_zero = false; + if (!_mi_os_commit(p, mi_arena_block_size(needed_bcount), &commit_zero, tld->stats)) { + memid->initially_committed = false; + } + else { + if (commit_zero) { memid->initially_zero = true; } + } + } + } + else { + // no need to commit, but check if already fully committed + memid->initially_committed = _mi_bitmap_is_claimed_across(arena->blocks_committed, arena->field_count, needed_bcount, bitmap_index); + } + + return p; +} + +// allocate in a speficic arena +static void* mi_arena_try_alloc_at_id(mi_arena_id_t arena_id, bool match_numa_node, int numa_node, size_t size, size_t alignment, + bool commit, bool allow_large, mi_arena_id_t req_arena_id, mi_memid_t* memid, mi_os_tld_t* tld ) +{ + MI_UNUSED_RELEASE(alignment); + mi_assert_internal(alignment <= MI_SEGMENT_ALIGN); + const size_t bcount = mi_block_count_of_size(size); + const size_t arena_index = mi_arena_id_index(arena_id); + mi_assert_internal(arena_index < mi_atomic_load_relaxed(&mi_arena_count)); + mi_assert_internal(size <= mi_arena_block_size(bcount)); + + // Check arena suitability + mi_arena_t* arena = mi_atomic_load_ptr_acquire(mi_arena_t, &mi_arenas[arena_index]); + if (arena == NULL) return NULL; + if (!allow_large && arena->is_large) return NULL; + if (!mi_arena_id_is_suitable(arena->id, arena->exclusive, req_arena_id)) return NULL; + if (req_arena_id == _mi_arena_id_none()) { // in not specific, check numa affinity + const bool numa_suitable = (numa_node < 0 || arena->numa_node < 0 || arena->numa_node == numa_node); + if (match_numa_node) { if (!numa_suitable) return NULL; } + else { if (numa_suitable) return NULL; } + } + + // try to allocate + void* p = mi_arena_try_alloc_at(arena, arena_index, bcount, commit, memid, tld); + mi_assert_internal(p == NULL || _mi_is_aligned(p, alignment)); + return p; +} + + +// allocate from an arena with fallback to the OS +static mi_decl_noinline void* mi_arena_try_alloc(int numa_node, size_t size, size_t alignment, + bool commit, bool allow_large, + mi_arena_id_t req_arena_id, mi_memid_t* memid, mi_os_tld_t* tld ) +{ + MI_UNUSED(alignment); + mi_assert_internal(alignment <= MI_SEGMENT_ALIGN); + const size_t max_arena = mi_atomic_load_relaxed(&mi_arena_count); + if mi_likely(max_arena == 0) return NULL; + + if (req_arena_id != _mi_arena_id_none()) { + // try a specific arena if requested + if (mi_arena_id_index(req_arena_id) < max_arena) { + void* p = mi_arena_try_alloc_at_id(req_arena_id, true, numa_node, size, alignment, commit, allow_large, req_arena_id, memid, tld); + if (p != NULL) return p; + } + } + else { + // try numa affine allocation + for (size_t i = 0; i < max_arena; i++) { + void* p = mi_arena_try_alloc_at_id(mi_arena_id_create(i), true, numa_node, size, alignment, commit, allow_large, req_arena_id, memid, tld); + if (p != NULL) return p; + } + + // try from another numa node instead.. + if (numa_node >= 0) { // if numa_node was < 0 (no specific affinity requested), all arena's have been tried already + for (size_t i = 0; i < max_arena; i++) { + void* p = mi_arena_try_alloc_at_id(mi_arena_id_create(i), false /* only proceed if not numa local */, numa_node, size, alignment, commit, allow_large, req_arena_id, memid, tld); + if (p != NULL) return p; + } + } + } + return NULL; +} + +// try to reserve a fresh arena space +static bool mi_arena_reserve(size_t req_size, bool allow_large, mi_arena_id_t req_arena_id, mi_arena_id_t *arena_id) +{ + if (_mi_preloading()) return false; // use OS only while pre loading + if (req_arena_id != _mi_arena_id_none()) return false; + + const size_t arena_count = mi_atomic_load_acquire(&mi_arena_count); + if (arena_count > (MI_MAX_ARENAS - 4)) return false; + + size_t arena_reserve = mi_option_get_size(mi_option_arena_reserve); + if (arena_reserve == 0) return false; + + if (!_mi_os_has_virtual_reserve()) { + arena_reserve = arena_reserve/4; // be conservative if virtual reserve is not supported (for some embedded systems for example) + } + arena_reserve = _mi_align_up(arena_reserve, MI_ARENA_BLOCK_SIZE); + if (arena_count >= 8 && arena_count <= 128) { + arena_reserve = ((size_t)1<<(arena_count/8)) * arena_reserve; // scale up the arena sizes exponentially + } + if (arena_reserve < req_size) return false; // should be able to at least handle the current allocation size + + // commit eagerly? + bool arena_commit = false; + if (mi_option_get(mi_option_arena_eager_commit) == 2) { arena_commit = _mi_os_has_overcommit(); } + else if (mi_option_get(mi_option_arena_eager_commit) == 1) { arena_commit = true; } + + return (mi_reserve_os_memory_ex(arena_reserve, arena_commit, allow_large, false /* exclusive */, arena_id) == 0); +} + + +void* _mi_arena_alloc_aligned(size_t size, size_t alignment, size_t align_offset, bool commit, bool allow_large, + mi_arena_id_t req_arena_id, mi_memid_t* memid, mi_os_tld_t* tld) +{ + mi_assert_internal(memid != NULL && tld != NULL); + mi_assert_internal(size > 0); + *memid = _mi_memid_none(); + + const int numa_node = _mi_os_numa_node(tld); // current numa node + + // try to allocate in an arena if the alignment is small enough and the object is not too small (as for heap meta data) + if (size >= MI_ARENA_MIN_OBJ_SIZE && alignment <= MI_SEGMENT_ALIGN && align_offset == 0) { + void* p = mi_arena_try_alloc(numa_node, size, alignment, commit, allow_large, req_arena_id, memid, tld); + if (p != NULL) return p; + + // otherwise, try to first eagerly reserve a new arena + if (req_arena_id == _mi_arena_id_none()) { + mi_arena_id_t arena_id = 0; + if (mi_arena_reserve(size, allow_large, req_arena_id, &arena_id)) { + // and try allocate in there + mi_assert_internal(req_arena_id == _mi_arena_id_none()); + p = mi_arena_try_alloc_at_id(arena_id, true, numa_node, size, alignment, commit, allow_large, req_arena_id, memid, tld); + if (p != NULL) return p; + } + } + } + + // if we cannot use OS allocation, return NULL + if (mi_option_is_enabled(mi_option_limit_os_alloc) || req_arena_id != _mi_arena_id_none()) { + errno = ENOMEM; + return NULL; + } + + // finally, fall back to the OS + if (align_offset > 0) { + return _mi_os_alloc_aligned_at_offset(size, alignment, align_offset, commit, allow_large, memid, tld->stats); + } + else { + return _mi_os_alloc_aligned(size, alignment, commit, allow_large, memid, tld->stats); + } +} + +void* _mi_arena_alloc(size_t size, bool commit, bool allow_large, mi_arena_id_t req_arena_id, mi_memid_t* memid, mi_os_tld_t* tld) +{ + return _mi_arena_alloc_aligned(size, MI_ARENA_BLOCK_SIZE, 0, commit, allow_large, req_arena_id, memid, tld); +} + + +void* mi_arena_area(mi_arena_id_t arena_id, size_t* size) { + if (size != NULL) *size = 0; + size_t arena_index = mi_arena_id_index(arena_id); + if (arena_index >= MI_MAX_ARENAS) return NULL; + mi_arena_t* arena = mi_atomic_load_ptr_acquire(mi_arena_t, &mi_arenas[arena_index]); + if (arena == NULL) return NULL; + if (size != NULL) { *size = mi_arena_block_size(arena->block_count); } + return arena->start; +} + + +/* ----------------------------------------------------------- + Arena purge +----------------------------------------------------------- */ + +static long mi_arena_purge_delay(void) { + // <0 = no purging allowed, 0=immediate purging, >0=milli-second delay + return (mi_option_get(mi_option_purge_delay) * mi_option_get(mi_option_arena_purge_mult)); +} + +// reset or decommit in an arena and update the committed/decommit bitmaps +// assumes we own the area (i.e. blocks_in_use is claimed by us) +static void mi_arena_purge(mi_arena_t* arena, size_t bitmap_idx, size_t blocks, mi_stats_t* stats) { + mi_assert_internal(arena->blocks_committed != NULL); + mi_assert_internal(arena->blocks_purge != NULL); + mi_assert_internal(!arena->memid.is_pinned); + const size_t size = mi_arena_block_size(blocks); + void* const p = mi_arena_block_start(arena, bitmap_idx); + bool needs_recommit; + if (_mi_bitmap_is_claimed_across(arena->blocks_committed, arena->field_count, blocks, bitmap_idx)) { + // all blocks are committed, we can purge freely + needs_recommit = _mi_os_purge(p, size, stats); + } + else { + // some blocks are not committed -- this can happen when a partially committed block is freed + // in `_mi_arena_free` and it is conservatively marked as uncommitted but still scheduled for a purge + // we need to ensure we do not try to reset (as that may be invalid for uncommitted memory), + // and also undo the decommit stats (as it was already adjusted) + mi_assert_internal(mi_option_is_enabled(mi_option_purge_decommits)); + needs_recommit = _mi_os_purge_ex(p, size, false /* allow reset? */, stats); + _mi_stat_increase(&stats->committed, size); + } + + // clear the purged blocks + _mi_bitmap_unclaim_across(arena->blocks_purge, arena->field_count, blocks, bitmap_idx); + // update committed bitmap + if (needs_recommit) { + _mi_bitmap_unclaim_across(arena->blocks_committed, arena->field_count, blocks, bitmap_idx); + } +} + +// Schedule a purge. This is usually delayed to avoid repeated decommit/commit calls. +// Note: assumes we (still) own the area as we may purge immediately +static void mi_arena_schedule_purge(mi_arena_t* arena, size_t bitmap_idx, size_t blocks, mi_stats_t* stats) { + mi_assert_internal(arena->blocks_purge != NULL); + const long delay = mi_arena_purge_delay(); + if (delay < 0) return; // is purging allowed at all? + + if (_mi_preloading() || delay == 0) { + // decommit directly + mi_arena_purge(arena, bitmap_idx, blocks, stats); + } + else { + // schedule decommit + mi_msecs_t expire = mi_atomic_loadi64_relaxed(&arena->purge_expire); + if (expire != 0) { + mi_atomic_addi64_acq_rel(&arena->purge_expire, delay/10); // add smallish extra delay + } + else { + mi_atomic_storei64_release(&arena->purge_expire, _mi_clock_now() + delay); + } + _mi_bitmap_claim_across(arena->blocks_purge, arena->field_count, blocks, bitmap_idx, NULL); + } +} + +// purge a range of blocks +// return true if the full range was purged. +// assumes we own the area (i.e. blocks_in_use is claimed by us) +static bool mi_arena_purge_range(mi_arena_t* arena, size_t idx, size_t startidx, size_t bitlen, size_t purge, mi_stats_t* stats) { + const size_t endidx = startidx + bitlen; + size_t bitidx = startidx; + bool all_purged = false; + while (bitidx < endidx) { + // count consequetive ones in the purge mask + size_t count = 0; + while (bitidx + count < endidx && (purge & ((size_t)1 << (bitidx + count))) != 0) { + count++; + } + if (count > 0) { + // found range to be purged + const mi_bitmap_index_t range_idx = mi_bitmap_index_create(idx, bitidx); + mi_arena_purge(arena, range_idx, count, stats); + if (count == bitlen) { + all_purged = true; + } + } + bitidx += (count+1); // +1 to skip the zero bit (or end) + } + return all_purged; +} + +// returns true if anything was purged +static bool mi_arena_try_purge(mi_arena_t* arena, mi_msecs_t now, bool force, mi_stats_t* stats) +{ + if (arena->memid.is_pinned || arena->blocks_purge == NULL) return false; + mi_msecs_t expire = mi_atomic_loadi64_relaxed(&arena->purge_expire); + if (expire == 0) return false; + if (!force && expire > now) return false; + + // reset expire (if not already set concurrently) + mi_atomic_casi64_strong_acq_rel(&arena->purge_expire, &expire, 0); + + // potential purges scheduled, walk through the bitmap + bool any_purged = false; + bool full_purge = true; + for (size_t i = 0; i < arena->field_count; i++) { + size_t purge = mi_atomic_load_relaxed(&arena->blocks_purge[i]); + if (purge != 0) { + size_t bitidx = 0; + while (bitidx < MI_BITMAP_FIELD_BITS) { + // find consequetive range of ones in the purge mask + size_t bitlen = 0; + while (bitidx + bitlen < MI_BITMAP_FIELD_BITS && (purge & ((size_t)1 << (bitidx + bitlen))) != 0) { + bitlen++; + } + // try to claim the longest range of corresponding in_use bits + const mi_bitmap_index_t bitmap_index = mi_bitmap_index_create(i, bitidx); + while( bitlen > 0 ) { + if (_mi_bitmap_try_claim(arena->blocks_inuse, arena->field_count, bitlen, bitmap_index)) { + break; + } + bitlen--; + } + // actual claimed bits at `in_use` + if (bitlen > 0) { + // read purge again now that we have the in_use bits + purge = mi_atomic_load_acquire(&arena->blocks_purge[i]); + if (!mi_arena_purge_range(arena, i, bitidx, bitlen, purge, stats)) { + full_purge = false; + } + any_purged = true; + // release the claimed `in_use` bits again + _mi_bitmap_unclaim(arena->blocks_inuse, arena->field_count, bitlen, bitmap_index); + } + bitidx += (bitlen+1); // +1 to skip the zero (or end) + } // while bitidx + } // purge != 0 + } + // if not fully purged, make sure to purge again in the future + if (!full_purge) { + const long delay = mi_arena_purge_delay(); + mi_msecs_t expected = 0; + mi_atomic_casi64_strong_acq_rel(&arena->purge_expire,&expected,_mi_clock_now() + delay); + } + return any_purged; +} + +static void mi_arenas_try_purge( bool force, bool visit_all, mi_stats_t* stats ) { + if (_mi_preloading() || mi_arena_purge_delay() <= 0) return; // nothing will be scheduled + + const size_t max_arena = mi_atomic_load_acquire(&mi_arena_count); + if (max_arena == 0) return; + + // allow only one thread to purge at a time + static mi_atomic_guard_t purge_guard; + mi_atomic_guard(&purge_guard) + { + mi_msecs_t now = _mi_clock_now(); + size_t max_purge_count = (visit_all ? max_arena : 1); + for (size_t i = 0; i < max_arena; i++) { + mi_arena_t* arena = mi_atomic_load_ptr_acquire(mi_arena_t, &mi_arenas[i]); + if (arena != NULL) { + if (mi_arena_try_purge(arena, now, force, stats)) { + if (max_purge_count <= 1) break; + max_purge_count--; + } + } + } + } +} + + +/* ----------------------------------------------------------- + Arena free +----------------------------------------------------------- */ + +void _mi_arena_free(void* p, size_t size, size_t committed_size, mi_memid_t memid, mi_stats_t* stats) { + mi_assert_internal(size > 0 && stats != NULL); + mi_assert_internal(committed_size <= size); + if (p==NULL) return; + if (size==0) return; + const bool all_committed = (committed_size == size); + + if (mi_memkind_is_os(memid.memkind)) { + // was a direct OS allocation, pass through + if (!all_committed && committed_size > 0) { + // if partially committed, adjust the committed stats (as `_mi_os_free` will increase decommit by the full size) + _mi_stat_decrease(&stats->committed, committed_size); + } + _mi_os_free(p, size, memid, stats); + } + else if (memid.memkind == MI_MEM_ARENA) { + // allocated in an arena + size_t arena_idx; + size_t bitmap_idx; + mi_arena_memid_indices(memid, &arena_idx, &bitmap_idx); + mi_assert_internal(arena_idx < MI_MAX_ARENAS); + mi_arena_t* arena = mi_atomic_load_ptr_acquire(mi_arena_t,&mi_arenas[arena_idx]); + mi_assert_internal(arena != NULL); + const size_t blocks = mi_block_count_of_size(size); + + // checks + if (arena == NULL) { + _mi_error_message(EINVAL, "trying to free from non-existent arena: %p, size %zu, memid: 0x%zx\n", p, size, memid); + return; + } + mi_assert_internal(arena->field_count > mi_bitmap_index_field(bitmap_idx)); + if (arena->field_count <= mi_bitmap_index_field(bitmap_idx)) { + _mi_error_message(EINVAL, "trying to free from non-existent arena block: %p, size %zu, memid: 0x%zx\n", p, size, memid); + return; + } + + // need to set all memory to undefined as some parts may still be marked as no_access (like padding etc.) + mi_track_mem_undefined(p,size); + + // potentially decommit + if (arena->memid.is_pinned || arena->blocks_committed == NULL) { + mi_assert_internal(all_committed); + } + else { + mi_assert_internal(arena->blocks_committed != NULL); + mi_assert_internal(arena->blocks_purge != NULL); + + if (!all_committed) { + // mark the entire range as no longer committed (so we recommit the full range when re-using) + _mi_bitmap_unclaim_across(arena->blocks_committed, arena->field_count, blocks, bitmap_idx); + mi_track_mem_noaccess(p,size); + if (committed_size > 0) { + // if partially committed, adjust the committed stats (is it will be recommitted when re-using) + // in the delayed purge, we now need to not count a decommit if the range is not marked as committed. + _mi_stat_decrease(&stats->committed, committed_size); + } + // note: if not all committed, it may be that the purge will reset/decommit the entire range + // that contains already decommitted parts. Since purge consistently uses reset or decommit that + // works (as we should never reset decommitted parts). + } + // (delay) purge the entire range + mi_arena_schedule_purge(arena, bitmap_idx, blocks, stats); + } + + // and make it available to others again + bool all_inuse = _mi_bitmap_unclaim_across(arena->blocks_inuse, arena->field_count, blocks, bitmap_idx); + if (!all_inuse) { + _mi_error_message(EAGAIN, "trying to free an already freed arena block: %p, size %zu\n", p, size); + return; + }; + } + else { + // arena was none, external, or static; nothing to do + mi_assert_internal(memid.memkind < MI_MEM_OS); + } + + // purge expired decommits + mi_arenas_try_purge(false, false, stats); +} + +// destroy owned arenas; this is unsafe and should only be done using `mi_option_destroy_on_exit` +// for dynamic libraries that are unloaded and need to release all their allocated memory. +static void mi_arenas_unsafe_destroy(void) { + const size_t max_arena = mi_atomic_load_relaxed(&mi_arena_count); + size_t new_max_arena = 0; + for (size_t i = 0; i < max_arena; i++) { + mi_arena_t* arena = mi_atomic_load_ptr_acquire(mi_arena_t, &mi_arenas[i]); + if (arena != NULL) { + if (arena->start != NULL && mi_memkind_is_os(arena->memid.memkind)) { + mi_atomic_store_ptr_release(mi_arena_t, &mi_arenas[i], NULL); + _mi_os_free(arena->start, mi_arena_size(arena), arena->memid, &_mi_stats_main); + } + else { + new_max_arena = i; + } + mi_arena_meta_free(arena, arena->meta_memid, arena->meta_size, &_mi_stats_main); + } + } + + // try to lower the max arena. + size_t expected = max_arena; + mi_atomic_cas_strong_acq_rel(&mi_arena_count, &expected, new_max_arena); +} + +// Purge the arenas; if `force_purge` is true, amenable parts are purged even if not yet expired +void _mi_arena_collect(bool force_purge, mi_stats_t* stats) { + mi_arenas_try_purge(force_purge, true /* visit all */, stats); +} + +// destroy owned arenas; this is unsafe and should only be done using `mi_option_destroy_on_exit` +// for dynamic libraries that are unloaded and need to release all their allocated memory. +void _mi_arena_unsafe_destroy_all(mi_stats_t* stats) { + mi_arenas_unsafe_destroy(); + _mi_arena_collect(true /* force purge */, stats); // purge non-owned arenas +} + +// Is a pointer inside any of our arenas? +bool _mi_arena_contains(const void* p) { + const size_t max_arena = mi_atomic_load_relaxed(&mi_arena_count); + for (size_t i = 0; i < max_arena; i++) { + mi_arena_t* arena = mi_atomic_load_ptr_acquire(mi_arena_t, &mi_arenas[i]); + if (arena != NULL && arena->start <= (const uint8_t*)p && arena->start + mi_arena_block_size(arena->block_count) > (const uint8_t*)p) { + return true; + } + } + return false; +} + + +/* ----------------------------------------------------------- + Add an arena. +----------------------------------------------------------- */ + +static bool mi_arena_add(mi_arena_t* arena, mi_arena_id_t* arena_id) { + mi_assert_internal(arena != NULL); + mi_assert_internal((uintptr_t)mi_atomic_load_ptr_relaxed(uint8_t,&arena->start) % MI_SEGMENT_ALIGN == 0); + mi_assert_internal(arena->block_count > 0); + if (arena_id != NULL) { *arena_id = -1; } + + size_t i = mi_atomic_increment_acq_rel(&mi_arena_count); + if (i >= MI_MAX_ARENAS) { + mi_atomic_decrement_acq_rel(&mi_arena_count); + return false; + } + arena->id = mi_arena_id_create(i); + mi_atomic_store_ptr_release(mi_arena_t,&mi_arenas[i], arena); + if (arena_id != NULL) { *arena_id = arena->id; } + return true; +} + +static bool mi_manage_os_memory_ex2(void* start, size_t size, bool is_large, int numa_node, bool exclusive, mi_memid_t memid, mi_arena_id_t* arena_id) mi_attr_noexcept +{ + if (arena_id != NULL) *arena_id = _mi_arena_id_none(); + if (size < MI_ARENA_BLOCK_SIZE) return false; + + if (is_large) { + mi_assert_internal(memid.initially_committed && memid.is_pinned); + } + + const size_t bcount = size / MI_ARENA_BLOCK_SIZE; + const size_t fields = _mi_divide_up(bcount, MI_BITMAP_FIELD_BITS); + const size_t bitmaps = (memid.is_pinned ? 2 : 4); + const size_t asize = sizeof(mi_arena_t) + (bitmaps*fields*sizeof(mi_bitmap_field_t)); + mi_memid_t meta_memid; + mi_arena_t* arena = (mi_arena_t*)mi_arena_meta_zalloc(asize, &meta_memid, &_mi_stats_main); // TODO: can we avoid allocating from the OS? + if (arena == NULL) return false; + + // already zero'd due to os_alloc + // _mi_memzero(arena, asize); + arena->id = _mi_arena_id_none(); + arena->memid = memid; + arena->exclusive = exclusive; + arena->meta_size = asize; + arena->meta_memid = meta_memid; + arena->block_count = bcount; + arena->field_count = fields; + arena->start = (uint8_t*)start; + arena->numa_node = numa_node; // TODO: or get the current numa node if -1? (now it allows anyone to allocate on -1) + arena->is_large = is_large; + arena->purge_expire = 0; + arena->search_idx = 0; + arena->blocks_dirty = &arena->blocks_inuse[fields]; // just after inuse bitmap + arena->blocks_committed = (arena->memid.is_pinned ? NULL : &arena->blocks_inuse[2*fields]); // just after dirty bitmap + arena->blocks_purge = (arena->memid.is_pinned ? NULL : &arena->blocks_inuse[3*fields]); // just after committed bitmap + // initialize committed bitmap? + if (arena->blocks_committed != NULL && arena->memid.initially_committed) { + memset((void*)arena->blocks_committed, 0xFF, fields*sizeof(mi_bitmap_field_t)); // cast to void* to avoid atomic warning + } + + // and claim leftover blocks if needed (so we never allocate there) + ptrdiff_t post = (fields * MI_BITMAP_FIELD_BITS) - bcount; + mi_assert_internal(post >= 0); + if (post > 0) { + // don't use leftover bits at the end + mi_bitmap_index_t postidx = mi_bitmap_index_create(fields - 1, MI_BITMAP_FIELD_BITS - post); + _mi_bitmap_claim(arena->blocks_inuse, fields, post, postidx, NULL); + } + return mi_arena_add(arena, arena_id); + +} + +bool mi_manage_os_memory_ex(void* start, size_t size, bool is_committed, bool is_large, bool is_zero, int numa_node, bool exclusive, mi_arena_id_t* arena_id) mi_attr_noexcept { + mi_memid_t memid = _mi_memid_create(MI_MEM_EXTERNAL); + memid.initially_committed = is_committed; + memid.initially_zero = is_zero; + memid.is_pinned = is_large; + return mi_manage_os_memory_ex2(start,size,is_large,numa_node,exclusive,memid, arena_id); +} + +// Reserve a range of regular OS memory +int mi_reserve_os_memory_ex(size_t size, bool commit, bool allow_large, bool exclusive, mi_arena_id_t* arena_id) mi_attr_noexcept { + if (arena_id != NULL) *arena_id = _mi_arena_id_none(); + size = _mi_align_up(size, MI_ARENA_BLOCK_SIZE); // at least one block + mi_memid_t memid; + void* start = _mi_os_alloc_aligned(size, MI_SEGMENT_ALIGN, commit, allow_large, &memid, &_mi_stats_main); + if (start == NULL) return ENOMEM; + const bool is_large = memid.is_pinned; // todo: use separate is_large field? + if (!mi_manage_os_memory_ex2(start, size, is_large, -1 /* numa node */, exclusive, memid, arena_id)) { + _mi_os_free_ex(start, size, commit, memid, &_mi_stats_main); + _mi_verbose_message("failed to reserve %zu k memory\n", _mi_divide_up(size, 1024)); + return ENOMEM; + } + _mi_verbose_message("reserved %zu KiB memory%s\n", _mi_divide_up(size, 1024), is_large ? " (in large os pages)" : ""); + return 0; +} + + +// Manage a range of regular OS memory +bool mi_manage_os_memory(void* start, size_t size, bool is_committed, bool is_large, bool is_zero, int numa_node) mi_attr_noexcept { + return mi_manage_os_memory_ex(start, size, is_committed, is_large, is_zero, numa_node, false /* exclusive? */, NULL); +} + +// Reserve a range of regular OS memory +int mi_reserve_os_memory(size_t size, bool commit, bool allow_large) mi_attr_noexcept { + return mi_reserve_os_memory_ex(size, commit, allow_large, false, NULL); +} + + +/* ----------------------------------------------------------- + Debugging +----------------------------------------------------------- */ + +static size_t mi_debug_show_bitmap(const char* prefix, mi_bitmap_field_t* fields, size_t field_count ) { + size_t inuse_count = 0; + for (size_t i = 0; i < field_count; i++) { + char buf[MI_BITMAP_FIELD_BITS + 1]; + uintptr_t field = mi_atomic_load_relaxed(&fields[i]); + for (size_t bit = 0; bit < MI_BITMAP_FIELD_BITS; bit++) { + bool inuse = ((((uintptr_t)1 << bit) & field) != 0); + if (inuse) inuse_count++; + buf[MI_BITMAP_FIELD_BITS - 1 - bit] = (inuse ? 'x' : '.'); + } + buf[MI_BITMAP_FIELD_BITS] = 0; + _mi_verbose_message("%s%s\n", prefix, buf); + } + return inuse_count; +} + +void mi_debug_show_arenas(void) mi_attr_noexcept { + size_t max_arenas = mi_atomic_load_relaxed(&mi_arena_count); + for (size_t i = 0; i < max_arenas; i++) { + mi_arena_t* arena = mi_atomic_load_ptr_relaxed(mi_arena_t, &mi_arenas[i]); + if (arena == NULL) break; + size_t inuse_count = 0; + _mi_verbose_message("arena %zu: %zu blocks with %zu fields\n", i, arena->block_count, arena->field_count); + inuse_count += mi_debug_show_bitmap(" ", arena->blocks_inuse, arena->field_count); + _mi_verbose_message(" blocks in use ('x'): %zu\n", inuse_count); + } +} + + +/* ----------------------------------------------------------- + Reserve a huge page arena. +----------------------------------------------------------- */ +// reserve at a specific numa node +int mi_reserve_huge_os_pages_at_ex(size_t pages, int numa_node, size_t timeout_msecs, bool exclusive, mi_arena_id_t* arena_id) mi_attr_noexcept { + if (arena_id != NULL) *arena_id = -1; + if (pages==0) return 0; + if (numa_node < -1) numa_node = -1; + if (numa_node >= 0) numa_node = numa_node % _mi_os_numa_node_count(); + size_t hsize = 0; + size_t pages_reserved = 0; + mi_memid_t memid; + void* p = _mi_os_alloc_huge_os_pages(pages, numa_node, timeout_msecs, &pages_reserved, &hsize, &memid); + if (p==NULL || pages_reserved==0) { + _mi_warning_message("failed to reserve %zu GiB huge pages\n", pages); + return ENOMEM; + } + _mi_verbose_message("numa node %i: reserved %zu GiB huge pages (of the %zu GiB requested)\n", numa_node, pages_reserved, pages); + + if (!mi_manage_os_memory_ex2(p, hsize, true, numa_node, exclusive, memid, arena_id)) { + _mi_os_free(p, hsize, memid, &_mi_stats_main); + return ENOMEM; + } + return 0; +} + +int mi_reserve_huge_os_pages_at(size_t pages, int numa_node, size_t timeout_msecs) mi_attr_noexcept { + return mi_reserve_huge_os_pages_at_ex(pages, numa_node, timeout_msecs, false, NULL); +} + +// reserve huge pages evenly among the given number of numa nodes (or use the available ones as detected) +int mi_reserve_huge_os_pages_interleave(size_t pages, size_t numa_nodes, size_t timeout_msecs) mi_attr_noexcept { + if (pages == 0) return 0; + + // pages per numa node + size_t numa_count = (numa_nodes > 0 ? numa_nodes : _mi_os_numa_node_count()); + if (numa_count <= 0) numa_count = 1; + const size_t pages_per = pages / numa_count; + const size_t pages_mod = pages % numa_count; + const size_t timeout_per = (timeout_msecs==0 ? 0 : (timeout_msecs / numa_count) + 50); + + // reserve evenly among numa nodes + for (size_t numa_node = 0; numa_node < numa_count && pages > 0; numa_node++) { + size_t node_pages = pages_per; // can be 0 + if (numa_node < pages_mod) node_pages++; + int err = mi_reserve_huge_os_pages_at(node_pages, (int)numa_node, timeout_per); + if (err) return err; + if (pages < node_pages) { + pages = 0; + } + else { + pages -= node_pages; + } + } + + return 0; +} + +int mi_reserve_huge_os_pages(size_t pages, double max_secs, size_t* pages_reserved) mi_attr_noexcept { + MI_UNUSED(max_secs); + _mi_warning_message("mi_reserve_huge_os_pages is deprecated: use mi_reserve_huge_os_pages_interleave/at instead\n"); + if (pages_reserved != NULL) *pages_reserved = 0; + int err = mi_reserve_huge_os_pages_interleave(pages, 0, (size_t)(max_secs * 1000.0)); + if (err==0 && pages_reserved!=NULL) *pages_reserved = pages; + return err; +} diff --git a/compat/mimalloc/bitmap.c b/compat/mimalloc/bitmap.c new file mode 100644 index 00000000000000..878f0ab3250a47 --- /dev/null +++ b/compat/mimalloc/bitmap.c @@ -0,0 +1,432 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2019-2023 Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +/* ---------------------------------------------------------------------------- +Concurrent bitmap that can set/reset sequences of bits atomically, +represeted as an array of fields where each field is a machine word (`size_t`) + +There are two api's; the standard one cannot have sequences that cross +between the bitmap fields (and a sequence must be <= MI_BITMAP_FIELD_BITS). + +The `_across` postfixed functions do allow sequences that can cross over +between the fields. (This is used in arena allocation) +---------------------------------------------------------------------------- */ + +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "bitmap.h" + +/* ----------------------------------------------------------- + Bitmap definition +----------------------------------------------------------- */ + +// The bit mask for a given number of blocks at a specified bit index. +static inline size_t mi_bitmap_mask_(size_t count, size_t bitidx) { + mi_assert_internal(count + bitidx <= MI_BITMAP_FIELD_BITS); + mi_assert_internal(count > 0); + if (count >= MI_BITMAP_FIELD_BITS) return MI_BITMAP_FIELD_FULL; + if (count == 0) return 0; + return ((((size_t)1 << count) - 1) << bitidx); +} + + +/* ----------------------------------------------------------- + Claim a bit sequence atomically +----------------------------------------------------------- */ + +// Try to atomically claim a sequence of `count` bits in a single +// field at `idx` in `bitmap`. Returns `true` on success. +inline bool _mi_bitmap_try_find_claim_field(mi_bitmap_t bitmap, size_t idx, const size_t count, mi_bitmap_index_t* bitmap_idx) +{ + mi_assert_internal(bitmap_idx != NULL); + mi_assert_internal(count <= MI_BITMAP_FIELD_BITS); + mi_assert_internal(count > 0); + mi_bitmap_field_t* field = &bitmap[idx]; + size_t map = mi_atomic_load_relaxed(field); + if (map==MI_BITMAP_FIELD_FULL) return false; // short cut + + // search for 0-bit sequence of length count + const size_t mask = mi_bitmap_mask_(count, 0); + const size_t bitidx_max = MI_BITMAP_FIELD_BITS - count; + +#ifdef MI_HAVE_FAST_BITSCAN + size_t bitidx = mi_ctz(~map); // quickly find the first zero bit if possible +#else + size_t bitidx = 0; // otherwise start at 0 +#endif + size_t m = (mask << bitidx); // invariant: m == mask shifted by bitidx + + // scan linearly for a free range of zero bits + while (bitidx <= bitidx_max) { + const size_t mapm = (map & m); + if (mapm == 0) { // are the mask bits free at bitidx? + mi_assert_internal((m >> bitidx) == mask); // no overflow? + const size_t newmap = (map | m); + mi_assert_internal((newmap^map) >> bitidx == mask); + if (!mi_atomic_cas_strong_acq_rel(field, &map, newmap)) { // TODO: use weak cas here? + // no success, another thread claimed concurrently.. keep going (with updated `map`) + continue; + } + else { + // success, we claimed the bits! + *bitmap_idx = mi_bitmap_index_create(idx, bitidx); + return true; + } + } + else { + // on to the next bit range +#ifdef MI_HAVE_FAST_BITSCAN + mi_assert_internal(mapm != 0); + const size_t shift = (count == 1 ? 1 : (MI_INTPTR_BITS - mi_clz(mapm) - bitidx)); + mi_assert_internal(shift > 0 && shift <= count); +#else + const size_t shift = 1; +#endif + bitidx += shift; + m <<= shift; + } + } + // no bits found + return false; +} + +// Find `count` bits of 0 and set them to 1 atomically; returns `true` on success. +// Starts at idx, and wraps around to search in all `bitmap_fields` fields. +// `count` can be at most MI_BITMAP_FIELD_BITS and will never cross fields. +bool _mi_bitmap_try_find_from_claim(mi_bitmap_t bitmap, const size_t bitmap_fields, const size_t start_field_idx, const size_t count, mi_bitmap_index_t* bitmap_idx) { + size_t idx = start_field_idx; + for (size_t visited = 0; visited < bitmap_fields; visited++, idx++) { + if (idx >= bitmap_fields) { idx = 0; } // wrap + if (_mi_bitmap_try_find_claim_field(bitmap, idx, count, bitmap_idx)) { + return true; + } + } + return false; +} + +// Like _mi_bitmap_try_find_from_claim but with an extra predicate that must be fullfilled +bool _mi_bitmap_try_find_from_claim_pred(mi_bitmap_t bitmap, const size_t bitmap_fields, + const size_t start_field_idx, const size_t count, + mi_bitmap_pred_fun_t pred_fun, void* pred_arg, + mi_bitmap_index_t* bitmap_idx) { + size_t idx = start_field_idx; + for (size_t visited = 0; visited < bitmap_fields; visited++, idx++) { + if (idx >= bitmap_fields) idx = 0; // wrap + if (_mi_bitmap_try_find_claim_field(bitmap, idx, count, bitmap_idx)) { + if (pred_fun == NULL || pred_fun(*bitmap_idx, pred_arg)) { + return true; + } + // predicate returned false, unclaim and look further + _mi_bitmap_unclaim(bitmap, bitmap_fields, count, *bitmap_idx); + } + } + return false; +} + +// Set `count` bits at `bitmap_idx` to 0 atomically +// Returns `true` if all `count` bits were 1 previously. +bool _mi_bitmap_unclaim(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx) { + const size_t idx = mi_bitmap_index_field(bitmap_idx); + const size_t bitidx = mi_bitmap_index_bit_in_field(bitmap_idx); + const size_t mask = mi_bitmap_mask_(count, bitidx); + mi_assert_internal(bitmap_fields > idx); MI_UNUSED(bitmap_fields); + // mi_assert_internal((bitmap[idx] & mask) == mask); + const size_t prev = mi_atomic_and_acq_rel(&bitmap[idx], ~mask); + return ((prev & mask) == mask); +} + + +// Set `count` bits at `bitmap_idx` to 1 atomically +// Returns `true` if all `count` bits were 0 previously. `any_zero` is `true` if there was at least one zero bit. +bool _mi_bitmap_claim(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx, bool* any_zero) { + const size_t idx = mi_bitmap_index_field(bitmap_idx); + const size_t bitidx = mi_bitmap_index_bit_in_field(bitmap_idx); + const size_t mask = mi_bitmap_mask_(count, bitidx); + mi_assert_internal(bitmap_fields > idx); MI_UNUSED(bitmap_fields); + //mi_assert_internal(any_zero != NULL || (bitmap[idx] & mask) == 0); + size_t prev = mi_atomic_or_acq_rel(&bitmap[idx], mask); + if (any_zero != NULL) { *any_zero = ((prev & mask) != mask); } + return ((prev & mask) == 0); +} + +// Returns `true` if all `count` bits were 1. `any_ones` is `true` if there was at least one bit set to one. +static bool mi_bitmap_is_claimedx(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx, bool* any_ones) { + const size_t idx = mi_bitmap_index_field(bitmap_idx); + const size_t bitidx = mi_bitmap_index_bit_in_field(bitmap_idx); + const size_t mask = mi_bitmap_mask_(count, bitidx); + mi_assert_internal(bitmap_fields > idx); MI_UNUSED(bitmap_fields); + const size_t field = mi_atomic_load_relaxed(&bitmap[idx]); + if (any_ones != NULL) { *any_ones = ((field & mask) != 0); } + return ((field & mask) == mask); +} + +// Try to set `count` bits at `bitmap_idx` from 0 to 1 atomically. +// Returns `true` if successful when all previous `count` bits were 0. +bool _mi_bitmap_try_claim(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx) { + const size_t idx = mi_bitmap_index_field(bitmap_idx); + const size_t bitidx = mi_bitmap_index_bit_in_field(bitmap_idx); + const size_t mask = mi_bitmap_mask_(count, bitidx); + mi_assert_internal(bitmap_fields > idx); MI_UNUSED(bitmap_fields); + size_t expected = mi_atomic_load_relaxed(&bitmap[idx]); + do { + if ((expected & mask) != 0) return false; + } + while (!mi_atomic_cas_strong_acq_rel(&bitmap[idx], &expected, expected | mask)); + mi_assert_internal((expected & mask) == 0); + return true; +} + + +bool _mi_bitmap_is_claimed(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx) { + return mi_bitmap_is_claimedx(bitmap, bitmap_fields, count, bitmap_idx, NULL); +} + +bool _mi_bitmap_is_any_claimed(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx) { + bool any_ones; + mi_bitmap_is_claimedx(bitmap, bitmap_fields, count, bitmap_idx, &any_ones); + return any_ones; +} + + +//-------------------------------------------------------------------------- +// the `_across` functions work on bitmaps where sequences can cross over +// between the fields. This is used in arena allocation +//-------------------------------------------------------------------------- + +// Try to atomically claim a sequence of `count` bits starting from the field +// at `idx` in `bitmap` and crossing into subsequent fields. Returns `true` on success. +// Only needs to consider crossing into the next fields (see `mi_bitmap_try_find_from_claim_across`) +static bool mi_bitmap_try_find_claim_field_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t idx, const size_t count, const size_t retries, mi_bitmap_index_t* bitmap_idx) +{ + mi_assert_internal(bitmap_idx != NULL); + + // check initial trailing zeros + mi_bitmap_field_t* field = &bitmap[idx]; + size_t map = mi_atomic_load_relaxed(field); + const size_t initial = mi_clz(map); // count of initial zeros starting at idx + mi_assert_internal(initial <= MI_BITMAP_FIELD_BITS); + if (initial == 0) return false; + if (initial >= count) return _mi_bitmap_try_find_claim_field(bitmap, idx, count, bitmap_idx); // no need to cross fields (this case won't happen for us) + if (_mi_divide_up(count - initial, MI_BITMAP_FIELD_BITS) >= (bitmap_fields - idx)) return false; // not enough entries + + // scan ahead + size_t found = initial; + size_t mask = 0; // mask bits for the final field + while(found < count) { + field++; + map = mi_atomic_load_relaxed(field); + const size_t mask_bits = (found + MI_BITMAP_FIELD_BITS <= count ? MI_BITMAP_FIELD_BITS : (count - found)); + mi_assert_internal(mask_bits > 0 && mask_bits <= MI_BITMAP_FIELD_BITS); + mask = mi_bitmap_mask_(mask_bits, 0); + if ((map & mask) != 0) return false; // some part is already claimed + found += mask_bits; + } + mi_assert_internal(field < &bitmap[bitmap_fields]); + + // we found a range of contiguous zeros up to the final field; mask contains mask in the final field + // now try to claim the range atomically + mi_bitmap_field_t* const final_field = field; + const size_t final_mask = mask; + mi_bitmap_field_t* const initial_field = &bitmap[idx]; + const size_t initial_idx = MI_BITMAP_FIELD_BITS - initial; + const size_t initial_mask = mi_bitmap_mask_(initial, initial_idx); + + // initial field + size_t newmap; + field = initial_field; + map = mi_atomic_load_relaxed(field); + do { + newmap = (map | initial_mask); + if ((map & initial_mask) != 0) { goto rollback; }; + } while (!mi_atomic_cas_strong_acq_rel(field, &map, newmap)); + + // intermediate fields + while (++field < final_field) { + newmap = MI_BITMAP_FIELD_FULL; + map = 0; + if (!mi_atomic_cas_strong_acq_rel(field, &map, newmap)) { goto rollback; } + } + + // final field + mi_assert_internal(field == final_field); + map = mi_atomic_load_relaxed(field); + do { + newmap = (map | final_mask); + if ((map & final_mask) != 0) { goto rollback; } + } while (!mi_atomic_cas_strong_acq_rel(field, &map, newmap)); + + // claimed! + *bitmap_idx = mi_bitmap_index_create(idx, initial_idx); + return true; + +rollback: + // roll back intermediate fields + // (we just failed to claim `field` so decrement first) + while (--field > initial_field) { + newmap = 0; + map = MI_BITMAP_FIELD_FULL; + mi_assert_internal(mi_atomic_load_relaxed(field) == map); + mi_atomic_store_release(field, newmap); + } + if (field == initial_field) { // (if we failed on the initial field, `field + 1 == initial_field`) + map = mi_atomic_load_relaxed(field); + do { + mi_assert_internal((map & initial_mask) == initial_mask); + newmap = (map & ~initial_mask); + } while (!mi_atomic_cas_strong_acq_rel(field, &map, newmap)); + } + // retry? (we make a recursive call instead of goto to be able to use const declarations) + if (retries <= 2) { + return mi_bitmap_try_find_claim_field_across(bitmap, bitmap_fields, idx, count, retries+1, bitmap_idx); + } + else { + return false; + } +} + + +// Find `count` bits of zeros and set them to 1 atomically; returns `true` on success. +// Starts at idx, and wraps around to search in all `bitmap_fields` fields. +bool _mi_bitmap_try_find_from_claim_across(mi_bitmap_t bitmap, const size_t bitmap_fields, const size_t start_field_idx, const size_t count, mi_bitmap_index_t* bitmap_idx) { + mi_assert_internal(count > 0); + if (count <= 2) { + // we don't bother with crossover fields for small counts + return _mi_bitmap_try_find_from_claim(bitmap, bitmap_fields, start_field_idx, count, bitmap_idx); + } + + // visit the fields + size_t idx = start_field_idx; + for (size_t visited = 0; visited < bitmap_fields; visited++, idx++) { + if (idx >= bitmap_fields) { idx = 0; } // wrap + // first try to claim inside a field + if (count <= MI_BITMAP_FIELD_BITS) { + if (_mi_bitmap_try_find_claim_field(bitmap, idx, count, bitmap_idx)) { + return true; + } + } + // if that fails, then try to claim across fields + if (mi_bitmap_try_find_claim_field_across(bitmap, bitmap_fields, idx, count, 0, bitmap_idx)) { + return true; + } + } + return false; +} + +// Helper for masks across fields; returns the mid count, post_mask may be 0 +static size_t mi_bitmap_mask_across(mi_bitmap_index_t bitmap_idx, size_t bitmap_fields, size_t count, size_t* pre_mask, size_t* mid_mask, size_t* post_mask) { + MI_UNUSED(bitmap_fields); + const size_t bitidx = mi_bitmap_index_bit_in_field(bitmap_idx); + if mi_likely(bitidx + count <= MI_BITMAP_FIELD_BITS) { + *pre_mask = mi_bitmap_mask_(count, bitidx); + *mid_mask = 0; + *post_mask = 0; + mi_assert_internal(mi_bitmap_index_field(bitmap_idx) < bitmap_fields); + return 0; + } + else { + const size_t pre_bits = MI_BITMAP_FIELD_BITS - bitidx; + mi_assert_internal(pre_bits < count); + *pre_mask = mi_bitmap_mask_(pre_bits, bitidx); + count -= pre_bits; + const size_t mid_count = (count / MI_BITMAP_FIELD_BITS); + *mid_mask = MI_BITMAP_FIELD_FULL; + count %= MI_BITMAP_FIELD_BITS; + *post_mask = (count==0 ? 0 : mi_bitmap_mask_(count, 0)); + mi_assert_internal(mi_bitmap_index_field(bitmap_idx) + mid_count + (count==0 ? 0 : 1) < bitmap_fields); + return mid_count; + } +} + +// Set `count` bits at `bitmap_idx` to 0 atomically +// Returns `true` if all `count` bits were 1 previously. +bool _mi_bitmap_unclaim_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx) { + size_t idx = mi_bitmap_index_field(bitmap_idx); + size_t pre_mask; + size_t mid_mask; + size_t post_mask; + size_t mid_count = mi_bitmap_mask_across(bitmap_idx, bitmap_fields, count, &pre_mask, &mid_mask, &post_mask); + bool all_one = true; + mi_bitmap_field_t* field = &bitmap[idx]; + size_t prev = mi_atomic_and_acq_rel(field++, ~pre_mask); // clear first part + if ((prev & pre_mask) != pre_mask) all_one = false; + while(mid_count-- > 0) { + prev = mi_atomic_and_acq_rel(field++, ~mid_mask); // clear mid part + if ((prev & mid_mask) != mid_mask) all_one = false; + } + if (post_mask!=0) { + prev = mi_atomic_and_acq_rel(field, ~post_mask); // clear end part + if ((prev & post_mask) != post_mask) all_one = false; + } + return all_one; +} + +// Set `count` bits at `bitmap_idx` to 1 atomically +// Returns `true` if all `count` bits were 0 previously. `any_zero` is `true` if there was at least one zero bit. +bool _mi_bitmap_claim_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx, bool* pany_zero) { + size_t idx = mi_bitmap_index_field(bitmap_idx); + size_t pre_mask; + size_t mid_mask; + size_t post_mask; + size_t mid_count = mi_bitmap_mask_across(bitmap_idx, bitmap_fields, count, &pre_mask, &mid_mask, &post_mask); + bool all_zero = true; + bool any_zero = false; + _Atomic(size_t)*field = &bitmap[idx]; + size_t prev = mi_atomic_or_acq_rel(field++, pre_mask); + if ((prev & pre_mask) != 0) all_zero = false; + if ((prev & pre_mask) != pre_mask) any_zero = true; + while (mid_count-- > 0) { + prev = mi_atomic_or_acq_rel(field++, mid_mask); + if ((prev & mid_mask) != 0) all_zero = false; + if ((prev & mid_mask) != mid_mask) any_zero = true; + } + if (post_mask!=0) { + prev = mi_atomic_or_acq_rel(field, post_mask); + if ((prev & post_mask) != 0) all_zero = false; + if ((prev & post_mask) != post_mask) any_zero = true; + } + if (pany_zero != NULL) { *pany_zero = any_zero; } + return all_zero; +} + + +// Returns `true` if all `count` bits were 1. +// `any_ones` is `true` if there was at least one bit set to one. +static bool mi_bitmap_is_claimedx_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx, bool* pany_ones) { + size_t idx = mi_bitmap_index_field(bitmap_idx); + size_t pre_mask; + size_t mid_mask; + size_t post_mask; + size_t mid_count = mi_bitmap_mask_across(bitmap_idx, bitmap_fields, count, &pre_mask, &mid_mask, &post_mask); + bool all_ones = true; + bool any_ones = false; + mi_bitmap_field_t* field = &bitmap[idx]; + size_t prev = mi_atomic_load_relaxed(field++); + if ((prev & pre_mask) != pre_mask) all_ones = false; + if ((prev & pre_mask) != 0) any_ones = true; + while (mid_count-- > 0) { + prev = mi_atomic_load_relaxed(field++); + if ((prev & mid_mask) != mid_mask) all_ones = false; + if ((prev & mid_mask) != 0) any_ones = true; + } + if (post_mask!=0) { + prev = mi_atomic_load_relaxed(field); + if ((prev & post_mask) != post_mask) all_ones = false; + if ((prev & post_mask) != 0) any_ones = true; + } + if (pany_ones != NULL) { *pany_ones = any_ones; } + return all_ones; +} + +bool _mi_bitmap_is_claimed_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx) { + return mi_bitmap_is_claimedx_across(bitmap, bitmap_fields, count, bitmap_idx, NULL); +} + +bool _mi_bitmap_is_any_claimed_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx) { + bool any_ones; + mi_bitmap_is_claimedx_across(bitmap, bitmap_fields, count, bitmap_idx, &any_ones); + return any_ones; +} diff --git a/compat/mimalloc/bitmap.h b/compat/mimalloc/bitmap.h new file mode 100644 index 00000000000000..9ba15d5d6f09ea --- /dev/null +++ b/compat/mimalloc/bitmap.h @@ -0,0 +1,115 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2019-2023 Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +/* ---------------------------------------------------------------------------- +Concurrent bitmap that can set/reset sequences of bits atomically, +represeted as an array of fields where each field is a machine word (`size_t`) + +There are two api's; the standard one cannot have sequences that cross +between the bitmap fields (and a sequence must be <= MI_BITMAP_FIELD_BITS). +(this is used in region allocation) + +The `_across` postfixed functions do allow sequences that can cross over +between the fields. (This is used in arena allocation) +---------------------------------------------------------------------------- */ +#pragma once +#ifndef MI_BITMAP_H +#define MI_BITMAP_H + +/* ----------------------------------------------------------- + Bitmap definition +----------------------------------------------------------- */ + +#define MI_BITMAP_FIELD_BITS (8*MI_SIZE_SIZE) +#define MI_BITMAP_FIELD_FULL (~((size_t)0)) // all bits set + +// An atomic bitmap of `size_t` fields +typedef _Atomic(size_t) mi_bitmap_field_t; +typedef mi_bitmap_field_t* mi_bitmap_t; + +// A bitmap index is the index of the bit in a bitmap. +typedef size_t mi_bitmap_index_t; + +// Create a bit index. +static inline mi_bitmap_index_t mi_bitmap_index_create(size_t idx, size_t bitidx) { + mi_assert_internal(bitidx < MI_BITMAP_FIELD_BITS); + return (idx*MI_BITMAP_FIELD_BITS) + bitidx; +} + +// Create a bit index. +static inline mi_bitmap_index_t mi_bitmap_index_create_from_bit(size_t full_bitidx) { + return mi_bitmap_index_create(full_bitidx / MI_BITMAP_FIELD_BITS, full_bitidx % MI_BITMAP_FIELD_BITS); +} + +// Get the field index from a bit index. +static inline size_t mi_bitmap_index_field(mi_bitmap_index_t bitmap_idx) { + return (bitmap_idx / MI_BITMAP_FIELD_BITS); +} + +// Get the bit index in a bitmap field +static inline size_t mi_bitmap_index_bit_in_field(mi_bitmap_index_t bitmap_idx) { + return (bitmap_idx % MI_BITMAP_FIELD_BITS); +} + +// Get the full bit index +static inline size_t mi_bitmap_index_bit(mi_bitmap_index_t bitmap_idx) { + return bitmap_idx; +} + +/* ----------------------------------------------------------- + Claim a bit sequence atomically +----------------------------------------------------------- */ + +// Try to atomically claim a sequence of `count` bits in a single +// field at `idx` in `bitmap`. Returns `true` on success. +bool _mi_bitmap_try_find_claim_field(mi_bitmap_t bitmap, size_t idx, const size_t count, mi_bitmap_index_t* bitmap_idx); + +// Starts at idx, and wraps around to search in all `bitmap_fields` fields. +// For now, `count` can be at most MI_BITMAP_FIELD_BITS and will never cross fields. +bool _mi_bitmap_try_find_from_claim(mi_bitmap_t bitmap, const size_t bitmap_fields, const size_t start_field_idx, const size_t count, mi_bitmap_index_t* bitmap_idx); + +// Like _mi_bitmap_try_find_from_claim but with an extra predicate that must be fullfilled +typedef bool (mi_cdecl *mi_bitmap_pred_fun_t)(mi_bitmap_index_t bitmap_idx, void* pred_arg); +bool _mi_bitmap_try_find_from_claim_pred(mi_bitmap_t bitmap, const size_t bitmap_fields, const size_t start_field_idx, const size_t count, mi_bitmap_pred_fun_t pred_fun, void* pred_arg, mi_bitmap_index_t* bitmap_idx); + +// Set `count` bits at `bitmap_idx` to 0 atomically +// Returns `true` if all `count` bits were 1 previously. +bool _mi_bitmap_unclaim(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx); + +// Try to set `count` bits at `bitmap_idx` from 0 to 1 atomically. +// Returns `true` if successful when all previous `count` bits were 0. +bool _mi_bitmap_try_claim(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx); + +// Set `count` bits at `bitmap_idx` to 1 atomically +// Returns `true` if all `count` bits were 0 previously. `any_zero` is `true` if there was at least one zero bit. +bool _mi_bitmap_claim(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx, bool* any_zero); + +bool _mi_bitmap_is_claimed(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx); +bool _mi_bitmap_is_any_claimed(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx); + + +//-------------------------------------------------------------------------- +// the `_across` functions work on bitmaps where sequences can cross over +// between the fields. This is used in arena allocation +//-------------------------------------------------------------------------- + +// Find `count` bits of zeros and set them to 1 atomically; returns `true` on success. +// Starts at idx, and wraps around to search in all `bitmap_fields` fields. +bool _mi_bitmap_try_find_from_claim_across(mi_bitmap_t bitmap, const size_t bitmap_fields, const size_t start_field_idx, const size_t count, mi_bitmap_index_t* bitmap_idx); + +// Set `count` bits at `bitmap_idx` to 0 atomically +// Returns `true` if all `count` bits were 1 previously. +bool _mi_bitmap_unclaim_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx); + +// Set `count` bits at `bitmap_idx` to 1 atomically +// Returns `true` if all `count` bits were 0 previously. `any_zero` is `true` if there was at least one zero bit. +bool _mi_bitmap_claim_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx, bool* pany_zero); + +bool _mi_bitmap_is_claimed_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx); +bool _mi_bitmap_is_any_claimed_across(mi_bitmap_t bitmap, size_t bitmap_fields, size_t count, mi_bitmap_index_t bitmap_idx); + +#endif diff --git a/compat/mimalloc/heap.c b/compat/mimalloc/heap.c new file mode 100644 index 00000000000000..dab8c4bf8ae388 --- /dev/null +++ b/compat/mimalloc/heap.c @@ -0,0 +1,626 @@ +/*---------------------------------------------------------------------------- +Copyright (c) 2018-2021, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" +#include "mimalloc/prim.h" // mi_prim_get_default_heap + +#include // memset, memcpy + +#if defined(_MSC_VER) && (_MSC_VER < 1920) +#pragma warning(disable:4204) // non-constant aggregate initializer +#endif + +/* ----------------------------------------------------------- + Helpers +----------------------------------------------------------- */ + +// return `true` if ok, `false` to break +typedef bool (heap_page_visitor_fun)(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_t* page, void* arg1, void* arg2); + +// Visit all pages in a heap; returns `false` if break was called. +static bool mi_heap_visit_pages(mi_heap_t* heap, heap_page_visitor_fun* fn, void* arg1, void* arg2) +{ + if (heap==NULL || heap->page_count==0) return 0; + + // visit all pages + #if MI_DEBUG>1 + size_t total = heap->page_count; + size_t count = 0; + #endif + + for (size_t i = 0; i <= MI_BIN_FULL; i++) { + mi_page_queue_t* pq = &heap->pages[i]; + mi_page_t* page = pq->first; + while(page != NULL) { + mi_page_t* next = page->next; // save next in case the page gets removed from the queue + mi_assert_internal(mi_page_heap(page) == heap); + #if MI_DEBUG>1 + count++; + #endif + if (!fn(heap, pq, page, arg1, arg2)) return false; + page = next; // and continue + } + } + mi_assert_internal(count == total); + return true; +} + + +#if MI_DEBUG>=2 +static bool mi_heap_page_is_valid(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_t* page, void* arg1, void* arg2) { + MI_UNUSED(arg1); + MI_UNUSED(arg2); + MI_UNUSED(pq); + mi_assert_internal(mi_page_heap(page) == heap); + mi_segment_t* segment = _mi_page_segment(page); + mi_assert_internal(segment->thread_id == heap->thread_id); + mi_assert_expensive(_mi_page_is_valid(page)); + return true; +} +#endif +#if MI_DEBUG>=3 +static bool mi_heap_is_valid(mi_heap_t* heap) { + mi_assert_internal(heap!=NULL); + mi_heap_visit_pages(heap, &mi_heap_page_is_valid, NULL, NULL); + return true; +} +#endif + + + + +/* ----------------------------------------------------------- + "Collect" pages by migrating `local_free` and `thread_free` + lists and freeing empty pages. This is done when a thread + stops (and in that case abandons pages if there are still + blocks alive) +----------------------------------------------------------- */ + +typedef enum mi_collect_e { + MI_NORMAL, + MI_FORCE, + MI_ABANDON +} mi_collect_t; + + +static bool mi_heap_page_collect(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_t* page, void* arg_collect, void* arg2 ) { + MI_UNUSED(arg2); + MI_UNUSED(heap); + mi_assert_internal(mi_heap_page_is_valid(heap, pq, page, NULL, NULL)); + mi_collect_t collect = *((mi_collect_t*)arg_collect); + _mi_page_free_collect(page, collect >= MI_FORCE); + if (mi_page_all_free(page)) { + // no more used blocks, free the page. + // note: this will free retired pages as well. + _mi_page_free(page, pq, collect >= MI_FORCE); + } + else if (collect == MI_ABANDON) { + // still used blocks but the thread is done; abandon the page + _mi_page_abandon(page, pq); + } + return true; // don't break +} + +static bool mi_heap_page_never_delayed_free(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_t* page, void* arg1, void* arg2) { + MI_UNUSED(arg1); + MI_UNUSED(arg2); + MI_UNUSED(heap); + MI_UNUSED(pq); + _mi_page_use_delayed_free(page, MI_NEVER_DELAYED_FREE, false); + return true; // don't break +} + +static void mi_heap_collect_ex(mi_heap_t* heap, mi_collect_t collect) +{ + if (heap==NULL || !mi_heap_is_initialized(heap)) return; + + const bool force = collect >= MI_FORCE; + _mi_deferred_free(heap, force); + + // note: never reclaim on collect but leave it to threads that need storage to reclaim + const bool force_main = + #ifdef NDEBUG + collect == MI_FORCE + #else + collect >= MI_FORCE + #endif + && _mi_is_main_thread() && mi_heap_is_backing(heap) && !heap->no_reclaim; + + if (force_main) { + // the main thread is abandoned (end-of-program), try to reclaim all abandoned segments. + // if all memory is freed by now, all segments should be freed. + _mi_abandoned_reclaim_all(heap, &heap->tld->segments); + } + + // if abandoning, mark all pages to no longer add to delayed_free + if (collect == MI_ABANDON) { + mi_heap_visit_pages(heap, &mi_heap_page_never_delayed_free, NULL, NULL); + } + + // free all current thread delayed blocks. + // (if abandoning, after this there are no more thread-delayed references into the pages.) + _mi_heap_delayed_free_all(heap); + + // collect retired pages + _mi_heap_collect_retired(heap, force); + + // collect all pages owned by this thread + mi_heap_visit_pages(heap, &mi_heap_page_collect, &collect, NULL); + mi_assert_internal( collect != MI_ABANDON || mi_atomic_load_ptr_acquire(mi_block_t,&heap->thread_delayed_free) == NULL ); + + // collect abandoned segments (in particular, purge expired parts of segments in the abandoned segment list) + // note: forced purge can be quite expensive if many threads are created/destroyed so we do not force on abandonment + _mi_abandoned_collect(heap, collect == MI_FORCE /* force? */, &heap->tld->segments); + + // collect segment local caches + if (force) { + _mi_segment_thread_collect(&heap->tld->segments); + } + + // collect regions on program-exit (or shared library unload) + if (force && _mi_is_main_thread() && mi_heap_is_backing(heap)) { + _mi_thread_data_collect(); // collect thread data cache + _mi_arena_collect(true /* force purge */, &heap->tld->stats); + } +} + +void _mi_heap_collect_abandon(mi_heap_t* heap) { + mi_heap_collect_ex(heap, MI_ABANDON); +} + +void mi_heap_collect(mi_heap_t* heap, bool force) mi_attr_noexcept { + mi_heap_collect_ex(heap, (force ? MI_FORCE : MI_NORMAL)); +} + +void mi_collect(bool force) mi_attr_noexcept { + mi_heap_collect(mi_prim_get_default_heap(), force); +} + + +/* ----------------------------------------------------------- + Heap new +----------------------------------------------------------- */ + +mi_heap_t* mi_heap_get_default(void) { + mi_thread_init(); + return mi_prim_get_default_heap(); +} + +static bool mi_heap_is_default(const mi_heap_t* heap) { + return (heap == mi_prim_get_default_heap()); +} + + +mi_heap_t* mi_heap_get_backing(void) { + mi_heap_t* heap = mi_heap_get_default(); + mi_assert_internal(heap!=NULL); + mi_heap_t* bheap = heap->tld->heap_backing; + mi_assert_internal(bheap!=NULL); + mi_assert_internal(bheap->thread_id == _mi_thread_id()); + return bheap; +} + +mi_decl_nodiscard mi_heap_t* mi_heap_new_in_arena(mi_arena_id_t arena_id) { + mi_heap_t* bheap = mi_heap_get_backing(); + mi_heap_t* heap = mi_heap_malloc_tp(bheap, mi_heap_t); // todo: OS allocate in secure mode? + if (heap == NULL) return NULL; + _mi_memcpy_aligned(heap, &_mi_heap_empty, sizeof(mi_heap_t)); + heap->tld = bheap->tld; + heap->thread_id = _mi_thread_id(); + heap->arena_id = arena_id; + _mi_random_split(&bheap->random, &heap->random); + heap->cookie = _mi_heap_random_next(heap) | 1; + heap->keys[0] = _mi_heap_random_next(heap); + heap->keys[1] = _mi_heap_random_next(heap); + heap->no_reclaim = true; // don't reclaim abandoned pages or otherwise destroy is unsafe + // push on the thread local heaps list + heap->next = heap->tld->heaps; + heap->tld->heaps = heap; + return heap; +} + +mi_decl_nodiscard mi_heap_t* mi_heap_new(void) { + return mi_heap_new_in_arena(_mi_arena_id_none()); +} + +bool _mi_heap_memid_is_suitable(mi_heap_t* heap, mi_memid_t memid) { + return _mi_arena_memid_is_suitable(memid, heap->arena_id); +} + +uintptr_t _mi_heap_random_next(mi_heap_t* heap) { + return _mi_random_next(&heap->random); +} + +// zero out the page queues +static void mi_heap_reset_pages(mi_heap_t* heap) { + mi_assert_internal(heap != NULL); + mi_assert_internal(mi_heap_is_initialized(heap)); + // TODO: copy full empty heap instead? + memset(&heap->pages_free_direct, 0, sizeof(heap->pages_free_direct)); + _mi_memcpy_aligned(&heap->pages, &_mi_heap_empty.pages, sizeof(heap->pages)); + heap->thread_delayed_free = NULL; + heap->page_count = 0; +} + +// called from `mi_heap_destroy` and `mi_heap_delete` to free the internal heap resources. +static void mi_heap_free(mi_heap_t* heap) { + mi_assert(heap != NULL); + mi_assert_internal(mi_heap_is_initialized(heap)); + if (heap==NULL || !mi_heap_is_initialized(heap)) return; + if (mi_heap_is_backing(heap)) return; // dont free the backing heap + + // reset default + if (mi_heap_is_default(heap)) { + _mi_heap_set_default_direct(heap->tld->heap_backing); + } + + // remove ourselves from the thread local heaps list + // linear search but we expect the number of heaps to be relatively small + mi_heap_t* prev = NULL; + mi_heap_t* curr = heap->tld->heaps; + while (curr != heap && curr != NULL) { + prev = curr; + curr = curr->next; + } + mi_assert_internal(curr == heap); + if (curr == heap) { + if (prev != NULL) { prev->next = heap->next; } + else { heap->tld->heaps = heap->next; } + } + mi_assert_internal(heap->tld->heaps != NULL); + + // and free the used memory + mi_free(heap); +} + + +/* ----------------------------------------------------------- + Heap destroy +----------------------------------------------------------- */ + +static bool _mi_heap_page_destroy(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_t* page, void* arg1, void* arg2) { + MI_UNUSED(arg1); + MI_UNUSED(arg2); + MI_UNUSED(heap); + MI_UNUSED(pq); + + // ensure no more thread_delayed_free will be added + _mi_page_use_delayed_free(page, MI_NEVER_DELAYED_FREE, false); + + // stats + const size_t bsize = mi_page_block_size(page); + if (bsize > MI_MEDIUM_OBJ_SIZE_MAX) { + if (bsize <= MI_LARGE_OBJ_SIZE_MAX) { + mi_heap_stat_decrease(heap, large, bsize); + } + else { + mi_heap_stat_decrease(heap, huge, bsize); + } + } +#if (MI_STAT) + _mi_page_free_collect(page, false); // update used count + const size_t inuse = page->used; + if (bsize <= MI_LARGE_OBJ_SIZE_MAX) { + mi_heap_stat_decrease(heap, normal, bsize * inuse); +#if (MI_STAT>1) + mi_heap_stat_decrease(heap, normal_bins[_mi_bin(bsize)], inuse); +#endif + } + mi_heap_stat_decrease(heap, malloc, bsize * inuse); // todo: off for aligned blocks... +#endif + + /// pretend it is all free now + mi_assert_internal(mi_page_thread_free(page) == NULL); + page->used = 0; + + // and free the page + // mi_page_free(page,false); + page->next = NULL; + page->prev = NULL; + _mi_segment_page_free(page,false /* no force? */, &heap->tld->segments); + + return true; // keep going +} + +void _mi_heap_destroy_pages(mi_heap_t* heap) { + mi_heap_visit_pages(heap, &_mi_heap_page_destroy, NULL, NULL); + mi_heap_reset_pages(heap); +} + +#if MI_TRACK_HEAP_DESTROY +static bool mi_cdecl mi_heap_track_block_free(const mi_heap_t* heap, const mi_heap_area_t* area, void* block, size_t block_size, void* arg) { + MI_UNUSED(heap); MI_UNUSED(area); MI_UNUSED(arg); MI_UNUSED(block_size); + mi_track_free_size(block,mi_usable_size(block)); + return true; +} +#endif + +void mi_heap_destroy(mi_heap_t* heap) { + mi_assert(heap != NULL); + mi_assert(mi_heap_is_initialized(heap)); + mi_assert(heap->no_reclaim); + mi_assert_expensive(mi_heap_is_valid(heap)); + if (heap==NULL || !mi_heap_is_initialized(heap)) return; + if (!heap->no_reclaim) { + // don't free in case it may contain reclaimed pages + mi_heap_delete(heap); + } + else { + // track all blocks as freed + #if MI_TRACK_HEAP_DESTROY + mi_heap_visit_blocks(heap, true, mi_heap_track_block_free, NULL); + #endif + // free all pages + _mi_heap_destroy_pages(heap); + mi_heap_free(heap); + } +} + +// forcefully destroy all heaps in the current thread +void _mi_heap_unsafe_destroy_all(void) { + mi_heap_t* bheap = mi_heap_get_backing(); + mi_heap_t* curr = bheap->tld->heaps; + while (curr != NULL) { + mi_heap_t* next = curr->next; + if (curr->no_reclaim) { + mi_heap_destroy(curr); + } + else { + _mi_heap_destroy_pages(curr); + } + curr = next; + } +} + +/* ----------------------------------------------------------- + Safe Heap delete +----------------------------------------------------------- */ + +// Transfer the pages from one heap to the other +static void mi_heap_absorb(mi_heap_t* heap, mi_heap_t* from) { + mi_assert_internal(heap!=NULL); + if (from==NULL || from->page_count == 0) return; + + // reduce the size of the delayed frees + _mi_heap_delayed_free_partial(from); + + // transfer all pages by appending the queues; this will set a new heap field + // so threads may do delayed frees in either heap for a while. + // note: appending waits for each page to not be in the `MI_DELAYED_FREEING` state + // so after this only the new heap will get delayed frees + for (size_t i = 0; i <= MI_BIN_FULL; i++) { + mi_page_queue_t* pq = &heap->pages[i]; + mi_page_queue_t* append = &from->pages[i]; + size_t pcount = _mi_page_queue_append(heap, pq, append); + heap->page_count += pcount; + from->page_count -= pcount; + } + mi_assert_internal(from->page_count == 0); + + // and do outstanding delayed frees in the `from` heap + // note: be careful here as the `heap` field in all those pages no longer point to `from`, + // turns out to be ok as `_mi_heap_delayed_free` only visits the list and calls a + // the regular `_mi_free_delayed_block` which is safe. + _mi_heap_delayed_free_all(from); + #if !defined(_MSC_VER) || (_MSC_VER > 1900) // somehow the following line gives an error in VS2015, issue #353 + mi_assert_internal(mi_atomic_load_ptr_relaxed(mi_block_t,&from->thread_delayed_free) == NULL); + #endif + + // and reset the `from` heap + mi_heap_reset_pages(from); +} + +// Safe delete a heap without freeing any still allocated blocks in that heap. +void mi_heap_delete(mi_heap_t* heap) +{ + mi_assert(heap != NULL); + mi_assert(mi_heap_is_initialized(heap)); + mi_assert_expensive(mi_heap_is_valid(heap)); + if (heap==NULL || !mi_heap_is_initialized(heap)) return; + + if (!mi_heap_is_backing(heap)) { + // tranfer still used pages to the backing heap + mi_heap_absorb(heap->tld->heap_backing, heap); + } + else { + // the backing heap abandons its pages + _mi_heap_collect_abandon(heap); + } + mi_assert_internal(heap->page_count==0); + mi_heap_free(heap); +} + +mi_heap_t* mi_heap_set_default(mi_heap_t* heap) { + mi_assert(heap != NULL); + mi_assert(mi_heap_is_initialized(heap)); + if (heap==NULL || !mi_heap_is_initialized(heap)) return NULL; + mi_assert_expensive(mi_heap_is_valid(heap)); + mi_heap_t* old = mi_prim_get_default_heap(); + _mi_heap_set_default_direct(heap); + return old; +} + + + + +/* ----------------------------------------------------------- + Analysis +----------------------------------------------------------- */ + +// static since it is not thread safe to access heaps from other threads. +static mi_heap_t* mi_heap_of_block(const void* p) { + if (p == NULL) return NULL; + mi_segment_t* segment = _mi_ptr_segment(p); + bool valid = (_mi_ptr_cookie(segment) == segment->cookie); + mi_assert_internal(valid); + if mi_unlikely(!valid) return NULL; + return mi_page_heap(_mi_segment_page_of(segment,p)); +} + +bool mi_heap_contains_block(mi_heap_t* heap, const void* p) { + mi_assert(heap != NULL); + if (heap==NULL || !mi_heap_is_initialized(heap)) return false; + return (heap == mi_heap_of_block(p)); +} + + +static bool mi_heap_page_check_owned(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_t* page, void* p, void* vfound) { + MI_UNUSED(heap); + MI_UNUSED(pq); + bool* found = (bool*)vfound; + mi_segment_t* segment = _mi_page_segment(page); + void* start = _mi_page_start(segment, page, NULL); + void* end = (uint8_t*)start + (page->capacity * mi_page_block_size(page)); + *found = (p >= start && p < end); + return (!*found); // continue if not found +} + +bool mi_heap_check_owned(mi_heap_t* heap, const void* p) { + mi_assert(heap != NULL); + if (heap==NULL || !mi_heap_is_initialized(heap)) return false; + if (((uintptr_t)p & (MI_INTPTR_SIZE - 1)) != 0) return false; // only aligned pointers + bool found = false; + mi_heap_visit_pages(heap, &mi_heap_page_check_owned, (void*)p, &found); + return found; +} + +bool mi_check_owned(const void* p) { + return mi_heap_check_owned(mi_prim_get_default_heap(), p); +} + +/* ----------------------------------------------------------- + Visit all heap blocks and areas + Todo: enable visiting abandoned pages, and + enable visiting all blocks of all heaps across threads +----------------------------------------------------------- */ + +// Separate struct to keep `mi_page_t` out of the public interface +typedef struct mi_heap_area_ex_s { + mi_heap_area_t area; + mi_page_t* page; +} mi_heap_area_ex_t; + +static bool mi_heap_area_visit_blocks(const mi_heap_area_ex_t* xarea, mi_block_visit_fun* visitor, void* arg) { + mi_assert(xarea != NULL); + if (xarea==NULL) return true; + const mi_heap_area_t* area = &xarea->area; + mi_page_t* page = xarea->page; + mi_assert(page != NULL); + if (page == NULL) return true; + + _mi_page_free_collect(page,true); + mi_assert_internal(page->local_free == NULL); + if (page->used == 0) return true; + + const size_t bsize = mi_page_block_size(page); + const size_t ubsize = mi_page_usable_block_size(page); // without padding + size_t psize; + uint8_t* pstart = _mi_page_start(_mi_page_segment(page), page, &psize); + + if (page->capacity == 1) { + // optimize page with one block + mi_assert_internal(page->used == 1 && page->free == NULL); + return visitor(mi_page_heap(page), area, pstart, ubsize, arg); + } + + // create a bitmap of free blocks. + #define MI_MAX_BLOCKS (MI_SMALL_PAGE_SIZE / sizeof(void*)) + uintptr_t free_map[MI_MAX_BLOCKS / sizeof(uintptr_t)]; + memset(free_map, 0, sizeof(free_map)); + + #if MI_DEBUG>1 + size_t free_count = 0; + #endif + for (mi_block_t* block = page->free; block != NULL; block = mi_block_next(page,block)) { + #if MI_DEBUG>1 + free_count++; + #endif + mi_assert_internal((uint8_t*)block >= pstart && (uint8_t*)block < (pstart + psize)); + size_t offset = (uint8_t*)block - pstart; + mi_assert_internal(offset % bsize == 0); + size_t blockidx = offset / bsize; // Todo: avoid division? + mi_assert_internal( blockidx < MI_MAX_BLOCKS); + size_t bitidx = (blockidx / sizeof(uintptr_t)); + size_t bit = blockidx - (bitidx * sizeof(uintptr_t)); + free_map[bitidx] |= ((uintptr_t)1 << bit); + } + mi_assert_internal(page->capacity == (free_count + page->used)); + + // walk through all blocks skipping the free ones + #if MI_DEBUG>1 + size_t used_count = 0; + #endif + for (size_t i = 0; i < page->capacity; i++) { + size_t bitidx = (i / sizeof(uintptr_t)); + size_t bit = i - (bitidx * sizeof(uintptr_t)); + uintptr_t m = free_map[bitidx]; + if (bit == 0 && m == UINTPTR_MAX) { + i += (sizeof(uintptr_t) - 1); // skip a run of free blocks + } + else if ((m & ((uintptr_t)1 << bit)) == 0) { + #if MI_DEBUG>1 + used_count++; + #endif + uint8_t* block = pstart + (i * bsize); + if (!visitor(mi_page_heap(page), area, block, ubsize, arg)) return false; + } + } + mi_assert_internal(page->used == used_count); + return true; +} + +typedef bool (mi_heap_area_visit_fun)(const mi_heap_t* heap, const mi_heap_area_ex_t* area, void* arg); + + +static bool mi_heap_visit_areas_page(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_t* page, void* vfun, void* arg) { + MI_UNUSED(heap); + MI_UNUSED(pq); + mi_heap_area_visit_fun* fun = (mi_heap_area_visit_fun*)vfun; + mi_heap_area_ex_t xarea; + const size_t bsize = mi_page_block_size(page); + const size_t ubsize = mi_page_usable_block_size(page); + xarea.page = page; + xarea.area.reserved = page->reserved * bsize; + xarea.area.committed = page->capacity * bsize; + xarea.area.blocks = _mi_page_start(_mi_page_segment(page), page, NULL); + xarea.area.used = page->used; // number of blocks in use (#553) + xarea.area.block_size = ubsize; + xarea.area.full_block_size = bsize; + return fun(heap, &xarea, arg); +} + +// Visit all heap pages as areas +static bool mi_heap_visit_areas(const mi_heap_t* heap, mi_heap_area_visit_fun* visitor, void* arg) { + if (visitor == NULL) return false; + return mi_heap_visit_pages((mi_heap_t*)heap, &mi_heap_visit_areas_page, (void*)(visitor), arg); // note: function pointer to void* :-{ +} + +// Just to pass arguments +typedef struct mi_visit_blocks_args_s { + bool visit_blocks; + mi_block_visit_fun* visitor; + void* arg; +} mi_visit_blocks_args_t; + +static bool mi_heap_area_visitor(const mi_heap_t* heap, const mi_heap_area_ex_t* xarea, void* arg) { + mi_visit_blocks_args_t* args = (mi_visit_blocks_args_t*)arg; + if (!args->visitor(heap, &xarea->area, NULL, xarea->area.block_size, args->arg)) return false; + if (args->visit_blocks) { + return mi_heap_area_visit_blocks(xarea, args->visitor, args->arg); + } + else { + return true; + } +} + +// Visit all blocks in a heap +bool mi_heap_visit_blocks(const mi_heap_t* heap, bool visit_blocks, mi_block_visit_fun* visitor, void* arg) { + mi_visit_blocks_args_t args = { visit_blocks, visitor, arg }; + return mi_heap_visit_areas(heap, &mi_heap_area_visitor, &args); +} diff --git a/compat/mimalloc/init.c b/compat/mimalloc/init.c new file mode 100644 index 00000000000000..4670d5510db187 --- /dev/null +++ b/compat/mimalloc/init.c @@ -0,0 +1,709 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2022, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/prim.h" + +#include // memcpy, memset +#include // atexit + + +// Empty page used to initialize the small free pages array +const mi_page_t _mi_page_empty = { + 0, false, false, false, + 0, // capacity + 0, // reserved capacity + { 0 }, // flags + false, // is_zero + 0, // retire_expire + NULL, // free + 0, // used + 0, // xblock_size + NULL, // local_free + #if (MI_PADDING || MI_ENCODE_FREELIST) + { 0, 0 }, + #endif + MI_ATOMIC_VAR_INIT(0), // xthread_free + MI_ATOMIC_VAR_INIT(0), // xheap + NULL, NULL + #if MI_INTPTR_SIZE==8 + , { 0 } // padding + #endif +}; + +#define MI_PAGE_EMPTY() ((mi_page_t*)&_mi_page_empty) + +#if (MI_SMALL_WSIZE_MAX==128) +#if (MI_PADDING>0) && (MI_INTPTR_SIZE >= 8) +#define MI_SMALL_PAGES_EMPTY { MI_INIT128(MI_PAGE_EMPTY), MI_PAGE_EMPTY(), MI_PAGE_EMPTY() } +#elif (MI_PADDING>0) +#define MI_SMALL_PAGES_EMPTY { MI_INIT128(MI_PAGE_EMPTY), MI_PAGE_EMPTY(), MI_PAGE_EMPTY(), MI_PAGE_EMPTY() } +#else +#define MI_SMALL_PAGES_EMPTY { MI_INIT128(MI_PAGE_EMPTY), MI_PAGE_EMPTY() } +#endif +#else +#error "define right initialization sizes corresponding to MI_SMALL_WSIZE_MAX" +#endif + +// Empty page queues for every bin +#define QNULL(sz) { NULL, NULL, (sz)*sizeof(uintptr_t) } +#define MI_PAGE_QUEUES_EMPTY \ + { QNULL(1), \ + QNULL( 1), QNULL( 2), QNULL( 3), QNULL( 4), QNULL( 5), QNULL( 6), QNULL( 7), QNULL( 8), /* 8 */ \ + QNULL( 10), QNULL( 12), QNULL( 14), QNULL( 16), QNULL( 20), QNULL( 24), QNULL( 28), QNULL( 32), /* 16 */ \ + QNULL( 40), QNULL( 48), QNULL( 56), QNULL( 64), QNULL( 80), QNULL( 96), QNULL( 112), QNULL( 128), /* 24 */ \ + QNULL( 160), QNULL( 192), QNULL( 224), QNULL( 256), QNULL( 320), QNULL( 384), QNULL( 448), QNULL( 512), /* 32 */ \ + QNULL( 640), QNULL( 768), QNULL( 896), QNULL( 1024), QNULL( 1280), QNULL( 1536), QNULL( 1792), QNULL( 2048), /* 40 */ \ + QNULL( 2560), QNULL( 3072), QNULL( 3584), QNULL( 4096), QNULL( 5120), QNULL( 6144), QNULL( 7168), QNULL( 8192), /* 48 */ \ + QNULL( 10240), QNULL( 12288), QNULL( 14336), QNULL( 16384), QNULL( 20480), QNULL( 24576), QNULL( 28672), QNULL( 32768), /* 56 */ \ + QNULL( 40960), QNULL( 49152), QNULL( 57344), QNULL( 65536), QNULL( 81920), QNULL( 98304), QNULL(114688), QNULL(131072), /* 64 */ \ + QNULL(163840), QNULL(196608), QNULL(229376), QNULL(262144), QNULL(327680), QNULL(393216), QNULL(458752), QNULL(524288), /* 72 */ \ + QNULL(MI_MEDIUM_OBJ_WSIZE_MAX + 1 /* 655360, Huge queue */), \ + QNULL(MI_MEDIUM_OBJ_WSIZE_MAX + 2) /* Full queue */ } + +#define MI_STAT_COUNT_NULL() {0,0,0,0} + +// Empty statistics +#if MI_STAT>1 +#define MI_STAT_COUNT_END_NULL() , { MI_STAT_COUNT_NULL(), MI_INIT32(MI_STAT_COUNT_NULL) } +#else +#define MI_STAT_COUNT_END_NULL() +#endif + +#define MI_STATS_NULL \ + MI_STAT_COUNT_NULL(), MI_STAT_COUNT_NULL(), \ + MI_STAT_COUNT_NULL(), MI_STAT_COUNT_NULL(), \ + MI_STAT_COUNT_NULL(), MI_STAT_COUNT_NULL(), \ + MI_STAT_COUNT_NULL(), MI_STAT_COUNT_NULL(), \ + MI_STAT_COUNT_NULL(), MI_STAT_COUNT_NULL(), \ + MI_STAT_COUNT_NULL(), MI_STAT_COUNT_NULL(), \ + MI_STAT_COUNT_NULL(), MI_STAT_COUNT_NULL(), \ + MI_STAT_COUNT_NULL(), \ + { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \ + { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 } \ + MI_STAT_COUNT_END_NULL() + + +// Empty slice span queues for every bin +#define SQNULL(sz) { NULL, NULL, sz } +#define MI_SEGMENT_SPAN_QUEUES_EMPTY \ + { SQNULL(1), \ + SQNULL( 1), SQNULL( 2), SQNULL( 3), SQNULL( 4), SQNULL( 5), SQNULL( 6), SQNULL( 7), SQNULL( 10), /* 8 */ \ + SQNULL( 12), SQNULL( 14), SQNULL( 16), SQNULL( 20), SQNULL( 24), SQNULL( 28), SQNULL( 32), SQNULL( 40), /* 16 */ \ + SQNULL( 48), SQNULL( 56), SQNULL( 64), SQNULL( 80), SQNULL( 96), SQNULL( 112), SQNULL( 128), SQNULL( 160), /* 24 */ \ + SQNULL( 192), SQNULL( 224), SQNULL( 256), SQNULL( 320), SQNULL( 384), SQNULL( 448), SQNULL( 512), SQNULL( 640), /* 32 */ \ + SQNULL( 768), SQNULL( 896), SQNULL( 1024) /* 35 */ } + + +// -------------------------------------------------------- +// Statically allocate an empty heap as the initial +// thread local value for the default heap, +// and statically allocate the backing heap for the main +// thread so it can function without doing any allocation +// itself (as accessing a thread local for the first time +// may lead to allocation itself on some platforms) +// -------------------------------------------------------- + +mi_decl_cache_align const mi_heap_t _mi_heap_empty = { + NULL, + MI_SMALL_PAGES_EMPTY, + MI_PAGE_QUEUES_EMPTY, + MI_ATOMIC_VAR_INIT(NULL), + 0, // tid + 0, // cookie + 0, // arena id + { 0, 0 }, // keys + { {0}, {0}, 0, true }, // random + 0, // page count + MI_BIN_FULL, 0, // page retired min/max + NULL, // next + false +}; + +#define tld_empty_stats ((mi_stats_t*)((uint8_t*)&tld_empty + offsetof(mi_tld_t,stats))) +#define tld_empty_os ((mi_os_tld_t*)((uint8_t*)&tld_empty + offsetof(mi_tld_t,os))) + +mi_decl_cache_align static const mi_tld_t tld_empty = { + 0, + false, + NULL, NULL, + { MI_SEGMENT_SPAN_QUEUES_EMPTY, 0, 0, 0, 0, tld_empty_stats, tld_empty_os }, // segments + { 0, tld_empty_stats }, // os + { MI_STATS_NULL } // stats +}; + +mi_threadid_t _mi_thread_id(void) mi_attr_noexcept { + return _mi_prim_thread_id(); +} + +// the thread-local default heap for allocation +mi_decl_thread mi_heap_t* _mi_heap_default = (mi_heap_t*)&_mi_heap_empty; + +extern mi_heap_t _mi_heap_main; + +static mi_tld_t tld_main = { + 0, false, + &_mi_heap_main, & _mi_heap_main, + { MI_SEGMENT_SPAN_QUEUES_EMPTY, 0, 0, 0, 0, &tld_main.stats, &tld_main.os }, // segments + { 0, &tld_main.stats }, // os + { MI_STATS_NULL } // stats +}; + +mi_heap_t _mi_heap_main = { + &tld_main, + MI_SMALL_PAGES_EMPTY, + MI_PAGE_QUEUES_EMPTY, + MI_ATOMIC_VAR_INIT(NULL), + 0, // thread id + 0, // initial cookie + 0, // arena id + { 0, 0 }, // the key of the main heap can be fixed (unlike page keys that need to be secure!) + { {0x846ca68b}, {0}, 0, true }, // random + 0, // page count + MI_BIN_FULL, 0, // page retired min/max + NULL, // next heap + false // can reclaim +}; + +bool _mi_process_is_initialized = false; // set to `true` in `mi_process_init`. + +mi_stats_t _mi_stats_main = { MI_STATS_NULL }; + + +static void mi_heap_main_init(void) { + if (_mi_heap_main.cookie == 0) { + _mi_heap_main.thread_id = _mi_thread_id(); + _mi_heap_main.cookie = 1; + #if defined(_WIN32) && !defined(MI_SHARED_LIB) + _mi_random_init_weak(&_mi_heap_main.random); // prevent allocation failure during bcrypt dll initialization with static linking + #else + _mi_random_init(&_mi_heap_main.random); + #endif + _mi_heap_main.cookie = _mi_heap_random_next(&_mi_heap_main); + _mi_heap_main.keys[0] = _mi_heap_random_next(&_mi_heap_main); + _mi_heap_main.keys[1] = _mi_heap_random_next(&_mi_heap_main); + } +} + +mi_heap_t* _mi_heap_main_get(void) { + mi_heap_main_init(); + return &_mi_heap_main; +} + + +/* ----------------------------------------------------------- + Initialization and freeing of the thread local heaps +----------------------------------------------------------- */ + +// note: in x64 in release build `sizeof(mi_thread_data_t)` is under 4KiB (= OS page size). +typedef struct mi_thread_data_s { + mi_heap_t heap; // must come first due to cast in `_mi_heap_done` + mi_tld_t tld; + mi_memid_t memid; +} mi_thread_data_t; + + +// Thread meta-data is allocated directly from the OS. For +// some programs that do not use thread pools and allocate and +// destroy many OS threads, this may causes too much overhead +// per thread so we maintain a small cache of recently freed metadata. + +#define TD_CACHE_SIZE (16) +static _Atomic(mi_thread_data_t*) td_cache[TD_CACHE_SIZE]; + +static mi_thread_data_t* mi_thread_data_zalloc(void) { + // try to find thread metadata in the cache + bool is_zero = false; + mi_thread_data_t* td = NULL; + for (int i = 0; i < TD_CACHE_SIZE; i++) { + td = mi_atomic_load_ptr_relaxed(mi_thread_data_t, &td_cache[i]); + if (td != NULL) { + // found cached allocation, try use it + td = mi_atomic_exchange_ptr_acq_rel(mi_thread_data_t, &td_cache[i], NULL); + if (td != NULL) { + break; + } + } + } + + // if that fails, allocate as meta data + if (td == NULL) { + mi_memid_t memid; + td = (mi_thread_data_t*)_mi_os_alloc(sizeof(mi_thread_data_t), &memid, &_mi_stats_main); + if (td == NULL) { + // if this fails, try once more. (issue #257) + td = (mi_thread_data_t*)_mi_os_alloc(sizeof(mi_thread_data_t), &memid, &_mi_stats_main); + if (td == NULL) { + // really out of memory + _mi_error_message(ENOMEM, "unable to allocate thread local heap metadata (%zu bytes)\n", sizeof(mi_thread_data_t)); + } + } + if (td != NULL) { + td->memid = memid; + is_zero = memid.initially_zero; + } + } + + if (td != NULL && !is_zero) { + _mi_memzero_aligned(td, sizeof(*td)); + } + return td; +} + +static void mi_thread_data_free( mi_thread_data_t* tdfree ) { + // try to add the thread metadata to the cache + for (int i = 0; i < TD_CACHE_SIZE; i++) { + mi_thread_data_t* td = mi_atomic_load_ptr_relaxed(mi_thread_data_t, &td_cache[i]); + if (td == NULL) { + mi_thread_data_t* expected = NULL; + if (mi_atomic_cas_ptr_weak_acq_rel(mi_thread_data_t, &td_cache[i], &expected, tdfree)) { + return; + } + } + } + // if that fails, just free it directly + _mi_os_free(tdfree, sizeof(mi_thread_data_t), tdfree->memid, &_mi_stats_main); +} + +void _mi_thread_data_collect(void) { + // free all thread metadata from the cache + for (int i = 0; i < TD_CACHE_SIZE; i++) { + mi_thread_data_t* td = mi_atomic_load_ptr_relaxed(mi_thread_data_t, &td_cache[i]); + if (td != NULL) { + td = mi_atomic_exchange_ptr_acq_rel(mi_thread_data_t, &td_cache[i], NULL); + if (td != NULL) { + _mi_os_free(td, sizeof(mi_thread_data_t), td->memid, &_mi_stats_main); + } + } + } +} + +// Initialize the thread local default heap, called from `mi_thread_init` +static bool _mi_heap_init(void) { + if (mi_heap_is_initialized(mi_prim_get_default_heap())) return true; + if (_mi_is_main_thread()) { + // mi_assert_internal(_mi_heap_main.thread_id != 0); // can happen on freeBSD where alloc is called before any initialization + // the main heap is statically allocated + mi_heap_main_init(); + _mi_heap_set_default_direct(&_mi_heap_main); + //mi_assert_internal(_mi_heap_default->tld->heap_backing == mi_prim_get_default_heap()); + } + else { + // use `_mi_os_alloc` to allocate directly from the OS + mi_thread_data_t* td = mi_thread_data_zalloc(); + if (td == NULL) return false; + + mi_tld_t* tld = &td->tld; + mi_heap_t* heap = &td->heap; + _mi_memcpy_aligned(tld, &tld_empty, sizeof(*tld)); + _mi_memcpy_aligned(heap, &_mi_heap_empty, sizeof(*heap)); + heap->thread_id = _mi_thread_id(); + _mi_random_init(&heap->random); + heap->cookie = _mi_heap_random_next(heap) | 1; + heap->keys[0] = _mi_heap_random_next(heap); + heap->keys[1] = _mi_heap_random_next(heap); + heap->tld = tld; + tld->heap_backing = heap; + tld->heaps = heap; + tld->segments.stats = &tld->stats; + tld->segments.os = &tld->os; + tld->os.stats = &tld->stats; + _mi_heap_set_default_direct(heap); + } + return false; +} + +// Free the thread local default heap (called from `mi_thread_done`) +static bool _mi_heap_done(mi_heap_t* heap) { + if (!mi_heap_is_initialized(heap)) return true; + + // reset default heap + _mi_heap_set_default_direct(_mi_is_main_thread() ? &_mi_heap_main : (mi_heap_t*)&_mi_heap_empty); + + // switch to backing heap + heap = heap->tld->heap_backing; + if (!mi_heap_is_initialized(heap)) return false; + + // delete all non-backing heaps in this thread + mi_heap_t* curr = heap->tld->heaps; + while (curr != NULL) { + mi_heap_t* next = curr->next; // save `next` as `curr` will be freed + if (curr != heap) { + mi_assert_internal(!mi_heap_is_backing(curr)); + mi_heap_delete(curr); + } + curr = next; + } + mi_assert_internal(heap->tld->heaps == heap && heap->next == NULL); + mi_assert_internal(mi_heap_is_backing(heap)); + + // collect if not the main thread + if (heap != &_mi_heap_main) { + _mi_heap_collect_abandon(heap); + } + + // merge stats + _mi_stats_done(&heap->tld->stats); + + // free if not the main thread + if (heap != &_mi_heap_main) { + // the following assertion does not always hold for huge segments as those are always treated + // as abondened: one may allocate it in one thread, but deallocate in another in which case + // the count can be too large or negative. todo: perhaps not count huge segments? see issue #363 + // mi_assert_internal(heap->tld->segments.count == 0 || heap->thread_id != _mi_thread_id()); + mi_thread_data_free((mi_thread_data_t*)heap); + } + else { + #if 0 + // never free the main thread even in debug mode; if a dll is linked statically with mimalloc, + // there may still be delete/free calls after the mi_fls_done is called. Issue #207 + _mi_heap_destroy_pages(heap); + mi_assert_internal(heap->tld->heap_backing == &_mi_heap_main); + #endif + } + return false; +} + + + +// -------------------------------------------------------- +// Try to run `mi_thread_done()` automatically so any memory +// owned by the thread but not yet released can be abandoned +// and re-owned by another thread. +// +// 1. windows dynamic library: +// call from DllMain on DLL_THREAD_DETACH +// 2. windows static library: +// use `FlsAlloc` to call a destructor when the thread is done +// 3. unix, pthreads: +// use a pthread key to call a destructor when a pthread is done +// +// In the last two cases we also need to call `mi_process_init` +// to set up the thread local keys. +// -------------------------------------------------------- + +// Set up handlers so `mi_thread_done` is called automatically +static void mi_process_setup_auto_thread_done(void) { + static bool tls_initialized = false; // fine if it races + if (tls_initialized) return; + tls_initialized = true; + _mi_prim_thread_init_auto_done(); + _mi_heap_set_default_direct(&_mi_heap_main); +} + + +bool _mi_is_main_thread(void) { + return (_mi_heap_main.thread_id==0 || _mi_heap_main.thread_id == _mi_thread_id()); +} + +static _Atomic(size_t) thread_count = MI_ATOMIC_VAR_INIT(1); + +size_t _mi_current_thread_count(void) { + return mi_atomic_load_relaxed(&thread_count); +} + +// This is called from the `mi_malloc_generic` +void mi_thread_init(void) mi_attr_noexcept +{ + // ensure our process has started already + mi_process_init(); + + // initialize the thread local default heap + // (this will call `_mi_heap_set_default_direct` and thus set the + // fiber/pthread key to a non-zero value, ensuring `_mi_thread_done` is called) + if (_mi_heap_init()) return; // returns true if already initialized + + _mi_stat_increase(&_mi_stats_main.threads, 1); + mi_atomic_increment_relaxed(&thread_count); + //_mi_verbose_message("thread init: 0x%zx\n", _mi_thread_id()); +} + +void mi_thread_done(void) mi_attr_noexcept { + _mi_thread_done(NULL); +} + +void _mi_thread_done(mi_heap_t* heap) +{ + // calling with NULL implies using the default heap + if (heap == NULL) { + heap = mi_prim_get_default_heap(); + if (heap == NULL) return; + } + + // prevent re-entrancy through heap_done/heap_set_default_direct (issue #699) + if (!mi_heap_is_initialized(heap)) { + return; + } + + // adjust stats + mi_atomic_decrement_relaxed(&thread_count); + _mi_stat_decrease(&_mi_stats_main.threads, 1); + + // check thread-id as on Windows shutdown with FLS the main (exit) thread may call this on thread-local heaps... + if (heap->thread_id != _mi_thread_id()) return; + + // abandon the thread local heap + if (_mi_heap_done(heap)) return; // returns true if already ran +} + +void _mi_heap_set_default_direct(mi_heap_t* heap) { + mi_assert_internal(heap != NULL); + #if defined(MI_TLS_SLOT) + mi_prim_tls_slot_set(MI_TLS_SLOT,heap); + #elif defined(MI_TLS_PTHREAD_SLOT_OFS) + *mi_tls_pthread_heap_slot() = heap; + #elif defined(MI_TLS_PTHREAD) + // we use _mi_heap_default_key + #else + _mi_heap_default = heap; + #endif + + // ensure the default heap is passed to `_mi_thread_done` + // setting to a non-NULL value also ensures `mi_thread_done` is called. + _mi_prim_thread_associate_default_heap(heap); +} + + +// -------------------------------------------------------- +// Run functions on process init/done, and thread init/done +// -------------------------------------------------------- +static void mi_cdecl mi_process_done(void); + +static bool os_preloading = true; // true until this module is initialized +static bool mi_redirected = false; // true if malloc redirects to mi_malloc + +// Returns true if this module has not been initialized; Don't use C runtime routines until it returns false. +bool mi_decl_noinline _mi_preloading(void) { + return os_preloading; +} + +mi_decl_nodiscard bool mi_is_redirected(void) mi_attr_noexcept { + return mi_redirected; +} + +// Communicate with the redirection module on Windows +#if defined(_WIN32) && defined(MI_SHARED_LIB) && !defined(MI_WIN_NOREDIRECT) +#ifdef __cplusplus +extern "C" { +#endif +mi_decl_export void _mi_redirect_entry(DWORD reason) { + // called on redirection; careful as this may be called before DllMain + if (reason == DLL_PROCESS_ATTACH) { + mi_redirected = true; + } + else if (reason == DLL_PROCESS_DETACH) { + mi_redirected = false; + } + else if (reason == DLL_THREAD_DETACH) { + mi_thread_done(); + } +} +__declspec(dllimport) bool mi_cdecl mi_allocator_init(const char** message); +__declspec(dllimport) void mi_cdecl mi_allocator_done(void); +#ifdef __cplusplus +} +#endif +#else +static bool mi_allocator_init(const char** message) { + if (message != NULL) *message = NULL; + return true; +} +static void mi_allocator_done(void) { + // nothing to do +} +#endif + +// Called once by the process loader +static void mi_process_load(void) { + mi_heap_main_init(); + #if defined(__APPLE__) || defined(MI_TLS_RECURSE_GUARD) + volatile mi_heap_t* dummy = _mi_heap_default; // access TLS to allocate it before setting tls_initialized to true; + if (dummy == NULL) return; // use dummy or otherwise the access may get optimized away (issue #697) + #endif + os_preloading = false; + mi_assert_internal(_mi_is_main_thread()); + #if !(defined(_WIN32) && defined(MI_SHARED_LIB)) // use Dll process detach (see below) instead of atexit (issue #521) + atexit(&mi_process_done); + #endif + _mi_options_init(); + mi_process_setup_auto_thread_done(); + mi_process_init(); + if (mi_redirected) _mi_verbose_message("malloc is redirected.\n"); + + // show message from the redirector (if present) + const char* msg = NULL; + mi_allocator_init(&msg); + if (msg != NULL && (mi_option_is_enabled(mi_option_verbose) || mi_option_is_enabled(mi_option_show_errors))) { + _mi_fputs(NULL,NULL,NULL,msg); + } + + // reseed random + _mi_random_reinit_if_weak(&_mi_heap_main.random); +} + +#if defined(_WIN32) && (defined(_M_IX86) || defined(_M_X64)) +#include +mi_decl_cache_align bool _mi_cpu_has_fsrm = false; + +static void mi_detect_cpu_features(void) { + // FSRM for fast rep movsb support (AMD Zen3+ (~2020) or Intel Ice Lake+ (~2017)) + int32_t cpu_info[4]; + __cpuid(cpu_info, 7); + _mi_cpu_has_fsrm = ((cpu_info[3] & (1 << 4)) != 0); // bit 4 of EDX : see +} +#else +static void mi_detect_cpu_features(void) { + // nothing +} +#endif + +// Initialize the process; called by thread_init or the process loader +void mi_process_init(void) mi_attr_noexcept { + // ensure we are called once + static mi_atomic_once_t process_init; + #if _MSC_VER < 1920 + mi_heap_main_init(); // vs2017 can dynamically re-initialize _mi_heap_main + #endif + if (!mi_atomic_once(&process_init)) return; + _mi_process_is_initialized = true; + _mi_verbose_message("process init: 0x%zx\n", _mi_thread_id()); + mi_process_setup_auto_thread_done(); + + mi_detect_cpu_features(); + _mi_os_init(); + mi_heap_main_init(); + #if MI_DEBUG + _mi_verbose_message("debug level : %d\n", MI_DEBUG); + #endif + _mi_verbose_message("secure level: %d\n", MI_SECURE); + _mi_verbose_message("mem tracking: %s\n", MI_TRACK_TOOL); + #if MI_TSAN + _mi_verbose_message("thread santizer enabled\n"); + #endif + mi_thread_init(); + + #if defined(_WIN32) + // On windows, when building as a static lib the FLS cleanup happens to early for the main thread. + // To avoid this, set the FLS value for the main thread to NULL so the fls cleanup + // will not call _mi_thread_done on the (still executing) main thread. See issue #508. + _mi_prim_thread_associate_default_heap(NULL); + #endif + + mi_stats_reset(); // only call stat reset *after* thread init (or the heap tld == NULL) + mi_track_init(); + + if (mi_option_is_enabled(mi_option_reserve_huge_os_pages)) { + size_t pages = mi_option_get_clamp(mi_option_reserve_huge_os_pages, 0, 128*1024); + long reserve_at = mi_option_get(mi_option_reserve_huge_os_pages_at); + if (reserve_at != -1) { + mi_reserve_huge_os_pages_at(pages, reserve_at, pages*500); + } else { + mi_reserve_huge_os_pages_interleave(pages, 0, pages*500); + } + } + if (mi_option_is_enabled(mi_option_reserve_os_memory)) { + long ksize = mi_option_get(mi_option_reserve_os_memory); + if (ksize > 0) { + mi_reserve_os_memory((size_t)ksize*MI_KiB, true /* commit? */, true /* allow large pages? */); + } + } +} + +// Called when the process is done (through `at_exit`) +static void mi_cdecl mi_process_done(void) { + // only shutdown if we were initialized + if (!_mi_process_is_initialized) return; + // ensure we are called once + static bool process_done = false; + if (process_done) return; + process_done = true; + + // release any thread specific resources and ensure _mi_thread_done is called on all but the main thread + _mi_prim_thread_done_auto_done(); + + #ifndef MI_SKIP_COLLECT_ON_EXIT + #if (MI_DEBUG || !defined(MI_SHARED_LIB)) + // free all memory if possible on process exit. This is not needed for a stand-alone process + // but should be done if mimalloc is statically linked into another shared library which + // is repeatedly loaded/unloaded, see issue #281. + mi_collect(true /* force */ ); + #endif + #endif + + // Forcefully release all retained memory; this can be dangerous in general if overriding regular malloc/free + // since after process_done there might still be other code running that calls `free` (like at_exit routines, + // or C-runtime termination code. + if (mi_option_is_enabled(mi_option_destroy_on_exit)) { + mi_collect(true /* force */); + _mi_heap_unsafe_destroy_all(); // forcefully release all memory held by all heaps (of this thread only!) + _mi_arena_unsafe_destroy_all(& _mi_heap_main_get()->tld->stats); + } + + if (mi_option_is_enabled(mi_option_show_stats) || mi_option_is_enabled(mi_option_verbose)) { + mi_stats_print(NULL); + } + mi_allocator_done(); + _mi_verbose_message("process done: 0x%zx\n", _mi_heap_main.thread_id); + os_preloading = true; // don't call the C runtime anymore +} + + + +#if defined(_WIN32) && defined(MI_SHARED_LIB) + // Windows DLL: easy to hook into process_init and thread_done + __declspec(dllexport) BOOL WINAPI DllMain(HINSTANCE inst, DWORD reason, LPVOID reserved) { + MI_UNUSED(reserved); + MI_UNUSED(inst); + if (reason==DLL_PROCESS_ATTACH) { + mi_process_load(); + } + else if (reason==DLL_PROCESS_DETACH) { + mi_process_done(); + } + else if (reason==DLL_THREAD_DETACH) { + if (!mi_is_redirected()) { + mi_thread_done(); + } + } + return TRUE; + } + +#elif defined(_MSC_VER) + // MSVC: use data section magic for static libraries + // See + static int _mi_process_init(void) { + mi_process_load(); + return 0; + } + typedef int(*_mi_crt_callback_t)(void); + #if defined(_M_X64) || defined(_M_ARM64) + __pragma(comment(linker, "/include:" "_mi_msvc_initu")) + #pragma section(".CRT$XIU", long, read) + #else + __pragma(comment(linker, "/include:" "__mi_msvc_initu")) + #endif + #pragma data_seg(".CRT$XIU") + mi_decl_externc _mi_crt_callback_t _mi_msvc_initu[] = { &_mi_process_init }; + #pragma data_seg() + +#elif defined(__cplusplus) + // C++: use static initialization to detect process start + static bool _mi_process_init(void) { + mi_process_load(); + return (_mi_heap_main.thread_id != 0); + } + static bool mi_initialized = _mi_process_init(); + +#elif defined(__GNUC__) || defined(__clang__) + // GCC,Clang: use the constructor attribute + static void __attribute__((constructor)) _mi_process_init(void) { + mi_process_load(); + } + +#else +#pragma message("define a way to call mi_process_load on your platform") +#endif diff --git a/compat/mimalloc/mimalloc.h b/compat/mimalloc/mimalloc.h new file mode 100644 index 00000000000000..c0f5e96e51e975 --- /dev/null +++ b/compat/mimalloc/mimalloc.h @@ -0,0 +1,565 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#pragma once +#ifndef MIMALLOC_H +#define MIMALLOC_H + +#define MI_MALLOC_VERSION 212 // major + 2 digits minor + +// ------------------------------------------------------ +// Compiler specific attributes +// ------------------------------------------------------ + +#ifdef __cplusplus + #if (__cplusplus >= 201103L) || (_MSC_VER > 1900) // C++11 + #define mi_attr_noexcept noexcept + #else + #define mi_attr_noexcept throw() + #endif +#else + #define mi_attr_noexcept +#endif + +#if defined(__cplusplus) && (__cplusplus >= 201703) + #define mi_decl_nodiscard [[nodiscard]] +#elif (defined(__GNUC__) && (__GNUC__ >= 4)) || defined(__clang__) // includes clang, icc, and clang-cl + #define mi_decl_nodiscard __attribute__((warn_unused_result)) +#elif defined(_HAS_NODISCARD) + #define mi_decl_nodiscard _NODISCARD +#elif (_MSC_VER >= 1700) + #define mi_decl_nodiscard _Check_return_ +#else + #define mi_decl_nodiscard +#endif + +#if defined(_MSC_VER) || defined(__MINGW32__) + #if !defined(MI_SHARED_LIB) + #define mi_decl_export + #elif defined(MI_SHARED_LIB_EXPORT) + #define mi_decl_export __declspec(dllexport) + #else + #define mi_decl_export __declspec(dllimport) + #endif + #if defined(__MINGW32__) + #define mi_decl_restrict + #define mi_attr_malloc __attribute__((malloc)) + #else + #if (_MSC_VER >= 1900) && !defined(__EDG__) + #define mi_decl_restrict __declspec(allocator) __declspec(restrict) + #else + #define mi_decl_restrict __declspec(restrict) + #endif + #define mi_attr_malloc + #endif + #define mi_cdecl __cdecl + #define mi_attr_alloc_size(s) + #define mi_attr_alloc_size2(s1,s2) + #define mi_attr_alloc_align(p) +#elif defined(__GNUC__) // includes clang and icc + #if defined(MI_SHARED_LIB) && defined(MI_SHARED_LIB_EXPORT) + #define mi_decl_export __attribute__((visibility("default"))) + #else + #define mi_decl_export + #endif + #define mi_cdecl // leads to warnings... __attribute__((cdecl)) + #define mi_decl_restrict + #define mi_attr_malloc __attribute__((malloc)) + #if (defined(__clang_major__) && (__clang_major__ < 4)) || (__GNUC__ < 5) + #define mi_attr_alloc_size(s) + #define mi_attr_alloc_size2(s1,s2) + #define mi_attr_alloc_align(p) + #elif defined(__INTEL_COMPILER) + #define mi_attr_alloc_size(s) __attribute__((alloc_size(s))) + #define mi_attr_alloc_size2(s1,s2) __attribute__((alloc_size(s1,s2))) + #define mi_attr_alloc_align(p) + #else + #define mi_attr_alloc_size(s) __attribute__((alloc_size(s))) + #define mi_attr_alloc_size2(s1,s2) __attribute__((alloc_size(s1,s2))) + #define mi_attr_alloc_align(p) __attribute__((alloc_align(p))) + #endif +#else + #define mi_cdecl + #define mi_decl_export + #define mi_decl_restrict + #define mi_attr_malloc + #define mi_attr_alloc_size(s) + #define mi_attr_alloc_size2(s1,s2) + #define mi_attr_alloc_align(p) +#endif + +// ------------------------------------------------------ +// Includes +// ------------------------------------------------------ + +#include // size_t +#include // bool +#include // INTPTR_MAX + +#ifdef __cplusplus +extern "C" { +#endif + +// ------------------------------------------------------ +// Standard malloc interface +// ------------------------------------------------------ + +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_malloc(size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_calloc(size_t count, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(1,2); +mi_decl_nodiscard mi_decl_export void* mi_realloc(void* p, size_t newsize) mi_attr_noexcept mi_attr_alloc_size(2); +mi_decl_export void* mi_expand(void* p, size_t newsize) mi_attr_noexcept mi_attr_alloc_size(2); + +mi_decl_export void mi_free(void* p) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export mi_decl_restrict char* mi_strdup(const char* s) mi_attr_noexcept mi_attr_malloc; +mi_decl_nodiscard mi_decl_export mi_decl_restrict char* mi_strndup(const char* s, size_t n) mi_attr_noexcept mi_attr_malloc; +mi_decl_nodiscard mi_decl_export mi_decl_restrict char* mi_realpath(const char* fname, char* resolved_name) mi_attr_noexcept mi_attr_malloc; + +// ------------------------------------------------------ +// Extended functionality +// ------------------------------------------------------ +#define MI_SMALL_WSIZE_MAX (128) +#define MI_SMALL_SIZE_MAX (MI_SMALL_WSIZE_MAX*sizeof(void*)) + +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_malloc_small(size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_zalloc_small(size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_zalloc(size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); + +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_mallocn(size_t count, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(1,2); +mi_decl_nodiscard mi_decl_export void* mi_reallocn(void* p, size_t count, size_t size) mi_attr_noexcept mi_attr_alloc_size2(2,3); +mi_decl_nodiscard mi_decl_export void* mi_reallocf(void* p, size_t newsize) mi_attr_noexcept mi_attr_alloc_size(2); + +mi_decl_nodiscard mi_decl_export size_t mi_usable_size(const void* p) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export size_t mi_good_size(size_t size) mi_attr_noexcept; + + +// ------------------------------------------------------ +// Internals +// ------------------------------------------------------ + +typedef void (mi_cdecl mi_deferred_free_fun)(bool force, unsigned long long heartbeat, void* arg); +mi_decl_export void mi_register_deferred_free(mi_deferred_free_fun* deferred_free, void* arg) mi_attr_noexcept; + +typedef void (mi_cdecl mi_output_fun)(const char* msg, void* arg); +mi_decl_export void mi_register_output(mi_output_fun* out, void* arg) mi_attr_noexcept; + +typedef void (mi_cdecl mi_error_fun)(int err, void* arg); +mi_decl_export void mi_register_error(mi_error_fun* fun, void* arg); + +mi_decl_export void mi_collect(bool force) mi_attr_noexcept; +mi_decl_export int mi_version(void) mi_attr_noexcept; +mi_decl_export void mi_stats_reset(void) mi_attr_noexcept; +mi_decl_export void mi_stats_merge(void) mi_attr_noexcept; +mi_decl_export void mi_stats_print(void* out) mi_attr_noexcept; // backward compatibility: `out` is ignored and should be NULL +mi_decl_export void mi_stats_print_out(mi_output_fun* out, void* arg) mi_attr_noexcept; + +mi_decl_export void mi_process_init(void) mi_attr_noexcept; +mi_decl_export void mi_thread_init(void) mi_attr_noexcept; +mi_decl_export void mi_thread_done(void) mi_attr_noexcept; +mi_decl_export void mi_thread_stats_print_out(mi_output_fun* out, void* arg) mi_attr_noexcept; + +mi_decl_export void mi_process_info(size_t* elapsed_msecs, size_t* user_msecs, size_t* system_msecs, + size_t* current_rss, size_t* peak_rss, + size_t* current_commit, size_t* peak_commit, size_t* page_faults) mi_attr_noexcept; + +// ------------------------------------------------------------------------------------- +// Aligned allocation +// Note that `alignment` always follows `size` for consistency with unaligned +// allocation, but unfortunately this differs from `posix_memalign` and `aligned_alloc`. +// ------------------------------------------------------------------------------------- + +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_malloc_aligned(size_t size, size_t alignment) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1) mi_attr_alloc_align(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_malloc_aligned_at(size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_zalloc_aligned(size_t size, size_t alignment) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1) mi_attr_alloc_align(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_zalloc_aligned_at(size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_calloc_aligned(size_t count, size_t size, size_t alignment) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(1,2) mi_attr_alloc_align(3); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_calloc_aligned_at(size_t count, size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(1,2); +mi_decl_nodiscard mi_decl_export void* mi_realloc_aligned(void* p, size_t newsize, size_t alignment) mi_attr_noexcept mi_attr_alloc_size(2) mi_attr_alloc_align(3); +mi_decl_nodiscard mi_decl_export void* mi_realloc_aligned_at(void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_alloc_size(2); + + +// ------------------------------------------------------------------------------------- +// Heaps: first-class, but can only allocate from the same thread that created it. +// ------------------------------------------------------------------------------------- + +struct mi_heap_s; +typedef struct mi_heap_s mi_heap_t; + +mi_decl_nodiscard mi_decl_export mi_heap_t* mi_heap_new(void); +mi_decl_export void mi_heap_delete(mi_heap_t* heap); +mi_decl_export void mi_heap_destroy(mi_heap_t* heap); +mi_decl_export mi_heap_t* mi_heap_set_default(mi_heap_t* heap); +mi_decl_export mi_heap_t* mi_heap_get_default(void); +mi_decl_export mi_heap_t* mi_heap_get_backing(void); +mi_decl_export void mi_heap_collect(mi_heap_t* heap, bool force) mi_attr_noexcept; + +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_malloc(mi_heap_t* heap, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_zalloc(mi_heap_t* heap, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_calloc(mi_heap_t* heap, size_t count, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(2, 3); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_mallocn(mi_heap_t* heap, size_t count, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(2, 3); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_malloc_small(mi_heap_t* heap, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2); + +mi_decl_nodiscard mi_decl_export void* mi_heap_realloc(mi_heap_t* heap, void* p, size_t newsize) mi_attr_noexcept mi_attr_alloc_size(3); +mi_decl_nodiscard mi_decl_export void* mi_heap_reallocn(mi_heap_t* heap, void* p, size_t count, size_t size) mi_attr_noexcept mi_attr_alloc_size2(3,4); +mi_decl_nodiscard mi_decl_export void* mi_heap_reallocf(mi_heap_t* heap, void* p, size_t newsize) mi_attr_noexcept mi_attr_alloc_size(3); + +mi_decl_nodiscard mi_decl_export mi_decl_restrict char* mi_heap_strdup(mi_heap_t* heap, const char* s) mi_attr_noexcept mi_attr_malloc; +mi_decl_nodiscard mi_decl_export mi_decl_restrict char* mi_heap_strndup(mi_heap_t* heap, const char* s, size_t n) mi_attr_noexcept mi_attr_malloc; +mi_decl_nodiscard mi_decl_export mi_decl_restrict char* mi_heap_realpath(mi_heap_t* heap, const char* fname, char* resolved_name) mi_attr_noexcept mi_attr_malloc; + +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_malloc_aligned(mi_heap_t* heap, size_t size, size_t alignment) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2) mi_attr_alloc_align(3); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_malloc_aligned_at(mi_heap_t* heap, size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_zalloc_aligned(mi_heap_t* heap, size_t size, size_t alignment) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2) mi_attr_alloc_align(3); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_zalloc_aligned_at(mi_heap_t* heap, size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_calloc_aligned(mi_heap_t* heap, size_t count, size_t size, size_t alignment) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(2, 3) mi_attr_alloc_align(4); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_calloc_aligned_at(mi_heap_t* heap, size_t count, size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size2(2, 3); +mi_decl_nodiscard mi_decl_export void* mi_heap_realloc_aligned(mi_heap_t* heap, void* p, size_t newsize, size_t alignment) mi_attr_noexcept mi_attr_alloc_size(3) mi_attr_alloc_align(4); +mi_decl_nodiscard mi_decl_export void* mi_heap_realloc_aligned_at(mi_heap_t* heap, void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_alloc_size(3); + + +// -------------------------------------------------------------------------------- +// Zero initialized re-allocation. +// Only valid on memory that was originally allocated with zero initialization too. +// e.g. `mi_calloc`, `mi_zalloc`, `mi_zalloc_aligned` etc. +// see +// -------------------------------------------------------------------------------- + +mi_decl_nodiscard mi_decl_export void* mi_rezalloc(void* p, size_t newsize) mi_attr_noexcept mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export void* mi_recalloc(void* p, size_t newcount, size_t size) mi_attr_noexcept mi_attr_alloc_size2(2,3); + +mi_decl_nodiscard mi_decl_export void* mi_rezalloc_aligned(void* p, size_t newsize, size_t alignment) mi_attr_noexcept mi_attr_alloc_size(2) mi_attr_alloc_align(3); +mi_decl_nodiscard mi_decl_export void* mi_rezalloc_aligned_at(void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export void* mi_recalloc_aligned(void* p, size_t newcount, size_t size, size_t alignment) mi_attr_noexcept mi_attr_alloc_size2(2,3) mi_attr_alloc_align(4); +mi_decl_nodiscard mi_decl_export void* mi_recalloc_aligned_at(void* p, size_t newcount, size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_alloc_size2(2,3); + +mi_decl_nodiscard mi_decl_export void* mi_heap_rezalloc(mi_heap_t* heap, void* p, size_t newsize) mi_attr_noexcept mi_attr_alloc_size(3); +mi_decl_nodiscard mi_decl_export void* mi_heap_recalloc(mi_heap_t* heap, void* p, size_t newcount, size_t size) mi_attr_noexcept mi_attr_alloc_size2(3,4); + +mi_decl_nodiscard mi_decl_export void* mi_heap_rezalloc_aligned(mi_heap_t* heap, void* p, size_t newsize, size_t alignment) mi_attr_noexcept mi_attr_alloc_size(3) mi_attr_alloc_align(4); +mi_decl_nodiscard mi_decl_export void* mi_heap_rezalloc_aligned_at(mi_heap_t* heap, void* p, size_t newsize, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_alloc_size(3); +mi_decl_nodiscard mi_decl_export void* mi_heap_recalloc_aligned(mi_heap_t* heap, void* p, size_t newcount, size_t size, size_t alignment) mi_attr_noexcept mi_attr_alloc_size2(3,4) mi_attr_alloc_align(5); +mi_decl_nodiscard mi_decl_export void* mi_heap_recalloc_aligned_at(mi_heap_t* heap, void* p, size_t newcount, size_t size, size_t alignment, size_t offset) mi_attr_noexcept mi_attr_alloc_size2(3,4); + + +// ------------------------------------------------------ +// Analysis +// ------------------------------------------------------ + +mi_decl_export bool mi_heap_contains_block(mi_heap_t* heap, const void* p); +mi_decl_export bool mi_heap_check_owned(mi_heap_t* heap, const void* p); +mi_decl_export bool mi_check_owned(const void* p); + +// An area of heap space contains blocks of a single size. +typedef struct mi_heap_area_s { + void* blocks; // start of the area containing heap blocks + size_t reserved; // bytes reserved for this area (virtual) + size_t committed; // current available bytes for this area + size_t used; // number of allocated blocks + size_t block_size; // size in bytes of each block + size_t full_block_size; // size in bytes of a full block including padding and metadata. +} mi_heap_area_t; + +typedef bool (mi_cdecl mi_block_visit_fun)(const mi_heap_t* heap, const mi_heap_area_t* area, void* block, size_t block_size, void* arg); + +mi_decl_export bool mi_heap_visit_blocks(const mi_heap_t* heap, bool visit_all_blocks, mi_block_visit_fun* visitor, void* arg); + +// Experimental +mi_decl_nodiscard mi_decl_export bool mi_is_in_heap_region(const void* p) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export bool mi_is_redirected(void) mi_attr_noexcept; + +mi_decl_export int mi_reserve_huge_os_pages_interleave(size_t pages, size_t numa_nodes, size_t timeout_msecs) mi_attr_noexcept; +mi_decl_export int mi_reserve_huge_os_pages_at(size_t pages, int numa_node, size_t timeout_msecs) mi_attr_noexcept; + +mi_decl_export int mi_reserve_os_memory(size_t size, bool commit, bool allow_large) mi_attr_noexcept; +mi_decl_export bool mi_manage_os_memory(void* start, size_t size, bool is_committed, bool is_large, bool is_zero, int numa_node) mi_attr_noexcept; + +mi_decl_export void mi_debug_show_arenas(void) mi_attr_noexcept; + +// Experimental: heaps associated with specific memory arena's +typedef int mi_arena_id_t; +mi_decl_export void* mi_arena_area(mi_arena_id_t arena_id, size_t* size); +mi_decl_export int mi_reserve_huge_os_pages_at_ex(size_t pages, int numa_node, size_t timeout_msecs, bool exclusive, mi_arena_id_t* arena_id) mi_attr_noexcept; +mi_decl_export int mi_reserve_os_memory_ex(size_t size, bool commit, bool allow_large, bool exclusive, mi_arena_id_t* arena_id) mi_attr_noexcept; +mi_decl_export bool mi_manage_os_memory_ex(void* start, size_t size, bool is_committed, bool is_large, bool is_zero, int numa_node, bool exclusive, mi_arena_id_t* arena_id) mi_attr_noexcept; + +#if MI_MALLOC_VERSION >= 182 +// Create a heap that only allocates in the specified arena +mi_decl_nodiscard mi_decl_export mi_heap_t* mi_heap_new_in_arena(mi_arena_id_t arena_id); +#endif + +// deprecated +mi_decl_export int mi_reserve_huge_os_pages(size_t pages, double max_secs, size_t* pages_reserved) mi_attr_noexcept; + + +// ------------------------------------------------------ +// Convenience +// ------------------------------------------------------ + +#define mi_malloc_tp(tp) ((tp*)mi_malloc(sizeof(tp))) +#define mi_zalloc_tp(tp) ((tp*)mi_zalloc(sizeof(tp))) +#define mi_calloc_tp(tp,n) ((tp*)mi_calloc(n,sizeof(tp))) +#define mi_mallocn_tp(tp,n) ((tp*)mi_mallocn(n,sizeof(tp))) +#define mi_reallocn_tp(p,tp,n) ((tp*)mi_reallocn(p,n,sizeof(tp))) +#define mi_recalloc_tp(p,tp,n) ((tp*)mi_recalloc(p,n,sizeof(tp))) + +#define mi_heap_malloc_tp(hp,tp) ((tp*)mi_heap_malloc(hp,sizeof(tp))) +#define mi_heap_zalloc_tp(hp,tp) ((tp*)mi_heap_zalloc(hp,sizeof(tp))) +#define mi_heap_calloc_tp(hp,tp,n) ((tp*)mi_heap_calloc(hp,n,sizeof(tp))) +#define mi_heap_mallocn_tp(hp,tp,n) ((tp*)mi_heap_mallocn(hp,n,sizeof(tp))) +#define mi_heap_reallocn_tp(hp,p,tp,n) ((tp*)mi_heap_reallocn(hp,p,n,sizeof(tp))) +#define mi_heap_recalloc_tp(hp,p,tp,n) ((tp*)mi_heap_recalloc(hp,p,n,sizeof(tp))) + + +// ------------------------------------------------------ +// Options +// ------------------------------------------------------ + +typedef enum mi_option_e { + // stable options + mi_option_show_errors, // print error messages + mi_option_show_stats, // print statistics on termination + mi_option_verbose, // print verbose messages + // the following options are experimental (see src/options.h) + mi_option_eager_commit, // eager commit segments? (after `eager_commit_delay` segments) (=1) + mi_option_arena_eager_commit, // eager commit arenas? Use 2 to enable just on overcommit systems (=2) + mi_option_purge_decommits, // should a memory purge decommit (or only reset) (=1) + mi_option_allow_large_os_pages, // allow large (2MiB) OS pages, implies eager commit + mi_option_reserve_huge_os_pages, // reserve N huge OS pages (1GiB/page) at startup + mi_option_reserve_huge_os_pages_at, // reserve huge OS pages at a specific NUMA node + mi_option_reserve_os_memory, // reserve specified amount of OS memory in an arena at startup + mi_option_deprecated_segment_cache, + mi_option_deprecated_page_reset, + mi_option_abandoned_page_purge, // immediately purge delayed purges on thread termination + mi_option_deprecated_segment_reset, + mi_option_eager_commit_delay, + mi_option_purge_delay, // memory purging is delayed by N milli seconds; use 0 for immediate purging or -1 for no purging at all. + mi_option_use_numa_nodes, // 0 = use all available numa nodes, otherwise use at most N nodes. + mi_option_limit_os_alloc, // 1 = do not use OS memory for allocation (but only programmatically reserved arenas) + mi_option_os_tag, // tag used for OS logging (macOS only for now) + mi_option_max_errors, // issue at most N error messages + mi_option_max_warnings, // issue at most N warning messages + mi_option_max_segment_reclaim, + mi_option_destroy_on_exit, // if set, release all memory on exit; sometimes used for dynamic unloading but can be unsafe. + mi_option_arena_reserve, // initial memory size in KiB for arena reservation (1GiB on 64-bit) + mi_option_arena_purge_mult, + mi_option_purge_extend_delay, + _mi_option_last, + // legacy option names + mi_option_large_os_pages = mi_option_allow_large_os_pages, + mi_option_eager_region_commit = mi_option_arena_eager_commit, + mi_option_reset_decommits = mi_option_purge_decommits, + mi_option_reset_delay = mi_option_purge_delay, + mi_option_abandoned_page_reset = mi_option_abandoned_page_purge +} mi_option_t; + + +mi_decl_nodiscard mi_decl_export bool mi_option_is_enabled(mi_option_t option); +mi_decl_export void mi_option_enable(mi_option_t option); +mi_decl_export void mi_option_disable(mi_option_t option); +mi_decl_export void mi_option_set_enabled(mi_option_t option, bool enable); +mi_decl_export void mi_option_set_enabled_default(mi_option_t option, bool enable); + +mi_decl_nodiscard mi_decl_export long mi_option_get(mi_option_t option); +mi_decl_nodiscard mi_decl_export long mi_option_get_clamp(mi_option_t option, long min, long max); +mi_decl_nodiscard mi_decl_export size_t mi_option_get_size(mi_option_t option); +mi_decl_export void mi_option_set(mi_option_t option, long value); +mi_decl_export void mi_option_set_default(mi_option_t option, long value); + + +// ------------------------------------------------------------------------------------------------------- +// "mi" prefixed implementations of various posix, Unix, Windows, and C++ allocation functions. +// (This can be convenient when providing overrides of these functions as done in `mimalloc-override.h`.) +// note: we use `mi_cfree` as "checked free" and it checks if the pointer is in our heap before free-ing. +// ------------------------------------------------------------------------------------------------------- + +mi_decl_export void mi_cfree(void* p) mi_attr_noexcept; +mi_decl_export void* mi__expand(void* p, size_t newsize) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export size_t mi_malloc_size(const void* p) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export size_t mi_malloc_good_size(size_t size) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export size_t mi_malloc_usable_size(const void *p) mi_attr_noexcept; + +mi_decl_export int mi_posix_memalign(void** p, size_t alignment, size_t size) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_memalign(size_t alignment, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2) mi_attr_alloc_align(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_valloc(size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_pvalloc(size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_aligned_alloc(size_t alignment, size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(2) mi_attr_alloc_align(1); + +mi_decl_nodiscard mi_decl_export void* mi_reallocarray(void* p, size_t count, size_t size) mi_attr_noexcept mi_attr_alloc_size2(2,3); +mi_decl_nodiscard mi_decl_export int mi_reallocarr(void* p, size_t count, size_t size) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export void* mi_aligned_recalloc(void* p, size_t newcount, size_t size, size_t alignment) mi_attr_noexcept; +mi_decl_nodiscard mi_decl_export void* mi_aligned_offset_recalloc(void* p, size_t newcount, size_t size, size_t alignment, size_t offset) mi_attr_noexcept; + +mi_decl_nodiscard mi_decl_export mi_decl_restrict unsigned short* mi_wcsdup(const unsigned short* s) mi_attr_noexcept mi_attr_malloc; +mi_decl_nodiscard mi_decl_export mi_decl_restrict unsigned char* mi_mbsdup(const unsigned char* s) mi_attr_noexcept mi_attr_malloc; +mi_decl_export int mi_dupenv_s(char** buf, size_t* size, const char* name) mi_attr_noexcept; +mi_decl_export int mi_wdupenv_s(unsigned short** buf, size_t* size, const unsigned short* name) mi_attr_noexcept; + +mi_decl_export void mi_free_size(void* p, size_t size) mi_attr_noexcept; +mi_decl_export void mi_free_size_aligned(void* p, size_t size, size_t alignment) mi_attr_noexcept; +mi_decl_export void mi_free_aligned(void* p, size_t alignment) mi_attr_noexcept; + +// The `mi_new` wrappers implement C++ semantics on out-of-memory instead of directly returning `NULL`. +// (and call `std::get_new_handler` and potentially raise a `std::bad_alloc` exception). +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_new(size_t size) mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_new_aligned(size_t size, size_t alignment) mi_attr_malloc mi_attr_alloc_size(1) mi_attr_alloc_align(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_new_nothrow(size_t size) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_new_aligned_nothrow(size_t size, size_t alignment) mi_attr_noexcept mi_attr_malloc mi_attr_alloc_size(1) mi_attr_alloc_align(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_new_n(size_t count, size_t size) mi_attr_malloc mi_attr_alloc_size2(1, 2); +mi_decl_nodiscard mi_decl_export void* mi_new_realloc(void* p, size_t newsize) mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export void* mi_new_reallocn(void* p, size_t newcount, size_t size) mi_attr_alloc_size2(2, 3); + +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_alloc_new(mi_heap_t* heap, size_t size) mi_attr_malloc mi_attr_alloc_size(2); +mi_decl_nodiscard mi_decl_export mi_decl_restrict void* mi_heap_alloc_new_n(mi_heap_t* heap, size_t count, size_t size) mi_attr_malloc mi_attr_alloc_size2(2, 3); + +#ifdef __cplusplus +} +#endif + +// --------------------------------------------------------------------------------------------- +// Implement the C++ std::allocator interface for use in STL containers. +// (note: see `mimalloc-new-delete.h` for overriding the new/delete operators globally) +// --------------------------------------------------------------------------------------------- +#ifdef __cplusplus + +#include // std::size_t +#include // PTRDIFF_MAX +#if (__cplusplus >= 201103L) || (_MSC_VER > 1900) // C++11 +#include // std::true_type +#include // std::forward +#endif + +template struct _mi_stl_allocator_common { + typedef T value_type; + typedef std::size_t size_type; + typedef std::ptrdiff_t difference_type; + typedef value_type& reference; + typedef value_type const& const_reference; + typedef value_type* pointer; + typedef value_type const* const_pointer; + + #if ((__cplusplus >= 201103L) || (_MSC_VER > 1900)) // C++11 + using propagate_on_container_copy_assignment = std::true_type; + using propagate_on_container_move_assignment = std::true_type; + using propagate_on_container_swap = std::true_type; + template void construct(U* p, Args&& ...args) { ::new(p) U(std::forward(args)...); } + template void destroy(U* p) mi_attr_noexcept { p->~U(); } + #else + void construct(pointer p, value_type const& val) { ::new(p) value_type(val); } + void destroy(pointer p) { p->~value_type(); } + #endif + + size_type max_size() const mi_attr_noexcept { return (PTRDIFF_MAX/sizeof(value_type)); } + pointer address(reference x) const { return &x; } + const_pointer address(const_reference x) const { return &x; } +}; + +template struct mi_stl_allocator : public _mi_stl_allocator_common { + using typename _mi_stl_allocator_common::size_type; + using typename _mi_stl_allocator_common::value_type; + using typename _mi_stl_allocator_common::pointer; + template struct rebind { typedef mi_stl_allocator other; }; + + mi_stl_allocator() mi_attr_noexcept = default; + mi_stl_allocator(const mi_stl_allocator&) mi_attr_noexcept = default; + template mi_stl_allocator(const mi_stl_allocator&) mi_attr_noexcept { } + mi_stl_allocator select_on_container_copy_construction() const { return *this; } + void deallocate(T* p, size_type) { mi_free(p); } + + #if (__cplusplus >= 201703L) // C++17 + mi_decl_nodiscard T* allocate(size_type count) { return static_cast(mi_new_n(count, sizeof(T))); } + mi_decl_nodiscard T* allocate(size_type count, const void*) { return allocate(count); } + #else + mi_decl_nodiscard pointer allocate(size_type count, const void* = 0) { return static_cast(mi_new_n(count, sizeof(value_type))); } + #endif + + #if ((__cplusplus >= 201103L) || (_MSC_VER > 1900)) // C++11 + using is_always_equal = std::true_type; + #endif +}; + +template bool operator==(const mi_stl_allocator& , const mi_stl_allocator& ) mi_attr_noexcept { return true; } +template bool operator!=(const mi_stl_allocator& , const mi_stl_allocator& ) mi_attr_noexcept { return false; } + + +#if (__cplusplus >= 201103L) || (_MSC_VER >= 1900) // C++11 +#define MI_HAS_HEAP_STL_ALLOCATOR 1 + +#include // std::shared_ptr + +// Common base class for STL allocators in a specific heap +template struct _mi_heap_stl_allocator_common : public _mi_stl_allocator_common { + using typename _mi_stl_allocator_common::size_type; + using typename _mi_stl_allocator_common::value_type; + using typename _mi_stl_allocator_common::pointer; + + _mi_heap_stl_allocator_common(mi_heap_t* hp) : heap(hp) { } /* will not delete nor destroy the passed in heap */ + + #if (__cplusplus >= 201703L) // C++17 + mi_decl_nodiscard T* allocate(size_type count) { return static_cast(mi_heap_alloc_new_n(this->heap.get(), count, sizeof(T))); } + mi_decl_nodiscard T* allocate(size_type count, const void*) { return allocate(count); } + #else + mi_decl_nodiscard pointer allocate(size_type count, const void* = 0) { return static_cast(mi_heap_alloc_new_n(this->heap.get(), count, sizeof(value_type))); } + #endif + + #if ((__cplusplus >= 201103L) || (_MSC_VER > 1900)) // C++11 + using is_always_equal = std::false_type; + #endif + + void collect(bool force) { mi_heap_collect(this->heap.get(), force); } + template bool is_equal(const _mi_heap_stl_allocator_common& x) const { return (this->heap == x.heap); } + +protected: + std::shared_ptr heap; + template friend struct _mi_heap_stl_allocator_common; + + _mi_heap_stl_allocator_common() { + mi_heap_t* hp = mi_heap_new(); + this->heap.reset(hp, (_mi_destroy ? &heap_destroy : &heap_delete)); /* calls heap_delete/destroy when the refcount drops to zero */ + } + _mi_heap_stl_allocator_common(const _mi_heap_stl_allocator_common& x) mi_attr_noexcept : heap(x.heap) { } + template _mi_heap_stl_allocator_common(const _mi_heap_stl_allocator_common& x) mi_attr_noexcept : heap(x.heap) { } + +private: + static void heap_delete(mi_heap_t* hp) { if (hp != NULL) { mi_heap_delete(hp); } } + static void heap_destroy(mi_heap_t* hp) { if (hp != NULL) { mi_heap_destroy(hp); } } +}; + +// STL allocator allocation in a specific heap +template struct mi_heap_stl_allocator : public _mi_heap_stl_allocator_common { + using typename _mi_heap_stl_allocator_common::size_type; + mi_heap_stl_allocator() : _mi_heap_stl_allocator_common() { } // creates fresh heap that is deleted when the destructor is called + mi_heap_stl_allocator(mi_heap_t* hp) : _mi_heap_stl_allocator_common(hp) { } // no delete nor destroy on the passed in heap + template mi_heap_stl_allocator(const mi_heap_stl_allocator& x) mi_attr_noexcept : _mi_heap_stl_allocator_common(x) { } + + mi_heap_stl_allocator select_on_container_copy_construction() const { return *this; } + void deallocate(T* p, size_type) { mi_free(p); } + template struct rebind { typedef mi_heap_stl_allocator other; }; +}; + +template bool operator==(const mi_heap_stl_allocator& x, const mi_heap_stl_allocator& y) mi_attr_noexcept { return (x.is_equal(y)); } +template bool operator!=(const mi_heap_stl_allocator& x, const mi_heap_stl_allocator& y) mi_attr_noexcept { return (!x.is_equal(y)); } + + +// STL allocator allocation in a specific heap, where `free` does nothing and +// the heap is destroyed in one go on destruction -- use with care! +template struct mi_heap_destroy_stl_allocator : public _mi_heap_stl_allocator_common { + using typename _mi_heap_stl_allocator_common::size_type; + mi_heap_destroy_stl_allocator() : _mi_heap_stl_allocator_common() { } // creates fresh heap that is destroyed when the destructor is called + mi_heap_destroy_stl_allocator(mi_heap_t* hp) : _mi_heap_stl_allocator_common(hp) { } // no delete nor destroy on the passed in heap + template mi_heap_destroy_stl_allocator(const mi_heap_destroy_stl_allocator& x) mi_attr_noexcept : _mi_heap_stl_allocator_common(x) { } + + mi_heap_destroy_stl_allocator select_on_container_copy_construction() const { return *this; } + void deallocate(T*, size_type) { /* do nothing as we destroy the heap on destruct. */ } + template struct rebind { typedef mi_heap_destroy_stl_allocator other; }; +}; + +template bool operator==(const mi_heap_destroy_stl_allocator& x, const mi_heap_destroy_stl_allocator& y) mi_attr_noexcept { return (x.is_equal(y)); } +template bool operator!=(const mi_heap_destroy_stl_allocator& x, const mi_heap_destroy_stl_allocator& y) mi_attr_noexcept { return (!x.is_equal(y)); } + +#endif // C++11 + +#endif // __cplusplus + +#endif diff --git a/compat/mimalloc/mimalloc/atomic.h b/compat/mimalloc/mimalloc/atomic.h new file mode 100644 index 00000000000000..c6b8146ffdb049 --- /dev/null +++ b/compat/mimalloc/mimalloc/atomic.h @@ -0,0 +1,385 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023 Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#pragma once +#ifndef MIMALLOC_ATOMIC_H +#define MIMALLOC_ATOMIC_H + +// -------------------------------------------------------------------------------------------- +// Atomics +// We need to be portable between C, C++, and MSVC. +// We base the primitives on the C/C++ atomics and create a mimimal wrapper for MSVC in C compilation mode. +// This is why we try to use only `uintptr_t` and `*` as atomic types. +// To gain better insight in the range of used atomics, we use explicitly named memory order operations +// instead of passing the memory order as a parameter. +// ----------------------------------------------------------------------------------------------- + +#if defined(__cplusplus) +// Use C++ atomics +#include +#define _Atomic(tp) std::atomic +#define mi_atomic(name) std::atomic_##name +#define mi_memory_order(name) std::memory_order_##name +#if !defined(ATOMIC_VAR_INIT) || (__cplusplus >= 202002L) // c++20, see issue #571 + #define MI_ATOMIC_VAR_INIT(x) x +#else + #define MI_ATOMIC_VAR_INIT(x) ATOMIC_VAR_INIT(x) +#endif +#elif defined(_MSC_VER) +// Use MSVC C wrapper for C11 atomics +#define _Atomic(tp) tp +#define MI_ATOMIC_VAR_INIT(x) x +#define mi_atomic(name) mi_atomic_##name +#define mi_memory_order(name) mi_memory_order_##name +#else +// Use C11 atomics +#include +#define mi_atomic(name) atomic_##name +#define mi_memory_order(name) memory_order_##name +#if !defined(ATOMIC_VAR_INIT) || (__STDC_VERSION__ >= 201710L) // c17, see issue #735 + #define MI_ATOMIC_VAR_INIT(x) x +#else + #define MI_ATOMIC_VAR_INIT(x) ATOMIC_VAR_INIT(x) +#endif +#endif + +// Various defines for all used memory orders in mimalloc +#define mi_atomic_cas_weak(p,expected,desired,mem_success,mem_fail) \ + mi_atomic(compare_exchange_weak_explicit)(p,expected,desired,mem_success,mem_fail) + +#define mi_atomic_cas_strong(p,expected,desired,mem_success,mem_fail) \ + mi_atomic(compare_exchange_strong_explicit)(p,expected,desired,mem_success,mem_fail) + +#define mi_atomic_load_acquire(p) mi_atomic(load_explicit)(p,mi_memory_order(acquire)) +#define mi_atomic_load_relaxed(p) mi_atomic(load_explicit)(p,mi_memory_order(relaxed)) +#define mi_atomic_store_release(p,x) mi_atomic(store_explicit)(p,x,mi_memory_order(release)) +#define mi_atomic_store_relaxed(p,x) mi_atomic(store_explicit)(p,x,mi_memory_order(relaxed)) +#define mi_atomic_exchange_release(p,x) mi_atomic(exchange_explicit)(p,x,mi_memory_order(release)) +#define mi_atomic_exchange_acq_rel(p,x) mi_atomic(exchange_explicit)(p,x,mi_memory_order(acq_rel)) +#define mi_atomic_cas_weak_release(p,exp,des) mi_atomic_cas_weak(p,exp,des,mi_memory_order(release),mi_memory_order(relaxed)) +#define mi_atomic_cas_weak_acq_rel(p,exp,des) mi_atomic_cas_weak(p,exp,des,mi_memory_order(acq_rel),mi_memory_order(acquire)) +#define mi_atomic_cas_strong_release(p,exp,des) mi_atomic_cas_strong(p,exp,des,mi_memory_order(release),mi_memory_order(relaxed)) +#define mi_atomic_cas_strong_acq_rel(p,exp,des) mi_atomic_cas_strong(p,exp,des,mi_memory_order(acq_rel),mi_memory_order(acquire)) + +#define mi_atomic_add_relaxed(p,x) mi_atomic(fetch_add_explicit)(p,x,mi_memory_order(relaxed)) +#define mi_atomic_sub_relaxed(p,x) mi_atomic(fetch_sub_explicit)(p,x,mi_memory_order(relaxed)) +#define mi_atomic_add_acq_rel(p,x) mi_atomic(fetch_add_explicit)(p,x,mi_memory_order(acq_rel)) +#define mi_atomic_sub_acq_rel(p,x) mi_atomic(fetch_sub_explicit)(p,x,mi_memory_order(acq_rel)) +#define mi_atomic_and_acq_rel(p,x) mi_atomic(fetch_and_explicit)(p,x,mi_memory_order(acq_rel)) +#define mi_atomic_or_acq_rel(p,x) mi_atomic(fetch_or_explicit)(p,x,mi_memory_order(acq_rel)) + +#define mi_atomic_increment_relaxed(p) mi_atomic_add_relaxed(p,(uintptr_t)1) +#define mi_atomic_decrement_relaxed(p) mi_atomic_sub_relaxed(p,(uintptr_t)1) +#define mi_atomic_increment_acq_rel(p) mi_atomic_add_acq_rel(p,(uintptr_t)1) +#define mi_atomic_decrement_acq_rel(p) mi_atomic_sub_acq_rel(p,(uintptr_t)1) + +static inline void mi_atomic_yield(void); +static inline intptr_t mi_atomic_addi(_Atomic(intptr_t)*p, intptr_t add); +static inline intptr_t mi_atomic_subi(_Atomic(intptr_t)*p, intptr_t sub); + + +#if defined(__cplusplus) || !defined(_MSC_VER) + +// In C++/C11 atomics we have polymorphic atomics so can use the typed `ptr` variants (where `tp` is the type of atomic value) +// We use these macros so we can provide a typed wrapper in MSVC in C compilation mode as well +#define mi_atomic_load_ptr_acquire(tp,p) mi_atomic_load_acquire(p) +#define mi_atomic_load_ptr_relaxed(tp,p) mi_atomic_load_relaxed(p) + +// In C++ we need to add casts to help resolve templates if NULL is passed +#if defined(__cplusplus) +#define mi_atomic_store_ptr_release(tp,p,x) mi_atomic_store_release(p,(tp*)x) +#define mi_atomic_store_ptr_relaxed(tp,p,x) mi_atomic_store_relaxed(p,(tp*)x) +#define mi_atomic_cas_ptr_weak_release(tp,p,exp,des) mi_atomic_cas_weak_release(p,exp,(tp*)des) +#define mi_atomic_cas_ptr_weak_acq_rel(tp,p,exp,des) mi_atomic_cas_weak_acq_rel(p,exp,(tp*)des) +#define mi_atomic_cas_ptr_strong_release(tp,p,exp,des) mi_atomic_cas_strong_release(p,exp,(tp*)des) +#define mi_atomic_exchange_ptr_release(tp,p,x) mi_atomic_exchange_release(p,(tp*)x) +#define mi_atomic_exchange_ptr_acq_rel(tp,p,x) mi_atomic_exchange_acq_rel(p,(tp*)x) +#else +#define mi_atomic_store_ptr_release(tp,p,x) mi_atomic_store_release(p,x) +#define mi_atomic_store_ptr_relaxed(tp,p,x) mi_atomic_store_relaxed(p,x) +#define mi_atomic_cas_ptr_weak_release(tp,p,exp,des) mi_atomic_cas_weak_release(p,exp,des) +#define mi_atomic_cas_ptr_weak_acq_rel(tp,p,exp,des) mi_atomic_cas_weak_acq_rel(p,exp,des) +#define mi_atomic_cas_ptr_strong_release(tp,p,exp,des) mi_atomic_cas_strong_release(p,exp,des) +#define mi_atomic_exchange_ptr_release(tp,p,x) mi_atomic_exchange_release(p,x) +#define mi_atomic_exchange_ptr_acq_rel(tp,p,x) mi_atomic_exchange_acq_rel(p,x) +#endif + +// These are used by the statistics +static inline int64_t mi_atomic_addi64_relaxed(volatile int64_t* p, int64_t add) { + return mi_atomic(fetch_add_explicit)((_Atomic(int64_t)*)p, add, mi_memory_order(relaxed)); +} +static inline void mi_atomic_maxi64_relaxed(volatile int64_t* p, int64_t x) { + int64_t current = mi_atomic_load_relaxed((_Atomic(int64_t)*)p); + while (current < x && !mi_atomic_cas_weak_release((_Atomic(int64_t)*)p, ¤t, x)) { /* nothing */ }; +} + +// Used by timers +#define mi_atomic_loadi64_acquire(p) mi_atomic(load_explicit)(p,mi_memory_order(acquire)) +#define mi_atomic_loadi64_relaxed(p) mi_atomic(load_explicit)(p,mi_memory_order(relaxed)) +#define mi_atomic_storei64_release(p,x) mi_atomic(store_explicit)(p,x,mi_memory_order(release)) +#define mi_atomic_storei64_relaxed(p,x) mi_atomic(store_explicit)(p,x,mi_memory_order(relaxed)) + +#define mi_atomic_casi64_strong_acq_rel(p,e,d) mi_atomic_cas_strong_acq_rel(p,e,d) +#define mi_atomic_addi64_acq_rel(p,i) mi_atomic_add_acq_rel(p,i) + + +#elif defined(_MSC_VER) + +// MSVC C compilation wrapper that uses Interlocked operations to model C11 atomics. +#define WIN32_LEAN_AND_MEAN +#include +#include +#ifdef _WIN64 +typedef LONG64 msc_intptr_t; +#define MI_64(f) f##64 +#else +typedef LONG msc_intptr_t; +#define MI_64(f) f +#endif + +typedef enum mi_memory_order_e { + mi_memory_order_relaxed, + mi_memory_order_consume, + mi_memory_order_acquire, + mi_memory_order_release, + mi_memory_order_acq_rel, + mi_memory_order_seq_cst +} mi_memory_order; + +static inline uintptr_t mi_atomic_fetch_add_explicit(_Atomic(uintptr_t)*p, uintptr_t add, mi_memory_order mo) { + (void)(mo); + return (uintptr_t)MI_64(_InterlockedExchangeAdd)((volatile msc_intptr_t*)p, (msc_intptr_t)add); +} +static inline uintptr_t mi_atomic_fetch_sub_explicit(_Atomic(uintptr_t)*p, uintptr_t sub, mi_memory_order mo) { + (void)(mo); + return (uintptr_t)MI_64(_InterlockedExchangeAdd)((volatile msc_intptr_t*)p, -((msc_intptr_t)sub)); +} +static inline uintptr_t mi_atomic_fetch_and_explicit(_Atomic(uintptr_t)*p, uintptr_t x, mi_memory_order mo) { + (void)(mo); + return (uintptr_t)MI_64(_InterlockedAnd)((volatile msc_intptr_t*)p, (msc_intptr_t)x); +} +static inline uintptr_t mi_atomic_fetch_or_explicit(_Atomic(uintptr_t)*p, uintptr_t x, mi_memory_order mo) { + (void)(mo); + return (uintptr_t)MI_64(_InterlockedOr)((volatile msc_intptr_t*)p, (msc_intptr_t)x); +} +static inline bool mi_atomic_compare_exchange_strong_explicit(_Atomic(uintptr_t)*p, uintptr_t* expected, uintptr_t desired, mi_memory_order mo1, mi_memory_order mo2) { + (void)(mo1); (void)(mo2); + uintptr_t read = (uintptr_t)MI_64(_InterlockedCompareExchange)((volatile msc_intptr_t*)p, (msc_intptr_t)desired, (msc_intptr_t)(*expected)); + if (read == *expected) { + return true; + } + else { + *expected = read; + return false; + } +} +static inline bool mi_atomic_compare_exchange_weak_explicit(_Atomic(uintptr_t)*p, uintptr_t* expected, uintptr_t desired, mi_memory_order mo1, mi_memory_order mo2) { + return mi_atomic_compare_exchange_strong_explicit(p, expected, desired, mo1, mo2); +} +static inline uintptr_t mi_atomic_exchange_explicit(_Atomic(uintptr_t)*p, uintptr_t exchange, mi_memory_order mo) { + (void)(mo); + return (uintptr_t)MI_64(_InterlockedExchange)((volatile msc_intptr_t*)p, (msc_intptr_t)exchange); +} +static inline void mi_atomic_thread_fence(mi_memory_order mo) { + (void)(mo); + _Atomic(uintptr_t) x = 0; + mi_atomic_exchange_explicit(&x, 1, mo); +} +static inline uintptr_t mi_atomic_load_explicit(_Atomic(uintptr_t) const* p, mi_memory_order mo) { + (void)(mo); +#if defined(_M_IX86) || defined(_M_X64) + return *p; +#else + uintptr_t x = *p; + if (mo > mi_memory_order_relaxed) { + while (!mi_atomic_compare_exchange_weak_explicit(p, &x, x, mo, mi_memory_order_relaxed)) { /* nothing */ }; + } + return x; +#endif +} +static inline void mi_atomic_store_explicit(_Atomic(uintptr_t)*p, uintptr_t x, mi_memory_order mo) { + (void)(mo); +#if defined(_M_IX86) || defined(_M_X64) + *p = x; +#else + mi_atomic_exchange_explicit(p, x, mo); +#endif +} +static inline int64_t mi_atomic_loadi64_explicit(_Atomic(int64_t)*p, mi_memory_order mo) { + (void)(mo); +#if defined(_M_X64) + return *p; +#else + int64_t old = *p; + int64_t x = old; + while ((old = InterlockedCompareExchange64(p, x, old)) != x) { + x = old; + } + return x; +#endif +} +static inline void mi_atomic_storei64_explicit(_Atomic(int64_t)*p, int64_t x, mi_memory_order mo) { + (void)(mo); +#if defined(x_M_IX86) || defined(_M_X64) + *p = x; +#else + InterlockedExchange64(p, x); +#endif +} + +// These are used by the statistics +static inline int64_t mi_atomic_addi64_relaxed(volatile _Atomic(int64_t)*p, int64_t add) { +#ifdef _WIN64 + return (int64_t)mi_atomic_addi((int64_t*)p, add); +#else + int64_t current; + int64_t sum; + do { + current = *p; + sum = current + add; + } while (_InterlockedCompareExchange64(p, sum, current) != current); + return current; +#endif +} +static inline void mi_atomic_maxi64_relaxed(volatile _Atomic(int64_t)*p, int64_t x) { + int64_t current; + do { + current = *p; + } while (current < x && _InterlockedCompareExchange64(p, x, current) != current); +} + +static inline void mi_atomic_addi64_acq_rel(volatile _Atomic(int64_t*)p, int64_t i) { + mi_atomic_addi64_relaxed(p, i); +} + +static inline bool mi_atomic_casi64_strong_acq_rel(volatile _Atomic(int64_t*)p, int64_t* exp, int64_t des) { + int64_t read = _InterlockedCompareExchange64(p, des, *exp); + if (read == *exp) { + return true; + } + else { + *exp = read; + return false; + } +} + +// The pointer macros cast to `uintptr_t`. +#define mi_atomic_load_ptr_acquire(tp,p) (tp*)mi_atomic_load_acquire((_Atomic(uintptr_t)*)(p)) +#define mi_atomic_load_ptr_relaxed(tp,p) (tp*)mi_atomic_load_relaxed((_Atomic(uintptr_t)*)(p)) +#define mi_atomic_store_ptr_release(tp,p,x) mi_atomic_store_release((_Atomic(uintptr_t)*)(p),(uintptr_t)(x)) +#define mi_atomic_store_ptr_relaxed(tp,p,x) mi_atomic_store_relaxed((_Atomic(uintptr_t)*)(p),(uintptr_t)(x)) +#define mi_atomic_cas_ptr_weak_release(tp,p,exp,des) mi_atomic_cas_weak_release((_Atomic(uintptr_t)*)(p),(uintptr_t*)exp,(uintptr_t)des) +#define mi_atomic_cas_ptr_weak_acq_rel(tp,p,exp,des) mi_atomic_cas_weak_acq_rel((_Atomic(uintptr_t)*)(p),(uintptr_t*)exp,(uintptr_t)des) +#define mi_atomic_cas_ptr_strong_release(tp,p,exp,des) mi_atomic_cas_strong_release((_Atomic(uintptr_t)*)(p),(uintptr_t*)exp,(uintptr_t)des) +#define mi_atomic_exchange_ptr_release(tp,p,x) (tp*)mi_atomic_exchange_release((_Atomic(uintptr_t)*)(p),(uintptr_t)x) +#define mi_atomic_exchange_ptr_acq_rel(tp,p,x) (tp*)mi_atomic_exchange_acq_rel((_Atomic(uintptr_t)*)(p),(uintptr_t)x) + +#define mi_atomic_loadi64_acquire(p) mi_atomic(loadi64_explicit)(p,mi_memory_order(acquire)) +#define mi_atomic_loadi64_relaxed(p) mi_atomic(loadi64_explicit)(p,mi_memory_order(relaxed)) +#define mi_atomic_storei64_release(p,x) mi_atomic(storei64_explicit)(p,x,mi_memory_order(release)) +#define mi_atomic_storei64_relaxed(p,x) mi_atomic(storei64_explicit)(p,x,mi_memory_order(relaxed)) + + +#endif + + +// Atomically add a signed value; returns the previous value. +static inline intptr_t mi_atomic_addi(_Atomic(intptr_t)*p, intptr_t add) { + return (intptr_t)mi_atomic_add_acq_rel((_Atomic(uintptr_t)*)p, (uintptr_t)add); +} + +// Atomically subtract a signed value; returns the previous value. +static inline intptr_t mi_atomic_subi(_Atomic(intptr_t)*p, intptr_t sub) { + return (intptr_t)mi_atomic_addi(p, -sub); +} + +typedef _Atomic(uintptr_t) mi_atomic_once_t; + +// Returns true only on the first invocation +static inline bool mi_atomic_once( mi_atomic_once_t* once ) { + if (mi_atomic_load_relaxed(once) != 0) return false; // quick test + uintptr_t expected = 0; + return mi_atomic_cas_strong_acq_rel(once, &expected, (uintptr_t)1); // try to set to 1 +} + +typedef _Atomic(uintptr_t) mi_atomic_guard_t; + +// Allows only one thread to execute at a time +#define mi_atomic_guard(guard) \ + uintptr_t _mi_guard_expected = 0; \ + for(bool _mi_guard_once = true; \ + _mi_guard_once && mi_atomic_cas_strong_acq_rel(guard,&_mi_guard_expected,(uintptr_t)1); \ + (mi_atomic_store_release(guard,(uintptr_t)0), _mi_guard_once = false) ) + + + +// Yield +#if defined(__cplusplus) +#include +static inline void mi_atomic_yield(void) { + std::this_thread::yield(); +} +#elif defined(_WIN32) +#define WIN32_LEAN_AND_MEAN +#include +static inline void mi_atomic_yield(void) { + YieldProcessor(); +} +#elif defined(__SSE2__) +#include +static inline void mi_atomic_yield(void) { + _mm_pause(); +} +#elif (defined(__GNUC__) || defined(__clang__)) && \ + (defined(__x86_64__) || defined(__i386__) || defined(__arm__) || defined(__armel__) || defined(__ARMEL__) || \ + defined(__aarch64__) || defined(__powerpc__) || defined(__ppc__) || defined(__PPC__)) || defined(__POWERPC__) +#if defined(__x86_64__) || defined(__i386__) +static inline void mi_atomic_yield(void) { + __asm__ volatile ("pause" ::: "memory"); +} +#elif defined(__aarch64__) +static inline void mi_atomic_yield(void) { + __asm__ volatile("wfe"); +} +#elif (defined(__arm__) && __ARM_ARCH__ >= 7) +static inline void mi_atomic_yield(void) { + __asm__ volatile("yield" ::: "memory"); +} +#elif defined(__powerpc__) || defined(__ppc__) || defined(__PPC__) || defined(__POWERPC__) +#ifdef __APPLE__ +static inline void mi_atomic_yield(void) { + __asm__ volatile ("or r27,r27,r27" ::: "memory"); +} +#else +static inline void mi_atomic_yield(void) { + __asm__ __volatile__ ("or 27,27,27" ::: "memory"); +} +#endif +#elif defined(__armel__) || defined(__ARMEL__) +static inline void mi_atomic_yield(void) { + __asm__ volatile ("nop" ::: "memory"); +} +#endif +#elif defined(__sun) +// Fallback for other archs +#include +static inline void mi_atomic_yield(void) { + smt_pause(); +} +#elif defined(__wasi__) +#include +static inline void mi_atomic_yield(void) { + sched_yield(); +} +#else +#include +static inline void mi_atomic_yield(void) { + sleep(0); +} +#endif + + +#endif // __MIMALLOC_ATOMIC_H diff --git a/compat/mimalloc/mimalloc/internal.h b/compat/mimalloc/mimalloc/internal.h new file mode 100644 index 00000000000000..f076bc6a40f977 --- /dev/null +++ b/compat/mimalloc/mimalloc/internal.h @@ -0,0 +1,979 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#pragma once +#ifndef MIMALLOC_INTERNAL_H +#define MIMALLOC_INTERNAL_H + + +// -------------------------------------------------------------------------- +// This file contains the interal API's of mimalloc and various utility +// functions and macros. +// -------------------------------------------------------------------------- + +#include "mimalloc/types.h" +#include "mimalloc/track.h" + +#if (MI_DEBUG>0) +#define mi_trace_message(...) _mi_trace_message(__VA_ARGS__) +#else +#define mi_trace_message(...) +#endif + +#define MI_CACHE_LINE 64 +#if defined(_MSC_VER) +#pragma warning(disable:4127) // suppress constant conditional warning (due to MI_SECURE paths) +#pragma warning(disable:26812) // unscoped enum warning +#define mi_decl_noinline __declspec(noinline) +#define mi_decl_thread __declspec(thread) +#define mi_decl_cache_align __declspec(align(MI_CACHE_LINE)) +#elif (defined(__GNUC__) && (__GNUC__ >= 3)) || defined(__clang__) // includes clang and icc +#define mi_decl_noinline __attribute__((noinline)) +#define mi_decl_thread __thread +#define mi_decl_cache_align __attribute__((aligned(MI_CACHE_LINE))) +#else +#define mi_decl_noinline +#define mi_decl_thread __thread // hope for the best :-) +#define mi_decl_cache_align +#endif + +#if defined(__EMSCRIPTEN__) && !defined(__wasi__) +#define __wasi__ +#endif + +#if defined(__cplusplus) +#define mi_decl_externc extern "C" +#else +#define mi_decl_externc +#endif + +// pthreads +#if !defined(_WIN32) && !defined(__wasi__) +#define MI_USE_PTHREADS +#include +#endif + +// "options.c" +void _mi_fputs(mi_output_fun* out, void* arg, const char* prefix, const char* message); +void _mi_fprintf(mi_output_fun* out, void* arg, const char* fmt, ...); +void _mi_warning_message(const char* fmt, ...); +void _mi_verbose_message(const char* fmt, ...); +void _mi_trace_message(const char* fmt, ...); +void _mi_options_init(void); +void _mi_error_message(int err, const char* fmt, ...); + +// random.c +void _mi_random_init(mi_random_ctx_t* ctx); +void _mi_random_init_weak(mi_random_ctx_t* ctx); +void _mi_random_reinit_if_weak(mi_random_ctx_t * ctx); +void _mi_random_split(mi_random_ctx_t* ctx, mi_random_ctx_t* new_ctx); +uintptr_t _mi_random_next(mi_random_ctx_t* ctx); +uintptr_t _mi_heap_random_next(mi_heap_t* heap); +uintptr_t _mi_os_random_weak(uintptr_t extra_seed); +static inline uintptr_t _mi_random_shuffle(uintptr_t x); + +// init.c +extern mi_decl_cache_align mi_stats_t _mi_stats_main; +extern mi_decl_cache_align const mi_page_t _mi_page_empty; +bool _mi_is_main_thread(void); +size_t _mi_current_thread_count(void); +bool _mi_preloading(void); // true while the C runtime is not initialized yet +mi_threadid_t _mi_thread_id(void) mi_attr_noexcept; +mi_heap_t* _mi_heap_main_get(void); // statically allocated main backing heap +void _mi_thread_done(mi_heap_t* heap); +void _mi_thread_data_collect(void); + +// os.c +void _mi_os_init(void); // called from process init +void* _mi_os_alloc(size_t size, mi_memid_t* memid, mi_stats_t* stats); +void _mi_os_free(void* p, size_t size, mi_memid_t memid, mi_stats_t* stats); +void _mi_os_free_ex(void* p, size_t size, bool still_committed, mi_memid_t memid, mi_stats_t* stats); + +size_t _mi_os_page_size(void); +size_t _mi_os_good_alloc_size(size_t size); +bool _mi_os_has_overcommit(void); +bool _mi_os_has_virtual_reserve(void); + +bool _mi_os_purge(void* p, size_t size, mi_stats_t* stats); +bool _mi_os_reset(void* addr, size_t size, mi_stats_t* tld_stats); +bool _mi_os_commit(void* p, size_t size, bool* is_zero, mi_stats_t* stats); +bool _mi_os_decommit(void* addr, size_t size, mi_stats_t* stats); +bool _mi_os_protect(void* addr, size_t size); +bool _mi_os_unprotect(void* addr, size_t size); +bool _mi_os_purge(void* p, size_t size, mi_stats_t* stats); +bool _mi_os_purge_ex(void* p, size_t size, bool allow_reset, mi_stats_t* stats); + +void* _mi_os_alloc_aligned(size_t size, size_t alignment, bool commit, bool allow_large, mi_memid_t* memid, mi_stats_t* stats); +void* _mi_os_alloc_aligned_at_offset(size_t size, size_t alignment, size_t align_offset, bool commit, bool allow_large, mi_memid_t* memid, mi_stats_t* tld_stats); + +void* _mi_os_get_aligned_hint(size_t try_alignment, size_t size); +bool _mi_os_use_large_page(size_t size, size_t alignment); +size_t _mi_os_large_page_size(void); + +void* _mi_os_alloc_huge_os_pages(size_t pages, int numa_node, mi_msecs_t max_secs, size_t* pages_reserved, size_t* psize, mi_memid_t* memid); + +// arena.c +mi_arena_id_t _mi_arena_id_none(void); +void _mi_arena_free(void* p, size_t size, size_t still_committed_size, mi_memid_t memid, mi_stats_t* stats); +void* _mi_arena_alloc(size_t size, bool commit, bool allow_large, mi_arena_id_t req_arena_id, mi_memid_t* memid, mi_os_tld_t* tld); +void* _mi_arena_alloc_aligned(size_t size, size_t alignment, size_t align_offset, bool commit, bool allow_large, mi_arena_id_t req_arena_id, mi_memid_t* memid, mi_os_tld_t* tld); +bool _mi_arena_memid_is_suitable(mi_memid_t memid, mi_arena_id_t request_arena_id); +bool _mi_arena_contains(const void* p); +void _mi_arena_collect(bool force_purge, mi_stats_t* stats); +void _mi_arena_unsafe_destroy_all(mi_stats_t* stats); + +// "segment-map.c" +void _mi_segment_map_allocated_at(const mi_segment_t* segment); +void _mi_segment_map_freed_at(const mi_segment_t* segment); + +// "segment.c" +mi_page_t* _mi_segment_page_alloc(mi_heap_t* heap, size_t block_size, size_t page_alignment, mi_segments_tld_t* tld, mi_os_tld_t* os_tld); +void _mi_segment_page_free(mi_page_t* page, bool force, mi_segments_tld_t* tld); +void _mi_segment_page_abandon(mi_page_t* page, mi_segments_tld_t* tld); +bool _mi_segment_try_reclaim_abandoned( mi_heap_t* heap, bool try_all, mi_segments_tld_t* tld); +void _mi_segment_thread_collect(mi_segments_tld_t* tld); + +#if MI_HUGE_PAGE_ABANDON +void _mi_segment_huge_page_free(mi_segment_t* segment, mi_page_t* page, mi_block_t* block); +#else +void _mi_segment_huge_page_reset(mi_segment_t* segment, mi_page_t* page, mi_block_t* block); +#endif + +uint8_t* _mi_segment_page_start(const mi_segment_t* segment, const mi_page_t* page, size_t* page_size); // page start for any page +void _mi_abandoned_reclaim_all(mi_heap_t* heap, mi_segments_tld_t* tld); +void _mi_abandoned_await_readers(void); +void _mi_abandoned_collect(mi_heap_t* heap, bool force, mi_segments_tld_t* tld); + +// "page.c" +void* _mi_malloc_generic(mi_heap_t* heap, size_t size, bool zero, size_t huge_alignment) mi_attr_noexcept mi_attr_malloc; + +void _mi_page_retire(mi_page_t* page) mi_attr_noexcept; // free the page if there are no other pages with many free blocks +void _mi_page_unfull(mi_page_t* page); +void _mi_page_free(mi_page_t* page, mi_page_queue_t* pq, bool force); // free the page +void _mi_page_abandon(mi_page_t* page, mi_page_queue_t* pq); // abandon the page, to be picked up by another thread... +void _mi_heap_delayed_free_all(mi_heap_t* heap); +bool _mi_heap_delayed_free_partial(mi_heap_t* heap); +void _mi_heap_collect_retired(mi_heap_t* heap, bool force); + +void _mi_page_use_delayed_free(mi_page_t* page, mi_delayed_t delay, bool override_never); +bool _mi_page_try_use_delayed_free(mi_page_t* page, mi_delayed_t delay, bool override_never); +size_t _mi_page_queue_append(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_queue_t* append); +void _mi_deferred_free(mi_heap_t* heap, bool force); + +void _mi_page_free_collect(mi_page_t* page,bool force); +void _mi_page_reclaim(mi_heap_t* heap, mi_page_t* page); // callback from segments + +size_t _mi_bin_size(uint8_t bin); // for stats +uint8_t _mi_bin(size_t size); // for stats + +// "heap.c" +void _mi_heap_destroy_pages(mi_heap_t* heap); +void _mi_heap_collect_abandon(mi_heap_t* heap); +void _mi_heap_set_default_direct(mi_heap_t* heap); +bool _mi_heap_memid_is_suitable(mi_heap_t* heap, mi_memid_t memid); +void _mi_heap_unsafe_destroy_all(void); + +// "stats.c" +void _mi_stats_done(mi_stats_t* stats); +mi_msecs_t _mi_clock_now(void); +mi_msecs_t _mi_clock_end(mi_msecs_t start); +mi_msecs_t _mi_clock_start(void); + +// "alloc.c" +void* _mi_page_malloc(mi_heap_t* heap, mi_page_t* page, size_t size, bool zero) mi_attr_noexcept; // called from `_mi_malloc_generic` +void* _mi_heap_malloc_zero(mi_heap_t* heap, size_t size, bool zero) mi_attr_noexcept; +void* _mi_heap_malloc_zero_ex(mi_heap_t* heap, size_t size, bool zero, size_t huge_alignment) mi_attr_noexcept; // called from `_mi_heap_malloc_aligned` +void* _mi_heap_realloc_zero(mi_heap_t* heap, void* p, size_t newsize, bool zero) mi_attr_noexcept; +mi_block_t* _mi_page_ptr_unalign(const mi_segment_t* segment, const mi_page_t* page, const void* p); +bool _mi_free_delayed_block(mi_block_t* block); +void _mi_free_generic(const mi_segment_t* segment, mi_page_t* page, bool is_local, void* p) mi_attr_noexcept; // for runtime integration +void _mi_padding_shrink(const mi_page_t* page, const mi_block_t* block, const size_t min_size); + +// option.c, c primitives +char _mi_toupper(char c); +int _mi_strnicmp(const char* s, const char* t, size_t n); +void _mi_strlcpy(char* dest, const char* src, size_t dest_size); +void _mi_strlcat(char* dest, const char* src, size_t dest_size); +size_t _mi_strlen(const char* s); +size_t _mi_strnlen(const char* s, size_t max_len); + + +#if MI_DEBUG>1 +bool _mi_page_is_valid(mi_page_t* page); +#endif + + +// ------------------------------------------------------ +// Branches +// ------------------------------------------------------ + +#if defined(__GNUC__) || defined(__clang__) +#define mi_unlikely(x) (__builtin_expect(!!(x),false)) +#define mi_likely(x) (__builtin_expect(!!(x),true)) +#elif (defined(__cplusplus) && (__cplusplus >= 202002L)) || (defined(_MSVC_LANG) && _MSVC_LANG >= 202002L) +#define mi_unlikely(x) (x) [[unlikely]] +#define mi_likely(x) (x) [[likely]] +#else +#define mi_unlikely(x) (x) +#define mi_likely(x) (x) +#endif + +#ifndef __has_builtin +#define __has_builtin(x) 0 +#endif + + +/* ----------------------------------------------------------- + Error codes passed to `_mi_fatal_error` + All are recoverable but EFAULT is a serious error and aborts by default in secure mode. + For portability define undefined error codes using common Unix codes: + +----------------------------------------------------------- */ +#include +#ifndef EAGAIN // double free +#define EAGAIN (11) +#endif +#ifndef ENOMEM // out of memory +#define ENOMEM (12) +#endif +#ifndef EFAULT // corrupted free-list or meta-data +#define EFAULT (14) +#endif +#ifndef EINVAL // trying to free an invalid pointer +#define EINVAL (22) +#endif +#ifndef EOVERFLOW // count*size overflow +#define EOVERFLOW (75) +#endif + + +/* ----------------------------------------------------------- + Inlined definitions +----------------------------------------------------------- */ +#define MI_UNUSED(x) (void)(x) +#if (MI_DEBUG>0) +#define MI_UNUSED_RELEASE(x) +#else +#define MI_UNUSED_RELEASE(x) MI_UNUSED(x) +#endif + +#define MI_INIT4(x) x(),x(),x(),x() +#define MI_INIT8(x) MI_INIT4(x),MI_INIT4(x) +#define MI_INIT16(x) MI_INIT8(x),MI_INIT8(x) +#define MI_INIT32(x) MI_INIT16(x),MI_INIT16(x) +#define MI_INIT64(x) MI_INIT32(x),MI_INIT32(x) +#define MI_INIT128(x) MI_INIT64(x),MI_INIT64(x) +#define MI_INIT256(x) MI_INIT128(x),MI_INIT128(x) + + +#include +// initialize a local variable to zero; use memset as compilers optimize constant sized memset's +#define _mi_memzero_var(x) memset(&x,0,sizeof(x)) + +// Is `x` a power of two? (0 is considered a power of two) +static inline bool _mi_is_power_of_two(uintptr_t x) { + return ((x & (x - 1)) == 0); +} + +// Is a pointer aligned? +static inline bool _mi_is_aligned(void* p, size_t alignment) { + mi_assert_internal(alignment != 0); + return (((uintptr_t)p % alignment) == 0); +} + +// Align upwards +static inline uintptr_t _mi_align_up(uintptr_t sz, size_t alignment) { + mi_assert_internal(alignment != 0); + uintptr_t mask = alignment - 1; + if ((alignment & mask) == 0) { // power of two? + return ((sz + mask) & ~mask); + } + else { + return (((sz + mask)/alignment)*alignment); + } +} + +// Align downwards +static inline uintptr_t _mi_align_down(uintptr_t sz, size_t alignment) { + mi_assert_internal(alignment != 0); + uintptr_t mask = alignment - 1; + if ((alignment & mask) == 0) { // power of two? + return (sz & ~mask); + } + else { + return ((sz / alignment) * alignment); + } +} + +// Divide upwards: `s <= _mi_divide_up(s,d)*d < s+d`. +static inline uintptr_t _mi_divide_up(uintptr_t size, size_t divider) { + mi_assert_internal(divider != 0); + return (divider == 0 ? size : ((size + divider - 1) / divider)); +} + +// Is memory zero initialized? +static inline bool mi_mem_is_zero(const void* p, size_t size) { + for (size_t i = 0; i < size; i++) { + if (((uint8_t*)p)[i] != 0) return false; + } + return true; +} + + +// Align a byte size to a size in _machine words_, +// i.e. byte size == `wsize*sizeof(void*)`. +static inline size_t _mi_wsize_from_size(size_t size) { + mi_assert_internal(size <= SIZE_MAX - sizeof(uintptr_t)); + return (size + sizeof(uintptr_t) - 1) / sizeof(uintptr_t); +} + +// Overflow detecting multiply +#if __has_builtin(__builtin_umul_overflow) || (defined(__GNUC__) && (__GNUC__ >= 5)) +#include // UINT_MAX, ULONG_MAX +#if defined(_CLOCK_T) // for Illumos +#undef _CLOCK_T +#endif +static inline bool mi_mul_overflow(size_t count, size_t size, size_t* total) { + #if (SIZE_MAX == ULONG_MAX) + return __builtin_umull_overflow(count, size, (unsigned long *)total); + #elif (SIZE_MAX == UINT_MAX) + return __builtin_umul_overflow(count, size, (unsigned int *)total); + #else + return __builtin_umulll_overflow(count, size, (unsigned long long *)total); + #endif +} +#else /* __builtin_umul_overflow is unavailable */ +static inline bool mi_mul_overflow(size_t count, size_t size, size_t* total) { + #define MI_MUL_NO_OVERFLOW ((size_t)1 << (4*sizeof(size_t))) // sqrt(SIZE_MAX) + *total = count * size; + // note: gcc/clang optimize this to directly check the overflow flag + return ((size >= MI_MUL_NO_OVERFLOW || count >= MI_MUL_NO_OVERFLOW) && size > 0 && (SIZE_MAX / size) < count); +} +#endif + +// Safe multiply `count*size` into `total`; return `true` on overflow. +static inline bool mi_count_size_overflow(size_t count, size_t size, size_t* total) { + if (count==1) { // quick check for the case where count is one (common for C++ allocators) + *total = size; + return false; + } + else if mi_unlikely(mi_mul_overflow(count, size, total)) { + #if MI_DEBUG > 0 + _mi_error_message(EOVERFLOW, "allocation request is too large (%zu * %zu bytes)\n", count, size); + #endif + *total = SIZE_MAX; + return true; + } + else return false; +} + + +/*---------------------------------------------------------------------------------------- + Heap functions +------------------------------------------------------------------------------------------- */ + +extern const mi_heap_t _mi_heap_empty; // read-only empty heap, initial value of the thread local default heap + +static inline bool mi_heap_is_backing(const mi_heap_t* heap) { + return (heap->tld->heap_backing == heap); +} + +static inline bool mi_heap_is_initialized(mi_heap_t* heap) { + mi_assert_internal(heap != NULL); + return (heap != &_mi_heap_empty); +} + +static inline uintptr_t _mi_ptr_cookie(const void* p) { + extern mi_heap_t _mi_heap_main; + mi_assert_internal(_mi_heap_main.cookie != 0); + return ((uintptr_t)p ^ _mi_heap_main.cookie); +} + +/* ----------------------------------------------------------- + Pages +----------------------------------------------------------- */ + +static inline mi_page_t* _mi_heap_get_free_small_page(mi_heap_t* heap, size_t size) { + mi_assert_internal(size <= (MI_SMALL_SIZE_MAX + MI_PADDING_SIZE)); + const size_t idx = _mi_wsize_from_size(size); + mi_assert_internal(idx < MI_PAGES_DIRECT); + return heap->pages_free_direct[idx]; +} + +// Segment that contains the pointer +// Large aligned blocks may be aligned at N*MI_SEGMENT_SIZE (inside a huge segment > MI_SEGMENT_SIZE), +// and we need align "down" to the segment info which is `MI_SEGMENT_SIZE` bytes before it; +// therefore we align one byte before `p`. +static inline mi_segment_t* _mi_ptr_segment(const void* p) { + mi_assert_internal(p != NULL); + return (mi_segment_t*)(((uintptr_t)p - 1) & ~MI_SEGMENT_MASK); +} + +static inline mi_page_t* mi_slice_to_page(mi_slice_t* s) { + mi_assert_internal(s->slice_offset== 0 && s->slice_count > 0); + return (mi_page_t*)(s); +} + +static inline mi_slice_t* mi_page_to_slice(mi_page_t* p) { + mi_assert_internal(p->slice_offset== 0 && p->slice_count > 0); + return (mi_slice_t*)(p); +} + +// Segment belonging to a page +static inline mi_segment_t* _mi_page_segment(const mi_page_t* page) { + mi_segment_t* segment = _mi_ptr_segment(page); + mi_assert_internal(segment == NULL || ((mi_slice_t*)page >= segment->slices && (mi_slice_t*)page < segment->slices + segment->slice_entries)); + return segment; +} + +static inline mi_slice_t* mi_slice_first(const mi_slice_t* slice) { + mi_slice_t* start = (mi_slice_t*)((uint8_t*)slice - slice->slice_offset); + mi_assert_internal(start >= _mi_ptr_segment(slice)->slices); + mi_assert_internal(start->slice_offset == 0); + mi_assert_internal(start + start->slice_count > slice); + return start; +} + +// Get the page containing the pointer (performance critical as it is called in mi_free) +static inline mi_page_t* _mi_segment_page_of(const mi_segment_t* segment, const void* p) { + mi_assert_internal(p > (void*)segment); + ptrdiff_t diff = (uint8_t*)p - (uint8_t*)segment; + mi_assert_internal(diff > 0 && diff <= (ptrdiff_t)MI_SEGMENT_SIZE); + size_t idx = (size_t)diff >> MI_SEGMENT_SLICE_SHIFT; + mi_assert_internal(idx <= segment->slice_entries); + mi_slice_t* slice0 = (mi_slice_t*)&segment->slices[idx]; + mi_slice_t* slice = mi_slice_first(slice0); // adjust to the block that holds the page data + mi_assert_internal(slice->slice_offset == 0); + mi_assert_internal(slice >= segment->slices && slice < segment->slices + segment->slice_entries); + return mi_slice_to_page(slice); +} + +// Quick page start for initialized pages +static inline uint8_t* _mi_page_start(const mi_segment_t* segment, const mi_page_t* page, size_t* page_size) { + return _mi_segment_page_start(segment, page, page_size); +} + +// Get the page containing the pointer +static inline mi_page_t* _mi_ptr_page(void* p) { + return _mi_segment_page_of(_mi_ptr_segment(p), p); +} + +// Get the block size of a page (special case for huge objects) +static inline size_t mi_page_block_size(const mi_page_t* page) { + const size_t bsize = page->xblock_size; + mi_assert_internal(bsize > 0); + if mi_likely(bsize < MI_HUGE_BLOCK_SIZE) { + return bsize; + } + else { + size_t psize; + _mi_segment_page_start(_mi_page_segment(page), page, &psize); + return psize; + } +} + +static inline bool mi_page_is_huge(const mi_page_t* page) { + return (_mi_page_segment(page)->kind == MI_SEGMENT_HUGE); +} + +// Get the usable block size of a page without fixed padding. +// This may still include internal padding due to alignment and rounding up size classes. +static inline size_t mi_page_usable_block_size(const mi_page_t* page) { + return mi_page_block_size(page) - MI_PADDING_SIZE; +} + +// size of a segment +static inline size_t mi_segment_size(mi_segment_t* segment) { + return segment->segment_slices * MI_SEGMENT_SLICE_SIZE; +} + +static inline uint8_t* mi_segment_end(mi_segment_t* segment) { + return (uint8_t*)segment + mi_segment_size(segment); +} + +// Thread free access +static inline mi_block_t* mi_page_thread_free(const mi_page_t* page) { + return (mi_block_t*)(mi_atomic_load_relaxed(&((mi_page_t*)page)->xthread_free) & ~3); +} + +static inline mi_delayed_t mi_page_thread_free_flag(const mi_page_t* page) { + return (mi_delayed_t)(mi_atomic_load_relaxed(&((mi_page_t*)page)->xthread_free) & 3); +} + +// Heap access +static inline mi_heap_t* mi_page_heap(const mi_page_t* page) { + return (mi_heap_t*)(mi_atomic_load_relaxed(&((mi_page_t*)page)->xheap)); +} + +static inline void mi_page_set_heap(mi_page_t* page, mi_heap_t* heap) { + mi_assert_internal(mi_page_thread_free_flag(page) != MI_DELAYED_FREEING); + mi_atomic_store_release(&page->xheap,(uintptr_t)heap); +} + +// Thread free flag helpers +static inline mi_block_t* mi_tf_block(mi_thread_free_t tf) { + return (mi_block_t*)(tf & ~0x03); +} +static inline mi_delayed_t mi_tf_delayed(mi_thread_free_t tf) { + return (mi_delayed_t)(tf & 0x03); +} +static inline mi_thread_free_t mi_tf_make(mi_block_t* block, mi_delayed_t delayed) { + return (mi_thread_free_t)((uintptr_t)block | (uintptr_t)delayed); +} +static inline mi_thread_free_t mi_tf_set_delayed(mi_thread_free_t tf, mi_delayed_t delayed) { + return mi_tf_make(mi_tf_block(tf),delayed); +} +static inline mi_thread_free_t mi_tf_set_block(mi_thread_free_t tf, mi_block_t* block) { + return mi_tf_make(block, mi_tf_delayed(tf)); +} + +// are all blocks in a page freed? +// note: needs up-to-date used count, (as the `xthread_free` list may not be empty). see `_mi_page_collect_free`. +static inline bool mi_page_all_free(const mi_page_t* page) { + mi_assert_internal(page != NULL); + return (page->used == 0); +} + +// are there any available blocks? +static inline bool mi_page_has_any_available(const mi_page_t* page) { + mi_assert_internal(page != NULL && page->reserved > 0); + return (page->used < page->reserved || (mi_page_thread_free(page) != NULL)); +} + +// are there immediately available blocks, i.e. blocks available on the free list. +static inline bool mi_page_immediate_available(const mi_page_t* page) { + mi_assert_internal(page != NULL); + return (page->free != NULL); +} + +// is more than 7/8th of a page in use? +static inline bool mi_page_mostly_used(const mi_page_t* page) { + if (page==NULL) return true; + uint16_t frac = page->reserved / 8U; + return (page->reserved - page->used <= frac); +} + +static inline mi_page_queue_t* mi_page_queue(const mi_heap_t* heap, size_t size) { + return &((mi_heap_t*)heap)->pages[_mi_bin(size)]; +} + + + +//----------------------------------------------------------- +// Page flags +//----------------------------------------------------------- +static inline bool mi_page_is_in_full(const mi_page_t* page) { + return page->flags.x.in_full; +} + +static inline void mi_page_set_in_full(mi_page_t* page, bool in_full) { + page->flags.x.in_full = in_full; +} + +static inline bool mi_page_has_aligned(const mi_page_t* page) { + return page->flags.x.has_aligned; +} + +static inline void mi_page_set_has_aligned(mi_page_t* page, bool has_aligned) { + page->flags.x.has_aligned = has_aligned; +} + + +/* ------------------------------------------------------------------- +Encoding/Decoding the free list next pointers + +This is to protect against buffer overflow exploits where the +free list is mutated. Many hardened allocators xor the next pointer `p` +with a secret key `k1`, as `p^k1`. This prevents overwriting with known +values but might be still too weak: if the attacker can guess +the pointer `p` this can reveal `k1` (since `p^k1^p == k1`). +Moreover, if multiple blocks can be read as well, the attacker can +xor both as `(p1^k1) ^ (p2^k1) == p1^p2` which may reveal a lot +about the pointers (and subsequently `k1`). + +Instead mimalloc uses an extra key `k2` and encodes as `((p^k2)<<> (MI_INTPTR_BITS - shift)))); +} +static inline uintptr_t mi_rotr(uintptr_t x, uintptr_t shift) { + shift %= MI_INTPTR_BITS; + return (shift==0 ? x : ((x >> shift) | (x << (MI_INTPTR_BITS - shift)))); +} + +static inline void* mi_ptr_decode(const void* null, const mi_encoded_t x, const uintptr_t* keys) { + void* p = (void*)(mi_rotr(x - keys[0], keys[0]) ^ keys[1]); + return (p==null ? NULL : p); +} + +static inline mi_encoded_t mi_ptr_encode(const void* null, const void* p, const uintptr_t* keys) { + uintptr_t x = (uintptr_t)(p==NULL ? null : p); + return mi_rotl(x ^ keys[1], keys[0]) + keys[0]; +} + +static inline mi_block_t* mi_block_nextx( const void* null, const mi_block_t* block, const uintptr_t* keys ) { + mi_track_mem_defined(block,sizeof(mi_block_t)); + mi_block_t* next; + #ifdef MI_ENCODE_FREELIST + next = (mi_block_t*)mi_ptr_decode(null, block->next, keys); + #else + MI_UNUSED(keys); MI_UNUSED(null); + next = (mi_block_t*)block->next; + #endif + mi_track_mem_noaccess(block,sizeof(mi_block_t)); + return next; +} + +static inline void mi_block_set_nextx(const void* null, mi_block_t* block, const mi_block_t* next, const uintptr_t* keys) { + mi_track_mem_undefined(block,sizeof(mi_block_t)); + #ifdef MI_ENCODE_FREELIST + block->next = mi_ptr_encode(null, next, keys); + #else + MI_UNUSED(keys); MI_UNUSED(null); + block->next = (mi_encoded_t)next; + #endif + mi_track_mem_noaccess(block,sizeof(mi_block_t)); +} + +static inline mi_block_t* mi_block_next(const mi_page_t* page, const mi_block_t* block) { + #ifdef MI_ENCODE_FREELIST + mi_block_t* next = mi_block_nextx(page,block,page->keys); + // check for free list corruption: is `next` at least in the same page? + // TODO: check if `next` is `page->block_size` aligned? + if mi_unlikely(next!=NULL && !mi_is_in_same_page(block, next)) { + _mi_error_message(EFAULT, "corrupted free list entry of size %zub at %p: value 0x%zx\n", mi_page_block_size(page), block, (uintptr_t)next); + next = NULL; + } + return next; + #else + MI_UNUSED(page); + return mi_block_nextx(page,block,NULL); + #endif +} + +static inline void mi_block_set_next(const mi_page_t* page, mi_block_t* block, const mi_block_t* next) { + #ifdef MI_ENCODE_FREELIST + mi_block_set_nextx(page,block,next, page->keys); + #else + MI_UNUSED(page); + mi_block_set_nextx(page,block,next,NULL); + #endif +} + + +// ------------------------------------------------------------------- +// commit mask +// ------------------------------------------------------------------- + +static inline void mi_commit_mask_create_empty(mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + cm->mask[i] = 0; + } +} + +static inline void mi_commit_mask_create_full(mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + cm->mask[i] = ~((size_t)0); + } +} + +static inline bool mi_commit_mask_is_empty(const mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + if (cm->mask[i] != 0) return false; + } + return true; +} + +static inline bool mi_commit_mask_is_full(const mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + if (cm->mask[i] != ~((size_t)0)) return false; + } + return true; +} + +// defined in `segment.c`: +size_t _mi_commit_mask_committed_size(const mi_commit_mask_t* cm, size_t total); +size_t _mi_commit_mask_next_run(const mi_commit_mask_t* cm, size_t* idx); + +#define mi_commit_mask_foreach(cm,idx,count) \ + idx = 0; \ + while ((count = _mi_commit_mask_next_run(cm,&idx)) > 0) { + +#define mi_commit_mask_foreach_end() \ + idx += count; \ + } + + + +/* ----------------------------------------------------------- + memory id's +----------------------------------------------------------- */ + +static inline mi_memid_t _mi_memid_create(mi_memkind_t memkind) { + mi_memid_t memid; + _mi_memzero_var(memid); + memid.memkind = memkind; + return memid; +} + +static inline mi_memid_t _mi_memid_none(void) { + return _mi_memid_create(MI_MEM_NONE); +} + +static inline mi_memid_t _mi_memid_create_os(bool committed, bool is_zero, bool is_large) { + mi_memid_t memid = _mi_memid_create(MI_MEM_OS); + memid.initially_committed = committed; + memid.initially_zero = is_zero; + memid.is_pinned = is_large; + return memid; +} + + +// ------------------------------------------------------------------- +// Fast "random" shuffle +// ------------------------------------------------------------------- + +static inline uintptr_t _mi_random_shuffle(uintptr_t x) { + if (x==0) { x = 17; } // ensure we don't get stuck in generating zeros +#if (MI_INTPTR_SIZE==8) + // by Sebastiano Vigna, see: + x ^= x >> 30; + x *= 0xbf58476d1ce4e5b9UL; + x ^= x >> 27; + x *= 0x94d049bb133111ebUL; + x ^= x >> 31; +#elif (MI_INTPTR_SIZE==4) + // by Chris Wellons, see: + x ^= x >> 16; + x *= 0x7feb352dUL; + x ^= x >> 15; + x *= 0x846ca68bUL; + x ^= x >> 16; +#endif + return x; +} + +// ------------------------------------------------------------------- +// Optimize numa node access for the common case (= one node) +// ------------------------------------------------------------------- + +int _mi_os_numa_node_get(mi_os_tld_t* tld); +size_t _mi_os_numa_node_count_get(void); + +extern _Atomic(size_t) _mi_numa_node_count; +static inline int _mi_os_numa_node(mi_os_tld_t* tld) { + if mi_likely(mi_atomic_load_relaxed(&_mi_numa_node_count) == 1) { return 0; } + else return _mi_os_numa_node_get(tld); +} +static inline size_t _mi_os_numa_node_count(void) { + const size_t count = mi_atomic_load_relaxed(&_mi_numa_node_count); + if mi_likely(count > 0) { return count; } + else return _mi_os_numa_node_count_get(); +} + + + +// ----------------------------------------------------------------------- +// Count bits: trailing or leading zeros (with MI_INTPTR_BITS on all zero) +// ----------------------------------------------------------------------- + +#if defined(__GNUC__) + +#include // LONG_MAX +#define MI_HAVE_FAST_BITSCAN +static inline size_t mi_clz(uintptr_t x) { + if (x==0) return MI_INTPTR_BITS; +#if (INTPTR_MAX == LONG_MAX) + return __builtin_clzl(x); +#else + return __builtin_clzll(x); +#endif +} +static inline size_t mi_ctz(uintptr_t x) { + if (x==0) return MI_INTPTR_BITS; +#if (INTPTR_MAX == LONG_MAX) + return __builtin_ctzl(x); +#else + return __builtin_ctzll(x); +#endif +} + +#elif defined(_MSC_VER) + +#include // LONG_MAX +#include // BitScanReverse64 +#define MI_HAVE_FAST_BITSCAN +static inline size_t mi_clz(uintptr_t x) { + if (x==0) return MI_INTPTR_BITS; + unsigned long idx; +#if (INTPTR_MAX == LONG_MAX) + _BitScanReverse(&idx, x); +#else + _BitScanReverse64(&idx, x); +#endif + return ((MI_INTPTR_BITS - 1) - idx); +} +static inline size_t mi_ctz(uintptr_t x) { + if (x==0) return MI_INTPTR_BITS; + unsigned long idx; +#if (INTPTR_MAX == LONG_MAX) + _BitScanForward(&idx, x); +#else + _BitScanForward64(&idx, x); +#endif + return idx; +} + +#else +static inline size_t mi_ctz32(uint32_t x) { + // de Bruijn multiplication, see + static const unsigned char debruijn[32] = { + 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8, + 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9 + }; + if (x==0) return 32; + return debruijn[((x & -(int32_t)x) * 0x077CB531UL) >> 27]; +} +static inline size_t mi_clz32(uint32_t x) { + // de Bruijn multiplication, see + static const uint8_t debruijn[32] = { + 31, 22, 30, 21, 18, 10, 29, 2, 20, 17, 15, 13, 9, 6, 28, 1, + 23, 19, 11, 3, 16, 14, 7, 24, 12, 4, 8, 25, 5, 26, 27, 0 + }; + if (x==0) return 32; + x |= x >> 1; + x |= x >> 2; + x |= x >> 4; + x |= x >> 8; + x |= x >> 16; + return debruijn[(uint32_t)(x * 0x07C4ACDDUL) >> 27]; +} + +static inline size_t mi_clz(uintptr_t x) { + if (x==0) return MI_INTPTR_BITS; +#if (MI_INTPTR_BITS <= 32) + return mi_clz32((uint32_t)x); +#else + size_t count = mi_clz32((uint32_t)(x >> 32)); + if (count < 32) return count; + return (32 + mi_clz32((uint32_t)x)); +#endif +} +static inline size_t mi_ctz(uintptr_t x) { + if (x==0) return MI_INTPTR_BITS; +#if (MI_INTPTR_BITS <= 32) + return mi_ctz32((uint32_t)x); +#else + size_t count = mi_ctz32((uint32_t)x); + if (count < 32) return count; + return (32 + mi_ctz32((uint32_t)(x>>32))); +#endif +} + +#endif + +// "bit scan reverse": Return index of the highest bit (or MI_INTPTR_BITS if `x` is zero) +static inline size_t mi_bsr(uintptr_t x) { + return (x==0 ? MI_INTPTR_BITS : MI_INTPTR_BITS - 1 - mi_clz(x)); +} + + +// --------------------------------------------------------------------------------- +// Provide our own `_mi_memcpy` for potential performance optimizations. +// +// For now, only on Windows with msvc/clang-cl we optimize to `rep movsb` if +// we happen to run on x86/x64 cpu's that have "fast short rep movsb" (FSRM) support +// (AMD Zen3+ (~2020) or Intel Ice Lake+ (~2017). See also issue #201 and pr #253. +// --------------------------------------------------------------------------------- + +#if !MI_TRACK_ENABLED && defined(_WIN32) && (defined(_M_IX86) || defined(_M_X64)) +#include +extern bool _mi_cpu_has_fsrm; +static inline void _mi_memcpy(void* dst, const void* src, size_t n) { + if (_mi_cpu_has_fsrm) { + __movsb((unsigned char*)dst, (const unsigned char*)src, n); + } + else { + memcpy(dst, src, n); + } +} +static inline void _mi_memzero(void* dst, size_t n) { + if (_mi_cpu_has_fsrm) { + __stosb((unsigned char*)dst, 0, n); + } + else { + memset(dst, 0, n); + } +} +#else +static inline void _mi_memcpy(void* dst, const void* src, size_t n) { + memcpy(dst, src, n); +} +static inline void _mi_memzero(void* dst, size_t n) { + memset(dst, 0, n); +} +#endif + +// ------------------------------------------------------------------------------- +// The `_mi_memcpy_aligned` can be used if the pointers are machine-word aligned +// This is used for example in `mi_realloc`. +// ------------------------------------------------------------------------------- + +#if (defined(__GNUC__) && (__GNUC__ >= 4)) || defined(__clang__) +// On GCC/CLang we provide a hint that the pointers are word aligned. +static inline void _mi_memcpy_aligned(void* dst, const void* src, size_t n) { + mi_assert_internal(((uintptr_t)dst % MI_INTPTR_SIZE == 0) && ((uintptr_t)src % MI_INTPTR_SIZE == 0)); + void* adst = __builtin_assume_aligned(dst, MI_INTPTR_SIZE); + const void* asrc = __builtin_assume_aligned(src, MI_INTPTR_SIZE); + _mi_memcpy(adst, asrc, n); +} + +static inline void _mi_memzero_aligned(void* dst, size_t n) { + mi_assert_internal((uintptr_t)dst % MI_INTPTR_SIZE == 0); + void* adst = __builtin_assume_aligned(dst, MI_INTPTR_SIZE); + _mi_memzero(adst, n); +} +#else +// Default fallback on `_mi_memcpy` +static inline void _mi_memcpy_aligned(void* dst, const void* src, size_t n) { + mi_assert_internal(((uintptr_t)dst % MI_INTPTR_SIZE == 0) && ((uintptr_t)src % MI_INTPTR_SIZE == 0)); + _mi_memcpy(dst, src, n); +} + +static inline void _mi_memzero_aligned(void* dst, size_t n) { + mi_assert_internal((uintptr_t)dst % MI_INTPTR_SIZE == 0); + _mi_memzero(dst, n); +} +#endif + + +#endif diff --git a/compat/mimalloc/mimalloc/prim.h b/compat/mimalloc/mimalloc/prim.h new file mode 100644 index 00000000000000..1e55cb5f8802d7 --- /dev/null +++ b/compat/mimalloc/mimalloc/prim.h @@ -0,0 +1,323 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#pragma once +#ifndef MIMALLOC_PRIM_H +#define MIMALLOC_PRIM_H + + +// -------------------------------------------------------------------------- +// This file specifies the primitive portability API. +// Each OS/host needs to implement these primitives, see `src/prim` +// for implementations on Window, macOS, WASI, and Linux/Unix. +// +// note: on all primitive functions, we always have result parameters != NUL, and: +// addr != NULL and page aligned +// size > 0 and page aligned +// return value is an error code an int where 0 is success. +// -------------------------------------------------------------------------- + +// OS memory configuration +typedef struct mi_os_mem_config_s { + size_t page_size; // 4KiB + size_t large_page_size; // 2MiB + size_t alloc_granularity; // smallest allocation size (on Windows 64KiB) + bool has_overcommit; // can we reserve more memory than can be actually committed? + bool must_free_whole; // must allocated blocks be freed as a whole (false for mmap, true for VirtualAlloc) + bool has_virtual_reserve; // supports virtual address space reservation? (if true we can reserve virtual address space without using commit or physical memory) +} mi_os_mem_config_t; + +// Initialize +void _mi_prim_mem_init( mi_os_mem_config_t* config ); + +// Free OS memory +int _mi_prim_free(void* addr, size_t size ); + +// Allocate OS memory. Return NULL on error. +// The `try_alignment` is just a hint and the returned pointer does not have to be aligned. +// If `commit` is false, the virtual memory range only needs to be reserved (with no access) +// which will later be committed explicitly using `_mi_prim_commit`. +// `is_zero` is set to true if the memory was zero initialized (as on most OS's) +// pre: !commit => !allow_large +// try_alignment >= _mi_os_page_size() and a power of 2 +int _mi_prim_alloc(size_t size, size_t try_alignment, bool commit, bool allow_large, bool* is_large, bool* is_zero, void** addr); + +// Commit memory. Returns error code or 0 on success. +// For example, on Linux this would make the memory PROT_READ|PROT_WRITE. +// `is_zero` is set to true if the memory was zero initialized (e.g. on Windows) +int _mi_prim_commit(void* addr, size_t size, bool* is_zero); + +// Decommit memory. Returns error code or 0 on success. The `needs_recommit` result is true +// if the memory would need to be re-committed. For example, on Windows this is always true, +// but on Linux we could use MADV_DONTNEED to decommit which does not need a recommit. +// pre: needs_recommit != NULL +int _mi_prim_decommit(void* addr, size_t size, bool* needs_recommit); + +// Reset memory. The range keeps being accessible but the content might be reset. +// Returns error code or 0 on success. +int _mi_prim_reset(void* addr, size_t size); + +// Protect memory. Returns error code or 0 on success. +int _mi_prim_protect(void* addr, size_t size, bool protect); + +// Allocate huge (1GiB) pages possibly associated with a NUMA node. +// `is_zero` is set to true if the memory was zero initialized (as on most OS's) +// pre: size > 0 and a multiple of 1GiB. +// numa_node is either negative (don't care), or a numa node number. +int _mi_prim_alloc_huge_os_pages(void* hint_addr, size_t size, int numa_node, bool* is_zero, void** addr); + +// Return the current NUMA node +size_t _mi_prim_numa_node(void); + +// Return the number of logical NUMA nodes +size_t _mi_prim_numa_node_count(void); + +// Clock ticks +mi_msecs_t _mi_prim_clock_now(void); + +// Return process information (only for statistics) +typedef struct mi_process_info_s { + mi_msecs_t elapsed; + mi_msecs_t utime; + mi_msecs_t stime; + size_t current_rss; + size_t peak_rss; + size_t current_commit; + size_t peak_commit; + size_t page_faults; +} mi_process_info_t; + +void _mi_prim_process_info(mi_process_info_t* pinfo); + +// Default stderr output. (only for warnings etc. with verbose enabled) +// msg != NULL && _mi_strlen(msg) > 0 +void _mi_prim_out_stderr( const char* msg ); + +// Get an environment variable. (only for options) +// name != NULL, result != NULL, result_size >= 64 +bool _mi_prim_getenv(const char* name, char* result, size_t result_size); + + +// Fill a buffer with strong randomness; return `false` on error or if +// there is no strong randomization available. +bool _mi_prim_random_buf(void* buf, size_t buf_len); + +// Called on the first thread start, and should ensure `_mi_thread_done` is called on thread termination. +void _mi_prim_thread_init_auto_done(void); + +// Called on process exit and may take action to clean up resources associated with the thread auto done. +void _mi_prim_thread_done_auto_done(void); + +// Called when the default heap for a thread changes +void _mi_prim_thread_associate_default_heap(mi_heap_t* heap); + + +//------------------------------------------------------------------- +// Thread id: `_mi_prim_thread_id()` +// +// Getting the thread id should be performant as it is called in the +// fast path of `_mi_free` and we specialize for various platforms as +// inlined definitions. Regular code should call `init.c:_mi_thread_id()`. +// We only require _mi_prim_thread_id() to return a unique id +// for each thread (unequal to zero). +//------------------------------------------------------------------- + +// defined in `init.c`; do not use these directly +extern mi_decl_thread mi_heap_t* _mi_heap_default; // default heap to allocate from +extern bool _mi_process_is_initialized; // has mi_process_init been called? + +static inline mi_threadid_t _mi_prim_thread_id(void) mi_attr_noexcept; + +#if defined(_WIN32) + +#define WIN32_LEAN_AND_MEAN +#include +static inline mi_threadid_t _mi_prim_thread_id(void) mi_attr_noexcept { + // Windows: works on Intel and ARM in both 32- and 64-bit + return (uintptr_t)NtCurrentTeb(); +} + +// We use assembly for a fast thread id on the main platforms. The TLS layout depends on +// both the OS and libc implementation so we use specific tests for each main platform. +// If you test on another platform and it works please send a PR :-) +// see also https://akkadia.org/drepper/tls.pdf for more info on the TLS register. +#elif defined(__GNUC__) && ( \ + (defined(__GLIBC__) && (defined(__x86_64__) || defined(__i386__) || defined(__arm__) || defined(__aarch64__))) \ + || (defined(__APPLE__) && (defined(__x86_64__) || defined(__aarch64__))) \ + || (defined(__BIONIC__) && (defined(__x86_64__) || defined(__i386__) || defined(__arm__) || defined(__aarch64__))) \ + || (defined(__FreeBSD__) && (defined(__x86_64__) || defined(__i386__) || defined(__aarch64__))) \ + || (defined(__OpenBSD__) && (defined(__x86_64__) || defined(__i386__) || defined(__aarch64__))) \ + ) + +static inline void* mi_prim_tls_slot(size_t slot) mi_attr_noexcept { + void* res; + const size_t ofs = (slot*sizeof(void*)); + #if defined(__i386__) + __asm__("movl %%gs:%1, %0" : "=r" (res) : "m" (*((void**)ofs)) : ); // x86 32-bit always uses GS + #elif defined(__APPLE__) && defined(__x86_64__) + __asm__("movq %%gs:%1, %0" : "=r" (res) : "m" (*((void**)ofs)) : ); // x86_64 macOSX uses GS + #elif defined(__x86_64__) && (MI_INTPTR_SIZE==4) + __asm__("movl %%fs:%1, %0" : "=r" (res) : "m" (*((void**)ofs)) : ); // x32 ABI + #elif defined(__x86_64__) + __asm__("movq %%fs:%1, %0" : "=r" (res) : "m" (*((void**)ofs)) : ); // x86_64 Linux, BSD uses FS + #elif defined(__arm__) + void** tcb; MI_UNUSED(ofs); + __asm__ volatile ("mrc p15, 0, %0, c13, c0, 3\nbic %0, %0, #3" : "=r" (tcb)); + res = tcb[slot]; + #elif defined(__aarch64__) + void** tcb; MI_UNUSED(ofs); + #if defined(__APPLE__) // M1, issue #343 + __asm__ volatile ("mrs %0, tpidrro_el0\nbic %0, %0, #7" : "=r" (tcb)); + #else + __asm__ volatile ("mrs %0, tpidr_el0" : "=r" (tcb)); + #endif + res = tcb[slot]; + #endif + return res; +} + +// setting a tls slot is only used on macOS for now +static inline void mi_prim_tls_slot_set(size_t slot, void* value) mi_attr_noexcept { + const size_t ofs = (slot*sizeof(void*)); + #if defined(__i386__) + __asm__("movl %1,%%gs:%0" : "=m" (*((void**)ofs)) : "rn" (value) : ); // 32-bit always uses GS + #elif defined(__APPLE__) && defined(__x86_64__) + __asm__("movq %1,%%gs:%0" : "=m" (*((void**)ofs)) : "rn" (value) : ); // x86_64 macOS uses GS + #elif defined(__x86_64__) && (MI_INTPTR_SIZE==4) + __asm__("movl %1,%%fs:%0" : "=m" (*((void**)ofs)) : "rn" (value) : ); // x32 ABI + #elif defined(__x86_64__) + __asm__("movq %1,%%fs:%0" : "=m" (*((void**)ofs)) : "rn" (value) : ); // x86_64 Linux, BSD uses FS + #elif defined(__arm__) + void** tcb; MI_UNUSED(ofs); + __asm__ volatile ("mrc p15, 0, %0, c13, c0, 3\nbic %0, %0, #3" : "=r" (tcb)); + tcb[slot] = value; + #elif defined(__aarch64__) + void** tcb; MI_UNUSED(ofs); + #if defined(__APPLE__) // M1, issue #343 + __asm__ volatile ("mrs %0, tpidrro_el0\nbic %0, %0, #7" : "=r" (tcb)); + #else + __asm__ volatile ("mrs %0, tpidr_el0" : "=r" (tcb)); + #endif + tcb[slot] = value; + #endif +} + +static inline mi_threadid_t _mi_prim_thread_id(void) mi_attr_noexcept { + #if defined(__BIONIC__) + // issue #384, #495: on the Bionic libc (Android), slot 1 is the thread id + // see: https://github.com/aosp-mirror/platform_bionic/blob/c44b1d0676ded732df4b3b21c5f798eacae93228/libc/platform/bionic/tls_defines.h#L86 + return (uintptr_t)mi_prim_tls_slot(1); + #else + // in all our other targets, slot 0 is the thread id + // glibc: https://sourceware.org/git/?p=glibc.git;a=blob_plain;f=sysdeps/x86_64/nptl/tls.h + // apple: https://github.com/apple/darwin-xnu/blob/main/libsyscall/os/tsd.h#L36 + return (uintptr_t)mi_prim_tls_slot(0); + #endif +} + +#else + +// otherwise use portable C, taking the address of a thread local variable (this is still very fast on most platforms). +static inline mi_threadid_t _mi_prim_thread_id(void) mi_attr_noexcept { + return (uintptr_t)&_mi_heap_default; +} + +#endif + + + +/* ---------------------------------------------------------------------------------------- +The thread local default heap: `_mi_prim_get_default_heap()` +This is inlined here as it is on the fast path for allocation functions. + +On most platforms (Windows, Linux, FreeBSD, NetBSD, etc), this just returns a +__thread local variable (`_mi_heap_default`). With the initial-exec TLS model this ensures +that the storage will always be available (allocated on the thread stacks). + +On some platforms though we cannot use that when overriding `malloc` since the underlying +TLS implementation (or the loader) will call itself `malloc` on a first access and recurse. +We try to circumvent this in an efficient way: +- macOSX : we use an unused TLS slot from the OS allocated slots (MI_TLS_SLOT). On OSX, the + loader itself calls `malloc` even before the modules are initialized. +- OpenBSD: we use an unused slot from the pthread block (MI_TLS_PTHREAD_SLOT_OFS). +- DragonFly: defaults are working but seem slow compared to freeBSD (see PR #323) +------------------------------------------------------------------------------------------- */ + +static inline mi_heap_t* mi_prim_get_default_heap(void); + +#if defined(MI_MALLOC_OVERRIDE) +#if defined(__APPLE__) // macOS + #define MI_TLS_SLOT 89 // seems unused? + // #define MI_TLS_RECURSE_GUARD 1 + // other possible unused ones are 9, 29, __PTK_FRAMEWORK_JAVASCRIPTCORE_KEY4 (94), __PTK_FRAMEWORK_GC_KEY9 (112) and __PTK_FRAMEWORK_OLDGC_KEY9 (89) + // see +#elif defined(__OpenBSD__) + // use end bytes of a name; goes wrong if anyone uses names > 23 characters (ptrhread specifies 16) + // see + #define MI_TLS_PTHREAD_SLOT_OFS (6*sizeof(int) + 4*sizeof(void*) + 24) + // #elif defined(__DragonFly__) + // #warning "mimalloc is not working correctly on DragonFly yet." + // #define MI_TLS_PTHREAD_SLOT_OFS (4 + 1*sizeof(void*)) // offset `uniqueid` (also used by gdb?) +#elif defined(__ANDROID__) + // See issue #381 + #define MI_TLS_PTHREAD +#endif +#endif + + +#if defined(MI_TLS_SLOT) + +static inline mi_heap_t* mi_prim_get_default_heap(void) { + mi_heap_t* heap = (mi_heap_t*)mi_prim_tls_slot(MI_TLS_SLOT); + if mi_unlikely(heap == NULL) { + #ifdef __GNUC__ + __asm(""); // prevent conditional load of the address of _mi_heap_empty + #endif + heap = (mi_heap_t*)&_mi_heap_empty; + } + return heap; +} + +#elif defined(MI_TLS_PTHREAD_SLOT_OFS) + +static inline mi_heap_t** mi_prim_tls_pthread_heap_slot(void) { + pthread_t self = pthread_self(); + #if defined(__DragonFly__) + if (self==NULL) return NULL; + #endif + return (mi_heap_t**)((uint8_t*)self + MI_TLS_PTHREAD_SLOT_OFS); +} + +static inline mi_heap_t* mi_prim_get_default_heap(void) { + mi_heap_t** pheap = mi_prim_tls_pthread_heap_slot(); + if mi_unlikely(pheap == NULL) return _mi_heap_main_get(); + mi_heap_t* heap = *pheap; + if mi_unlikely(heap == NULL) return (mi_heap_t*)&_mi_heap_empty; + return heap; +} + +#elif defined(MI_TLS_PTHREAD) + +extern pthread_key_t _mi_heap_default_key; +static inline mi_heap_t* mi_prim_get_default_heap(void) { + mi_heap_t* heap = (mi_unlikely(_mi_heap_default_key == (pthread_key_t)(-1)) ? _mi_heap_main_get() : (mi_heap_t*)pthread_getspecific(_mi_heap_default_key)); + return (mi_unlikely(heap == NULL) ? (mi_heap_t*)&_mi_heap_empty : heap); +} + +#else // default using a thread local variable; used on most platforms. + +static inline mi_heap_t* mi_prim_get_default_heap(void) { + #if defined(MI_TLS_RECURSE_GUARD) + if (mi_unlikely(!_mi_process_is_initialized)) return _mi_heap_main_get(); + #endif + return _mi_heap_default; +} + +#endif // mi_prim_get_default_heap() + + + +#endif // MIMALLOC_PRIM_H diff --git a/compat/mimalloc/mimalloc/track.h b/compat/mimalloc/mimalloc/track.h new file mode 100644 index 00000000000000..fa1a048d846a9c --- /dev/null +++ b/compat/mimalloc/mimalloc/track.h @@ -0,0 +1,147 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#pragma once +#ifndef MIMALLOC_TRACK_H +#define MIMALLOC_TRACK_H + +/* ------------------------------------------------------------------------------------------------------ +Track memory ranges with macros for tools like Valgrind address sanitizer, or other memory checkers. +These can be defined for tracking allocation: + + #define mi_track_malloc_size(p,reqsize,size,zero) + #define mi_track_free_size(p,_size) + +The macros are set up such that the size passed to `mi_track_free_size` +always matches the size of `mi_track_malloc_size`. (currently, `size == mi_usable_size(p)`). +The `reqsize` is what the user requested, and `size >= reqsize`. +The `size` is either byte precise (and `size==reqsize`) if `MI_PADDING` is enabled, +or otherwise it is the usable block size which may be larger than the original request. +Use `_mi_block_size_of(void* p)` to get the full block size that was allocated (including padding etc). +The `zero` parameter is `true` if the allocated block is zero initialized. + +Optional: + + #define mi_track_align(p,alignedp,offset,size) + #define mi_track_resize(p,oldsize,newsize) + #define mi_track_init() + +The `mi_track_align` is called right after a `mi_track_malloc` for aligned pointers in a block. +The corresponding `mi_track_free` still uses the block start pointer and original size (corresponding to the `mi_track_malloc`). +The `mi_track_resize` is currently unused but could be called on reallocations within a block. +`mi_track_init` is called at program start. + +The following macros are for tools like asan and valgrind to track whether memory is +defined, undefined, or not accessible at all: + + #define mi_track_mem_defined(p,size) + #define mi_track_mem_undefined(p,size) + #define mi_track_mem_noaccess(p,size) + +-------------------------------------------------------------------------------------------------------*/ + +#if MI_TRACK_VALGRIND +// valgrind tool + +#define MI_TRACK_ENABLED 1 +#define MI_TRACK_HEAP_DESTROY 1 // track free of individual blocks on heap_destroy +#define MI_TRACK_TOOL "valgrind" + +#include +#include + +#define mi_track_malloc_size(p,reqsize,size,zero) VALGRIND_MALLOCLIKE_BLOCK(p,size,MI_PADDING_SIZE /*red zone*/,zero) +#define mi_track_free_size(p,_size) VALGRIND_FREELIKE_BLOCK(p,MI_PADDING_SIZE /*red zone*/) +#define mi_track_resize(p,oldsize,newsize) VALGRIND_RESIZEINPLACE_BLOCK(p,oldsize,newsize,MI_PADDING_SIZE /*red zone*/) +#define mi_track_mem_defined(p,size) VALGRIND_MAKE_MEM_DEFINED(p,size) +#define mi_track_mem_undefined(p,size) VALGRIND_MAKE_MEM_UNDEFINED(p,size) +#define mi_track_mem_noaccess(p,size) VALGRIND_MAKE_MEM_NOACCESS(p,size) + +#elif MI_TRACK_ASAN +// address sanitizer + +#define MI_TRACK_ENABLED 1 +#define MI_TRACK_HEAP_DESTROY 0 +#define MI_TRACK_TOOL "asan" + +#include + +#define mi_track_malloc_size(p,reqsize,size,zero) ASAN_UNPOISON_MEMORY_REGION(p,size) +#define mi_track_free_size(p,size) ASAN_POISON_MEMORY_REGION(p,size) +#define mi_track_mem_defined(p,size) ASAN_UNPOISON_MEMORY_REGION(p,size) +#define mi_track_mem_undefined(p,size) ASAN_UNPOISON_MEMORY_REGION(p,size) +#define mi_track_mem_noaccess(p,size) ASAN_POISON_MEMORY_REGION(p,size) + +#elif MI_TRACK_ETW +// windows event tracing + +#define MI_TRACK_ENABLED 1 +#define MI_TRACK_HEAP_DESTROY 1 +#define MI_TRACK_TOOL "ETW" + +#define WIN32_LEAN_AND_MEAN +#include +#include "../src/prim/windows/etw.h" + +#define mi_track_init() EventRegistermicrosoft_windows_mimalloc(); +#define mi_track_malloc_size(p,reqsize,size,zero) EventWriteETW_MI_ALLOC((UINT64)(p), size) +#define mi_track_free_size(p,size) EventWriteETW_MI_FREE((UINT64)(p), size) + +#else +// no tracking + +#define MI_TRACK_ENABLED 0 +#define MI_TRACK_HEAP_DESTROY 0 +#define MI_TRACK_TOOL "none" + +#define mi_track_malloc_size(p,reqsize,size,zero) +#define mi_track_free_size(p,_size) + +#endif + +// ------------------- +// Utility definitions + +#ifndef mi_track_resize +#define mi_track_resize(p,oldsize,newsize) mi_track_free_size(p,oldsize); mi_track_malloc(p,newsize,false) +#endif + +#ifndef mi_track_align +#define mi_track_align(p,alignedp,offset,size) mi_track_mem_noaccess(p,offset) +#endif + +#ifndef mi_track_init +#define mi_track_init() +#endif + +#ifndef mi_track_mem_defined +#define mi_track_mem_defined(p,size) +#endif + +#ifndef mi_track_mem_undefined +#define mi_track_mem_undefined(p,size) +#endif + +#ifndef mi_track_mem_noaccess +#define mi_track_mem_noaccess(p,size) +#endif + + +#if MI_PADDING +#define mi_track_malloc(p,reqsize,zero) \ + if ((p)!=NULL) { \ + mi_assert_internal(mi_usable_size(p)==(reqsize)); \ + mi_track_malloc_size(p,reqsize,reqsize,zero); \ + } +#else +#define mi_track_malloc(p,reqsize,zero) \ + if ((p)!=NULL) { \ + mi_assert_internal(mi_usable_size(p)>=(reqsize)); \ + mi_track_malloc_size(p,reqsize,mi_usable_size(p),zero); \ + } +#endif + +#endif diff --git a/compat/mimalloc/mimalloc/types.h b/compat/mimalloc/mimalloc/types.h new file mode 100644 index 00000000000000..7616f37e4b978f --- /dev/null +++ b/compat/mimalloc/mimalloc/types.h @@ -0,0 +1,670 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#pragma once +#ifndef MIMALLOC_TYPES_H +#define MIMALLOC_TYPES_H + +// -------------------------------------------------------------------------- +// This file contains the main type definitions for mimalloc: +// mi_heap_t : all data for a thread-local heap, contains +// lists of all managed heap pages. +// mi_segment_t : a larger chunk of memory (32GiB) from where pages +// are allocated. +// mi_page_t : a mimalloc page (usually 64KiB or 512KiB) from +// where objects are allocated. +// -------------------------------------------------------------------------- + + +#include // ptrdiff_t +#include // uintptr_t, uint16_t, etc +#include "mimalloc/atomic.h" // _Atomic + +#ifdef _MSC_VER +#pragma warning(disable:4214) // bitfield is not int +#endif + +// Minimal alignment necessary. On most platforms 16 bytes are needed +// due to SSE registers for example. This must be at least `sizeof(void*)` +#ifndef MI_MAX_ALIGN_SIZE +#define MI_MAX_ALIGN_SIZE 16 // sizeof(max_align_t) +#endif + +// ------------------------------------------------------ +// Variants +// ------------------------------------------------------ + +// Define NDEBUG in the release version to disable assertions. +// #define NDEBUG + +// Define MI_TRACK_ to enable tracking support +// #define MI_TRACK_VALGRIND 1 +// #define MI_TRACK_ASAN 1 +// #define MI_TRACK_ETW 1 + +// Define MI_STAT as 1 to maintain statistics; set it to 2 to have detailed statistics (but costs some performance). +// #define MI_STAT 1 + +// Define MI_SECURE to enable security mitigations +// #define MI_SECURE 1 // guard page around metadata +// #define MI_SECURE 2 // guard page around each mimalloc page +// #define MI_SECURE 3 // encode free lists (detect corrupted free list (buffer overflow), and invalid pointer free) +// #define MI_SECURE 4 // checks for double free. (may be more expensive) + +#if !defined(MI_SECURE) +#define MI_SECURE 0 +#endif + +// Define MI_DEBUG for debug mode +// #define MI_DEBUG 1 // basic assertion checks and statistics, check double free, corrupted free list, and invalid pointer free. +// #define MI_DEBUG 2 // + internal assertion checks +// #define MI_DEBUG 3 // + extensive internal invariant checking (cmake -DMI_DEBUG_FULL=ON) +#if !defined(MI_DEBUG) +#if !defined(NDEBUG) || defined(_DEBUG) +#define MI_DEBUG 2 +#else +#define MI_DEBUG 0 +#endif +#endif + +// Reserve extra padding at the end of each block to be more resilient against heap block overflows. +// The padding can detect buffer overflow on free. +#if !defined(MI_PADDING) && (MI_SECURE>=3 || MI_DEBUG>=1 || (MI_TRACK_VALGRIND || MI_TRACK_ASAN || MI_TRACK_ETW)) +#define MI_PADDING 1 +#endif + +// Check padding bytes; allows byte-precise buffer overflow detection +#if !defined(MI_PADDING_CHECK) && MI_PADDING && (MI_SECURE>=3 || MI_DEBUG>=1) +#define MI_PADDING_CHECK 1 +#endif + + +// Encoded free lists allow detection of corrupted free lists +// and can detect buffer overflows, modify after free, and double `free`s. +#if (MI_SECURE>=3 || MI_DEBUG>=1) +#define MI_ENCODE_FREELIST 1 +#endif + + +// We used to abandon huge pages but to eagerly deallocate if freed from another thread, +// but that makes it not possible to visit them during a heap walk or include them in a +// `mi_heap_destroy`. We therefore instead reset/decommit the huge blocks if freed from +// another thread so most memory is available until it gets properly freed by the owning thread. +// #define MI_HUGE_PAGE_ABANDON 1 + + +// ------------------------------------------------------ +// Platform specific values +// ------------------------------------------------------ + +// ------------------------------------------------------ +// Size of a pointer. +// We assume that `sizeof(void*)==sizeof(intptr_t)` +// and it holds for all platforms we know of. +// +// However, the C standard only requires that: +// p == (void*)((intptr_t)p)) +// but we also need: +// i == (intptr_t)((void*)i) +// or otherwise one might define an intptr_t type that is larger than a pointer... +// ------------------------------------------------------ + +#if INTPTR_MAX > INT64_MAX +# define MI_INTPTR_SHIFT (4) // assume 128-bit (as on arm CHERI for example) +#elif INTPTR_MAX == INT64_MAX +# define MI_INTPTR_SHIFT (3) +#elif INTPTR_MAX == INT32_MAX +# define MI_INTPTR_SHIFT (2) +#else +#error platform pointers must be 32, 64, or 128 bits +#endif + +#if SIZE_MAX == UINT64_MAX +# define MI_SIZE_SHIFT (3) +typedef int64_t mi_ssize_t; +#elif SIZE_MAX == UINT32_MAX +# define MI_SIZE_SHIFT (2) +typedef int32_t mi_ssize_t; +#else +#error platform objects must be 32 or 64 bits +#endif + +#if (SIZE_MAX/2) > LONG_MAX +# define MI_ZU(x) x##ULL +# define MI_ZI(x) x##LL +#else +# define MI_ZU(x) x##UL +# define MI_ZI(x) x##L +#endif + +#define MI_INTPTR_SIZE (1< 4 +#define MI_SEGMENT_SHIFT ( 9 + MI_SEGMENT_SLICE_SHIFT) // 32MiB +#else +#define MI_SEGMENT_SHIFT ( 7 + MI_SEGMENT_SLICE_SHIFT) // 4MiB on 32-bit +#endif + +#define MI_SMALL_PAGE_SHIFT (MI_SEGMENT_SLICE_SHIFT) // 64KiB +#define MI_MEDIUM_PAGE_SHIFT ( 3 + MI_SMALL_PAGE_SHIFT) // 512KiB + + +// Derived constants +#define MI_SEGMENT_SIZE (MI_ZU(1)<= 655360) +#error "mimalloc internal: define more bins" +#endif + +// Maximum slice offset (15) +#define MI_MAX_SLICE_OFFSET ((MI_ALIGNMENT_MAX / MI_SEGMENT_SLICE_SIZE) - 1) + +// Used as a special value to encode block sizes in 32 bits. +#define MI_HUGE_BLOCK_SIZE ((uint32_t)(2*MI_GiB)) + +// blocks up to this size are always allocated aligned +#define MI_MAX_ALIGN_GUARANTEE (8*MI_MAX_ALIGN_SIZE) + +// Alignments over MI_ALIGNMENT_MAX are allocated in dedicated huge page segments +#define MI_ALIGNMENT_MAX (MI_SEGMENT_SIZE >> 1) + + +// ------------------------------------------------------ +// Mimalloc pages contain allocated blocks +// ------------------------------------------------------ + +// The free lists use encoded next fields +// (Only actually encodes when MI_ENCODED_FREELIST is defined.) +typedef uintptr_t mi_encoded_t; + +// thread id's +typedef size_t mi_threadid_t; + +// free lists contain blocks +typedef struct mi_block_s { + mi_encoded_t next; +} mi_block_t; + + +// The delayed flags are used for efficient multi-threaded free-ing +typedef enum mi_delayed_e { + MI_USE_DELAYED_FREE = 0, // push on the owning heap thread delayed list + MI_DELAYED_FREEING = 1, // temporary: another thread is accessing the owning heap + MI_NO_DELAYED_FREE = 2, // optimize: push on page local thread free queue if another block is already in the heap thread delayed free list + MI_NEVER_DELAYED_FREE = 3 // sticky, only resets on page reclaim +} mi_delayed_t; + + +// The `in_full` and `has_aligned` page flags are put in a union to efficiently +// test if both are false (`full_aligned == 0`) in the `mi_free` routine. +#if !MI_TSAN +typedef union mi_page_flags_s { + uint8_t full_aligned; + struct { + uint8_t in_full : 1; + uint8_t has_aligned : 1; + } x; +} mi_page_flags_t; +#else +// under thread sanitizer, use a byte for each flag to suppress warning, issue #130 +typedef union mi_page_flags_s { + uint16_t full_aligned; + struct { + uint8_t in_full; + uint8_t has_aligned; + } x; +} mi_page_flags_t; +#endif + +// Thread free list. +// We use the bottom 2 bits of the pointer for mi_delayed_t flags +typedef uintptr_t mi_thread_free_t; + +// A page contains blocks of one specific size (`block_size`). +// Each page has three list of free blocks: +// `free` for blocks that can be allocated, +// `local_free` for freed blocks that are not yet available to `mi_malloc` +// `thread_free` for freed blocks by other threads +// The `local_free` and `thread_free` lists are migrated to the `free` list +// when it is exhausted. The separate `local_free` list is necessary to +// implement a monotonic heartbeat. The `thread_free` list is needed for +// avoiding atomic operations in the common case. +// +// +// `used - |thread_free|` == actual blocks that are in use (alive) +// `used - |thread_free| + |free| + |local_free| == capacity` +// +// We don't count `freed` (as |free|) but use `used` to reduce +// the number of memory accesses in the `mi_page_all_free` function(s). +// +// Notes: +// - Access is optimized for `mi_free` and `mi_page_alloc` (in `alloc.c`) +// - Using `uint16_t` does not seem to slow things down +// - The size is 8 words on 64-bit which helps the page index calculations +// (and 10 words on 32-bit, and encoded free lists add 2 words. Sizes 10 +// and 12 are still good for address calculation) +// - To limit the structure size, the `xblock_size` is 32-bits only; for +// blocks > MI_HUGE_BLOCK_SIZE the size is determined from the segment page size +// - `thread_free` uses the bottom bits as a delayed-free flags to optimize +// concurrent frees where only the first concurrent free adds to the owning +// heap `thread_delayed_free` list (see `alloc.c:mi_free_block_mt`). +// The invariant is that no-delayed-free is only set if there is +// at least one block that will be added, or as already been added, to +// the owning heap `thread_delayed_free` list. This guarantees that pages +// will be freed correctly even if only other threads free blocks. +typedef struct mi_page_s { + // "owned" by the segment + uint32_t slice_count; // slices in this page (0 if not a page) + uint32_t slice_offset; // distance from the actual page data slice (0 if a page) + uint8_t is_committed : 1; // `true` if the page virtual memory is committed + uint8_t is_zero_init : 1; // `true` if the page was initially zero initialized + + // layout like this to optimize access in `mi_malloc` and `mi_free` + uint16_t capacity; // number of blocks committed, must be the first field, see `segment.c:page_clear` + uint16_t reserved; // number of blocks reserved in memory + mi_page_flags_t flags; // `in_full` and `has_aligned` flags (8 bits) + uint8_t free_is_zero : 1; // `true` if the blocks in the free list are zero initialized + uint8_t retire_expire : 7; // expiration count for retired blocks + + mi_block_t* free; // list of available free blocks (`malloc` allocates from this list) + uint32_t used; // number of blocks in use (including blocks in `local_free` and `thread_free`) + uint32_t xblock_size; // size available in each block (always `>0`) + mi_block_t* local_free; // list of deferred free blocks by this thread (migrates to `free`) + + #if (MI_ENCODE_FREELIST || MI_PADDING) + uintptr_t keys[2]; // two random keys to encode the free lists (see `_mi_block_next`) or padding canary + #endif + + _Atomic(mi_thread_free_t) xthread_free; // list of deferred free blocks freed by other threads + _Atomic(uintptr_t) xheap; + + struct mi_page_s* next; // next page owned by this thread with the same `block_size` + struct mi_page_s* prev; // previous page owned by this thread with the same `block_size` + + // 64-bit 9 words, 32-bit 12 words, (+2 for secure) + #if MI_INTPTR_SIZE==8 + uintptr_t padding[1]; + #endif +} mi_page_t; + + + +// ------------------------------------------------------ +// Mimalloc segments contain mimalloc pages +// ------------------------------------------------------ + +typedef enum mi_page_kind_e { + MI_PAGE_SMALL, // small blocks go into 64KiB pages inside a segment + MI_PAGE_MEDIUM, // medium blocks go into medium pages inside a segment + MI_PAGE_LARGE, // larger blocks go into a page of just one block + MI_PAGE_HUGE, // huge blocks (> 16 MiB) are put into a single page in a single segment. +} mi_page_kind_t; + +typedef enum mi_segment_kind_e { + MI_SEGMENT_NORMAL, // MI_SEGMENT_SIZE size with pages inside. + MI_SEGMENT_HUGE, // > MI_LARGE_SIZE_MAX segment with just one huge page inside. +} mi_segment_kind_t; + +// ------------------------------------------------------ +// A segment holds a commit mask where a bit is set if +// the corresponding MI_COMMIT_SIZE area is committed. +// The MI_COMMIT_SIZE must be a multiple of the slice +// size. If it is equal we have the most fine grained +// decommit (but setting it higher can be more efficient). +// The MI_MINIMAL_COMMIT_SIZE is the minimal amount that will +// be committed in one go which can be set higher than +// MI_COMMIT_SIZE for efficiency (while the decommit mask +// is still tracked in fine-grained MI_COMMIT_SIZE chunks) +// ------------------------------------------------------ + +#define MI_MINIMAL_COMMIT_SIZE (1*MI_SEGMENT_SLICE_SIZE) +#define MI_COMMIT_SIZE (MI_SEGMENT_SLICE_SIZE) // 64KiB +#define MI_COMMIT_MASK_BITS (MI_SEGMENT_SIZE / MI_COMMIT_SIZE) +#define MI_COMMIT_MASK_FIELD_BITS MI_SIZE_BITS +#define MI_COMMIT_MASK_FIELD_COUNT (MI_COMMIT_MASK_BITS / MI_COMMIT_MASK_FIELD_BITS) + +#if (MI_COMMIT_MASK_BITS != (MI_COMMIT_MASK_FIELD_COUNT * MI_COMMIT_MASK_FIELD_BITS)) +#error "the segment size must be exactly divisible by the (commit size * size_t bits)" +#endif + +typedef struct mi_commit_mask_s { + size_t mask[MI_COMMIT_MASK_FIELD_COUNT]; +} mi_commit_mask_t; + +typedef mi_page_t mi_slice_t; +typedef int64_t mi_msecs_t; + + +// Memory can reside in arena's, direct OS allocated, or statically allocated. The memid keeps track of this. +typedef enum mi_memkind_e { + MI_MEM_NONE, // not allocated + MI_MEM_EXTERNAL, // not owned by mimalloc but provided externally (via `mi_manage_os_memory` for example) + MI_MEM_STATIC, // allocated in a static area and should not be freed (for arena meta data for example) + MI_MEM_OS, // allocated from the OS + MI_MEM_OS_HUGE, // allocated as huge os pages + MI_MEM_OS_REMAP, // allocated in a remapable area (i.e. using `mremap`) + MI_MEM_ARENA // allocated from an arena (the usual case) +} mi_memkind_t; + +static inline bool mi_memkind_is_os(mi_memkind_t memkind) { + return (memkind >= MI_MEM_OS && memkind <= MI_MEM_OS_REMAP); +} + +typedef struct mi_memid_os_info { + void* base; // actual base address of the block (used for offset aligned allocations) + size_t alignment; // alignment at allocation +} mi_memid_os_info_t; + +typedef struct mi_memid_arena_info { + size_t block_index; // index in the arena + mi_arena_id_t id; // arena id (>= 1) + bool is_exclusive; // the arena can only be used for specific arena allocations +} mi_memid_arena_info_t; + +typedef struct mi_memid_s { + union { + mi_memid_os_info_t os; // only used for MI_MEM_OS + mi_memid_arena_info_t arena; // only used for MI_MEM_ARENA + } mem; + bool is_pinned; // `true` if we cannot decommit/reset/protect in this memory (e.g. when allocated using large OS pages) + bool initially_committed;// `true` if the memory was originally allocated as committed + bool initially_zero; // `true` if the memory was originally zero initialized + mi_memkind_t memkind; +} mi_memid_t; + + +// Segments are large allocated memory blocks (8mb on 64 bit) from +// the OS. Inside segments we allocated fixed size _pages_ that +// contain blocks. +typedef struct mi_segment_s { + // constant fields + mi_memid_t memid; // memory id for arena allocation + bool allow_decommit; + bool allow_purge; + size_t segment_size; + + // segment fields + mi_msecs_t purge_expire; + mi_commit_mask_t purge_mask; + mi_commit_mask_t commit_mask; + + _Atomic(struct mi_segment_s*) abandoned_next; + + // from here is zero initialized + struct mi_segment_s* next; // the list of freed segments in the cache (must be first field, see `segment.c:mi_segment_init`) + + size_t abandoned; // abandoned pages (i.e. the original owning thread stopped) (`abandoned <= used`) + size_t abandoned_visits; // count how often this segment is visited in the abandoned list (to force reclaim it it is too long) + size_t used; // count of pages in use + uintptr_t cookie; // verify addresses in debug mode: `mi_ptr_cookie(segment) == segment->cookie` + + size_t segment_slices; // for huge segments this may be different from `MI_SLICES_PER_SEGMENT` + size_t segment_info_slices; // initial slices we are using segment info and possible guard pages. + + // layout like this to optimize access in `mi_free` + mi_segment_kind_t kind; + size_t slice_entries; // entries in the `slices` array, at most `MI_SLICES_PER_SEGMENT` + _Atomic(mi_threadid_t) thread_id; // unique id of the thread owning this segment + + mi_slice_t slices[MI_SLICES_PER_SEGMENT+1]; // one more for huge blocks with large alignment +} mi_segment_t; + + +// ------------------------------------------------------ +// Heaps +// Provide first-class heaps to allocate from. +// A heap just owns a set of pages for allocation and +// can only be allocate/reallocate from the thread that created it. +// Freeing blocks can be done from any thread though. +// Per thread, the segments are shared among its heaps. +// Per thread, there is always a default heap that is +// used for allocation; it is initialized to statically +// point to an empty heap to avoid initialization checks +// in the fast path. +// ------------------------------------------------------ + +// Thread local data +typedef struct mi_tld_s mi_tld_t; + +// Pages of a certain block size are held in a queue. +typedef struct mi_page_queue_s { + mi_page_t* first; + mi_page_t* last; + size_t block_size; +} mi_page_queue_t; + +#define MI_BIN_FULL (MI_BIN_HUGE+1) + +// Random context +typedef struct mi_random_cxt_s { + uint32_t input[16]; + uint32_t output[16]; + int output_available; + bool weak; +} mi_random_ctx_t; + + +// In debug mode there is a padding structure at the end of the blocks to check for buffer overflows +#if (MI_PADDING) +typedef struct mi_padding_s { + uint32_t canary; // encoded block value to check validity of the padding (in case of overflow) + uint32_t delta; // padding bytes before the block. (mi_usable_size(p) - delta == exact allocated bytes) +} mi_padding_t; +#define MI_PADDING_SIZE (sizeof(mi_padding_t)) +#define MI_PADDING_WSIZE ((MI_PADDING_SIZE + MI_INTPTR_SIZE - 1) / MI_INTPTR_SIZE) +#else +#define MI_PADDING_SIZE 0 +#define MI_PADDING_WSIZE 0 +#endif + +#define MI_PAGES_DIRECT (MI_SMALL_WSIZE_MAX + MI_PADDING_WSIZE + 1) + + +// A heap owns a set of pages. +struct mi_heap_s { + mi_tld_t* tld; + mi_page_t* pages_free_direct[MI_PAGES_DIRECT]; // optimize: array where every entry points a page with possibly free blocks in the corresponding queue for that size. + mi_page_queue_t pages[MI_BIN_FULL + 1]; // queue of pages for each size class (or "bin") + _Atomic(mi_block_t*) thread_delayed_free; + mi_threadid_t thread_id; // thread this heap belongs too + mi_arena_id_t arena_id; // arena id if the heap belongs to a specific arena (or 0) + uintptr_t cookie; // random cookie to verify pointers (see `_mi_ptr_cookie`) + uintptr_t keys[2]; // two random keys used to encode the `thread_delayed_free` list + mi_random_ctx_t random; // random number context used for secure allocation + size_t page_count; // total number of pages in the `pages` queues. + size_t page_retired_min; // smallest retired index (retired pages are fully free, but still in the page queues) + size_t page_retired_max; // largest retired index into the `pages` array. + mi_heap_t* next; // list of heaps per thread + bool no_reclaim; // `true` if this heap should not reclaim abandoned pages +}; + + + +// ------------------------------------------------------ +// Debug +// ------------------------------------------------------ + +#if !defined(MI_DEBUG_UNINIT) +#define MI_DEBUG_UNINIT (0xD0) +#endif +#if !defined(MI_DEBUG_FREED) +#define MI_DEBUG_FREED (0xDF) +#endif +#if !defined(MI_DEBUG_PADDING) +#define MI_DEBUG_PADDING (0xDE) +#endif + +#if (MI_DEBUG) +// use our own assertion to print without memory allocation +void _mi_assert_fail(const char* assertion, const char* fname, unsigned int line, const char* func ); +#define mi_assert(expr) ((expr) ? (void)0 : _mi_assert_fail(#expr,__FILE__,__LINE__,__func__)) +#else +#define mi_assert(x) +#endif + +#if (MI_DEBUG>1) +#define mi_assert_internal mi_assert +#else +#define mi_assert_internal(x) +#endif + +#if (MI_DEBUG>2) +#define mi_assert_expensive mi_assert +#else +#define mi_assert_expensive(x) +#endif + +// ------------------------------------------------------ +// Statistics +// ------------------------------------------------------ + +#ifndef MI_STAT +#if (MI_DEBUG>0) +#define MI_STAT 2 +#else +#define MI_STAT 0 +#endif +#endif + +typedef struct mi_stat_count_s { + int64_t allocated; + int64_t freed; + int64_t peak; + int64_t current; +} mi_stat_count_t; + +typedef struct mi_stat_counter_s { + int64_t total; + int64_t count; +} mi_stat_counter_t; + +typedef struct mi_stats_s { + mi_stat_count_t segments; + mi_stat_count_t pages; + mi_stat_count_t reserved; + mi_stat_count_t committed; + mi_stat_count_t reset; + mi_stat_count_t purged; + mi_stat_count_t page_committed; + mi_stat_count_t segments_abandoned; + mi_stat_count_t pages_abandoned; + mi_stat_count_t threads; + mi_stat_count_t normal; + mi_stat_count_t huge; + mi_stat_count_t large; + mi_stat_count_t malloc; + mi_stat_count_t segments_cache; + mi_stat_counter_t pages_extended; + mi_stat_counter_t mmap_calls; + mi_stat_counter_t commit_calls; + mi_stat_counter_t reset_calls; + mi_stat_counter_t purge_calls; + mi_stat_counter_t page_no_retire; + mi_stat_counter_t searches; + mi_stat_counter_t normal_count; + mi_stat_counter_t huge_count; + mi_stat_counter_t large_count; +#if MI_STAT>1 + mi_stat_count_t normal_bins[MI_BIN_HUGE+1]; +#endif +} mi_stats_t; + + +void _mi_stat_increase(mi_stat_count_t* stat, size_t amount); +void _mi_stat_decrease(mi_stat_count_t* stat, size_t amount); +void _mi_stat_counter_increase(mi_stat_counter_t* stat, size_t amount); + +#if (MI_STAT) +#define mi_stat_increase(stat,amount) _mi_stat_increase( &(stat), amount) +#define mi_stat_decrease(stat,amount) _mi_stat_decrease( &(stat), amount) +#define mi_stat_counter_increase(stat,amount) _mi_stat_counter_increase( &(stat), amount) +#else +#define mi_stat_increase(stat,amount) (void)0 +#define mi_stat_decrease(stat,amount) (void)0 +#define mi_stat_counter_increase(stat,amount) (void)0 +#endif + +#define mi_heap_stat_counter_increase(heap,stat,amount) mi_stat_counter_increase( (heap)->tld->stats.stat, amount) +#define mi_heap_stat_increase(heap,stat,amount) mi_stat_increase( (heap)->tld->stats.stat, amount) +#define mi_heap_stat_decrease(heap,stat,amount) mi_stat_decrease( (heap)->tld->stats.stat, amount) + +// ------------------------------------------------------ +// Thread Local data +// ------------------------------------------------------ + +// A "span" is is an available range of slices. The span queues keep +// track of slice spans of at most the given `slice_count` (but more than the previous size class). +typedef struct mi_span_queue_s { + mi_slice_t* first; + mi_slice_t* last; + size_t slice_count; +} mi_span_queue_t; + +#define MI_SEGMENT_BIN_MAX (35) // 35 == mi_segment_bin(MI_SLICES_PER_SEGMENT) + +// OS thread local data +typedef struct mi_os_tld_s { + size_t region_idx; // start point for next allocation + mi_stats_t* stats; // points to tld stats +} mi_os_tld_t; + + +// Segments thread local data +typedef struct mi_segments_tld_s { + mi_span_queue_t spans[MI_SEGMENT_BIN_MAX+1]; // free slice spans inside segments + size_t count; // current number of segments; + size_t peak_count; // peak number of segments + size_t current_size; // current size of all segments + size_t peak_size; // peak size of all segments + mi_stats_t* stats; // points to tld stats + mi_os_tld_t* os; // points to os stats +} mi_segments_tld_t; + +// Thread local data +struct mi_tld_s { + unsigned long long heartbeat; // monotonic heartbeat count + bool recurse; // true if deferred was called; used to prevent infinite recursion. + mi_heap_t* heap_backing; // backing heap of this thread (cannot be deleted) + mi_heap_t* heaps; // list of heaps in this thread (so we can abandon all when the thread terminates) + mi_segments_tld_t segments; // segment tld + mi_os_tld_t os; // os tld + mi_stats_t stats; // statistics +}; + +#endif diff --git a/compat/mimalloc/options.c b/compat/mimalloc/options.c new file mode 100644 index 00000000000000..3a3090d9acfc94 --- /dev/null +++ b/compat/mimalloc/options.c @@ -0,0 +1,571 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2021, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" +#include "mimalloc/prim.h" // mi_prim_out_stderr + +#include // FILE +#include // abort +#include + + +static long mi_max_error_count = 16; // stop outputting errors after this (use < 0 for no limit) +static long mi_max_warning_count = 16; // stop outputting warnings after this (use < 0 for no limit) + +static void mi_add_stderr_output(void); + +int mi_version(void) mi_attr_noexcept { + return MI_MALLOC_VERSION; +} + + +// -------------------------------------------------------- +// Options +// These can be accessed by multiple threads and may be +// concurrently initialized, but an initializing data race +// is ok since they resolve to the same value. +// -------------------------------------------------------- +typedef enum mi_init_e { + UNINIT, // not yet initialized + DEFAULTED, // not found in the environment, use default value + INITIALIZED // found in environment or set explicitly +} mi_init_t; + +typedef struct mi_option_desc_s { + long value; // the value + mi_init_t init; // is it initialized yet? (from the environment) + mi_option_t option; // for debugging: the option index should match the option + const char* name; // option name without `mimalloc_` prefix + const char* legacy_name; // potential legacy option name +} mi_option_desc_t; + +#define MI_OPTION(opt) mi_option_##opt, #opt, NULL +#define MI_OPTION_LEGACY(opt,legacy) mi_option_##opt, #opt, #legacy + +static mi_option_desc_t options[_mi_option_last] = +{ + // stable options + #if MI_DEBUG || defined(MI_SHOW_ERRORS) + { 1, UNINIT, MI_OPTION(show_errors) }, + #else + { 0, UNINIT, MI_OPTION(show_errors) }, + #endif + { 0, UNINIT, MI_OPTION(show_stats) }, + { 0, UNINIT, MI_OPTION(verbose) }, + + // the following options are experimental and not all combinations make sense. + { 1, UNINIT, MI_OPTION(eager_commit) }, // commit per segment directly (4MiB) (but see also `eager_commit_delay`) + { 2, UNINIT, MI_OPTION_LEGACY(arena_eager_commit,eager_region_commit) }, // eager commit arena's? 2 is used to enable this only on an OS that has overcommit (i.e. linux) + { 1, UNINIT, MI_OPTION_LEGACY(purge_decommits,reset_decommits) }, // purge decommits memory (instead of reset) (note: on linux this uses MADV_DONTNEED for decommit) + { 0, UNINIT, MI_OPTION_LEGACY(allow_large_os_pages,large_os_pages) }, // use large OS pages, use only with eager commit to prevent fragmentation of VMA's + { 0, UNINIT, MI_OPTION(reserve_huge_os_pages) }, // per 1GiB huge pages + {-1, UNINIT, MI_OPTION(reserve_huge_os_pages_at) }, // reserve huge pages at node N + { 0, UNINIT, MI_OPTION(reserve_os_memory) }, + { 0, UNINIT, MI_OPTION(deprecated_segment_cache) }, // cache N segments per thread + { 0, UNINIT, MI_OPTION(deprecated_page_reset) }, // reset page memory on free + { 0, UNINIT, MI_OPTION_LEGACY(abandoned_page_purge,abandoned_page_reset) }, // reset free page memory when a thread terminates + { 0, UNINIT, MI_OPTION(deprecated_segment_reset) }, // reset segment memory on free (needs eager commit) +#if defined(__NetBSD__) + { 0, UNINIT, MI_OPTION(eager_commit_delay) }, // the first N segments per thread are not eagerly committed +#else + { 1, UNINIT, MI_OPTION(eager_commit_delay) }, // the first N segments per thread are not eagerly committed (but per page in the segment on demand) +#endif + { 10, UNINIT, MI_OPTION_LEGACY(purge_delay,reset_delay) }, // purge delay in milli-seconds + { 0, UNINIT, MI_OPTION(use_numa_nodes) }, // 0 = use available numa nodes, otherwise use at most N nodes. + { 0, UNINIT, MI_OPTION(limit_os_alloc) }, // 1 = do not use OS memory for allocation (but only reserved arenas) + { 100, UNINIT, MI_OPTION(os_tag) }, // only apple specific for now but might serve more or less related purpose + { 16, UNINIT, MI_OPTION(max_errors) }, // maximum errors that are output + { 16, UNINIT, MI_OPTION(max_warnings) }, // maximum warnings that are output + { 8, UNINIT, MI_OPTION(max_segment_reclaim)}, // max. number of segment reclaims from the abandoned segments per try. + { 0, UNINIT, MI_OPTION(destroy_on_exit)}, // release all OS memory on process exit; careful with dangling pointer or after-exit frees! + #if (MI_INTPTR_SIZE>4) + { 1024L * 1024L, UNINIT, MI_OPTION(arena_reserve) }, // reserve memory N KiB at a time + #else + { 128L * 1024L, UNINIT, MI_OPTION(arena_reserve) }, + #endif + { 10, UNINIT, MI_OPTION(arena_purge_mult) }, // purge delay multiplier for arena's + { 1, UNINIT, MI_OPTION_LEGACY(purge_extend_delay, decommit_extend_delay) }, +}; + +static void mi_option_init(mi_option_desc_t* desc); + +void _mi_options_init(void) { + // called on process load; should not be called before the CRT is initialized! + // (e.g. do not call this from process_init as that may run before CRT initialization) + mi_add_stderr_output(); // now it safe to use stderr for output + for(int i = 0; i < _mi_option_last; i++ ) { + mi_option_t option = (mi_option_t)i; + long l = mi_option_get(option); MI_UNUSED(l); // initialize + // if (option != mi_option_verbose) + { + mi_option_desc_t* desc = &options[option]; + _mi_verbose_message("option '%s': %ld\n", desc->name, desc->value); + } + } + mi_max_error_count = mi_option_get(mi_option_max_errors); + mi_max_warning_count = mi_option_get(mi_option_max_warnings); +} + +mi_decl_nodiscard long mi_option_get(mi_option_t option) { + mi_assert(option >= 0 && option < _mi_option_last); + if (option < 0 || option >= _mi_option_last) return 0; + mi_option_desc_t* desc = &options[option]; + mi_assert(desc->option == option); // index should match the option + if mi_unlikely(desc->init == UNINIT) { + mi_option_init(desc); + } + return desc->value; +} + +mi_decl_nodiscard long mi_option_get_clamp(mi_option_t option, long min, long max) { + long x = mi_option_get(option); + return (x < min ? min : (x > max ? max : x)); +} + +mi_decl_nodiscard size_t mi_option_get_size(mi_option_t option) { + mi_assert_internal(option == mi_option_reserve_os_memory || option == mi_option_arena_reserve); + long x = mi_option_get(option); + return (x < 0 ? 0 : (size_t)x * MI_KiB); +} + +void mi_option_set(mi_option_t option, long value) { + mi_assert(option >= 0 && option < _mi_option_last); + if (option < 0 || option >= _mi_option_last) return; + mi_option_desc_t* desc = &options[option]; + mi_assert(desc->option == option); // index should match the option + desc->value = value; + desc->init = INITIALIZED; +} + +void mi_option_set_default(mi_option_t option, long value) { + mi_assert(option >= 0 && option < _mi_option_last); + if (option < 0 || option >= _mi_option_last) return; + mi_option_desc_t* desc = &options[option]; + if (desc->init != INITIALIZED) { + desc->value = value; + } +} + +mi_decl_nodiscard bool mi_option_is_enabled(mi_option_t option) { + return (mi_option_get(option) != 0); +} + +void mi_option_set_enabled(mi_option_t option, bool enable) { + mi_option_set(option, (enable ? 1 : 0)); +} + +void mi_option_set_enabled_default(mi_option_t option, bool enable) { + mi_option_set_default(option, (enable ? 1 : 0)); +} + +void mi_option_enable(mi_option_t option) { + mi_option_set_enabled(option,true); +} + +void mi_option_disable(mi_option_t option) { + mi_option_set_enabled(option,false); +} + +static void mi_cdecl mi_out_stderr(const char* msg, void* arg) { + MI_UNUSED(arg); + if (msg != NULL && msg[0] != 0) { + _mi_prim_out_stderr(msg); + } +} + +// Since an output function can be registered earliest in the `main` +// function we also buffer output that happens earlier. When +// an output function is registered it is called immediately with +// the output up to that point. +#ifndef MI_MAX_DELAY_OUTPUT +#define MI_MAX_DELAY_OUTPUT ((size_t)(32*1024)) +#endif +static char out_buf[MI_MAX_DELAY_OUTPUT+1]; +static _Atomic(size_t) out_len; + +static void mi_cdecl mi_out_buf(const char* msg, void* arg) { + MI_UNUSED(arg); + if (msg==NULL) return; + if (mi_atomic_load_relaxed(&out_len)>=MI_MAX_DELAY_OUTPUT) return; + size_t n = _mi_strlen(msg); + if (n==0) return; + // claim space + size_t start = mi_atomic_add_acq_rel(&out_len, n); + if (start >= MI_MAX_DELAY_OUTPUT) return; + // check bound + if (start+n >= MI_MAX_DELAY_OUTPUT) { + n = MI_MAX_DELAY_OUTPUT-start-1; + } + _mi_memcpy(&out_buf[start], msg, n); +} + +static void mi_out_buf_flush(mi_output_fun* out, bool no_more_buf, void* arg) { + if (out==NULL) return; + // claim (if `no_more_buf == true`, no more output will be added after this point) + size_t count = mi_atomic_add_acq_rel(&out_len, (no_more_buf ? MI_MAX_DELAY_OUTPUT : 1)); + // and output the current contents + if (count>MI_MAX_DELAY_OUTPUT) count = MI_MAX_DELAY_OUTPUT; + out_buf[count] = 0; + out(out_buf,arg); + if (!no_more_buf) { + out_buf[count] = '\n'; // if continue with the buffer, insert a newline + } +} + + +// Once this module is loaded, switch to this routine +// which outputs to stderr and the delayed output buffer. +static void mi_cdecl mi_out_buf_stderr(const char* msg, void* arg) { + mi_out_stderr(msg,arg); + mi_out_buf(msg,arg); +} + + + +// -------------------------------------------------------- +// Default output handler +// -------------------------------------------------------- + +// Should be atomic but gives errors on many platforms as generally we cannot cast a function pointer to a uintptr_t. +// For now, don't register output from multiple threads. +static mi_output_fun* volatile mi_out_default; // = NULL +static _Atomic(void*) mi_out_arg; // = NULL + +static mi_output_fun* mi_out_get_default(void** parg) { + if (parg != NULL) { *parg = mi_atomic_load_ptr_acquire(void,&mi_out_arg); } + mi_output_fun* out = mi_out_default; + return (out == NULL ? &mi_out_buf : out); +} + +void mi_register_output(mi_output_fun* out, void* arg) mi_attr_noexcept { + mi_out_default = (out == NULL ? &mi_out_stderr : out); // stop using the delayed output buffer + mi_atomic_store_ptr_release(void,&mi_out_arg, arg); + if (out!=NULL) mi_out_buf_flush(out,true,arg); // output all the delayed output now +} + +// add stderr to the delayed output after the module is loaded +static void mi_add_stderr_output(void) { + mi_assert_internal(mi_out_default == NULL); + mi_out_buf_flush(&mi_out_stderr, false, NULL); // flush current contents to stderr + mi_out_default = &mi_out_buf_stderr; // and add stderr to the delayed output +} + +// -------------------------------------------------------- +// Messages, all end up calling `_mi_fputs`. +// -------------------------------------------------------- +static _Atomic(size_t) error_count; // = 0; // when >= max_error_count stop emitting errors +static _Atomic(size_t) warning_count; // = 0; // when >= max_warning_count stop emitting warnings + +// When overriding malloc, we may recurse into mi_vfprintf if an allocation +// inside the C runtime causes another message. +// In some cases (like on macOS) the loader already allocates which +// calls into mimalloc; if we then access thread locals (like `recurse`) +// this may crash as the access may call _tlv_bootstrap that tries to +// (recursively) invoke malloc again to allocate space for the thread local +// variables on demand. This is why we use a _mi_preloading test on such +// platforms. However, C code generator may move the initial thread local address +// load before the `if` and we therefore split it out in a separate funcion. +static mi_decl_thread bool recurse = false; + +static mi_decl_noinline bool mi_recurse_enter_prim(void) { + if (recurse) return false; + recurse = true; + return true; +} + +static mi_decl_noinline void mi_recurse_exit_prim(void) { + recurse = false; +} + +static bool mi_recurse_enter(void) { + #if defined(__APPLE__) || defined(MI_TLS_RECURSE_GUARD) + if (_mi_preloading()) return false; + #endif + return mi_recurse_enter_prim(); +} + +static void mi_recurse_exit(void) { + #if defined(__APPLE__) || defined(MI_TLS_RECURSE_GUARD) + if (_mi_preloading()) return; + #endif + mi_recurse_exit_prim(); +} + +void _mi_fputs(mi_output_fun* out, void* arg, const char* prefix, const char* message) { + if (out==NULL || (void*)out==(void*)stdout || (void*)out==(void*)stderr) { // TODO: use mi_out_stderr for stderr? + if (!mi_recurse_enter()) return; + out = mi_out_get_default(&arg); + if (prefix != NULL) out(prefix, arg); + out(message, arg); + mi_recurse_exit(); + } + else { + if (prefix != NULL) out(prefix, arg); + out(message, arg); + } +} + +// Define our own limited `fprintf` that avoids memory allocation. +// We do this using `snprintf` with a limited buffer. +static void mi_vfprintf( mi_output_fun* out, void* arg, const char* prefix, const char* fmt, va_list args ) { + char buf[512]; + if (fmt==NULL) return; + if (!mi_recurse_enter()) return; + vsnprintf(buf,sizeof(buf)-1,fmt,args); + mi_recurse_exit(); + _mi_fputs(out,arg,prefix,buf); +} + +void _mi_fprintf( mi_output_fun* out, void* arg, const char* fmt, ... ) { + va_list args; + va_start(args,fmt); + mi_vfprintf(out,arg,NULL,fmt,args); + va_end(args); +} + +static void mi_vfprintf_thread(mi_output_fun* out, void* arg, const char* prefix, const char* fmt, va_list args) { + if (prefix != NULL && _mi_strnlen(prefix,33) <= 32 && !_mi_is_main_thread()) { + char tprefix[64]; + snprintf(tprefix, sizeof(tprefix), "%sthread 0x%llx: ", prefix, (unsigned long long)_mi_thread_id()); + mi_vfprintf(out, arg, tprefix, fmt, args); + } + else { + mi_vfprintf(out, arg, prefix, fmt, args); + } +} + +void _mi_trace_message(const char* fmt, ...) { + if (mi_option_get(mi_option_verbose) <= 1) return; // only with verbose level 2 or higher + va_list args; + va_start(args, fmt); + mi_vfprintf_thread(NULL, NULL, "mimalloc: ", fmt, args); + va_end(args); +} + +void _mi_verbose_message(const char* fmt, ...) { + if (!mi_option_is_enabled(mi_option_verbose)) return; + va_list args; + va_start(args,fmt); + mi_vfprintf(NULL, NULL, "mimalloc: ", fmt, args); + va_end(args); +} + +static void mi_show_error_message(const char* fmt, va_list args) { + if (!mi_option_is_enabled(mi_option_verbose)) { + if (!mi_option_is_enabled(mi_option_show_errors)) return; + if (mi_max_error_count >= 0 && (long)mi_atomic_increment_acq_rel(&error_count) > mi_max_error_count) return; + } + mi_vfprintf_thread(NULL, NULL, "mimalloc: error: ", fmt, args); +} + +void _mi_warning_message(const char* fmt, ...) { + if (!mi_option_is_enabled(mi_option_verbose)) { + if (!mi_option_is_enabled(mi_option_show_errors)) return; + if (mi_max_warning_count >= 0 && (long)mi_atomic_increment_acq_rel(&warning_count) > mi_max_warning_count) return; + } + va_list args; + va_start(args,fmt); + mi_vfprintf_thread(NULL, NULL, "mimalloc: warning: ", fmt, args); + va_end(args); +} + + +#if MI_DEBUG +void _mi_assert_fail(const char* assertion, const char* fname, unsigned line, const char* func ) { + _mi_fprintf(NULL, NULL, "mimalloc: assertion failed: at \"%s\":%u, %s\n assertion: \"%s\"\n", fname, line, (func==NULL?"":func), assertion); + abort(); +} +#endif + +// -------------------------------------------------------- +// Errors +// -------------------------------------------------------- + +static mi_error_fun* volatile mi_error_handler; // = NULL +static _Atomic(void*) mi_error_arg; // = NULL + +static void mi_error_default(int err) { + MI_UNUSED(err); +#if (MI_DEBUG>0) + if (err==EFAULT) { + #ifdef _MSC_VER + __debugbreak(); + #endif + abort(); + } +#endif +#if (MI_SECURE>0) + if (err==EFAULT) { // abort on serious errors in secure mode (corrupted meta-data) + abort(); + } +#endif +#if defined(MI_XMALLOC) + if (err==ENOMEM || err==EOVERFLOW) { // abort on memory allocation fails in xmalloc mode + abort(); + } +#endif +} + +void mi_register_error(mi_error_fun* fun, void* arg) { + mi_error_handler = fun; // can be NULL + mi_atomic_store_ptr_release(void,&mi_error_arg, arg); +} + +void _mi_error_message(int err, const char* fmt, ...) { + // show detailed error message + va_list args; + va_start(args, fmt); + mi_show_error_message(fmt, args); + va_end(args); + // and call the error handler which may abort (or return normally) + if (mi_error_handler != NULL) { + mi_error_handler(err, mi_atomic_load_ptr_acquire(void,&mi_error_arg)); + } + else { + mi_error_default(err); + } +} + +// -------------------------------------------------------- +// Initialize options by checking the environment +// -------------------------------------------------------- +char _mi_toupper(char c) { + if (c >= 'a' && c <= 'z') return (c - 'a' + 'A'); + else return c; +} + +int _mi_strnicmp(const char* s, const char* t, size_t n) { + if (n == 0) return 0; + for (; *s != 0 && *t != 0 && n > 0; s++, t++, n--) { + if (_mi_toupper(*s) != _mi_toupper(*t)) break; + } + return (n == 0 ? 0 : *s - *t); +} + +void _mi_strlcpy(char* dest, const char* src, size_t dest_size) { + if (dest==NULL || src==NULL || dest_size == 0) return; + // copy until end of src, or when dest is (almost) full + while (*src != 0 && dest_size > 1) { + *dest++ = *src++; + dest_size--; + } + // always zero terminate + *dest = 0; +} + +void _mi_strlcat(char* dest, const char* src, size_t dest_size) { + if (dest==NULL || src==NULL || dest_size == 0) return; + // find end of string in the dest buffer + while (*dest != 0 && dest_size > 1) { + dest++; + dest_size--; + } + // and catenate + _mi_strlcpy(dest, src, dest_size); +} + +size_t _mi_strlen(const char* s) { + if (s==NULL) return 0; + size_t len = 0; + while(s[len] != 0) { len++; } + return len; +} + +size_t _mi_strnlen(const char* s, size_t max_len) { + if (s==NULL) return 0; + size_t len = 0; + while(s[len] != 0 && len < max_len) { len++; } + return len; +} + +#ifdef MI_NO_GETENV +static bool mi_getenv(const char* name, char* result, size_t result_size) { + MI_UNUSED(name); + MI_UNUSED(result); + MI_UNUSED(result_size); + return false; +} +#else +static bool mi_getenv(const char* name, char* result, size_t result_size) { + if (name==NULL || result == NULL || result_size < 64) return false; + return _mi_prim_getenv(name,result,result_size); +} +#endif + +// TODO: implement ourselves to reduce dependencies on the C runtime +#include // strtol +#include // strstr + + +static void mi_option_init(mi_option_desc_t* desc) { + // Read option value from the environment + char s[64 + 1]; + char buf[64+1]; + _mi_strlcpy(buf, "mimalloc_", sizeof(buf)); + _mi_strlcat(buf, desc->name, sizeof(buf)); + bool found = mi_getenv(buf, s, sizeof(s)); + if (!found && desc->legacy_name != NULL) { + _mi_strlcpy(buf, "mimalloc_", sizeof(buf)); + _mi_strlcat(buf, desc->legacy_name, sizeof(buf)); + found = mi_getenv(buf, s, sizeof(s)); + if (found) { + _mi_warning_message("environment option \"mimalloc_%s\" is deprecated -- use \"mimalloc_%s\" instead.\n", desc->legacy_name, desc->name); + } + } + + if (found) { + size_t len = _mi_strnlen(s, sizeof(buf) - 1); + for (size_t i = 0; i < len; i++) { + buf[i] = _mi_toupper(s[i]); + } + buf[len] = 0; + if (buf[0] == 0 || strstr("1;TRUE;YES;ON", buf) != NULL) { + desc->value = 1; + desc->init = INITIALIZED; + } + else if (strstr("0;FALSE;NO;OFF", buf) != NULL) { + desc->value = 0; + desc->init = INITIALIZED; + } + else { + char* end = buf; + long value = strtol(buf, &end, 10); + if (desc->option == mi_option_reserve_os_memory || desc->option == mi_option_arena_reserve) { + // this option is interpreted in KiB to prevent overflow of `long` + if (*end == 'K') { end++; } + else if (*end == 'M') { value *= MI_KiB; end++; } + else if (*end == 'G') { value *= MI_MiB; end++; } + else { value = (value + MI_KiB - 1) / MI_KiB; } + if (end[0] == 'I' && end[1] == 'B') { end += 2; } + else if (*end == 'B') { end++; } + } + if (*end == 0) { + desc->value = value; + desc->init = INITIALIZED; + } + else { + // set `init` first to avoid recursion through _mi_warning_message on mimalloc_verbose. + desc->init = DEFAULTED; + if (desc->option == mi_option_verbose && desc->value == 0) { + // if the 'mimalloc_verbose' env var has a bogus value we'd never know + // (since the value defaults to 'off') so in that case briefly enable verbose + desc->value = 1; + _mi_warning_message("environment option mimalloc_%s has an invalid value.\n", desc->name); + desc->value = 0; + } + else { + _mi_warning_message("environment option mimalloc_%s has an invalid value.\n", desc->name); + } + } + } + mi_assert_internal(desc->init != UNINIT); + } + else if (!_mi_preloading()) { + desc->init = DEFAULTED; + } +} diff --git a/compat/mimalloc/os.c b/compat/mimalloc/os.c new file mode 100644 index 00000000000000..bf9de1be0fdb49 --- /dev/null +++ b/compat/mimalloc/os.c @@ -0,0 +1,689 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" +#include "mimalloc/prim.h" + + +/* ----------------------------------------------------------- + Initialization. + On windows initializes support for aligned allocation and + large OS pages (if MIMALLOC_LARGE_OS_PAGES is true). +----------------------------------------------------------- */ + +static mi_os_mem_config_t mi_os_mem_config = { + 4096, // page size + 0, // large page size (usually 2MiB) + 4096, // allocation granularity + true, // has overcommit? (if true we use MAP_NORESERVE on mmap systems) + false, // must free whole? (on mmap systems we can free anywhere in a mapped range, but on Windows we must free the entire span) + true // has virtual reserve? (if true we can reserve virtual address space without using commit or physical memory) +}; + +bool _mi_os_has_overcommit(void) { + return mi_os_mem_config.has_overcommit; +} + +bool _mi_os_has_virtual_reserve(void) { + return mi_os_mem_config.has_virtual_reserve; +} + + +// OS (small) page size +size_t _mi_os_page_size(void) { + return mi_os_mem_config.page_size; +} + +// if large OS pages are supported (2 or 4MiB), then return the size, otherwise return the small page size (4KiB) +size_t _mi_os_large_page_size(void) { + return (mi_os_mem_config.large_page_size != 0 ? mi_os_mem_config.large_page_size : _mi_os_page_size()); +} + +bool _mi_os_use_large_page(size_t size, size_t alignment) { + // if we have access, check the size and alignment requirements + if (mi_os_mem_config.large_page_size == 0 || !mi_option_is_enabled(mi_option_allow_large_os_pages)) return false; + return ((size % mi_os_mem_config.large_page_size) == 0 && (alignment % mi_os_mem_config.large_page_size) == 0); +} + +// round to a good OS allocation size (bounded by max 12.5% waste) +size_t _mi_os_good_alloc_size(size_t size) { + size_t align_size; + if (size < 512*MI_KiB) align_size = _mi_os_page_size(); + else if (size < 2*MI_MiB) align_size = 64*MI_KiB; + else if (size < 8*MI_MiB) align_size = 256*MI_KiB; + else if (size < 32*MI_MiB) align_size = 1*MI_MiB; + else align_size = 4*MI_MiB; + if mi_unlikely(size >= (SIZE_MAX - align_size)) return size; // possible overflow? + return _mi_align_up(size, align_size); +} + +void _mi_os_init(void) { + _mi_prim_mem_init(&mi_os_mem_config); +} + + +/* ----------------------------------------------------------- + Util +-------------------------------------------------------------- */ +bool _mi_os_decommit(void* addr, size_t size, mi_stats_t* stats); +bool _mi_os_commit(void* addr, size_t size, bool* is_zero, mi_stats_t* tld_stats); + +static void* mi_align_up_ptr(void* p, size_t alignment) { + return (void*)_mi_align_up((uintptr_t)p, alignment); +} + +static void* mi_align_down_ptr(void* p, size_t alignment) { + return (void*)_mi_align_down((uintptr_t)p, alignment); +} + + +/* ----------------------------------------------------------- + aligned hinting +-------------------------------------------------------------- */ + +// On 64-bit systems, we can do efficient aligned allocation by using +// the 2TiB to 30TiB area to allocate those. +#if (MI_INTPTR_SIZE >= 8) +static mi_decl_cache_align _Atomic(uintptr_t)aligned_base; + +// Return a MI_SEGMENT_SIZE aligned address that is probably available. +// If this returns NULL, the OS will determine the address but on some OS's that may not be +// properly aligned which can be more costly as it needs to be adjusted afterwards. +// For a size > 1GiB this always returns NULL in order to guarantee good ASLR randomization; +// (otherwise an initial large allocation of say 2TiB has a 50% chance to include (known) addresses +// in the middle of the 2TiB - 6TiB address range (see issue #372)) + +#define MI_HINT_BASE ((uintptr_t)2 << 40) // 2TiB start +#define MI_HINT_AREA ((uintptr_t)4 << 40) // upto 6TiB (since before win8 there is "only" 8TiB available to processes) +#define MI_HINT_MAX ((uintptr_t)30 << 40) // wrap after 30TiB (area after 32TiB is used for huge OS pages) + +void* _mi_os_get_aligned_hint(size_t try_alignment, size_t size) +{ + if (try_alignment <= 1 || try_alignment > MI_SEGMENT_SIZE) return NULL; + size = _mi_align_up(size, MI_SEGMENT_SIZE); + if (size > 1*MI_GiB) return NULL; // guarantee the chance of fixed valid address is at most 1/(MI_HINT_AREA / 1<<30) = 1/4096. + #if (MI_SECURE>0) + size += MI_SEGMENT_SIZE; // put in `MI_SEGMENT_SIZE` virtual gaps between hinted blocks; this splits VLA's but increases guarded areas. + #endif + + uintptr_t hint = mi_atomic_add_acq_rel(&aligned_base, size); + if (hint == 0 || hint > MI_HINT_MAX) { // wrap or initialize + uintptr_t init = MI_HINT_BASE; + #if (MI_SECURE>0 || MI_DEBUG==0) // security: randomize start of aligned allocations unless in debug mode + uintptr_t r = _mi_heap_random_next(mi_prim_get_default_heap()); + init = init + ((MI_SEGMENT_SIZE * ((r>>17) & 0xFFFFF)) % MI_HINT_AREA); // (randomly 20 bits)*4MiB == 0 to 4TiB + #endif + uintptr_t expected = hint + size; + mi_atomic_cas_strong_acq_rel(&aligned_base, &expected, init); + hint = mi_atomic_add_acq_rel(&aligned_base, size); // this may still give 0 or > MI_HINT_MAX but that is ok, it is a hint after all + } + if (hint%try_alignment != 0) return NULL; + return (void*)hint; +} +#else +void* _mi_os_get_aligned_hint(size_t try_alignment, size_t size) { + MI_UNUSED(try_alignment); MI_UNUSED(size); + return NULL; +} +#endif + + +/* ----------------------------------------------------------- + Free memory +-------------------------------------------------------------- */ + +static void mi_os_free_huge_os_pages(void* p, size_t size, mi_stats_t* stats); + +static void mi_os_prim_free(void* addr, size_t size, bool still_committed, mi_stats_t* tld_stats) { + MI_UNUSED(tld_stats); + mi_assert_internal((size % _mi_os_page_size()) == 0); + if (addr == NULL || size == 0) return; // || _mi_os_is_huge_reserved(addr) + int err = _mi_prim_free(addr, size); + if (err != 0) { + _mi_warning_message("unable to free OS memory (error: %d (0x%x), size: 0x%zx bytes, address: %p)\n", err, err, size, addr); + } + mi_stats_t* stats = &_mi_stats_main; + if (still_committed) { _mi_stat_decrease(&stats->committed, size); } + _mi_stat_decrease(&stats->reserved, size); +} + +void _mi_os_free_ex(void* addr, size_t size, bool still_committed, mi_memid_t memid, mi_stats_t* tld_stats) { + if (mi_memkind_is_os(memid.memkind)) { + size_t csize = _mi_os_good_alloc_size(size); + void* base = addr; + // different base? (due to alignment) + if (memid.mem.os.base != NULL) { + mi_assert(memid.mem.os.base <= addr); + mi_assert((uint8_t*)memid.mem.os.base + memid.mem.os.alignment >= (uint8_t*)addr); + base = memid.mem.os.base; + csize += ((uint8_t*)addr - (uint8_t*)memid.mem.os.base); + } + // free it + if (memid.memkind == MI_MEM_OS_HUGE) { + mi_assert(memid.is_pinned); + mi_os_free_huge_os_pages(base, csize, tld_stats); + } + else { + mi_os_prim_free(base, csize, still_committed, tld_stats); + } + } + else { + // nothing to do + mi_assert(memid.memkind < MI_MEM_OS); + } +} + +void _mi_os_free(void* p, size_t size, mi_memid_t memid, mi_stats_t* tld_stats) { + _mi_os_free_ex(p, size, true, memid, tld_stats); +} + + +/* ----------------------------------------------------------- + Primitive allocation from the OS. +-------------------------------------------------------------- */ + +// Note: the `try_alignment` is just a hint and the returned pointer is not guaranteed to be aligned. +static void* mi_os_prim_alloc(size_t size, size_t try_alignment, bool commit, bool allow_large, bool* is_large, bool* is_zero, mi_stats_t* stats) { + mi_assert_internal(size > 0 && (size % _mi_os_page_size()) == 0); + mi_assert_internal(is_zero != NULL); + mi_assert_internal(is_large != NULL); + if (size == 0) return NULL; + if (!commit) { allow_large = false; } + if (try_alignment == 0) { try_alignment = 1; } // avoid 0 to ensure there will be no divide by zero when aligning + + *is_zero = false; + void* p = NULL; + int err = _mi_prim_alloc(size, try_alignment, commit, allow_large, is_large, is_zero, &p); + if (err != 0) { + _mi_warning_message("unable to allocate OS memory (error: %d (0x%x), size: 0x%zx bytes, align: 0x%zx, commit: %d, allow large: %d)\n", err, err, size, try_alignment, commit, allow_large); + } + mi_stat_counter_increase(stats->mmap_calls, 1); + if (p != NULL) { + _mi_stat_increase(&stats->reserved, size); + if (commit) { + _mi_stat_increase(&stats->committed, size); + // seems needed for asan (or `mimalloc-test-api` fails) + #ifdef MI_TRACK_ASAN + if (*is_zero) { mi_track_mem_defined(p,size); } + else { mi_track_mem_undefined(p,size); } + #endif + } + } + return p; +} + + +// Primitive aligned allocation from the OS. +// This function guarantees the allocated memory is aligned. +static void* mi_os_prim_alloc_aligned(size_t size, size_t alignment, bool commit, bool allow_large, bool* is_large, bool* is_zero, void** base, mi_stats_t* stats) { + mi_assert_internal(alignment >= _mi_os_page_size() && ((alignment & (alignment - 1)) == 0)); + mi_assert_internal(size > 0 && (size % _mi_os_page_size()) == 0); + mi_assert_internal(is_large != NULL); + mi_assert_internal(is_zero != NULL); + mi_assert_internal(base != NULL); + if (!commit) allow_large = false; + if (!(alignment >= _mi_os_page_size() && ((alignment & (alignment - 1)) == 0))) return NULL; + size = _mi_align_up(size, _mi_os_page_size()); + + // try first with a hint (this will be aligned directly on Win 10+ or BSD) + void* p = mi_os_prim_alloc(size, alignment, commit, allow_large, is_large, is_zero, stats); + if (p == NULL) return NULL; + + // aligned already? + if (((uintptr_t)p % alignment) == 0) { + *base = p; + } + else { + // if not aligned, free it, overallocate, and unmap around it + _mi_warning_message("unable to allocate aligned OS memory directly, fall back to over-allocation (size: 0x%zx bytes, address: %p, alignment: 0x%zx, commit: %d)\n", size, p, alignment, commit); + mi_os_prim_free(p, size, commit, stats); + if (size >= (SIZE_MAX - alignment)) return NULL; // overflow + const size_t over_size = size + alignment; + + if (mi_os_mem_config.must_free_whole) { // win32 virtualAlloc cannot free parts of an allocate block + // over-allocate uncommitted (virtual) memory + p = mi_os_prim_alloc(over_size, 1 /*alignment*/, false /* commit? */, false /* allow_large */, is_large, is_zero, stats); + if (p == NULL) return NULL; + + // set p to the aligned part in the full region + // note: this is dangerous on Windows as VirtualFree needs the actual base pointer + // this is handled though by having the `base` field in the memid's + *base = p; // remember the base + p = mi_align_up_ptr(p, alignment); + + // explicitly commit only the aligned part + if (commit) { + _mi_os_commit(p, size, NULL, stats); + } + } + else { // mmap can free inside an allocation + // overallocate... + p = mi_os_prim_alloc(over_size, 1, commit, false, is_large, is_zero, stats); + if (p == NULL) return NULL; + + // and selectively unmap parts around the over-allocated area. (noop on sbrk) + void* aligned_p = mi_align_up_ptr(p, alignment); + size_t pre_size = (uint8_t*)aligned_p - (uint8_t*)p; + size_t mid_size = _mi_align_up(size, _mi_os_page_size()); + size_t post_size = over_size - pre_size - mid_size; + mi_assert_internal(pre_size < over_size&& post_size < over_size&& mid_size >= size); + if (pre_size > 0) { mi_os_prim_free(p, pre_size, commit, stats); } + if (post_size > 0) { mi_os_prim_free((uint8_t*)aligned_p + mid_size, post_size, commit, stats); } + // we can return the aligned pointer on `mmap` (and sbrk) systems + p = aligned_p; + *base = aligned_p; // since we freed the pre part, `*base == p`. + } + } + + mi_assert_internal(p == NULL || (p != NULL && *base != NULL && ((uintptr_t)p % alignment) == 0)); + return p; +} + + +/* ----------------------------------------------------------- + OS API: alloc and alloc_aligned +----------------------------------------------------------- */ + +void* _mi_os_alloc(size_t size, mi_memid_t* memid, mi_stats_t* tld_stats) { + MI_UNUSED(tld_stats); + *memid = _mi_memid_none(); + mi_stats_t* stats = &_mi_stats_main; + if (size == 0) return NULL; + size = _mi_os_good_alloc_size(size); + bool os_is_large = false; + bool os_is_zero = false; + void* p = mi_os_prim_alloc(size, 0, true, false, &os_is_large, &os_is_zero, stats); + if (p != NULL) { + *memid = _mi_memid_create_os(true, os_is_zero, os_is_large); + } + return p; +} + +void* _mi_os_alloc_aligned(size_t size, size_t alignment, bool commit, bool allow_large, mi_memid_t* memid, mi_stats_t* tld_stats) +{ + MI_UNUSED(&_mi_os_get_aligned_hint); // suppress unused warnings + MI_UNUSED(tld_stats); + *memid = _mi_memid_none(); + if (size == 0) return NULL; + size = _mi_os_good_alloc_size(size); + alignment = _mi_align_up(alignment, _mi_os_page_size()); + + bool os_is_large = false; + bool os_is_zero = false; + void* os_base = NULL; + void* p = mi_os_prim_alloc_aligned(size, alignment, commit, allow_large, &os_is_large, &os_is_zero, &os_base, &_mi_stats_main /*tld->stats*/ ); + if (p != NULL) { + *memid = _mi_memid_create_os(commit, os_is_zero, os_is_large); + memid->mem.os.base = os_base; + memid->mem.os.alignment = alignment; + } + return p; +} + +/* ----------------------------------------------------------- + OS aligned allocation with an offset. This is used + for large alignments > MI_ALIGNMENT_MAX. We use a large mimalloc + page where the object can be aligned at an offset from the start of the segment. + As we may need to overallocate, we need to free such pointers using `mi_free_aligned` + to use the actual start of the memory region. +----------------------------------------------------------- */ + +void* _mi_os_alloc_aligned_at_offset(size_t size, size_t alignment, size_t offset, bool commit, bool allow_large, mi_memid_t* memid, mi_stats_t* tld_stats) { + mi_assert(offset <= MI_SEGMENT_SIZE); + mi_assert(offset <= size); + mi_assert((alignment % _mi_os_page_size()) == 0); + *memid = _mi_memid_none(); + if (offset > MI_SEGMENT_SIZE) return NULL; + if (offset == 0) { + // regular aligned allocation + return _mi_os_alloc_aligned(size, alignment, commit, allow_large, memid, tld_stats); + } + else { + // overallocate to align at an offset + const size_t extra = _mi_align_up(offset, alignment) - offset; + const size_t oversize = size + extra; + void* const start = _mi_os_alloc_aligned(oversize, alignment, commit, allow_large, memid, tld_stats); + if (start == NULL) return NULL; + + void* const p = (uint8_t*)start + extra; + mi_assert(_mi_is_aligned((uint8_t*)p + offset, alignment)); + // decommit the overallocation at the start + if (commit && extra > _mi_os_page_size()) { + _mi_os_decommit(start, extra, tld_stats); + } + return p; + } +} + +/* ----------------------------------------------------------- + OS memory API: reset, commit, decommit, protect, unprotect. +----------------------------------------------------------- */ + +// OS page align within a given area, either conservative (pages inside the area only), +// or not (straddling pages outside the area is possible) +static void* mi_os_page_align_areax(bool conservative, void* addr, size_t size, size_t* newsize) { + mi_assert(addr != NULL && size > 0); + if (newsize != NULL) *newsize = 0; + if (size == 0 || addr == NULL) return NULL; + + // page align conservatively within the range + void* start = (conservative ? mi_align_up_ptr(addr, _mi_os_page_size()) + : mi_align_down_ptr(addr, _mi_os_page_size())); + void* end = (conservative ? mi_align_down_ptr((uint8_t*)addr + size, _mi_os_page_size()) + : mi_align_up_ptr((uint8_t*)addr + size, _mi_os_page_size())); + ptrdiff_t diff = (uint8_t*)end - (uint8_t*)start; + if (diff <= 0) return NULL; + + mi_assert_internal((conservative && (size_t)diff <= size) || (!conservative && (size_t)diff >= size)); + if (newsize != NULL) *newsize = (size_t)diff; + return start; +} + +static void* mi_os_page_align_area_conservative(void* addr, size_t size, size_t* newsize) { + return mi_os_page_align_areax(true, addr, size, newsize); +} + +bool _mi_os_commit(void* addr, size_t size, bool* is_zero, mi_stats_t* tld_stats) { + MI_UNUSED(tld_stats); + mi_stats_t* stats = &_mi_stats_main; + if (is_zero != NULL) { *is_zero = false; } + _mi_stat_increase(&stats->committed, size); // use size for precise commit vs. decommit + _mi_stat_counter_increase(&stats->commit_calls, 1); + + // page align range + size_t csize; + void* start = mi_os_page_align_areax(false /* conservative? */, addr, size, &csize); + if (csize == 0) return true; + + // commit + bool os_is_zero = false; + int err = _mi_prim_commit(start, csize, &os_is_zero); + if (err != 0) { + _mi_warning_message("cannot commit OS memory (error: %d (0x%x), address: %p, size: 0x%zx bytes)\n", err, err, start, csize); + return false; + } + if (os_is_zero && is_zero != NULL) { + *is_zero = true; + mi_assert_expensive(mi_mem_is_zero(start, csize)); + } + // note: the following seems required for asan (otherwise `mimalloc-test-stress` fails) + #ifdef MI_TRACK_ASAN + if (os_is_zero) { mi_track_mem_defined(start,csize); } + else { mi_track_mem_undefined(start,csize); } + #endif + return true; +} + +static bool mi_os_decommit_ex(void* addr, size_t size, bool* needs_recommit, mi_stats_t* tld_stats) { + MI_UNUSED(tld_stats); + mi_stats_t* stats = &_mi_stats_main; + mi_assert_internal(needs_recommit!=NULL); + _mi_stat_decrease(&stats->committed, size); + + // page align + size_t csize; + void* start = mi_os_page_align_area_conservative(addr, size, &csize); + if (csize == 0) return true; + + // decommit + *needs_recommit = true; + int err = _mi_prim_decommit(start,csize,needs_recommit); + if (err != 0) { + _mi_warning_message("cannot decommit OS memory (error: %d (0x%x), address: %p, size: 0x%zx bytes)\n", err, err, start, csize); + } + mi_assert_internal(err == 0); + return (err == 0); +} + +bool _mi_os_decommit(void* addr, size_t size, mi_stats_t* tld_stats) { + bool needs_recommit; + return mi_os_decommit_ex(addr, size, &needs_recommit, tld_stats); +} + + +// Signal to the OS that the address range is no longer in use +// but may be used later again. This will release physical memory +// pages and reduce swapping while keeping the memory committed. +// We page align to a conservative area inside the range to reset. +bool _mi_os_reset(void* addr, size_t size, mi_stats_t* stats) { + // page align conservatively within the range + size_t csize; + void* start = mi_os_page_align_area_conservative(addr, size, &csize); + if (csize == 0) return true; // || _mi_os_is_huge_reserved(addr) + _mi_stat_increase(&stats->reset, csize); + _mi_stat_counter_increase(&stats->reset_calls, 1); + + #if (MI_DEBUG>1) && !MI_SECURE && !MI_TRACK_ENABLED // && !MI_TSAN + memset(start, 0, csize); // pretend it is eagerly reset + #endif + + int err = _mi_prim_reset(start, csize); + if (err != 0) { + _mi_warning_message("cannot reset OS memory (error: %d (0x%x), address: %p, size: 0x%zx bytes)\n", err, err, start, csize); + } + return (err == 0); +} + + +// either resets or decommits memory, returns true if the memory needs +// to be recommitted if it is to be re-used later on. +bool _mi_os_purge_ex(void* p, size_t size, bool allow_reset, mi_stats_t* stats) +{ + if (mi_option_get(mi_option_purge_delay) < 0) return false; // is purging allowed? + _mi_stat_counter_increase(&stats->purge_calls, 1); + _mi_stat_increase(&stats->purged, size); + + if (mi_option_is_enabled(mi_option_purge_decommits) && // should decommit? + !_mi_preloading()) // don't decommit during preloading (unsafe) + { + bool needs_recommit = true; + mi_os_decommit_ex(p, size, &needs_recommit, stats); + return needs_recommit; + } + else { + if (allow_reset) { // this can sometimes be not allowed if the range is not fully committed + _mi_os_reset(p, size, stats); + } + return false; // needs no recommit + } +} + +// either resets or decommits memory, returns true if the memory needs +// to be recommitted if it is to be re-used later on. +bool _mi_os_purge(void* p, size_t size, mi_stats_t * stats) { + return _mi_os_purge_ex(p, size, true, stats); +} + +// Protect a region in memory to be not accessible. +static bool mi_os_protectx(void* addr, size_t size, bool protect) { + // page align conservatively within the range + size_t csize = 0; + void* start = mi_os_page_align_area_conservative(addr, size, &csize); + if (csize == 0) return false; + /* + if (_mi_os_is_huge_reserved(addr)) { + _mi_warning_message("cannot mprotect memory allocated in huge OS pages\n"); + } + */ + int err = _mi_prim_protect(start,csize,protect); + if (err != 0) { + _mi_warning_message("cannot %s OS memory (error: %d (0x%x), address: %p, size: 0x%zx bytes)\n", (protect ? "protect" : "unprotect"), err, err, start, csize); + } + return (err == 0); +} + +bool _mi_os_protect(void* addr, size_t size) { + return mi_os_protectx(addr, size, true); +} + +bool _mi_os_unprotect(void* addr, size_t size) { + return mi_os_protectx(addr, size, false); +} + + + +/* ---------------------------------------------------------------------------- +Support for allocating huge OS pages (1Gib) that are reserved up-front +and possibly associated with a specific NUMA node. (use `numa_node>=0`) +-----------------------------------------------------------------------------*/ +#define MI_HUGE_OS_PAGE_SIZE (MI_GiB) + + +#if (MI_INTPTR_SIZE >= 8) +// To ensure proper alignment, use our own area for huge OS pages +static mi_decl_cache_align _Atomic(uintptr_t) mi_huge_start; // = 0 + +// Claim an aligned address range for huge pages +static uint8_t* mi_os_claim_huge_pages(size_t pages, size_t* total_size) { + if (total_size != NULL) *total_size = 0; + const size_t size = pages * MI_HUGE_OS_PAGE_SIZE; + + uintptr_t start = 0; + uintptr_t end = 0; + uintptr_t huge_start = mi_atomic_load_relaxed(&mi_huge_start); + do { + start = huge_start; + if (start == 0) { + // Initialize the start address after the 32TiB area + start = ((uintptr_t)32 << 40); // 32TiB virtual start address + #if (MI_SECURE>0 || MI_DEBUG==0) // security: randomize start of huge pages unless in debug mode + uintptr_t r = _mi_heap_random_next(mi_prim_get_default_heap()); + start = start + ((uintptr_t)MI_HUGE_OS_PAGE_SIZE * ((r>>17) & 0x0FFF)); // (randomly 12bits)*1GiB == between 0 to 4TiB + #endif + } + end = start + size; + mi_assert_internal(end % MI_SEGMENT_SIZE == 0); + } while (!mi_atomic_cas_strong_acq_rel(&mi_huge_start, &huge_start, end)); + + if (total_size != NULL) *total_size = size; + return (uint8_t*)start; +} +#else +static uint8_t* mi_os_claim_huge_pages(size_t pages, size_t* total_size) { + MI_UNUSED(pages); + if (total_size != NULL) *total_size = 0; + return NULL; +} +#endif + +// Allocate MI_SEGMENT_SIZE aligned huge pages +void* _mi_os_alloc_huge_os_pages(size_t pages, int numa_node, mi_msecs_t max_msecs, size_t* pages_reserved, size_t* psize, mi_memid_t* memid) { + *memid = _mi_memid_none(); + if (psize != NULL) *psize = 0; + if (pages_reserved != NULL) *pages_reserved = 0; + size_t size = 0; + uint8_t* start = mi_os_claim_huge_pages(pages, &size); + if (start == NULL) return NULL; // or 32-bit systems + + // Allocate one page at the time but try to place them contiguously + // We allocate one page at the time to be able to abort if it takes too long + // or to at least allocate as many as available on the system. + mi_msecs_t start_t = _mi_clock_start(); + size_t page = 0; + bool all_zero = true; + while (page < pages) { + // allocate a page + bool is_zero = false; + void* addr = start + (page * MI_HUGE_OS_PAGE_SIZE); + void* p = NULL; + int err = _mi_prim_alloc_huge_os_pages(addr, MI_HUGE_OS_PAGE_SIZE, numa_node, &is_zero, &p); + if (!is_zero) { all_zero = false; } + if (err != 0) { + _mi_warning_message("unable to allocate huge OS page (error: %d (0x%x), address: %p, size: %zx bytes)\n", err, err, addr, MI_HUGE_OS_PAGE_SIZE); + break; + } + + // Did we succeed at a contiguous address? + if (p != addr) { + // no success, issue a warning and break + if (p != NULL) { + _mi_warning_message("could not allocate contiguous huge OS page %zu at %p\n", page, addr); + mi_os_prim_free(p, MI_HUGE_OS_PAGE_SIZE, true, &_mi_stats_main); + } + break; + } + + // success, record it + page++; // increase before timeout check (see issue #711) + _mi_stat_increase(&_mi_stats_main.committed, MI_HUGE_OS_PAGE_SIZE); + _mi_stat_increase(&_mi_stats_main.reserved, MI_HUGE_OS_PAGE_SIZE); + + // check for timeout + if (max_msecs > 0) { + mi_msecs_t elapsed = _mi_clock_end(start_t); + if (page >= 1) { + mi_msecs_t estimate = ((elapsed / (page+1)) * pages); + if (estimate > 2*max_msecs) { // seems like we are going to timeout, break + elapsed = max_msecs + 1; + } + } + if (elapsed > max_msecs) { + _mi_warning_message("huge OS page allocation timed out (after allocating %zu page(s))\n", page); + break; + } + } + } + mi_assert_internal(page*MI_HUGE_OS_PAGE_SIZE <= size); + if (pages_reserved != NULL) { *pages_reserved = page; } + if (psize != NULL) { *psize = page * MI_HUGE_OS_PAGE_SIZE; } + if (page != 0) { + mi_assert(start != NULL); + *memid = _mi_memid_create_os(true /* is committed */, all_zero, true /* is_large */); + memid->memkind = MI_MEM_OS_HUGE; + mi_assert(memid->is_pinned); + #ifdef MI_TRACK_ASAN + if (all_zero) { mi_track_mem_defined(start,size); } + #endif + } + return (page == 0 ? NULL : start); +} + +// free every huge page in a range individually (as we allocated per page) +// note: needed with VirtualAlloc but could potentially be done in one go on mmap'd systems. +static void mi_os_free_huge_os_pages(void* p, size_t size, mi_stats_t* stats) { + if (p==NULL || size==0) return; + uint8_t* base = (uint8_t*)p; + while (size >= MI_HUGE_OS_PAGE_SIZE) { + mi_os_prim_free(base, MI_HUGE_OS_PAGE_SIZE, true, stats); + size -= MI_HUGE_OS_PAGE_SIZE; + base += MI_HUGE_OS_PAGE_SIZE; + } +} + +/* ---------------------------------------------------------------------------- +Support NUMA aware allocation +-----------------------------------------------------------------------------*/ + +_Atomic(size_t) _mi_numa_node_count; // = 0 // cache the node count + +size_t _mi_os_numa_node_count_get(void) { + size_t count = mi_atomic_load_acquire(&_mi_numa_node_count); + if (count <= 0) { + long ncount = mi_option_get(mi_option_use_numa_nodes); // given explicitly? + if (ncount > 0) { + count = (size_t)ncount; + } + else { + count = _mi_prim_numa_node_count(); // or detect dynamically + if (count == 0) count = 1; + } + mi_atomic_store_release(&_mi_numa_node_count, count); // save it + _mi_verbose_message("using %zd numa regions\n", count); + } + return count; +} + +int _mi_os_numa_node_get(mi_os_tld_t* tld) { + MI_UNUSED(tld); + size_t numa_count = _mi_os_numa_node_count(); + if (numa_count<=1) return 0; // optimize on single numa node systems: always node 0 + // never more than the node count and >= 0 + size_t numa_node = _mi_prim_numa_node(); + if (numa_node >= numa_count) { numa_node = numa_node % numa_count; } + return (int)numa_node; +} diff --git a/compat/mimalloc/page-queue.c b/compat/mimalloc/page-queue.c new file mode 100644 index 00000000000000..5619a81f9917fe --- /dev/null +++ b/compat/mimalloc/page-queue.c @@ -0,0 +1,332 @@ +/*---------------------------------------------------------------------------- +Copyright (c) 2018-2020, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +/* ----------------------------------------------------------- + Definition of page queues for each block size +----------------------------------------------------------- */ + +#ifndef MI_IN_PAGE_C +#error "this file should be included from 'page.c'" +#endif + +/* ----------------------------------------------------------- + Minimal alignment in machine words (i.e. `sizeof(void*)`) +----------------------------------------------------------- */ + +#if (MI_MAX_ALIGN_SIZE > 4*MI_INTPTR_SIZE) + #error "define alignment for more than 4x word size for this platform" +#elif (MI_MAX_ALIGN_SIZE > 2*MI_INTPTR_SIZE) + #define MI_ALIGN4W // 4 machine words minimal alignment +#elif (MI_MAX_ALIGN_SIZE > MI_INTPTR_SIZE) + #define MI_ALIGN2W // 2 machine words minimal alignment +#else + // ok, default alignment is 1 word +#endif + + +/* ----------------------------------------------------------- + Queue query +----------------------------------------------------------- */ + + +static inline bool mi_page_queue_is_huge(const mi_page_queue_t* pq) { + return (pq->block_size == (MI_MEDIUM_OBJ_SIZE_MAX+sizeof(uintptr_t))); +} + +static inline bool mi_page_queue_is_full(const mi_page_queue_t* pq) { + return (pq->block_size == (MI_MEDIUM_OBJ_SIZE_MAX+(2*sizeof(uintptr_t)))); +} + +static inline bool mi_page_queue_is_special(const mi_page_queue_t* pq) { + return (pq->block_size > MI_MEDIUM_OBJ_SIZE_MAX); +} + +/* ----------------------------------------------------------- + Bins +----------------------------------------------------------- */ + +// Return the bin for a given field size. +// Returns MI_BIN_HUGE if the size is too large. +// We use `wsize` for the size in "machine word sizes", +// i.e. byte size == `wsize*sizeof(void*)`. +static inline uint8_t mi_bin(size_t size) { + size_t wsize = _mi_wsize_from_size(size); + uint8_t bin; + if (wsize <= 1) { + bin = 1; + } + #if defined(MI_ALIGN4W) + else if (wsize <= 4) { + bin = (uint8_t)((wsize+1)&~1); // round to double word sizes + } + #elif defined(MI_ALIGN2W) + else if (wsize <= 8) { + bin = (uint8_t)((wsize+1)&~1); // round to double word sizes + } + #else + else if (wsize <= 8) { + bin = (uint8_t)wsize; + } + #endif + else if (wsize > MI_MEDIUM_OBJ_WSIZE_MAX) { + bin = MI_BIN_HUGE; + } + else { + #if defined(MI_ALIGN4W) + if (wsize <= 16) { wsize = (wsize+3)&~3; } // round to 4x word sizes + #endif + wsize--; + // find the highest bit + uint8_t b = (uint8_t)mi_bsr(wsize); // note: wsize != 0 + // and use the top 3 bits to determine the bin (~12.5% worst internal fragmentation). + // - adjust with 3 because we use do not round the first 8 sizes + // which each get an exact bin + bin = ((b << 2) + (uint8_t)((wsize >> (b - 2)) & 0x03)) - 3; + mi_assert_internal(bin < MI_BIN_HUGE); + } + mi_assert_internal(bin > 0 && bin <= MI_BIN_HUGE); + return bin; +} + + + +/* ----------------------------------------------------------- + Queue of pages with free blocks +----------------------------------------------------------- */ + +uint8_t _mi_bin(size_t size) { + return mi_bin(size); +} + +size_t _mi_bin_size(uint8_t bin) { + return _mi_heap_empty.pages[bin].block_size; +} + +// Good size for allocation +size_t mi_good_size(size_t size) mi_attr_noexcept { + if (size <= MI_MEDIUM_OBJ_SIZE_MAX) { + return _mi_bin_size(mi_bin(size)); + } + else { + return _mi_align_up(size,_mi_os_page_size()); + } +} + +#if (MI_DEBUG>1) +static bool mi_page_queue_contains(mi_page_queue_t* queue, const mi_page_t* page) { + mi_assert_internal(page != NULL); + mi_page_t* list = queue->first; + while (list != NULL) { + mi_assert_internal(list->next == NULL || list->next->prev == list); + mi_assert_internal(list->prev == NULL || list->prev->next == list); + if (list == page) break; + list = list->next; + } + return (list == page); +} + +#endif + +#if (MI_DEBUG>1) +static bool mi_heap_contains_queue(const mi_heap_t* heap, const mi_page_queue_t* pq) { + return (pq >= &heap->pages[0] && pq <= &heap->pages[MI_BIN_FULL]); +} +#endif + +static mi_page_queue_t* mi_page_queue_of(const mi_page_t* page) { + uint8_t bin = (mi_page_is_in_full(page) ? MI_BIN_FULL : mi_bin(page->xblock_size)); + mi_heap_t* heap = mi_page_heap(page); + mi_assert_internal(heap != NULL && bin <= MI_BIN_FULL); + mi_page_queue_t* pq = &heap->pages[bin]; + mi_assert_internal(bin >= MI_BIN_HUGE || page->xblock_size == pq->block_size); + mi_assert_expensive(mi_page_queue_contains(pq, page)); + return pq; +} + +static mi_page_queue_t* mi_heap_page_queue_of(mi_heap_t* heap, const mi_page_t* page) { + uint8_t bin = (mi_page_is_in_full(page) ? MI_BIN_FULL : mi_bin(page->xblock_size)); + mi_assert_internal(bin <= MI_BIN_FULL); + mi_page_queue_t* pq = &heap->pages[bin]; + mi_assert_internal(mi_page_is_in_full(page) || page->xblock_size == pq->block_size); + return pq; +} + +// The current small page array is for efficiency and for each +// small size (up to 256) it points directly to the page for that +// size without having to compute the bin. This means when the +// current free page queue is updated for a small bin, we need to update a +// range of entries in `_mi_page_small_free`. +static inline void mi_heap_queue_first_update(mi_heap_t* heap, const mi_page_queue_t* pq) { + mi_assert_internal(mi_heap_contains_queue(heap,pq)); + size_t size = pq->block_size; + if (size > MI_SMALL_SIZE_MAX) return; + + mi_page_t* page = pq->first; + if (pq->first == NULL) page = (mi_page_t*)&_mi_page_empty; + + // find index in the right direct page array + size_t start; + size_t idx = _mi_wsize_from_size(size); + mi_page_t** pages_free = heap->pages_free_direct; + + if (pages_free[idx] == page) return; // already set + + // find start slot + if (idx<=1) { + start = 0; + } + else { + // find previous size; due to minimal alignment upto 3 previous bins may need to be skipped + uint8_t bin = mi_bin(size); + const mi_page_queue_t* prev = pq - 1; + while( bin == mi_bin(prev->block_size) && prev > &heap->pages[0]) { + prev--; + } + start = 1 + _mi_wsize_from_size(prev->block_size); + if (start > idx) start = idx; + } + + // set size range to the right page + mi_assert(start <= idx); + for (size_t sz = start; sz <= idx; sz++) { + pages_free[sz] = page; + } +} + +/* +static bool mi_page_queue_is_empty(mi_page_queue_t* queue) { + return (queue->first == NULL); +} +*/ + +static void mi_page_queue_remove(mi_page_queue_t* queue, mi_page_t* page) { + mi_assert_internal(page != NULL); + mi_assert_expensive(mi_page_queue_contains(queue, page)); + mi_assert_internal(page->xblock_size == queue->block_size || (page->xblock_size > MI_MEDIUM_OBJ_SIZE_MAX && mi_page_queue_is_huge(queue)) || (mi_page_is_in_full(page) && mi_page_queue_is_full(queue))); + mi_heap_t* heap = mi_page_heap(page); + + if (page->prev != NULL) page->prev->next = page->next; + if (page->next != NULL) page->next->prev = page->prev; + if (page == queue->last) queue->last = page->prev; + if (page == queue->first) { + queue->first = page->next; + // update first + mi_assert_internal(mi_heap_contains_queue(heap, queue)); + mi_heap_queue_first_update(heap,queue); + } + heap->page_count--; + page->next = NULL; + page->prev = NULL; + // mi_atomic_store_ptr_release(mi_atomic_cast(void*, &page->heap), NULL); + mi_page_set_in_full(page,false); +} + + +static void mi_page_queue_push(mi_heap_t* heap, mi_page_queue_t* queue, mi_page_t* page) { + mi_assert_internal(mi_page_heap(page) == heap); + mi_assert_internal(!mi_page_queue_contains(queue, page)); + #if MI_HUGE_PAGE_ABANDON + mi_assert_internal(_mi_page_segment(page)->kind != MI_SEGMENT_HUGE); + #endif + mi_assert_internal(page->xblock_size == queue->block_size || + (page->xblock_size > MI_MEDIUM_OBJ_SIZE_MAX) || + (mi_page_is_in_full(page) && mi_page_queue_is_full(queue))); + + mi_page_set_in_full(page, mi_page_queue_is_full(queue)); + // mi_atomic_store_ptr_release(mi_atomic_cast(void*, &page->heap), heap); + page->next = queue->first; + page->prev = NULL; + if (queue->first != NULL) { + mi_assert_internal(queue->first->prev == NULL); + queue->first->prev = page; + queue->first = page; + } + else { + queue->first = queue->last = page; + } + + // update direct + mi_heap_queue_first_update(heap, queue); + heap->page_count++; +} + + +static void mi_page_queue_enqueue_from(mi_page_queue_t* to, mi_page_queue_t* from, mi_page_t* page) { + mi_assert_internal(page != NULL); + mi_assert_expensive(mi_page_queue_contains(from, page)); + mi_assert_expensive(!mi_page_queue_contains(to, page)); + + mi_assert_internal((page->xblock_size == to->block_size && page->xblock_size == from->block_size) || + (page->xblock_size == to->block_size && mi_page_queue_is_full(from)) || + (page->xblock_size == from->block_size && mi_page_queue_is_full(to)) || + (page->xblock_size > MI_LARGE_OBJ_SIZE_MAX && mi_page_queue_is_huge(to)) || + (page->xblock_size > MI_LARGE_OBJ_SIZE_MAX && mi_page_queue_is_full(to))); + + mi_heap_t* heap = mi_page_heap(page); + if (page->prev != NULL) page->prev->next = page->next; + if (page->next != NULL) page->next->prev = page->prev; + if (page == from->last) from->last = page->prev; + if (page == from->first) { + from->first = page->next; + // update first + mi_assert_internal(mi_heap_contains_queue(heap, from)); + mi_heap_queue_first_update(heap, from); + } + + page->prev = to->last; + page->next = NULL; + if (to->last != NULL) { + mi_assert_internal(heap == mi_page_heap(to->last)); + to->last->next = page; + to->last = page; + } + else { + to->first = page; + to->last = page; + mi_heap_queue_first_update(heap, to); + } + + mi_page_set_in_full(page, mi_page_queue_is_full(to)); +} + +// Only called from `mi_heap_absorb`. +size_t _mi_page_queue_append(mi_heap_t* heap, mi_page_queue_t* pq, mi_page_queue_t* append) { + mi_assert_internal(mi_heap_contains_queue(heap,pq)); + mi_assert_internal(pq->block_size == append->block_size); + + if (append->first==NULL) return 0; + + // set append pages to new heap and count + size_t count = 0; + for (mi_page_t* page = append->first; page != NULL; page = page->next) { + // inline `mi_page_set_heap` to avoid wrong assertion during absorption; + // in this case it is ok to be delayed freeing since both "to" and "from" heap are still alive. + mi_atomic_store_release(&page->xheap, (uintptr_t)heap); + // set the flag to delayed free (not overriding NEVER_DELAYED_FREE) which has as a + // side effect that it spins until any DELAYED_FREEING is finished. This ensures + // that after appending only the new heap will be used for delayed free operations. + _mi_page_use_delayed_free(page, MI_USE_DELAYED_FREE, false); + count++; + } + + if (pq->last==NULL) { + // take over afresh + mi_assert_internal(pq->first==NULL); + pq->first = append->first; + pq->last = append->last; + mi_heap_queue_first_update(heap, pq); + } + else { + // append to end + mi_assert_internal(pq->last!=NULL); + mi_assert_internal(append->first!=NULL); + pq->last->next = append->first; + append->first->prev = pq->last; + pq->last = append->last; + } + return count; +} diff --git a/compat/mimalloc/page.c b/compat/mimalloc/page.c new file mode 100644 index 00000000000000..211204aa79e59d --- /dev/null +++ b/compat/mimalloc/page.c @@ -0,0 +1,939 @@ +/*---------------------------------------------------------------------------- +Copyright (c) 2018-2020, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +/* ----------------------------------------------------------- + The core of the allocator. Every segment contains + pages of a certain block size. The main function + exported is `mi_malloc_generic`. +----------------------------------------------------------- */ + +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" + +/* ----------------------------------------------------------- + Definition of page queues for each block size +----------------------------------------------------------- */ + +#define MI_IN_PAGE_C +#include "page-queue.c" +#undef MI_IN_PAGE_C + + +/* ----------------------------------------------------------- + Page helpers +----------------------------------------------------------- */ + +// Index a block in a page +static inline mi_block_t* mi_page_block_at(const mi_page_t* page, void* page_start, size_t block_size, size_t i) { + MI_UNUSED(page); + mi_assert_internal(page != NULL); + mi_assert_internal(i <= page->reserved); + return (mi_block_t*)((uint8_t*)page_start + (i * block_size)); +} + +static void mi_page_init(mi_heap_t* heap, mi_page_t* page, size_t size, mi_tld_t* tld); +static void mi_page_extend_free(mi_heap_t* heap, mi_page_t* page, mi_tld_t* tld); + +#if (MI_DEBUG>=3) +static size_t mi_page_list_count(mi_page_t* page, mi_block_t* head) { + size_t count = 0; + while (head != NULL) { + mi_assert_internal(page == _mi_ptr_page(head)); + count++; + head = mi_block_next(page, head); + } + return count; +} + +/* +// Start of the page available memory +static inline uint8_t* mi_page_area(const mi_page_t* page) { + return _mi_page_start(_mi_page_segment(page), page, NULL); +} +*/ + +static bool mi_page_list_is_valid(mi_page_t* page, mi_block_t* p) { + size_t psize; + uint8_t* page_area = _mi_page_start(_mi_page_segment(page), page, &psize); + mi_block_t* start = (mi_block_t*)page_area; + mi_block_t* end = (mi_block_t*)(page_area + psize); + while(p != NULL) { + if (p < start || p >= end) return false; + p = mi_block_next(page, p); + } +#if MI_DEBUG>3 // generally too expensive to check this + if (page->free_is_zero) { + const size_t ubsize = mi_page_usable_block_size(page); + for (mi_block_t* block = page->free; block != NULL; block = mi_block_next(page, block)) { + mi_assert_expensive(mi_mem_is_zero(block + 1, ubsize - sizeof(mi_block_t))); + } + } +#endif + return true; +} + +static bool mi_page_is_valid_init(mi_page_t* page) { + mi_assert_internal(page->xblock_size > 0); + mi_assert_internal(page->used <= page->capacity); + mi_assert_internal(page->capacity <= page->reserved); + + mi_segment_t* segment = _mi_page_segment(page); + uint8_t* start = _mi_page_start(segment,page,NULL); + mi_assert_internal(start == _mi_segment_page_start(segment,page,NULL)); + //const size_t bsize = mi_page_block_size(page); + //mi_assert_internal(start + page->capacity*page->block_size == page->top); + + mi_assert_internal(mi_page_list_is_valid(page,page->free)); + mi_assert_internal(mi_page_list_is_valid(page,page->local_free)); + + #if MI_DEBUG>3 // generally too expensive to check this + if (page->free_is_zero) { + const size_t ubsize = mi_page_usable_block_size(page); + for(mi_block_t* block = page->free; block != NULL; block = mi_block_next(page,block)) { + mi_assert_expensive(mi_mem_is_zero(block + 1, ubsize - sizeof(mi_block_t))); + } + } + #endif + + #if !MI_TRACK_ENABLED && !MI_TSAN + mi_block_t* tfree = mi_page_thread_free(page); + mi_assert_internal(mi_page_list_is_valid(page, tfree)); + //size_t tfree_count = mi_page_list_count(page, tfree); + //mi_assert_internal(tfree_count <= page->thread_freed + 1); + #endif + + size_t free_count = mi_page_list_count(page, page->free) + mi_page_list_count(page, page->local_free); + mi_assert_internal(page->used + free_count == page->capacity); + + return true; +} + +extern bool _mi_process_is_initialized; // has mi_process_init been called? + +bool _mi_page_is_valid(mi_page_t* page) { + mi_assert_internal(mi_page_is_valid_init(page)); + #if MI_SECURE + mi_assert_internal(page->keys[0] != 0); + #endif + if (mi_page_heap(page)!=NULL) { + mi_segment_t* segment = _mi_page_segment(page); + + mi_assert_internal(!_mi_process_is_initialized || segment->thread_id==0 || segment->thread_id == mi_page_heap(page)->thread_id); + #if MI_HUGE_PAGE_ABANDON + if (segment->kind != MI_SEGMENT_HUGE) + #endif + { + mi_page_queue_t* pq = mi_page_queue_of(page); + mi_assert_internal(mi_page_queue_contains(pq, page)); + mi_assert_internal(pq->block_size==mi_page_block_size(page) || mi_page_block_size(page) > MI_MEDIUM_OBJ_SIZE_MAX || mi_page_is_in_full(page)); + mi_assert_internal(mi_heap_contains_queue(mi_page_heap(page),pq)); + } + } + return true; +} +#endif + +void _mi_page_use_delayed_free(mi_page_t* page, mi_delayed_t delay, bool override_never) { + while (!_mi_page_try_use_delayed_free(page, delay, override_never)) { + mi_atomic_yield(); + } +} + +bool _mi_page_try_use_delayed_free(mi_page_t* page, mi_delayed_t delay, bool override_never) { + mi_thread_free_t tfreex; + mi_delayed_t old_delay; + mi_thread_free_t tfree; + size_t yield_count = 0; + do { + tfree = mi_atomic_load_acquire(&page->xthread_free); // note: must acquire as we can break/repeat this loop and not do a CAS; + tfreex = mi_tf_set_delayed(tfree, delay); + old_delay = mi_tf_delayed(tfree); + if mi_unlikely(old_delay == MI_DELAYED_FREEING) { + if (yield_count >= 4) return false; // give up after 4 tries + yield_count++; + mi_atomic_yield(); // delay until outstanding MI_DELAYED_FREEING are done. + // tfree = mi_tf_set_delayed(tfree, MI_NO_DELAYED_FREE); // will cause CAS to busy fail + } + else if (delay == old_delay) { + break; // avoid atomic operation if already equal + } + else if (!override_never && old_delay == MI_NEVER_DELAYED_FREE) { + break; // leave never-delayed flag set + } + } while ((old_delay == MI_DELAYED_FREEING) || + !mi_atomic_cas_weak_release(&page->xthread_free, &tfree, tfreex)); + + return true; // success +} + +/* ----------------------------------------------------------- + Page collect the `local_free` and `thread_free` lists +----------------------------------------------------------- */ + +// Collect the local `thread_free` list using an atomic exchange. +// Note: The exchange must be done atomically as this is used right after +// moving to the full list in `mi_page_collect_ex` and we need to +// ensure that there was no race where the page became unfull just before the move. +static void _mi_page_thread_free_collect(mi_page_t* page) +{ + mi_block_t* head; + mi_thread_free_t tfreex; + mi_thread_free_t tfree = mi_atomic_load_relaxed(&page->xthread_free); + do { + head = mi_tf_block(tfree); + tfreex = mi_tf_set_block(tfree,NULL); + } while (!mi_atomic_cas_weak_acq_rel(&page->xthread_free, &tfree, tfreex)); + + // return if the list is empty + if (head == NULL) return; + + // find the tail -- also to get a proper count (without data races) + uint32_t max_count = page->capacity; // cannot collect more than capacity + uint32_t count = 1; + mi_block_t* tail = head; + mi_block_t* next; + while ((next = mi_block_next(page,tail)) != NULL && count <= max_count) { + count++; + tail = next; + } + // if `count > max_count` there was a memory corruption (possibly infinite list due to double multi-threaded free) + if (count > max_count) { + _mi_error_message(EFAULT, "corrupted thread-free list\n"); + return; // the thread-free items cannot be freed + } + + // and append the current local free list + mi_block_set_next(page,tail, page->local_free); + page->local_free = head; + + // update counts now + page->used -= count; +} + +void _mi_page_free_collect(mi_page_t* page, bool force) { + mi_assert_internal(page!=NULL); + + // collect the thread free list + if (force || mi_page_thread_free(page) != NULL) { // quick test to avoid an atomic operation + _mi_page_thread_free_collect(page); + } + + // and the local free list + if (page->local_free != NULL) { + if mi_likely(page->free == NULL) { + // usual case + page->free = page->local_free; + page->local_free = NULL; + page->free_is_zero = false; + } + else if (force) { + // append -- only on shutdown (force) as this is a linear operation + mi_block_t* tail = page->local_free; + mi_block_t* next; + while ((next = mi_block_next(page, tail)) != NULL) { + tail = next; + } + mi_block_set_next(page, tail, page->free); + page->free = page->local_free; + page->local_free = NULL; + page->free_is_zero = false; + } + } + + mi_assert_internal(!force || page->local_free == NULL); +} + + + +/* ----------------------------------------------------------- + Page fresh and retire +----------------------------------------------------------- */ + +// called from segments when reclaiming abandoned pages +void _mi_page_reclaim(mi_heap_t* heap, mi_page_t* page) { + mi_assert_expensive(mi_page_is_valid_init(page)); + + mi_assert_internal(mi_page_heap(page) == heap); + mi_assert_internal(mi_page_thread_free_flag(page) != MI_NEVER_DELAYED_FREE); + #if MI_HUGE_PAGE_ABANDON + mi_assert_internal(_mi_page_segment(page)->kind != MI_SEGMENT_HUGE); + #endif + + // TODO: push on full queue immediately if it is full? + mi_page_queue_t* pq = mi_page_queue(heap, mi_page_block_size(page)); + mi_page_queue_push(heap, pq, page); + mi_assert_expensive(_mi_page_is_valid(page)); +} + +// allocate a fresh page from a segment +static mi_page_t* mi_page_fresh_alloc(mi_heap_t* heap, mi_page_queue_t* pq, size_t block_size, size_t page_alignment) { + #if !MI_HUGE_PAGE_ABANDON + mi_assert_internal(pq != NULL); + mi_assert_internal(mi_heap_contains_queue(heap, pq)); + mi_assert_internal(page_alignment > 0 || block_size > MI_MEDIUM_OBJ_SIZE_MAX || block_size == pq->block_size); + #endif + mi_page_t* page = _mi_segment_page_alloc(heap, block_size, page_alignment, &heap->tld->segments, &heap->tld->os); + if (page == NULL) { + // this may be out-of-memory, or an abandoned page was reclaimed (and in our queue) + return NULL; + } + mi_assert_internal(page_alignment >0 || block_size > MI_MEDIUM_OBJ_SIZE_MAX || _mi_page_segment(page)->kind != MI_SEGMENT_HUGE); + mi_assert_internal(pq!=NULL || page->xblock_size != 0); + mi_assert_internal(pq!=NULL || mi_page_block_size(page) >= block_size); + // a fresh page was found, initialize it + const size_t full_block_size = ((pq == NULL || mi_page_queue_is_huge(pq)) ? mi_page_block_size(page) : block_size); // see also: mi_segment_huge_page_alloc + mi_assert_internal(full_block_size >= block_size); + mi_page_init(heap, page, full_block_size, heap->tld); + mi_heap_stat_increase(heap, pages, 1); + if (pq != NULL) { mi_page_queue_push(heap, pq, page); } + mi_assert_expensive(_mi_page_is_valid(page)); + return page; +} + +// Get a fresh page to use +static mi_page_t* mi_page_fresh(mi_heap_t* heap, mi_page_queue_t* pq) { + mi_assert_internal(mi_heap_contains_queue(heap, pq)); + mi_page_t* page = mi_page_fresh_alloc(heap, pq, pq->block_size, 0); + if (page==NULL) return NULL; + mi_assert_internal(pq->block_size==mi_page_block_size(page)); + mi_assert_internal(pq==mi_page_queue(heap, mi_page_block_size(page))); + return page; +} + +/* ----------------------------------------------------------- + Do any delayed frees + (put there by other threads if they deallocated in a full page) +----------------------------------------------------------- */ +void _mi_heap_delayed_free_all(mi_heap_t* heap) { + while (!_mi_heap_delayed_free_partial(heap)) { + mi_atomic_yield(); + } +} + +// returns true if all delayed frees were processed +bool _mi_heap_delayed_free_partial(mi_heap_t* heap) { + // take over the list (note: no atomic exchange since it is often NULL) + mi_block_t* block = mi_atomic_load_ptr_relaxed(mi_block_t, &heap->thread_delayed_free); + while (block != NULL && !mi_atomic_cas_ptr_weak_acq_rel(mi_block_t, &heap->thread_delayed_free, &block, NULL)) { /* nothing */ }; + bool all_freed = true; + + // and free them all + while(block != NULL) { + mi_block_t* next = mi_block_nextx(heap,block, heap->keys); + // use internal free instead of regular one to keep stats etc correct + if (!_mi_free_delayed_block(block)) { + // we might already start delayed freeing while another thread has not yet + // reset the delayed_freeing flag; in that case delay it further by reinserting the current block + // into the delayed free list + all_freed = false; + mi_block_t* dfree = mi_atomic_load_ptr_relaxed(mi_block_t, &heap->thread_delayed_free); + do { + mi_block_set_nextx(heap, block, dfree, heap->keys); + } while (!mi_atomic_cas_ptr_weak_release(mi_block_t,&heap->thread_delayed_free, &dfree, block)); + } + block = next; + } + return all_freed; +} + +/* ----------------------------------------------------------- + Unfull, abandon, free and retire +----------------------------------------------------------- */ + +// Move a page from the full list back to a regular list +void _mi_page_unfull(mi_page_t* page) { + mi_assert_internal(page != NULL); + mi_assert_expensive(_mi_page_is_valid(page)); + mi_assert_internal(mi_page_is_in_full(page)); + if (!mi_page_is_in_full(page)) return; + + mi_heap_t* heap = mi_page_heap(page); + mi_page_queue_t* pqfull = &heap->pages[MI_BIN_FULL]; + mi_page_set_in_full(page, false); // to get the right queue + mi_page_queue_t* pq = mi_heap_page_queue_of(heap, page); + mi_page_set_in_full(page, true); + mi_page_queue_enqueue_from(pq, pqfull, page); +} + +static void mi_page_to_full(mi_page_t* page, mi_page_queue_t* pq) { + mi_assert_internal(pq == mi_page_queue_of(page)); + mi_assert_internal(!mi_page_immediate_available(page)); + mi_assert_internal(!mi_page_is_in_full(page)); + + if (mi_page_is_in_full(page)) return; + mi_page_queue_enqueue_from(&mi_page_heap(page)->pages[MI_BIN_FULL], pq, page); + _mi_page_free_collect(page,false); // try to collect right away in case another thread freed just before MI_USE_DELAYED_FREE was set +} + + +// Abandon a page with used blocks at the end of a thread. +// Note: only call if it is ensured that no references exist from +// the `page->heap->thread_delayed_free` into this page. +// Currently only called through `mi_heap_collect_ex` which ensures this. +void _mi_page_abandon(mi_page_t* page, mi_page_queue_t* pq) { + mi_assert_internal(page != NULL); + mi_assert_expensive(_mi_page_is_valid(page)); + mi_assert_internal(pq == mi_page_queue_of(page)); + mi_assert_internal(mi_page_heap(page) != NULL); + + mi_heap_t* pheap = mi_page_heap(page); + + // remove from our page list + mi_segments_tld_t* segments_tld = &pheap->tld->segments; + mi_page_queue_remove(pq, page); + + // page is no longer associated with our heap + mi_assert_internal(mi_page_thread_free_flag(page)==MI_NEVER_DELAYED_FREE); + mi_page_set_heap(page, NULL); + +#if (MI_DEBUG>1) && !MI_TRACK_ENABLED + // check there are no references left.. + for (mi_block_t* block = (mi_block_t*)pheap->thread_delayed_free; block != NULL; block = mi_block_nextx(pheap, block, pheap->keys)) { + mi_assert_internal(_mi_ptr_page(block) != page); + } +#endif + + // and abandon it + mi_assert_internal(mi_page_heap(page) == NULL); + _mi_segment_page_abandon(page,segments_tld); +} + + +// Free a page with no more free blocks +void _mi_page_free(mi_page_t* page, mi_page_queue_t* pq, bool force) { + mi_assert_internal(page != NULL); + mi_assert_expensive(_mi_page_is_valid(page)); + mi_assert_internal(pq == mi_page_queue_of(page)); + mi_assert_internal(mi_page_all_free(page)); + mi_assert_internal(mi_page_thread_free_flag(page)!=MI_DELAYED_FREEING); + + // no more aligned blocks in here + mi_page_set_has_aligned(page, false); + + mi_heap_t* heap = mi_page_heap(page); + + // remove from the page list + // (no need to do _mi_heap_delayed_free first as all blocks are already free) + mi_segments_tld_t* segments_tld = &heap->tld->segments; + mi_page_queue_remove(pq, page); + + // and free it + mi_page_set_heap(page,NULL); + _mi_segment_page_free(page, force, segments_tld); +} + +// Retire parameters +#define MI_MAX_RETIRE_SIZE (MI_MEDIUM_OBJ_SIZE_MAX) +#define MI_RETIRE_CYCLES (16) + +// Retire a page with no more used blocks +// Important to not retire too quickly though as new +// allocations might coming. +// Note: called from `mi_free` and benchmarks often +// trigger this due to freeing everything and then +// allocating again so careful when changing this. +void _mi_page_retire(mi_page_t* page) mi_attr_noexcept { + mi_assert_internal(page != NULL); + mi_assert_expensive(_mi_page_is_valid(page)); + mi_assert_internal(mi_page_all_free(page)); + + mi_page_set_has_aligned(page, false); + + // don't retire too often.. + // (or we end up retiring and re-allocating most of the time) + // NOTE: refine this more: we should not retire if this + // is the only page left with free blocks. It is not clear + // how to check this efficiently though... + // for now, we don't retire if it is the only page left of this size class. + mi_page_queue_t* pq = mi_page_queue_of(page); + if mi_likely(page->xblock_size <= MI_MAX_RETIRE_SIZE && !mi_page_queue_is_special(pq)) { // not too large && not full or huge queue? + if (pq->last==page && pq->first==page) { // the only page in the queue? + mi_stat_counter_increase(_mi_stats_main.page_no_retire,1); + page->retire_expire = 1 + (page->xblock_size <= MI_SMALL_OBJ_SIZE_MAX ? MI_RETIRE_CYCLES : MI_RETIRE_CYCLES/4); + mi_heap_t* heap = mi_page_heap(page); + mi_assert_internal(pq >= heap->pages); + const size_t index = pq - heap->pages; + mi_assert_internal(index < MI_BIN_FULL && index < MI_BIN_HUGE); + if (index < heap->page_retired_min) heap->page_retired_min = index; + if (index > heap->page_retired_max) heap->page_retired_max = index; + mi_assert_internal(mi_page_all_free(page)); + return; // dont't free after all + } + } + _mi_page_free(page, pq, false); +} + +// free retired pages: we don't need to look at the entire queues +// since we only retire pages that are at the head position in a queue. +void _mi_heap_collect_retired(mi_heap_t* heap, bool force) { + size_t min = MI_BIN_FULL; + size_t max = 0; + for(size_t bin = heap->page_retired_min; bin <= heap->page_retired_max; bin++) { + mi_page_queue_t* pq = &heap->pages[bin]; + mi_page_t* page = pq->first; + if (page != NULL && page->retire_expire != 0) { + if (mi_page_all_free(page)) { + page->retire_expire--; + if (force || page->retire_expire == 0) { + _mi_page_free(pq->first, pq, force); + } + else { + // keep retired, update min/max + if (bin < min) min = bin; + if (bin > max) max = bin; + } + } + else { + page->retire_expire = 0; + } + } + } + heap->page_retired_min = min; + heap->page_retired_max = max; +} + + +/* ----------------------------------------------------------- + Initialize the initial free list in a page. + In secure mode we initialize a randomized list by + alternating between slices. +----------------------------------------------------------- */ + +#define MI_MAX_SLICE_SHIFT (6) // at most 64 slices +#define MI_MAX_SLICES (1UL << MI_MAX_SLICE_SHIFT) +#define MI_MIN_SLICES (2) + +static void mi_page_free_list_extend_secure(mi_heap_t* const heap, mi_page_t* const page, const size_t bsize, const size_t extend, mi_stats_t* const stats) { + MI_UNUSED(stats); + #if (MI_SECURE<=2) + mi_assert_internal(page->free == NULL); + mi_assert_internal(page->local_free == NULL); + #endif + mi_assert_internal(page->capacity + extend <= page->reserved); + mi_assert_internal(bsize == mi_page_block_size(page)); + void* const page_area = _mi_page_start(_mi_page_segment(page), page, NULL); + + // initialize a randomized free list + // set up `slice_count` slices to alternate between + size_t shift = MI_MAX_SLICE_SHIFT; + while ((extend >> shift) == 0) { + shift--; + } + const size_t slice_count = (size_t)1U << shift; + const size_t slice_extend = extend / slice_count; + mi_assert_internal(slice_extend >= 1); + mi_block_t* blocks[MI_MAX_SLICES]; // current start of the slice + size_t counts[MI_MAX_SLICES]; // available objects in the slice + for (size_t i = 0; i < slice_count; i++) { + blocks[i] = mi_page_block_at(page, page_area, bsize, page->capacity + i*slice_extend); + counts[i] = slice_extend; + } + counts[slice_count-1] += (extend % slice_count); // final slice holds the modulus too (todo: distribute evenly?) + + // and initialize the free list by randomly threading through them + // set up first element + const uintptr_t r = _mi_heap_random_next(heap); + size_t current = r % slice_count; + counts[current]--; + mi_block_t* const free_start = blocks[current]; + // and iterate through the rest; use `random_shuffle` for performance + uintptr_t rnd = _mi_random_shuffle(r|1); // ensure not 0 + for (size_t i = 1; i < extend; i++) { + // call random_shuffle only every INTPTR_SIZE rounds + const size_t round = i%MI_INTPTR_SIZE; + if (round == 0) rnd = _mi_random_shuffle(rnd); + // select a random next slice index + size_t next = ((rnd >> 8*round) & (slice_count-1)); + while (counts[next]==0) { // ensure it still has space + next++; + if (next==slice_count) next = 0; + } + // and link the current block to it + counts[next]--; + mi_block_t* const block = blocks[current]; + blocks[current] = (mi_block_t*)((uint8_t*)block + bsize); // bump to the following block + mi_block_set_next(page, block, blocks[next]); // and set next; note: we may have `current == next` + current = next; + } + // prepend to the free list (usually NULL) + mi_block_set_next(page, blocks[current], page->free); // end of the list + page->free = free_start; +} + +static mi_decl_noinline void mi_page_free_list_extend( mi_page_t* const page, const size_t bsize, const size_t extend, mi_stats_t* const stats) +{ + MI_UNUSED(stats); + #if (MI_SECURE <= 2) + mi_assert_internal(page->free == NULL); + mi_assert_internal(page->local_free == NULL); + #endif + mi_assert_internal(page->capacity + extend <= page->reserved); + mi_assert_internal(bsize == mi_page_block_size(page)); + void* const page_area = _mi_page_start(_mi_page_segment(page), page, NULL ); + + mi_block_t* const start = mi_page_block_at(page, page_area, bsize, page->capacity); + + // initialize a sequential free list + mi_block_t* const last = mi_page_block_at(page, page_area, bsize, page->capacity + extend - 1); + mi_block_t* block = start; + while(block <= last) { + mi_block_t* next = (mi_block_t*)((uint8_t*)block + bsize); + mi_block_set_next(page,block,next); + block = next; + } + // prepend to free list (usually `NULL`) + mi_block_set_next(page, last, page->free); + page->free = start; +} + +/* ----------------------------------------------------------- + Page initialize and extend the capacity +----------------------------------------------------------- */ + +#define MI_MAX_EXTEND_SIZE (4*1024) // heuristic, one OS page seems to work well. +#if (MI_SECURE>0) +#define MI_MIN_EXTEND (8*MI_SECURE) // extend at least by this many +#else +#define MI_MIN_EXTEND (4) +#endif + +// Extend the capacity (up to reserved) by initializing a free list +// We do at most `MI_MAX_EXTEND` to avoid touching too much memory +// Note: we also experimented with "bump" allocation on the first +// allocations but this did not speed up any benchmark (due to an +// extra test in malloc? or cache effects?) +static void mi_page_extend_free(mi_heap_t* heap, mi_page_t* page, mi_tld_t* tld) { + MI_UNUSED(tld); + mi_assert_expensive(mi_page_is_valid_init(page)); + #if (MI_SECURE<=2) + mi_assert(page->free == NULL); + mi_assert(page->local_free == NULL); + if (page->free != NULL) return; + #endif + if (page->capacity >= page->reserved) return; + + size_t page_size; + _mi_page_start(_mi_page_segment(page), page, &page_size); + mi_stat_counter_increase(tld->stats.pages_extended, 1); + + // calculate the extend count + const size_t bsize = (page->xblock_size < MI_HUGE_BLOCK_SIZE ? page->xblock_size : page_size); + size_t extend = page->reserved - page->capacity; + mi_assert_internal(extend > 0); + + size_t max_extend = (bsize >= MI_MAX_EXTEND_SIZE ? MI_MIN_EXTEND : MI_MAX_EXTEND_SIZE/(uint32_t)bsize); + if (max_extend < MI_MIN_EXTEND) { max_extend = MI_MIN_EXTEND; } + mi_assert_internal(max_extend > 0); + + if (extend > max_extend) { + // ensure we don't touch memory beyond the page to reduce page commit. + // the `lean` benchmark tests this. Going from 1 to 8 increases rss by 50%. + extend = max_extend; + } + + mi_assert_internal(extend > 0 && extend + page->capacity <= page->reserved); + mi_assert_internal(extend < (1UL<<16)); + + // and append the extend the free list + if (extend < MI_MIN_SLICES || MI_SECURE==0) { //!mi_option_is_enabled(mi_option_secure)) { + mi_page_free_list_extend(page, bsize, extend, &tld->stats ); + } + else { + mi_page_free_list_extend_secure(heap, page, bsize, extend, &tld->stats); + } + // enable the new free list + page->capacity += (uint16_t)extend; + mi_stat_increase(tld->stats.page_committed, extend * bsize); + mi_assert_expensive(mi_page_is_valid_init(page)); +} + +// Initialize a fresh page +static void mi_page_init(mi_heap_t* heap, mi_page_t* page, size_t block_size, mi_tld_t* tld) { + mi_assert(page != NULL); + mi_segment_t* segment = _mi_page_segment(page); + mi_assert(segment != NULL); + mi_assert_internal(block_size > 0); + // set fields + mi_page_set_heap(page, heap); + page->xblock_size = (block_size < MI_HUGE_BLOCK_SIZE ? (uint32_t)block_size : MI_HUGE_BLOCK_SIZE); // initialize before _mi_segment_page_start + size_t page_size; + const void* page_start = _mi_segment_page_start(segment, page, &page_size); + MI_UNUSED(page_start); + mi_track_mem_noaccess(page_start,page_size); + mi_assert_internal(mi_page_block_size(page) <= page_size); + mi_assert_internal(page_size <= page->slice_count*MI_SEGMENT_SLICE_SIZE); + mi_assert_internal(page_size / block_size < (1L<<16)); + page->reserved = (uint16_t)(page_size / block_size); + mi_assert_internal(page->reserved > 0); + #if (MI_PADDING || MI_ENCODE_FREELIST) + page->keys[0] = _mi_heap_random_next(heap); + page->keys[1] = _mi_heap_random_next(heap); + #endif + page->free_is_zero = page->is_zero_init; + #if MI_DEBUG>2 + if (page->is_zero_init) { + mi_track_mem_defined(page_start, page_size); + mi_assert_expensive(mi_mem_is_zero(page_start, page_size)); + } + #endif + + mi_assert_internal(page->is_committed); + mi_assert_internal(page->capacity == 0); + mi_assert_internal(page->free == NULL); + mi_assert_internal(page->used == 0); + mi_assert_internal(page->xthread_free == 0); + mi_assert_internal(page->next == NULL); + mi_assert_internal(page->prev == NULL); + mi_assert_internal(page->retire_expire == 0); + mi_assert_internal(!mi_page_has_aligned(page)); + #if (MI_PADDING || MI_ENCODE_FREELIST) + mi_assert_internal(page->keys[0] != 0); + mi_assert_internal(page->keys[1] != 0); + #endif + mi_assert_expensive(mi_page_is_valid_init(page)); + + // initialize an initial free list + mi_page_extend_free(heap,page,tld); + mi_assert(mi_page_immediate_available(page)); +} + + +/* ----------------------------------------------------------- + Find pages with free blocks +-------------------------------------------------------------*/ + +// Find a page with free blocks of `page->block_size`. +static mi_page_t* mi_page_queue_find_free_ex(mi_heap_t* heap, mi_page_queue_t* pq, bool first_try) +{ + // search through the pages in "next fit" order + #if MI_STAT + size_t count = 0; + #endif + mi_page_t* page = pq->first; + while (page != NULL) + { + mi_page_t* next = page->next; // remember next + #if MI_STAT + count++; + #endif + + // 0. collect freed blocks by us and other threads + _mi_page_free_collect(page, false); + + // 1. if the page contains free blocks, we are done + if (mi_page_immediate_available(page)) { + break; // pick this one + } + + // 2. Try to extend + if (page->capacity < page->reserved) { + mi_page_extend_free(heap, page, heap->tld); + mi_assert_internal(mi_page_immediate_available(page)); + break; + } + + // 3. If the page is completely full, move it to the `mi_pages_full` + // queue so we don't visit long-lived pages too often. + mi_assert_internal(!mi_page_is_in_full(page) && !mi_page_immediate_available(page)); + mi_page_to_full(page, pq); + + page = next; + } // for each page + + mi_heap_stat_counter_increase(heap, searches, count); + + if (page == NULL) { + _mi_heap_collect_retired(heap, false); // perhaps make a page available? + page = mi_page_fresh(heap, pq); + if (page == NULL && first_try) { + // out-of-memory _or_ an abandoned page with free blocks was reclaimed, try once again + page = mi_page_queue_find_free_ex(heap, pq, false); + } + } + else { + mi_assert(pq->first == page); + page->retire_expire = 0; + } + mi_assert_internal(page == NULL || mi_page_immediate_available(page)); + return page; +} + + + +// Find a page with free blocks of `size`. +static inline mi_page_t* mi_find_free_page(mi_heap_t* heap, size_t size) { + mi_page_queue_t* pq = mi_page_queue(heap,size); + mi_page_t* page = pq->first; + if (page != NULL) { + #if (MI_SECURE>=3) // in secure mode, we extend half the time to increase randomness + if (page->capacity < page->reserved && ((_mi_heap_random_next(heap) & 1) == 1)) { + mi_page_extend_free(heap, page, heap->tld); + mi_assert_internal(mi_page_immediate_available(page)); + } + else + #endif + { + _mi_page_free_collect(page,false); + } + + if (mi_page_immediate_available(page)) { + page->retire_expire = 0; + return page; // fast path + } + } + return mi_page_queue_find_free_ex(heap, pq, true); +} + + +/* ----------------------------------------------------------- + Users can register a deferred free function called + when the `free` list is empty. Since the `local_free` + is separate this is deterministically called after + a certain number of allocations. +----------------------------------------------------------- */ + +static mi_deferred_free_fun* volatile deferred_free = NULL; +static _Atomic(void*) deferred_arg; // = NULL + +void _mi_deferred_free(mi_heap_t* heap, bool force) { + heap->tld->heartbeat++; + if (deferred_free != NULL && !heap->tld->recurse) { + heap->tld->recurse = true; + deferred_free(force, heap->tld->heartbeat, mi_atomic_load_ptr_relaxed(void,&deferred_arg)); + heap->tld->recurse = false; + } +} + +void mi_register_deferred_free(mi_deferred_free_fun* fn, void* arg) mi_attr_noexcept { + deferred_free = fn; + mi_atomic_store_ptr_release(void,&deferred_arg, arg); +} + + +/* ----------------------------------------------------------- + General allocation +----------------------------------------------------------- */ + +// Large and huge page allocation. +// Huge pages are allocated directly without being in a queue. +// Because huge pages contain just one block, and the segment contains +// just that page, we always treat them as abandoned and any thread +// that frees the block can free the whole page and segment directly. +// Huge pages are also use if the requested alignment is very large (> MI_ALIGNMENT_MAX). +static mi_page_t* mi_large_huge_page_alloc(mi_heap_t* heap, size_t size, size_t page_alignment) { + size_t block_size = _mi_os_good_alloc_size(size); + mi_assert_internal(mi_bin(block_size) == MI_BIN_HUGE || page_alignment > 0); + bool is_huge = (block_size > MI_LARGE_OBJ_SIZE_MAX || page_alignment > 0); + #if MI_HUGE_PAGE_ABANDON + mi_page_queue_t* pq = (is_huge ? NULL : mi_page_queue(heap, block_size)); + #else + mi_page_queue_t* pq = mi_page_queue(heap, is_huge ? MI_HUGE_BLOCK_SIZE : block_size); // not block_size as that can be low if the page_alignment > 0 + mi_assert_internal(!is_huge || mi_page_queue_is_huge(pq)); + #endif + mi_page_t* page = mi_page_fresh_alloc(heap, pq, block_size, page_alignment); + if (page != NULL) { + mi_assert_internal(mi_page_immediate_available(page)); + + if (is_huge) { + mi_assert_internal(_mi_page_segment(page)->kind == MI_SEGMENT_HUGE); + mi_assert_internal(_mi_page_segment(page)->used==1); + #if MI_HUGE_PAGE_ABANDON + mi_assert_internal(_mi_page_segment(page)->thread_id==0); // abandoned, not in the huge queue + mi_page_set_heap(page, NULL); + #endif + } + else { + mi_assert_internal(_mi_page_segment(page)->kind != MI_SEGMENT_HUGE); + } + + const size_t bsize = mi_page_usable_block_size(page); // note: not `mi_page_block_size` to account for padding + if (bsize <= MI_LARGE_OBJ_SIZE_MAX) { + mi_heap_stat_increase(heap, large, bsize); + mi_heap_stat_counter_increase(heap, large_count, 1); + } + else { + mi_heap_stat_increase(heap, huge, bsize); + mi_heap_stat_counter_increase(heap, huge_count, 1); + } + } + return page; +} + + +// Allocate a page +// Note: in debug mode the size includes MI_PADDING_SIZE and might have overflowed. +static mi_page_t* mi_find_page(mi_heap_t* heap, size_t size, size_t huge_alignment) mi_attr_noexcept { + // huge allocation? + const size_t req_size = size - MI_PADDING_SIZE; // correct for padding_size in case of an overflow on `size` + if mi_unlikely(req_size > (MI_MEDIUM_OBJ_SIZE_MAX - MI_PADDING_SIZE) || huge_alignment > 0) { + if mi_unlikely(req_size > PTRDIFF_MAX) { // we don't allocate more than PTRDIFF_MAX (see ) + _mi_error_message(EOVERFLOW, "allocation request is too large (%zu bytes)\n", req_size); + return NULL; + } + else { + return mi_large_huge_page_alloc(heap,size,huge_alignment); + } + } + else { + // otherwise find a page with free blocks in our size segregated queues + #if MI_PADDING + mi_assert_internal(size >= MI_PADDING_SIZE); + #endif + return mi_find_free_page(heap, size); + } +} + +// Generic allocation routine if the fast path (`alloc.c:mi_page_malloc`) does not succeed. +// Note: in debug mode the size includes MI_PADDING_SIZE and might have overflowed. +// The `huge_alignment` is normally 0 but is set to a multiple of MI_SEGMENT_SIZE for +// very large requested alignments in which case we use a huge segment. +void* _mi_malloc_generic(mi_heap_t* heap, size_t size, bool zero, size_t huge_alignment) mi_attr_noexcept +{ + mi_assert_internal(heap != NULL); + + // initialize if necessary + if mi_unlikely(!mi_heap_is_initialized(heap)) { + heap = mi_heap_get_default(); // calls mi_thread_init + if mi_unlikely(!mi_heap_is_initialized(heap)) { return NULL; } + } + mi_assert_internal(mi_heap_is_initialized(heap)); + + // call potential deferred free routines + _mi_deferred_free(heap, false); + + // free delayed frees from other threads (but skip contended ones) + _mi_heap_delayed_free_partial(heap); + + // find (or allocate) a page of the right size + mi_page_t* page = mi_find_page(heap, size, huge_alignment); + if mi_unlikely(page == NULL) { // first time out of memory, try to collect and retry the allocation once more + mi_heap_collect(heap, true /* force */); + page = mi_find_page(heap, size, huge_alignment); + } + + if mi_unlikely(page == NULL) { // out of memory + const size_t req_size = size - MI_PADDING_SIZE; // correct for padding_size in case of an overflow on `size` + _mi_error_message(ENOMEM, "unable to allocate memory (%zu bytes)\n", req_size); + return NULL; + } + + mi_assert_internal(mi_page_immediate_available(page)); + mi_assert_internal(mi_page_block_size(page) >= size); + + // and try again, this time succeeding! (i.e. this should never recurse through _mi_page_malloc) + if mi_unlikely(zero && page->xblock_size == 0) { + // note: we cannot call _mi_page_malloc with zeroing for huge blocks; we zero it afterwards in that case. + void* p = _mi_page_malloc(heap, page, size, false); + mi_assert_internal(p != NULL); + _mi_memzero_aligned(p, mi_page_usable_block_size(page)); + return p; + } + else { + return _mi_page_malloc(heap, page, size, zero); + } +} diff --git a/compat/mimalloc/prim/windows/prim.c b/compat/mimalloc/prim/windows/prim.c new file mode 100644 index 00000000000000..d060833c5b644d --- /dev/null +++ b/compat/mimalloc/prim/windows/prim.c @@ -0,0 +1,622 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +// This file is included in `src/prim/prim.c` + +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" +#include "mimalloc/prim.h" +#include // fputs, stderr + + +//--------------------------------------------- +// Dynamically bind Windows API points for portability +//--------------------------------------------- + +// We use VirtualAlloc2 for aligned allocation, but it is only supported on Windows 10 and Windows Server 2016. +// So, we need to look it up dynamically to run on older systems. (use __stdcall for 32-bit compatibility) +// NtAllocateVirtualAllocEx is used for huge OS page allocation (1GiB) +// We define a minimal MEM_EXTENDED_PARAMETER ourselves in order to be able to compile with older SDK's. +typedef enum MI_MEM_EXTENDED_PARAMETER_TYPE_E { + MiMemExtendedParameterInvalidType = 0, + MiMemExtendedParameterAddressRequirements, + MiMemExtendedParameterNumaNode, + MiMemExtendedParameterPartitionHandle, + MiMemExtendedParameterUserPhysicalHandle, + MiMemExtendedParameterAttributeFlags, + MiMemExtendedParameterMax +} MI_MEM_EXTENDED_PARAMETER_TYPE; + +typedef struct DECLSPEC_ALIGN(8) MI_MEM_EXTENDED_PARAMETER_S { + struct { DWORD64 Type : 8; DWORD64 Reserved : 56; } Type; + union { DWORD64 ULong64; PVOID Pointer; SIZE_T Size; HANDLE Handle; DWORD ULong; } Arg; +} MI_MEM_EXTENDED_PARAMETER; + +typedef struct MI_MEM_ADDRESS_REQUIREMENTS_S { + PVOID LowestStartingAddress; + PVOID HighestEndingAddress; + SIZE_T Alignment; +} MI_MEM_ADDRESS_REQUIREMENTS; + +#define MI_MEM_EXTENDED_PARAMETER_NONPAGED_HUGE 0x00000010 + +#include +typedef PVOID (__stdcall *PVirtualAlloc2)(HANDLE, PVOID, SIZE_T, ULONG, ULONG, MI_MEM_EXTENDED_PARAMETER*, ULONG); +typedef NTSTATUS (__stdcall *PNtAllocateVirtualMemoryEx)(HANDLE, PVOID*, SIZE_T*, ULONG, ULONG, MI_MEM_EXTENDED_PARAMETER*, ULONG); +static PVirtualAlloc2 pVirtualAlloc2 = NULL; +static PNtAllocateVirtualMemoryEx pNtAllocateVirtualMemoryEx = NULL; + +// Similarly, GetNumaProcesorNodeEx is only supported since Windows 7 +typedef struct MI_PROCESSOR_NUMBER_S { WORD Group; BYTE Number; BYTE Reserved; } MI_PROCESSOR_NUMBER; + +typedef VOID (__stdcall *PGetCurrentProcessorNumberEx)(MI_PROCESSOR_NUMBER* ProcNumber); +typedef BOOL (__stdcall *PGetNumaProcessorNodeEx)(MI_PROCESSOR_NUMBER* Processor, PUSHORT NodeNumber); +typedef BOOL (__stdcall* PGetNumaNodeProcessorMaskEx)(USHORT Node, PGROUP_AFFINITY ProcessorMask); +typedef BOOL (__stdcall *PGetNumaProcessorNode)(UCHAR Processor, PUCHAR NodeNumber); +static PGetCurrentProcessorNumberEx pGetCurrentProcessorNumberEx = NULL; +static PGetNumaProcessorNodeEx pGetNumaProcessorNodeEx = NULL; +static PGetNumaNodeProcessorMaskEx pGetNumaNodeProcessorMaskEx = NULL; +static PGetNumaProcessorNode pGetNumaProcessorNode = NULL; + +//--------------------------------------------- +// Enable large page support dynamically (if possible) +//--------------------------------------------- + +static bool win_enable_large_os_pages(size_t* large_page_size) +{ + static bool large_initialized = false; + if (large_initialized) return (_mi_os_large_page_size() > 0); + large_initialized = true; + + // Try to see if large OS pages are supported + // To use large pages on Windows, we first need access permission + // Set "Lock pages in memory" permission in the group policy editor + // + unsigned long err = 0; + HANDLE token = NULL; + BOOL ok = OpenProcessToken(GetCurrentProcess(), TOKEN_ADJUST_PRIVILEGES | TOKEN_QUERY, &token); + if (ok) { + TOKEN_PRIVILEGES tp; + ok = LookupPrivilegeValue(NULL, TEXT("SeLockMemoryPrivilege"), &tp.Privileges[0].Luid); + if (ok) { + tp.PrivilegeCount = 1; + tp.Privileges[0].Attributes = SE_PRIVILEGE_ENABLED; + ok = AdjustTokenPrivileges(token, FALSE, &tp, 0, (PTOKEN_PRIVILEGES)NULL, 0); + if (ok) { + err = GetLastError(); + ok = (err == ERROR_SUCCESS); + if (ok && large_page_size != NULL) { + *large_page_size = GetLargePageMinimum(); + } + } + } + CloseHandle(token); + } + if (!ok) { + if (err == 0) err = GetLastError(); + _mi_warning_message("cannot enable large OS page support, error %lu\n", err); + } + return (ok!=0); +} + + +//--------------------------------------------- +// Initialize +//--------------------------------------------- + +void _mi_prim_mem_init( mi_os_mem_config_t* config ) +{ + config->has_overcommit = false; + config->must_free_whole = true; + config->has_virtual_reserve = true; + // get the page size + SYSTEM_INFO si; + GetSystemInfo(&si); + if (si.dwPageSize > 0) { config->page_size = si.dwPageSize; } + if (si.dwAllocationGranularity > 0) { config->alloc_granularity = si.dwAllocationGranularity; } + // get the VirtualAlloc2 function + HINSTANCE hDll; + hDll = LoadLibrary(TEXT("kernelbase.dll")); + if (hDll != NULL) { + // use VirtualAlloc2FromApp if possible as it is available to Windows store apps + pVirtualAlloc2 = (PVirtualAlloc2)(void (*)(void))GetProcAddress(hDll, "VirtualAlloc2FromApp"); + if (pVirtualAlloc2==NULL) pVirtualAlloc2 = (PVirtualAlloc2)(void (*)(void))GetProcAddress(hDll, "VirtualAlloc2"); + FreeLibrary(hDll); + } + // NtAllocateVirtualMemoryEx is used for huge page allocation + hDll = LoadLibrary(TEXT("ntdll.dll")); + if (hDll != NULL) { + pNtAllocateVirtualMemoryEx = (PNtAllocateVirtualMemoryEx)(void (*)(void))GetProcAddress(hDll, "NtAllocateVirtualMemoryEx"); + FreeLibrary(hDll); + } + // Try to use Win7+ numa API + hDll = LoadLibrary(TEXT("kernel32.dll")); + if (hDll != NULL) { + pGetCurrentProcessorNumberEx = (PGetCurrentProcessorNumberEx)(void (*)(void))GetProcAddress(hDll, "GetCurrentProcessorNumberEx"); + pGetNumaProcessorNodeEx = (PGetNumaProcessorNodeEx)(void (*)(void))GetProcAddress(hDll, "GetNumaProcessorNodeEx"); + pGetNumaNodeProcessorMaskEx = (PGetNumaNodeProcessorMaskEx)(void (*)(void))GetProcAddress(hDll, "GetNumaNodeProcessorMaskEx"); + pGetNumaProcessorNode = (PGetNumaProcessorNode)(void (*)(void))GetProcAddress(hDll, "GetNumaProcessorNode"); + FreeLibrary(hDll); + } + if (mi_option_is_enabled(mi_option_allow_large_os_pages) || mi_option_is_enabled(mi_option_reserve_huge_os_pages)) { + win_enable_large_os_pages(&config->large_page_size); + } +} + + +//--------------------------------------------- +// Free +//--------------------------------------------- + +int _mi_prim_free(void* addr, size_t size ) { + MI_UNUSED(size); + DWORD errcode = 0; + bool err = (VirtualFree(addr, 0, MEM_RELEASE) == 0); + if (err) { errcode = GetLastError(); } + if (errcode == ERROR_INVALID_ADDRESS) { + // In mi_os_mem_alloc_aligned the fallback path may have returned a pointer inside + // the memory region returned by VirtualAlloc; in that case we need to free using + // the start of the region. + MEMORY_BASIC_INFORMATION info = { 0 }; + VirtualQuery(addr, &info, sizeof(info)); + if (info.AllocationBase < addr && ((uint8_t*)addr - (uint8_t*)info.AllocationBase) < (ptrdiff_t)MI_SEGMENT_SIZE) { + errcode = 0; + err = (VirtualFree(info.AllocationBase, 0, MEM_RELEASE) == 0); + if (err) { errcode = GetLastError(); } + } + } + return (int)errcode; +} + + +//--------------------------------------------- +// VirtualAlloc +//--------------------------------------------- + +static void* win_virtual_alloc_prim(void* addr, size_t size, size_t try_alignment, DWORD flags) { + #if (MI_INTPTR_SIZE >= 8) + // on 64-bit systems, try to use the virtual address area after 2TiB for 4MiB aligned allocations + if (addr == NULL) { + void* hint = _mi_os_get_aligned_hint(try_alignment,size); + if (hint != NULL) { + void* p = VirtualAlloc(hint, size, flags, PAGE_READWRITE); + if (p != NULL) return p; + _mi_verbose_message("warning: unable to allocate hinted aligned OS memory (%zu bytes, error code: 0x%x, address: %p, alignment: %zu, flags: 0x%x)\n", size, GetLastError(), hint, try_alignment, flags); + // fall through on error + } + } + #endif + // on modern Windows try use VirtualAlloc2 for aligned allocation + if (try_alignment > 1 && (try_alignment % _mi_os_page_size()) == 0 && pVirtualAlloc2 != NULL) { + MI_MEM_ADDRESS_REQUIREMENTS reqs = { 0, 0, 0 }; + reqs.Alignment = try_alignment; + MI_MEM_EXTENDED_PARAMETER param = { {0, 0}, {0} }; + param.Type.Type = MiMemExtendedParameterAddressRequirements; + param.Arg.Pointer = &reqs; + void* p = (*pVirtualAlloc2)(GetCurrentProcess(), addr, size, flags, PAGE_READWRITE, ¶m, 1); + if (p != NULL) return p; + _mi_warning_message("unable to allocate aligned OS memory (%zu bytes, error code: 0x%x, address: %p, alignment: %zu, flags: 0x%x)\n", size, GetLastError(), addr, try_alignment, flags); + // fall through on error + } + // last resort + return VirtualAlloc(addr, size, flags, PAGE_READWRITE); +} + +static void* win_virtual_alloc(void* addr, size_t size, size_t try_alignment, DWORD flags, bool large_only, bool allow_large, bool* is_large) { + mi_assert_internal(!(large_only && !allow_large)); + static _Atomic(size_t) large_page_try_ok; // = 0; + void* p = NULL; + // Try to allocate large OS pages (2MiB) if allowed or required. + if ((large_only || _mi_os_use_large_page(size, try_alignment)) + && allow_large && (flags&MEM_COMMIT)!=0 && (flags&MEM_RESERVE)!=0) { + size_t try_ok = mi_atomic_load_acquire(&large_page_try_ok); + if (!large_only && try_ok > 0) { + // if a large page allocation fails, it seems the calls to VirtualAlloc get very expensive. + // therefore, once a large page allocation failed, we don't try again for `large_page_try_ok` times. + mi_atomic_cas_strong_acq_rel(&large_page_try_ok, &try_ok, try_ok - 1); + } + else { + // large OS pages must always reserve and commit. + *is_large = true; + p = win_virtual_alloc_prim(addr, size, try_alignment, flags | MEM_LARGE_PAGES); + if (large_only) return p; + // fall back to non-large page allocation on error (`p == NULL`). + if (p == NULL) { + mi_atomic_store_release(&large_page_try_ok,10UL); // on error, don't try again for the next N allocations + } + } + } + // Fall back to regular page allocation + if (p == NULL) { + *is_large = ((flags&MEM_LARGE_PAGES) != 0); + p = win_virtual_alloc_prim(addr, size, try_alignment, flags); + } + //if (p == NULL) { _mi_warning_message("unable to allocate OS memory (%zu bytes, error code: 0x%x, address: %p, alignment: %zu, flags: 0x%x, large only: %d, allow large: %d)\n", size, GetLastError(), addr, try_alignment, flags, large_only, allow_large); } + return p; +} + +int _mi_prim_alloc(size_t size, size_t try_alignment, bool commit, bool allow_large, bool* is_large, bool* is_zero, void** addr) { + mi_assert_internal(size > 0 && (size % _mi_os_page_size()) == 0); + mi_assert_internal(commit || !allow_large); + mi_assert_internal(try_alignment > 0); + *is_zero = true; + int flags = MEM_RESERVE; + if (commit) { flags |= MEM_COMMIT; } + *addr = win_virtual_alloc(NULL, size, try_alignment, flags, false, allow_large, is_large); + return (*addr != NULL ? 0 : (int)GetLastError()); +} + + +//--------------------------------------------- +// Commit/Reset/Protect +//--------------------------------------------- +#ifdef _MSC_VER +#pragma warning(disable:6250) // suppress warning calling VirtualFree without MEM_RELEASE (for decommit) +#endif + +int _mi_prim_commit(void* addr, size_t size, bool* is_zero) { + *is_zero = false; + /* + // zero'ing only happens on an initial commit... but checking upfront seems expensive.. + _MEMORY_BASIC_INFORMATION meminfo; _mi_memzero_var(meminfo); + if (VirtualQuery(addr, &meminfo, size) > 0) { + if ((meminfo.State & MEM_COMMIT) == 0) { + *is_zero = true; + } + } + */ + // commit + void* p = VirtualAlloc(addr, size, MEM_COMMIT, PAGE_READWRITE); + if (p == NULL) return (int)GetLastError(); + return 0; +} + +int _mi_prim_decommit(void* addr, size_t size, bool* needs_recommit) { + BOOL ok = VirtualFree(addr, size, MEM_DECOMMIT); + *needs_recommit = true; // for safety, assume always decommitted even in the case of an error. + return (ok ? 0 : (int)GetLastError()); +} + +int _mi_prim_reset(void* addr, size_t size) { + void* p = VirtualAlloc(addr, size, MEM_RESET, PAGE_READWRITE); + mi_assert_internal(p == addr); + #if 0 + if (p != NULL) { + VirtualUnlock(addr,size); // VirtualUnlock after MEM_RESET removes the memory directly from the working set + } + #endif + return (p != NULL ? 0 : (int)GetLastError()); +} + +int _mi_prim_protect(void* addr, size_t size, bool protect) { + DWORD oldprotect = 0; + BOOL ok = VirtualProtect(addr, size, protect ? PAGE_NOACCESS : PAGE_READWRITE, &oldprotect); + return (ok ? 0 : (int)GetLastError()); +} + + +//--------------------------------------------- +// Huge page allocation +//--------------------------------------------- + +static void* _mi_prim_alloc_huge_os_pagesx(void* hint_addr, size_t size, int numa_node) +{ + const DWORD flags = MEM_LARGE_PAGES | MEM_COMMIT | MEM_RESERVE; + + win_enable_large_os_pages(NULL); + + MI_MEM_EXTENDED_PARAMETER params[3] = { {{0,0},{0}},{{0,0},{0}},{{0,0},{0}} }; + // on modern Windows try use NtAllocateVirtualMemoryEx for 1GiB huge pages + static bool mi_huge_pages_available = true; + if (pNtAllocateVirtualMemoryEx != NULL && mi_huge_pages_available) { + params[0].Type.Type = MiMemExtendedParameterAttributeFlags; + params[0].Arg.ULong64 = MI_MEM_EXTENDED_PARAMETER_NONPAGED_HUGE; + ULONG param_count = 1; + if (numa_node >= 0) { + param_count++; + params[1].Type.Type = MiMemExtendedParameterNumaNode; + params[1].Arg.ULong = (unsigned)numa_node; + } + SIZE_T psize = size; + void* base = hint_addr; + NTSTATUS err = (*pNtAllocateVirtualMemoryEx)(GetCurrentProcess(), &base, &psize, flags, PAGE_READWRITE, params, param_count); + if (err == 0 && base != NULL) { + return base; + } + else { + // fall back to regular large pages + mi_huge_pages_available = false; // don't try further huge pages + _mi_warning_message("unable to allocate using huge (1GiB) pages, trying large (2MiB) pages instead (status 0x%lx)\n", err); + } + } + // on modern Windows try use VirtualAlloc2 for numa aware large OS page allocation + if (pVirtualAlloc2 != NULL && numa_node >= 0) { + params[0].Type.Type = MiMemExtendedParameterNumaNode; + params[0].Arg.ULong = (unsigned)numa_node; + return (*pVirtualAlloc2)(GetCurrentProcess(), hint_addr, size, flags, PAGE_READWRITE, params, 1); + } + + // otherwise use regular virtual alloc on older windows + return VirtualAlloc(hint_addr, size, flags, PAGE_READWRITE); +} + +int _mi_prim_alloc_huge_os_pages(void* hint_addr, size_t size, int numa_node, bool* is_zero, void** addr) { + *is_zero = true; + *addr = _mi_prim_alloc_huge_os_pagesx(hint_addr,size,numa_node); + return (*addr != NULL ? 0 : (int)GetLastError()); +} + + +//--------------------------------------------- +// Numa nodes +//--------------------------------------------- + +size_t _mi_prim_numa_node(void) { + USHORT numa_node = 0; + if (pGetCurrentProcessorNumberEx != NULL && pGetNumaProcessorNodeEx != NULL) { + // Extended API is supported + MI_PROCESSOR_NUMBER pnum; + (*pGetCurrentProcessorNumberEx)(&pnum); + USHORT nnode = 0; + BOOL ok = (*pGetNumaProcessorNodeEx)(&pnum, &nnode); + if (ok) { numa_node = nnode; } + } + else if (pGetNumaProcessorNode != NULL) { + // Vista or earlier, use older API that is limited to 64 processors. Issue #277 + DWORD pnum = GetCurrentProcessorNumber(); + UCHAR nnode = 0; + BOOL ok = pGetNumaProcessorNode((UCHAR)pnum, &nnode); + if (ok) { numa_node = nnode; } + } + return numa_node; +} + +size_t _mi_prim_numa_node_count(void) { + ULONG numa_max = 0; + GetNumaHighestNodeNumber(&numa_max); + // find the highest node number that has actual processors assigned to it. Issue #282 + while(numa_max > 0) { + if (pGetNumaNodeProcessorMaskEx != NULL) { + // Extended API is supported + GROUP_AFFINITY affinity; + if ((*pGetNumaNodeProcessorMaskEx)((USHORT)numa_max, &affinity)) { + if (affinity.Mask != 0) break; // found the maximum non-empty node + } + } + else { + // Vista or earlier, use older API that is limited to 64 processors. + ULONGLONG mask; + if (GetNumaNodeProcessorMask((UCHAR)numa_max, &mask)) { + if (mask != 0) break; // found the maximum non-empty node + }; + } + // max node was invalid or had no processor assigned, try again + numa_max--; + } + return ((size_t)numa_max + 1); +} + + +//---------------------------------------------------------------- +// Clock +//---------------------------------------------------------------- + +static mi_msecs_t mi_to_msecs(LARGE_INTEGER t) { + static LARGE_INTEGER mfreq; // = 0 + if (mfreq.QuadPart == 0LL) { + LARGE_INTEGER f; + QueryPerformanceFrequency(&f); + mfreq.QuadPart = f.QuadPart/1000LL; + if (mfreq.QuadPart == 0) mfreq.QuadPart = 1; + } + return (mi_msecs_t)(t.QuadPart / mfreq.QuadPart); +} + +mi_msecs_t _mi_prim_clock_now(void) { + LARGE_INTEGER t; + QueryPerformanceCounter(&t); + return mi_to_msecs(t); +} + + +//---------------------------------------------------------------- +// Process Info +//---------------------------------------------------------------- + +#include +#include + +static mi_msecs_t filetime_msecs(const FILETIME* ftime) { + ULARGE_INTEGER i; + i.LowPart = ftime->dwLowDateTime; + i.HighPart = ftime->dwHighDateTime; + mi_msecs_t msecs = (i.QuadPart / 10000); // FILETIME is in 100 nano seconds + return msecs; +} + +typedef BOOL (WINAPI *PGetProcessMemoryInfo)(HANDLE, PPROCESS_MEMORY_COUNTERS, DWORD); +static PGetProcessMemoryInfo pGetProcessMemoryInfo = NULL; + +void _mi_prim_process_info(mi_process_info_t* pinfo) +{ + FILETIME ct; + FILETIME ut; + FILETIME st; + FILETIME et; + GetProcessTimes(GetCurrentProcess(), &ct, &et, &st, &ut); + pinfo->utime = filetime_msecs(&ut); + pinfo->stime = filetime_msecs(&st); + + // load psapi on demand + if (pGetProcessMemoryInfo == NULL) { + HINSTANCE hDll = LoadLibrary(TEXT("psapi.dll")); + if (hDll != NULL) { + pGetProcessMemoryInfo = (PGetProcessMemoryInfo)(void (*)(void))GetProcAddress(hDll, "GetProcessMemoryInfo"); + } + } + + // get process info + PROCESS_MEMORY_COUNTERS info; + memset(&info, 0, sizeof(info)); + if (pGetProcessMemoryInfo != NULL) { + pGetProcessMemoryInfo(GetCurrentProcess(), &info, sizeof(info)); + } + pinfo->current_rss = (size_t)info.WorkingSetSize; + pinfo->peak_rss = (size_t)info.PeakWorkingSetSize; + pinfo->current_commit = (size_t)info.PagefileUsage; + pinfo->peak_commit = (size_t)info.PeakPagefileUsage; + pinfo->page_faults = (size_t)info.PageFaultCount; +} + +//---------------------------------------------------------------- +// Output +//---------------------------------------------------------------- + +void _mi_prim_out_stderr( const char* msg ) +{ + // on windows with redirection, the C runtime cannot handle locale dependent output + // after the main thread closes so we use direct console output. + if (!_mi_preloading()) { + // _cputs(msg); // _cputs cannot be used at is aborts if it fails to lock the console + static HANDLE hcon = INVALID_HANDLE_VALUE; + static bool hconIsConsole; + if (hcon == INVALID_HANDLE_VALUE) { + CONSOLE_SCREEN_BUFFER_INFO sbi; + hcon = GetStdHandle(STD_ERROR_HANDLE); + hconIsConsole = ((hcon != INVALID_HANDLE_VALUE) && GetConsoleScreenBufferInfo(hcon, &sbi)); + } + const size_t len = _mi_strlen(msg); + if (len > 0 && len < UINT32_MAX) { + DWORD written = 0; + if (hconIsConsole) { + WriteConsoleA(hcon, msg, (DWORD)len, &written, NULL); + } + else if (hcon != INVALID_HANDLE_VALUE) { + // use direct write if stderr was redirected + WriteFile(hcon, msg, (DWORD)len, &written, NULL); + } + else { + // finally fall back to fputs after all + fputs(msg, stderr); + } + } + } +} + + +//---------------------------------------------------------------- +// Environment +//---------------------------------------------------------------- + +// On Windows use GetEnvironmentVariable instead of getenv to work +// reliably even when this is invoked before the C runtime is initialized. +// i.e. when `_mi_preloading() == true`. +// Note: on windows, environment names are not case sensitive. +bool _mi_prim_getenv(const char* name, char* result, size_t result_size) { + result[0] = 0; + size_t len = GetEnvironmentVariableA(name, result, (DWORD)result_size); + return (len > 0 && len < result_size); +} + + + +//---------------------------------------------------------------- +// Random +//---------------------------------------------------------------- + +#if defined(MI_USE_RTLGENRANDOM) // || defined(__cplusplus) +// We prefer to use BCryptGenRandom instead of (the unofficial) RtlGenRandom but when using +// dynamic overriding, we observed it can raise an exception when compiled with C++, and +// sometimes deadlocks when also running under the VS debugger. +// In contrast, issue #623 implies that on Windows Server 2019 we need to use BCryptGenRandom. +// To be continued.. +#pragma comment (lib,"advapi32.lib") +#define RtlGenRandom SystemFunction036 +mi_decl_externc BOOLEAN NTAPI RtlGenRandom(PVOID RandomBuffer, ULONG RandomBufferLength); + +bool _mi_prim_random_buf(void* buf, size_t buf_len) { + return (RtlGenRandom(buf, (ULONG)buf_len) != 0); +} + +#else + +#ifndef BCRYPT_USE_SYSTEM_PREFERRED_RNG +#define BCRYPT_USE_SYSTEM_PREFERRED_RNG 0x00000002 +#endif + +typedef LONG (NTAPI *PBCryptGenRandom)(HANDLE, PUCHAR, ULONG, ULONG); +static PBCryptGenRandom pBCryptGenRandom = NULL; + +bool _mi_prim_random_buf(void* buf, size_t buf_len) { + if (pBCryptGenRandom == NULL) { + HINSTANCE hDll = LoadLibrary(TEXT("bcrypt.dll")); + if (hDll != NULL) { + pBCryptGenRandom = (PBCryptGenRandom)(void (*)(void))GetProcAddress(hDll, "BCryptGenRandom"); + } + if (pBCryptGenRandom == NULL) return false; + } + return (pBCryptGenRandom(NULL, (PUCHAR)buf, (ULONG)buf_len, BCRYPT_USE_SYSTEM_PREFERRED_RNG) >= 0); +} + +#endif // MI_USE_RTLGENRANDOM + +//---------------------------------------------------------------- +// Thread init/done +//---------------------------------------------------------------- + +#if !defined(MI_SHARED_LIB) + +// use thread local storage keys to detect thread ending +#include +#if (_WIN32_WINNT < 0x600) // before Windows Vista +WINBASEAPI DWORD WINAPI FlsAlloc( _In_opt_ PFLS_CALLBACK_FUNCTION lpCallback ); +WINBASEAPI PVOID WINAPI FlsGetValue( _In_ DWORD dwFlsIndex ); +WINBASEAPI BOOL WINAPI FlsSetValue( _In_ DWORD dwFlsIndex, _In_opt_ PVOID lpFlsData ); +WINBASEAPI BOOL WINAPI FlsFree(_In_ DWORD dwFlsIndex); +#endif + +static DWORD mi_fls_key = (DWORD)(-1); + +static void NTAPI mi_fls_done(PVOID value) { + mi_heap_t* heap = (mi_heap_t*)value; + if (heap != NULL) { + _mi_thread_done(heap); + FlsSetValue(mi_fls_key, NULL); // prevent recursion as _mi_thread_done may set it back to the main heap, issue #672 + } +} + +void _mi_prim_thread_init_auto_done(void) { + mi_fls_key = FlsAlloc(&mi_fls_done); +} + +void _mi_prim_thread_done_auto_done(void) { + // call thread-done on all threads (except the main thread) to prevent + // dangling callback pointer if statically linked with a DLL; Issue #208 + FlsFree(mi_fls_key); +} + +void _mi_prim_thread_associate_default_heap(mi_heap_t* heap) { + mi_assert_internal(mi_fls_key != (DWORD)(-1)); + FlsSetValue(mi_fls_key, heap); +} + +#else + +// Dll; nothing to do as in that case thread_done is handled through the DLL_THREAD_DETACH event. + +void _mi_prim_thread_init_auto_done(void) { +} + +void _mi_prim_thread_done_auto_done(void) { +} + +void _mi_prim_thread_associate_default_heap(mi_heap_t* heap) { + MI_UNUSED(heap); +} + +#endif diff --git a/compat/mimalloc/random.c b/compat/mimalloc/random.c new file mode 100644 index 00000000000000..2a18b5aa992dad --- /dev/null +++ b/compat/mimalloc/random.c @@ -0,0 +1,254 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2019-2021, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/prim.h" // _mi_prim_random_buf +#include // memset + +/* ---------------------------------------------------------------------------- +We use our own PRNG to keep predictable performance of random number generation +and to avoid implementations that use a lock. We only use the OS provided +random source to initialize the initial seeds. Since we do not need ultimate +performance but we do rely on the security (for secret cookies in secure mode) +we use a cryptographically secure generator (chacha20). +-----------------------------------------------------------------------------*/ + +#define MI_CHACHA_ROUNDS (20) // perhaps use 12 for better performance? + + +/* ---------------------------------------------------------------------------- +Chacha20 implementation as the original algorithm with a 64-bit nonce +and counter: https://en.wikipedia.org/wiki/Salsa20 +The input matrix has sixteen 32-bit values: +Position 0 to 3: constant key +Position 4 to 11: the key +Position 12 to 13: the counter. +Position 14 to 15: the nonce. + +The implementation uses regular C code which compiles very well on modern compilers. +(gcc x64 has no register spills, and clang 6+ uses SSE instructions) +-----------------------------------------------------------------------------*/ + +static inline uint32_t rotl(uint32_t x, uint32_t shift) { + return (x << shift) | (x >> (32 - shift)); +} + +static inline void qround(uint32_t x[16], size_t a, size_t b, size_t c, size_t d) { + x[a] += x[b]; x[d] = rotl(x[d] ^ x[a], 16); + x[c] += x[d]; x[b] = rotl(x[b] ^ x[c], 12); + x[a] += x[b]; x[d] = rotl(x[d] ^ x[a], 8); + x[c] += x[d]; x[b] = rotl(x[b] ^ x[c], 7); +} + +static void chacha_block(mi_random_ctx_t* ctx) +{ + // scramble into `x` + uint32_t x[16]; + for (size_t i = 0; i < 16; i++) { + x[i] = ctx->input[i]; + } + for (size_t i = 0; i < MI_CHACHA_ROUNDS; i += 2) { + qround(x, 0, 4, 8, 12); + qround(x, 1, 5, 9, 13); + qround(x, 2, 6, 10, 14); + qround(x, 3, 7, 11, 15); + qround(x, 0, 5, 10, 15); + qround(x, 1, 6, 11, 12); + qround(x, 2, 7, 8, 13); + qround(x, 3, 4, 9, 14); + } + + // add scrambled data to the initial state + for (size_t i = 0; i < 16; i++) { + ctx->output[i] = x[i] + ctx->input[i]; + } + ctx->output_available = 16; + + // increment the counter for the next round + ctx->input[12] += 1; + if (ctx->input[12] == 0) { + ctx->input[13] += 1; + if (ctx->input[13] == 0) { // and keep increasing into the nonce + ctx->input[14] += 1; + } + } +} + +static uint32_t chacha_next32(mi_random_ctx_t* ctx) { + if (ctx->output_available <= 0) { + chacha_block(ctx); + ctx->output_available = 16; // (assign again to suppress static analysis warning) + } + const uint32_t x = ctx->output[16 - ctx->output_available]; + ctx->output[16 - ctx->output_available] = 0; // reset once the data is handed out + ctx->output_available--; + return x; +} + +static inline uint32_t read32(const uint8_t* p, size_t idx32) { + const size_t i = 4*idx32; + return ((uint32_t)p[i+0] | (uint32_t)p[i+1] << 8 | (uint32_t)p[i+2] << 16 | (uint32_t)p[i+3] << 24); +} + +static void chacha_init(mi_random_ctx_t* ctx, const uint8_t key[32], uint64_t nonce) +{ + // since we only use chacha for randomness (and not encryption) we + // do not _need_ to read 32-bit values as little endian but we do anyways + // just for being compatible :-) + memset(ctx, 0, sizeof(*ctx)); + for (size_t i = 0; i < 4; i++) { + const uint8_t* sigma = (uint8_t*)"expand 32-byte k"; + ctx->input[i] = read32(sigma,i); + } + for (size_t i = 0; i < 8; i++) { + ctx->input[i + 4] = read32(key,i); + } + ctx->input[12] = 0; + ctx->input[13] = 0; + ctx->input[14] = (uint32_t)nonce; + ctx->input[15] = (uint32_t)(nonce >> 32); +} + +static void chacha_split(mi_random_ctx_t* ctx, uint64_t nonce, mi_random_ctx_t* ctx_new) { + memset(ctx_new, 0, sizeof(*ctx_new)); + _mi_memcpy(ctx_new->input, ctx->input, sizeof(ctx_new->input)); + ctx_new->input[12] = 0; + ctx_new->input[13] = 0; + ctx_new->input[14] = (uint32_t)nonce; + ctx_new->input[15] = (uint32_t)(nonce >> 32); + mi_assert_internal(ctx->input[14] != ctx_new->input[14] || ctx->input[15] != ctx_new->input[15]); // do not reuse nonces! + chacha_block(ctx_new); +} + + +/* ---------------------------------------------------------------------------- +Random interface +-----------------------------------------------------------------------------*/ + +#if MI_DEBUG>1 +static bool mi_random_is_initialized(mi_random_ctx_t* ctx) { + return (ctx != NULL && ctx->input[0] != 0); +} +#endif + +void _mi_random_split(mi_random_ctx_t* ctx, mi_random_ctx_t* ctx_new) { + mi_assert_internal(mi_random_is_initialized(ctx)); + mi_assert_internal(ctx != ctx_new); + chacha_split(ctx, (uintptr_t)ctx_new /*nonce*/, ctx_new); +} + +uintptr_t _mi_random_next(mi_random_ctx_t* ctx) { + mi_assert_internal(mi_random_is_initialized(ctx)); + #if MI_INTPTR_SIZE <= 4 + return chacha_next32(ctx); + #elif MI_INTPTR_SIZE == 8 + return (((uintptr_t)chacha_next32(ctx) << 32) | chacha_next32(ctx)); + #else + # error "define mi_random_next for this platform" + #endif +} + + +/* ---------------------------------------------------------------------------- +To initialize a fresh random context. +If we cannot get good randomness, we fall back to weak randomness based on a timer and ASLR. +-----------------------------------------------------------------------------*/ + +uintptr_t _mi_os_random_weak(uintptr_t extra_seed) { + uintptr_t x = (uintptr_t)&_mi_os_random_weak ^ extra_seed; // ASLR makes the address random + x ^= _mi_prim_clock_now(); + // and do a few randomization steps + uintptr_t max = ((x ^ (x >> 17)) & 0x0F) + 1; + for (uintptr_t i = 0; i < max; i++) { + x = _mi_random_shuffle(x); + } + mi_assert_internal(x != 0); + return x; +} + +static void mi_random_init_ex(mi_random_ctx_t* ctx, bool use_weak) { + uint8_t key[32]; + if (use_weak || !_mi_prim_random_buf(key, sizeof(key))) { + // if we fail to get random data from the OS, we fall back to a + // weak random source based on the current time + #if !defined(__wasi__) + if (!use_weak) { _mi_warning_message("unable to use secure randomness\n"); } + #endif + uintptr_t x = _mi_os_random_weak(0); + for (size_t i = 0; i < 8; i++) { // key is eight 32-bit words. + x = _mi_random_shuffle(x); + ((uint32_t*)key)[i] = (uint32_t)x; + } + ctx->weak = true; + } + else { + ctx->weak = false; + } + chacha_init(ctx, key, (uintptr_t)ctx /*nonce*/ ); +} + +void _mi_random_init(mi_random_ctx_t* ctx) { + mi_random_init_ex(ctx, false); +} + +void _mi_random_init_weak(mi_random_ctx_t * ctx) { + mi_random_init_ex(ctx, true); +} + +void _mi_random_reinit_if_weak(mi_random_ctx_t * ctx) { + if (ctx->weak) { + _mi_random_init(ctx); + } +} + +/* -------------------------------------------------------- +test vectors from +----------------------------------------------------------- */ +/* +static bool array_equals(uint32_t* x, uint32_t* y, size_t n) { + for (size_t i = 0; i < n; i++) { + if (x[i] != y[i]) return false; + } + return true; +} +static void chacha_test(void) +{ + uint32_t x[4] = { 0x11111111, 0x01020304, 0x9b8d6f43, 0x01234567 }; + uint32_t x_out[4] = { 0xea2a92f4, 0xcb1cf8ce, 0x4581472e, 0x5881c4bb }; + qround(x, 0, 1, 2, 3); + mi_assert_internal(array_equals(x, x_out, 4)); + + uint32_t y[16] = { + 0x879531e0, 0xc5ecf37d, 0x516461b1, 0xc9a62f8a, + 0x44c20ef3, 0x3390af7f, 0xd9fc690b, 0x2a5f714c, + 0x53372767, 0xb00a5631, 0x974c541a, 0x359e9963, + 0x5c971061, 0x3d631689, 0x2098d9d6, 0x91dbd320 }; + uint32_t y_out[16] = { + 0x879531e0, 0xc5ecf37d, 0xbdb886dc, 0xc9a62f8a, + 0x44c20ef3, 0x3390af7f, 0xd9fc690b, 0xcfacafd2, + 0xe46bea80, 0xb00a5631, 0x974c541a, 0x359e9963, + 0x5c971061, 0xccc07c79, 0x2098d9d6, 0x91dbd320 }; + qround(y, 2, 7, 8, 13); + mi_assert_internal(array_equals(y, y_out, 16)); + + mi_random_ctx_t r = { + { 0x61707865, 0x3320646e, 0x79622d32, 0x6b206574, + 0x03020100, 0x07060504, 0x0b0a0908, 0x0f0e0d0c, + 0x13121110, 0x17161514, 0x1b1a1918, 0x1f1e1d1c, + 0x00000001, 0x09000000, 0x4a000000, 0x00000000 }, + {0}, + 0 + }; + uint32_t r_out[16] = { + 0xe4e7f110, 0x15593bd1, 0x1fdd0f50, 0xc47120a3, + 0xc7f4d1c7, 0x0368c033, 0x9aaa2204, 0x4e6cd4c3, + 0x466482d2, 0x09aa9f07, 0x05d7c214, 0xa2028bd9, + 0xd19c12b5, 0xb94e16de, 0xe883d0cb, 0x4e3c50a2 }; + chacha_block(&r); + mi_assert_internal(array_equals(r.output, r_out, 16)); +} +*/ diff --git a/compat/mimalloc/segment-cache.c b/compat/mimalloc/segment-cache.c new file mode 100644 index 00000000000000..e69de29bb2d1d6 diff --git a/compat/mimalloc/segment-map.c b/compat/mimalloc/segment-map.c new file mode 100644 index 00000000000000..3cd2127e56c1a7 --- /dev/null +++ b/compat/mimalloc/segment-map.c @@ -0,0 +1,153 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2019-2023, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ + +/* ----------------------------------------------------------- + The following functions are to reliably find the segment or + block that encompasses any pointer p (or NULL if it is not + in any of our segments). + We maintain a bitmap of all memory with 1 bit per MI_SEGMENT_SIZE (64MiB) + set to 1 if it contains the segment meta data. +----------------------------------------------------------- */ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" + +#if (MI_INTPTR_SIZE==8) +#define MI_MAX_ADDRESS ((size_t)40 << 40) // 40TB (to include huge page areas) +#else +#define MI_MAX_ADDRESS ((size_t)2 << 30) // 2Gb +#endif + +#define MI_SEGMENT_MAP_BITS (MI_MAX_ADDRESS / MI_SEGMENT_SIZE) +#define MI_SEGMENT_MAP_SIZE (MI_SEGMENT_MAP_BITS / 8) +#define MI_SEGMENT_MAP_WSIZE (MI_SEGMENT_MAP_SIZE / MI_INTPTR_SIZE) + +static _Atomic(uintptr_t) mi_segment_map[MI_SEGMENT_MAP_WSIZE + 1]; // 2KiB per TB with 64MiB segments + +static size_t mi_segment_map_index_of(const mi_segment_t* segment, size_t* bitidx) { + mi_assert_internal(_mi_ptr_segment(segment + 1) == segment); // is it aligned on MI_SEGMENT_SIZE? + if ((uintptr_t)segment >= MI_MAX_ADDRESS) { + *bitidx = 0; + return MI_SEGMENT_MAP_WSIZE; + } + else { + const uintptr_t segindex = ((uintptr_t)segment) / MI_SEGMENT_SIZE; + *bitidx = segindex % MI_INTPTR_BITS; + const size_t mapindex = segindex / MI_INTPTR_BITS; + mi_assert_internal(mapindex < MI_SEGMENT_MAP_WSIZE); + return mapindex; + } +} + +void _mi_segment_map_allocated_at(const mi_segment_t* segment) { + size_t bitidx; + size_t index = mi_segment_map_index_of(segment, &bitidx); + mi_assert_internal(index <= MI_SEGMENT_MAP_WSIZE); + if (index==MI_SEGMENT_MAP_WSIZE) return; + uintptr_t mask = mi_atomic_load_relaxed(&mi_segment_map[index]); + uintptr_t newmask; + do { + newmask = (mask | ((uintptr_t)1 << bitidx)); + } while (!mi_atomic_cas_weak_release(&mi_segment_map[index], &mask, newmask)); +} + +void _mi_segment_map_freed_at(const mi_segment_t* segment) { + size_t bitidx; + size_t index = mi_segment_map_index_of(segment, &bitidx); + mi_assert_internal(index <= MI_SEGMENT_MAP_WSIZE); + if (index == MI_SEGMENT_MAP_WSIZE) return; + uintptr_t mask = mi_atomic_load_relaxed(&mi_segment_map[index]); + uintptr_t newmask; + do { + newmask = (mask & ~((uintptr_t)1 << bitidx)); + } while (!mi_atomic_cas_weak_release(&mi_segment_map[index], &mask, newmask)); +} + +// Determine the segment belonging to a pointer or NULL if it is not in a valid segment. +static mi_segment_t* _mi_segment_of(const void* p) { + if (p == NULL) return NULL; + mi_segment_t* segment = _mi_ptr_segment(p); + mi_assert_internal(segment != NULL); + size_t bitidx; + size_t index = mi_segment_map_index_of(segment, &bitidx); + // fast path: for any pointer to valid small/medium/large object or first MI_SEGMENT_SIZE in huge + const uintptr_t mask = mi_atomic_load_relaxed(&mi_segment_map[index]); + if mi_likely((mask & ((uintptr_t)1 << bitidx)) != 0) { + return segment; // yes, allocated by us + } + if (index==MI_SEGMENT_MAP_WSIZE) return NULL; + + // TODO: maintain max/min allocated range for efficiency for more efficient rejection of invalid pointers? + + // search downwards for the first segment in case it is an interior pointer + // could be slow but searches in MI_INTPTR_SIZE * MI_SEGMENT_SIZE (512MiB) steps trough + // valid huge objects + // note: we could maintain a lowest index to speed up the path for invalid pointers? + size_t lobitidx; + size_t loindex; + uintptr_t lobits = mask & (((uintptr_t)1 << bitidx) - 1); + if (lobits != 0) { + loindex = index; + lobitidx = mi_bsr(lobits); // lobits != 0 + } + else if (index == 0) { + return NULL; + } + else { + mi_assert_internal(index > 0); + uintptr_t lomask = mask; + loindex = index; + do { + loindex--; + lomask = mi_atomic_load_relaxed(&mi_segment_map[loindex]); + } while (lomask != 0 && loindex > 0); + if (lomask == 0) return NULL; + lobitidx = mi_bsr(lomask); // lomask != 0 + } + mi_assert_internal(loindex < MI_SEGMENT_MAP_WSIZE); + // take difference as the addresses could be larger than the MAX_ADDRESS space. + size_t diff = (((index - loindex) * (8*MI_INTPTR_SIZE)) + bitidx - lobitidx) * MI_SEGMENT_SIZE; + segment = (mi_segment_t*)((uint8_t*)segment - diff); + + if (segment == NULL) return NULL; + mi_assert_internal((void*)segment < p); + bool cookie_ok = (_mi_ptr_cookie(segment) == segment->cookie); + mi_assert_internal(cookie_ok); + if mi_unlikely(!cookie_ok) return NULL; + if (((uint8_t*)segment + mi_segment_size(segment)) <= (uint8_t*)p) return NULL; // outside the range + mi_assert_internal(p >= (void*)segment && (uint8_t*)p < (uint8_t*)segment + mi_segment_size(segment)); + return segment; +} + +// Is this a valid pointer in our heap? +static bool mi_is_valid_pointer(const void* p) { + return ((_mi_segment_of(p) != NULL) || (_mi_arena_contains(p))); +} + +mi_decl_nodiscard mi_decl_export bool mi_is_in_heap_region(const void* p) mi_attr_noexcept { + return mi_is_valid_pointer(p); +} + +/* +// Return the full segment range belonging to a pointer +static void* mi_segment_range_of(const void* p, size_t* size) { + mi_segment_t* segment = _mi_segment_of(p); + if (segment == NULL) { + if (size != NULL) *size = 0; + return NULL; + } + else { + if (size != NULL) *size = segment->segment_size; + return segment; + } + mi_assert_expensive(page == NULL || mi_segment_is_valid(_mi_page_segment(page),tld)); + mi_assert_internal(page == NULL || (mi_segment_page_size(_mi_page_segment(page)) - (MI_SECURE == 0 ? 0 : _mi_os_page_size())) >= block_size); + mi_reset_delayed(tld); + mi_assert_internal(page == NULL || mi_page_not_in_queue(page, tld)); + return page; +} +*/ diff --git a/compat/mimalloc/segment.c b/compat/mimalloc/segment.c new file mode 100644 index 00000000000000..6b901f6cc80f13 --- /dev/null +++ b/compat/mimalloc/segment.c @@ -0,0 +1,1617 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2020, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" + +#include // memset +#include + +#define MI_PAGE_HUGE_ALIGN (256*1024) + +static void mi_segment_try_purge(mi_segment_t* segment, bool force, mi_stats_t* stats); + + +// ------------------------------------------------------------------- +// commit mask +// ------------------------------------------------------------------- + +static bool mi_commit_mask_all_set(const mi_commit_mask_t* commit, const mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + if ((commit->mask[i] & cm->mask[i]) != cm->mask[i]) return false; + } + return true; +} + +static bool mi_commit_mask_any_set(const mi_commit_mask_t* commit, const mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + if ((commit->mask[i] & cm->mask[i]) != 0) return true; + } + return false; +} + +static void mi_commit_mask_create_intersect(const mi_commit_mask_t* commit, const mi_commit_mask_t* cm, mi_commit_mask_t* res) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + res->mask[i] = (commit->mask[i] & cm->mask[i]); + } +} + +static void mi_commit_mask_clear(mi_commit_mask_t* res, const mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + res->mask[i] &= ~(cm->mask[i]); + } +} + +static void mi_commit_mask_set(mi_commit_mask_t* res, const mi_commit_mask_t* cm) { + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + res->mask[i] |= cm->mask[i]; + } +} + +static void mi_commit_mask_create(size_t bitidx, size_t bitcount, mi_commit_mask_t* cm) { + mi_assert_internal(bitidx < MI_COMMIT_MASK_BITS); + mi_assert_internal((bitidx + bitcount) <= MI_COMMIT_MASK_BITS); + if (bitcount == MI_COMMIT_MASK_BITS) { + mi_assert_internal(bitidx==0); + mi_commit_mask_create_full(cm); + } + else if (bitcount == 0) { + mi_commit_mask_create_empty(cm); + } + else { + mi_commit_mask_create_empty(cm); + size_t i = bitidx / MI_COMMIT_MASK_FIELD_BITS; + size_t ofs = bitidx % MI_COMMIT_MASK_FIELD_BITS; + while (bitcount > 0) { + mi_assert_internal(i < MI_COMMIT_MASK_FIELD_COUNT); + size_t avail = MI_COMMIT_MASK_FIELD_BITS - ofs; + size_t count = (bitcount > avail ? avail : bitcount); + size_t mask = (count >= MI_COMMIT_MASK_FIELD_BITS ? ~((size_t)0) : (((size_t)1 << count) - 1) << ofs); + cm->mask[i] = mask; + bitcount -= count; + ofs = 0; + i++; + } + } +} + +size_t _mi_commit_mask_committed_size(const mi_commit_mask_t* cm, size_t total) { + mi_assert_internal((total%MI_COMMIT_MASK_BITS)==0); + size_t count = 0; + for (size_t i = 0; i < MI_COMMIT_MASK_FIELD_COUNT; i++) { + size_t mask = cm->mask[i]; + if (~mask == 0) { + count += MI_COMMIT_MASK_FIELD_BITS; + } + else { + for (; mask != 0; mask >>= 1) { // todo: use popcount + if ((mask&1)!=0) count++; + } + } + } + // we use total since for huge segments each commit bit may represent a larger size + return ((total / MI_COMMIT_MASK_BITS) * count); +} + + +size_t _mi_commit_mask_next_run(const mi_commit_mask_t* cm, size_t* idx) { + size_t i = (*idx) / MI_COMMIT_MASK_FIELD_BITS; + size_t ofs = (*idx) % MI_COMMIT_MASK_FIELD_BITS; + size_t mask = 0; + // find first ones + while (i < MI_COMMIT_MASK_FIELD_COUNT) { + mask = cm->mask[i]; + mask >>= ofs; + if (mask != 0) { + while ((mask&1) == 0) { + mask >>= 1; + ofs++; + } + break; + } + i++; + ofs = 0; + } + if (i >= MI_COMMIT_MASK_FIELD_COUNT) { + // not found + *idx = MI_COMMIT_MASK_BITS; + return 0; + } + else { + // found, count ones + size_t count = 0; + *idx = (i*MI_COMMIT_MASK_FIELD_BITS) + ofs; + do { + mi_assert_internal(ofs < MI_COMMIT_MASK_FIELD_BITS && (mask&1) == 1); + do { + count++; + mask >>= 1; + } while ((mask&1) == 1); + if ((((*idx + count) % MI_COMMIT_MASK_FIELD_BITS) == 0)) { + i++; + if (i >= MI_COMMIT_MASK_FIELD_COUNT) break; + mask = cm->mask[i]; + ofs = 0; + } + } while ((mask&1) == 1); + mi_assert_internal(count > 0); + return count; + } +} + + +/* -------------------------------------------------------------------------------- + Segment allocation + + If a thread ends, it "abandons" pages with used blocks + and there is an abandoned segment list whose segments can + be reclaimed by still running threads, much like work-stealing. +-------------------------------------------------------------------------------- */ + + +/* ----------------------------------------------------------- + Slices +----------------------------------------------------------- */ + + +static const mi_slice_t* mi_segment_slices_end(const mi_segment_t* segment) { + return &segment->slices[segment->slice_entries]; +} + +static uint8_t* mi_slice_start(const mi_slice_t* slice) { + mi_segment_t* segment = _mi_ptr_segment(slice); + mi_assert_internal(slice >= segment->slices && slice < mi_segment_slices_end(segment)); + return ((uint8_t*)segment + ((slice - segment->slices)*MI_SEGMENT_SLICE_SIZE)); +} + + +/* ----------------------------------------------------------- + Bins +----------------------------------------------------------- */ +// Use bit scan forward to quickly find the first zero bit if it is available + +static inline size_t mi_slice_bin8(size_t slice_count) { + if (slice_count<=1) return slice_count; + mi_assert_internal(slice_count <= MI_SLICES_PER_SEGMENT); + slice_count--; + size_t s = mi_bsr(slice_count); // slice_count > 1 + if (s <= 2) return slice_count + 1; + size_t bin = ((s << 2) | ((slice_count >> (s - 2))&0x03)) - 4; + return bin; +} + +static inline size_t mi_slice_bin(size_t slice_count) { + mi_assert_internal(slice_count*MI_SEGMENT_SLICE_SIZE <= MI_SEGMENT_SIZE); + mi_assert_internal(mi_slice_bin8(MI_SLICES_PER_SEGMENT) <= MI_SEGMENT_BIN_MAX); + size_t bin = mi_slice_bin8(slice_count); + mi_assert_internal(bin <= MI_SEGMENT_BIN_MAX); + return bin; +} + +static inline size_t mi_slice_index(const mi_slice_t* slice) { + mi_segment_t* segment = _mi_ptr_segment(slice); + ptrdiff_t index = slice - segment->slices; + mi_assert_internal(index >= 0 && index < (ptrdiff_t)segment->slice_entries); + return index; +} + + +/* ----------------------------------------------------------- + Slice span queues +----------------------------------------------------------- */ + +static void mi_span_queue_push(mi_span_queue_t* sq, mi_slice_t* slice) { + // todo: or push to the end? + mi_assert_internal(slice->prev == NULL && slice->next==NULL); + slice->prev = NULL; // paranoia + slice->next = sq->first; + sq->first = slice; + if (slice->next != NULL) slice->next->prev = slice; + else sq->last = slice; + slice->xblock_size = 0; // free +} + +static mi_span_queue_t* mi_span_queue_for(size_t slice_count, mi_segments_tld_t* tld) { + size_t bin = mi_slice_bin(slice_count); + mi_span_queue_t* sq = &tld->spans[bin]; + mi_assert_internal(sq->slice_count >= slice_count); + return sq; +} + +static void mi_span_queue_delete(mi_span_queue_t* sq, mi_slice_t* slice) { + mi_assert_internal(slice->xblock_size==0 && slice->slice_count>0 && slice->slice_offset==0); + // should work too if the queue does not contain slice (which can happen during reclaim) + if (slice->prev != NULL) slice->prev->next = slice->next; + if (slice == sq->first) sq->first = slice->next; + if (slice->next != NULL) slice->next->prev = slice->prev; + if (slice == sq->last) sq->last = slice->prev; + slice->prev = NULL; + slice->next = NULL; + slice->xblock_size = 1; // no more free +} + + +/* ----------------------------------------------------------- + Invariant checking +----------------------------------------------------------- */ + +static bool mi_slice_is_used(const mi_slice_t* slice) { + return (slice->xblock_size > 0); +} + + +#if (MI_DEBUG>=3) +static bool mi_span_queue_contains(mi_span_queue_t* sq, mi_slice_t* slice) { + for (mi_slice_t* s = sq->first; s != NULL; s = s->next) { + if (s==slice) return true; + } + return false; +} + +static bool mi_segment_is_valid(mi_segment_t* segment, mi_segments_tld_t* tld) { + mi_assert_internal(segment != NULL); + mi_assert_internal(_mi_ptr_cookie(segment) == segment->cookie); + mi_assert_internal(segment->abandoned <= segment->used); + mi_assert_internal(segment->thread_id == 0 || segment->thread_id == _mi_thread_id()); + mi_assert_internal(mi_commit_mask_all_set(&segment->commit_mask, &segment->purge_mask)); // can only decommit committed blocks + //mi_assert_internal(segment->segment_info_size % MI_SEGMENT_SLICE_SIZE == 0); + mi_slice_t* slice = &segment->slices[0]; + const mi_slice_t* end = mi_segment_slices_end(segment); + size_t used_count = 0; + mi_span_queue_t* sq; + while(slice < end) { + mi_assert_internal(slice->slice_count > 0); + mi_assert_internal(slice->slice_offset == 0); + size_t index = mi_slice_index(slice); + size_t maxindex = (index + slice->slice_count >= segment->slice_entries ? segment->slice_entries : index + slice->slice_count) - 1; + if (mi_slice_is_used(slice)) { // a page in use, we need at least MAX_SLICE_OFFSET valid back offsets + used_count++; + for (size_t i = 0; i <= MI_MAX_SLICE_OFFSET && index + i <= maxindex; i++) { + mi_assert_internal(segment->slices[index + i].slice_offset == i*sizeof(mi_slice_t)); + mi_assert_internal(i==0 || segment->slices[index + i].slice_count == 0); + mi_assert_internal(i==0 || segment->slices[index + i].xblock_size == 1); + } + // and the last entry as well (for coalescing) + const mi_slice_t* last = slice + slice->slice_count - 1; + if (last > slice && last < mi_segment_slices_end(segment)) { + mi_assert_internal(last->slice_offset == (slice->slice_count-1)*sizeof(mi_slice_t)); + mi_assert_internal(last->slice_count == 0); + mi_assert_internal(last->xblock_size == 1); + } + } + else { // free range of slices; only last slice needs a valid back offset + mi_slice_t* last = &segment->slices[maxindex]; + if (segment->kind != MI_SEGMENT_HUGE || slice->slice_count <= (segment->slice_entries - segment->segment_info_slices)) { + mi_assert_internal((uint8_t*)slice == (uint8_t*)last - last->slice_offset); + } + mi_assert_internal(slice == last || last->slice_count == 0 ); + mi_assert_internal(last->xblock_size == 0 || (segment->kind==MI_SEGMENT_HUGE && last->xblock_size==1)); + if (segment->kind != MI_SEGMENT_HUGE && segment->thread_id != 0) { // segment is not huge or abandoned + sq = mi_span_queue_for(slice->slice_count,tld); + mi_assert_internal(mi_span_queue_contains(sq,slice)); + } + } + slice = &segment->slices[maxindex+1]; + } + mi_assert_internal(slice == end); + mi_assert_internal(used_count == segment->used + 1); + return true; +} +#endif + +/* ----------------------------------------------------------- + Segment size calculations +----------------------------------------------------------- */ + +static size_t mi_segment_info_size(mi_segment_t* segment) { + return segment->segment_info_slices * MI_SEGMENT_SLICE_SIZE; +} + +static uint8_t* _mi_segment_page_start_from_slice(const mi_segment_t* segment, const mi_slice_t* slice, size_t xblock_size, size_t* page_size) +{ + ptrdiff_t idx = slice - segment->slices; + size_t psize = (size_t)slice->slice_count * MI_SEGMENT_SLICE_SIZE; + // make the start not OS page aligned for smaller blocks to avoid page/cache effects + // note: the offset must always be an xblock_size multiple since we assume small allocations + // are aligned (see `mi_heap_malloc_aligned`). + size_t start_offset = 0; + if (xblock_size >= MI_INTPTR_SIZE) { + if (xblock_size <= 64) { start_offset = 3*xblock_size; } + else if (xblock_size <= 512) { start_offset = xblock_size; } + } + if (page_size != NULL) { *page_size = psize - start_offset; } + return (uint8_t*)segment + ((idx*MI_SEGMENT_SLICE_SIZE) + start_offset); +} + +// Start of the page available memory; can be used on uninitialized pages +uint8_t* _mi_segment_page_start(const mi_segment_t* segment, const mi_page_t* page, size_t* page_size) +{ + const mi_slice_t* slice = mi_page_to_slice((mi_page_t*)page); + uint8_t* p = _mi_segment_page_start_from_slice(segment, slice, page->xblock_size, page_size); + mi_assert_internal(page->xblock_size > 0 || _mi_ptr_page(p) == page); + mi_assert_internal(_mi_ptr_segment(p) == segment); + return p; +} + + +static size_t mi_segment_calculate_slices(size_t required, size_t* pre_size, size_t* info_slices) { + size_t page_size = _mi_os_page_size(); + size_t isize = _mi_align_up(sizeof(mi_segment_t), page_size); + size_t guardsize = 0; + + if (MI_SECURE>0) { + // in secure mode, we set up a protected page in between the segment info + // and the page data (and one at the end of the segment) + guardsize = page_size; + if (required > 0) { + required = _mi_align_up(required, MI_SEGMENT_SLICE_SIZE) + page_size; + } + } + + if (pre_size != NULL) *pre_size = isize; + isize = _mi_align_up(isize + guardsize, MI_SEGMENT_SLICE_SIZE); + if (info_slices != NULL) *info_slices = isize / MI_SEGMENT_SLICE_SIZE; + size_t segment_size = (required==0 ? MI_SEGMENT_SIZE : _mi_align_up( required + isize + guardsize, MI_SEGMENT_SLICE_SIZE) ); + mi_assert_internal(segment_size % MI_SEGMENT_SLICE_SIZE == 0); + return (segment_size / MI_SEGMENT_SLICE_SIZE); +} + + +/* ---------------------------------------------------------------------------- +Segment caches +We keep a small segment cache per thread to increase local +reuse and avoid setting/clearing guard pages in secure mode. +------------------------------------------------------------------------------- */ + +static void mi_segments_track_size(long segment_size, mi_segments_tld_t* tld) { + if (segment_size>=0) _mi_stat_increase(&tld->stats->segments,1); + else _mi_stat_decrease(&tld->stats->segments,1); + tld->count += (segment_size >= 0 ? 1 : -1); + if (tld->count > tld->peak_count) tld->peak_count = tld->count; + tld->current_size += segment_size; + if (tld->current_size > tld->peak_size) tld->peak_size = tld->current_size; +} + +static void mi_segment_os_free(mi_segment_t* segment, mi_segments_tld_t* tld) { + segment->thread_id = 0; + _mi_segment_map_freed_at(segment); + mi_segments_track_size(-((long)mi_segment_size(segment)),tld); + if (MI_SECURE>0) { + // _mi_os_unprotect(segment, mi_segment_size(segment)); // ensure no more guard pages are set + // unprotect the guard pages; we cannot just unprotect the whole segment size as part may be decommitted + size_t os_pagesize = _mi_os_page_size(); + _mi_os_unprotect((uint8_t*)segment + mi_segment_info_size(segment) - os_pagesize, os_pagesize); + uint8_t* end = (uint8_t*)segment + mi_segment_size(segment) - os_pagesize; + _mi_os_unprotect(end, os_pagesize); + } + + // purge delayed decommits now? (no, leave it to the arena) + // mi_segment_try_purge(segment,true,tld->stats); + + const size_t size = mi_segment_size(segment); + const size_t csize = _mi_commit_mask_committed_size(&segment->commit_mask, size); + + _mi_abandoned_await_readers(); // wait until safe to free + _mi_arena_free(segment, mi_segment_size(segment), csize, segment->memid, tld->stats); +} + +// called by threads that are terminating +void _mi_segment_thread_collect(mi_segments_tld_t* tld) { + MI_UNUSED(tld); + // nothing to do +} + + +/* ----------------------------------------------------------- + Commit/Decommit ranges +----------------------------------------------------------- */ + +static void mi_segment_commit_mask(mi_segment_t* segment, bool conservative, uint8_t* p, size_t size, uint8_t** start_p, size_t* full_size, mi_commit_mask_t* cm) { + mi_assert_internal(_mi_ptr_segment(p + 1) == segment); + mi_assert_internal(segment->kind != MI_SEGMENT_HUGE); + mi_commit_mask_create_empty(cm); + if (size == 0 || size > MI_SEGMENT_SIZE || segment->kind == MI_SEGMENT_HUGE) return; + const size_t segstart = mi_segment_info_size(segment); + const size_t segsize = mi_segment_size(segment); + if (p >= (uint8_t*)segment + segsize) return; + + size_t pstart = (p - (uint8_t*)segment); + mi_assert_internal(pstart + size <= segsize); + + size_t start; + size_t end; + if (conservative) { + // decommit conservative + start = _mi_align_up(pstart, MI_COMMIT_SIZE); + end = _mi_align_down(pstart + size, MI_COMMIT_SIZE); + mi_assert_internal(start >= segstart); + mi_assert_internal(end <= segsize); + } + else { + // commit liberal + start = _mi_align_down(pstart, MI_MINIMAL_COMMIT_SIZE); + end = _mi_align_up(pstart + size, MI_MINIMAL_COMMIT_SIZE); + } + if (pstart >= segstart && start < segstart) { // note: the mask is also calculated for an initial commit of the info area + start = segstart; + } + if (end > segsize) { + end = segsize; + } + + mi_assert_internal(start <= pstart && (pstart + size) <= end); + mi_assert_internal(start % MI_COMMIT_SIZE==0 && end % MI_COMMIT_SIZE == 0); + *start_p = (uint8_t*)segment + start; + *full_size = (end > start ? end - start : 0); + if (*full_size == 0) return; + + size_t bitidx = start / MI_COMMIT_SIZE; + mi_assert_internal(bitidx < MI_COMMIT_MASK_BITS); + + size_t bitcount = *full_size / MI_COMMIT_SIZE; // can be 0 + if (bitidx + bitcount > MI_COMMIT_MASK_BITS) { + _mi_warning_message("commit mask overflow: idx=%zu count=%zu start=%zx end=%zx p=0x%p size=%zu fullsize=%zu\n", bitidx, bitcount, start, end, p, size, *full_size); + } + mi_assert_internal((bitidx + bitcount) <= MI_COMMIT_MASK_BITS); + mi_commit_mask_create(bitidx, bitcount, cm); +} + +static bool mi_segment_commit(mi_segment_t* segment, uint8_t* p, size_t size, mi_stats_t* stats) { + mi_assert_internal(mi_commit_mask_all_set(&segment->commit_mask, &segment->purge_mask)); + + // commit liberal + uint8_t* start = NULL; + size_t full_size = 0; + mi_commit_mask_t mask; + mi_segment_commit_mask(segment, false /* conservative? */, p, size, &start, &full_size, &mask); + if (mi_commit_mask_is_empty(&mask) || full_size == 0) return true; + + if (!mi_commit_mask_all_set(&segment->commit_mask, &mask)) { + // committing + bool is_zero = false; + mi_commit_mask_t cmask; + mi_commit_mask_create_intersect(&segment->commit_mask, &mask, &cmask); + _mi_stat_decrease(&_mi_stats_main.committed, _mi_commit_mask_committed_size(&cmask, MI_SEGMENT_SIZE)); // adjust for overlap + if (!_mi_os_commit(start, full_size, &is_zero, stats)) return false; + mi_commit_mask_set(&segment->commit_mask, &mask); + } + + // increase purge expiration when using part of delayed purges -- we assume more allocations are coming soon. + if (mi_commit_mask_any_set(&segment->purge_mask, &mask)) { + segment->purge_expire = _mi_clock_now() + mi_option_get(mi_option_purge_delay); + } + + // always clear any delayed purges in our range (as they are either committed now) + mi_commit_mask_clear(&segment->purge_mask, &mask); + return true; +} + +static bool mi_segment_ensure_committed(mi_segment_t* segment, uint8_t* p, size_t size, mi_stats_t* stats) { + mi_assert_internal(mi_commit_mask_all_set(&segment->commit_mask, &segment->purge_mask)); + // note: assumes commit_mask is always full for huge segments as otherwise the commit mask bits can overflow + if (mi_commit_mask_is_full(&segment->commit_mask) && mi_commit_mask_is_empty(&segment->purge_mask)) return true; // fully committed + mi_assert_internal(segment->kind != MI_SEGMENT_HUGE); + return mi_segment_commit(segment, p, size, stats); +} + +static bool mi_segment_purge(mi_segment_t* segment, uint8_t* p, size_t size, mi_stats_t* stats) { + mi_assert_internal(mi_commit_mask_all_set(&segment->commit_mask, &segment->purge_mask)); + if (!segment->allow_purge) return true; + + // purge conservative + uint8_t* start = NULL; + size_t full_size = 0; + mi_commit_mask_t mask; + mi_segment_commit_mask(segment, true /* conservative? */, p, size, &start, &full_size, &mask); + if (mi_commit_mask_is_empty(&mask) || full_size==0) return true; + + if (mi_commit_mask_any_set(&segment->commit_mask, &mask)) { + // purging + mi_assert_internal((void*)start != (void*)segment); + mi_assert_internal(segment->allow_decommit); + const bool decommitted = _mi_os_purge(start, full_size, stats); // reset or decommit + if (decommitted) { + mi_commit_mask_t cmask; + mi_commit_mask_create_intersect(&segment->commit_mask, &mask, &cmask); + _mi_stat_increase(&_mi_stats_main.committed, full_size - _mi_commit_mask_committed_size(&cmask, MI_SEGMENT_SIZE)); // adjust for double counting + mi_commit_mask_clear(&segment->commit_mask, &mask); + } + } + + // always clear any scheduled purges in our range + mi_commit_mask_clear(&segment->purge_mask, &mask); + return true; +} + +static void mi_segment_schedule_purge(mi_segment_t* segment, uint8_t* p, size_t size, mi_stats_t* stats) { + if (!segment->allow_purge) return; + + if (mi_option_get(mi_option_purge_delay) == 0) { + mi_segment_purge(segment, p, size, stats); + } + else { + // register for future purge in the purge mask + uint8_t* start = NULL; + size_t full_size = 0; + mi_commit_mask_t mask; + mi_segment_commit_mask(segment, true /*conservative*/, p, size, &start, &full_size, &mask); + if (mi_commit_mask_is_empty(&mask) || full_size==0) return; + + // update delayed commit + mi_assert_internal(segment->purge_expire > 0 || mi_commit_mask_is_empty(&segment->purge_mask)); + mi_commit_mask_t cmask; + mi_commit_mask_create_intersect(&segment->commit_mask, &mask, &cmask); // only purge what is committed; span_free may try to decommit more + mi_commit_mask_set(&segment->purge_mask, &cmask); + mi_msecs_t now = _mi_clock_now(); + if (segment->purge_expire == 0) { + // no previous purgess, initialize now + segment->purge_expire = now + mi_option_get(mi_option_purge_delay); + } + else if (segment->purge_expire <= now) { + // previous purge mask already expired + if (segment->purge_expire + mi_option_get(mi_option_purge_extend_delay) <= now) { + mi_segment_try_purge(segment, true, stats); + } + else { + segment->purge_expire = now + mi_option_get(mi_option_purge_extend_delay); // (mi_option_get(mi_option_purge_delay) / 8); // wait a tiny bit longer in case there is a series of free's + } + } + else { + // previous purge mask is not yet expired, increase the expiration by a bit. + segment->purge_expire += mi_option_get(mi_option_purge_extend_delay); + } + } +} + +static void mi_segment_try_purge(mi_segment_t* segment, bool force, mi_stats_t* stats) { + if (!segment->allow_purge || mi_commit_mask_is_empty(&segment->purge_mask)) return; + mi_msecs_t now = _mi_clock_now(); + if (!force && now < segment->purge_expire) return; + + mi_commit_mask_t mask = segment->purge_mask; + segment->purge_expire = 0; + mi_commit_mask_create_empty(&segment->purge_mask); + + size_t idx; + size_t count; + mi_commit_mask_foreach(&mask, idx, count) { + // if found, decommit that sequence + if (count > 0) { + uint8_t* p = (uint8_t*)segment + (idx*MI_COMMIT_SIZE); + size_t size = count * MI_COMMIT_SIZE; + mi_segment_purge(segment, p, size, stats); + } + } + mi_commit_mask_foreach_end() + mi_assert_internal(mi_commit_mask_is_empty(&segment->purge_mask)); +} + + +/* ----------------------------------------------------------- + Span free +----------------------------------------------------------- */ + +static bool mi_segment_is_abandoned(mi_segment_t* segment) { + return (segment->thread_id == 0); +} + +// note: can be called on abandoned segments +static void mi_segment_span_free(mi_segment_t* segment, size_t slice_index, size_t slice_count, bool allow_purge, mi_segments_tld_t* tld) { + mi_assert_internal(slice_index < segment->slice_entries); + mi_span_queue_t* sq = (segment->kind == MI_SEGMENT_HUGE || mi_segment_is_abandoned(segment) + ? NULL : mi_span_queue_for(slice_count,tld)); + if (slice_count==0) slice_count = 1; + mi_assert_internal(slice_index + slice_count - 1 < segment->slice_entries); + + // set first and last slice (the intermediates can be undetermined) + mi_slice_t* slice = &segment->slices[slice_index]; + slice->slice_count = (uint32_t)slice_count; + mi_assert_internal(slice->slice_count == slice_count); // no overflow? + slice->slice_offset = 0; + if (slice_count > 1) { + mi_slice_t* last = &segment->slices[slice_index + slice_count - 1]; + last->slice_count = 0; + last->slice_offset = (uint32_t)(sizeof(mi_page_t)*(slice_count - 1)); + last->xblock_size = 0; + } + + // perhaps decommit + if (allow_purge) { + mi_segment_schedule_purge(segment, mi_slice_start(slice), slice_count * MI_SEGMENT_SLICE_SIZE, tld->stats); + } + + // and push it on the free page queue (if it was not a huge page) + if (sq != NULL) mi_span_queue_push( sq, slice ); + else slice->xblock_size = 0; // mark huge page as free anyways +} + +/* +// called from reclaim to add existing free spans +static void mi_segment_span_add_free(mi_slice_t* slice, mi_segments_tld_t* tld) { + mi_segment_t* segment = _mi_ptr_segment(slice); + mi_assert_internal(slice->xblock_size==0 && slice->slice_count>0 && slice->slice_offset==0); + size_t slice_index = mi_slice_index(slice); + mi_segment_span_free(segment,slice_index,slice->slice_count,tld); +} +*/ + +static void mi_segment_span_remove_from_queue(mi_slice_t* slice, mi_segments_tld_t* tld) { + mi_assert_internal(slice->slice_count > 0 && slice->slice_offset==0 && slice->xblock_size==0); + mi_assert_internal(_mi_ptr_segment(slice)->kind != MI_SEGMENT_HUGE); + mi_span_queue_t* sq = mi_span_queue_for(slice->slice_count, tld); + mi_span_queue_delete(sq, slice); +} + +// note: can be called on abandoned segments +static mi_slice_t* mi_segment_span_free_coalesce(mi_slice_t* slice, mi_segments_tld_t* tld) { + mi_assert_internal(slice != NULL && slice->slice_count > 0 && slice->slice_offset == 0); + mi_segment_t* segment = _mi_ptr_segment(slice); + bool is_abandoned = mi_segment_is_abandoned(segment); + + // for huge pages, just mark as free but don't add to the queues + if (segment->kind == MI_SEGMENT_HUGE) { + // issue #691: segment->used can be 0 if the huge page block was freed while abandoned (reclaim will get here in that case) + mi_assert_internal((segment->used==0 && slice->xblock_size==0) || segment->used == 1); // decreased right after this call in `mi_segment_page_clear` + slice->xblock_size = 0; // mark as free anyways + // we should mark the last slice `xblock_size=0` now to maintain invariants but we skip it to + // avoid a possible cache miss (and the segment is about to be freed) + return slice; + } + + // otherwise coalesce the span and add to the free span queues + size_t slice_count = slice->slice_count; + mi_slice_t* next = slice + slice->slice_count; + mi_assert_internal(next <= mi_segment_slices_end(segment)); + if (next < mi_segment_slices_end(segment) && next->xblock_size==0) { + // free next block -- remove it from free and merge + mi_assert_internal(next->slice_count > 0 && next->slice_offset==0); + slice_count += next->slice_count; // extend + if (!is_abandoned) { mi_segment_span_remove_from_queue(next, tld); } + } + if (slice > segment->slices) { + mi_slice_t* prev = mi_slice_first(slice - 1); + mi_assert_internal(prev >= segment->slices); + if (prev->xblock_size==0) { + // free previous slice -- remove it from free and merge + mi_assert_internal(prev->slice_count > 0 && prev->slice_offset==0); + slice_count += prev->slice_count; + if (!is_abandoned) { mi_segment_span_remove_from_queue(prev, tld); } + slice = prev; + } + } + + // and add the new free page + mi_segment_span_free(segment, mi_slice_index(slice), slice_count, true, tld); + return slice; +} + + + +/* ----------------------------------------------------------- + Page allocation +----------------------------------------------------------- */ + +// Note: may still return NULL if committing the memory failed +static mi_page_t* mi_segment_span_allocate(mi_segment_t* segment, size_t slice_index, size_t slice_count, mi_segments_tld_t* tld) { + mi_assert_internal(slice_index < segment->slice_entries); + mi_slice_t* const slice = &segment->slices[slice_index]; + mi_assert_internal(slice->xblock_size==0 || slice->xblock_size==1); + + // commit before changing the slice data + if (!mi_segment_ensure_committed(segment, _mi_segment_page_start_from_slice(segment, slice, 0, NULL), slice_count * MI_SEGMENT_SLICE_SIZE, tld->stats)) { + return NULL; // commit failed! + } + + // convert the slices to a page + slice->slice_offset = 0; + slice->slice_count = (uint32_t)slice_count; + mi_assert_internal(slice->slice_count == slice_count); + const size_t bsize = slice_count * MI_SEGMENT_SLICE_SIZE; + slice->xblock_size = (uint32_t)(bsize >= MI_HUGE_BLOCK_SIZE ? MI_HUGE_BLOCK_SIZE : bsize); + mi_page_t* page = mi_slice_to_page(slice); + mi_assert_internal(mi_page_block_size(page) == bsize); + + // set slice back pointers for the first MI_MAX_SLICE_OFFSET entries + size_t extra = slice_count-1; + if (extra > MI_MAX_SLICE_OFFSET) extra = MI_MAX_SLICE_OFFSET; + if (slice_index + extra >= segment->slice_entries) extra = segment->slice_entries - slice_index - 1; // huge objects may have more slices than avaiable entries in the segment->slices + + mi_slice_t* slice_next = slice + 1; + for (size_t i = 1; i <= extra; i++, slice_next++) { + slice_next->slice_offset = (uint32_t)(sizeof(mi_slice_t)*i); + slice_next->slice_count = 0; + slice_next->xblock_size = 1; + } + + // and also for the last one (if not set already) (the last one is needed for coalescing and for large alignments) + // note: the cast is needed for ubsan since the index can be larger than MI_SLICES_PER_SEGMENT for huge allocations (see #543) + mi_slice_t* last = slice + slice_count - 1; + mi_slice_t* end = (mi_slice_t*)mi_segment_slices_end(segment); + if (last > end) last = end; + if (last > slice) { + last->slice_offset = (uint32_t)(sizeof(mi_slice_t) * (last - slice)); + last->slice_count = 0; + last->xblock_size = 1; + } + + // and initialize the page + page->is_committed = true; + segment->used++; + return page; +} + +static void mi_segment_slice_split(mi_segment_t* segment, mi_slice_t* slice, size_t slice_count, mi_segments_tld_t* tld) { + mi_assert_internal(_mi_ptr_segment(slice) == segment); + mi_assert_internal(slice->slice_count >= slice_count); + mi_assert_internal(slice->xblock_size > 0); // no more in free queue + if (slice->slice_count <= slice_count) return; + mi_assert_internal(segment->kind != MI_SEGMENT_HUGE); + size_t next_index = mi_slice_index(slice) + slice_count; + size_t next_count = slice->slice_count - slice_count; + mi_segment_span_free(segment, next_index, next_count, false /* don't purge left-over part */, tld); + slice->slice_count = (uint32_t)slice_count; +} + +static mi_page_t* mi_segments_page_find_and_allocate(size_t slice_count, mi_arena_id_t req_arena_id, mi_segments_tld_t* tld) { + mi_assert_internal(slice_count*MI_SEGMENT_SLICE_SIZE <= MI_LARGE_OBJ_SIZE_MAX); + // search from best fit up + mi_span_queue_t* sq = mi_span_queue_for(slice_count, tld); + if (slice_count == 0) slice_count = 1; + while (sq <= &tld->spans[MI_SEGMENT_BIN_MAX]) { + for (mi_slice_t* slice = sq->first; slice != NULL; slice = slice->next) { + if (slice->slice_count >= slice_count) { + // found one + mi_segment_t* segment = _mi_ptr_segment(slice); + if (_mi_arena_memid_is_suitable(segment->memid, req_arena_id)) { + // found a suitable page span + mi_span_queue_delete(sq, slice); + + if (slice->slice_count > slice_count) { + mi_segment_slice_split(segment, slice, slice_count, tld); + } + mi_assert_internal(slice != NULL && slice->slice_count == slice_count && slice->xblock_size > 0); + mi_page_t* page = mi_segment_span_allocate(segment, mi_slice_index(slice), slice->slice_count, tld); + if (page == NULL) { + // commit failed; return NULL but first restore the slice + mi_segment_span_free_coalesce(slice, tld); + return NULL; + } + return page; + } + } + } + sq++; + } + // could not find a page.. + return NULL; +} + + +/* ----------------------------------------------------------- + Segment allocation +----------------------------------------------------------- */ + +static mi_segment_t* mi_segment_os_alloc( size_t required, size_t page_alignment, bool eager_delayed, mi_arena_id_t req_arena_id, + size_t* psegment_slices, size_t* ppre_size, size_t* pinfo_slices, + bool commit, mi_segments_tld_t* tld, mi_os_tld_t* os_tld) + +{ + mi_memid_t memid; + bool allow_large = (!eager_delayed && (MI_SECURE == 0)); // only allow large OS pages once we are no longer lazy + size_t align_offset = 0; + size_t alignment = MI_SEGMENT_ALIGN; + + if (page_alignment > 0) { + // mi_assert_internal(huge_page != NULL); + mi_assert_internal(page_alignment >= MI_SEGMENT_ALIGN); + alignment = page_alignment; + const size_t info_size = (*pinfo_slices) * MI_SEGMENT_SLICE_SIZE; + align_offset = _mi_align_up( info_size, MI_SEGMENT_ALIGN ); + const size_t extra = align_offset - info_size; + // recalculate due to potential guard pages + *psegment_slices = mi_segment_calculate_slices(required + extra, ppre_size, pinfo_slices); + } + + const size_t segment_size = (*psegment_slices) * MI_SEGMENT_SLICE_SIZE; + mi_segment_t* segment = (mi_segment_t*)_mi_arena_alloc_aligned(segment_size, alignment, align_offset, commit, allow_large, req_arena_id, &memid, os_tld); + if (segment == NULL) { + return NULL; // failed to allocate + } + + // ensure metadata part of the segment is committed + mi_commit_mask_t commit_mask; + if (memid.initially_committed) { + mi_commit_mask_create_full(&commit_mask); + } + else { + // at least commit the info slices + const size_t commit_needed = _mi_divide_up((*pinfo_slices)*MI_SEGMENT_SLICE_SIZE, MI_COMMIT_SIZE); + mi_assert_internal(commit_needed>0); + mi_commit_mask_create(0, commit_needed, &commit_mask); + mi_assert_internal(commit_needed*MI_COMMIT_SIZE >= (*pinfo_slices)*MI_SEGMENT_SLICE_SIZE); + if (!_mi_os_commit(segment, commit_needed*MI_COMMIT_SIZE, NULL, tld->stats)) { + _mi_arena_free(segment,segment_size,0,memid,tld->stats); + return NULL; + } + } + mi_assert_internal(segment != NULL && (uintptr_t)segment % MI_SEGMENT_SIZE == 0); + + segment->memid = memid; + segment->allow_decommit = !memid.is_pinned; + segment->allow_purge = segment->allow_decommit && (mi_option_get(mi_option_purge_delay) >= 0); + segment->segment_size = segment_size; + segment->commit_mask = commit_mask; + segment->purge_expire = 0; + mi_commit_mask_create_empty(&segment->purge_mask); + mi_atomic_store_ptr_release(mi_segment_t, &segment->abandoned_next, NULL); // tsan + + mi_segments_track_size((long)(segment_size), tld); + _mi_segment_map_allocated_at(segment); + return segment; +} + + +// Allocate a segment from the OS aligned to `MI_SEGMENT_SIZE` . +static mi_segment_t* mi_segment_alloc(size_t required, size_t page_alignment, mi_arena_id_t req_arena_id, mi_segments_tld_t* tld, mi_os_tld_t* os_tld, mi_page_t** huge_page) +{ + mi_assert_internal((required==0 && huge_page==NULL) || (required>0 && huge_page != NULL)); + + // calculate needed sizes first + size_t info_slices; + size_t pre_size; + size_t segment_slices = mi_segment_calculate_slices(required, &pre_size, &info_slices); + + // Commit eagerly only if not the first N lazy segments (to reduce impact of many threads that allocate just a little) + const bool eager_delay = (// !_mi_os_has_overcommit() && // never delay on overcommit systems + _mi_current_thread_count() > 1 && // do not delay for the first N threads + tld->count < (size_t)mi_option_get(mi_option_eager_commit_delay)); + const bool eager = !eager_delay && mi_option_is_enabled(mi_option_eager_commit); + bool commit = eager || (required > 0); + + // Allocate the segment from the OS + mi_segment_t* segment = mi_segment_os_alloc(required, page_alignment, eager_delay, req_arena_id, + &segment_slices, &pre_size, &info_slices, commit, tld, os_tld); + if (segment == NULL) return NULL; + + // zero the segment info? -- not always needed as it may be zero initialized from the OS + if (!segment->memid.initially_zero) { + ptrdiff_t ofs = offsetof(mi_segment_t, next); + size_t prefix = offsetof(mi_segment_t, slices) - ofs; + size_t zsize = prefix + (sizeof(mi_slice_t) * (segment_slices + 1)); // one more + _mi_memzero((uint8_t*)segment + ofs, zsize); + } + + // initialize the rest of the segment info + const size_t slice_entries = (segment_slices > MI_SLICES_PER_SEGMENT ? MI_SLICES_PER_SEGMENT : segment_slices); + segment->segment_slices = segment_slices; + segment->segment_info_slices = info_slices; + segment->thread_id = _mi_thread_id(); + segment->cookie = _mi_ptr_cookie(segment); + segment->slice_entries = slice_entries; + segment->kind = (required == 0 ? MI_SEGMENT_NORMAL : MI_SEGMENT_HUGE); + + // _mi_memzero(segment->slices, sizeof(mi_slice_t)*(info_slices+1)); + _mi_stat_increase(&tld->stats->page_committed, mi_segment_info_size(segment)); + + // set up guard pages + size_t guard_slices = 0; + if (MI_SECURE>0) { + // in secure mode, we set up a protected page in between the segment info + // and the page data, and at the end of the segment. + size_t os_pagesize = _mi_os_page_size(); + mi_assert_internal(mi_segment_info_size(segment) - os_pagesize >= pre_size); + _mi_os_protect((uint8_t*)segment + mi_segment_info_size(segment) - os_pagesize, os_pagesize); + uint8_t* end = (uint8_t*)segment + mi_segment_size(segment) - os_pagesize; + mi_segment_ensure_committed(segment, end, os_pagesize, tld->stats); + _mi_os_protect(end, os_pagesize); + if (slice_entries == segment_slices) segment->slice_entries--; // don't use the last slice :-( + guard_slices = 1; + } + + // reserve first slices for segment info + mi_page_t* page0 = mi_segment_span_allocate(segment, 0, info_slices, tld); + mi_assert_internal(page0!=NULL); if (page0==NULL) return NULL; // cannot fail as we always commit in advance + mi_assert_internal(segment->used == 1); + segment->used = 0; // don't count our internal slices towards usage + + // initialize initial free pages + if (segment->kind == MI_SEGMENT_NORMAL) { // not a huge page + mi_assert_internal(huge_page==NULL); + mi_segment_span_free(segment, info_slices, segment->slice_entries - info_slices, false /* don't purge */, tld); + } + else { + mi_assert_internal(huge_page!=NULL); + mi_assert_internal(mi_commit_mask_is_empty(&segment->purge_mask)); + mi_assert_internal(mi_commit_mask_is_full(&segment->commit_mask)); + *huge_page = mi_segment_span_allocate(segment, info_slices, segment_slices - info_slices - guard_slices, tld); + mi_assert_internal(*huge_page != NULL); // cannot fail as we commit in advance + } + + mi_assert_expensive(mi_segment_is_valid(segment,tld)); + return segment; +} + + +static void mi_segment_free(mi_segment_t* segment, bool force, mi_segments_tld_t* tld) { + MI_UNUSED(force); + mi_assert_internal(segment != NULL); + mi_assert_internal(segment->next == NULL); + mi_assert_internal(segment->used == 0); + + // Remove the free pages + mi_slice_t* slice = &segment->slices[0]; + const mi_slice_t* end = mi_segment_slices_end(segment); + #if MI_DEBUG>1 + size_t page_count = 0; + #endif + while (slice < end) { + mi_assert_internal(slice->slice_count > 0); + mi_assert_internal(slice->slice_offset == 0); + mi_assert_internal(mi_slice_index(slice)==0 || slice->xblock_size == 0); // no more used pages .. + if (slice->xblock_size == 0 && segment->kind != MI_SEGMENT_HUGE) { + mi_segment_span_remove_from_queue(slice, tld); + } + #if MI_DEBUG>1 + page_count++; + #endif + slice = slice + slice->slice_count; + } + mi_assert_internal(page_count == 2); // first page is allocated by the segment itself + + // stats + _mi_stat_decrease(&tld->stats->page_committed, mi_segment_info_size(segment)); + + // return it to the OS + mi_segment_os_free(segment, tld); +} + + +/* ----------------------------------------------------------- + Page Free +----------------------------------------------------------- */ + +static void mi_segment_abandon(mi_segment_t* segment, mi_segments_tld_t* tld); + +// note: can be called on abandoned pages +static mi_slice_t* mi_segment_page_clear(mi_page_t* page, mi_segments_tld_t* tld) { + mi_assert_internal(page->xblock_size > 0); + mi_assert_internal(mi_page_all_free(page)); + mi_segment_t* segment = _mi_ptr_segment(page); + mi_assert_internal(segment->used > 0); + + size_t inuse = page->capacity * mi_page_block_size(page); + _mi_stat_decrease(&tld->stats->page_committed, inuse); + _mi_stat_decrease(&tld->stats->pages, 1); + + // reset the page memory to reduce memory pressure? + if (segment->allow_decommit && mi_option_is_enabled(mi_option_deprecated_page_reset)) { + size_t psize; + uint8_t* start = _mi_page_start(segment, page, &psize); + _mi_os_reset(start, psize, tld->stats); + } + + // zero the page data, but not the segment fields + page->is_zero_init = false; + ptrdiff_t ofs = offsetof(mi_page_t, capacity); + _mi_memzero((uint8_t*)page + ofs, sizeof(*page) - ofs); + page->xblock_size = 1; + + // and free it + mi_slice_t* slice = mi_segment_span_free_coalesce(mi_page_to_slice(page), tld); + segment->used--; + // cannot assert segment valid as it is called during reclaim + // mi_assert_expensive(mi_segment_is_valid(segment, tld)); + return slice; +} + +void _mi_segment_page_free(mi_page_t* page, bool force, mi_segments_tld_t* tld) +{ + mi_assert(page != NULL); + + mi_segment_t* segment = _mi_page_segment(page); + mi_assert_expensive(mi_segment_is_valid(segment,tld)); + + // mark it as free now + mi_segment_page_clear(page, tld); + mi_assert_expensive(mi_segment_is_valid(segment, tld)); + + if (segment->used == 0) { + // no more used pages; remove from the free list and free the segment + mi_segment_free(segment, force, tld); + } + else if (segment->used == segment->abandoned) { + // only abandoned pages; remove from free list and abandon + mi_segment_abandon(segment,tld); + } +} + + +/* ----------------------------------------------------------- +Abandonment + +When threads terminate, they can leave segments with +live blocks (reachable through other threads). Such segments +are "abandoned" and will be reclaimed by other threads to +reuse their pages and/or free them eventually + +We maintain a global list of abandoned segments that are +reclaimed on demand. Since this is shared among threads +the implementation needs to avoid the A-B-A problem on +popping abandoned segments: +We use tagged pointers to avoid accidentally identifying +reused segments, much like stamped references in Java. +Secondly, we maintain a reader counter to avoid resetting +or decommitting segments that have a pending read operation. + +Note: the current implementation is one possible design; +another way might be to keep track of abandoned segments +in the arenas/segment_cache's. This would have the advantage of keeping +all concurrent code in one place and not needing to deal +with ABA issues. The drawback is that it is unclear how to +scan abandoned segments efficiently in that case as they +would be spread among all other segments in the arenas. +----------------------------------------------------------- */ + +// Use the bottom 20-bits (on 64-bit) of the aligned segment pointers +// to put in a tag that increments on update to avoid the A-B-A problem. +#define MI_TAGGED_MASK MI_SEGMENT_MASK +typedef uintptr_t mi_tagged_segment_t; + +static mi_segment_t* mi_tagged_segment_ptr(mi_tagged_segment_t ts) { + return (mi_segment_t*)(ts & ~MI_TAGGED_MASK); +} + +static mi_tagged_segment_t mi_tagged_segment(mi_segment_t* segment, mi_tagged_segment_t ts) { + mi_assert_internal(((uintptr_t)segment & MI_TAGGED_MASK) == 0); + uintptr_t tag = ((ts & MI_TAGGED_MASK) + 1) & MI_TAGGED_MASK; + return ((uintptr_t)segment | tag); +} + +// This is a list of visited abandoned pages that were full at the time. +// this list migrates to `abandoned` when that becomes NULL. The use of +// this list reduces contention and the rate at which segments are visited. +static mi_decl_cache_align _Atomic(mi_segment_t*) abandoned_visited; // = NULL + +// The abandoned page list (tagged as it supports pop) +static mi_decl_cache_align _Atomic(mi_tagged_segment_t) abandoned; // = NULL + +// Maintain these for debug purposes (these counts may be a bit off) +static mi_decl_cache_align _Atomic(size_t) abandoned_count; +static mi_decl_cache_align _Atomic(size_t) abandoned_visited_count; + +// We also maintain a count of current readers of the abandoned list +// in order to prevent resetting/decommitting segment memory if it might +// still be read. +static mi_decl_cache_align _Atomic(size_t) abandoned_readers; // = 0 + +// Push on the visited list +static void mi_abandoned_visited_push(mi_segment_t* segment) { + mi_assert_internal(segment->thread_id == 0); + mi_assert_internal(mi_atomic_load_ptr_relaxed(mi_segment_t,&segment->abandoned_next) == NULL); + mi_assert_internal(segment->next == NULL); + mi_assert_internal(segment->used > 0); + mi_segment_t* anext = mi_atomic_load_ptr_relaxed(mi_segment_t, &abandoned_visited); + do { + mi_atomic_store_ptr_release(mi_segment_t, &segment->abandoned_next, anext); + } while (!mi_atomic_cas_ptr_weak_release(mi_segment_t, &abandoned_visited, &anext, segment)); + mi_atomic_increment_relaxed(&abandoned_visited_count); +} + +// Move the visited list to the abandoned list. +static bool mi_abandoned_visited_revisit(void) +{ + // quick check if the visited list is empty + if (mi_atomic_load_ptr_relaxed(mi_segment_t, &abandoned_visited) == NULL) return false; + + // grab the whole visited list + mi_segment_t* first = mi_atomic_exchange_ptr_acq_rel(mi_segment_t, &abandoned_visited, NULL); + if (first == NULL) return false; + + // first try to swap directly if the abandoned list happens to be NULL + mi_tagged_segment_t afirst; + mi_tagged_segment_t ts = mi_atomic_load_relaxed(&abandoned); + if (mi_tagged_segment_ptr(ts)==NULL) { + size_t count = mi_atomic_load_relaxed(&abandoned_visited_count); + afirst = mi_tagged_segment(first, ts); + if (mi_atomic_cas_strong_acq_rel(&abandoned, &ts, afirst)) { + mi_atomic_add_relaxed(&abandoned_count, count); + mi_atomic_sub_relaxed(&abandoned_visited_count, count); + return true; + } + } + + // find the last element of the visited list: O(n) + mi_segment_t* last = first; + mi_segment_t* next; + while ((next = mi_atomic_load_ptr_relaxed(mi_segment_t, &last->abandoned_next)) != NULL) { + last = next; + } + + // and atomically prepend to the abandoned list + // (no need to increase the readers as we don't access the abandoned segments) + mi_tagged_segment_t anext = mi_atomic_load_relaxed(&abandoned); + size_t count; + do { + count = mi_atomic_load_relaxed(&abandoned_visited_count); + mi_atomic_store_ptr_release(mi_segment_t, &last->abandoned_next, mi_tagged_segment_ptr(anext)); + afirst = mi_tagged_segment(first, anext); + } while (!mi_atomic_cas_weak_release(&abandoned, &anext, afirst)); + mi_atomic_add_relaxed(&abandoned_count, count); + mi_atomic_sub_relaxed(&abandoned_visited_count, count); + return true; +} + +// Push on the abandoned list. +static void mi_abandoned_push(mi_segment_t* segment) { + mi_assert_internal(segment->thread_id == 0); + mi_assert_internal(mi_atomic_load_ptr_relaxed(mi_segment_t, &segment->abandoned_next) == NULL); + mi_assert_internal(segment->next == NULL); + mi_assert_internal(segment->used > 0); + mi_tagged_segment_t next; + mi_tagged_segment_t ts = mi_atomic_load_relaxed(&abandoned); + do { + mi_atomic_store_ptr_release(mi_segment_t, &segment->abandoned_next, mi_tagged_segment_ptr(ts)); + next = mi_tagged_segment(segment, ts); + } while (!mi_atomic_cas_weak_release(&abandoned, &ts, next)); + mi_atomic_increment_relaxed(&abandoned_count); +} + +// Wait until there are no more pending reads on segments that used to be in the abandoned list +// called for example from `arena.c` before decommitting +void _mi_abandoned_await_readers(void) { + size_t n; + do { + n = mi_atomic_load_acquire(&abandoned_readers); + if (n != 0) mi_atomic_yield(); + } while (n != 0); +} + +// Pop from the abandoned list +static mi_segment_t* mi_abandoned_pop(void) { + mi_segment_t* segment; + // Check efficiently if it is empty (or if the visited list needs to be moved) + mi_tagged_segment_t ts = mi_atomic_load_relaxed(&abandoned); + segment = mi_tagged_segment_ptr(ts); + if mi_likely(segment == NULL) { + if mi_likely(!mi_abandoned_visited_revisit()) { // try to swap in the visited list on NULL + return NULL; + } + } + + // Do a pop. We use a reader count to prevent + // a segment to be decommitted while a read is still pending, + // and a tagged pointer to prevent A-B-A link corruption. + // (this is called from `region.c:_mi_mem_free` for example) + mi_atomic_increment_relaxed(&abandoned_readers); // ensure no segment gets decommitted + mi_tagged_segment_t next = 0; + ts = mi_atomic_load_acquire(&abandoned); + do { + segment = mi_tagged_segment_ptr(ts); + if (segment != NULL) { + mi_segment_t* anext = mi_atomic_load_ptr_relaxed(mi_segment_t, &segment->abandoned_next); + next = mi_tagged_segment(anext, ts); // note: reads the segment's `abandoned_next` field so should not be decommitted + } + } while (segment != NULL && !mi_atomic_cas_weak_acq_rel(&abandoned, &ts, next)); + mi_atomic_decrement_relaxed(&abandoned_readers); // release reader lock + if (segment != NULL) { + mi_atomic_store_ptr_release(mi_segment_t, &segment->abandoned_next, NULL); + mi_atomic_decrement_relaxed(&abandoned_count); + } + return segment; +} + +/* ----------------------------------------------------------- + Abandon segment/page +----------------------------------------------------------- */ + +static void mi_segment_abandon(mi_segment_t* segment, mi_segments_tld_t* tld) { + mi_assert_internal(segment->used == segment->abandoned); + mi_assert_internal(segment->used > 0); + mi_assert_internal(mi_atomic_load_ptr_relaxed(mi_segment_t, &segment->abandoned_next) == NULL); + mi_assert_internal(segment->abandoned_visits == 0); + mi_assert_expensive(mi_segment_is_valid(segment,tld)); + + // remove the free pages from the free page queues + mi_slice_t* slice = &segment->slices[0]; + const mi_slice_t* end = mi_segment_slices_end(segment); + while (slice < end) { + mi_assert_internal(slice->slice_count > 0); + mi_assert_internal(slice->slice_offset == 0); + if (slice->xblock_size == 0) { // a free page + mi_segment_span_remove_from_queue(slice,tld); + slice->xblock_size = 0; // but keep it free + } + slice = slice + slice->slice_count; + } + + // perform delayed decommits (forcing is much slower on mstress) + mi_segment_try_purge(segment, mi_option_is_enabled(mi_option_abandoned_page_purge) /* force? */, tld->stats); + + // all pages in the segment are abandoned; add it to the abandoned list + _mi_stat_increase(&tld->stats->segments_abandoned, 1); + mi_segments_track_size(-((long)mi_segment_size(segment)), tld); + segment->thread_id = 0; + mi_atomic_store_ptr_release(mi_segment_t, &segment->abandoned_next, NULL); + segment->abandoned_visits = 1; // from 0 to 1 to signify it is abandoned + mi_abandoned_push(segment); +} + +void _mi_segment_page_abandon(mi_page_t* page, mi_segments_tld_t* tld) { + mi_assert(page != NULL); + mi_assert_internal(mi_page_thread_free_flag(page)==MI_NEVER_DELAYED_FREE); + mi_assert_internal(mi_page_heap(page) == NULL); + mi_segment_t* segment = _mi_page_segment(page); + + mi_assert_expensive(mi_segment_is_valid(segment,tld)); + segment->abandoned++; + + _mi_stat_increase(&tld->stats->pages_abandoned, 1); + mi_assert_internal(segment->abandoned <= segment->used); + if (segment->used == segment->abandoned) { + // all pages are abandoned, abandon the entire segment + mi_segment_abandon(segment, tld); + } +} + +/* ----------------------------------------------------------- + Reclaim abandoned pages +----------------------------------------------------------- */ + +static mi_slice_t* mi_slices_start_iterate(mi_segment_t* segment, const mi_slice_t** end) { + mi_slice_t* slice = &segment->slices[0]; + *end = mi_segment_slices_end(segment); + mi_assert_internal(slice->slice_count>0 && slice->xblock_size>0); // segment allocated page + slice = slice + slice->slice_count; // skip the first segment allocated page + return slice; +} + +// Possibly free pages and check if free space is available +static bool mi_segment_check_free(mi_segment_t* segment, size_t slices_needed, size_t block_size, mi_segments_tld_t* tld) +{ + mi_assert_internal(block_size < MI_HUGE_BLOCK_SIZE); + mi_assert_internal(mi_segment_is_abandoned(segment)); + bool has_page = false; + + // for all slices + const mi_slice_t* end; + mi_slice_t* slice = mi_slices_start_iterate(segment, &end); + while (slice < end) { + mi_assert_internal(slice->slice_count > 0); + mi_assert_internal(slice->slice_offset == 0); + if (mi_slice_is_used(slice)) { // used page + // ensure used count is up to date and collect potential concurrent frees + mi_page_t* const page = mi_slice_to_page(slice); + _mi_page_free_collect(page, false); + if (mi_page_all_free(page)) { + // if this page is all free now, free it without adding to any queues (yet) + mi_assert_internal(page->next == NULL && page->prev==NULL); + _mi_stat_decrease(&tld->stats->pages_abandoned, 1); + segment->abandoned--; + slice = mi_segment_page_clear(page, tld); // re-assign slice due to coalesce! + mi_assert_internal(!mi_slice_is_used(slice)); + if (slice->slice_count >= slices_needed) { + has_page = true; + } + } + else { + if (page->xblock_size == block_size && mi_page_has_any_available(page)) { + // a page has available free blocks of the right size + has_page = true; + } + } + } + else { + // empty span + if (slice->slice_count >= slices_needed) { + has_page = true; + } + } + slice = slice + slice->slice_count; + } + return has_page; +} + +// Reclaim an abandoned segment; returns NULL if the segment was freed +// set `right_page_reclaimed` to `true` if it reclaimed a page of the right `block_size` that was not full. +static mi_segment_t* mi_segment_reclaim(mi_segment_t* segment, mi_heap_t* heap, size_t requested_block_size, bool* right_page_reclaimed, mi_segments_tld_t* tld) { + mi_assert_internal(mi_atomic_load_ptr_relaxed(mi_segment_t, &segment->abandoned_next) == NULL); + mi_assert_expensive(mi_segment_is_valid(segment, tld)); + if (right_page_reclaimed != NULL) { *right_page_reclaimed = false; } + + segment->thread_id = _mi_thread_id(); + segment->abandoned_visits = 0; + mi_segments_track_size((long)mi_segment_size(segment), tld); + mi_assert_internal(segment->next == NULL); + _mi_stat_decrease(&tld->stats->segments_abandoned, 1); + + // for all slices + const mi_slice_t* end; + mi_slice_t* slice = mi_slices_start_iterate(segment, &end); + while (slice < end) { + mi_assert_internal(slice->slice_count > 0); + mi_assert_internal(slice->slice_offset == 0); + if (mi_slice_is_used(slice)) { + // in use: reclaim the page in our heap + mi_page_t* page = mi_slice_to_page(slice); + mi_assert_internal(page->is_committed); + mi_assert_internal(mi_page_thread_free_flag(page)==MI_NEVER_DELAYED_FREE); + mi_assert_internal(mi_page_heap(page) == NULL); + mi_assert_internal(page->next == NULL && page->prev==NULL); + _mi_stat_decrease(&tld->stats->pages_abandoned, 1); + segment->abandoned--; + // set the heap again and allow delayed free again + mi_page_set_heap(page, heap); + _mi_page_use_delayed_free(page, MI_USE_DELAYED_FREE, true); // override never (after heap is set) + _mi_page_free_collect(page, false); // ensure used count is up to date + if (mi_page_all_free(page)) { + // if everything free by now, free the page + slice = mi_segment_page_clear(page, tld); // set slice again due to coalesceing + } + else { + // otherwise reclaim it into the heap + _mi_page_reclaim(heap, page); + if (requested_block_size == page->xblock_size && mi_page_has_any_available(page)) { + if (right_page_reclaimed != NULL) { *right_page_reclaimed = true; } + } + } + } + else { + // the span is free, add it to our page queues + slice = mi_segment_span_free_coalesce(slice, tld); // set slice again due to coalesceing + } + mi_assert_internal(slice->slice_count>0 && slice->slice_offset==0); + slice = slice + slice->slice_count; + } + + mi_assert(segment->abandoned == 0); + if (segment->used == 0) { // due to page_clear + mi_assert_internal(right_page_reclaimed == NULL || !(*right_page_reclaimed)); + mi_segment_free(segment, false, tld); + return NULL; + } + else { + return segment; + } +} + + +void _mi_abandoned_reclaim_all(mi_heap_t* heap, mi_segments_tld_t* tld) { + mi_segment_t* segment; + while ((segment = mi_abandoned_pop()) != NULL) { + mi_segment_reclaim(segment, heap, 0, NULL, tld); + } +} + +static mi_segment_t* mi_segment_try_reclaim(mi_heap_t* heap, size_t needed_slices, size_t block_size, bool* reclaimed, mi_segments_tld_t* tld) +{ + *reclaimed = false; + mi_segment_t* segment; + long max_tries = mi_option_get_clamp(mi_option_max_segment_reclaim, 8, 1024); // limit the work to bound allocation times + while ((max_tries-- > 0) && ((segment = mi_abandoned_pop()) != NULL)) { + segment->abandoned_visits++; + // todo: an arena exclusive heap will potentially visit many abandoned unsuitable segments + // and push them into the visited list and use many tries. Perhaps we can skip non-suitable ones in a better way? + bool is_suitable = _mi_heap_memid_is_suitable(heap, segment->memid); + bool has_page = mi_segment_check_free(segment,needed_slices,block_size,tld); // try to free up pages (due to concurrent frees) + if (segment->used == 0) { + // free the segment (by forced reclaim) to make it available to other threads. + // note1: we prefer to free a segment as that might lead to reclaiming another + // segment that is still partially used. + // note2: we could in principle optimize this by skipping reclaim and directly + // freeing but that would violate some invariants temporarily) + mi_segment_reclaim(segment, heap, 0, NULL, tld); + } + else if (has_page && is_suitable) { + // found a large enough free span, or a page of the right block_size with free space + // we return the result of reclaim (which is usually `segment`) as it might free + // the segment due to concurrent frees (in which case `NULL` is returned). + return mi_segment_reclaim(segment, heap, block_size, reclaimed, tld); + } + else if (segment->abandoned_visits > 3 && is_suitable) { + // always reclaim on 3rd visit to limit the abandoned queue length. + mi_segment_reclaim(segment, heap, 0, NULL, tld); + } + else { + // otherwise, push on the visited list so it gets not looked at too quickly again + mi_segment_try_purge(segment, true /* force? */, tld->stats); // force purge if needed as we may not visit soon again + mi_abandoned_visited_push(segment); + } + } + return NULL; +} + + +void _mi_abandoned_collect(mi_heap_t* heap, bool force, mi_segments_tld_t* tld) +{ + mi_segment_t* segment; + int max_tries = (force ? 16*1024 : 1024); // limit latency + if (force) { + mi_abandoned_visited_revisit(); + } + while ((max_tries-- > 0) && ((segment = mi_abandoned_pop()) != NULL)) { + mi_segment_check_free(segment,0,0,tld); // try to free up pages (due to concurrent frees) + if (segment->used == 0) { + // free the segment (by forced reclaim) to make it available to other threads. + // note: we could in principle optimize this by skipping reclaim and directly + // freeing but that would violate some invariants temporarily) + mi_segment_reclaim(segment, heap, 0, NULL, tld); + } + else { + // otherwise, purge if needed and push on the visited list + // note: forced purge can be expensive if many threads are destroyed/created as in mstress. + mi_segment_try_purge(segment, force, tld->stats); + mi_abandoned_visited_push(segment); + } + } +} + +/* ----------------------------------------------------------- + Reclaim or allocate +----------------------------------------------------------- */ + +static mi_segment_t* mi_segment_reclaim_or_alloc(mi_heap_t* heap, size_t needed_slices, size_t block_size, mi_segments_tld_t* tld, mi_os_tld_t* os_tld) +{ + mi_assert_internal(block_size < MI_HUGE_BLOCK_SIZE); + mi_assert_internal(block_size <= MI_LARGE_OBJ_SIZE_MAX); + + // 1. try to reclaim an abandoned segment + bool reclaimed; + mi_segment_t* segment = mi_segment_try_reclaim(heap, needed_slices, block_size, &reclaimed, tld); + if (reclaimed) { + // reclaimed the right page right into the heap + mi_assert_internal(segment != NULL); + return NULL; // pretend out-of-memory as the page will be in the page queue of the heap with available blocks + } + else if (segment != NULL) { + // reclaimed a segment with a large enough empty span in it + return segment; + } + // 2. otherwise allocate a fresh segment + return mi_segment_alloc(0, 0, heap->arena_id, tld, os_tld, NULL); +} + + +/* ----------------------------------------------------------- + Page allocation +----------------------------------------------------------- */ + +static mi_page_t* mi_segments_page_alloc(mi_heap_t* heap, mi_page_kind_t page_kind, size_t required, size_t block_size, mi_segments_tld_t* tld, mi_os_tld_t* os_tld) +{ + mi_assert_internal(required <= MI_LARGE_OBJ_SIZE_MAX && page_kind <= MI_PAGE_LARGE); + + // find a free page + size_t page_size = _mi_align_up(required, (required > MI_MEDIUM_PAGE_SIZE ? MI_MEDIUM_PAGE_SIZE : MI_SEGMENT_SLICE_SIZE)); + size_t slices_needed = page_size / MI_SEGMENT_SLICE_SIZE; + mi_assert_internal(slices_needed * MI_SEGMENT_SLICE_SIZE == page_size); + mi_page_t* page = mi_segments_page_find_and_allocate(slices_needed, heap->arena_id, tld); //(required <= MI_SMALL_SIZE_MAX ? 0 : slices_needed), tld); + if (page==NULL) { + // no free page, allocate a new segment and try again + if (mi_segment_reclaim_or_alloc(heap, slices_needed, block_size, tld, os_tld) == NULL) { + // OOM or reclaimed a good page in the heap + return NULL; + } + else { + // otherwise try again + return mi_segments_page_alloc(heap, page_kind, required, block_size, tld, os_tld); + } + } + mi_assert_internal(page != NULL && page->slice_count*MI_SEGMENT_SLICE_SIZE == page_size); + mi_assert_internal(_mi_ptr_segment(page)->thread_id == _mi_thread_id()); + mi_segment_try_purge(_mi_ptr_segment(page), false, tld->stats); + return page; +} + + + +/* ----------------------------------------------------------- + Huge page allocation +----------------------------------------------------------- */ + +static mi_page_t* mi_segment_huge_page_alloc(size_t size, size_t page_alignment, mi_arena_id_t req_arena_id, mi_segments_tld_t* tld, mi_os_tld_t* os_tld) +{ + mi_page_t* page = NULL; + mi_segment_t* segment = mi_segment_alloc(size,page_alignment,req_arena_id,tld,os_tld,&page); + if (segment == NULL || page==NULL) return NULL; + mi_assert_internal(segment->used==1); + mi_assert_internal(mi_page_block_size(page) >= size); + #if MI_HUGE_PAGE_ABANDON + segment->thread_id = 0; // huge segments are immediately abandoned + #endif + + // for huge pages we initialize the xblock_size as we may + // overallocate to accommodate large alignments. + size_t psize; + uint8_t* start = _mi_segment_page_start(segment, page, &psize); + page->xblock_size = (psize > MI_HUGE_BLOCK_SIZE ? MI_HUGE_BLOCK_SIZE : (uint32_t)psize); + + // decommit the part of the prefix of a page that will not be used; this can be quite large (close to MI_SEGMENT_SIZE) + if (page_alignment > 0 && segment->allow_decommit) { + uint8_t* aligned_p = (uint8_t*)_mi_align_up((uintptr_t)start, page_alignment); + mi_assert_internal(_mi_is_aligned(aligned_p, page_alignment)); + mi_assert_internal(psize - (aligned_p - start) >= size); + uint8_t* decommit_start = start + sizeof(mi_block_t); // for the free list + ptrdiff_t decommit_size = aligned_p - decommit_start; + _mi_os_reset(decommit_start, decommit_size, &_mi_stats_main); // note: cannot use segment_decommit on huge segments + } + + return page; +} + +#if MI_HUGE_PAGE_ABANDON +// free huge block from another thread +void _mi_segment_huge_page_free(mi_segment_t* segment, mi_page_t* page, mi_block_t* block) { + // huge page segments are always abandoned and can be freed immediately by any thread + mi_assert_internal(segment->kind==MI_SEGMENT_HUGE); + mi_assert_internal(segment == _mi_page_segment(page)); + mi_assert_internal(mi_atomic_load_relaxed(&segment->thread_id)==0); + + // claim it and free + mi_heap_t* heap = mi_heap_get_default(); // issue #221; don't use the internal get_default_heap as we need to ensure the thread is initialized. + // paranoia: if this it the last reference, the cas should always succeed + size_t expected_tid = 0; + if (mi_atomic_cas_strong_acq_rel(&segment->thread_id, &expected_tid, heap->thread_id)) { + mi_block_set_next(page, block, page->free); + page->free = block; + page->used--; + page->is_zero = false; + mi_assert(page->used == 0); + mi_tld_t* tld = heap->tld; + _mi_segment_page_free(page, true, &tld->segments); + } +#if (MI_DEBUG!=0) + else { + mi_assert_internal(false); + } +#endif +} + +#else +// reset memory of a huge block from another thread +void _mi_segment_huge_page_reset(mi_segment_t* segment, mi_page_t* page, mi_block_t* block) { + MI_UNUSED(page); + mi_assert_internal(segment->kind == MI_SEGMENT_HUGE); + mi_assert_internal(segment == _mi_page_segment(page)); + mi_assert_internal(page->used == 1); // this is called just before the free + mi_assert_internal(page->free == NULL); + if (segment->allow_decommit) { + size_t csize = mi_usable_size(block); + if (csize > sizeof(mi_block_t)) { + csize = csize - sizeof(mi_block_t); + uint8_t* p = (uint8_t*)block + sizeof(mi_block_t); + _mi_os_reset(p, csize, &_mi_stats_main); // note: cannot use segment_decommit on huge segments + } + } +} +#endif + +/* ----------------------------------------------------------- + Page allocation and free +----------------------------------------------------------- */ +mi_page_t* _mi_segment_page_alloc(mi_heap_t* heap, size_t block_size, size_t page_alignment, mi_segments_tld_t* tld, mi_os_tld_t* os_tld) { + mi_page_t* page; + if mi_unlikely(page_alignment > MI_ALIGNMENT_MAX) { + mi_assert_internal(_mi_is_power_of_two(page_alignment)); + mi_assert_internal(page_alignment >= MI_SEGMENT_SIZE); + if (page_alignment < MI_SEGMENT_SIZE) { page_alignment = MI_SEGMENT_SIZE; } + page = mi_segment_huge_page_alloc(block_size,page_alignment,heap->arena_id,tld,os_tld); + } + else if (block_size <= MI_SMALL_OBJ_SIZE_MAX) { + page = mi_segments_page_alloc(heap,MI_PAGE_SMALL,block_size,block_size,tld,os_tld); + } + else if (block_size <= MI_MEDIUM_OBJ_SIZE_MAX) { + page = mi_segments_page_alloc(heap,MI_PAGE_MEDIUM,MI_MEDIUM_PAGE_SIZE,block_size,tld, os_tld); + } + else if (block_size <= MI_LARGE_OBJ_SIZE_MAX) { + page = mi_segments_page_alloc(heap,MI_PAGE_LARGE,block_size,block_size,tld, os_tld); + } + else { + page = mi_segment_huge_page_alloc(block_size,page_alignment,heap->arena_id,tld,os_tld); + } + mi_assert_internal(page == NULL || _mi_heap_memid_is_suitable(heap, _mi_page_segment(page)->memid)); + mi_assert_expensive(page == NULL || mi_segment_is_valid(_mi_page_segment(page),tld)); + return page; +} diff --git a/compat/mimalloc/stats.c b/compat/mimalloc/stats.c new file mode 100644 index 00000000000000..6817e07aa1ee9f --- /dev/null +++ b/compat/mimalloc/stats.c @@ -0,0 +1,467 @@ +/* ---------------------------------------------------------------------------- +Copyright (c) 2018-2021, Microsoft Research, Daan Leijen +This is free software; you can redistribute it and/or modify it under the +terms of the MIT license. A copy of the license can be found in the file +"LICENSE" at the root of this distribution. +-----------------------------------------------------------------------------*/ +#include "mimalloc.h" +#include "mimalloc/internal.h" +#include "mimalloc/atomic.h" +#include "mimalloc/prim.h" + +#include // snprintf +#include // memset + +#if defined(_MSC_VER) && (_MSC_VER < 1920) +#pragma warning(disable:4204) // non-constant aggregate initializer +#endif + +/* ----------------------------------------------------------- + Statistics operations +----------------------------------------------------------- */ + +static bool mi_is_in_main(void* stat) { + return ((uint8_t*)stat >= (uint8_t*)&_mi_stats_main + && (uint8_t*)stat < ((uint8_t*)&_mi_stats_main + sizeof(mi_stats_t))); +} + +static void mi_stat_update(mi_stat_count_t* stat, int64_t amount) { + if (amount == 0) return; + if (mi_is_in_main(stat)) + { + // add atomically (for abandoned pages) + int64_t current = mi_atomic_addi64_relaxed(&stat->current, amount); + mi_atomic_maxi64_relaxed(&stat->peak, current + amount); + if (amount > 0) { + mi_atomic_addi64_relaxed(&stat->allocated,amount); + } + else { + mi_atomic_addi64_relaxed(&stat->freed, -amount); + } + } + else { + // add thread local + stat->current += amount; + if (stat->current > stat->peak) stat->peak = stat->current; + if (amount > 0) { + stat->allocated += amount; + } + else { + stat->freed += -amount; + } + } +} + +void _mi_stat_counter_increase(mi_stat_counter_t* stat, size_t amount) { + if (mi_is_in_main(stat)) { + mi_atomic_addi64_relaxed( &stat->count, 1 ); + mi_atomic_addi64_relaxed( &stat->total, (int64_t)amount ); + } + else { + stat->count++; + stat->total += amount; + } +} + +void _mi_stat_increase(mi_stat_count_t* stat, size_t amount) { + mi_stat_update(stat, (int64_t)amount); +} + +void _mi_stat_decrease(mi_stat_count_t* stat, size_t amount) { + mi_stat_update(stat, -((int64_t)amount)); +} + +// must be thread safe as it is called from stats_merge +static void mi_stat_add(mi_stat_count_t* stat, const mi_stat_count_t* src, int64_t unit) { + if (stat==src) return; + if (src->allocated==0 && src->freed==0) return; + mi_atomic_addi64_relaxed( &stat->allocated, src->allocated * unit); + mi_atomic_addi64_relaxed( &stat->current, src->current * unit); + mi_atomic_addi64_relaxed( &stat->freed, src->freed * unit); + // peak scores do not work across threads.. + mi_atomic_addi64_relaxed( &stat->peak, src->peak * unit); +} + +static void mi_stat_counter_add(mi_stat_counter_t* stat, const mi_stat_counter_t* src, int64_t unit) { + if (stat==src) return; + mi_atomic_addi64_relaxed( &stat->total, src->total * unit); + mi_atomic_addi64_relaxed( &stat->count, src->count * unit); +} + +// must be thread safe as it is called from stats_merge +static void mi_stats_add(mi_stats_t* stats, const mi_stats_t* src) { + if (stats==src) return; + mi_stat_add(&stats->segments, &src->segments,1); + mi_stat_add(&stats->pages, &src->pages,1); + mi_stat_add(&stats->reserved, &src->reserved, 1); + mi_stat_add(&stats->committed, &src->committed, 1); + mi_stat_add(&stats->reset, &src->reset, 1); + mi_stat_add(&stats->purged, &src->purged, 1); + mi_stat_add(&stats->page_committed, &src->page_committed, 1); + + mi_stat_add(&stats->pages_abandoned, &src->pages_abandoned, 1); + mi_stat_add(&stats->segments_abandoned, &src->segments_abandoned, 1); + mi_stat_add(&stats->threads, &src->threads, 1); + + mi_stat_add(&stats->malloc, &src->malloc, 1); + mi_stat_add(&stats->segments_cache, &src->segments_cache, 1); + mi_stat_add(&stats->normal, &src->normal, 1); + mi_stat_add(&stats->huge, &src->huge, 1); + mi_stat_add(&stats->large, &src->large, 1); + + mi_stat_counter_add(&stats->pages_extended, &src->pages_extended, 1); + mi_stat_counter_add(&stats->mmap_calls, &src->mmap_calls, 1); + mi_stat_counter_add(&stats->commit_calls, &src->commit_calls, 1); + mi_stat_counter_add(&stats->reset_calls, &src->reset_calls, 1); + mi_stat_counter_add(&stats->purge_calls, &src->purge_calls, 1); + + mi_stat_counter_add(&stats->page_no_retire, &src->page_no_retire, 1); + mi_stat_counter_add(&stats->searches, &src->searches, 1); + mi_stat_counter_add(&stats->normal_count, &src->normal_count, 1); + mi_stat_counter_add(&stats->huge_count, &src->huge_count, 1); + mi_stat_counter_add(&stats->large_count, &src->large_count, 1); +#if MI_STAT>1 + for (size_t i = 0; i <= MI_BIN_HUGE; i++) { + if (src->normal_bins[i].allocated > 0 || src->normal_bins[i].freed > 0) { + mi_stat_add(&stats->normal_bins[i], &src->normal_bins[i], 1); + } + } +#endif +} + +/* ----------------------------------------------------------- + Display statistics +----------------------------------------------------------- */ + +// unit > 0 : size in binary bytes +// unit == 0: count as decimal +// unit < 0 : count in binary +static void mi_printf_amount(int64_t n, int64_t unit, mi_output_fun* out, void* arg, const char* fmt) { + char buf[32]; buf[0] = 0; + int len = 32; + const char* suffix = (unit <= 0 ? " " : "B"); + const int64_t base = (unit == 0 ? 1000 : 1024); + if (unit>0) n *= unit; + + const int64_t pos = (n < 0 ? -n : n); + if (pos < base) { + if (n!=1 || suffix[0] != 'B') { // skip printing 1 B for the unit column + snprintf(buf, len, "%d %-3s", (int)n, (n==0 ? "" : suffix)); + } + } + else { + int64_t divider = base; + const char* magnitude = "K"; + if (pos >= divider*base) { divider *= base; magnitude = "M"; } + if (pos >= divider*base) { divider *= base; magnitude = "G"; } + const int64_t tens = (n / (divider/10)); + const long whole = (long)(tens/10); + const long frac1 = (long)(tens%10); + char unitdesc[8]; + snprintf(unitdesc, 8, "%s%s%s", magnitude, (base==1024 ? "i" : ""), suffix); + snprintf(buf, len, "%ld.%ld %-3s", whole, (frac1 < 0 ? -frac1 : frac1), unitdesc); + } + _mi_fprintf(out, arg, (fmt==NULL ? "%12s" : fmt), buf); +} + + +static void mi_print_amount(int64_t n, int64_t unit, mi_output_fun* out, void* arg) { + mi_printf_amount(n,unit,out,arg,NULL); +} + +static void mi_print_count(int64_t n, int64_t unit, mi_output_fun* out, void* arg) { + if (unit==1) _mi_fprintf(out, arg, "%12s"," "); + else mi_print_amount(n,0,out,arg); +} + +static void mi_stat_print_ex(const mi_stat_count_t* stat, const char* msg, int64_t unit, mi_output_fun* out, void* arg, const char* notok ) { + _mi_fprintf(out, arg,"%10s:", msg); + if (unit > 0) { + mi_print_amount(stat->peak, unit, out, arg); + mi_print_amount(stat->allocated, unit, out, arg); + mi_print_amount(stat->freed, unit, out, arg); + mi_print_amount(stat->current, unit, out, arg); + mi_print_amount(unit, 1, out, arg); + mi_print_count(stat->allocated, unit, out, arg); + if (stat->allocated > stat->freed) { + _mi_fprintf(out, arg, " "); + _mi_fprintf(out, arg, (notok == NULL ? "not all freed" : notok)); + _mi_fprintf(out, arg, "\n"); + } + else { + _mi_fprintf(out, arg, " ok\n"); + } + } + else if (unit<0) { + mi_print_amount(stat->peak, -1, out, arg); + mi_print_amount(stat->allocated, -1, out, arg); + mi_print_amount(stat->freed, -1, out, arg); + mi_print_amount(stat->current, -1, out, arg); + if (unit==-1) { + _mi_fprintf(out, arg, "%24s", ""); + } + else { + mi_print_amount(-unit, 1, out, arg); + mi_print_count((stat->allocated / -unit), 0, out, arg); + } + if (stat->allocated > stat->freed) + _mi_fprintf(out, arg, " not all freed!\n"); + else + _mi_fprintf(out, arg, " ok\n"); + } + else { + mi_print_amount(stat->peak, 1, out, arg); + mi_print_amount(stat->allocated, 1, out, arg); + _mi_fprintf(out, arg, "%11s", " "); // no freed + mi_print_amount(stat->current, 1, out, arg); + _mi_fprintf(out, arg, "\n"); + } +} + +static void mi_stat_print(const mi_stat_count_t* stat, const char* msg, int64_t unit, mi_output_fun* out, void* arg) { + mi_stat_print_ex(stat, msg, unit, out, arg, NULL); +} + +static void mi_stat_peak_print(const mi_stat_count_t* stat, const char* msg, int64_t unit, mi_output_fun* out, void* arg) { + _mi_fprintf(out, arg, "%10s:", msg); + mi_print_amount(stat->peak, unit, out, arg); + _mi_fprintf(out, arg, "\n"); +} + +static void mi_stat_counter_print(const mi_stat_counter_t* stat, const char* msg, mi_output_fun* out, void* arg ) { + _mi_fprintf(out, arg, "%10s:", msg); + mi_print_amount(stat->total, -1, out, arg); + _mi_fprintf(out, arg, "\n"); +} + + +static void mi_stat_counter_print_avg(const mi_stat_counter_t* stat, const char* msg, mi_output_fun* out, void* arg) { + const int64_t avg_tens = (stat->count == 0 ? 0 : (stat->total*10 / stat->count)); + const long avg_whole = (long)(avg_tens/10); + const long avg_frac1 = (long)(avg_tens%10); + _mi_fprintf(out, arg, "%10s: %5ld.%ld avg\n", msg, avg_whole, avg_frac1); +} + + +static void mi_print_header(mi_output_fun* out, void* arg ) { + _mi_fprintf(out, arg, "%10s: %11s %11s %11s %11s %11s %11s\n", "heap stats", "peak ", "total ", "freed ", "current ", "unit ", "count "); +} + +#if MI_STAT>1 +static void mi_stats_print_bins(const mi_stat_count_t* bins, size_t max, const char* fmt, mi_output_fun* out, void* arg) { + bool found = false; + char buf[64]; + for (size_t i = 0; i <= max; i++) { + if (bins[i].allocated > 0) { + found = true; + int64_t unit = _mi_bin_size((uint8_t)i); + snprintf(buf, 64, "%s %3lu", fmt, (long)i); + mi_stat_print(&bins[i], buf, unit, out, arg); + } + } + if (found) { + _mi_fprintf(out, arg, "\n"); + mi_print_header(out, arg); + } +} +#endif + + + +//------------------------------------------------------------ +// Use an output wrapper for line-buffered output +// (which is nice when using loggers etc.) +//------------------------------------------------------------ +typedef struct buffered_s { + mi_output_fun* out; // original output function + void* arg; // and state + char* buf; // local buffer of at least size `count+1` + size_t used; // currently used chars `used <= count` + size_t count; // total chars available for output +} buffered_t; + +static void mi_buffered_flush(buffered_t* buf) { + buf->buf[buf->used] = 0; + _mi_fputs(buf->out, buf->arg, NULL, buf->buf); + buf->used = 0; +} + +static void mi_cdecl mi_buffered_out(const char* msg, void* arg) { + buffered_t* buf = (buffered_t*)arg; + if (msg==NULL || buf==NULL) return; + for (const char* src = msg; *src != 0; src++) { + char c = *src; + if (buf->used >= buf->count) mi_buffered_flush(buf); + mi_assert_internal(buf->used < buf->count); + buf->buf[buf->used++] = c; + if (c == '\n') mi_buffered_flush(buf); + } +} + +//------------------------------------------------------------ +// Print statistics +//------------------------------------------------------------ + +static void _mi_stats_print(mi_stats_t* stats, mi_output_fun* out0, void* arg0) mi_attr_noexcept { + // wrap the output function to be line buffered + char buf[256]; + buffered_t buffer = { out0, arg0, NULL, 0, 255 }; + buffer.buf = buf; + mi_output_fun* out = &mi_buffered_out; + void* arg = &buffer; + + // and print using that + mi_print_header(out,arg); + #if MI_STAT>1 + mi_stats_print_bins(stats->normal_bins, MI_BIN_HUGE, "normal",out,arg); + #endif + #if MI_STAT + mi_stat_print(&stats->normal, "normal", (stats->normal_count.count == 0 ? 1 : -(stats->normal.allocated / stats->normal_count.count)), out, arg); + mi_stat_print(&stats->large, "large", (stats->large_count.count == 0 ? 1 : -(stats->large.allocated / stats->large_count.count)), out, arg); + mi_stat_print(&stats->huge, "huge", (stats->huge_count.count == 0 ? 1 : -(stats->huge.allocated / stats->huge_count.count)), out, arg); + mi_stat_count_t total = { 0,0,0,0 }; + mi_stat_add(&total, &stats->normal, 1); + mi_stat_add(&total, &stats->large, 1); + mi_stat_add(&total, &stats->huge, 1); + mi_stat_print(&total, "total", 1, out, arg); + #endif + #if MI_STAT>1 + mi_stat_print(&stats->malloc, "malloc req", 1, out, arg); + _mi_fprintf(out, arg, "\n"); + #endif + mi_stat_print_ex(&stats->reserved, "reserved", 1, out, arg, ""); + mi_stat_print_ex(&stats->committed, "committed", 1, out, arg, ""); + mi_stat_peak_print(&stats->reset, "reset", 1, out, arg ); + mi_stat_peak_print(&stats->purged, "purged", 1, out, arg ); + mi_stat_print(&stats->page_committed, "touched", 1, out, arg); + mi_stat_print(&stats->segments, "segments", -1, out, arg); + mi_stat_print(&stats->segments_abandoned, "-abandoned", -1, out, arg); + mi_stat_print(&stats->segments_cache, "-cached", -1, out, arg); + mi_stat_print(&stats->pages, "pages", -1, out, arg); + mi_stat_print(&stats->pages_abandoned, "-abandoned", -1, out, arg); + mi_stat_counter_print(&stats->pages_extended, "-extended", out, arg); + mi_stat_counter_print(&stats->page_no_retire, "-noretire", out, arg); + mi_stat_counter_print(&stats->mmap_calls, "mmaps", out, arg); + mi_stat_counter_print(&stats->commit_calls, "commits", out, arg); + mi_stat_counter_print(&stats->reset_calls, "resets", out, arg); + mi_stat_counter_print(&stats->purge_calls, "purges", out, arg); + mi_stat_print(&stats->threads, "threads", -1, out, arg); + mi_stat_counter_print_avg(&stats->searches, "searches", out, arg); + _mi_fprintf(out, arg, "%10s: %5zu\n", "numa nodes", _mi_os_numa_node_count()); + + size_t elapsed; + size_t user_time; + size_t sys_time; + size_t current_rss; + size_t peak_rss; + size_t current_commit; + size_t peak_commit; + size_t page_faults; + mi_process_info(&elapsed, &user_time, &sys_time, ¤t_rss, &peak_rss, ¤t_commit, &peak_commit, &page_faults); + _mi_fprintf(out, arg, "%10s: %5ld.%03ld s\n", "elapsed", elapsed/1000, elapsed%1000); + _mi_fprintf(out, arg, "%10s: user: %ld.%03ld s, system: %ld.%03ld s, faults: %lu, rss: ", "process", + user_time/1000, user_time%1000, sys_time/1000, sys_time%1000, (unsigned long)page_faults ); + mi_printf_amount((int64_t)peak_rss, 1, out, arg, "%s"); + if (peak_commit > 0) { + _mi_fprintf(out, arg, ", commit: "); + mi_printf_amount((int64_t)peak_commit, 1, out, arg, "%s"); + } + _mi_fprintf(out, arg, "\n"); +} + +static mi_msecs_t mi_process_start; // = 0 + +static mi_stats_t* mi_stats_get_default(void) { + mi_heap_t* heap = mi_heap_get_default(); + return &heap->tld->stats; +} + +static void mi_stats_merge_from(mi_stats_t* stats) { + if (stats != &_mi_stats_main) { + mi_stats_add(&_mi_stats_main, stats); + memset(stats, 0, sizeof(mi_stats_t)); + } +} + +void mi_stats_reset(void) mi_attr_noexcept { + mi_stats_t* stats = mi_stats_get_default(); + if (stats != &_mi_stats_main) { memset(stats, 0, sizeof(mi_stats_t)); } + memset(&_mi_stats_main, 0, sizeof(mi_stats_t)); + if (mi_process_start == 0) { mi_process_start = _mi_clock_start(); }; +} + +void mi_stats_merge(void) mi_attr_noexcept { + mi_stats_merge_from( mi_stats_get_default() ); +} + +void _mi_stats_done(mi_stats_t* stats) { // called from `mi_thread_done` + mi_stats_merge_from(stats); +} + +void mi_stats_print_out(mi_output_fun* out, void* arg) mi_attr_noexcept { + mi_stats_merge_from(mi_stats_get_default()); + _mi_stats_print(&_mi_stats_main, out, arg); +} + +void mi_stats_print(void* out) mi_attr_noexcept { + // for compatibility there is an `out` parameter (which can be `stdout` or `stderr`) + mi_stats_print_out((mi_output_fun*)out, NULL); +} + +void mi_thread_stats_print_out(mi_output_fun* out, void* arg) mi_attr_noexcept { + _mi_stats_print(mi_stats_get_default(), out, arg); +} + + +// ---------------------------------------------------------------- +// Basic timer for convenience; use milli-seconds to avoid doubles +// ---------------------------------------------------------------- + +static mi_msecs_t mi_clock_diff; + +mi_msecs_t _mi_clock_now(void) { + return _mi_prim_clock_now(); +} + +mi_msecs_t _mi_clock_start(void) { + if (mi_clock_diff == 0.0) { + mi_msecs_t t0 = _mi_clock_now(); + mi_clock_diff = _mi_clock_now() - t0; + } + return _mi_clock_now(); +} + +mi_msecs_t _mi_clock_end(mi_msecs_t start) { + mi_msecs_t end = _mi_clock_now(); + return (end - start - mi_clock_diff); +} + + +// -------------------------------------------------------- +// Basic process statistics +// -------------------------------------------------------- + +mi_decl_export void mi_process_info(size_t* elapsed_msecs, size_t* user_msecs, size_t* system_msecs, size_t* current_rss, size_t* peak_rss, size_t* current_commit, size_t* peak_commit, size_t* page_faults) mi_attr_noexcept +{ + mi_process_info_t pinfo; + _mi_memzero_var(pinfo); + pinfo.elapsed = _mi_clock_end(mi_process_start); + pinfo.current_commit = (size_t)(mi_atomic_loadi64_relaxed((_Atomic(int64_t)*)&_mi_stats_main.committed.current)); + pinfo.peak_commit = (size_t)(mi_atomic_loadi64_relaxed((_Atomic(int64_t)*)&_mi_stats_main.committed.peak)); + pinfo.current_rss = pinfo.current_commit; + pinfo.peak_rss = pinfo.peak_commit; + pinfo.utime = 0; + pinfo.stime = 0; + pinfo.page_faults = 0; + + _mi_prim_process_info(&pinfo); + + if (elapsed_msecs!=NULL) *elapsed_msecs = (pinfo.elapsed < 0 ? 0 : (pinfo.elapsed < (mi_msecs_t)PTRDIFF_MAX ? (size_t)pinfo.elapsed : PTRDIFF_MAX)); + if (user_msecs!=NULL) *user_msecs = (pinfo.utime < 0 ? 0 : (pinfo.utime < (mi_msecs_t)PTRDIFF_MAX ? (size_t)pinfo.utime : PTRDIFF_MAX)); + if (system_msecs!=NULL) *system_msecs = (pinfo.stime < 0 ? 0 : (pinfo.stime < (mi_msecs_t)PTRDIFF_MAX ? (size_t)pinfo.stime : PTRDIFF_MAX)); + if (current_rss!=NULL) *current_rss = pinfo.current_rss; + if (peak_rss!=NULL) *peak_rss = pinfo.peak_rss; + if (current_commit!=NULL) *current_commit = pinfo.current_commit; + if (peak_commit!=NULL) *peak_commit = pinfo.peak_commit; + if (page_faults!=NULL) *page_faults = pinfo.page_faults; +} From fb6af4f6944004845b3a73f68189c708ff28f530 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 18 Apr 2017 12:09:08 +0200 Subject: [PATCH 036/259] mingw: demonstrate a problem with certain absolute paths On Windows, there are several categories of absolute paths. One such category starts with a backslash and is implicitly relative to the drive associated with the current working directory. Example: c: git clone https://github.com/git-for-windows/git \G4W should clone into C:\G4W. There is currently a problem with that, in that mingw_mktemp() does not expect the _wmktemp() function to prefix the absolute path with the drive prefix, and as a consequence, the resulting path does not fit into the originally-passed string buffer. The symptom is a "Result too large" error. Reported by Juan Carlos Arevalo Baeza. Signed-off-by: Johannes Schindelin --- t/t5580-unc-paths.sh | 19 ++++++++++++++----- 1 file changed, 14 insertions(+), 5 deletions(-) diff --git a/t/t5580-unc-paths.sh b/t/t5580-unc-paths.sh index 65ef1a3628ee94..2c30f3b51d3183 100755 --- a/t/t5580-unc-paths.sh +++ b/t/t5580-unc-paths.sh @@ -20,14 +20,11 @@ fi UNCPATH="$(winpwd)" case "$UNCPATH" in [A-Z]:*) + WITHOUTDRIVE="${UNCPATH#?:}" # Use administrative share e.g. \\localhost\C$\git-sdk-64\usr\src\git # (we use forward slashes here because MSYS2 and Git accept them, and # they are easier on the eyes) - UNCPATH="//localhost/${UNCPATH%%:*}\$/${UNCPATH#?:}" - test -d "$UNCPATH" || { - skip_all='could not access administrative share; skipping' - test_done - } + UNCPATH="//localhost/${UNCPATH%%:*}\$$WITHOUTDRIVE" ;; *) skip_all='skipping UNC path tests, cannot determine current path as UNC' @@ -35,6 +32,18 @@ case "$UNCPATH" in ;; esac +test_expect_failure 'clone into absolute path lacking a drive prefix' ' + USINGBACKSLASHES="$(echo "$WITHOUTDRIVE"/without-drive-prefix | + tr / \\\\)" && + git clone . "$USINGBACKSLASHES" && + test -f without-drive-prefix/.git/HEAD +' + +test -d "$UNCPATH" || { + skip_all='could not access administrative share; skipping' + test_done +} + test_expect_success setup ' test_commit initial ' From 03560065f0c2089646db3a6f0fff5923cc34e7c3 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 7 Dec 2018 13:39:30 +0100 Subject: [PATCH 037/259] clean: do not traverse mount points It seems to be not exactly rare on Windows to install NTFS junction points (the equivalent of "bind mounts" on Linux/Unix) in worktrees, e.g. to map some development tools into a subdirectory. In such a scenario, it is pretty horrible if `git clean -dfx` traverses into the mapped directory and starts to "clean up". Let's just not do that. Let's make sure before we traverse into a directory that it is not a mount point (or junction). This addresses https://github.com/git-for-windows/git/issues/607 Signed-off-by: Johannes Schindelin --- builtin/clean.c | 14 ++++++++++++++ compat/mingw.c | 22 ++++++++++++++++++++++ compat/mingw.h | 3 +++ git-compat-util.h | 4 ++++ path.c | 39 +++++++++++++++++++++++++++++++++++++++ path.h | 1 + t/t7300-clean.sh | 9 +++++++++ 7 files changed, 92 insertions(+) diff --git a/builtin/clean.c b/builtin/clean.c index 053c94fc6bd12a..aeb7eb2d1ed2c3 100644 --- a/builtin/clean.c +++ b/builtin/clean.c @@ -40,6 +40,8 @@ static const char *msg_remove = N_("Removing %s\n"); static const char *msg_would_remove = N_("Would remove %s\n"); static const char *msg_skip_git_dir = N_("Skipping repository %s\n"); static const char *msg_would_skip_git_dir = N_("Would skip repository %s\n"); +static const char *msg_skip_mount_point = N_("Skipping mount point %s\n"); +static const char *msg_would_skip_mount_point = N_("Would skip mount point %s\n"); static const char *msg_warn_remove_failed = N_("failed to remove %s"); static const char *msg_warn_lstat_failed = N_("could not lstat %s\n"); static const char *msg_skip_cwd = N_("Refusing to remove current working directory\n"); @@ -184,6 +186,18 @@ static int remove_dirs(struct strbuf *path, const char *prefix, int force_flag, goto out; } + if (is_mount_point(path)) { + if (!quiet) { + quote_path(path->buf, prefix, "ed, 0); + printf(dry_run ? + _(msg_would_skip_mount_point) : + _(msg_skip_mount_point), quoted.buf); + } + *dir_gone = 0; + + goto out; + } + dir = opendir(path->buf); if (!dir) { /* an empty dir could be removed even if it is unreadble */ diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..5824a9162a6d19 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -2688,6 +2688,28 @@ pid_t waitpid(pid_t pid, int *status, int options) return -1; } +int mingw_is_mount_point(struct strbuf *path) +{ + WIN32_FIND_DATAW findbuf = { 0 }; + HANDLE handle; + wchar_t wfilename[MAX_PATH]; + int wlen = xutftowcs_path(wfilename, path->buf); + if (wlen < 0) + die(_("could not get long path for '%s'"), path->buf); + + /* remove trailing slash, if any */ + if (wlen > 0 && wfilename[wlen - 1] == L'/') + wfilename[--wlen] = L'\0'; + + handle = FindFirstFileW(wfilename, &findbuf); + if (handle == INVALID_HANDLE_VALUE) + return 0; + FindClose(handle); + + return (findbuf.dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) && + (findbuf.dwReserved0 == IO_REPARSE_TAG_MOUNT_POINT); +} + int xutftowcsn(wchar_t *wcs, const char *utfs, size_t wcslen, int utflen) { int upos = 0, wpos = 0; diff --git a/compat/mingw.h b/compat/mingw.h index ebfb8ba423b754..92830de32a40e3 100644 --- a/compat/mingw.h +++ b/compat/mingw.h @@ -454,6 +454,9 @@ static inline void convert_slashes(char *path) if (*path == '\\') *path = '/'; } +struct strbuf; +int mingw_is_mount_point(struct strbuf *path); +#define is_mount_point mingw_is_mount_point #define PATH_SEP ';' char *mingw_query_user_email(void); #define query_user_email mingw_query_user_email diff --git a/git-compat-util.h b/git-compat-util.h index e123288e8f1393..df7f4437536dcf 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -626,6 +626,10 @@ static inline int git_has_dir_sep(const char *path) #define has_dir_sep(path) git_has_dir_sep(path) #endif +#ifndef is_mount_point +#define is_mount_point is_mount_point_via_stat +#endif + #ifndef query_user_email #define query_user_email() NULL #endif diff --git a/path.c b/path.c index 910756c8b3249c..f60c07b76b6c29 100644 --- a/path.c +++ b/path.c @@ -1300,6 +1300,45 @@ char *strip_path_suffix(const char *path, const char *suffix) return offset == -1 ? NULL : xstrndup(path, offset); } +int is_mount_point_via_stat(struct strbuf *path) +{ + size_t len = path->len; + dev_t current_dev; + struct stat st; + + if (!strcmp("/", path->buf)) + return 1; + + strbuf_addstr(path, "/."); + if (lstat(path->buf, &st)) { + /* + * If we cannot access the current directory, we cannot say + * that it is a bind mount. + */ + strbuf_setlen(path, len); + return 0; + } + current_dev = st.st_dev; + + /* Now look at the parent directory */ + strbuf_addch(path, '.'); + if (lstat(path->buf, &st)) { + /* + * If we cannot access the parent directory, we cannot say + * that it is a bind mount. + */ + strbuf_setlen(path, len); + return 0; + } + strbuf_setlen(path, len); + + /* + * If the device ID differs between current and parent directory, + * then it is a bind mount. + */ + return current_dev != st.st_dev; +} + int daemon_avoid_alias(const char *p) { int sl, ndot; diff --git a/path.h b/path.h index 65fe968a13a191..535b8456c646f2 100644 --- a/path.h +++ b/path.h @@ -172,6 +172,7 @@ int normalize_path_copy(char *dst, const char *src); int strbuf_normalize_path(struct strbuf *src); int longest_ancestor_length(const char *path, struct string_list *prefixes); char *strip_path_suffix(const char *path, const char *suffix); +int is_mount_point_via_stat(struct strbuf *path); int daemon_avoid_alias(const char *path); /* diff --git a/t/t7300-clean.sh b/t/t7300-clean.sh index 00d4070156243b..7c3a1ca91df534 100755 --- a/t/t7300-clean.sh +++ b/t/t7300-clean.sh @@ -800,4 +800,13 @@ test_expect_success 'traverse into directories that may have ignored entries' ' ) ' +test_expect_success MINGW 'clean does not traverse mount points' ' + mkdir target && + >target/dont-clean-me && + git init with-mountpoint && + cmd //c "mklink /j with-mountpoint\\mountpoint target" && + git -C with-mountpoint clean -dfx && + test_path_is_file target/dont-clean-me +' + test_done From d7b93492f495ccbcfe8775be78577dbfd2303ad2 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 24 Jun 2019 23:41:27 +0200 Subject: [PATCH 038/259] mimalloc: adjust for building inside Git We want to compile mimalloc's source code as part of Git, rather than requiring the code to be built as an external library: mimalloc uses a CMake-based build, which is not necessarily easy to integrate into the flavors of Git for Windows (which will be the main benefitting port). Signed-off-by: Johannes Schindelin --- compat/mimalloc/alloc.c | 4 ---- compat/mimalloc/mimalloc.h | 3 ++- 2 files changed, 2 insertions(+), 5 deletions(-) diff --git a/compat/mimalloc/alloc.c b/compat/mimalloc/alloc.c index 961f6d53d0f2c7..ae272c1fb54504 100644 --- a/compat/mimalloc/alloc.c +++ b/compat/mimalloc/alloc.c @@ -16,10 +16,6 @@ terms of the MIT license. A copy of the license can be found in the file #include // memset, strlen (for mi_strdup) #include // malloc, abort -#define MI_IN_ALLOC_C -#include "alloc-override.c" -#undef MI_IN_ALLOC_C - // ------------------------------------------------------ // Allocation // ------------------------------------------------------ diff --git a/compat/mimalloc/mimalloc.h b/compat/mimalloc/mimalloc.h index c0f5e96e51e975..7e3b5dd66e91a0 100644 --- a/compat/mimalloc/mimalloc.h +++ b/compat/mimalloc/mimalloc.h @@ -95,7 +95,8 @@ terms of the MIT license. A copy of the license can be found in the file // Includes // ------------------------------------------------------ -#include // size_t +#include "git-compat-util.h" + #include // bool #include // INTPTR_MAX From 595129783ccfd52a3a52611cb4b91529a766befa Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 18 Apr 2017 12:38:30 +0200 Subject: [PATCH 039/259] mingw: allow absolute paths without drive prefix When specifying an absolute path without a drive prefix, we convert that path internally. Let's make sure that we handle that case properly, too ;-) This fixes the command git clone https://github.com/git-for-windows/git \G4W Signed-off-by: Johannes Schindelin --- compat/mingw.c | 10 +++++++++- t/t5580-unc-paths.sh | 2 +- 2 files changed, 10 insertions(+), 2 deletions(-) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..12adc941381aa6 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -1141,11 +1141,19 @@ unsigned int sleep (unsigned int seconds) char *mingw_mktemp(char *template) { wchar_t wtemplate[MAX_PATH]; + int offset = 0; + if (xutftowcs_path(wtemplate, template) < 0) return NULL; + + if (is_dir_sep(template[0]) && !is_dir_sep(template[1]) && + iswalpha(wtemplate[0]) && wtemplate[1] == L':') { + /* We have an absolute path missing the drive prefix */ + offset = 2; + } if (!_wmktemp(wtemplate)) return NULL; - if (xwcstoutf(template, wtemplate, strlen(template) + 1) < 0) + if (xwcstoutf(template, wtemplate + offset, strlen(template) + 1) < 0) return NULL; return template; } diff --git a/t/t5580-unc-paths.sh b/t/t5580-unc-paths.sh index 2c30f3b51d3183..e9df367d5777fd 100755 --- a/t/t5580-unc-paths.sh +++ b/t/t5580-unc-paths.sh @@ -32,7 +32,7 @@ case "$UNCPATH" in ;; esac -test_expect_failure 'clone into absolute path lacking a drive prefix' ' +test_expect_success 'clone into absolute path lacking a drive prefix' ' USINGBACKSLASHES="$(echo "$WITHOUTDRIVE"/without-drive-prefix | tr / \\\\)" && git clone . "$USINGBACKSLASHES" && From 0baebe3ce9909253673bd79ef82b4e0e81f161a3 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 11 Dec 2018 12:55:26 +0100 Subject: [PATCH 040/259] clean: remove mount points when possible Windows' equivalent to "bind mounts", NTFS junction points, can be unlinked without affecting the mount target. This is clearly what users expect to happen when they call `git clean -dfx` in a worktree that contains NTFS junction points: the junction should be removed, and the target directory of said junction should be left alone (unless it is inside the worktree). Signed-off-by: Johannes Schindelin --- builtin/clean.c | 13 +++++++++++++ compat/mingw.h | 1 + t/t7300-clean.sh | 1 + 3 files changed, 15 insertions(+) diff --git a/builtin/clean.c b/builtin/clean.c index aeb7eb2d1ed2c3..df39efcc50fbc4 100644 --- a/builtin/clean.c +++ b/builtin/clean.c @@ -40,8 +40,10 @@ static const char *msg_remove = N_("Removing %s\n"); static const char *msg_would_remove = N_("Would remove %s\n"); static const char *msg_skip_git_dir = N_("Skipping repository %s\n"); static const char *msg_would_skip_git_dir = N_("Would skip repository %s\n"); +#ifndef CAN_UNLINK_MOUNT_POINTS static const char *msg_skip_mount_point = N_("Skipping mount point %s\n"); static const char *msg_would_skip_mount_point = N_("Would skip mount point %s\n"); +#endif static const char *msg_warn_remove_failed = N_("failed to remove %s"); static const char *msg_warn_lstat_failed = N_("could not lstat %s\n"); static const char *msg_skip_cwd = N_("Refusing to remove current working directory\n"); @@ -187,6 +189,7 @@ static int remove_dirs(struct strbuf *path, const char *prefix, int force_flag, } if (is_mount_point(path)) { +#ifndef CAN_UNLINK_MOUNT_POINTS if (!quiet) { quote_path(path->buf, prefix, "ed, 0); printf(dry_run ? @@ -194,6 +197,16 @@ static int remove_dirs(struct strbuf *path, const char *prefix, int force_flag, _(msg_skip_mount_point), quoted.buf); } *dir_gone = 0; +#else + if (!dry_run && unlink(path->buf)) { + int saved_errno = errno; + quote_path(path->buf, prefix, "ed, 0); + errno = saved_errno; + warning_errno(_(msg_warn_remove_failed), quoted.buf); + *dir_gone = 0; + ret = -1; + } +#endif goto out; } diff --git a/compat/mingw.h b/compat/mingw.h index 92830de32a40e3..92aa29bb8bb790 100644 --- a/compat/mingw.h +++ b/compat/mingw.h @@ -457,6 +457,7 @@ static inline void convert_slashes(char *path) struct strbuf; int mingw_is_mount_point(struct strbuf *path); #define is_mount_point mingw_is_mount_point +#define CAN_UNLINK_MOUNT_POINTS 1 #define PATH_SEP ';' char *mingw_query_user_email(void); #define query_user_email mingw_query_user_email diff --git a/t/t7300-clean.sh b/t/t7300-clean.sh index 7c3a1ca91df534..6f16f3893191e7 100755 --- a/t/t7300-clean.sh +++ b/t/t7300-clean.sh @@ -806,6 +806,7 @@ test_expect_success MINGW 'clean does not traverse mount points' ' git init with-mountpoint && cmd //c "mklink /j with-mountpoint\\mountpoint target" && git -C with-mountpoint clean -dfx && + test_path_is_missing with-mountpoint/mountpoint && test_path_is_file target/dont-clean-me ' From 5aee7119a8c9f954d46530c4c166b8307054cdf3 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 16 Feb 2015 14:06:59 +0100 Subject: [PATCH 041/259] mingw: include the Python parts in the build While Git for Windows does not _ship_ Python (in order to save on bandwidth), MSYS2 provides very fine Python interpreters that users can easily take advantage of, by using Git for Windows within its SDK. Signed-off-by: Johannes Schindelin --- config.mak.uname | 1 + 1 file changed, 1 insertion(+) diff --git a/config.mak.uname b/config.mak.uname index b12d4e168ae119..7fba3e595c5cd5 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -739,6 +739,7 @@ ifeq ($(uname_S),MINGW) USE_GETTEXT_SCHEME = fallthrough USE_LIBPCRE = YesPlease USE_NED_ALLOCATOR = YesPlease + NO_PYTHON = ifeq (/mingw64,$(subst 32,64,$(prefix))) # Move system config into top-level /etc/ ETC_GITCONFIG = ../etc/gitconfig From 0cd275e959826737171a9882535d5fbf30371c75 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 24 Jun 2019 23:43:06 +0200 Subject: [PATCH 042/259] mimalloc: offer a build-time option to enable it By defining `USE_MIMALLOC`, Git can now be compiled with that nicely-fast and small allocator. Note that we have to disable a couple `DEVELOPER` options to build mimalloc's source code, as it makes heavy use of declarations after statements, among other things that disagree with Git's conventions. We even have to silence some GCC warnings in non-DEVELOPER mode. For example, the `-Wno-array-bounds` flag is needed because in `-O2` builds, trying to call `NtCurrentTeb()` (which `_mi_thread_id()` does on Windows) causes the bogus warning about a system header, likely related to https://sourceforge.net/p/mingw-w64/mailman/message/37674519/ and to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99578: C:/git-sdk-64-minimal/mingw64/include/psdk_inc/intrin-impl.h:838:1: error: array subscript 0 is outside array bounds of 'long long unsigned int[0]' [-Werror=array-bounds] 838 | __buildreadseg(__readgsqword, unsigned __int64, "gs", "q") | ^~~~~~~~~~~~~~ Also: The `mimalloc` library uses C11-style atomics, therefore we must require that standard when compiling with GCC if we want to use `mimalloc` (instead of requiring "only" C99). This is what we do in the CMake definition already, therefore this commit does not need to touch `contrib/buildsystems/`. Signed-off-by: Johannes Schindelin --- Makefile | 37 +++++++++++++++++++++++++++++++++++++ config.mak.dev | 2 ++ config.mak.uname | 2 +- git-compat-util.h | 10 ++++++++++ 4 files changed, 50 insertions(+), 1 deletion(-) diff --git a/Makefile b/Makefile index fbb9227e7d7909..ec5951d5c8a057 100644 --- a/Makefile +++ b/Makefile @@ -2147,6 +2147,43 @@ ifdef USE_NED_ALLOCATOR OVERRIDE_STRDUP = YesPlease endif +ifdef USE_MIMALLOC + MIMALLOC_OBJS = \ + compat/mimalloc/alloc-aligned.o \ + compat/mimalloc/alloc.o \ + compat/mimalloc/arena.o \ + compat/mimalloc/bitmap.o \ + compat/mimalloc/heap.o \ + compat/mimalloc/init.o \ + compat/mimalloc/options.o \ + compat/mimalloc/os.o \ + compat/mimalloc/page.o \ + compat/mimalloc/random.o \ + compat/mimalloc/prim/windows/prim.o \ + compat/mimalloc/segment.o \ + compat/mimalloc/segment-cache.o \ + compat/mimalloc/segment-map.o \ + compat/mimalloc/stats.o + + COMPAT_CFLAGS += -Icompat/mimalloc -DMI_DEBUG=0 -DUSE_MIMALLOC --std=gnu11 + COMPAT_OBJS += $(MIMALLOC_OBJS) + +$(MIMALLOC_OBJS): COMPAT_CFLAGS += -DBANNED_H + +$(MIMALLOC_OBJS): COMPAT_CFLAGS += \ + -Wno-attributes \ + -Wno-unknown-pragmas \ + -Wno-array-bounds + +ifdef DEVELOPER +$(MIMALLOC_OBJS): COMPAT_CFLAGS += \ + -Wno-pedantic \ + -Wno-declaration-after-statement \ + -Wno-old-style-definition \ + -Wno-missing-prototypes +endif +endif + ifdef OVERRIDE_STRDUP COMPAT_CFLAGS += -DOVERRIDE_STRDUP COMPAT_OBJS += compat/strdup.o diff --git a/config.mak.dev b/config.mak.dev index 0fd8cc4d355ebb..a5d6dd67dfd77e 100644 --- a/config.mak.dev +++ b/config.mak.dev @@ -22,8 +22,10 @@ endif ifneq ($(uname_S),FreeBSD) ifneq ($(or $(filter gcc6,$(COMPILER_FEATURES)),$(filter clang7,$(COMPILER_FEATURES))),) +ifndef USE_MIMALLOC DEVELOPER_CFLAGS += -std=gnu99 endif +endif else # FreeBSD cannot limit to C99 because its system headers unconditionally # rely on C11 features. diff --git a/config.mak.uname b/config.mak.uname index 7fba3e595c5cd5..cdfb606f010ba4 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -494,7 +494,7 @@ endif CC = compat/vcbuild/scripts/clink.pl AR = compat/vcbuild/scripts/lib.pl CFLAGS = - BASIC_CFLAGS = -nologo -I. -Icompat/vcbuild/include -DWIN32 -D_CONSOLE -DHAVE_STRING_H -D_CRT_SECURE_NO_WARNINGS -D_CRT_NONSTDC_NO_DEPRECATE + BASIC_CFLAGS = -nologo -I. -Icompat/vcbuild/include -DWIN32 -D_CONSOLE -DHAVE_STRING_H -D_CRT_SECURE_NO_WARNINGS -D_CRT_NONSTDC_NO_DEPRECATE -MP -std:c11 COMPAT_OBJS = compat/msvc.o compat/winansi.o \ compat/win32/flush.o \ compat/win32/path-utils.o \ diff --git a/git-compat-util.h b/git-compat-util.h index 6aee7e1b27a26b..0505966dd8b02a 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -443,6 +443,16 @@ char *gitdirname(char *); # include #endif +#ifdef USE_MIMALLOC +#include "mimalloc.h" +#define malloc mi_malloc +#define calloc mi_calloc +#define realloc mi_realloc +#define free mi_free +#define strdup mi_strdup +#define strndup mi_strndup +#endif + /* On most systems would have given us this, but * not on some systems (e.g. z/OS). */ From 15788f1e936bc3cf1bbf61afb6552742c1553e01 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 31 Jan 2020 12:02:47 +0100 Subject: [PATCH 043/259] mingw: demonstrate a `git add` issue with NTFS junctions NTFS junctions are somewhat similar in spirit to Unix bind mounts: they point to a different directory and are resolved by the filesystem driver. As such, they appear to `lstat()` as if they are directories, not as if they are symbolic links. _Any_ user can create junctions, while symbolic links can only be created by non-administrators in Developer Mode on Windows 10. Hence NTFS junctions are much more common "in the wild" than NTFS symbolic links. It was reported in https://github.com/git-for-windows/git/issues/2481 that adding files via an absolute path that traverses an NTFS junction: since 1e64d18 (mingw: do resolve symlinks in `getcwd()`), we resolve not only symbolic links but also NTFS junctions when determining the absolute path of the current directory. The same is not true for `git add `, where symbolic links are resolved in ``, but not NTFS junctions. Signed-off-by: Johannes Schindelin --- t/t3700-add.sh | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/t/t3700-add.sh b/t/t3700-add.sh index df580a5806b4f1..d934f3417faa45 100755 --- a/t/t3700-add.sh +++ b/t/t3700-add.sh @@ -548,4 +548,15 @@ test_expect_success CASE_INSENSITIVE_FS 'path is case-insensitive' ' git add "$downcased" ' +test_expect_failure MINGW 'can add files via NTFS junctions' ' + test_when_finished "cmd //c rmdir junction && rm -rf target" && + test_create_repo target && + cmd //c "mklink /j junction target" && + >target/via-junction && + git -C junction add "$(pwd)/junction/via-junction" && + echo via-junction >expect && + git -C target diff --cached --name-only >actual && + test_cmp expect actual +' + test_done From efd11ee276482617f70c6c4bc3eb4b2da35269da Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 21 Feb 2017 13:28:58 +0100 Subject: [PATCH 044/259] mingw: ensure valid CTYPE A change between versions 2.4.1 and 2.6.0 of the MSYS2 runtime modified how Cygwin's runtime (and hence Git for Windows' MSYS2 runtime derivative) handles locales: d16a56306d (Consolidate wctomb/mbtowc calls for POSIX-1.2008, 2016-07-20). An unintended side-effect is that "cold-calling" into the POSIX emulation will start with a locale based on the current code page, something that Git for Windows is very ill-prepared for, as it expects to be able to pass a command-line containing non-ASCII characters to the shell without having those characters munged. One symptom of this behavior: when `git clone` or `git fetch` shell out to call `git-upload-pack` with a path that contains non-ASCII characters, the shell tried to interpret the entire command-line (including command-line parameters) as executable path, which obviously must fail. This fixes https://github.com/git-for-windows/git/issues/1036 Signed-off-by: Johannes Schindelin --- compat/mingw.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..ad00d56ed0cc58 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -2824,6 +2824,9 @@ static void setup_windows_environment(void) if (!tmp && (tmp = getenv("USERPROFILE"))) setenv("HOME", tmp, 1); } + + if (!getenv("LC_ALL") && !getenv("LC_CTYPE") && !getenv("LANG")) + setenv("LC_CTYPE", "C.UTF-8", 1); } static PSID get_current_user_sid(void) From b29ef5975bcb4e9afe7108650959fd1c7f8c2678 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Fri, 12 May 2023 15:54:11 -0400 Subject: [PATCH 045/259] mimalloc: use "weak" random seed when statically linked Always use the internal "use_weak" random seed when initializing the "mimalloc" heap when statically linked on Windows. The imported "mimalloc" routines support several random sources to seed the heap data structures, including BCrypt.dll and RtlGenRandom. Crashes have been reported when using BCrypt.dll if it initialized during an `atexit()` handler function. Granted, such DLL initialization should not happen in an atexit handler, but yet the crashes remain. It should be noted that on Windows when statically linked, the mimalloc startup code (called by the GCC CRT to initialize static data prior to calling `main()`) always uses the internal "weak" random seed. "mimalloc" does not try to load an alternate random source until after the OS initialization has completed. Heap data is stored in `__declspec(thread)` TLS data and in theory each Git thread will have its own heap data. However, testing shows that the "mimalloc" library doesn't actually call `os_random_buf()` (to load a new random source) when creating these new per-thread heap structures. However, if an atexit handler is forced to run on a non-main thread, the "mimalloc" library *WILL* try to create a new heap and seed it with `os_random_buf()`. (The reason for this is still a mystery to this author.) The `os_random_buf()` call can cause the (previously uninitialized BCrypt.dll library) to be dynamically loaded and a call made into it. Crashes have been reported in v2.40.1.vfs.0.0 while in this call. As a workaround, the fix here forces the use of the internal "use_weak" random code for the subsequent `os_random_buf()` calls. Since we have been using that random generator for the majority of the program, it seems safe to use it for the final few mallocs in the atexit handler (of which there really shouldn't be that many. Signed-off-by: Jeff Hostetler Signed-off-by: Johannes Schindelin --- compat/mimalloc/init.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/compat/mimalloc/init.c b/compat/mimalloc/init.c index 4670d5510db187..4ec5812e3ce1d0 100644 --- a/compat/mimalloc/init.c +++ b/compat/mimalloc/init.c @@ -302,7 +302,11 @@ static bool _mi_heap_init(void) { _mi_memcpy_aligned(tld, &tld_empty, sizeof(*tld)); _mi_memcpy_aligned(heap, &_mi_heap_empty, sizeof(*heap)); heap->thread_id = _mi_thread_id(); + #if defined(_WIN32) && !defined(MI_SHARED_LIB) + _mi_random_init_weak(&heap->random); // match mi_heap_main_init() + #else _mi_random_init(&heap->random); + #endif heap->cookie = _mi_heap_random_next(heap) | 1; heap->keys[0] = _mi_heap_random_next(heap); heap->keys[1] = _mi_heap_random_next(heap); From a507335528dfcdc723f40c7e3e64a69ef1c6a214 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 31 Jan 2020 11:44:31 +0100 Subject: [PATCH 046/259] strbuf_realpath(): use platform-dependent API if available Some platforms (e.g. Windows) provide API functions to resolve paths much quicker. Let's offer a way to short-cut `strbuf_realpath()` on those platforms. Signed-off-by: Johannes Schindelin --- abspath.c | 3 +++ git-compat-util.h | 4 ++++ 2 files changed, 7 insertions(+) diff --git a/abspath.c b/abspath.c index 1202cde23dbc9b..0c17e98654e4b0 100644 --- a/abspath.c +++ b/abspath.c @@ -93,6 +93,9 @@ static char *strbuf_realpath_1(struct strbuf *resolved, const char *path, goto error_out; } + if (platform_strbuf_realpath(resolved, path)) + return resolved->buf; + strbuf_addstr(&remaining, path); get_root_part(resolved, &remaining); diff --git a/git-compat-util.h b/git-compat-util.h index e123288e8f1393..079b837e5a1f52 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -630,6 +630,10 @@ static inline int git_has_dir_sep(const char *path) #define query_user_email() NULL #endif +#ifndef platform_strbuf_realpath +#define platform_strbuf_realpath(resolved, path) NULL +#endif + #ifdef __TANDEM #include #include From e7cafe91313d2c64c78998e2444da604138f96c6 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sat, 1 Feb 2020 00:31:16 +0100 Subject: [PATCH 047/259] mingw: allow `git.exe` to be used instead of the "Git wrapper" Git for Windows wants to add `git.exe` to the users' `PATH`, without cluttering the latter with unnecessary executables such as `wish.exe`. To that end, it invented the concept of its "Git wrapper", i.e. a tiny executable located in `C:\Program Files\Git\cmd\git.exe` (originally a CMD script) whose sole purpose is to set up a couple of environment variables and then spawn the _actual_ `git.exe` (which nowadays lives in `C:\Program Files\Git\mingw64\bin\git.exe` for 64-bit, and the obvious equivalent for 32-bit installations). Currently, the following environment variables are set unless already initialized: - `MSYSTEM`, to make sure that the MSYS2 Bash and the MSYS2 Perl interpreter behave as expected, and - `PLINK_PROTOCOL`, to force PuTTY's `plink.exe` to use the SSH protocol instead of Telnet, - `PATH`, to make sure that the `bin` folder in the user's home directory, as well as the `/mingw64/bin` and the `/usr/bin` directories are included. The trick here is that the `/mingw64/bin/` and `/usr/bin/` directories are relative to the top-level installation directory of Git for Windows (which the included Bash interprets as `/`, i.e. as the MSYS pseudo root directory). Using the absence of `MSYSTEM` as a tell-tale, we can detect in `git.exe` whether these environment variables have been initialized properly. Therefore we can call `C:\Program Files\Git\mingw64\bin\git` in-place after this change, without having to call Git through the Git wrapper. Obviously, above-mentioned directories must be _prepended_ to the `PATH` variable, otherwise we risk picking up executables from unrelated Git installations. We do that by constructing the new `PATH` value from scratch, appending `$HOME/bin` (if `HOME` is set), then the MSYS2 system directories, and then appending the original `PATH`. Side note: this modification of the `PATH` variable is independent of the modification necessary to reach the executables and scripts in `/mingw64/libexec/git-core/`, i.e. the `GIT_EXEC_PATH`. That modification is still performed by Git, elsewhere, long after making the changes described above. While we _still_ cannot simply hard-link `mingw64\bin\git.exe` to `cmd` (because the former depends on a couple of `.dll` files that are only in `mingw64\bin`, i.e. calling `...\cmd\git.exe` would fail to load due to missing dependencies), at least we can now avoid that extra process of running the Git wrapper (which then has to wait for the spawned `git.exe` to finish) by calling `...\mingw64\bin\git.exe` directly, via its absolute path. Testing this is in Git's test suite tricky: we set up a "new" MSYS pseudo-root and copy the `git.exe` file into the appropriate location, then verify that `MSYSTEM` is set properly, and also that the `PATH` is modified so that scripts can be found in `$HOME/bin`, `/mingw64/bin/` and `/usr/bin/`. This addresses https://github.com/git-for-windows/git/issues/2283 Signed-off-by: Johannes Schindelin --- compat/mingw.c | 72 +++++++++++++++++++++++++++++++++++++++++++ config.mak.uname | 4 +-- t/t0060-path-utils.sh | 31 ++++++++++++++++++- 3 files changed, 104 insertions(+), 3 deletions(-) diff --git a/compat/mingw.c b/compat/mingw.c index ad00d56ed0cc58..a3e5baf32e8e87 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -2773,6 +2773,50 @@ int xwcstoutf(char *utf, const wchar_t *wcs, size_t utflen) return -1; } +#ifdef ENSURE_MSYSTEM_IS_SET +#if !defined(RUNTIME_PREFIX) || !defined(HAVE_WPGMPTR) +static size_t append_system_bin_dirs(char *path UNUSED, size_t size UNUSED) +{ + return 0; +} +#else +static size_t append_system_bin_dirs(char *path, size_t size) +{ + char prefix[32768]; + const char *slash; + size_t len = xwcstoutf(prefix, _wpgmptr, sizeof(prefix)), off = 0; + + if (len == 0 || len >= sizeof(prefix) || + !(slash = find_last_dir_sep(prefix))) + return 0; + /* strip trailing `git.exe` */ + len = slash - prefix; + + /* strip trailing `cmd` or `mingw64\bin` or `mingw32\bin` or `bin` or `libexec\git-core` */ + if (strip_suffix_mem(prefix, &len, "\\mingw64\\libexec\\git-core") || + strip_suffix_mem(prefix, &len, "\\mingw64\\bin")) + off += xsnprintf(path + off, size - off, + "%.*s\\mingw64\\bin;", (int)len, prefix); + else if (strip_suffix_mem(prefix, &len, "\\mingw32\\libexec\\git-core") || + strip_suffix_mem(prefix, &len, "\\mingw32\\bin")) + off += xsnprintf(path + off, size - off, + "%.*s\\mingw32\\bin;", (int)len, prefix); + else if (strip_suffix_mem(prefix, &len, "\\cmd") || + strip_suffix_mem(prefix, &len, "\\bin") || + strip_suffix_mem(prefix, &len, "\\libexec\\git-core")) + off += xsnprintf(path + off, size - off, + "%.*s\\mingw%d\\bin;", (int)len, prefix, + (int)(sizeof(void *) * 8)); + else + return 0; + + off += xsnprintf(path + off, size - off, + "%.*s\\usr\\bin;", (int)len, prefix); + return off; +} +#endif +#endif + static void setup_windows_environment(void) { char *tmp = getenv("TMPDIR"); @@ -2825,6 +2869,34 @@ static void setup_windows_environment(void) setenv("HOME", tmp, 1); } + if (!getenv("PLINK_PROTOCOL")) + setenv("PLINK_PROTOCOL", "ssh", 0); + +#ifdef ENSURE_MSYSTEM_IS_SET + if (!(tmp = getenv("MSYSTEM")) || !tmp[0]) { + const char *home = getenv("HOME"), *path = getenv("PATH"); + char buf[32768]; + size_t off = 0; + + xsnprintf(buf, sizeof(buf), + "MINGW%d", (int)(sizeof(void *) * 8)); + setenv("MSYSTEM", buf, 1); + + if (home) + off += xsnprintf(buf + off, sizeof(buf) - off, + "%s\\bin;", home); + off += append_system_bin_dirs(buf + off, sizeof(buf) - off); + if (path) + off += xsnprintf(buf + off, sizeof(buf) - off, + "%s", path); + else if (off > 0) + buf[off - 1] = '\0'; + else + buf[0] = '\0'; + setenv("PATH", buf, 1); + } +#endif + if (!getenv("LC_ALL") && !getenv("LC_CTYPE") && !getenv("LANG")) setenv("LC_CTYPE", "C.UTF-8", 1); } diff --git a/config.mak.uname b/config.mak.uname index b12d4e168ae119..23748b65a8e480 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -501,7 +501,7 @@ endif compat/win32/pthread.o compat/win32/syslog.o \ compat/win32/trace2_win32_process_info.o \ compat/win32/dirent.o - COMPAT_CFLAGS = -D__USE_MINGW_ACCESS -DDETECT_MSYS_TTY -DNOGDI -DHAVE_STRING_H -Icompat -Icompat/regex -Icompat/win32 -DSTRIP_EXTENSION=\".exe\" + COMPAT_CFLAGS = -D__USE_MINGW_ACCESS -DDETECT_MSYS_TTY -DENSURE_MSYSTEM_IS_SET -DNOGDI -DHAVE_STRING_H -Icompat -Icompat/regex -Icompat/win32 -DSTRIP_EXTENSION=\".exe\" BASIC_LDFLAGS = -IGNORE:4217 -IGNORE:4049 -NOLOGO -ENTRY:wmainCRTStartup -SUBSYSTEM:CONSOLE # invalidcontinue.obj allows Git's source code to close the same file # handle twice, or to access the osfhandle of an already-closed stdout @@ -730,7 +730,7 @@ ifeq ($(uname_S),MINGW) endif CC = gcc COMPAT_CFLAGS += -D__USE_MINGW_ANSI_STDIO=0 -DDETECT_MSYS_TTY \ - -fstack-protector-strong + -DENSURE_MSYSTEM_IS_SET -fstack-protector-strong EXTLIBS += -lntdll EXTRA_PROGRAMS += headless-git$X INSTALL = /bin/install diff --git a/t/t0060-path-utils.sh b/t/t0060-path-utils.sh index 8545cdfab559b4..b25b6da58ec4be 100755 --- a/t/t0060-path-utils.sh +++ b/t/t0060-path-utils.sh @@ -602,7 +602,8 @@ test_expect_success !VALGRIND,RUNTIME_PREFIX,CAN_EXEC_IN_PWD 'RUNTIME_PREFIX wor echo "echo HERE" | write_script pretend/libexec/git-core/git-here && GIT_EXEC_PATH= ./pretend/bin/git here >actual && echo HERE >expect && - test_cmp expect actual' + test_cmp expect actual +' test_expect_success !VALGRIND,RUNTIME_PREFIX,CAN_EXEC_IN_PWD '%(prefix)/ works' ' git config yes.path "%(prefix)/yes" && @@ -611,4 +612,32 @@ test_expect_success !VALGRIND,RUNTIME_PREFIX,CAN_EXEC_IN_PWD '%(prefix)/ works' test_cmp expect actual ' +test_expect_success MINGW,RUNTIME_PREFIX 'MSYSTEM/PATH is adjusted if necessary' ' + mkdir -p "$HOME"/bin pretend/mingw64/bin \ + pretend/mingw64/libexec/git-core pretend/usr/bin && + cp "$GIT_EXEC_PATH"/git.exe pretend/mingw64/bin/ && + cp "$GIT_EXEC_PATH"/git.exe pretend/mingw64/libexec/git-core/ && + # copy the .dll files, if any (happens when building via CMake) + case "$GIT_EXEC_PATH"/*.dll in + */"*.dll") ;; # no `.dll` files to be copied + *) + cp "$GIT_EXEC_PATH"/*.dll pretend/mingw64/bin/ && + cp "$GIT_EXEC_PATH"/*.dll pretend/mingw64/libexec/git-core/ + ;; + esac && + echo "env | grep MSYSTEM=" | write_script "$HOME"/bin/git-test-home && + echo "echo mingw64" | write_script pretend/mingw64/bin/git-test-bin && + echo "echo usr" | write_script pretend/usr/bin/git-test-bin2 && + + ( + MSYSTEM= && + GIT_EXEC_PATH= && + pretend/mingw64/libexec/git-core/git.exe test-home >actual && + pretend/mingw64/libexec/git-core/git.exe test-bin >>actual && + pretend/mingw64/bin/git.exe test-bin2 >>actual + ) && + test_write_lines MSYSTEM=$MSYSTEM mingw64 usr >expect && + test_cmp expect actual +' + test_done From e142c7d75df029ffe07d8bb04fcba5d0d28aefe8 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 24 Jun 2019 23:45:21 +0200 Subject: [PATCH 048/259] mingw: use mimalloc Thorough benchmarking with repacking a subset of linux.git (the commit history reachable from 93a6fefe2f ([PATCH] fix the SYSCTL=n compilation, 2007-02-28), to be precise) suggest that this allocator is on par, in multi-threaded situations maybe even better than nedmalloc: `git repack -adfq` with mimalloc, 8 threads: 31.166991900 27.576763800 28.712311000 27.373859000 27.163141900 `git repack -adfq` with nedmalloc, 8 threads: 31.915032900 27.149883100 28.244933700 27.240188800 28.580849500 In a different test using GitHub Actions build agents (probably single-threaded, a core-strength of nedmalloc)): `git repack -q -d -l -A --unpack-unreachable=2.weeks.ago` with mimalloc: 943.426 978.500 939.709 959.811 954.605 `git repack -q -d -l -A --unpack-unreachable=2.weeks.ago` with nedmalloc: 995.383 952.179 943.253 963.043 980.468 While these measurements were not executed with complete scientific rigor, as no hardware was set aside specifically for these benchmarks, it shows that mimalloc and nedmalloc perform almost the same, nedmalloc with a bit higher variance and also slightly higher average (further testing suggests that nedmalloc performs worse in multi-threaded situations than in single-threaded ones). In short: mimalloc seems to be slightly better suited for our purposes than nedmalloc. Seeing that mimalloc is developed actively, while nedmalloc ceased to see any updates in eight years, let's use mimalloc on Windows instead. Signed-off-by: Johannes Schindelin --- config.mak.uname | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/config.mak.uname b/config.mak.uname index cdfb606f010ba4..2ebb82feb9f1b2 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -738,7 +738,7 @@ ifeq ($(uname_S),MINGW) HAVE_LIBCHARSET_H = YesPlease USE_GETTEXT_SCHEME = fallthrough USE_LIBPCRE = YesPlease - USE_NED_ALLOCATOR = YesPlease + USE_MIMALLOC = YesPlease NO_PYTHON = ifeq (/mingw64,$(subst 32,64,$(prefix))) # Move system config into top-level /etc/ From 50141b6a3c9f716823c286d9dfcd57257fa42db2 Mon Sep 17 00:00:00 2001 From: Thomas Braun Date: Thu, 8 May 2014 21:43:24 +0200 Subject: [PATCH 049/259] transport: optionally disable side-band-64k Since commit 0c499ea60fda (send-pack: demultiplex a sideband stream with status data, 2010-02-05) the send-pack builtin uses the side-band-64k capability if advertised by the server. Unfortunately this breaks pushing over the dump git protocol if used over a network connection. The detailed reasons for this breakage are (by courtesy of Jeff Preshing, quoted from https://groups.google.com/d/msg/msysgit/at8D7J-h7mw/eaLujILGUWoJ): MinGW wraps Windows sockets in CRT file descriptors in order to mimic the functionality of POSIX sockets. This causes msvcrt.dll to treat sockets as Installable File System (IFS) handles, calling ReadFile, WriteFile, DuplicateHandle and CloseHandle on them. This approach works well in simple cases on recent versions of Windows, but does not support all usage patterns. In particular, using this approach, any attempt to read & write concurrently on the same socket (from one or more processes) will deadlock in a scenario where the read waits for a response from the server which is only invoked after the write. This is what send_pack currently attempts to do in the use_sideband codepath. The new config option `sendpack.sideband` allows to override the side-band-64k capability of the server, and thus makes the dumb git protocol work. Other transportation methods like ssh and http/https still benefit from the sideband channel, therefore the default value of `sendpack.sideband` is still true. Signed-off-by: Thomas Braun Signed-off-by: Oliver Schneider Signed-off-by: Johannes Schindelin --- Documentation/config.adoc | 2 ++ Documentation/config/sendpack.adoc | 5 +++++ send-pack.c | 6 +++--- 3 files changed, 10 insertions(+), 3 deletions(-) create mode 100644 Documentation/config/sendpack.adoc diff --git a/Documentation/config.adoc b/Documentation/config.adoc index a8b04c4e5183d2..644216f02a322a 100644 --- a/Documentation/config.adoc +++ b/Documentation/config.adoc @@ -518,6 +518,8 @@ include::config/safe.adoc[] include::config/sendemail.adoc[] +include::config/sendpack.adoc[] + include::config/sequencer.adoc[] include::config/showbranch.adoc[] diff --git a/Documentation/config/sendpack.adoc b/Documentation/config/sendpack.adoc new file mode 100644 index 00000000000000..e306f657fba7dd --- /dev/null +++ b/Documentation/config/sendpack.adoc @@ -0,0 +1,5 @@ +sendpack.sideband:: + Allows to disable the side-band-64k capability for send-pack even + when it is advertised by the server. Makes it possible to work + around a limitation in the git for windows implementation together + with the dump git protocol. Defaults to true. diff --git a/send-pack.c b/send-pack.c index 856a65d5f5abce..5ae6d9f8e7408a 100644 --- a/send-pack.c +++ b/send-pack.c @@ -497,7 +497,7 @@ int send_pack(struct repository *r, int need_pack_data = 0; int allow_deleting_refs = 0; int status_report = 0; - int use_sideband = 0; + int use_sideband = 1; int quiet_supported = 0; int agent_supported = 0; int advertise_sid = 0; @@ -521,6 +521,7 @@ int send_pack(struct repository *r, goto out; } + repo_config_get_bool(r, "sendpack.sideband", &use_sideband); repo_config_get_bool(r, "push.negotiate", &push_negotiate); if (push_negotiate) { trace2_region_enter("send_pack", "push_negotiate", r); @@ -542,8 +543,7 @@ int send_pack(struct repository *r, allow_deleting_refs = 1; if (server_supports("ofs-delta")) args->use_ofs_delta = 1; - if (server_supports("side-band-64k")) - use_sideband = 1; + use_sideband = use_sideband && server_supports("side-band-64k"); if (server_supports("quiet")) quiet_supported = 1; if (server_supports("agent")) From dfae88ef5841618a0c4a8593f7cb07c32c52c532 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 1 Jan 2020 21:07:22 +0100 Subject: [PATCH 050/259] mingw: do resolve symlinks in `getcwd()` As pointed out in https://github.com/git-for-windows/git/issues/1676, the `git rev-parse --is-inside-work-tree` command currently fails when the current directory's path contains symbolic links. The underlying reason for this bug is that `getcwd()` is supposed to resolve symbolic links, but our `mingw_getcwd()` implementation did not. We do have all the building blocks for that, though: the `GetFinalPathByHandleW()` function will resolve symbolic links. However, we only called that function if `GetLongPathNameW()` failed, for historical reasons: the latter function was supported for a long time, but the former API function was introduced only with Windows Vista, and we used to support also Windows XP. With that support having been dropped, we are free to call the symbolic link-resolving function right away. Signed-off-by: Johannes Schindelin --- compat/mingw.c | 18 +++++++----------- 1 file changed, 7 insertions(+), 11 deletions(-) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..579a6a815f777a 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -1211,18 +1211,16 @@ char *mingw_getcwd(char *pointer, int len) { wchar_t cwd[MAX_PATH], wpointer[MAX_PATH]; DWORD ret = GetCurrentDirectoryW(ARRAY_SIZE(cwd), cwd); + HANDLE hnd; if (!ret || ret >= ARRAY_SIZE(cwd)) { errno = ret ? ENAMETOOLONG : err_win_to_posix(GetLastError()); return NULL; } - ret = GetLongPathNameW(cwd, wpointer, ARRAY_SIZE(wpointer)); - if (!ret && GetLastError() == ERROR_ACCESS_DENIED) { - HANDLE hnd = CreateFileW(cwd, 0, - FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, - OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); - if (hnd == INVALID_HANDLE_VALUE) - return NULL; + hnd = CreateFileW(cwd, 0, + FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, + OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); + if (hnd != INVALID_HANDLE_VALUE) { ret = GetFinalPathNameByHandleW(hnd, wpointer, ARRAY_SIZE(wpointer), 0); CloseHandle(hnd); if (!ret || ret >= ARRAY_SIZE(wpointer)) @@ -1231,13 +1229,11 @@ char *mingw_getcwd(char *pointer, int len) return NULL; return pointer; } - if (!ret || ret >= ARRAY_SIZE(wpointer)) - return NULL; - if (GetFileAttributesW(wpointer) == INVALID_FILE_ATTRIBUTES) { + if (GetFileAttributesW(cwd) == INVALID_FILE_ATTRIBUTES) { errno = ENOENT; return NULL; } - if (xwcstoutf(pointer, wpointer, len) < 0) + if (xwcstoutf(pointer, cwd, len) < 0) return NULL; convert_slashes(pointer); return pointer; From e5ed42706b7bb6bab56184916de417284b44c97d Mon Sep 17 00:00:00 2001 From: Bjoern Mueller Date: Wed, 22 Jan 2020 13:49:13 +0100 Subject: [PATCH 051/259] mingw: fix fatal error working on mapped network drives on Windows In 1e64d18 (mingw: do resolve symlinks in `getcwd()`) a problem was introduced that causes git for Windows to stop working with certain mapped network drives (in particular, drives that are mapped to locations with long path names). Error message was "fatal: Unable to read current working directory: No such file or directory". Present change fixes this issue as discussed in https://github.com/git-for-windows/git/issues/2480 Signed-off-by: Bjoern Mueller --- compat/mingw.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..41943760f8c8fa 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -1225,8 +1225,13 @@ char *mingw_getcwd(char *pointer, int len) return NULL; ret = GetFinalPathNameByHandleW(hnd, wpointer, ARRAY_SIZE(wpointer), 0); CloseHandle(hnd); - if (!ret || ret >= ARRAY_SIZE(wpointer)) - return NULL; + if (!ret || ret >= ARRAY_SIZE(wpointer)) { + ret = GetLongPathNameW(cwd, wpointer, ARRAY_SIZE(wpointer)); + if (!ret || ret >= ARRAY_SIZE(wpointer)) { + errno = ret ? ENAMETOOLONG : err_win_to_posix(GetLastError()); + return NULL; + } + } if (xwcstoutf(pointer, normalize_ntpath(wpointer), len) < 0) return NULL; return pointer; From 7f26170b90b57c660abdfa51cec3c52a8bfcb0c6 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Thu, 30 Jan 2020 14:22:27 -0500 Subject: [PATCH 052/259] clink.pl: fix MSVC compile script to handle libcurl-d.lib Update clink.pl to link with either libcurl.lib or libcurl-d.lib depending on whether DEBUG=1 is set. Signed-off-by: Jeff Hostetler Signed-off-by: Johannes Schindelin --- compat/vcbuild/scripts/clink.pl | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/compat/vcbuild/scripts/clink.pl b/compat/vcbuild/scripts/clink.pl index 3bd824154be381..c4c99d1a11f18c 100755 --- a/compat/vcbuild/scripts/clink.pl +++ b/compat/vcbuild/scripts/clink.pl @@ -56,7 +56,8 @@ # need to use that instead? foreach my $flag (@lflags) { if ($flag =~ /^-LIBPATH:(.*)/) { - foreach my $l ("libcurl_imp.lib", "libcurl.lib") { + my $libcurl = $is_debug ? "libcurl-d.lib" : "libcurl.lib"; + foreach my $l ("libcurl_imp.lib", $libcurl) { if (-f "$1/$l") { $lib = $l; last; From 71f46917f25ba36cf5adb5a5e09fcdb2fb70db06 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 31 Jan 2020 11:49:04 +0100 Subject: [PATCH 053/259] mingw: implement a platform-specific `strbuf_realpath()` There is a Win32 API function to resolve symbolic links, and we can use that instead of resolving them manually. Even better, this function also resolves NTFS junction points (which are somewhat similar to bind mounts). This fixes https://github.com/git-for-windows/git/issues/2481. Signed-off-by: Johannes Schindelin --- compat/mingw.c | 76 +++++++++++++++++++++++++++++++++++++++++++ compat/mingw.h | 3 ++ t/t0060-path-utils.sh | 8 +++++ t/t3700-add.sh | 2 +- t/t5601-clone.sh | 7 ++++ 5 files changed, 95 insertions(+), 1 deletion(-) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..7a4dc87fa17e10 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -1207,6 +1207,82 @@ struct tm *localtime_r(const time_t *timep, struct tm *result) } #endif +char *mingw_strbuf_realpath(struct strbuf *resolved, const char *path) +{ + wchar_t wpath[MAX_PATH]; + HANDLE h; + DWORD ret; + int len; + const char *last_component = NULL; + char *append = NULL; + + if (xutftowcs_path(wpath, path) < 0) + return NULL; + + h = CreateFileW(wpath, 0, + FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, + OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); + + /* + * strbuf_realpath() allows the last path component to not exist. If + * that is the case, now it's time to try without last component. + */ + if (h == INVALID_HANDLE_VALUE && + GetLastError() == ERROR_FILE_NOT_FOUND) { + /* cut last component off of `wpath` */ + wchar_t *p = wpath + wcslen(wpath); + + while (p != wpath) + if (*(--p) == L'/' || *p == L'\\') + break; /* found start of last component */ + + if (p != wpath && (last_component = find_last_dir_sep(path))) { + append = xstrdup(last_component + 1); /* skip directory separator */ + /* + * Do not strip the trailing slash at the drive root, otherwise + * the path would be e.g. `C:` (which resolves to the + * _current_ directory on that drive). + */ + if (p[-1] == L':') + p[1] = L'\0'; + else + *p = L'\0'; + h = CreateFileW(wpath, 0, FILE_SHARE_READ | + FILE_SHARE_WRITE | FILE_SHARE_DELETE, + NULL, OPEN_EXISTING, + FILE_FLAG_BACKUP_SEMANTICS, NULL); + } + } + + if (h == INVALID_HANDLE_VALUE) { +realpath_failed: + FREE_AND_NULL(append); + return NULL; + } + + ret = GetFinalPathNameByHandleW(h, wpath, ARRAY_SIZE(wpath), 0); + CloseHandle(h); + if (!ret || ret >= ARRAY_SIZE(wpath)) + goto realpath_failed; + + len = wcslen(wpath) * 3; + strbuf_grow(resolved, len); + len = xwcstoutf(resolved->buf, normalize_ntpath(wpath), len); + if (len < 0) + goto realpath_failed; + resolved->len = len; + + if (append) { + /* Use forward-slash, like `normalize_ntpath()` */ + strbuf_complete(resolved, '/'); + strbuf_addstr(resolved, append); + FREE_AND_NULL(append); + } + + return resolved->buf; + +} + char *mingw_getcwd(char *pointer, int len) { wchar_t cwd[MAX_PATH], wpointer[MAX_PATH]; diff --git a/compat/mingw.h b/compat/mingw.h index ebfb8ba423b754..35c427376cbc6d 100644 --- a/compat/mingw.h +++ b/compat/mingw.h @@ -457,6 +457,9 @@ static inline void convert_slashes(char *path) #define PATH_SEP ';' char *mingw_query_user_email(void); #define query_user_email mingw_query_user_email +struct strbuf; +char *mingw_strbuf_realpath(struct strbuf *resolved, const char *path); +#define platform_strbuf_realpath mingw_strbuf_realpath #if !defined(__MINGW64_VERSION_MAJOR) && (!defined(_MSC_VER) || _MSC_VER < 1800) #define PRIuMAX "I64u" #define PRId64 "I64d" diff --git a/t/t0060-path-utils.sh b/t/t0060-path-utils.sh index 8545cdfab559b4..eb2ab9d437ea8e 100755 --- a/t/t0060-path-utils.sh +++ b/t/t0060-path-utils.sh @@ -281,6 +281,14 @@ test_expect_success SYMLINKS 'real path works on symlinks' ' test_cmp expect actual ' +test_expect_success MINGW 'real path works near drive root' ' + # we need a non-existing path at the drive root; simply skip if C:/xyz exists + if test ! -e C:/xyz + then + test C:/xyz = $(test-tool path-utils real_path C:/xyz) + fi +' + test_expect_success SYMLINKS 'prefix_path works with absolute paths to work tree symlinks' ' ln -s target symlink && echo "symlink" >expect && diff --git a/t/t3700-add.sh b/t/t3700-add.sh index d934f3417faa45..10185f31f019a9 100755 --- a/t/t3700-add.sh +++ b/t/t3700-add.sh @@ -548,7 +548,7 @@ test_expect_success CASE_INSENSITIVE_FS 'path is case-insensitive' ' git add "$downcased" ' -test_expect_failure MINGW 'can add files via NTFS junctions' ' +test_expect_success MINGW 'can add files via NTFS junctions' ' test_when_finished "cmd //c rmdir junction && rm -rf target" && test_create_repo target && cmd //c "mklink /j junction target" && diff --git a/t/t5601-clone.sh b/t/t5601-clone.sh index d0c18660e33113..b3ecd6b8b06c33 100755 --- a/t/t5601-clone.sh +++ b/t/t5601-clone.sh @@ -78,6 +78,13 @@ test_expect_success 'clone respects GIT_WORK_TREE' ' ' +test_expect_success CASE_INSENSITIVE_FS 'core.worktree is not added due to path case' ' + + mkdir UPPERCASE && + git clone src "$(pwd)/uppercase" && + test "unset" = "$(git -C UPPERCASE config --default unset core.worktree)" +' + test_expect_success 'clone from hooks' ' test_create_repo r0 && From e68d06573799c11b6f074c68c4c7079f6c6ff65d Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sat, 9 May 2020 16:19:06 +0200 Subject: [PATCH 054/259] t5505/t5516: allow running without `.git/branches/` in the templates When we commit the template directory as part of `make vcxproj`, the `branches/` directory is not actually commited, as it is empty. Two tests were not prepared for that situation. This developer tried to get rid of the support for `.git/branches/` a long time ago, but that effort did not bear fruit, so the best we can do is work around in these here tests. Signed-off-by: Johannes Schindelin --- t/t5505-remote.sh | 4 ++-- t/t5516-fetch-push.sh | 8 ++++---- 2 files changed, 6 insertions(+), 6 deletions(-) diff --git a/t/t5505-remote.sh b/t/t5505-remote.sh index bb7e0c6879ed8a..75e2177e8ce254 100755 --- a/t/t5505-remote.sh +++ b/t/t5505-remote.sh @@ -1155,7 +1155,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'migrate a remote from named file i ( cd six && git remote rm origin && - mkdir .git/branches && + mkdir -p .git/branches && echo "$origin_url#main" >.git/branches/origin && git remote rename origin origin && test_path_is_missing .git/branches/origin && @@ -1170,7 +1170,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'migrate a remote from named file i ( cd seven && git remote rm origin && - mkdir .git/branches && + mkdir -p .git/branches && echo "quux#foom" > .git/branches/origin && git remote rename origin origin && test_path_is_missing .git/branches/origin && diff --git a/t/t5516-fetch-push.sh b/t/t5516-fetch-push.sh index 85ed049627d2e6..3aa7cc14c032a7 100755 --- a/t/t5516-fetch-push.sh +++ b/t/t5516-fetch-push.sh @@ -979,7 +979,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'fetch with branches' ' mk_empty testrepo && git branch second $the_first_commit && git checkout second && - mkdir testrepo/.git/branches && + mkdir -p testrepo/.git/branches && echo ".." > testrepo/.git/branches/branch1 && ( cd testrepo && @@ -993,7 +993,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'fetch with branches' ' test_expect_success WITHOUT_BREAKING_CHANGES 'fetch with branches containing #' ' mk_empty testrepo && - mkdir testrepo/.git/branches && + mkdir -p testrepo/.git/branches && echo "..#second" > testrepo/.git/branches/branch2 && ( cd testrepo && @@ -1010,7 +1010,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'push with branches' ' git checkout second && test_when_finished "rm -rf .git/branches" && - mkdir .git/branches && + mkdir -p .git/branches && echo "testrepo" > .git/branches/branch1 && git push branch1 && @@ -1026,7 +1026,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'push with branches containing #' ' mk_empty testrepo && test_when_finished "rm -rf .git/branches" && - mkdir .git/branches && + mkdir -p .git/branches && echo "testrepo#branch3" > .git/branches/branch2 && git push branch2 && From 4f063aa4bde5cc6ab6397a019b19c92cda9b2af6 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 25 Aug 2020 12:13:26 +0200 Subject: [PATCH 055/259] mingw: ignore HOMEDRIVE/HOMEPATH if it points to Windows' system directory Internally, Git expects the environment variable `HOME` to be set, and to point to the current user's home directory. This environment variable is not set by default on Windows, and therefore Git tries its best to construct one if it finds `HOME` unset. There are actually two different approaches Git tries: first, it looks at `HOMEDRIVE`/`HOMEPATH` because this is widely used in corporate environments with roaming profiles, and a user generally wants their global Git settings to be in a roaming profile. Only when `HOMEDRIVE`/`HOMEPATH` is either unset or does not point to a valid location, Git will fall back to using `USERPROFILE` instead. However, starting with Windows Vista, for secondary logons and services, the environment variables `HOMEDRIVE`/`HOMEPATH` point to Windows' system directory (usually `C:\Windows\system32`). That is undesirable, and that location is usually write-protected anyway. So let's verify that the `HOMEDRIVE`/`HOMEPATH` combo does not point to Windows' system directory before using it, falling back to `USERPROFILE` if it does. This fixes git-for-windows#2709 Initial-Path-by: Ivan Pozdeev Signed-off-by: Johannes Schindelin --- compat/mingw.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/compat/mingw.c b/compat/mingw.c index a3e5baf32e8e87..2b18f81c38e206 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -2817,6 +2817,18 @@ static size_t append_system_bin_dirs(char *path, size_t size) #endif #endif +static int is_system32_path(const char *path) +{ + WCHAR system32[MAX_PATH], wpath[MAX_PATH]; + + if (xutftowcs_path(wpath, path) < 0 || + !GetSystemDirectoryW(system32, ARRAY_SIZE(system32)) || + _wcsicmp(system32, wpath)) + return 0; + + return 1; +} + static void setup_windows_environment(void) { char *tmp = getenv("TMPDIR"); @@ -2857,7 +2869,8 @@ static void setup_windows_environment(void) strbuf_addstr(&buf, tmp); if ((tmp = getenv("HOMEPATH"))) { strbuf_addstr(&buf, tmp); - if (is_directory(buf.buf)) + if (!is_system32_path(buf.buf) && + is_directory(buf.buf)) setenv("HOME", buf.buf, 1); else tmp = NULL; /* use $USERPROFILE */ From b4859e8fc36a1b10fdc62663318ea2e6d5220d8b Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 4 Mar 2020 21:55:28 +0100 Subject: [PATCH 056/259] http: use new "best effort" strategy for Secure Channel revoke checking MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The native Windows HTTPS backend is based on Secure Channel which lets the caller decide how to handle revocation checking problems caused by missing information in the certificate or offline CRL distribution points. Unfortunately, cURL chose to handle these problems differently than OpenSSL by default: while OpenSSL happily ignores those problems (essentially saying "¯\_(ツ)_/¯"), the Secure Channel backend will error out instead. As a remedy, the "no revoke" mode was introduced, which turns off revocation checking altogether. This is a bit heavy-handed. We support this via the `http.schannelCheckRevoke` setting. In https://github.com/curl/curl/pull/4981, we contributed an opt-in "best effort" strategy that emulates what OpenSSL seems to do. In Git for Windows, we actually want this to be the default. This patch makes it so, introducing it as a new value for the `http.schannelCheckRevoke" setting, which now becmes a tristate: it accepts the values "false", "true" or "best-effort" (defaulting to the last one). Signed-off-by: Johannes Schindelin --- Documentation/config/http.adoc | 12 +++++++----- http.c | 26 ++++++++++++++++++++++---- 2 files changed, 29 insertions(+), 9 deletions(-) diff --git a/Documentation/config/http.adoc b/Documentation/config/http.adoc index 22a8803deaf964..de52a1c44fbf14 100644 --- a/Documentation/config/http.adoc +++ b/Documentation/config/http.adoc @@ -233,11 +233,13 @@ http.sslKeyType:: http.schannelCheckRevoke:: Used to enforce or disable certificate revocation checks in cURL - when http.sslBackend is set to "schannel". Defaults to `true` if - unset. Only necessary to disable this if Git consistently errors - and the message is about checking the revocation status of a - certificate. This option is ignored if cURL lacks support for - setting the relevant SSL option at runtime. + when http.sslBackend is set to "schannel" via "true" and "false", + respectively. Another accepted value is "best-effort" (the default) + in which case revocation checks are performed, but errors due to + revocation list distribution points that are offline are silently + ignored, as well as errors due to certificates missing revocation + list distribution points. This option is ignored if cURL lacks + support for setting the relevant SSL option at runtime. http.schannelUseSSLCAInfo:: As of cURL v7.60.0, the Secure Channel backend can use the diff --git a/http.c b/http.c index 0c9a872809f872..657c3e1dc5e2c2 100644 --- a/http.c +++ b/http.c @@ -143,7 +143,13 @@ static char *cached_accept_language; static char *http_ssl_backend; -static int http_schannel_check_revoke = 1; +static int http_schannel_check_revoke_mode = +#ifdef CURLSSLOPT_REVOKE_BEST_EFFORT + CURLSSLOPT_REVOKE_BEST_EFFORT; +#else + CURLSSLOPT_NO_REVOKE; +#endif + /* * With the backend being set to `schannel`, setting sslCAinfo would override * the Certificate Store in cURL v7.60.0 and later, which is not what we want @@ -418,7 +424,19 @@ static int http_options(const char *var, const char *value, } if (!strcmp("http.schannelcheckrevoke", var)) { - http_schannel_check_revoke = git_config_bool(var, value); + if (value && !strcmp(value, "best-effort")) { + http_schannel_check_revoke_mode = +#ifdef CURLSSLOPT_REVOKE_BEST_EFFORT + CURLSSLOPT_REVOKE_BEST_EFFORT; +#else + CURLSSLOPT_NO_REVOKE; + warning(_("%s=%s unsupported by current cURL"), + var, value); +#endif + } else + http_schannel_check_revoke_mode = + (git_config_bool(var, value) ? + 0 : CURLSSLOPT_NO_REVOKE); return 0; } @@ -1044,8 +1062,8 @@ static CURL *get_curl_handle(void) #endif if (http_ssl_backend && !strcmp("schannel", http_ssl_backend) && - !http_schannel_check_revoke) { - curl_easy_setopt(result, CURLOPT_SSL_OPTIONS, CURLSSLOPT_NO_REVOKE); + http_schannel_check_revoke_mode) { + curl_easy_setopt(result, CURLOPT_SSL_OPTIONS, http_schannel_check_revoke_mode); } if (http_proactive_auth != PROACTIVE_AUTH_NONE) From 0d93c661645caab63ea2024b9026961e0e74fb70 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sat, 9 May 2020 19:24:23 +0200 Subject: [PATCH 057/259] t5505/t5516: fix white-space around redirectors The convention in Git project's shell scripts is to have white-space _before_, but not _after_ the `>` (or `<`). Signed-off-by: Johannes Schindelin --- t/t5505-remote.sh | 6 +++--- t/t5516-fetch-push.sh | 10 +++++----- 2 files changed, 8 insertions(+), 8 deletions(-) diff --git a/t/t5505-remote.sh b/t/t5505-remote.sh index 75e2177e8ce254..a89b2883f35079 100755 --- a/t/t5505-remote.sh +++ b/t/t5505-remote.sh @@ -951,8 +951,8 @@ test_expect_success '"remote show" does not show symbolic refs' ' ( cd three && git remote show origin >output && - ! grep "^ *HEAD$" < output && - ! grep -i stale < output + ! grep "^ *HEAD$" .git/branches/origin && + echo "quux#foom" >.git/branches/origin && git remote rename origin origin && test_path_is_missing .git/branches/origin && test "$(git config remote.origin.url)" = "quux" && diff --git a/t/t5516-fetch-push.sh b/t/t5516-fetch-push.sh index 3aa7cc14c032a7..ecca661e58a44c 100755 --- a/t/t5516-fetch-push.sh +++ b/t/t5516-fetch-push.sh @@ -980,7 +980,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'fetch with branches' ' git branch second $the_first_commit && git checkout second && mkdir -p testrepo/.git/branches && - echo ".." > testrepo/.git/branches/branch1 && + echo ".." >testrepo/.git/branches/branch1 && ( cd testrepo && git fetch branch1 && @@ -994,7 +994,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'fetch with branches' ' test_expect_success WITHOUT_BREAKING_CHANGES 'fetch with branches containing #' ' mk_empty testrepo && mkdir -p testrepo/.git/branches && - echo "..#second" > testrepo/.git/branches/branch2 && + echo "..#second" >testrepo/.git/branches/branch2 && ( cd testrepo && git fetch branch2 && @@ -1011,7 +1011,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'push with branches' ' test_when_finished "rm -rf .git/branches" && mkdir -p .git/branches && - echo "testrepo" > .git/branches/branch1 && + echo "testrepo" >.git/branches/branch1 && git push branch1 && ( @@ -1027,7 +1027,7 @@ test_expect_success WITHOUT_BREAKING_CHANGES 'push with branches containing #' ' test_when_finished "rm -rf .git/branches" && mkdir -p .git/branches && - echo "testrepo#branch3" > .git/branches/branch2 && + echo "testrepo#branch3" >.git/branches/branch2 && git push branch2 && ( @@ -1555,7 +1555,7 @@ EOF git init no-thin && git --git-dir=no-thin/.git config receive.unpacklimit 0 && git push no-thin/.git refs/heads/main:refs/heads/foo && - echo modified >> path1 && + echo modified >>path1 && git commit -am modified && git repack -adf && rcvpck="git receive-pack --reject-thin-pack-for-testing" && From 3503c1ca5fb0c78f56e771b6a0422ce5b9a75198 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sat, 12 Sep 2015 12:25:47 +0200 Subject: [PATCH 058/259] t3701: verify that we can add *lots* of files interactively Signed-off-by: Johannes Schindelin --- t/t3701-add-interactive.sh | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) diff --git a/t/t3701-add-interactive.sh b/t/t3701-add-interactive.sh index b8a05d95f3f130..81d0487b3b90eb 100755 --- a/t/t3701-add-interactive.sh +++ b/t/t3701-add-interactive.sh @@ -1166,6 +1166,27 @@ test_expect_success 'checkout -p patch editing of added file' ' ) ' +test_expect_success EXPENSIVE 'add -i with a lot of files' ' + git reset --hard && + x160=0123456789012345678901234567890123456789 && + x160=$x160$x160$x160$x160 && + y= && + i=0 && + while test $i -le 200 + do + name=$(printf "%s%03d" $x160 $i) && + echo $name >$name && + git add -N $name && + y="${y}y$LF" && + i=$(($i+1)) || + exit 1 + done && + echo "$y" | git add -p -- . && + git diff --cached >staged && + test_line_count = 1407 staged && + git reset --hard +' + test_expect_success 'show help from add--helper' ' git reset --hard && cat >expect <<-EOF && From fe3e223f027a4d1038c2d6e39eca904077d41e50 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Thu, 2 Jul 2020 16:35:05 +0200 Subject: [PATCH 059/259] git add -i: handle CR/LF line endings in the interactive input As of Git for Windows v2.27.0, there is an option to use Windows' newly-introduced Pseudo Console support. When running an interactive add operation with this support enabled, Git will receive CR/LF line endings. Therefore, let's not pretend that we are expecting Unix line endings. This fixes https://github.com/git-for-windows/git/issues/2729 Signed-off-by: Johannes Schindelin --- prompt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/prompt.c b/prompt.c index f21c5bf1c7e42f..a07b135050b9c2 100644 --- a/prompt.c +++ b/prompt.c @@ -80,7 +80,7 @@ int git_read_line_interactively(struct strbuf *line) int ret; fflush(stdout); - ret = strbuf_getline_lf(line, stdin); + ret = strbuf_getline(line, stdin); if (ret != EOF) strbuf_trim_trailing_newline(line); From efcc9f8134bd485a718023a5d0bf87396289099e Mon Sep 17 00:00:00 2001 From: Luke Bonanomi Date: Wed, 24 Jun 2020 07:45:52 -0400 Subject: [PATCH 060/259] commit: accept "scissors" with CR/LF line endings This change enhances `git commit --cleanup=scissors` by detecting scissors lines ending in either LF (UNIX-style) or CR/LF (DOS-style). Regression tests are included to specifically test for trailing comments after a CR/LF-terminated scissors line. Signed-off-by: Luke Bonanomi Signed-off-by: Johannes Schindelin --- t/t7502-commit-porcelain.sh | 42 +++++++++++++++++++++++++++++++++++++ wt-status.c | 13 +++++++++--- 2 files changed, 52 insertions(+), 3 deletions(-) diff --git a/t/t7502-commit-porcelain.sh b/t/t7502-commit-porcelain.sh index b37e2018a74a7b..c38b96b66cd20a 100755 --- a/t/t7502-commit-porcelain.sh +++ b/t/t7502-commit-porcelain.sh @@ -623,6 +623,48 @@ test_expect_success 'cleanup commit messages (scissors option,-F,-e, scissors on test_must_be_empty actual ' +test_expect_success 'helper-editor' ' + + write_script lf-to-crlf.sh <<-\EOF + sed "s/\$/Q/" <"$1" | tr Q "\\015" >"$1".new && + mv -f "$1".new "$1" + EOF +' + +test_expect_success 'cleanup commit messages (scissors option,-F,-e, CR/LF line endings)' ' + + test_config core.editor "\"$PWD/lf-to-crlf.sh\"" && + scissors="# ------------------------ >8 ------------------------" && + + test_write_lines >text \ + "# Keep this comment" "" " $scissors" \ + "# Keep this comment, too" "$scissors" \ + "# Remove this comment" "$scissors" \ + "Remove this comment, too" && + + test_write_lines >expect \ + "# Keep this comment" "" " $scissors" \ + "# Keep this comment, too" && + + git commit --cleanup=scissors -e -F text --allow-empty && + git cat-file -p HEAD >raw && + sed -e "1,/^\$/d" raw >actual && + test_cmp expect actual +' + +test_expect_success 'cleanup commit messages (scissors option,-F,-e, scissors on first line, CR/LF line endings)' ' + + scissors="# ------------------------ >8 ------------------------" && + test_write_lines >text \ + "$scissors" \ + "# Remove this comment and any following lines" && + cp text /tmp/test2-text && + git commit --cleanup=scissors -e -F text --allow-empty --allow-empty-message && + git cat-file -p HEAD >raw && + sed -e "1,/^\$/d" raw >actual && + test_must_be_empty actual +' + test_expect_success 'cleanup commit messages (strip option,-F)' ' echo >>negative && diff --git a/wt-status.c b/wt-status.c index 1da5732f57b115..50bace8378bd15 100644 --- a/wt-status.c +++ b/wt-status.c @@ -40,7 +40,7 @@ #define UF_DELAY_WARNING_IN_MS (2 * 1000) static const char cut_line[] = -"------------------------ >8 ------------------------\n"; +"------------------------ >8 ------------------------"; static char default_wt_status_colors[][COLOR_MAXLEN] = { GIT_COLOR_NORMAL, /* WT_STATUS_HEADER */ @@ -1096,15 +1096,22 @@ static void wt_longstatus_print_other(struct wt_status *s, status_printf_ln(s, GIT_COLOR_NORMAL, "%s", ""); } +static inline int starts_with_newline(const char *p) +{ + return *p == '\n' || (*p == '\r' && p[1] == '\n'); +} + size_t wt_status_locate_end(const char *s, size_t len) { const char *p; struct strbuf pattern = STRBUF_INIT; strbuf_addf(&pattern, "\n%s %s", comment_line_str, cut_line); - if (starts_with(s, pattern.buf + 1)) + if (starts_with(s, pattern.buf + 1) && + starts_with_newline(s + pattern.len - 1)) len = 0; - else if ((p = strstr(s, pattern.buf))) { + else if ((p = strstr(s, pattern.buf)) && + starts_with_newline(p + pattern.len)) { size_t newlen = p - s + 1; if (newlen < len) len = newlen; From 56773d3513ddfbd043ba5b869b6360d213211614 Mon Sep 17 00:00:00 2001 From: Jens Glathe Date: Tue, 2 Jun 2020 12:12:25 +0200 Subject: [PATCH 061/259] t0014: fix indentation For some reason, this test case was indented with 4 spaces instead of 1 horizontal tab. The other test cases in the same test script are fine. Signed-off-by: Jens Glathe Signed-off-by: Johannes Schindelin --- t/t0014-alias.sh | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/t/t0014-alias.sh b/t/t0014-alias.sh index 854d59ec58c25a..30708146887d19 100755 --- a/t/t0014-alias.sh +++ b/t/t0014-alias.sh @@ -38,10 +38,10 @@ test_expect_success 'looping aliases - internal execution' ' #' test_expect_success 'run-command formats empty args properly' ' - test_must_fail env GIT_TRACE=1 git frotz a "" b " " c 2>actual.raw && - sed -ne "/run_command:/s/.*trace: run_command: //p" actual.raw >actual && - echo "git-frotz a '\'''\'' b '\'' '\'' c" >expect && - test_cmp expect actual + test_must_fail env GIT_TRACE=1 git frotz a "" b " " c 2>actual.raw && + sed -ne "/run_command:/s/.*trace: run_command: //p" actual.raw >actual && + echo "git-frotz a '\'''\'' b '\'' '\'' c" >expect && + test_cmp expect actual ' test_expect_success 'tracing a shell alias with arguments shows trace of prepared command' ' From 8b8b511eacfcf24ea01ebeaa23d3abc49a4a4084 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Tue, 30 Mar 2021 14:25:31 -0400 Subject: [PATCH 062/259] clink.pl: fix libexpatd.lib link error when using MSVC When building with `make MSVC=1 DEBUG=1`, link to `libexpatd.lib` rather than `libexpat.lib`. It appears that the `vcpkg` package for "libexpat" has changed and now creates `libexpatd.lib` for debug mode builds. Previously, both debug and release builds created a ".lib" with the same basename. Signed-off-by: Jeff Hostetler --- compat/vcbuild/scripts/clink.pl | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/compat/vcbuild/scripts/clink.pl b/compat/vcbuild/scripts/clink.pl index 3bd824154be381..2768ae15f1879f 100755 --- a/compat/vcbuild/scripts/clink.pl +++ b/compat/vcbuild/scripts/clink.pl @@ -66,7 +66,11 @@ } push(@args, $lib); } elsif ("$arg" eq "-lexpat") { + if ($is_debug) { + push(@args, "libexpatd.lib"); + } else { push(@args, "libexpat.lib"); + } } elsif ("$arg" =~ /^-L/ && "$arg" ne "-LTCG") { $arg =~ s/^-L/-LIBPATH:/; push(@lflags, $arg); From 5df9a4955dc5a8179285df2018e0851208c8cf6c Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 5 Apr 2021 15:27:38 -0400 Subject: [PATCH 063/259] Makefile: clean up .ilk files when MSVC=1 Signed-off-by: Jeff Hostetler --- Makefile | 3 +++ 1 file changed, 3 insertions(+) diff --git a/Makefile b/Makefile index 7315507381e146..e77a81be291f35 100644 --- a/Makefile +++ b/Makefile @@ -3781,12 +3781,15 @@ ifdef MSVC $(RM) $(patsubst %.o,%.o.pdb,$(OBJECTS)) $(RM) headless-git.o.pdb $(RM) $(patsubst %.exe,%.pdb,$(OTHER_PROGRAMS)) + $(RM) $(patsubst %.exe,%.ilk,$(OTHER_PROGRAMS)) $(RM) $(patsubst %.exe,%.iobj,$(OTHER_PROGRAMS)) $(RM) $(patsubst %.exe,%.ipdb,$(OTHER_PROGRAMS)) $(RM) $(patsubst %.exe,%.pdb,$(PROGRAMS)) + $(RM) $(patsubst %.exe,%.ilk,$(PROGRAMS)) $(RM) $(patsubst %.exe,%.iobj,$(PROGRAMS)) $(RM) $(patsubst %.exe,%.ipdb,$(PROGRAMS)) $(RM) $(patsubst %.exe,%.pdb,$(TEST_PROGRAMS)) + $(RM) $(patsubst %.exe,%.ilk,$(TEST_PROGRAMS)) $(RM) $(patsubst %.exe,%.iobj,$(TEST_PROGRAMS)) $(RM) $(patsubst %.exe,%.ipdb,$(TEST_PROGRAMS)) $(RM) compat/vcbuild/MSVC-DEFS-GEN From 019f3118e573f024fbce6bbc41708608e44d0d11 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 5 Apr 2021 14:08:22 -0400 Subject: [PATCH 064/259] vcbuild: add support for compiling Windows resource files Create a wrapper for the Windows Resource Compiler (RC.EXE) for use by the MSVC=1 builds. This is similar to the CL.EXE and LIB.EXE wrappers used for the MSVC=1 builds. Signed-off-by: Jeff Hostetler --- compat/vcbuild/find_vs_env.bat | 7 ++++++ compat/vcbuild/scripts/rc.pl | 46 ++++++++++++++++++++++++++++++++++ config.mak.uname | 3 ++- 3 files changed, 55 insertions(+), 1 deletion(-) create mode 100644 compat/vcbuild/scripts/rc.pl diff --git a/compat/vcbuild/find_vs_env.bat b/compat/vcbuild/find_vs_env.bat index b35d264c0e6bed..379b16296e09c2 100644 --- a/compat/vcbuild/find_vs_env.bat +++ b/compat/vcbuild/find_vs_env.bat @@ -99,6 +99,7 @@ REM ================================================================ SET sdk_dir=%WindowsSdkDir% SET sdk_ver=%WindowsSDKVersion% + SET sdk_ver_bin_dir=%WindowsSdkVerBinPath%%tgt% SET si=%sdk_dir%Include\%sdk_ver% SET sdk_includes=-I"%si%ucrt" -I"%si%um" -I"%si%shared" SET sl=%sdk_dir%lib\%sdk_ver% @@ -130,6 +131,7 @@ REM ================================================================ SET sdk_dir=%WindowsSdkDir% SET sdk_ver=%WindowsSDKVersion% + SET sdk_ver_bin_dir=%WindowsSdkVerBinPath%bin\amd64 SET si=%sdk_dir%Include\%sdk_ver% SET sdk_includes=-I"%si%ucrt" -I"%si%um" -I"%si%shared" -I"%si%winrt" SET sl=%sdk_dir%lib\%sdk_ver% @@ -160,6 +162,11 @@ REM ================================================================ echo msvc_includes=%msvc_includes% echo msvc_libs=%msvc_libs% + echo sdk_ver_bin_dir=%sdk_ver_bin_dir% + SET X1=%sdk_ver_bin_dir:C:=/C% + SET X2=%X1:\=/% + echo sdk_ver_bin_dir_msys=%X2% + echo sdk_includes=%sdk_includes% echo sdk_libs=%sdk_libs% diff --git a/compat/vcbuild/scripts/rc.pl b/compat/vcbuild/scripts/rc.pl new file mode 100644 index 00000000000000..7bca4cd81c6c63 --- /dev/null +++ b/compat/vcbuild/scripts/rc.pl @@ -0,0 +1,46 @@ +#!/usr/bin/perl -w +###################################################################### +# Compile Resources on Windows +# +# This is a wrapper to facilitate the compilation of Git with MSVC +# using GNU Make as the build system. So, instead of manipulating the +# Makefile into something nasty, just to support non-space arguments +# etc, we use this wrapper to fix the command line options +# +###################################################################### +use strict; +my @args = (); +my @input = (); + +while (@ARGV) { + my $arg = shift @ARGV; + if ("$arg" =~ /^-[dD]/) { + # GIT_VERSION gets passed with too many + # layers of dquote escaping. + $arg =~ s/\\"/"/g; + + push(@args, $arg); + + } elsif ("$arg" eq "-i") { + my $arg = shift @ARGV; + # TODO complain if NULL or is dashed ?? + push(@input, $arg); + + } elsif ("$arg" eq "-o") { + my $arg = shift @ARGV; + # TODO complain if NULL or is dashed ?? + push(@args, "-fo$arg"); + + } else { + push(@args, $arg); + } +} + +push(@args, "-nologo"); +push(@args, "-v"); +push(@args, @input); + +unshift(@args, "rc.exe"); +printf("**** @args\n"); + +exit (system(@args) != 0); diff --git a/config.mak.uname b/config.mak.uname index 8e30da5c411715..8e0ed259ab715b 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -440,7 +440,7 @@ ifeq ($(uname_S),Windows) # link.exe next to, and required by, cl.exe, we have to prepend this # onto the existing $PATH. # - SANE_TOOL_PATH ?= $(msvc_bin_dir_msys) + SANE_TOOL_PATH ?= $(msvc_bin_dir_msys):$(sdk_ver_bin_dir_msys) HAVE_ALLOCA_H = YesPlease NO_PREAD = YesPlease NEEDS_CRYPTO_WITH_SSL = YesPlease @@ -508,6 +508,7 @@ endif # See https://msdn.microsoft.com/en-us/library/ms235330.aspx EXTLIBS = user32.lib advapi32.lib shell32.lib wininet.lib ws2_32.lib invalidcontinue.obj kernel32.lib ntdll.lib PTHREAD_LIBS = + RC = compat/vcbuild/scripts/rc.pl lib = BASIC_CFLAGS += $(vcpkg_inc) $(sdk_includes) $(msvc_includes) ifndef DEBUG From 58ec679a70f75202c446f57c8e8962c1bc2a2423 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 5 Apr 2021 14:12:14 -0400 Subject: [PATCH 065/259] config.mak.uname: add git.rc to MSVC builds Teach MSVC=1 builds to depend on the `git.rc` file so that the resulting executables have Windows-style resources and version number information within them. Signed-off-by: Jeff Hostetler --- config.mak.uname | 1 + 1 file changed, 1 insertion(+) diff --git a/config.mak.uname b/config.mak.uname index 8e0ed259ab715b..8ef58660aa3ecf 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -507,6 +507,7 @@ endif # handle twice, or to access the osfhandle of an already-closed stdout # See https://msdn.microsoft.com/en-us/library/ms235330.aspx EXTLIBS = user32.lib advapi32.lib shell32.lib wininet.lib ws2_32.lib invalidcontinue.obj kernel32.lib ntdll.lib + GITLIBS += git.res PTHREAD_LIBS = RC = compat/vcbuild/scripts/rc.pl lib = From db6fa8d6e80263632276d60e7575ed9d50301d27 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 5 Apr 2021 14:24:52 -0400 Subject: [PATCH 066/259] clink.pl: ignore no-stack-protector arg on MSVC=1 builds Ignore the `-fno-stack-protector` compiler argument when building with MSVC. This will be used in a later commit that needs to build a Win32 GUI app. Signed-off-by: Jeff Hostetler --- compat/vcbuild/scripts/clink.pl | 2 ++ 1 file changed, 2 insertions(+) diff --git a/compat/vcbuild/scripts/clink.pl b/compat/vcbuild/scripts/clink.pl index 2768ae15f1879f..73c8a2b184f38b 100755 --- a/compat/vcbuild/scripts/clink.pl +++ b/compat/vcbuild/scripts/clink.pl @@ -122,6 +122,8 @@ push(@cflags, "-wd4996"); } elsif ("$arg" =~ /^-W[a-z]/) { # let's ignore those + } elsif ("$arg" eq "-fno-stack-protector") { + # eat this } else { push(@args, $arg); } From dc823236c7231d0915d141e41641c68c9eb3aa80 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 12 Aug 2020 15:06:17 +0000 Subject: [PATCH 067/259] git-gui: accommodate for intent-to-add files As of Git v2.28.0, the diff for files staged via `git add -N` marks them as new files. Git GUI was ill-prepared for that, and this patch teaches Git GUI about them. Please note that this will not even fix things with v2.28.0, as the `rp/apply-cached-with-i-t-a` patches are required on Git's side, too. This fixes https://github.com/git-for-windows/git/issues/2779 Signed-off-by: Johannes Schindelin Signed-off-by: Pratyush Yadav --- git-gui/git-gui.sh | 2 ++ git-gui/lib/diff.tcl | 12 ++++++++---- 2 files changed, 10 insertions(+), 4 deletions(-) diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh index 887d6d596c16fc..3390dc15ccaffe 100755 --- a/git-gui/git-gui.sh +++ b/git-gui/git-gui.sh @@ -2078,6 +2078,7 @@ set all_icons(U$ui_index) file_merge set all_icons(T$ui_index) file_statechange set all_icons(_$ui_workdir) file_plain +set all_icons(A$ui_workdir) file_plain set all_icons(M$ui_workdir) file_mod set all_icons(D$ui_workdir) file_question set all_icons(U$ui_workdir) file_merge @@ -2104,6 +2105,7 @@ foreach i { {A_ {mc "Staged for commit"}} {AM {mc "Portions staged for commit"}} {AD {mc "Staged for commit, missing"}} + {AA {mc "Intended to be added"}} {_D {mc "Missing"}} {D_ {mc "Staged for removal"}} diff --git a/git-gui/lib/diff.tcl b/git-gui/lib/diff.tcl index d657bfec05b498..9a593a3f6e1547 100644 --- a/git-gui/lib/diff.tcl +++ b/git-gui/lib/diff.tcl @@ -568,7 +568,8 @@ proc apply_or_revert_hunk {x y revert} { if {$current_diff_side eq $ui_index} { set failed_msg [mc "Failed to unstage selected hunk."] lappend apply_cmd --reverse --cached - if {[string index $mi 0] ne {M}} { + set file_state [string index $mi 0] + if {$file_state ne {M} && $file_state ne {A}} { unlock_index return } @@ -581,7 +582,8 @@ proc apply_or_revert_hunk {x y revert} { lappend apply_cmd --cached } - if {[string index $mi 1] ne {M}} { + set file_state [string index $mi 1] + if {$file_state ne {M} && $file_state ne {A}} { unlock_index return } @@ -673,7 +675,8 @@ proc apply_or_revert_range_or_line {x y revert} { set failed_msg [mc "Failed to unstage selected line."] set to_context {+} lappend apply_cmd --reverse --cached - if {[string index $mi 0] ne {M}} { + set file_state [string index $mi 0] + if {$file_state ne {M} && $file_state ne {A}} { unlock_index return } @@ -688,7 +691,8 @@ proc apply_or_revert_range_or_line {x y revert} { lappend apply_cmd --cached } - if {[string index $mi 1] ne {M}} { + set file_state [string index $mi 1] + if {$file_state ne {M} && $file_state ne {A}} { unlock_index return } From e002ab3ef32067f81c672fd97e8af119fa0364d4 Mon Sep 17 00:00:00 2001 From: Jeff Hostetler Date: Mon, 5 Apr 2021 14:39:33 -0400 Subject: [PATCH 068/259] clink.pl: move default linker options for MSVC=1 builds Move the default `-ENTRY` and `-SUBSYSTEM` arguments for MSVC=1 builds from `config.mak.uname` into `clink.pl`. These args are constant for console-mode executables. Add support to `clink.pl` for generating a Win32 GUI application using the `-mwindows` argument (to match how GCC does it). This changes the `-ENTRY` and `-SUBSYSTEM` arguments accordingly. Signed-off-by: Jeff Hostetler --- compat/vcbuild/scripts/clink.pl | 11 +++++++++++ config.mak.uname | 2 +- 2 files changed, 12 insertions(+), 1 deletion(-) diff --git a/compat/vcbuild/scripts/clink.pl b/compat/vcbuild/scripts/clink.pl index 73c8a2b184f38b..a38b360015ece9 100755 --- a/compat/vcbuild/scripts/clink.pl +++ b/compat/vcbuild/scripts/clink.pl @@ -15,6 +15,7 @@ my @lflags = (); my $is_linking = 0; my $is_debug = 0; +my $is_gui = 0; while (@ARGV) { my $arg = shift @ARGV; if ("$arg" eq "-DDEBUG") { @@ -124,11 +125,21 @@ # let's ignore those } elsif ("$arg" eq "-fno-stack-protector") { # eat this + } elsif ("$arg" eq "-mwindows") { + $is_gui = 1; } else { push(@args, $arg); } } if ($is_linking) { + if ($is_gui) { + push(@args, "-ENTRY:wWinMainCRTStartup"); + push(@args, "-SUBSYSTEM:WINDOWS"); + } else { + push(@args, "-ENTRY:wmainCRTStartup"); + push(@args, "-SUBSYSTEM:CONSOLE"); + } + push(@args, @lflags); unshift(@args, "link.exe"); } else { diff --git a/config.mak.uname b/config.mak.uname index 8ef58660aa3ecf..8839dc511c426f 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -502,7 +502,7 @@ endif compat/win32/trace2_win32_process_info.o \ compat/win32/dirent.o COMPAT_CFLAGS = -D__USE_MINGW_ACCESS -DDETECT_MSYS_TTY -DENSURE_MSYSTEM_IS_SET -DNOGDI -DHAVE_STRING_H -Icompat -Icompat/regex -Icompat/win32 -DSTRIP_EXTENSION=\".exe\" - BASIC_LDFLAGS = -IGNORE:4217 -IGNORE:4049 -NOLOGO -ENTRY:wmainCRTStartup -SUBSYSTEM:CONSOLE + BASIC_LDFLAGS = -IGNORE:4217 -IGNORE:4049 -NOLOGO # invalidcontinue.obj allows Git's source code to close the same file # handle twice, or to access the osfhandle of an already-closed stdout # See https://msdn.microsoft.com/en-us/library/ms235330.aspx From 4c94afca19dd42e7e9f02823d0e3e6f0dd5b520d Mon Sep 17 00:00:00 2001 From: Yuyi Wang Date: Sat, 11 Mar 2023 17:51:18 +0800 Subject: [PATCH 069/259] cmake: install headless-git. headless-git is a git executable without opening a console window. It is useful when other GUI executables want to call git. We should install it together with git on Windows. Signed-off-by: Yuyi Wang --- contrib/buildsystems/CMakeLists.txt | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt index 25b495fa737d07..709f51120a683a 100644 --- a/contrib/buildsystems/CMakeLists.txt +++ b/contrib/buildsystems/CMakeLists.txt @@ -745,6 +745,7 @@ if(WIN32) endif() add_executable(headless-git ${CMAKE_SOURCE_DIR}/compat/win32/headless.c) + list(APPEND PROGRAMS_BUILT headless-git) if(CMAKE_C_COMPILER_ID STREQUAL "GNU" OR CMAKE_C_COMPILER_ID STREQUAL "Clang") target_link_options(headless-git PUBLIC -municode -Wl,-subsystem,windows) elseif(CMAKE_C_COMPILER_ID STREQUAL "MSVC") @@ -945,7 +946,7 @@ list(TRANSFORM git_perl_scripts PREPEND "${CMAKE_BINARY_DIR}/") #install foreach(program ${PROGRAMS_BUILT}) -if(program MATCHES "^(git|git-shell|scalar)$") +if(program MATCHES "^(git|git-shell|headless-git|scalar)$") install(TARGETS ${program} RUNTIME DESTINATION bin) else() From 14beae64d252904aac6ce3e719269198b972b888 Mon Sep 17 00:00:00 2001 From: Ian Bearman Date: Fri, 31 Jan 2020 16:00:25 -0800 Subject: [PATCH 070/259] vcbuild: install ARM64 dependencies when building ARM64 binaries Co-authored-by: Dennis Ameling Signed-off-by: Ian Bearman Signed-off-by: Dennis Ameling Signed-off-by: Johannes Schindelin --- compat/vcbuild/README | 6 +++++- compat/vcbuild/vcpkg_copy_dlls.bat | 7 ++++++- compat/vcbuild/vcpkg_install.bat | 9 +++++++-- 3 files changed, 18 insertions(+), 4 deletions(-) diff --git a/compat/vcbuild/README b/compat/vcbuild/README index 29ec1d0f104b80..1df1cabb1ebbbd 100644 --- a/compat/vcbuild/README +++ b/compat/vcbuild/README @@ -6,7 +6,11 @@ The Steps to Build Git with VS2015 or VS2017 from the command line. Prompt or from an SDK bash window: $ cd - $ ./compat/vcbuild/vcpkg_install.bat + $ ./compat/vcbuild/vcpkg_install.bat x64-windows + + or + + $ ./compat/vcbuild/vcpkg_install.bat arm64-windows The vcpkg tools and all of the third-party sources will be installed in this folder: diff --git a/compat/vcbuild/vcpkg_copy_dlls.bat b/compat/vcbuild/vcpkg_copy_dlls.bat index 13661c14f8705c..8bea0cbf83b6cf 100644 --- a/compat/vcbuild/vcpkg_copy_dlls.bat +++ b/compat/vcbuild/vcpkg_copy_dlls.bat @@ -15,7 +15,12 @@ REM ================================================================ @FOR /F "delims=" %%D IN ("%~dp0") DO @SET cwd=%%~fD cd %cwd% - SET arch=x64-windows + SET arch=%2 + IF NOT DEFINED arch ( + echo defaulting to 'x64-windows`. Invoke %0 with 'x86-windows', 'x64-windows', or 'arm64-windows' + set arch=x64-windows + ) + SET inst=%cwd%vcpkg\installed\%arch% IF [%1]==[release] ( diff --git a/compat/vcbuild/vcpkg_install.bat b/compat/vcbuild/vcpkg_install.bat index 8330d8120fb511..cacef18c11dc79 100644 --- a/compat/vcbuild/vcpkg_install.bat +++ b/compat/vcbuild/vcpkg_install.bat @@ -31,6 +31,12 @@ REM ================================================================ SETLOCAL EnableDelayedExpansion + SET arch=%1 + IF NOT DEFINED arch ( + echo defaulting to 'x64-windows`. Invoke %0 with 'x86-windows', 'x64-windows', or 'arm64-windows' + set arch=x64-windows + ) + @FOR /F "delims=" %%D IN ("%~dp0") DO @SET cwd=%%~fD cd %cwd% @@ -55,9 +61,8 @@ REM ================================================================ echo Successfully installed %cwd%vcpkg\vcpkg.exe :install_libraries - SET arch=x64-windows - echo Installing third-party libraries... + echo Installing third-party libraries(%arch%)... FOR %%i IN (zlib expat libiconv openssl libssh2 curl) DO ( cd %cwd%vcpkg IF NOT EXIST "packages\%%i_%arch%" CALL :sub__install_one %%i From 1d88f5f8826a8a7b908a3c4271d37f7f4e25216e Mon Sep 17 00:00:00 2001 From: Ian Bearman Date: Tue, 4 Feb 2020 10:34:40 -0800 Subject: [PATCH 071/259] vcbuild: add an option to install individual 'features' In this context, a "feature" is a dependency combined with its own dependencies. Signed-off-by: Ian Bearman Signed-off-by: Johannes Schindelin --- compat/vcbuild/vcpkg_install.bat | 35 +++++++++++++++++++++++++++++++- 1 file changed, 34 insertions(+), 1 deletion(-) diff --git a/compat/vcbuild/vcpkg_install.bat b/compat/vcbuild/vcpkg_install.bat index cacef18c11dc79..8da212487ae97d 100644 --- a/compat/vcbuild/vcpkg_install.bat +++ b/compat/vcbuild/vcpkg_install.bat @@ -85,14 +85,47 @@ REM ================================================================ :sub__install_one echo Installing package %1... + call :%1_features + REM vcpkg may not be reliable on slow, intermittent or proxy REM connections, see e.g. REM https://social.msdn.microsoft.com/Forums/windowsdesktop/en-US/4a8f7be5-5e15-4213-a7bb-ddf424a954e6/winhttpsendrequest-ends-with-12002-errorhttptimeout-after-21-seconds-no-matter-what-timeout?forum=windowssdk REM which explains the hidden 21 second timeout REM (last post by Dave : Microsoft - Windows Networking team) - .\vcpkg.exe install %1:%arch% + .\vcpkg.exe install %1%features%:%arch% IF ERRORLEVEL 1 ( EXIT /B 1 ) echo Finished %1 goto :EOF + +:: +:: features for each vcpkg to install +:: there should be an entry here for each package to install +:: 'set features=' means use the default otherwise +:: 'set features=[comma-delimited-feature-set]' is the syntax +:: + +:zlib_features +set features= +goto :EOF + +:expat_features +set features= +goto :EOF + +:libiconv_features +set features= +goto :EOF + +:openssl_features +set features= +goto :EOF + +:libssh2_features +set features= +goto :EOF + +:curl_features +set features=[core,openssl] +goto :EOF From 33f017b7e45d6aa69f22fab50c2e61046529365a Mon Sep 17 00:00:00 2001 From: Dennis Ameling Date: Fri, 4 Dec 2020 14:11:34 +0100 Subject: [PATCH 072/259] cmake: allow building for Windows/ARM64 Signed-off-by: Dennis Ameling Signed-off-by: Johannes Schindelin --- contrib/buildsystems/CMakeLists.txt | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt index 25b495fa737d07..aaaee1bc4b7943 100644 --- a/contrib/buildsystems/CMakeLists.txt +++ b/contrib/buildsystems/CMakeLists.txt @@ -65,9 +65,9 @@ if(USE_VCPKG) set(VCPKG_DIR "${CMAKE_SOURCE_DIR}/compat/vcbuild/vcpkg") if(NOT EXISTS ${VCPKG_DIR}) message("Initializing vcpkg and building the Git's dependencies (this will take a while...)") - execute_process(COMMAND ${CMAKE_SOURCE_DIR}/compat/vcbuild/vcpkg_install.bat) + execute_process(COMMAND ${CMAKE_SOURCE_DIR}/compat/vcbuild/vcpkg_install.bat ${VCPKG_ARCH}) endif() - list(APPEND CMAKE_PREFIX_PATH "${VCPKG_DIR}/installed/x64-windows") + list(APPEND CMAKE_PREFIX_PATH "${VCPKG_DIR}/installed/${VCPKG_ARCH}") # In the vcpkg edition, we need this to be able to link to libcurl set(CURL_NO_CURL_CMAKE ON) @@ -1211,7 +1211,7 @@ string(REPLACE "@USE_LIBPCRE2@" "" git_build_options "${git_build_options}") string(REPLACE "@WITH_BREAKING_CHANGES@" "" git_build_options "${git_build_options}") string(REPLACE "@X@" "${EXE_EXTENSION}" git_build_options "${git_build_options}") if(USE_VCPKG) - string(APPEND git_build_options "PATH=\"$PATH:$TEST_DIRECTORY/../compat/vcbuild/vcpkg/installed/x64-windows/bin\"\n") + string(APPEND git_build_options "PATH=\"$PATH:$TEST_DIRECTORY/../compat/vcbuild/vcpkg/installed/${VCPKG_ARCH}/bin\"\n") endif() file(WRITE ${CMAKE_BINARY_DIR}/GIT-BUILD-OPTIONS ${git_build_options}) From cb368c9b77f652ae1c343ce0818a5e73d3ebbb46 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Sun, 6 Oct 2019 18:40:55 +0100 Subject: [PATCH 073/259] vcpkg_install: detect lack of Git The vcpkg_install batch file depends on the availability of a working Git on the CMD path. This may not be present if the user has selected the 'bash only' option during Git-for-Windows install. Detect and tell the user about their lack of a working Git in the CMD window. Fixes #2348. A separate PR https://github.com/git-for-windows/build-extra/pull/258 now highlights the recommended path setting during install. Signed-off-by: Philip Oakley --- compat/vcbuild/vcpkg_install.bat | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/compat/vcbuild/vcpkg_install.bat b/compat/vcbuild/vcpkg_install.bat index ebd0bad242a8ca..bcbbf536af3141 100644 --- a/compat/vcbuild/vcpkg_install.bat +++ b/compat/vcbuild/vcpkg_install.bat @@ -36,6 +36,13 @@ REM ================================================================ dir vcpkg\vcpkg.exe >nul 2>nul && GOTO :install_libraries + git.exe version 2>nul + IF ERRORLEVEL 1 ( + echo "***" + echo "Git not found. Please adjust your CMD path or Git install option." + echo "***" + EXIT /B 1 ) + echo Fetching vcpkg in %cwd%vcpkg git.exe clone https://github.com/Microsoft/vcpkg vcpkg IF ERRORLEVEL 1 ( EXIT /B 1 ) From df54c4aeac5145c2044050f63c358e507cbb80c9 Mon Sep 17 00:00:00 2001 From: Dennis Ameling Date: Sun, 29 Nov 2020 00:12:26 +0100 Subject: [PATCH 074/259] ci(vs-build) also build Windows/ARM64 artifacts There are no Windows/ARM64 agents in GitHub Actions yet, therefore we just skip adjusting the `vs-test` job for now. Signed-off-by: Dennis Ameling Signed-off-by: Johannes Schindelin --- .github/workflows/main.yml | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml index 9959b61ece2bcc..6ee3c5b42b85c4 100644 --- a/.github/workflows/main.yml +++ b/.github/workflows/main.yml @@ -169,8 +169,11 @@ jobs: NO_PERL: 1 GIT_CONFIG_PARAMETERS: "'user.name=CI' 'user.email=ci@git'" runs-on: windows-latest + strategy: + matrix: + arch: [x64, arm64] concurrency: - group: vs-build-${{ github.ref }} + group: vs-build-${{ github.ref }}-${{ matrix.arch }} cancel-in-progress: ${{ needs.ci-config.outputs.skip_concurrent == 'yes' }} steps: - uses: actions/checkout@v4 @@ -189,14 +192,14 @@ jobs: uses: microsoft/setup-msbuild@v2 - name: copy dlls to root shell: cmd - run: compat\vcbuild\vcpkg_copy_dlls.bat release + run: compat\vcbuild\vcpkg_copy_dlls.bat release ${{ matrix.arch }}-windows - name: generate Visual Studio solution shell: bash run: | - cmake `pwd`/contrib/buildsystems/ -DCMAKE_PREFIX_PATH=`pwd`/compat/vcbuild/vcpkg/installed/x64-windows \ - -DNO_GETTEXT=YesPlease -DPERL_TESTS=OFF -DPYTHON_TESTS=OFF -DCURL_NO_CURL_CMAKE=ON + cmake `pwd`/contrib/buildsystems/ -DCMAKE_PREFIX_PATH=`pwd`/compat/vcbuild/vcpkg/installed/${{ matrix.arch }}-windows \ + -DNO_GETTEXT=YesPlease -DPERL_TESTS=OFF -DPYTHON_TESTS=OFF -DCURL_NO_CURL_CMAKE=ON -DCMAKE_GENERATOR_PLATFORM=${{ matrix.arch }} -DVCPKG_ARCH=${{ matrix.arch }}-windows - name: MSBuild - run: msbuild git.sln -property:Configuration=Release -property:Platform=x64 -maxCpuCount:4 -property:PlatformToolset=v142 + run: msbuild git.sln -property:Configuration=Release -property:Platform=${{ matrix.arch }} -maxCpuCount:4 -property:PlatformToolset=v142 - name: bundle artifact tar shell: bash env: @@ -210,7 +213,7 @@ jobs: - name: upload tracked files and build artifacts uses: actions/upload-artifact@v4 with: - name: vs-artifacts + name: vs-artifacts-${{ matrix.arch }} path: artifacts vs-test: name: win+VS test @@ -228,7 +231,7 @@ jobs: - name: download tracked files and build artifacts uses: actions/download-artifact@v4 with: - name: vs-artifacts + name: vs-artifacts-x64 path: ${{github.workspace}} - name: extract tracked files and build artifacts shell: bash From cd6fa378c5b18828aea5e45edc52e497f7257ef3 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Sun, 6 Oct 2019 18:43:57 +0100 Subject: [PATCH 075/259] vcpkg_install: add comment regarding slow network connections The vcpkg downloads may not succeed. Warn careful readers of the time out. A simple retry will usually resolve the issue. Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- compat/vcbuild/vcpkg_install.bat | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/compat/vcbuild/vcpkg_install.bat b/compat/vcbuild/vcpkg_install.bat index bcbbf536af3141..8330d8120fb511 100644 --- a/compat/vcbuild/vcpkg_install.bat +++ b/compat/vcbuild/vcpkg_install.bat @@ -80,6 +80,12 @@ REM ================================================================ :sub__install_one echo Installing package %1... + REM vcpkg may not be reliable on slow, intermittent or proxy + REM connections, see e.g. + REM https://social.msdn.microsoft.com/Forums/windowsdesktop/en-US/4a8f7be5-5e15-4213-a7bb-ddf424a954e6/winhttpsendrequest-ends-with-12002-errorhttptimeout-after-21-seconds-no-matter-what-timeout?forum=windowssdk + REM which explains the hidden 21 second timeout + REM (last post by Dave : Microsoft - Windows Networking team) + .\vcpkg.exe install %1:%arch% IF ERRORLEVEL 1 ( EXIT /B 1 ) From 9aeff06555a08d2a4521a13b7908f10e2f42cf2d Mon Sep 17 00:00:00 2001 From: Dennis Ameling Date: Sun, 6 Dec 2020 18:39:26 +0100 Subject: [PATCH 076/259] Add schannel to curl installation Signed-off-by: Dennis Ameling --- compat/vcbuild/vcpkg_install.bat | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/compat/vcbuild/vcpkg_install.bat b/compat/vcbuild/vcpkg_install.bat index 8da212487ae97d..575c65c20ba307 100644 --- a/compat/vcbuild/vcpkg_install.bat +++ b/compat/vcbuild/vcpkg_install.bat @@ -127,5 +127,5 @@ set features= goto :EOF :curl_features -set features=[core,openssl] +set features=[core,openssl,schannel] goto :EOF From 1cbc98ed4b884f826e2e8488caa2e851f460e942 Mon Sep 17 00:00:00 2001 From: Dennis Ameling Date: Mon, 19 Jul 2021 13:02:16 +0200 Subject: [PATCH 077/259] cmake(): allow setting HOST_CPU for cross-compilation Git's regular Makefile mentions that HOST_CPU should be defined when cross-compiling Git: https://github.com/git-for-windows/git/blob/37796bca76ef4180c39ee508ca3e42c0777ba444/Makefile#L438-L439 This is then used to set the GIT_HOST_CPU variable when compiling Git: https://github.com/git-for-windows/git/blob/37796bca76ef4180c39ee508ca3e42c0777ba444/Makefile#L1337-L1341 Then, when the user runs `git version --build-options`, it returns that value: https://github.com/git-for-windows/git/blob/37796bca76ef4180c39ee508ca3e42c0777ba444/help.c#L658 This commit adds the same functionality to the CMake configuration. Users can now set -DHOST_CPU= to set the target architecture. Signed-off-by: Dennis Ameling --- .github/workflows/main.yml | 2 +- contrib/buildsystems/CMakeLists.txt | 9 ++++++++- 2 files changed, 9 insertions(+), 2 deletions(-) diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml index 6ee3c5b42b85c4..4073c8b076eb24 100644 --- a/.github/workflows/main.yml +++ b/.github/workflows/main.yml @@ -197,7 +197,7 @@ jobs: shell: bash run: | cmake `pwd`/contrib/buildsystems/ -DCMAKE_PREFIX_PATH=`pwd`/compat/vcbuild/vcpkg/installed/${{ matrix.arch }}-windows \ - -DNO_GETTEXT=YesPlease -DPERL_TESTS=OFF -DPYTHON_TESTS=OFF -DCURL_NO_CURL_CMAKE=ON -DCMAKE_GENERATOR_PLATFORM=${{ matrix.arch }} -DVCPKG_ARCH=${{ matrix.arch }}-windows + -DNO_GETTEXT=YesPlease -DPERL_TESTS=OFF -DPYTHON_TESTS=OFF -DCURL_NO_CURL_CMAKE=ON -DCMAKE_GENERATOR_PLATFORM=${{ matrix.arch }} -DVCPKG_ARCH=${{ matrix.arch }}-windows -DHOST_CPU=${{ matrix.arch }} - name: MSBuild run: msbuild git.sln -property:Configuration=Release -property:Platform=${{ matrix.arch }} -maxCpuCount:4 -property:PlatformToolset=v142 - name: bundle artifact tar diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt index aaaee1bc4b7943..d61e54a7f6dd4e 100644 --- a/contrib/buildsystems/CMakeLists.txt +++ b/contrib/buildsystems/CMakeLists.txt @@ -212,7 +212,14 @@ endif() #default behaviour include_directories(${CMAKE_SOURCE_DIR}) -add_compile_definitions(GIT_HOST_CPU="${CMAKE_SYSTEM_PROCESSOR}") + +# When cross-compiling, define HOST_CPU as the canonical name of the CPU on +# which the built Git will run (for instance "x86_64"). +if(NOT HOST_CPU) + add_compile_definitions(GIT_HOST_CPU="${CMAKE_SYSTEM_PROCESSOR}") +else() + add_compile_definitions(GIT_HOST_CPU="${HOST_CPU}") +endif() add_compile_definitions(SHA256_BLK INTERNAL_QSORT RUNTIME_PREFIX) add_compile_definitions(NO_OPENSSL SHA1_DC SHA1DC_NO_STANDARD_INCLUDES SHA1DC_INIT_SAFE_HASH_DEFAULT=0 From 165c32d28bffaba29e6997806221f1fb6c7c8263 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Fri, 2 Jul 2021 00:30:24 +0100 Subject: [PATCH 078/259] CMake: default Visual Studio generator has changed Correct some wording and inform users regarding the Visual Studio changes (from V16.6) to the default generator. Subsequent commits ensure that Git for Windows can be directly opened in modern Visual Studio without needing special configuration of the CMakeLists settings. It appeares that internally Visual Studio creates it's own version of the .sln file (etc.) for extension tools that expect them. The large number of references below document the shifting of Visual Studio default and CMake setting options. refs: https://docs.microsoft.com/en-us/search/?scope=C%2B%2B&view=msvc-150&terms=Ninja 1. https://docs.microsoft.com/en-us/cpp/linux/cmake-linux-configure?view=msvc-160 (note the linux bit) "In Visual Studio 2019 version 16.6 or later ***, Ninja is the default generator for configurations targeting a remote system or WSL. For more information, see this post on the C++ Team Blog [https://devblogs.microsoft.com/cppblog/linux-development-with-visual-studio-first-class-support-for-gdbserver-improved-build-times-with-ninja-and-updates-to-the-connection-manager/]. For more information about these settings, see CMakeSettings.json reference [https://docs.microsoft.com/en-us/cpp/build/cmakesettings-reference?view=msvc-160]." 2. https://docs.microsoft.com/en-us/cpp/build/cmake-presets-vs?view=msvc-160 "CMake supports two files that allow users to specify common configure, build, and test options and share them with others: CMakePresets.json and CMakeUserPresets.json." " Both files are supported in Visual Studio 2019 version 16.10 or later. ***" 3. https://devblogs.microsoft.com/cppblog/linux-development-with-visual-studio-first-class-support-for-gdbserver-improved-build-times-with-ninja-and-updates-to-the-connection-manager/ " Ninja has been the default generator (underlying build system) for CMake configurations targeting Windows for some time***, but in Visual Studio 2019 version 16.6 Preview 3*** we added support for Ninja on Linux." 4. https://docs.microsoft.com/en-us/cpp/build/cmakesettings-reference?view=msvc-160 " `generator`: specifies CMake generator to use for this configuration. May be one of: Visual Studio 2019 only: Visual Studio 16 2019 Visual Studio 16 2019 Win64 Visual Studio 16 2019 ARM Visual Studio 2017 and later: Visual Studio 15 2017 Visual Studio 15 2017 Win64 Visual Studio 15 2017 ARM Visual Studio 14 2015 Visual Studio 14 2015 Win64 Visual Studio 14 2015 ARM Unix Makefiles Ninja Because Ninja is designed for fast build speeds instead of flexibility and function, it is set as the default. However, some CMake projects may be unable to correctly build using Ninja. If this occurs, you can instruct CMake to generate Visual Studio projects instead. To specify a Visual Studio generator in Visual Studio 2017, open the settings editor from the main menu by choosing CMake | Change CMake Settings. Delete "Ninja" and type "V". This activates IntelliSense, which enables you to choose the generator you want." "To specify a Visual Studio generator in Visual Studio 2019, right-click on the CMakeLists.txt file in Solution Explorer and choose CMake Settings for project > Show Advanced Settings > CMake Generator. When the active configuration specifies a Visual Studio generator, by default MSBuild.exe is invoked with` -m -v:minimal` arguments." 5. https://docs.microsoft.com/en-us/cpp/build/cmake-presets-vs?view=msvc-160#enable-cmakepresetsjson-integration-in-visual-studio-2019 "Enable CMakePresets.json integration in Visual Studio 2019 CMakePresets.json integration isn't enabled by default in Visual Studio 2019. You can enable it for all CMake projects in Tools > Options > CMake > General: (tick a box)" ... see more. 6. https://docs.microsoft.com/en-us/cpp/build/cmakesettings-reference?view=msvc-140 (whichever v140 is..) "CMake projects are supported in Visual Studio 2017 and later." 7. https://docs.microsoft.com/en-us/cpp/overview/what-s-new-for-cpp-2017?view=msvc-150 "Support added for the CMake Ninja generator." 8. https://docs.microsoft.com/en-us/cpp/overview/what-s-new-for-cpp-2017?view=msvc-150#cmake-support-via-open-folder "CMake support via Open Folder Visual Studio 2017 introduces support for using CMake projects without converting to MSBuild project files (.vcxproj). For more information, see CMake projects in Visual Studio[https://docs.microsoft.com/en-us/cpp/build/cmake-projects-in-visual-studio?view=msvc-150]. Opening CMake projects with Open Folder automatically configures the environment for C++ editing, building, and debugging." ... +more! 9. https://docs.microsoft.com/en-us/cpp/build/cmake-presets-vs?view=msvc-160#supported-cmake-and-cmakepresetsjson-versions "Visual Studio reads and evaluates CMakePresets.json and CMakeUserPresets.json itself and doesn't invoke CMake directly with the --preset option. So, CMake version 3.20 or later isn't strictly required when you're building with CMakePresets.json inside Visual Studio. We recommend using CMake version 3.14 or later." 10. https://docs.microsoft.com/en-us/cpp/build/cmake-presets-vs?view=msvc-160#enable-cmakepresetsjson-integration-in-visual-studio-2019 "If you don't want to enable CMakePresets.json integration for all CMake projects, you can enable CMakePresets.json integration for a single CMake project by adding a CMakePresets.json file to the root of the open folder. You must close and reopen the folder in Visual Studio to activate the integration. 11. https://docs.microsoft.com/en-us/cpp/build/cmake-presets-vs?view=msvc-160#default-configure-presets ***(doesn't actually say which version..) "Default Configure Presets If no CMakePresets.json or CMakeUserPresets.json file exists, or if CMakePresets.json or CMakeUserPresets.json is invalid, Visual Studio will fall back*** on the following default Configure Presets: Windows example JSON { "name": "windows-default", "displayName": "Windows x64 Debug", "description": "Sets Ninja generator, compilers, x64 architecture, build and install directory, debug build type", "generator": "Ninja", "binaryDir": "${sourceDir}/out/build/${presetName}", "architecture": { "value": "x64", "strategy": "external" }, "cacheVariables": { "CMAKE_BUILD_TYPE": "Debug", "CMAKE_INSTALL_PREFIX": "${sourceDir}/out/install/${presetName}" }, "vendor": { "microsoft.com/VisualStudioSettings/CMake/1.0": { "hostOS": [ "Windows" ] } } }, " Signed-off-by: Philip Oakley --- contrib/buildsystems/CMakeLists.txt | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt index d61e54a7f6dd4e..30fa3250c75bf1 100644 --- a/contrib/buildsystems/CMakeLists.txt +++ b/contrib/buildsystems/CMakeLists.txt @@ -14,6 +14,11 @@ Note: Visual Studio also has the option of opening `CMakeLists.txt` directly; Using this option, Visual Studio will not find the source code, though, therefore the `File>Open>Folder...` option is preferred. +Visual Studio does not produce a .sln solution file nor the .vcxproj files +that may be required by VS extension tools. + +To generate the .sln/.vcxproj files run CMake manually, as described below. + Instructions to run CMake manually: mkdir -p contrib/buildsystems/out @@ -22,7 +27,7 @@ Instructions to run CMake manually: This will build the git binaries in contrib/buildsystems/out directory (our top-level .gitignore file knows to ignore contents of -this directory). +this directory). The project .sln and .vcxproj files are also generated. Possible build configurations(-DCMAKE_BUILD_TYPE) with corresponding compiler flags @@ -35,17 +40,16 @@ empty(default) : NOTE: -DCMAKE_BUILD_TYPE is optional. For multi-config generators like Visual Studio this option is ignored -This process generates a Makefile(Linux/*BSD/MacOS) , Visual Studio solution(Windows) by default. +This process generates a Makefile(Linux/*BSD/MacOS), Visual Studio solution(Windows) by default. Run `make` to build Git on Linux/*BSD/MacOS. Open git.sln on Windows and build Git. -NOTE: By default CMake uses Makefile as the build tool on Linux and Visual Studio in Windows, -to use another tool say `ninja` add this to the command line when configuring. -`-G Ninja` - NOTE: By default CMake will install vcpkg locally to your source tree on configuration, to avoid this, add `-DNO_VCPKG=TRUE` to the command line when configuring. +The Visual Studio default generator changed in v16.6 from its Visual Studio +implemenation to `Ninja` This required changes to many CMake scripts. + ]] cmake_minimum_required(VERSION 3.14) From c59ac5b23799f58f546c43351c544187e7e9bdaa Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Sat, 24 Apr 2021 11:09:58 +0100 Subject: [PATCH 079/259] .gitignore: add Visual Studio CMakeSetting.json file The CMakeSettings.json file is tool generated. Developers may track it should they provide additional settings. Signed-off-by: Philip Oakley --- .gitignore | 1 + 1 file changed, 1 insertion(+) diff --git a/.gitignore b/.gitignore index 08a66ca50844c5..b633ef317214b8 100644 --- a/.gitignore +++ b/.gitignore @@ -251,5 +251,6 @@ Release/ /git.VC.db *.dSYM /contrib/buildsystems/out +CMakeSettings.json /contrib/libgit-rs/target /contrib/libgit-sys/target From 9080f212b903b4075c8fd3dedfb8d95408e3d702 Mon Sep 17 00:00:00 2001 From: Victoria Dye Date: Thu, 5 Aug 2021 19:04:13 -0400 Subject: [PATCH 080/259] subtree: update `contrib/subtree` `test` target The intention of this change is to align with how the top-level git `Makefile` defines its own test target (which also internally calls `$(MAKE) -C t/ all`). This change also ensures the consistency of `make -C contrib/subtree test` with other testing in CI executions (which rely on `$DEFAULT_TEST_TARGET` being defined as `prove`). Signed-off-by: Victoria Dye --- contrib/subtree/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/contrib/subtree/Makefile b/contrib/subtree/Makefile index c0c9f21cb78022..dab2dfc08ee222 100644 --- a/contrib/subtree/Makefile +++ b/contrib/subtree/Makefile @@ -95,7 +95,7 @@ $(GIT_SUBTREE_TEST): $(GIT_SUBTREE) cp $< $@ test: $(GIT_SUBTREE_TEST) - $(MAKE) -C t/ test + $(MAKE) -C t/ all clean: $(RM) $(GIT_SUBTREE) From f1a9157e20bf6550a9ee6177c2fed9f220d31aa9 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Thu, 22 Apr 2021 11:11:38 +0100 Subject: [PATCH 081/259] CMakeLists: add default "x64-windows" arch for Visual Studio In Git-for-Windows, work on using ARM64 has progressed. The commit 2d94b77b27 (cmake: allow building for Windows/ARM64, 2020-12-04) failed to notice that /compat/vcbuild/vcpkg_install.bat will default to using the "x64-windows" architecture for the vcpkg installation if not set, but CMake is not told of this default. Commit 635b6d99b3 (vcbuild: install ARM64 dependencies when building ARM64 binaries, 2020-01-31) later updated vcpkg_install.bat to accept an arch (%1) parameter, but retained the default. This default is neccessary for the use case where the project directory is opened directly in Visual Studio, which will find and build a CMakeLists.txt file without any parameters, thus expecting use of the default setting. Also Visual studio will generate internal .sln solution and .vcxproj project files needed for some extension tools. Inform users of the additional .sln/.vcxproj generation. ** How to test: rm -rf '.vs' # remove old visual studio settings rm -rf 'compat/vcbuild/vcpkg' # remove any vcpkg downloads rm -rf 'contrib/buildsystems/out' # remove builds & CMake artifacts with a fresh Visual Studio Community Edition, File>>Open>>(git *folder*) to load the project (which will take some time!). check for successful compilation. The implicit .sln (etc.) are in the hidden .vs directory created by Visual Studio. Signed-off-by: Philip Oakley --- contrib/buildsystems/CMakeLists.txt | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt index 30fa3250c75bf1..d21e22835643d0 100644 --- a/contrib/buildsystems/CMakeLists.txt +++ b/contrib/buildsystems/CMakeLists.txt @@ -71,6 +71,10 @@ if(USE_VCPKG) message("Initializing vcpkg and building the Git's dependencies (this will take a while...)") execute_process(COMMAND ${CMAKE_SOURCE_DIR}/compat/vcbuild/vcpkg_install.bat ${VCPKG_ARCH}) endif() + if(NOT EXISTS ${VCPKG_ARCH}) + message("VCPKG_ARCH: unset, using 'x64-windows'") + set(VCPKG_ARCH "x64-windows") # default from vcpkg_install.bat + endif() list(APPEND CMAKE_PREFIX_PATH "${VCPKG_DIR}/installed/${VCPKG_ARCH}") # In the vcpkg edition, we need this to be able to link to libcurl From 765ae0ee984c45ffe7e9f84af47c56e76f6c6652 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 2 Apr 2021 22:50:54 +0200 Subject: [PATCH 082/259] mingw: allow for longer paths in `parse_interpreter()` MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit As reported in https://github.com/newren/git-filter-repo/pull/225, it looks like 99 bytes is not really sufficient to represent e.g. the full path to Python when installed via Windows Store (and this path is used in the hasb bang line when installing scripts via `pip`). Let's increase it to what is probably the maximum sensible path size: MAX_PATH. This makes `parse_interpreter()` in line with what `lookup_prog()` handles. Signed-off-by: Johannes Schindelin Signed-off-by: Vilius Šumskas --- compat/mingw.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..f65d8e2bc0a576 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -1339,7 +1339,7 @@ static const char *quote_arg_msys2(const char *arg) static const char *parse_interpreter(const char *cmd) { - static char buf[100]; + static char buf[MAX_PATH]; char *p, *opt; ssize_t n; /* read() can return negative values */ int fd; From 898e9a5319064a71e1644d05a87e4c75a182ff9b Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 17 May 2021 10:46:52 +0200 Subject: [PATCH 083/259] compat/vcbuild: document preferred way to build in Visual Studio We used to have that `make vcxproj` hack, but a hack it is. In the meantime, we have a much cleaner solution: using CMake, either explicitly, or even more conveniently via Visual Studio's built-in CMake support (simply open Git's top-level directory via File>Open>Folder...). Let's let the `README` reflect this. Signed-off-by: Johannes Schindelin --- compat/vcbuild/README | 28 +++++++++------------------- 1 file changed, 9 insertions(+), 19 deletions(-) diff --git a/compat/vcbuild/README b/compat/vcbuild/README index 29ec1d0f104b80..5c71ea2daa4017 100644 --- a/compat/vcbuild/README +++ b/compat/vcbuild/README @@ -37,27 +37,17 @@ The Steps to Build Git with VS2015 or VS2017 from the command line. ================================================================ -Alternatively, run `make vcxproj` and then load the generated `git.sln` in -Visual Studio. The initial build will install the vcpkg system and build the +Alternatively, just open Git's top-level directory in Visual Studio, via +`File>Open>Folder...`. This will use CMake internally to generate the +project definitions. It will also install the vcpkg system and build the dependencies automatically. This will take a while. -Instead of generating the `git.sln` file yourself (which requires a full Git -for Windows SDK), you may want to consider fetching the `vs/master` branch of -https://github.com/git-for-windows/git instead (which is updated automatically -via CI running `make vcxproj`). The `vs/master` branch does not require a Git -for Windows to build, but you can run the test scripts in a regular Git Bash. - -Note that `make vcxproj` will automatically add and commit the generated `.sln` -and `.vcxproj` files to the repo. This is necessary to allow building a -fully-testable Git in Visual Studio, where a regular Git Bash can be used to -run the test scripts (as opposed to a full Git for Windows SDK): a number of -build targets, such as Git commands implemented as Unix shell scripts (where -`@@SHELL_PATH@@` and other placeholders are interpolated) require a full-blown -Git for Windows SDK (which is about 10x the size of a regular Git for Windows -installation). - -If your plan is to open a Pull Request with Git for Windows, it is a good idea -to drop this commit before submitting. +You can also generate the Visual Studio solution manually by downloading +and running CMake explicitly rather than letting Visual Studio doing +that implicitly. + +Another, deprecated option is to run `make vcxproj`. This option is +superseded by the CMake-based build, and will be removed at some point. ================================================================ The Steps of Build Git with VS2008 From e73268d3eda585761c81127bfe8f29de94da2108 Mon Sep 17 00:00:00 2001 From: Pascal Muller Date: Wed, 23 Jun 2021 21:21:10 +0200 Subject: [PATCH 084/259] http: optionally send SSL client certificate This adds support for a new http.sslAutoClientCert config value. In cURL 7.77 or later the schannel backend does not automatically send client certificates from the Windows Certificate Store anymore. This config value is only used if http.sslBackend is set to "schannel", and can be used to opt in to the old behavior and force cURL to send client certificates. This fixes https://github.com/git-for-windows/git/issues/3292 Signed-off-by: Pascal Muller --- Documentation/config/http.adoc | 5 +++++ git-curl-compat.h | 8 ++++++++ http.c | 24 +++++++++++++++++++++--- 3 files changed, 34 insertions(+), 3 deletions(-) diff --git a/Documentation/config/http.adoc b/Documentation/config/http.adoc index de52a1c44fbf14..484a6b574473e7 100644 --- a/Documentation/config/http.adoc +++ b/Documentation/config/http.adoc @@ -249,6 +249,11 @@ http.schannelUseSSLCAInfo:: when the `schannel` backend was configured via `http.sslBackend`, unless `http.schannelUseSSLCAInfo` overrides this behavior. +http.sslAutoClientCert:: + As of cURL v7.77.0, the Secure Channel backend won't automatically + send client certificates from the Windows Certificate Store anymore. + To opt in to the old behavior, http.sslAutoClientCert can be set. + http.pinnedPubkey:: Public key of the https service. It may either be the filename of a PEM or DER encoded public key file or a string starting with diff --git a/git-curl-compat.h b/git-curl-compat.h index 703756ba851d5b..8291755f50454f 100644 --- a/git-curl-compat.h +++ b/git-curl-compat.h @@ -45,4 +45,12 @@ #define GIT_CURL_HAVE_CURLOPT_PROTOCOLS_STR 1 #endif +/** + * CURLSSLOPT_AUTO_CLIENT_CERT was added in 7.77.0, released in May + * 2021. + */ +#if LIBCURL_VERSION_NUM >= 0x074d00 +#define GIT_CURL_HAVE_CURLSSLOPT_AUTO_CLIENT_CERT +#endif + #endif diff --git a/http.c b/http.c index 657c3e1dc5e2c2..d5396a3ce2ba34 100644 --- a/http.c +++ b/http.c @@ -157,6 +157,8 @@ static int http_schannel_check_revoke_mode = */ static int http_schannel_use_ssl_cainfo; +static int http_auto_client_cert; + static int always_auth_proactively(void) { return http_proactive_auth != PROACTIVE_AUTH_NONE && @@ -445,6 +447,11 @@ static int http_options(const char *var, const char *value, return 0; } + if (!strcmp("http.sslautoclientcert", var)) { + http_auto_client_cert = git_config_bool(var, value); + return 0; + } + if (!strcmp("http.minsessions", var)) { min_curl_sessions = git_config_int(var, value, ctx->kvi); if (min_curl_sessions > 1) @@ -1061,9 +1068,20 @@ static CURL *get_curl_handle(void) } #endif - if (http_ssl_backend && !strcmp("schannel", http_ssl_backend) && - http_schannel_check_revoke_mode) { - curl_easy_setopt(result, CURLOPT_SSL_OPTIONS, http_schannel_check_revoke_mode); + if (http_ssl_backend && !strcmp("schannel", http_ssl_backend)) { + long ssl_options = 0; + if (http_schannel_check_revoke_mode) { + ssl_options |= http_schannel_check_revoke_mode; + } + + if (http_auto_client_cert) { +#ifdef GIT_CURL_HAVE_CURLSSLOPT_AUTO_CLIENT_CERT + ssl_options |= CURLSSLOPT_AUTO_CLIENT_CERT; +#endif + } + + if (ssl_options) + curl_easy_setopt(result, CURLOPT_SSL_OPTIONS, ssl_options); } if (http_proactive_auth != PROACTIVE_AUTH_NONE) From b45aa3f194423c1f550bcecf46e0b037ac78e232 Mon Sep 17 00:00:00 2001 From: Victoria Dye Date: Thu, 5 Aug 2021 19:11:59 -0400 Subject: [PATCH 085/259] ci: run `contrib/subtree` tests in CI builds Because `git subtree` (unlike most other `contrib` modules) is included as part of the standard release of Git for Windows, its stability should be verified as consistently as it is for the rest of git. By including the `git subtree` tests in the CI workflow, these tests are as much of a gate to merging and indicator of stability as the standard test suite. Signed-off-by: Victoria Dye --- ci/run-build-and-tests.sh | 4 ++++ ci/run-test-slice.sh | 3 +++ 2 files changed, 7 insertions(+) diff --git a/ci/run-build-and-tests.sh b/ci/run-build-and-tests.sh index 1c698467235943..a7b7f2b846a579 100755 --- a/ci/run-build-and-tests.sh +++ b/ci/run-build-and-tests.sh @@ -77,5 +77,9 @@ case "$jobname" in ;; esac +case " $MAKE_TARGETS " in +*" all "*) make -C contrib/subtree test;; +esac + check_unignored_build_artifacts save_good_tree diff --git a/ci/run-test-slice.sh b/ci/run-test-slice.sh index e167e646f79e3d..8cb0038d07ca64 100755 --- a/ci/run-test-slice.sh +++ b/ci/run-test-slice.sh @@ -20,4 +20,7 @@ if [ "$1" == "0" ] ; then group "Run unit tests" make --quiet -C t unit-tests-test-tool fi +# Run the git subtree tests only if main tests succeeded +test 0 != "$1" || make -C contrib/subtree test + check_unignored_build_artifacts From 5b5124910c89a6c936865875b43b6bf42170155a Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Mon, 10 May 2021 16:47:40 +0100 Subject: [PATCH 086/259] CMake: show Win32 and Generator_platform build-option values Ensure key CMake option values are part of the CMake output to facilitate user support when tool updates impact the wider CMake actions, particularly ongoing 'improvements' in Visual Studio. These CMake displays perform the same function as the build-options.txt provided in the main Git for Windows. CMake is already chatty. The setting of CMAKE_EXPORT_COMPILE_COMMANDS is also reported. Include the environment's CMAKE_EXPORT_COMPILE_COMMANDS value which may have been propogated to CMake's internal value. Testing the CMAKE_EXPORT_COMPILE_COMMANDS processing can be difficult in the Visual Studio environment, as it may be cached in many places. The 'environment' may include the OS, the user shell, CMake's own environment, along with the Visual Studio presets and caches. See previous commit for arefacts that need removing for a clean test. Signed-off-by: Philip Oakley --- contrib/buildsystems/CMakeLists.txt | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt index d21e22835643d0..fa3885b9bc99b0 100644 --- a/contrib/buildsystems/CMakeLists.txt +++ b/contrib/buildsystems/CMakeLists.txt @@ -63,10 +63,20 @@ endif() if(NOT DEFINED CMAKE_EXPORT_COMPILE_COMMANDS) set(CMAKE_EXPORT_COMPILE_COMMANDS TRUE) + message("settting CMAKE_EXPORT_COMPILE_COMMANDS: ${CMAKE_EXPORT_COMPILE_COMMANDS}") endif() if(USE_VCPKG) set(VCPKG_DIR "${CMAKE_SOURCE_DIR}/compat/vcbuild/vcpkg") + message("WIN32: ${WIN32}") # show its underlying text values + message("VCPKG_DIR: ${VCPKG_DIR}") + message("VCPKG_ARCH: ${VCPKG_ARCH}") # maybe unset + message("MSVC: ${MSVC}") + message("CMAKE_GENERATOR: ${CMAKE_GENERATOR}") + message("CMAKE_CXX_COMPILER_ID: ${CMAKE_CXX_COMPILER_ID}") + message("CMAKE_GENERATOR_PLATFORM: ${CMAKE_GENERATOR_PLATFORM}") + message("CMAKE_EXPORT_COMPILE_COMMANDS: ${CMAKE_EXPORT_COMPILE_COMMANDS}") + message("ENV(CMAKE_EXPORT_COMPILE_COMMANDS): $ENV{CMAKE_EXPORT_COMPILE_COMMANDS}") if(NOT EXISTS ${VCPKG_DIR}) message("Initializing vcpkg and building the Git's dependencies (this will take a while...)") execute_process(COMMAND ${CMAKE_SOURCE_DIR}/compat/vcbuild/vcpkg_install.bat ${VCPKG_ARCH}) From 24571ca4629df6e84c0908c74d095bb42fe205c2 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Sun, 31 Oct 2021 23:15:13 +0000 Subject: [PATCH 087/259] hash-object: demonstrate a >4GB/LLP64 problem On LLP64 systems, such as Windows, the size of `long`, `int`, etc. is only 32 bits (for backward compatibility). Git's use of `unsigned long` for file memory sizes in many places, rather than size_t, limits the handling of large files on LLP64 systems (commonly given as `>4GB`). Provide a minimum test for handling a >4GB file. The `hash-object` command, with the `--literally` and without `-w` option avoids writing the object, either loose or packed. This avoids the code paths hitting the `bigFileThreshold` config test code, the zlib code, and the pack code. Subsequent patches will walk the test's call chain, converting types to `size_t` (which is larger in LLP64 data models) where appropriate. Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- t/t1007-hash-object.sh | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/t/t1007-hash-object.sh b/t/t1007-hash-object.sh index a0481139de5519..51604797e98b35 100755 --- a/t/t1007-hash-object.sh +++ b/t/t1007-hash-object.sh @@ -49,6 +49,9 @@ test_expect_success 'setup' ' example sha1:ddd3f836d3e3fbb7ae289aa9ae83536f76956399 example sha256:b44fe1fe65589848253737db859bd490453510719d7424daab03daf0767b85ae + + large5GB sha1:0be2be10a4c8764f32c4bf372a98edc731a4b204 + large5GB sha256:dc18ca621300c8d3cfa505a275641ebab00de189859e022a975056882d313e64 EOF ' @@ -265,4 +268,12 @@ test_expect_success '--stdin outside of repository (uses SHA-1)' ' test_cmp expect actual ' +test_expect_failure EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ + 'files over 4GB hash literally' ' + test-tool genzeros $((5*1024*1024*1024)) >big && + test_oid large5GB >expect && + git hash-object --stdin --literally actual && + test_cmp expect actual +' + test_done From 62a86b8f89c21e16146b6457914438075cf58ce2 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Fri, 12 Nov 2021 21:07:03 +0000 Subject: [PATCH 088/259] write_object_file_literally(): use size_t The previous commit adds a test that demonstrates a problem in the `hash-object --literally` command, manifesting in an unnecessary file size limit on systems using the LLP64 data model (which includes Windows). Walking the affected code path is `cmd_hash_object()` >> `hash_fd()` >> `hash_literally()` >> `hash_object_file_literally()`. The function `hash_object_file_literally()` is the first with a file length parameter (via a mem buffer). This commit changes the type of that parameter to the LLP64 compatible `size_t` type. There are no other uses of the function. The `strbuf` type is already `size_t` compatible. Note: The hash-object test does not yet pass. Subsequent commits will continue to walk the call tree's lower level functions to identify further fixes. Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- object-file.c | 4 ++-- object-store-ll.h | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/object-file.c b/object-file.c index 726e41a0475b43..d9c1b2baa5cfc9 100644 --- a/object-file.c +++ b/object-file.c @@ -1977,7 +1977,7 @@ static void write_object_file_prepare(const struct git_hash_algo *algo, } static void write_object_file_prepare_literally(const struct git_hash_algo *algo, - const void *buf, unsigned long len, + const void *buf, size_t len, const char *type, struct object_id *oid, char *hdr, int *hdrlen) { @@ -2531,7 +2531,7 @@ int write_object_file_flags(const void *buf, unsigned long len, return 0; } -int write_object_file_literally(const void *buf, unsigned long len, +int write_object_file_literally(const void *buf, size_t len, const char *type, struct object_id *oid, unsigned flags) { diff --git a/object-store-ll.h b/object-store-ll.h index cd3bd5bd99f78c..d54534742f97df 100644 --- a/object-store-ll.h +++ b/object-store-ll.h @@ -282,7 +282,7 @@ static inline int write_object_file(const void *buf, unsigned long len, return write_object_file_flags(buf, len, type, oid, NULL, 0); } -int write_object_file_literally(const void *buf, unsigned long len, +int write_object_file_literally(const void *buf, size_t len, const char *type, struct object_id *oid, unsigned flags); int stream_loose_object(struct input_stream *in_stream, size_t len, From 5ef640773c6badb0f9cba65164a56c9bad2f55dd Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Fri, 12 Nov 2021 21:14:50 +0000 Subject: [PATCH 089/259] object-file.c: use size_t for header lengths Continue walking the code path for the >4GB `hash-object --literally` test. The `hash_object_file_literally()` function internally uses both `hash_object_file()` and `write_object_file_prepare()`. Both function signatures use `unsigned long` rather than `size_t` for the mem buffer sizes. Use `size_t` instead, for LLP64 compatibility. While at it, convert those function's object's header buffer length to `size_t` for consistency. The value is already upcast to `uintmax_t` for print format compatibility. Note: The hash-object test still does not pass. A subsequent commit continues to walk the call tree's lower level hash functions to identify further fixes. Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- object-file.c | 22 +++++++++++----------- object-store-ll.h | 4 ++-- 2 files changed, 13 insertions(+), 13 deletions(-) diff --git a/object-file.c b/object-file.c index d9c1b2baa5cfc9..5c8870c2a7d20a 100644 --- a/object-file.c +++ b/object-file.c @@ -1954,7 +1954,7 @@ void *read_object_with_reference(struct repository *r, static void hash_object_body(const struct git_hash_algo *algo, struct git_hash_ctx *c, const void *buf, unsigned long len, struct object_id *oid, - char *hdr, int *hdrlen) + char *hdr, size_t *hdrlen) { algo->init_fn(c); git_hash_update(c, hdr, *hdrlen); @@ -1963,9 +1963,9 @@ static void hash_object_body(const struct git_hash_algo *algo, struct git_hash_c } static void write_object_file_prepare(const struct git_hash_algo *algo, - const void *buf, unsigned long len, + const void *buf, size_t len, enum object_type type, struct object_id *oid, - char *hdr, int *hdrlen) + char *hdr, size_t *hdrlen) { struct git_hash_ctx c; @@ -1979,7 +1979,7 @@ static void write_object_file_prepare(const struct git_hash_algo *algo, static void write_object_file_prepare_literally(const struct git_hash_algo *algo, const void *buf, size_t len, const char *type, struct object_id *oid, - char *hdr, int *hdrlen) + char *hdr, size_t *hdrlen) { struct git_hash_ctx c; @@ -2117,17 +2117,17 @@ int finalize_object_file_flags(const char *tmpfile, const char *filename, } static void hash_object_file_literally(const struct git_hash_algo *algo, - const void *buf, unsigned long len, + const void *buf, size_t len, const char *type, struct object_id *oid) { char hdr[MAX_HEADER_LEN]; - int hdrlen = sizeof(hdr); + size_t hdrlen = sizeof(hdr); write_object_file_prepare_literally(algo, buf, len, type, oid, hdr, &hdrlen); } void hash_object_file(const struct git_hash_algo *algo, const void *buf, - unsigned long len, enum object_type type, + size_t len, enum object_type type, struct object_id *oid) { hash_object_file_literally(algo, buf, len, type_name(type), oid); @@ -2491,7 +2491,7 @@ int stream_loose_object(struct input_stream *in_stream, size_t len, return err; } -int write_object_file_flags(const void *buf, unsigned long len, +int write_object_file_flags(const void *buf, size_t len, enum object_type type, struct object_id *oid, struct object_id *compat_oid_in, unsigned flags) { @@ -2500,7 +2500,7 @@ int write_object_file_flags(const void *buf, unsigned long len, const struct git_hash_algo *compat = repo->compat_hash_algo; struct object_id compat_oid; char hdr[MAX_HEADER_LEN]; - int hdrlen = sizeof(hdr); + size_t hdrlen = sizeof(hdr); /* Generate compat_oid */ if (compat) { @@ -2540,8 +2540,8 @@ int write_object_file_literally(const void *buf, size_t len, const struct git_hash_algo *algo = repo->hash_algo; const struct git_hash_algo *compat = repo->compat_hash_algo; struct object_id compat_oid; - int hdrlen, status = 0; - int compat_type = -1; + size_t hdrlen; + int status = 0, compat_type = -1; if (compat) { compat_type = type_from_string_gently(type, -1, 1); diff --git a/object-store-ll.h b/object-store-ll.h index d54534742f97df..f9eee3433a112c 100644 --- a/object-store-ll.h +++ b/object-store-ll.h @@ -270,10 +270,10 @@ void *repo_read_object_file(struct repository *r, int oid_object_info(struct repository *r, const struct object_id *, unsigned long *); void hash_object_file(const struct git_hash_algo *algo, const void *buf, - unsigned long len, enum object_type type, + size_t len, enum object_type type, struct object_id *oid); -int write_object_file_flags(const void *buf, unsigned long len, +int write_object_file_flags(const void *buf, size_t len, enum object_type type, struct object_id *oid, struct object_id *comapt_oid_in, unsigned flags); static inline int write_object_file(const void *buf, unsigned long len, From d611ed842d2fb94dd4f0ebdd5ea3bac82f31f8bf Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Fri, 12 Nov 2021 21:16:51 +0000 Subject: [PATCH 090/259] hash algorithms: use size_t for section lengths Continue walking the code path for the >4GB `hash-object --literally` test to the hash algorithm step for LLP64 systems. This patch lets the SHA1DC code use `size_t`, making it compatible with LLP64 data models (as used e.g. by Windows). The interested reader of this patch will note that we adjust the signature of the `git_SHA1DCUpdate()` function without updating _any_ call site. This certainly puzzled at least one reviewer already, so here is an explanation: This function is never called directly, but always via the macro `platform_SHA1_Update`, which is usually called via the macro `git_SHA1_Update`. However, we never call `git_SHA1_Update()` directly in `struct git_hash_algo`. Instead, we call `git_hash_sha1_update()`, which is defined thusly: static void git_hash_sha1_update(git_hash_ctx *ctx, const void *data, size_t len) { git_SHA1_Update(&ctx->sha1, data, len); } i.e. it contains an implicit downcast from `size_t` to `unsigned long` (before this here patch). With this patch, there is no downcast anymore. With this patch, finally, the t1007-hash-object.sh "files over 4GB hash literally" test case is fixed. Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- object-file.c | 4 ++-- sha1dc_git.c | 3 +-- sha1dc_git.h | 2 +- t/t1007-hash-object.sh | 2 +- 4 files changed, 5 insertions(+), 6 deletions(-) diff --git a/object-file.c b/object-file.c index 5c8870c2a7d20a..4d29bf2e3d2415 100644 --- a/object-file.c +++ b/object-file.c @@ -1952,7 +1952,7 @@ void *read_object_with_reference(struct repository *r, } static void hash_object_body(const struct git_hash_algo *algo, struct git_hash_ctx *c, - const void *buf, unsigned long len, + const void *buf, size_t len, struct object_id *oid, char *hdr, size_t *hdrlen) { @@ -1972,7 +1972,7 @@ static void write_object_file_prepare(const struct git_hash_algo *algo, /* Generate the header */ *hdrlen = format_object_header(hdr, *hdrlen, type, len); - /* Sha1.. */ + /* Hash (function pointers) computation */ hash_object_body(algo, &c, buf, len, oid, hdr, hdrlen); } diff --git a/sha1dc_git.c b/sha1dc_git.c index 9b675a046ee699..fe58d7962a30c9 100644 --- a/sha1dc_git.c +++ b/sha1dc_git.c @@ -27,10 +27,9 @@ void git_SHA1DCFinal(unsigned char hash[20], SHA1_CTX *ctx) /* * Same as SHA1DCUpdate, but adjust types to match git's usual interface. */ -void git_SHA1DCUpdate(SHA1_CTX *ctx, const void *vdata, unsigned long len) +void git_SHA1DCUpdate(SHA1_CTX *ctx, const void *vdata, size_t len) { const char *data = vdata; - /* We expect an unsigned long, but sha1dc only takes an int */ while (len > INT_MAX) { SHA1DCUpdate(ctx, data, INT_MAX); data += INT_MAX; diff --git a/sha1dc_git.h b/sha1dc_git.h index f6f880cabea382..0bcf1aa84b7241 100644 --- a/sha1dc_git.h +++ b/sha1dc_git.h @@ -15,7 +15,7 @@ void git_SHA1DCInit(SHA1_CTX *); #endif void git_SHA1DCFinal(unsigned char [20], SHA1_CTX *); -void git_SHA1DCUpdate(SHA1_CTX *ctx, const void *data, unsigned long len); +void git_SHA1DCUpdate(SHA1_CTX *ctx, const void *data, size_t len); #define platform_SHA_IS_SHA1DC /* used by "test-tool sha1-is-sha1dc" */ diff --git a/t/t1007-hash-object.sh b/t/t1007-hash-object.sh index 51604797e98b35..710139837e7a91 100755 --- a/t/t1007-hash-object.sh +++ b/t/t1007-hash-object.sh @@ -268,7 +268,7 @@ test_expect_success '--stdin outside of repository (uses SHA-1)' ' test_cmp expect actual ' -test_expect_failure EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ +test_expect_success EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ 'files over 4GB hash literally' ' test-tool genzeros $((5*1024*1024*1024)) >big && test_oid large5GB >expect && From 3f79b30dfe751968c524e4aaaf38263a8c8d7001 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Mon, 6 Dec 2021 22:26:50 +0000 Subject: [PATCH 091/259] hash-object --stdin: verify that it works with >4GB/LLP64 Just like the `hash-object --literally` code path, the `--stdin` code path also needs to use `size_t` instead of `unsigned long` to represent memory sizes, otherwise it would cause problems on platforms using the LLP64 data model (such as Windows). To limit the scope of the test case, the object is explicitly not written to the object store, nor are any filters applied. The `big` file from the previous test case is reused to save setup time; To avoid relying on that side effect, it is generated if it does not exist (e.g. when running via `sh t1007-*.sh --long --run=1,41`). Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- t/t1007-hash-object.sh | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/t/t1007-hash-object.sh b/t/t1007-hash-object.sh index 710139837e7a91..aa1dd51709d5cc 100755 --- a/t/t1007-hash-object.sh +++ b/t/t1007-hash-object.sh @@ -276,4 +276,12 @@ test_expect_success EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ test_cmp expect actual ' +test_expect_success EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ + 'files over 4GB hash correctly via --stdin' ' + { test -f big || test-tool genzeros $((5*1024*1024*1024)) >big; } && + test_oid large5GB >expect && + git hash-object --stdin actual && + test_cmp expect actual +' + test_done From 43f9bbdce4f9bf66023923b4f219858f5a1eebe0 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Mon, 6 Dec 2021 22:42:46 +0000 Subject: [PATCH 092/259] hash-object: add another >4GB/LLP64 test case To complement the `--stdin` and `--literally` test cases that verify that we can hash files larger than 4GB on 64-bit platforms using the LLP64 data model, here is a test case that exercises `hash-object` _without_ any options. Just as before, we use the `big` file from the previous test case if it exists to save on setup time, otherwise generate it. Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- t/t1007-hash-object.sh | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/t/t1007-hash-object.sh b/t/t1007-hash-object.sh index aa1dd51709d5cc..7c6b7913d4f23d 100755 --- a/t/t1007-hash-object.sh +++ b/t/t1007-hash-object.sh @@ -284,4 +284,12 @@ test_expect_success EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ test_cmp expect actual ' +test_expect_success EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ + 'files over 4GB hash correctly' ' + { test -f big || test-tool genzeros $((5*1024*1024*1024)) >big; } && + test_oid large5GB >expect && + git hash-object -- big >actual && + test_cmp expect actual +' + test_done From 37089cbc0814e7a2f8b9701689e7541dafa8e55c Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Wed, 13 Apr 2022 14:49:17 -0400 Subject: [PATCH 093/259] setup: properly use "%(prefix)/" when in WSL Signed-off-by: Derrick Stolee --- setup.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) diff --git a/setup.c b/setup.c index f93bd6a24a5d9c..6791111992c5bb 100644 --- a/setup.c +++ b/setup.c @@ -1781,10 +1781,19 @@ const char *setup_git_directory_gently(int *nongit_ok) break; case GIT_DIR_INVALID_OWNERSHIP: if (!nongit_ok) { + struct strbuf prequoted = STRBUF_INIT; struct strbuf quoted = STRBUF_INIT; strbuf_complete(&report, '\n'); - sq_quote_buf_pretty("ed, dir.buf); + +#ifdef __MINGW32__ + if (dir.buf[0] == '/') + strbuf_addstr(&prequoted, "%(prefix)/"); +#endif + + strbuf_add(&prequoted, dir.buf, dir.len); + sq_quote_buf_pretty("ed, prequoted.buf); + die(_("detected dubious ownership in repository at '%s'\n" "%s" "To add an exception for this directory, call:\n" From eb2c4b4c81a26353421ed54821a614e370c6ca75 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 8 Sep 2021 13:05:42 +0200 Subject: [PATCH 094/259] init: do parse _all_ core.* settings early In Git for Windows, `has_symlinks` is set to 0 by default. Therefore, we need to parse the config setting `core.symlinks` to know if it has been set to `true`. In `git init`, we must do that before copying the templates because they might contain symbolic links. Even if the support for symbolic links on Windows has not made it to upstream Git yet, we really should make sure that all the `core.*` settings are parsed before proceeding, as they might very well change the behavior of `git init` in a way the user intended. This fixes https://github.com/git-for-windows/git/issues/3414 Signed-off-by: Johannes Schindelin --- config.c | 4 ++-- config.h | 2 ++ setup.c | 2 +- 3 files changed, 5 insertions(+), 3 deletions(-) diff --git a/config.c b/config.c index 658569af084897..e1a9e9aa7aa6ec 100644 --- a/config.c +++ b/config.c @@ -1393,8 +1393,8 @@ int git_config_color(char *dest, const char *var, const char *value) return 0; } -static int git_default_core_config(const char *var, const char *value, - const struct config_context *ctx, void *cb) +int git_default_core_config(const char *var, const char *value, + const struct config_context *ctx, void *cb) { /* This needs a better name */ if (!strcmp(var, "core.filemode")) { diff --git a/config.h b/config.h index 5c730c4f899160..e4199bbdc07685 100644 --- a/config.h +++ b/config.h @@ -167,6 +167,8 @@ typedef int (*config_fn_t)(const char *, const char *, int git_default_config(const char *, const char *, const struct config_context *, void *); +int git_default_core_config(const char *var, const char *value, + const struct config_context *ctx, void *cb); /** * Read a specific file in git-config format. diff --git a/setup.c b/setup.c index f93bd6a24a5d9c..554b3e3751390d 100644 --- a/setup.c +++ b/setup.c @@ -2590,7 +2590,7 @@ int init_db(const char *git_dir, const char *real_git_dir, * have set up the repository format such that we can evaluate * includeIf conditions correctly in the case of re-initialization. */ - git_config(platform_core_config, NULL); + git_config(git_default_core_config, NULL); safe_create_dir(the_repository, git_dir, 0); From 721f99aefe944f02024ad3abc7b642b934ac92d1 Mon Sep 17 00:00:00 2001 From: Philip Oakley Date: Tue, 7 Dec 2021 09:53:41 +0000 Subject: [PATCH 095/259] hash-object: add a >4GB/LLP64 test case using filtered input To verify that the `clean` side of the `clean`/`smudge` filter code is correct with regards to LLP64 (read: to ensure that `size_t` is used instead of `unsigned long`), here is a test case using a trivial filter, specifically _not_ writing anything to the object store to limit the scope of the test case. As in previous commits, the `big` file from previous test cases is reused if available, to save setup time, otherwise re-generated. Signed-off-by: Philip Oakley Signed-off-by: Johannes Schindelin --- t/t1007-hash-object.sh | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/t/t1007-hash-object.sh b/t/t1007-hash-object.sh index 7c6b7913d4f23d..5850c27a9cc315 100755 --- a/t/t1007-hash-object.sh +++ b/t/t1007-hash-object.sh @@ -292,4 +292,16 @@ test_expect_success EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ test_cmp expect actual ' +# This clean filter does nothing, other than excercising the interface. +# We ensure that cleaning doesn't mangle large files on 64-bit Windows. +test_expect_success EXPENSIVE,SIZE_T_IS_64BIT,!LONG_IS_64BIT \ + 'hash filtered files over 4GB correctly' ' + { test -f big || test-tool genzeros $((5*1024*1024*1024)) >big; } && + test_oid large5GB >expect && + test_config filter.null-filter.clean "cat" && + echo "big filter=null-filter" >.gitattributes && + git hash-object -- big >actual && + test_cmp expect actual +' + test_done From 8537e4eba9f41780510763bc4b89377da47d52c6 Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Wed, 13 Apr 2022 14:54:43 -0400 Subject: [PATCH 096/259] compat/mingw.c: do not warn when failing to get owner In the case of Git for Windows (say, in a Git Bash window) running in a Windows Subsystem for Linux (WSL) directory, the GetNamedSecurityInfoW() call in is_path_owned_By_current_side() returns an error code other than ERROR_SUCCESS. This is consistent behavior across this boundary. In these cases, the owner would always be different because the WSL owner is a different entity than the Windows user. The change here is to suppress the error message that looks like this: error: failed to get owner for '//wsl.localhost/...' (1) Before this change, this warning happens for every Git command, regardless of whether the directory is marked with safe.directory. Signed-off-by: Derrick Stolee --- compat/mingw.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..4f0833b3523208 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -2926,9 +2926,7 @@ int is_path_owned_by_current_sid(const char *path, struct strbuf *report) DACL_SECURITY_INFORMATION, &sid, NULL, NULL, NULL, &descriptor); - if (err != ERROR_SUCCESS) - error(_("failed to get owner for '%s' (%ld)"), path, err); - else if (sid && IsValidSid(sid)) { + if (err == ERROR_SUCCESS && sid && IsValidSid(sid)) { /* Now, verify that the SID matches the current user's */ static PSID current_user_sid; BOOL is_member; From 11f6fbd9704307227ded437d3ae17f7f4d82717b Mon Sep 17 00:00:00 2001 From: Rafael Kitover Date: Tue, 12 Apr 2022 19:53:33 +0000 Subject: [PATCH 097/259] mingw: $env:TERM="xterm-256color" for newer OSes For Windows builds >= 15063 set $env:TERM to "xterm-256color" instead of "cygwin" because they have a more capable console system that supports this. Also set $env:COLORTERM="truecolor" if unset. $env:TERM is initialized so that ANSI colors in color.c work, see 29a3963484 (Win32: patch Windows environment on startup, 2012-01-15). See git-for-windows/git#3629 regarding problems caused by always setting $env:TERM="cygwin". This is the same heuristic used by the Cygwin runtime. Signed-off-by: Rafael Kitover Signed-off-by: Johannes Schindelin --- compat/mingw.c | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..301d28b8d173a9 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -2797,9 +2797,20 @@ static void setup_windows_environment(void) convert_slashes(tmp); } - /* simulate TERM to enable auto-color (see color.c) */ - if (!getenv("TERM")) - setenv("TERM", "cygwin", 1); + + /* + * Make sure TERM is set up correctly to enable auto-color + * (see color.c .) Use "cygwin" for older OS releases which + * works correctly with MSYS2 utilities on older consoles. + */ + if (!getenv("TERM")) { + if ((GetVersion() >> 16) < 15063) + setenv("TERM", "cygwin", 0); + else { + setenv("TERM", "xterm-256color", 0); + setenv("COLORTERM", "truecolor", 0); + } + } /* calculate HOME if not set */ if (!getenv("HOME")) { From beeb830e04058a03193a50992c3079027549f636 Mon Sep 17 00:00:00 2001 From: Christopher Degawa Date: Sat, 28 May 2022 14:53:54 -0500 Subject: [PATCH 098/259] winansi: check result and Buffer before using Name NtQueryObject under Wine can return a success but fill out no name. In those situations, Wine will set Buffer to NULL, and set result to the sizeof(OBJECT_NAME_INFORMATION). Running a command such as echo "$(git.exe --version 2>/dev/null)" will crash due to a NULL pointer dereference when the code attempts to null terminate the buffer, although, weirdly, removing the subshell or redirecting stdout to a file will not trigger the crash. Code has been added to also check Buffer and Length to ensure the check is as robust as possible due to the current behavior being fragile at best, and could potentially change in the future This code is based on the behavior of NtQueryObject under wine and reactos. Signed-off-by: Christopher Degawa --- compat/winansi.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/compat/winansi.c b/compat/winansi.c index ac2ffb78691a7d..d28137a20b0bcc 100644 --- a/compat/winansi.c +++ b/compat/winansi.c @@ -575,6 +575,9 @@ static void detect_msys_tty(int fd) if (!NT_SUCCESS(NtQueryObject(h, ObjectNameInformation, buffer, sizeof(buffer) - 2, &result))) return; + if (result < sizeof(*nameinfo) || !nameinfo->Name.Buffer || + !nameinfo->Name.Length) + return; name = nameinfo->Name.Buffer; name[nameinfo->Name.Length / sizeof(*name)] = 0; From 551cfe61ff1d4650a5eec5b7bf58b1a67e7b6083 Mon Sep 17 00:00:00 2001 From: Dennis Ameling Date: Tue, 4 Oct 2022 09:59:32 +0200 Subject: [PATCH 099/259] config.mak.uname: add support for clangarm64 CLANGARM64 is a relatively new MSYSTEM added by the MSYS2 team. In order to have Git build correctly for this platform, let's add some configuration for it to config.mak.uname. Signed-off-by: Dennis Ameling --- config.mak.uname | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/config.mak.uname b/config.mak.uname index 2ebb82feb9f1b2..bf0247c3ecc841 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -724,6 +724,10 @@ ifeq ($(uname_S),MINGW) prefix = /mingw64 HOST_CPU = x86_64 BASIC_LDFLAGS += -Wl,--pic-executable,-e,mainCRTStartup + else ifeq (CLANGARM64,$(MSYSTEM)) + prefix = /clangarm64 + HOST_CPU = aarch64 + BASIC_LDFLAGS += -Wl,--pic-executable,-e,mainCRTStartup else COMPAT_CFLAGS += -D_USE_32BIT_TIME_T BASIC_LDFLAGS += -Wl,--large-address-aware From 6d3e5cfd4947270e8028cac78e0ca1c268898884 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 25 Feb 2025 10:17:12 +0000 Subject: [PATCH 100/259] msvc: do handle builds on Windows/ARM64 Git for Windows/ARM64 settled on using `clang` to compile `git.exe`, and hence needs to run in a system where `MSYSTEM` is set to `CLANGARM64` and the prefix to use is `/clangarm64`. We already did that in the `MINGW` arm, i.e. for regular Git for Windows builds using MINGW GCC (or `clang`'s shim pretending to be GCC), now it is time to do the same in the MS Visual C part. Signed-off-by: Johannes Schindelin --- config.mak.uname | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/config.mak.uname b/config.mak.uname index bf0247c3ecc841..a955b69527bffb 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -432,7 +432,11 @@ ifeq ($(uname_S),Windows) ifeq (MINGW32,$(MSYSTEM)) prefix = /mingw32 else - prefix = /mingw64 + ifeq (CLANGARM64,$(MSYSTEM)) + prefix = /clangarm64 + else + prefix = /mingw64 + endif endif # Prepend MSVC 64-bit tool-chain to PATH. # From e88076e046d87ee07403e40c8fca26620293499f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E5=AD=99=E5=8D=93=E8=AF=86?= Date: Sun, 16 Jan 2022 03:38:33 +0800 Subject: [PATCH 101/259] Add config option `windows.appendAtomically` MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Atomic append on windows is only supported on local disk files, and it may cause errors in other situations, e.g. network file system. If that is the case, this config option should be used to turn atomic append off. Co-Authored-By: Johannes Schindelin Signed-off-by: 孙卓识 Signed-off-by: Johannes Schindelin --- Documentation/config.adoc | 2 ++ Documentation/config/windows.adoc | 4 ++++ compat/mingw.c | 36 ++++++++++++++++++++++++++++--- 3 files changed, 39 insertions(+), 3 deletions(-) create mode 100644 Documentation/config/windows.adoc diff --git a/Documentation/config.adoc b/Documentation/config.adoc index a8b04c4e5183d2..30e11203c87d1c 100644 --- a/Documentation/config.adoc +++ b/Documentation/config.adoc @@ -558,4 +558,6 @@ include::config/versionsort.adoc[] include::config/web.adoc[] +include::config/windows.adoc[] + include::config/worktree.adoc[] diff --git a/Documentation/config/windows.adoc b/Documentation/config/windows.adoc new file mode 100644 index 00000000000000..fdaaf1c65504f3 --- /dev/null +++ b/Documentation/config/windows.adoc @@ -0,0 +1,4 @@ +windows.appendAtomically:: + By default, append atomic API is used on windows. But it works only with + local disk files, if you're working on a network file system, you should + set it false to turn it off. diff --git a/compat/mingw.c b/compat/mingw.c index f524c54d06d965..0658cdfc3c733f 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -21,6 +21,7 @@ #include "gettext.h" #define SECURITY_WIN32 #include +#include "../repository.h" #define HCAST(type, handle) ((type)(intptr_t)handle) @@ -614,6 +615,7 @@ static int is_local_named_pipe_path(const char *filename) int mingw_open (const char *filename, int oflags, ...) { + static int append_atomically = -1; typedef int (*open_fn_t)(wchar_t const *wfilename, int oflags, ...); va_list args; unsigned mode; @@ -630,7 +632,16 @@ int mingw_open (const char *filename, int oflags, ...) return -1; } - if ((oflags & O_APPEND) && !is_local_named_pipe_path(filename)) + /* + * Only set append_atomically to default value(1) when repo is initialized + * and fail to get config value + */ + if (append_atomically < 0 && the_repository && the_repository->commondir && + git_config_get_bool("windows.appendatomically", &append_atomically)) + append_atomically = 1; + + if (append_atomically && (oflags & O_APPEND) && + !is_local_named_pipe_path(filename)) open_fn = mingw_open_append; else if (!(oflags & ~(O_ACCMODE | O_NOINHERIT))) open_fn = mingw_open_existing; @@ -781,9 +792,28 @@ ssize_t mingw_write(int fd, const void *buf, size_t len) /* check if fd is a pipe */ HANDLE h = (HANDLE) _get_osfhandle(fd); - if (GetFileType(h) != FILE_TYPE_PIPE) + if (GetFileType(h) != FILE_TYPE_PIPE) { + if (orig == EINVAL) { + wchar_t path[MAX_PATH]; + DWORD ret = GetFinalPathNameByHandleW(h, path, + ARRAY_SIZE(path), 0); + UINT drive_type = ret > 0 && ret < ARRAY_SIZE(path) ? + GetDriveTypeW(path) : DRIVE_UNKNOWN; + + /* + * The default atomic append causes such an error on + * network file systems, in such a case, it should be + * turned off via config. + * + * `drive_type` of UNC path: DRIVE_NO_ROOT_DIR + */ + if (DRIVE_NO_ROOT_DIR == drive_type || DRIVE_REMOTE == drive_type) + warning("invalid write operation detected; you may try:\n" + "\n\tgit config windows.appendAtomically false"); + } + errno = orig; - else if (orig == EINVAL) + } else if (orig == EINVAL) errno = EPIPE; else { DWORD buf_size; From 5398cd9fe7c15b4739352b8d10c9245493c51b80 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Mon, 4 Sep 2017 11:59:45 +0200 Subject: [PATCH 102/259] mingw: change core.fsyncObjectFiles = 1 by default MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From the documentation of said setting: This boolean will enable fsync() when writing object files. This is a total waste of time and effort on a filesystem that orders data writes properly, but can be useful for filesystems that do not use journalling (traditional UNIX filesystems) or that only journal metadata and not file contents (OS X’s HFS+, or Linux ext3 with "data=writeback"). The most common file system on Windows (NTFS) does not guarantee that order, therefore a sudden loss of power (or any other event causing an unclean shutdown) would cause corrupt files (i.e. files filled with NULs). Therefore we need to change the default. Note that the documentation makes it sound as if this causes really bad performance. In reality, writing loose objects is something that is done only rarely, and only a handful of files at a time. Signed-off-by: Johannes Schindelin --- compat/mingw.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/compat/mingw.c b/compat/mingw.c index 0658cdfc3c733f..7a19f84e246f1a 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -21,6 +21,7 @@ #include "gettext.h" #define SECURITY_WIN32 #include +#include "../write-or-die.h" #include "../repository.h" #define HCAST(type, handle) ((type)(intptr_t)handle) @@ -3285,6 +3286,7 @@ int wmain(int argc, const wchar_t **wargv) #endif maybe_redirect_std_handles(); + fsync_object_files = 1; /* determine size of argv and environ conversion buffer */ maxlen = wcslen(wargv[0]); From 5046722503217c7fc2038141671e7da9eeb8ebc6 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Matthias=20A=C3=9Fhauer?= Date: Sun, 10 Jul 2022 11:27:25 +0200 Subject: [PATCH 103/259] MinGW: link as terminal server aware MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Whith Windows 2000, Microsoft introduced a flag to the PE header to mark executables as "terminal server aware". Windows terminal servers provide a redirected Windows directory and redirected registry hives when launching legacy applications without this flag set. Since we do not use any INI files in the Windows directory and don't write to the registry, we don't need this additional preparation. Telling the OS that we don't need this should provide slightly improved startup times in terminal server environments. When building for supported Windows Versions with MSVC the /TSAWARE linker flag is automatically set, but MinGW requires us to set the --tsaware flag manually. This partially addresses https://github.com/git-for-windows/git/issues/3935. Signed-off-by: Matthias Aßhauer --- config.mak.uname | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/config.mak.uname b/config.mak.uname index b12d4e168ae119..1736c4f0116e75 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -691,7 +691,7 @@ ifeq ($(uname_S),MINGW) DEFAULT_HELP_FORMAT = html HAVE_PLATFORM_PROCINFO = YesPlease CSPRNG_METHOD = rtlgenrandom - BASIC_LDFLAGS += -municode + BASIC_LDFLAGS += -municode -Wl,--tsaware COMPAT_CFLAGS += -DNOGDI -Icompat -Icompat/win32 COMPAT_CFLAGS += -DSTRIP_EXTENSION=\".exe\" COMPAT_OBJS += compat/mingw.o compat/winansi.o \ From e9e45cb28f55dba00116b0745f4b0cf6340e206b Mon Sep 17 00:00:00 2001 From: Kiel Hurley Date: Wed, 2 Nov 2022 22:56:16 +1300 Subject: [PATCH 104/259] Fix Windows version resources Add FileVersion, which is a required field As not all required fields were present, none were being included Fixes #4090 Signed-off-by: Kiel Hurley --- git.rc.in | 1 + 1 file changed, 1 insertion(+) diff --git a/git.rc.in b/git.rc.in index e69444eef3f0c5..460ea39561b87f 100644 --- a/git.rc.in +++ b/git.rc.in @@ -12,6 +12,7 @@ BEGIN VALUE "OriginalFilename", "git.exe\0" VALUE "ProductName", "Git\0" VALUE "ProductVersion", "@GIT_VERSION@\0" + VALUE "FileVersion", "@GIT_VERSION@\0" END END From e6aed4b51d1bfe2351507e50a6052080ff871975 Mon Sep 17 00:00:00 2001 From: Dennis Ameling Date: Tue, 4 Oct 2022 09:58:10 +0200 Subject: [PATCH 105/259] bswap.h: add support for built-in bswap functions Newer compiler versions, like GCC 10 and Clang 12, have built-in functions for bswap32 and bswap64. This comes in handy, for example, when targeting CLANGARM64 on Windows, which would not be supported without this logic. Signed-off-by: Dennis Ameling --- compat/bswap.h | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/compat/bswap.h b/compat/bswap.h index b34054f2bd7284..9e0f98e00b93a4 100644 --- a/compat/bswap.h +++ b/compat/bswap.h @@ -35,7 +35,19 @@ static inline uint64_t default_bswap64(uint64_t val) #undef bswap32 #undef bswap64 -#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) +/** + * __has_builtin is available since Clang 10 and GCC 10. + * Below is a fallback for older compilers. + */ +#ifndef __has_builtin + #define __has_builtin(x) 0 +#endif + +#if __has_builtin(__builtin_bswap32) && __has_builtin(__builtin_bswap64) +#define bswap32(x) __builtin_bswap32((x)) +#define bswap64(x) __builtin_bswap64((x)) + +#elif defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) #define bswap32 git_bswap32 static inline uint32_t git_bswap32(uint32_t x) From d31e375c1bf887480a9ab622b55f0ec1a00ae2fe Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 25 Feb 2025 10:21:52 +0000 Subject: [PATCH 106/259] mingw(arm64): do move the `/etc/git*` location In fb5e3378f8 (mingw: move Git for Windows' system config where users expect it, 2021-06-22), I moved the location of Git for Windows' system config and system Git attributes file to the top-level `/etc/` directory (because it is a much more obvious location than, say, `/mingw64/etc/`). The patch relied on a very specific scenario that the newly-supported Windows/ARM64 builds of `git.exe` fails to fall into. So let's broaden the condition a bit, so that Windows/ARM64 builds also use that location (instead of the even more obscure `/clangarm64/etc/` directory). This fixes https://github.com/git-for-windows/git/issues/5431. Signed-off-by: Johannes Schindelin --- config.mak.uname | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/config.mak.uname b/config.mak.uname index a955b69527bffb..0229eb3b8fe21f 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -489,7 +489,7 @@ ifeq ($(uname_S),Windows) NO_POSIX_GOODIES = UnfortunatelyYes NATIVE_CRLF = YesPlease DEFAULT_HELP_FORMAT = html -ifeq (/mingw64,$(subst 32,64,$(prefix))) +ifeq (/mingw64,$(subst 32,64,$(subst clangarm,mingw,$(prefix)))) # Move system config into top-level /etc/ ETC_GITCONFIG = ../etc/gitconfig ETC_GITATTRIBUTES = ../etc/gitattributes @@ -748,7 +748,7 @@ ifeq ($(uname_S),MINGW) USE_LIBPCRE = YesPlease USE_MIMALLOC = YesPlease NO_PYTHON = - ifeq (/mingw64,$(subst 32,64,$(prefix))) + ifeq (/mingw64,$(subst 32,64,$(subst clangarm,mingw,$(prefix)))) # Move system config into top-level /etc/ ETC_GITCONFIG = ../etc/gitconfig ETC_GITATTRIBUTES = ../etc/gitattributes From 905c9fdf9e76a4924a3bdefd7a8f41b03d9e3b46 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sat, 6 May 2023 22:26:15 +0200 Subject: [PATCH 107/259] http: optionally load libcurl lazily This compile-time option allows to ask Git to load libcurl dynamically at runtime. Together with a follow-up patch that optionally overrides the file name depending on the `http.sslBackend` setting, this kicks open the door for installing multiple libcurl flavors side by side, and load the one corresponding to the (runtime-)configured SSL/TLS backend. Signed-off-by: Johannes Schindelin --- Makefile | 28 +++- compat/lazyload-curl.c | 354 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 375 insertions(+), 7 deletions(-) create mode 100644 compat/lazyload-curl.c diff --git a/Makefile b/Makefile index 7315507381e146..59689d92fd0261 100644 --- a/Makefile +++ b/Makefile @@ -475,6 +475,11 @@ include shared.mak # # CURL_LDFLAGS=-lcurl # +# Define LAZYLOAD_LIBCURL to dynamically load the libcurl; This can be useful +# if Multiple libcurl versions exist (with different file names) that link to +# various SSL/TLS backends, to support the `http.sslBackend` runtime switch in +# such a scenario. +# # === Optional library: libpcre2 === # # Define USE_LIBPCRE if you have and want to use libpcre. Various @@ -1661,10 +1666,19 @@ else CURL_LIBCURL = endif - ifndef CURL_LDFLAGS - CURL_LDFLAGS = $(eval CURL_LDFLAGS := $$(shell $$(CURL_CONFIG) --libs))$(CURL_LDFLAGS) + ifdef LAZYLOAD_LIBCURL + LAZYLOAD_LIBCURL_OBJ = compat/lazyload-curl.o + OBJECTS += $(LAZYLOAD_LIBCURL_OBJ) + # The `CURL_STATICLIB` constant must be defined to avoid seeing the functions + # declared as DLL imports + CURL_CFLAGS = -DCURL_STATICLIB + CURL_LIBCURL = -ldl + else + ifndef CURL_LDFLAGS + CURL_LDFLAGS = $(eval CURL_LDFLAGS := $$(shell $$(CURL_CONFIG) --libs))$(CURL_LDFLAGS) + endif + CURL_LIBCURL += $(CURL_LDFLAGS) endif - CURL_LIBCURL += $(CURL_LDFLAGS) ifndef CURL_CFLAGS CURL_CFLAGS = $(eval CURL_CFLAGS := $$(shell $$(CURL_CONFIG) --cflags))$(CURL_CFLAGS) @@ -1685,7 +1699,7 @@ else endif ifdef USE_CURL_FOR_IMAP_SEND BASIC_CFLAGS += -DUSE_CURL_FOR_IMAP_SEND - IMAP_SEND_BUILDDEPS = http.o + IMAP_SEND_BUILDDEPS = http.o $(LAZYLOAD_LIBCURL_OBJ) IMAP_SEND_LDFLAGS += $(CURL_LIBCURL) endif ifndef NO_EXPAT @@ -2886,10 +2900,10 @@ git-imap-send$X: imap-send.o $(IMAP_SEND_BUILDDEPS) GIT-LDFLAGS $(GITLIBS) $(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) \ $(IMAP_SEND_LDFLAGS) $(LIBS) -git-http-fetch$X: http.o http-walker.o http-fetch.o GIT-LDFLAGS $(GITLIBS) +git-http-fetch$X: http.o http-walker.o http-fetch.o $(LAZYLOAD_LIBCURL_OBJ) GIT-LDFLAGS $(GITLIBS) $(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) \ $(CURL_LIBCURL) $(LIBS) -git-http-push$X: http.o http-push.o GIT-LDFLAGS $(GITLIBS) +git-http-push$X: http.o http-push.o $(LAZYLOAD_LIBCURL_OBJ) GIT-LDFLAGS $(GITLIBS) $(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) \ $(CURL_LIBCURL) $(EXPAT_LIBEXPAT) $(LIBS) @@ -2899,7 +2913,7 @@ $(REMOTE_CURL_ALIASES): $(REMOTE_CURL_PRIMARY) ln -s $< $@ 2>/dev/null || \ cp $< $@ -$(REMOTE_CURL_PRIMARY): remote-curl.o http.o http-walker.o GIT-LDFLAGS $(GITLIBS) +$(REMOTE_CURL_PRIMARY): remote-curl.o http.o http-walker.o $(LAZYLOAD_LIBCURL_OBJ) GIT-LDFLAGS $(GITLIBS) $(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) \ $(CURL_LIBCURL) $(EXPAT_LIBEXPAT) $(LIBS) diff --git a/compat/lazyload-curl.c b/compat/lazyload-curl.c new file mode 100644 index 00000000000000..19aa2b6d4b6942 --- /dev/null +++ b/compat/lazyload-curl.c @@ -0,0 +1,354 @@ +#include "../git-compat-util.h" +#include "../git-curl-compat.h" +#include + +/* + * The ABI version of libcurl is encoded in its shared libraries' file names. + * This ABI version has not changed since October 2006 and is unlikely to be + * changed in the future. See https://curl.se/libcurl/abi.html for details. + */ +#define LIBCURL_ABI_VERSION "4" + +typedef void (*func_t)(void); + +#ifdef __APPLE__ +#define LIBCURL_FILE_NAME(base) base "." LIBCURL_ABI_VERSION ".dylib" +#else +#define LIBCURL_FILE_NAME(base) base ".so." LIBCURL_ABI_VERSION +#endif + +static void *load_library(const char *name) +{ + return dlopen(name, RTLD_LAZY); +} + +static func_t load_function(void *handle, const char *name) +{ + /* + * Casting the return value of `dlsym()` to a function pointer is + * explicitly allowed in recent POSIX standards, but GCC complains + * about this in pedantic mode nevertheless. For more about this issue, + * see https://stackoverflow.com/q/31526876/1860823 and + * http://stackoverflow.com/a/36385690/1905491. + */ + func_t f; + *(void **)&f = dlsym(handle, name); + return f; +} + +typedef struct curl_version_info_data *(*curl_version_info_type)(CURLversion version); +static curl_version_info_type curl_version_info_func; + +typedef char *(*curl_easy_escape_type)(CURL *handle, const char *string, int length); +static curl_easy_escape_type curl_easy_escape_func; + +typedef void (*curl_free_type)(void *p); +static curl_free_type curl_free_func; + +typedef CURLcode (*curl_global_init_type)(long flags); +static curl_global_init_type curl_global_init_func; + +typedef CURLsslset (*curl_global_sslset_type)(curl_sslbackend id, const char *name, const curl_ssl_backend ***avail); +static curl_global_sslset_type curl_global_sslset_func; + +typedef void (*curl_global_cleanup_type)(void); +static curl_global_cleanup_type curl_global_cleanup_func; + +typedef struct curl_slist *(*curl_slist_append_type)(struct curl_slist *list, const char *data); +static curl_slist_append_type curl_slist_append_func; + +typedef void (*curl_slist_free_all_type)(struct curl_slist *list); +static curl_slist_free_all_type curl_slist_free_all_func; + +typedef const char *(*curl_easy_strerror_type)(CURLcode error); +static curl_easy_strerror_type curl_easy_strerror_func; + +typedef CURLM *(*curl_multi_init_type)(void); +static curl_multi_init_type curl_multi_init_func; + +typedef CURLMcode (*curl_multi_add_handle_type)(CURLM *multi_handle, CURL *curl_handle); +static curl_multi_add_handle_type curl_multi_add_handle_func; + +typedef CURLMcode (*curl_multi_remove_handle_type)(CURLM *multi_handle, CURL *curl_handle); +static curl_multi_remove_handle_type curl_multi_remove_handle_func; + +typedef CURLMcode (*curl_multi_fdset_type)(CURLM *multi_handle, fd_set *read_fd_set, fd_set *write_fd_set, fd_set *exc_fd_set, int *max_fd); +static curl_multi_fdset_type curl_multi_fdset_func; + +typedef CURLMcode (*curl_multi_perform_type)(CURLM *multi_handle, int *running_handles); +static curl_multi_perform_type curl_multi_perform_func; + +typedef CURLMcode (*curl_multi_cleanup_type)(CURLM *multi_handle); +static curl_multi_cleanup_type curl_multi_cleanup_func; + +typedef CURLMsg *(*curl_multi_info_read_type)(CURLM *multi_handle, int *msgs_in_queue); +static curl_multi_info_read_type curl_multi_info_read_func; + +typedef const char *(*curl_multi_strerror_type)(CURLMcode error); +static curl_multi_strerror_type curl_multi_strerror_func; + +typedef CURLMcode (*curl_multi_timeout_type)(CURLM *multi_handle, long *milliseconds); +static curl_multi_timeout_type curl_multi_timeout_func; + +typedef CURL *(*curl_easy_init_type)(void); +static curl_easy_init_type curl_easy_init_func; + +typedef CURLcode (*curl_easy_perform_type)(CURL *curl); +static curl_easy_perform_type curl_easy_perform_func; + +typedef void (*curl_easy_cleanup_type)(CURL *curl); +static curl_easy_cleanup_type curl_easy_cleanup_func; + +typedef CURL *(*curl_easy_duphandle_type)(CURL *curl); +static curl_easy_duphandle_type curl_easy_duphandle_func; + +typedef CURLcode (*curl_easy_getinfo_long_type)(CURL *curl, CURLINFO info, long *value); +static curl_easy_getinfo_long_type curl_easy_getinfo_long_func; + +typedef CURLcode (*curl_easy_getinfo_pointer_type)(CURL *curl, CURLINFO info, void **value); +static curl_easy_getinfo_pointer_type curl_easy_getinfo_pointer_func; + +typedef CURLcode (*curl_easy_getinfo_off_t_type)(CURL *curl, CURLINFO info, curl_off_t *value); +static curl_easy_getinfo_off_t_type curl_easy_getinfo_off_t_func; + +typedef CURLcode (*curl_easy_setopt_long_type)(CURL *curl, CURLoption opt, long value); +static curl_easy_setopt_long_type curl_easy_setopt_long_func; + +typedef CURLcode (*curl_easy_setopt_pointer_type)(CURL *curl, CURLoption opt, void *value); +static curl_easy_setopt_pointer_type curl_easy_setopt_pointer_func; + +typedef CURLcode (*curl_easy_setopt_off_t_type)(CURL *curl, CURLoption opt, curl_off_t value); +static curl_easy_setopt_off_t_type curl_easy_setopt_off_t_func; + +static void lazy_load_curl(void) +{ + static int initialized; + void *libcurl; + func_t curl_easy_getinfo_func, curl_easy_setopt_func; + + if (initialized) + return; + + initialized = 1; + libcurl = load_library(LIBCURL_FILE_NAME("libcurl")); + if (!libcurl) + die("failed to load library '%s'", LIBCURL_FILE_NAME("libcurl")); + + curl_version_info_func = (curl_version_info_type)load_function(libcurl, "curl_version_info"); + curl_easy_escape_func = (curl_easy_escape_type)load_function(libcurl, "curl_easy_escape"); + curl_free_func = (curl_free_type)load_function(libcurl, "curl_free"); + curl_global_init_func = (curl_global_init_type)load_function(libcurl, "curl_global_init"); + curl_global_sslset_func = (curl_global_sslset_type)load_function(libcurl, "curl_global_sslset"); + curl_global_cleanup_func = (curl_global_cleanup_type)load_function(libcurl, "curl_global_cleanup"); + curl_slist_append_func = (curl_slist_append_type)load_function(libcurl, "curl_slist_append"); + curl_slist_free_all_func = (curl_slist_free_all_type)load_function(libcurl, "curl_slist_free_all"); + curl_easy_strerror_func = (curl_easy_strerror_type)load_function(libcurl, "curl_easy_strerror"); + curl_multi_init_func = (curl_multi_init_type)load_function(libcurl, "curl_multi_init"); + curl_multi_add_handle_func = (curl_multi_add_handle_type)load_function(libcurl, "curl_multi_add_handle"); + curl_multi_remove_handle_func = (curl_multi_remove_handle_type)load_function(libcurl, "curl_multi_remove_handle"); + curl_multi_fdset_func = (curl_multi_fdset_type)load_function(libcurl, "curl_multi_fdset"); + curl_multi_perform_func = (curl_multi_perform_type)load_function(libcurl, "curl_multi_perform"); + curl_multi_cleanup_func = (curl_multi_cleanup_type)load_function(libcurl, "curl_multi_cleanup"); + curl_multi_info_read_func = (curl_multi_info_read_type)load_function(libcurl, "curl_multi_info_read"); + curl_multi_strerror_func = (curl_multi_strerror_type)load_function(libcurl, "curl_multi_strerror"); + curl_multi_timeout_func = (curl_multi_timeout_type)load_function(libcurl, "curl_multi_timeout"); + curl_easy_init_func = (curl_easy_init_type)load_function(libcurl, "curl_easy_init"); + curl_easy_perform_func = (curl_easy_perform_type)load_function(libcurl, "curl_easy_perform"); + curl_easy_cleanup_func = (curl_easy_cleanup_type)load_function(libcurl, "curl_easy_cleanup"); + curl_easy_duphandle_func = (curl_easy_duphandle_type)load_function(libcurl, "curl_easy_duphandle"); + + curl_easy_getinfo_func = load_function(libcurl, "curl_easy_getinfo"); + curl_easy_getinfo_long_func = (curl_easy_getinfo_long_type)curl_easy_getinfo_func; + curl_easy_getinfo_pointer_func = (curl_easy_getinfo_pointer_type)curl_easy_getinfo_func; + curl_easy_getinfo_off_t_func = (curl_easy_getinfo_off_t_type)curl_easy_getinfo_func; + + curl_easy_setopt_func = load_function(libcurl, "curl_easy_setopt"); + curl_easy_setopt_long_func = (curl_easy_setopt_long_type)curl_easy_setopt_func; + curl_easy_setopt_pointer_func = (curl_easy_setopt_pointer_type)curl_easy_setopt_func; + curl_easy_setopt_off_t_func = (curl_easy_setopt_off_t_type)curl_easy_setopt_func; +} + +struct curl_version_info_data *curl_version_info(CURLversion version) +{ + lazy_load_curl(); + return curl_version_info_func(version); +} + +char *curl_easy_escape(CURL *handle, const char *string, int length) +{ + lazy_load_curl(); + return curl_easy_escape_func(handle, string, length); +} + +void curl_free(void *p) +{ + lazy_load_curl(); + curl_free_func(p); +} + +CURLcode curl_global_init(long flags) +{ + lazy_load_curl(); + return curl_global_init_func(flags); +} + +CURLsslset curl_global_sslset(curl_sslbackend id, const char *name, const curl_ssl_backend ***avail) +{ + lazy_load_curl(); + return curl_global_sslset_func(id, name, avail); +} + +void curl_global_cleanup(void) +{ + lazy_load_curl(); + curl_global_cleanup_func(); +} + +struct curl_slist *curl_slist_append(struct curl_slist *list, const char *data) +{ + lazy_load_curl(); + return curl_slist_append_func(list, data); +} + +void curl_slist_free_all(struct curl_slist *list) +{ + lazy_load_curl(); + curl_slist_free_all_func(list); +} + +const char *curl_easy_strerror(CURLcode error) +{ + lazy_load_curl(); + return curl_easy_strerror_func(error); +} + +CURLM *curl_multi_init(void) +{ + lazy_load_curl(); + return curl_multi_init_func(); +} + +CURLMcode curl_multi_add_handle(CURLM *multi_handle, CURL *curl_handle) +{ + lazy_load_curl(); + return curl_multi_add_handle_func(multi_handle, curl_handle); +} + +CURLMcode curl_multi_remove_handle(CURLM *multi_handle, CURL *curl_handle) +{ + lazy_load_curl(); + return curl_multi_remove_handle_func(multi_handle, curl_handle); +} + +CURLMcode curl_multi_fdset(CURLM *multi_handle, fd_set *read_fd_set, fd_set *write_fd_set, fd_set *exc_fd_set, int *max_fd) +{ + lazy_load_curl(); + return curl_multi_fdset_func(multi_handle, read_fd_set, write_fd_set, exc_fd_set, max_fd); +} + +CURLMcode curl_multi_perform(CURLM *multi_handle, int *running_handles) +{ + lazy_load_curl(); + return curl_multi_perform_func(multi_handle, running_handles); +} + +CURLMcode curl_multi_cleanup(CURLM *multi_handle) +{ + lazy_load_curl(); + return curl_multi_cleanup_func(multi_handle); +} + +CURLMsg *curl_multi_info_read(CURLM *multi_handle, int *msgs_in_queue) +{ + lazy_load_curl(); + return curl_multi_info_read_func(multi_handle, msgs_in_queue); +} + +const char *curl_multi_strerror(CURLMcode error) +{ + lazy_load_curl(); + return curl_multi_strerror_func(error); +} + +CURLMcode curl_multi_timeout(CURLM *multi_handle, long *milliseconds) +{ + lazy_load_curl(); + return curl_multi_timeout_func(multi_handle, milliseconds); +} + +CURL *curl_easy_init(void) +{ + lazy_load_curl(); + return curl_easy_init_func(); +} + +CURLcode curl_easy_perform(CURL *curl) +{ + lazy_load_curl(); + return curl_easy_perform_func(curl); +} + +void curl_easy_cleanup(CURL *curl) +{ + lazy_load_curl(); + curl_easy_cleanup_func(curl); +} + +CURL *curl_easy_duphandle(CURL *curl) +{ + lazy_load_curl(); + return curl_easy_duphandle_func(curl); +} + +#ifndef CURL_IGNORE_DEPRECATION +#define CURL_IGNORE_DEPRECATION(x) x +#endif + +#ifndef CURLOPTTYPE_BLOB +#define CURLOPTTYPE_BLOB 40000 +#endif + +#undef curl_easy_getinfo +CURLcode curl_easy_getinfo(CURL *curl, CURLINFO info, ...) +{ + va_list ap; + CURLcode res; + + va_start(ap, info); + lazy_load_curl(); + CURL_IGNORE_DEPRECATION( + if (info >= CURLINFO_LONG && info < CURLINFO_DOUBLE) + res = curl_easy_getinfo_long_func(curl, info, va_arg(ap, long *)); + else if ((info >= CURLINFO_STRING && info < CURLINFO_LONG) || + (info >= CURLINFO_SLIST && info < CURLINFO_SOCKET)) + res = curl_easy_getinfo_pointer_func(curl, info, va_arg(ap, void **)); + else if (info >= CURLINFO_OFF_T) + res = curl_easy_getinfo_off_t_func(curl, info, va_arg(ap, curl_off_t *)); + else + die("%s:%d: TODO (info: %d)!", __FILE__, __LINE__, info); + ) + va_end(ap); + return res; +} + +#undef curl_easy_setopt +CURLcode curl_easy_setopt(CURL *curl, CURLoption opt, ...) +{ + va_list ap; + CURLcode res; + + va_start(ap, opt); + lazy_load_curl(); + CURL_IGNORE_DEPRECATION( + if (opt >= CURLOPTTYPE_LONG && opt < CURLOPTTYPE_OBJECTPOINT) + res = curl_easy_setopt_long_func(curl, opt, va_arg(ap, long)); + else if (opt >= CURLOPTTYPE_OBJECTPOINT && opt < CURLOPTTYPE_OFF_T) + res = curl_easy_setopt_pointer_func(curl, opt, va_arg(ap, void *)); + else if (opt >= CURLOPTTYPE_OFF_T && opt < CURLOPTTYPE_BLOB) + res = curl_easy_setopt_off_t_func(curl, opt, va_arg(ap, curl_off_t)); + else + die("%s:%d: TODO (opt: %d)!", __FILE__, __LINE__, opt); + ) + va_end(ap); + return res; +} From 14f7f3324ac896d5ffb5e767026989d9d85ead47 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sun, 7 May 2023 22:51:52 +0200 Subject: [PATCH 108/259] http: support lazy-loading libcurl also on Windows This implements the Windows-specific support code, because everything is slightly different on Windows, even loading shared libraries. Note: I specifically do _not_ use the code from `compat/win32/lazyload.h` here because that code is optimized for loading individual functions from various system DLLs, while we specifically want to load _many_ functions from _one_ DLL here, and distinctly not a system DLL (we expect libcurl to be located outside `C:\Windows\system32`, something `INIT_PROC_ADDR` refuses to work with). Also, the `curl_easy_getinfo()`/`curl_easy_setopt()` functions are declared as vararg functions, which `lazyload.h` cannot handle. Finally, we are about to optionally override the exact file name that is to be loaded, which is a goal contrary to `lazyload.h`'s design. Signed-off-by: Johannes Schindelin --- Makefile | 4 ++++ compat/lazyload-curl.c | 52 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 56 insertions(+) diff --git a/Makefile b/Makefile index 59689d92fd0261..234a709593c254 100644 --- a/Makefile +++ b/Makefile @@ -1672,7 +1672,11 @@ else # The `CURL_STATICLIB` constant must be defined to avoid seeing the functions # declared as DLL imports CURL_CFLAGS = -DCURL_STATICLIB +ifneq ($(uname_S),MINGW) +ifneq ($(uname_S),Windows) CURL_LIBCURL = -ldl +endif +endif else ifndef CURL_LDFLAGS CURL_LDFLAGS = $(eval CURL_LDFLAGS := $$(shell $$(CURL_CONFIG) --libs))$(CURL_LDFLAGS) diff --git a/compat/lazyload-curl.c b/compat/lazyload-curl.c index 19aa2b6d4b6942..98d73fb0f2a66f 100644 --- a/compat/lazyload-curl.c +++ b/compat/lazyload-curl.c @@ -1,6 +1,8 @@ #include "../git-compat-util.h" #include "../git-curl-compat.h" +#ifndef WIN32 #include +#endif /* * The ABI version of libcurl is encoded in its shared libraries' file names. @@ -11,6 +13,7 @@ typedef void (*func_t)(void); +#ifndef WIN32 #ifdef __APPLE__ #define LIBCURL_FILE_NAME(base) base "." LIBCURL_ABI_VERSION ".dylib" #else @@ -35,6 +38,55 @@ static func_t load_function(void *handle, const char *name) *(void **)&f = dlsym(handle, name); return f; } +#else +#define LIBCURL_FILE_NAME(base) base "-" LIBCURL_ABI_VERSION ".dll" + +static void *load_library(const char *name) +{ + size_t name_size = strlen(name) + 1; + const char *path = getenv("PATH"); + char dll_path[MAX_PATH]; + + while (path && *path) { + const char *sep = strchrnul(path, ';'); + size_t len = sep - path; + + if (len && len + name_size < sizeof(dll_path)) { + memcpy(dll_path, path, len); + dll_path[len] = '/'; + memcpy(dll_path + len + 1, name, name_size); + + if (!access(dll_path, R_OK)) { + wchar_t wpath[MAX_PATH]; + int wlen = MultiByteToWideChar(CP_UTF8, 0, dll_path, -1, wpath, ARRAY_SIZE(wpath)); + void *res = wlen ? (void *)LoadLibraryExW(wpath, NULL, 0) : NULL; + if (!res) { + DWORD err = GetLastError(); + char buf[1024]; + + if (!FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM | + FORMAT_MESSAGE_ARGUMENT_ARRAY | + FORMAT_MESSAGE_IGNORE_INSERTS, + NULL, err, LANG_NEUTRAL, + buf, sizeof(buf) - 1, NULL)) + xsnprintf(buf, sizeof(buf), "last error: %ld", err); + error("LoadLibraryExW() failed with: %s", buf); + } + return res; + } + } + + path = *sep ? sep + 1 : NULL; + } + + return NULL; +} + +static func_t load_function(void *handle, const char *name) +{ + return (func_t)GetProcAddress((HANDLE)handle, name); +} +#endif typedef struct curl_version_info_data *(*curl_version_info_type)(CURLversion version); static curl_version_info_type curl_version_info_func; From de2514d628203e7b4c83dcbafcf7a8a36e9d7da0 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sun, 7 May 2023 22:05:33 +0200 Subject: [PATCH 109/259] http: when loading libcurl lazily, allow for multiple SSL backends The previous commits introduced a compile-time option to load libcurl lazily, but it uses the hard-coded name "libcurl-4.dll" (or equivalent on platforms other than Windows). To allow for installing multiple libcurl flavors side by side, where each supports one specific SSL/TLS backend, let's first look whether `libcurl--4.dll` exists, and only use `libcurl-4.dll` as a fall back. That will allow us to ship with a libcurl by default that only supports the Secure Channel backend for the `https://` protocol. This libcurl won't suffer from any dependency problem when upgrading OpenSSL to a new major version (which will change the DLL name, and hence break every program and library that depends on it). This is crucial because Git for Windows relies on libcurl to keep working when building and deploying a new OpenSSL package because that library is used by `git fetch` and `git clone`. Note that this feature is by no means specific to Windows. On Ubuntu, for example, a `git` built using `LAZY_LOAD_LIBCURL` will use `libcurl.so.4` for `http.sslbackend=openssl` and `libcurl-gnutls.so.4` for `http.sslbackend=gnutls`. Signed-off-by: Johannes Schindelin --- compat/lazyload-curl.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/compat/lazyload-curl.c b/compat/lazyload-curl.c index 98d73fb0f2a66f..2f2b2ebc27993e 100644 --- a/compat/lazyload-curl.c +++ b/compat/lazyload-curl.c @@ -172,17 +172,26 @@ static curl_easy_setopt_pointer_type curl_easy_setopt_pointer_func; typedef CURLcode (*curl_easy_setopt_off_t_type)(CURL *curl, CURLoption opt, curl_off_t value); static curl_easy_setopt_off_t_type curl_easy_setopt_off_t_func; +static char ssl_backend[64]; + static void lazy_load_curl(void) { static int initialized; - void *libcurl; + void *libcurl = NULL; func_t curl_easy_getinfo_func, curl_easy_setopt_func; if (initialized) return; initialized = 1; - libcurl = load_library(LIBCURL_FILE_NAME("libcurl")); + if (ssl_backend[0]) { + char dll_name[64 + 16]; + snprintf(dll_name, sizeof(dll_name) - 1, + LIBCURL_FILE_NAME("libcurl-%s"), ssl_backend); + libcurl = load_library(dll_name); + } + if (!libcurl) + libcurl = load_library(LIBCURL_FILE_NAME("libcurl")); if (!libcurl) die("failed to load library '%s'", LIBCURL_FILE_NAME("libcurl")); @@ -246,6 +255,9 @@ CURLcode curl_global_init(long flags) CURLsslset curl_global_sslset(curl_sslbackend id, const char *name, const curl_ssl_backend ***avail) { + if (name && strlen(name) < sizeof(ssl_backend)) + strlcpy(ssl_backend, name, sizeof(ssl_backend)); + lazy_load_curl(); return curl_global_sslset_func(id, name, avail); } From 418bbe6136e9ca223c89c54f011366c245b37f0b Mon Sep 17 00:00:00 2001 From: Andrey Zabavnikov Date: Fri, 28 Oct 2022 17:12:06 +0300 Subject: [PATCH 110/259] status: fix for old-style submodules with commondir In f9b7573f6b00 (repository: free fields before overwriting them, 2017-09-05), Git was taught to release memory before overwriting it, but 357a03ebe9e0 (repository.c: move env-related setup code back to environment.c, 2018-03-03) changed the code so that it would not _always_ be overwritten. As a consequence, the `commondir` attribute would point to already-free()d memory. This seems not to cause problems in core Git, but there are add-on patches in Git for Windows where the `commondir` attribute is subsequently used and causing invalid memory accesses e.g. in setups containing old-style submodules (i.e. the ones with a `.git` directory within theirs worktrees) that have `commondir` configured. This fixes https://github.com/git-for-windows/git/pull/4083. Signed-off-by: Andrey Zabavnikov Signed-off-by: Johannes Schindelin --- repository.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/repository.c b/repository.c index 6cbaf2e3daa93a..6cf423da50adec 100644 --- a/repository.c +++ b/repository.c @@ -136,7 +136,7 @@ static void repo_set_commondir(struct repository *repo, { struct strbuf sb = STRBUF_INIT; - free(repo->commondir); + FREE_AND_NULL(repo->commondir); if (commondir) { repo->different_commondir = 1; From f7f17bc96ff9843873237a9318630608c3ef5da7 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Fri, 27 Jan 2023 08:55:21 +0100 Subject: [PATCH 111/259] windows: skip linking `git-` for built-ins It is merely a historical wart that, say, `git-commit` exists in the `libexec/git-core/` directory, a tribute to the original idea to let Git be essentially a bunch of Unix shell scripts revolving around very few "plumbing" (AKA low-level) commands. Git has evolved a lot from there. These days, most of Git's functionality is contained within the `git` executable, in the form of "built-in" commands. To accommodate for scripts that use the "dashed" form of Git commands, even today, Git provides hard-links that make the `git` executable available as, say, `git-commit`, just in case that an old script has not been updated to invoke `git commit`. Those hard-links do not come cheap: they take about half a minute for every build of Git on Windows, they are mistaken for taking up huge amounts of space by some Windows Explorer versions that do not understand hard-links, and therefore many a "bug" report had to be addressed. The "dashed form" has been officially deprecated in Git version 1.5.4, which was released on February 2nd, 2008, i.e. a very long time ago. This deprecation was never finalized by skipping these hard-links, but we can start the process now, in Git for Windows. Signed-off-by: Johannes Schindelin --- config.mak.uname | 2 ++ 1 file changed, 2 insertions(+) diff --git a/config.mak.uname b/config.mak.uname index 0229eb3b8fe21f..1f78fbad5f4067 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -489,6 +489,7 @@ ifeq ($(uname_S),Windows) NO_POSIX_GOODIES = UnfortunatelyYes NATIVE_CRLF = YesPlease DEFAULT_HELP_FORMAT = html + SKIP_DASHED_BUILT_INS = YabbaDabbaDoo ifeq (/mingw64,$(subst 32,64,$(subst clangarm,mingw,$(prefix)))) # Move system config into top-level /etc/ ETC_GITCONFIG = ../etc/gitconfig @@ -681,6 +682,7 @@ ifeq ($(uname_S),MINGW) FSMONITOR_DAEMON_BACKEND = win32 FSMONITOR_OS_SETTINGS = win32 + SKIP_DASHED_BUILT_INS = YabbaDabbaDoo RUNTIME_PREFIX = YesPlease HAVE_WPGMPTR = YesWeDo NO_ST_BLOCKS_IN_STRUCT_STAT = YesPlease From 285ca8a7e6c93e61d2a0a6373297902a018528c5 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 21 Mar 2023 16:14:44 +0100 Subject: [PATCH 112/259] windows: fix Repository>Explore Working Copy Since Git v2.39.1, we are a bit more stringent in searching the PATH. In particular, we specifically require the `.exe` suffix. However, the `Repository>Explore Working Copy` command asks for `explorer.exe` to be found on the `PATH`, which _already_ has that suffix. Let's unstartle the PATH-finding logic about this scenario. This fixes https://github.com/git-for-windows/git/issues/4356 Signed-off-by: Johannes Schindelin --- git-gui/git-gui.sh | 3 +++ 1 file changed, 3 insertions(+) diff --git a/git-gui/git-gui.sh b/git-gui/git-gui.sh index 887d6d596c16fc..c913a08566542e 100755 --- a/git-gui/git-gui.sh +++ b/git-gui/git-gui.sh @@ -101,6 +101,9 @@ proc _which {what args} { if {[is_Windows] && [lsearch -exact $args -script] >= 0} { set suffix {} + } elseif {[string match *$_search_exe $what]} { + # The search string already has the file extension + set suffix {} } else { set suffix $_search_exe } From f5418816915d1cee3c2f00b560f807f126476d11 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Sun, 7 May 2023 22:43:37 +0200 Subject: [PATCH 113/259] mingw: do load libcurl dynamically by default This will help with Git for Windows' maintenance going forward: It allows Git for Windows to switch its primary libcurl to a variant without the OpenSSL backend, while still loading an alternate when setting `http.sslBackend = openssl`. This is necessary to avoid maintenance headaches with upgrading OpenSSL: its major version name is encoded in the shared library's file name and hence major version updates (temporarily) break libraries that are linked against the OpenSSL library. Signed-off-by: Johannes Schindelin --- config.mak.uname | 1 + 1 file changed, 1 insertion(+) diff --git a/config.mak.uname b/config.mak.uname index 1736c4f0116e75..76b60c2fba8d57 100644 --- a/config.mak.uname +++ b/config.mak.uname @@ -692,6 +692,7 @@ ifeq ($(uname_S),MINGW) HAVE_PLATFORM_PROCINFO = YesPlease CSPRNG_METHOD = rtlgenrandom BASIC_LDFLAGS += -municode -Wl,--tsaware + LAZYLOAD_LIBCURL = YesDoThatPlease COMPAT_CFLAGS += -DNOGDI -Icompat -Icompat/win32 COMPAT_CFLAGS += -DSTRIP_EXTENSION=\".exe\" COMPAT_OBJS += compat/mingw.o compat/winansi.o \ From 11333966ea18ceb42e187ce162c96e3c4f315306 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 2 Nov 2022 16:23:58 +0100 Subject: [PATCH 114/259] Add a GitHub workflow to verify that Git/Scalar work in Nano Server In Git for Windows v2.39.0, we fixed a regression where `git.exe` would no longer work in Windows Nano Server (frequently used in Docker containers). This GitHub workflow can be used to verify manually that the Git/Scalar executables work in Nano Server. Signed-off-by: Johannes Schindelin --- .github/workflows/nano-server.yml | 76 +++++++++++++++++++++++++++++++ 1 file changed, 76 insertions(+) create mode 100644 .github/workflows/nano-server.yml diff --git a/.github/workflows/nano-server.yml b/.github/workflows/nano-server.yml new file mode 100644 index 00000000000000..3d943c23d2616d --- /dev/null +++ b/.github/workflows/nano-server.yml @@ -0,0 +1,76 @@ +name: Windows Nano Server tests + +on: + workflow_dispatch: + +env: + DEVELOPER: 1 + +jobs: + test-nano-server: + runs-on: windows-2022 + env: + WINDBG_DIR: "C:/Program Files (x86)/Windows Kits/10/Debuggers/x64" + IMAGE: mcr.microsoft.com/powershell:nanoserver-ltsc2022 + + steps: + - uses: actions/checkout@v4 + - uses: git-for-windows/setup-git-for-windows-sdk@v1 + - name: build Git + shell: bash + run: make -j15 + - name: pull nanoserver image + shell: bash + run: docker pull $IMAGE + - name: run nano-server test + shell: bash + run: | + docker run \ + --user "ContainerAdministrator" \ + -v "$WINDBG_DIR:C:/dbg" \ + -v "$(cygpath -aw /mingw64/bin):C:/mingw64-bin" \ + -v "$(cygpath -aw .):C:/test" \ + $IMAGE pwsh.exe -Command ' + # Extend the PATH to include the `.dll` files in /mingw64/bin/ + $env:PATH += ";C:\mingw64-bin" + + # For each executable to test pick some no-operation set of + # flags/subcommands or something that should quickly result in an + # error with known exit code that is not a negative 32-bit + # number, and set the expected return code appropriately. + # + # Only test executables that could be expected to run in a UI + # less environment. + # + # ( Executable path, arguments, expected return code ) + # also note space is required before close parenthesis (a + # powershell quirk when defining nested arrays like this) + + $executables_to_test = @( + ("C:\test\git.exe", "", 1 ), + ("C:\test\scalar.exe", "version", 0 ) + ) + + foreach ($executable in $executables_to_test) + { + Write-Output "Now testing $($executable[0])" + &$executable[0] $executable[1] + if ($LASTEXITCODE -ne $executable[2]) { + # if we failed, run the debugger to find out what function + # or DLL could not be found and then exit the script with + # failure The missing DLL or EXE will be referenced near + # the end of the output + + # Set a flag to have the debugger show loader stub + # diagnostics. This requires running as administrator, + # otherwise the flag will be ignored. + C:\dbg\gflags -i $executable[0] +SLS + + C:\dbg\cdb.exe -c "g" -c "q" $executable[0] $executable[1] + + exit 1 + } + } + + exit 0 + ' From 738b64da657a27dd437161739a949dcc5cd98b6a Mon Sep 17 00:00:00 2001 From: David Lomas Date: Fri, 28 Jul 2023 15:31:25 +0100 Subject: [PATCH 115/259] mingw: suggest `windows.appendAtomically` in more cases When running Git for Windows on a remote APFS filesystem, it would appear that the `mingw_open_append()`/`write()` combination would fail almost exactly like on some CIFS-mounted shares as had been reported in https://github.com/git-for-windows/git/issues/2753, albeit with a different `errno` value. Let's handle that `errno` value just the same, by suggesting to set `windows.appendAtomically=false`. Signed-off-by: David Lomas Signed-off-by: Johannes Schindelin --- compat/mingw.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/compat/mingw.c b/compat/mingw.c index 0658cdfc3c733f..adf2e423da539a 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -787,7 +787,7 @@ ssize_t mingw_write(int fd, const void *buf, size_t len) { ssize_t result = write(fd, buf, len); - if (result < 0 && (errno == EINVAL || errno == ENOSPC) && buf) { + if (result < 0 && (errno == EINVAL || errno == EBADF || errno == ENOSPC) && buf) { int orig = errno; /* check if fd is a pipe */ @@ -813,7 +813,7 @@ ssize_t mingw_write(int fd, const void *buf, size_t len) } errno = orig; - } else if (orig == EINVAL) + } else if (orig == EINVAL || errno == EBADF) errno = EPIPE; else { DWORD buf_size; From b554f2bb7a6da298a2ce73acf5b1a8c41348f41b Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Wed, 22 Nov 2023 22:57:38 +0100 Subject: [PATCH 116/259] win32: use native ANSI sequence processing, if possible Windows 10 version 1511 (also known as Anniversary Update), according to https://learn.microsoft.com/en-us/windows/console/console-virtual-terminal-sequences introduced native support for ANSI sequence processing. This allows using colors from the entire 24-bit color range. All we need to do is test whether the console's "virtual processing support" can be enabled. If it can, we do not even need to start the `console_thread` to handle ANSI sequences. Or, almost all we need to do: When `console_thread()` does its work, it uses the Unicode-aware `write_console()` function to write to the Win32 Console, which supports Git for Windows' implicit convention that all text that is written is encoded in UTF-8. The same is not necessarily true if native ANSI sequence processing is used, as the output is then subject to the current code page. Let's ensure that the code page is set to `CP_UTF8` as long as Git writes to it. Signed-off-by: Johannes Schindelin --- compat/winansi.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 46 insertions(+) diff --git a/compat/winansi.c b/compat/winansi.c index ac2ffb78691a7d..a83a7f47ada9b2 100644 --- a/compat/winansi.c +++ b/compat/winansi.c @@ -593,6 +593,49 @@ static void detect_msys_tty(int fd) #endif +static HANDLE std_console_handle; +static DWORD std_console_mode = ENABLE_VIRTUAL_TERMINAL_PROCESSING; +static UINT std_console_code_page = CP_UTF8; + +static void reset_std_console(void) +{ + if (std_console_mode != ENABLE_VIRTUAL_TERMINAL_PROCESSING) + SetConsoleMode(std_console_handle, std_console_mode); + if (std_console_code_page != CP_UTF8) + SetConsoleOutputCP(std_console_code_page); +} + +static int enable_virtual_processing(void) +{ + std_console_handle = GetStdHandle(STD_OUTPUT_HANDLE); + if (std_console_handle == INVALID_HANDLE_VALUE || + !GetConsoleMode(std_console_handle, &std_console_mode)) { + std_console_handle = GetStdHandle(STD_ERROR_HANDLE); + if (std_console_handle == INVALID_HANDLE_VALUE || + !GetConsoleMode(std_console_handle, &std_console_mode)) + return 0; + } + + std_console_code_page = GetConsoleOutputCP(); + if (std_console_code_page != CP_UTF8) + SetConsoleOutputCP(CP_UTF8); + if (!std_console_code_page) + std_console_code_page = CP_UTF8; + + atexit(reset_std_console); + + if (std_console_mode & ENABLE_VIRTUAL_TERMINAL_PROCESSING) + return 1; + + if (!SetConsoleMode(std_console_handle, + std_console_mode | + ENABLE_PROCESSED_OUTPUT | + ENABLE_VIRTUAL_TERMINAL_PROCESSING)) + return 0; + + return 1; +} + /* * Wrapper for isatty(). Most calls in the main git code * call isatty(1 or 2) to see if the instance is interactive @@ -631,6 +674,9 @@ void winansi_init(void) return; } + if (enable_virtual_processing()) + return; + /* create a named pipe to communicate with the console thread */ if (swprintf(name, ARRAY_SIZE(name) - 1, L"\\\\.\\pipe\\winansi%lu", GetCurrentProcessId()) < 0) From 0c971083088c815001385bd5f0f2039ffc79d8e8 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Matthias=20A=C3=9Fhauer?= Date: Sat, 2 Dec 2023 12:10:00 +0100 Subject: [PATCH 117/259] git.rc: include winuser.h MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit winuser.h contains the definition of RT_MANIFEST that our LLVM based toolchain needs to understand that we want to embed compat/win32/git.manifest as an application manifest. It currently just embeds it as additional data that Windows doesn't understand. This also helps our GCC based toolchain understand that we only want one copy embedded. It currently embeds one working assembly manifest and one nearly identical, but useless copy as additional data. This also teaches our Visual Studio based buildsystems to pick up the manifest file from git.rc. This means we don't have to explicitly specify it in contrib/buildsystems/Generators/Vcxproj.pm anymore. Slightly counter-intuitively this also means we have to explicitly tell Cmake not to embed a default manifest. This fixes https://github.com/git-for-windows/git/issues/4707 Signed-off-by: Matthias Aßhauer Signed-off-by: Johannes Schindelin --- contrib/buildsystems/CMakeLists.txt | 1 + git.rc.in | 1 + 2 files changed, 2 insertions(+) diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt index 25b495fa737d07..5ac5850100033c 100644 --- a/contrib/buildsystems/CMakeLists.txt +++ b/contrib/buildsystems/CMakeLists.txt @@ -208,6 +208,7 @@ if(CMAKE_C_COMPILER_ID STREQUAL "MSVC") set(CMAKE_RUNTIME_OUTPUT_DIRECTORY_DEBUG ${CMAKE_BINARY_DIR}) set(CMAKE_RUNTIME_OUTPUT_DIRECTORY_RELEASE ${CMAKE_BINARY_DIR}) add_compile_options(/MP /std:c11) + add_link_options(/MANIFEST:NO) endif() #default behaviour diff --git a/git.rc.in b/git.rc.in index e69444eef3f0c5..1d5b627b610549 100644 --- a/git.rc.in +++ b/git.rc.in @@ -1,3 +1,4 @@ +#include 1 VERSIONINFO FILEVERSION @GIT_MAJOR_VERSION@,@GIT_MINOR_VERSION@,@GIT_MICRO_VERSION@,@GIT_PATCH_LEVEL@ PRODUCTVERSION @GIT_MAJOR_VERSION@,@GIT_MINOR_VERSION@,@GIT_MICRO_VERSION@,@GIT_PATCH_LEVEL@ From 32e659362c94169a8e759a296213c6202db064ab Mon Sep 17 00:00:00 2001 From: MinarKotonoha Date: Mon, 8 Apr 2024 16:41:10 +0800 Subject: [PATCH 118/259] common-main.c: fflush stdout buffer upon exit By default, the buffer type of Windows' `stdout` is unbuffered (_IONBF), and there is no need to manually fflush `stdout`. But some programs, such as the Windows Filtering Platform driver provided by the security software, may change the buffer type of `stdout` to full buffering. This nees `fflush(stdout)` to be called manually, otherwise there will be no output to `stdout`. Signed-off-by: MinarKotonoha Signed-off-by: Johannes Schindelin --- common-exit.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/common-exit.c b/common-exit.c index 1aaa538be3ed67..609f32abed8b53 100644 --- a/common-exit.c +++ b/common-exit.c @@ -11,6 +11,13 @@ static void check_bug_if_BUG(void) /* We wrap exit() to call common_exit() in git-compat-util.h */ int common_exit(const char *file, int line, int code) { + /* + * Windows Filtering Platform driver provided by the security software + * may change buffer type of stdout from _IONBF to _IOFBF. + * It will no output without fflush manually. + */ + fflush(stdout); + /* * For non-POSIX systems: Take the lowest 8 bits of the "code" * to e.g. turn -1 into 255. On a POSIX system this is From 83013cc868a9a646f3db91b18f43c91b1b4be715 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 9 Apr 2024 16:50:56 +0200 Subject: [PATCH 119/259] t5601/t7406(mingw): do run tests with symlink support A long time ago, we decided to run tests in Git for Windows' SDK with the default `winsymlinks` mode: copying instead of linking. This is still the default mode of MSYS2 to this day. However, this is not how most users run Git for Windows: As the majority of Git for Windows' users seem to be on Windows 10 and newer, likely having enabled Developer Mode (which allows creating symbolic links without administrator privileges), they will run with symlink support enabled. This is the reason why it is crucial to get the fixes for CVE-2024-? to the users, and also why it is crucial to ensure that the test suite exercises the related test cases. This commit ensures the latter. Signed-off-by: Johannes Schindelin --- t/t5601-clone.sh | 10 ++++++++++ t/t7406-submodule-update.sh | 9 +++++++++ 2 files changed, 19 insertions(+) diff --git a/t/t5601-clone.sh b/t/t5601-clone.sh index d0c18660e33113..a529916e71a4de 100755 --- a/t/t5601-clone.sh +++ b/t/t5601-clone.sh @@ -7,6 +7,16 @@ export GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME . ./test-lib.sh +# This test script contains test cases that need to create symbolic links. To +# make sure that these test cases are exercised in Git for Windows, where (for +# historical reasons) `ln -s` creates copies by default, let's specifically ask +# for `ln -s` to create symbolic links whenever possible. +if test_have_prereq MINGW +then + MSYS=${MSYS+$MSYS }winsymlinks:nativestrict + export MSYS +fi + X= test_have_prereq !MINGW || X=.exe diff --git a/t/t7406-submodule-update.sh b/t/t7406-submodule-update.sh index c562bad042ab2d..38bfaeb510a2fb 100755 --- a/t/t7406-submodule-update.sh +++ b/t/t7406-submodule-update.sh @@ -14,6 +14,15 @@ export GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME . ./test-lib.sh +# This test script contains test cases that need to create symbolic links. To +# make sure that these test cases are exercised in Git for Windows, where (for +# historical reasons) `ln -s` creates copies by default, let's specifically ask +# for `ln -s` to create symbolic links whenever possible. +if test_have_prereq MINGW +then + MSYS=${MSYS+$MSYS }winsymlinks:nativestrict + export MSYS +fi compare_head() { From 31f14a71f78f585b8cec5ae74868e952bbe02eb0 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 21 May 2024 13:55:26 +0200 Subject: [PATCH 120/259] win32: ensure that `localtime_r()` is declared even in i686 builds The `__MINGW64__` constant is defined, surprise, surprise, only when building for a 64-bit CPU architecture. Therefore using it as a guard to define `_POSIX_C_SOURCE` (so that `localtime_r()` is declared, among other functions) is not enough, we also need to check `__MINGW32__`. Technically, the latter constant is defined even for 64-bit builds. But let's make things a bit easier to understand by testing for both constants. Making it so fixes this compile warning (turned error in GCC v14.1): archive-zip.c: In function 'dos_time': archive-zip.c:612:9: error: implicit declaration of function 'localtime_r'; did you mean 'localtime_s'? [-Wimplicit-function-declaration] 612 | localtime_r(&time, &tm); | ^~~~~~~~~~~ | localtime_s Signed-off-by: Johannes Schindelin --- git-compat-util.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/git-compat-util.h b/git-compat-util.h index e123288e8f1393..e63517630582ac 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -176,7 +176,7 @@ DISABLE_WARNING(-Wsign-compare) /* Approximation of the length of the decimal representation of this type. */ #define decimal_length(x) ((int)(sizeof(x) * 2.56 + 0.5) + 1) -#ifdef __MINGW64__ +#if defined(__MINGW32__) || defined(__MINGW64__) #define _POSIX_C_SOURCE 1 #elif defined(__sun__) /* From d6aa1bbe43eff3c9a9f36d9b7443306316356bec Mon Sep 17 00:00:00 2001 From: Ariel Lourenco Date: Tue, 2 Jul 2024 18:09:43 -0300 Subject: [PATCH 121/259] Fallback to AppData if XDG_CONFIG_HOME is unset In order to be a better Windows citizenship, Git should save its configuration files on AppData folder. This can enables git configuration files be replicated between machines using the same Microsoft account logon which would reduce the friction of setting up Git on new systems. Therefore, if %APPDATA%\Git\config exists, we use it; otherwise $HOME/.config/git/config is used. Signed-off-by: Ariel Lourenco --- path.c | 23 ++++++++++++++++++++--- 1 file changed, 20 insertions(+), 3 deletions(-) diff --git a/path.c b/path.c index 910756c8b3249c..51295e3ade7531 100644 --- a/path.c +++ b/path.c @@ -1517,6 +1517,7 @@ int looks_like_command_line_option(const char *str) char *xdg_config_home_for(const char *subdir, const char *filename) { const char *home, *config_home; + char *home_config = NULL; assert(subdir); assert(filename); @@ -1525,10 +1526,26 @@ char *xdg_config_home_for(const char *subdir, const char *filename) return mkpathdup("%s/%s/%s", config_home, subdir, filename); home = getenv("HOME"); - if (home) - return mkpathdup("%s/.config/%s/%s", home, subdir, filename); + if (home && *home) + home_config = mkpathdup("%s/.config/%s/%s", home, subdir, filename); + + #ifdef WIN32 + { + const char *appdata = getenv("APPDATA"); + if (appdata && *appdata) { + char *appdata_config = mkpathdup("%s/Git/%s", appdata, filename); + if (file_exists(appdata_config)) { + if (home_config && file_exists(home_config)) + warning("'%s' was ignored because '%s' exists.", home_config, appdata_config); + free(home_config); + return appdata_config; + } + free(appdata_config); + } + } + #endif - return NULL; + return home_config; } char *xdg_config_home(const char *filename) From e28c964c975fc13bd81c122a36fa0a318c0af5ae Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Thu, 4 Jul 2024 22:41:56 +0200 Subject: [PATCH 122/259] run-command: be helpful with Git LFS fails on Windows 7 Git LFS is now built with Go 1.21 which no longer supports Windows 7. However, Git for Windows still wants to support Windows 7. Ideally, Git LFS would re-introduce Windows 7 support until Git for Windows drops support for Windows 7, but that's not going to happen: https://github.com/git-for-windows/git/issues/4996#issuecomment-2176152565 The next best thing we can do is to let the users know what is happening, and how to get out of their fix, at least. This is not quite as easy as it would first seem because programs compiled with Go 1.21 or newer will simply throw an exception and fail with an Access Violation on Windows 7. The only way I found to address this is to replicate the logic from Go's very own `version` command (which can determine the Go version with which a given executable was built) to detect the situation, and in that case offer a helpful error message. This addresses https://github.com/git-for-windows/git/issues/4996. Signed-off-by: Johannes Schindelin --- compat/win32/path-utils.c | 199 ++++++++++++++++++++++++++++++++++++++ compat/win32/path-utils.h | 3 + git-compat-util.h | 7 ++ run-command.c | 1 + 4 files changed, 210 insertions(+) diff --git a/compat/win32/path-utils.c b/compat/win32/path-utils.c index 966ef779b9ca9b..c4fea0301b5ecc 100644 --- a/compat/win32/path-utils.c +++ b/compat/win32/path-utils.c @@ -2,6 +2,9 @@ #include "../../git-compat-util.h" #include "../../environment.h" +#include "../../wrapper.h" +#include "../../strbuf.h" +#include "../../versioncmp.h" int win32_has_dos_drive_prefix(const char *path) { @@ -89,3 +92,199 @@ int win32_fspathcmp(const char *a, const char *b) { return win32_fspathncmp(a, b, (size_t)-1); } + +static int read_at(int fd, char *buffer, size_t offset, size_t size) +{ + if (lseek(fd, offset, SEEK_SET) < 0) { + fprintf(stderr, "could not seek to 0x%x\n", (unsigned int)offset); + return -1; + } + + return read_in_full(fd, buffer, size); +} + +static size_t le16(const char *buffer) +{ + unsigned char *u = (unsigned char *)buffer; + return u[0] | (u[1] << 8); +} + +static size_t le32(const char *buffer) +{ + return le16(buffer) | (le16(buffer + 2) << 16); +} + +/* + * Determine the Go version of a given executable, if it was built with Go. + * + * This recapitulates the logic from + * https://github.com/golang/go/blob/master/src/cmd/go/internal/version/version.go + * (without requiring the user to install `go.exe` to find out). + */ +static ssize_t get_go_version(const char *path, char *go_version, size_t go_version_size) +{ + int fd = open(path, O_RDONLY); + char buffer[1024]; + off_t offset; + size_t num_sections, opt_header_size, i; + char *p = NULL, *q; + ssize_t res = -1; + + if (fd < 0) + return -1; + + if (read_in_full(fd, buffer, 2) < 0) + goto fail; + + /* + * Parse the PE file format, for more details, see + * https://en.wikipedia.org/wiki/Portable_Executable#Layout and + * https://learn.microsoft.com/en-us/windows/win32/debug/pe-format + */ + if (buffer[0] != 'M' || buffer[1] != 'Z') + goto fail; + + if (read_at(fd, buffer, 0x3c, 4) < 0) + goto fail; + + /* Read the `PE\0\0` signature and the COFF file header */ + offset = le32(buffer); + if (read_at(fd, buffer, offset, 24) < 0) + goto fail; + + if (buffer[0] != 'P' || buffer[1] != 'E' || buffer[2] != '\0' || buffer[3] != '\0') + goto fail; + + num_sections = le16(buffer + 6); + opt_header_size = le16(buffer + 20); + offset += 24; /* skip file header */ + + /* + * Validate magic number 0x10b or 0x20b, for full details see + * https://learn.microsoft.com/en-us/windows/win32/debug/pe-format#optional-header-standard-fields-image-only + */ + if (read_at(fd, buffer, offset, 2) < 0 || + ((i = le16(buffer)) != 0x10b && i != 0x20b)) + goto fail; + + offset += opt_header_size; + + for (i = 0; i < num_sections; i++) { + if (read_at(fd, buffer, offset + i * 40, 40) < 0) + goto fail; + + /* + * For full details about the section headers, see + * https://learn.microsoft.com/en-us/windows/win32/debug/pe-format#section-table-section-headers + */ + if ((le32(buffer + 36) /* characteristics */ & ~0x600000) /* IMAGE_SCN_ALIGN_32BYTES */ == + (/* IMAGE_SCN_CNT_INITIALIZED_DATA */ 0x00000040 | + /* IMAGE_SCN_MEM_READ */ 0x40000000 | + /* IMAGE_SCN_MEM_WRITE */ 0x80000000)) { + size_t size = le32(buffer + 16); /* "SizeOfRawData " */ + size_t pointer = le32(buffer + 20); /* "PointerToRawData " */ + + /* + * Skip the section if either size or pointer is 0, see + * https://github.com/golang/go/blob/go1.21.0/src/debug/buildinfo/buildinfo.go#L333 + * for full details. + * + * Merely seeing a non-zero size will not actually do, + * though: he size must be at least `buildInfoSize`, + * i.e. 32, and we expect a UVarint (at least another + * byte) _and_ the bytes representing the string, + * which we expect to start with the letters "go" and + * continue with the Go version number. + */ + if (size < 32 + 1 + 2 + 1 || !pointer) + continue; + + p = malloc(size); + + if (!p || read_at(fd, p, pointer, size) < 0) + goto fail; + + /* + * Look for the build information embedded by Go, see + * https://github.com/golang/go/blob/go1.21.0/src/debug/buildinfo/buildinfo.go#L165-L175 + * for full details. + * + * Note: Go contains code to enforce alignment along a + * 16-byte boundary. In practice, no `.exe` has been + * observed that required any adjustment, therefore + * this here code skips that logic for simplicity. + */ + q = memmem(p, size - 18, "\xff Go buildinf:", 14); + if (!q) + goto fail; + /* + * Decode the build blob. For full details, see + * https://github.com/golang/go/blob/go1.21.0/src/debug/buildinfo/buildinfo.go#L177-L191 + * + * Note: The `endianness` values observed in practice + * were always 2, therefore the complex logic to handle + * any other value is skipped for simplicty. + */ + if ((q[14] == 8 || q[14] == 4) && q[15] == 2) { + /* + * Only handle a Go version string with fewer + * than 128 characters, so the Go UVarint at + * q[32] that indicates the string's length must + * be only one byte (without the high bit set). + */ + if ((q[32] & 0x80) || + !q[32] || + (q + 33 + q[32] - p) > (ssize_t)size || + q[32] + 1 > (ssize_t)go_version_size) + goto fail; + res = q[32]; + memcpy(go_version, q + 33, res); + go_version[res] = '\0'; + break; + } + } + } + +fail: + free(p); + close(fd); + return res; +} + +void win32_warn_about_git_lfs_on_windows7(int exit_code, const char *argv0) +{ + char buffer[128], *git_lfs = NULL; + const char *p; + + /* + * Git LFS v3.5.1 fails with an Access Violation on Windows 7; That + * would usually show up as an exit code 0xc0000005. For some reason + * (probably because at this point, we no longer have the _original_ + * HANDLE that was returned by `CreateProcess()`) we observe other + * values like 0xb00 and 0x2 instead. Since the exact exit code + * seems to be inconsistent, we check for a non-zero exit status. + */ + if (exit_code == 0) + return; + if (GetVersion() >> 16 > 7601) + return; /* Warn only on Windows 7 or older */ + if (!istarts_with(argv0, "git-lfs ") && + strcasecmp(argv0, "git-lfs")) + return; + if (!(git_lfs = locate_in_PATH("git-lfs"))) + return; + if (get_go_version(git_lfs, buffer, sizeof(buffer)) > 0 && + skip_prefix(buffer, "go", &p) && + versioncmp("1.21.0", p) <= 0) + warning("This program was built with Go v%s\n" + "i.e. without support for this Windows version:\n" + "\n\t%s\n" + "\n" + "To work around this, you can download and install a " + "working version from\n" + "\n" + "\thttps://github.com/git-lfs/git-lfs/releases/tag/" + "v3.4.1\n", + p, git_lfs); + free(git_lfs); +} diff --git a/compat/win32/path-utils.h b/compat/win32/path-utils.h index a561c700e75713..a69483c332c1a7 100644 --- a/compat/win32/path-utils.h +++ b/compat/win32/path-utils.h @@ -34,4 +34,7 @@ int win32_fspathcmp(const char *a, const char *b); int win32_fspathncmp(const char *a, const char *b, size_t count); #define fspathncmp win32_fspathncmp +void win32_warn_about_git_lfs_on_windows7(int exit_code, const char *argv0); +#define warn_about_git_lfs_on_windows7 win32_warn_about_git_lfs_on_windows7 + #endif diff --git a/git-compat-util.h b/git-compat-util.h index e123288e8f1393..8a4381fda9eb04 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -537,6 +537,13 @@ static inline int git_offset_1st_component(const char *path) #define fspathncmp git_fspathncmp #endif +#ifndef warn_about_git_lfs_on_windows7 +static inline void warn_about_git_lfs_on_windows7(int exit_code UNUSED, + const char *argv0 UNUSED) +{ +} +#endif + #ifndef is_valid_path #define is_valid_path(path) 1 #endif diff --git a/run-command.c b/run-command.c index 402138b8b53ae8..ca546563f985c3 100644 --- a/run-command.c +++ b/run-command.c @@ -576,6 +576,7 @@ static int wait_or_whine(pid_t pid, const char *argv0, int in_signal) */ code += 128; } else if (WIFEXITED(status)) { + warn_about_git_lfs_on_windows7(status, argv0); code = WEXITSTATUS(status); } else { if (!in_signal) From 9c7962f36a284943f3bb0a50093eff0a8f001c82 Mon Sep 17 00:00:00 2001 From: Johannes Schindelin Date: Tue, 24 Sep 2024 08:47:39 +0200 Subject: [PATCH 123/259] ci: work around a problem with HTTP/2 vs libcurl v8.10.0 As reported in https://lore.kernel.org/git/ZuPKvYP9ZZ2mhb4m@pks.im/, libcurl v8.10.0 had a regression that was picked up by Git's t5559.30 "large fetch-pack requests can be sent using chunked encoding". This bug was fixed in libcurl v8.10.1. Sadly, the macos-13 runner image was updated in the brief window between these two libcurl versions, breaking each and every CI build, as reported at https://github.com/git-for-windows/git/issues/5159. This would usually not matter, we would just ignore the failing CI builds until the macos-13 runner image is rebuilt in a couple of days, and then the CI builds would succeed again. However. As has become the custom, a surprise Git version was released, and now that Git for Windows wants to follow suit, since Git for Windows has this custom of trying to never release a version with a failing CI build, we _must_ work around it. This patch implements this work-around, basically for the sake of Git for Windows v2.46.2's CI build. Signed-off-by: Johannes Schindelin --- t/t5551-http-fetch-smart.sh | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/t/t5551-http-fetch-smart.sh b/t/t5551-http-fetch-smart.sh index 761fdfcfe6c49d..13ac44099d03c0 100755 --- a/t/t5551-http-fetch-smart.sh +++ b/t/t5551-http-fetch-smart.sh @@ -403,7 +403,15 @@ test_expect_success CMDLINE_LIMIT \ ) ' -test_expect_success 'large fetch-pack requests can be sent using chunked encoding' ' +# This is a temporary work-around for libcurl v8.10.0 on the macos-* runners; +# see https://github.com/git-for-windows/git/issues/5159 for full details +test_lazy_prereq UNBROKEN_HTTP2 ' + test "$HTTP_PROTO" = HTTP/2 && + test -z "$(brew info -q curl 2>/dev/null | + sed -n "/^Installed/{N;s/.*8\\.10\\.0.*/BROKEN HTTP2/p;}")" +' + +test_expect_success UNBROKEN_HTTP2 'large fetch-pack requests can be sent using chunked encoding' ' GIT_TRACE_CURL=true git -c http.postbuffer=65536 \ clone --bare "$HTTPD_URL/smart/repo.git" split.git 2>err && { From 8bc5a2898a42c9d91735fbd844ca7f59a169e3ba Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Fri, 6 Sep 2024 14:16:13 -0400 Subject: [PATCH 124/259] revision: create mark_trees_uninteresting_dense() The sparse tree walk algorithm was created in d5d2e93577e (revision: implement sparse algorithm, 2019-01-16) and involves using the mark_trees_uninteresting_sparse() method. This method takes a repository and an oidset of tree IDs, some of which have the UNINTERESTING flag and some of which do not. Create a method that has an equivalent set of preconditions but uses a "dense" walk (recursively visits all reachable trees, as long as they have not previously been marked UNINTERESTING). This is an important difference from mark_tree_uninteresting(), which short-circuits if the given tree has the UNINTERESTING flag. A use of this method will be added in a later change, with a condition set whether the sparse or dense approach should be used. Signed-off-by: Derrick Stolee --- revision.c | 15 +++++++++++++++ revision.h | 1 + 2 files changed, 16 insertions(+) diff --git a/revision.c b/revision.c index c4390f0938cbde..6c88841dcca7a4 100644 --- a/revision.c +++ b/revision.c @@ -220,6 +220,21 @@ static void add_children_by_path(struct repository *r, free_tree_buffer(tree); } +void mark_trees_uninteresting_dense(struct repository *r, + struct oidset *trees) +{ + struct object_id *oid; + struct oidset_iter iter; + + oidset_iter_init(trees, &iter); + while ((oid = oidset_iter_next(&iter))) { + struct tree *tree = lookup_tree(r, oid); + + if (tree->object.flags & UNINTERESTING) + mark_tree_contents_uninteresting(r, tree); + } +} + void mark_trees_uninteresting_sparse(struct repository *r, struct oidset *trees) { diff --git a/revision.h b/revision.h index 71e984c452b8d7..8938b2db112e2a 100644 --- a/revision.h +++ b/revision.h @@ -487,6 +487,7 @@ void put_revision_mark(const struct rev_info *revs, void mark_parents_uninteresting(struct rev_info *revs, struct commit *commit); void mark_tree_uninteresting(struct repository *r, struct tree *tree); +void mark_trees_uninteresting_dense(struct repository *r, struct oidset *trees); void mark_trees_uninteresting_sparse(struct repository *r, struct oidset *trees); void show_object_with_name(FILE *, struct object *, const char *); From 977c6d6744c9a06e7c3e39f3f6b4a71dc3d41885 Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Tue, 24 Sep 2024 13:45:19 -0400 Subject: [PATCH 125/259] pack-objects: extract should_attempt_deltas() This will be helpful in a future change. Signed-off-by: Derrick Stolee --- builtin/pack-objects.c | 53 +++++++++++++++++++++++------------------- 1 file changed, 29 insertions(+), 24 deletions(-) diff --git a/builtin/pack-objects.c b/builtin/pack-objects.c index 58a9b1612626e0..1d0992a8dac089 100644 --- a/builtin/pack-objects.c +++ b/builtin/pack-objects.c @@ -3196,6 +3196,33 @@ static int add_ref_tag(const char *tag UNUSED, const char *referent UNUSED, cons return 0; } +static int should_attempt_deltas(struct object_entry *entry) +{ + if (DELTA(entry)) + return 0; + + if (!entry->type_valid || + oe_size_less_than(&to_pack, entry, 50)) + return 0; + + if (entry->no_try_delta) + return 0; + + if (!entry->preferred_base) { + if (oe_type(entry) < 0) + die(_("unable to get type of object %s"), + oid_to_hex(&entry->idx.oid)); + } else if (oe_type(entry) < 0) { + /* + * This object is not found, but we + * don't have to include it anyway. + */ + return 0; + } + + return 1; +} + static void prepare_pack(int window, int depth) { struct object_entry **delta_list; @@ -3226,33 +3253,11 @@ static void prepare_pack(int window, int depth) for (i = 0; i < to_pack.nr_objects; i++) { struct object_entry *entry = to_pack.objects + i; - if (DELTA(entry)) - /* This happens if we decided to reuse existing - * delta from a pack. "reuse_delta &&" is implied. - */ - continue; - - if (!entry->type_valid || - oe_size_less_than(&to_pack, entry, 50)) + if (!should_attempt_deltas(entry)) continue; - if (entry->no_try_delta) - continue; - - if (!entry->preferred_base) { + if (!entry->preferred_base) nr_deltas++; - if (oe_type(entry) < 0) - die(_("unable to get type of object %s"), - oid_to_hex(&entry->idx.oid)); - } else { - if (oe_type(entry) < 0) { - /* - * This object is not found, but we - * don't have to include it anyway. - */ - continue; - } - } delta_list[n++] = entry; } From f9dec4144a01f8e1b1f13e06c1b9d1843e32835d Mon Sep 17 00:00:00 2001 From: Derrick Stolee Date: Thu, 5 Sep 2024 10:04:51 -0400 Subject: [PATCH 126/259] pack-objects: add --path-walk option In order to more easily compute delta bases among objects that appear at the exact same path, add a --path-walk option to 'git pack-objects'. This option will use the path-walk API instead of the object walk given by the revision machinery. Since objects will be provided in batches representing a common path, those objects can be tested for delta bases immediately instead of waiting for a sort of the full object list by name-hash. This has multiple benefits, including avoiding collisions by name-hash. The objects marked as UNINTERESTING are included in these batches, so we are guaranteeing some locality to find good delta bases. After the individual passes are done on a per-path basis, the default name-hash is used to find other opportunistic delta bases that did not match exactly by the full path name. RFC TODO: It is important to note that this option is inherently incompatible with using a bitmap index. This walk probably also does not work with other advanced features, such as delta islands. Getting ahead of myself, this option compares well with --full-name-hash when the packfile is large enough, but also performs at least as well as the default in all cases that I've seen. RFC TODO: this should probably be recording the batch locations to another list so they could be processed in a second phase using threads. RFC TODO: list some examples of how this outperforms previous pack-objects strategies. (This is coming in later commits that include performance test changes.) Signed-off-by: Derrick Stolee --- Documentation/git-pack-objects.adoc | 12 +- Documentation/technical/api-path-walk.adoc | 1 + builtin/pack-objects.c | 147 +++++++++++++++++++-- t/t5300-pack-object.sh | 17 +++ 4 files changed, 167 insertions(+), 10 deletions(-) diff --git a/Documentation/git-pack-objects.adoc b/Documentation/git-pack-objects.adoc index 7f69ae4855f6ce..a91aa7989cbeff 100644 --- a/Documentation/git-pack-objects.adoc +++ b/Documentation/git-pack-objects.adoc @@ -16,7 +16,7 @@ SYNOPSIS [--cruft] [--cruft-expiration=