DWARF input format #55450

al13n321 · 2023-10-10T07:04:33Z

Changelog category (leave one):

New Feature

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

Added DWARF input format that reads debug symbols from an ELF executable/library/object file.

Documentation entry for user-facing changes

Documentation is written (mandatory for new features)

Because grepping dwarfdump output is very slow and inconvenient.

robot-clickhouse · 2023-10-10T07:07:25Z

This is an automated comment for commit abd35e0 with description of existing statuses. It's updated for the latest CI running

✅ Click here to open a full report in a separate page

Successful checks

Check name	Description	Status
AST fuzzer	Runs randomly generated queries to catch program errors. The build type is optionally given in parenthesis. If it fails, ask a maintainer for help	✅ success
CI running	A meta-check that indicates the running CI. Normally, it's in success or pending state. The failed status indicates some problems with the PR	✅ success
ClickHouse build check	Builds ClickHouse in various configurations for use in further steps. You have to fix the builds that fail. Build logs often has enough information to fix the error, but you might have to reproduce the failure locally. The cmake options can be found in the build log, grepping for cmake. Use these options and follow the general build process	✅ success
Compatibility check	Checks that clickhouse binary runs on distributions with old libc versions. If it fails, ask a maintainer for help	✅ success
Docker image for servers	The check to build and optionally push the mentioned image to docker hub	✅ success
Docs Check	Builds and tests the documentation	✅ success
Fast test	Normally this is the first check that is ran for a PR. It builds ClickHouse and runs most of stateless functional tests, omitting some. If it fails, further checks are not started until it is fixed. Look at the report to see which tests fail, then reproduce the failure locally as described here	✅ success
Flaky tests	Checks if new added or modified tests are flaky by running them repeatedly, in parallel, with more randomization. Functional tests are run 100 times with address sanitizer, and additional randomization of thread scheduling. Integrational tests are run up to 10 times. If at least once a new test has failed, or was too long, this check will be red. We don't allow flaky tests, read the doc	✅ success
Install packages	Checks that the built packages are installable in a clear environment	✅ success
Integration tests	The integration tests report. In parenthesis the package type is given, and in square brackets are the optional part/total tests	✅ success
Mergeable Check	Checks if all other necessary checks are successful	✅ success
Performance Comparison	Measure changes in query performance. The performance test report is described in detail here. In square brackets are the optional part/total tests	✅ success
Push to Dockerhub	The check for building and pushing the CI related docker images to docker hub	✅ success
SQLTest	There's no description for the check yet, please add it to tests/ci/ci_config.py:CHECK_DESCRIPTIONS	✅ success
SQLancer	Fuzzing tests that detect logical bugs with SQLancer tool	✅ success
Sqllogic	Run clickhouse on the sqllogic test set against sqlite and checks that all statements are passed	✅ success
Stateful tests	Runs stateful functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stateless tests	Runs stateless functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stress test	Runs stateless functional tests concurrently from several clients to detect concurrency-related errors	✅ success
Style Check	Runs a set of checks to keep the code style clean. If some of tests failed, see the related log from the report	✅ success
Unit tests	Runs the unit tests for different release types	✅ success
Upgrade check	Runs stress tests on server version from last release and then tries to upgrade it to the version from the PR. It checks if the new server can successfully startup without any errors, crashes or sanitizer asserts	✅ success

src/Processors/Formats/Impl/DWARFBlockInputFormat.h

antonio2368 · 2023-10-11T14:49:33Z

src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp

+    auto tag_names = ColumnString::create();
+    /// Note: TagString() returns empty string for tags that don't exist, and tag 0 doesn't exist.
+    for (uint32_t tag = 0; tag <= UINT16_MAX; ++tag)
+        append(tag_names, removePrefix(llvm::dwarf::TagString(tag), strlen("DW_TAG_")));


we will repeat strlen for each tag, no? why not store the prefix_size in variable?

Compilers usually optimize it out: https://godbolt.org/z/M7dTqPvKK . Even if not (e.g. in debug build), doing a short strlen 64 K times is cheap enough.

So I thought it's not worth the extra code to avoid this... except I didn't realize that it's only 1 line of extra code. Guess I'll add it, just to prevent the next reader of the code from wondering about this.

src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp

antonio2368 · 2023-10-12T08:27:38Z

src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp

+    if (need[COL_ATTR_FORM] || need[COL_ATTR_INT] || need[COL_ATTR_STR])
+        need[COL_ATTR_NAME] = true;
+    if (need[COL_ANCESTOR_OFFSETS])
+        need[COL_ANCESTOR_TAGS] = true;


can these be a simple assignment?

Suggested change

if (need[COL_ATTR_FORM] || need[COL_ATTR_INT] || need[COL_ATTR_STR])

need[COL_ATTR_NAME] = true;

if (need[COL_ANCESTOR_OFFSETS])

need[COL_ANCESTOR_TAGS] = true;

need[COL_ATTR_NAME] = need[COL_ATTR_FORM] || need[COL_ATTR_INT] || need[COL_ATTR_STR];

need[COL_ANCESTOR_TAGS] = need[COL_ANCESTOR_OFFSETS];

(That would be |=, which annoyingly produces compiler warnings-treated-as-errors for bool. Will change to x = x || y, which still looks nicer than the ifs, I guess.)

antonio2368 · 2023-10-12T08:36:52Z

src/Processors/Formats/Impl/DWARFBlockInputFormat.cpp

+                        {
+                            if (attr.Attr == llvm::dwarf::DW_AT_language) // programming language
+                                append(col_attr_str, removePrefix(llvm::dwarf::LanguageString(static_cast<uint32_t>(val.getRawUValue())),
+                                    strlen("DW_LANG_")));


does it make sense to create static constexpr std::string_view for these common prefixes so you can simply call .size() method on them?

Oh, nice trick, constexpr size_t x = strlen("foo") doesn't work (lame) but constexpr std::string_view x = "foo"; ... x.size() does.

al13n321 · 2023-10-12T16:36:00Z

Uh oh, MSAN fails with this nonsense:

Oct 12 05:06:26   ORIGIN: invalid (0). Might be a bug in MemorySanitizer origin tracking.
Oct 12 05:06:26     This could still be a bug in your code, too!

They can't distinguish between a compiler bug and a program bug! What kind of design is that. Why do I keep running into sanitizer-related compiler bugs.

https://s3.amazonaws.com/clickhouse-builds/PRs/55450/637822e163aac09f0fed7badcf609d0fae03c2e4/package_msan/build_log.log

al13n321 · 2023-10-12T17:48:59Z

cmake/sanitize.cmake

@@ -54,9 +54,6 @@ if (SANITIZE)
        set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${SAN_FLAGS} ${UBSAN_FLAGS}")
        set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${SAN_FLAGS} ${UBSAN_FLAGS}")

-        # llvm-tblgen, that is used during LLVM build, doesn't work with UBSan.
-        set (ENABLE_EMBEDDED_COMPILER 0 CACHE BOOL "")


This seemed redundant with the check in contrib/llvm-project-cmake/CMakeLists.txt, moved the comment into there.

al13n321 · 2023-10-12T17:49:58Z

contrib/llvm-project-cmake/CMakeLists.txt

 else()
-   set (ENABLE_EMBEDDED_COMPILER_DEFAULT ON)
+    set (ENABLE_EMBEDDED_COMPILER_DEFAULT ${ENABLE_LIBRARIES})


Made ENABLE_LIBRARIES=0 imply ENABLE_EMBEDDED_COMPILER=0 by default. Hope there isn't a good reason why it wasn't done this way originally.

al13n321 · 2023-10-12T17:55:51Z

tests/instructions/syntax.txt

@@ -1,6 +1,5 @@
-# Quick syntax check (2 minutes on 16-core server)
+# Relatively quick syntax check (20 minutes on 16-core server)


20 minutes is how long it took for me (on master branch without this PR). Did clang get 10x slower in 5 years? :o (More than that because computers got faster. But less because the code got bigger. Let's guess that these two effects roughly cancel out.) I heard that llvm gets slower with each version, but would have expected less than 1.6x/year.

al13n321 · 2023-10-16T23:59:48Z

Some possible features for future:

Decompressing ELF sections. E.g. libc debug symbols binaries at /usr/lib/debug/.build-id/*/*.debug have compressed .debug_* sections (flag ELFCOMPRESS_ZLIB), at least on Ubuntu on my machine.
Parse .gnu_debuglink section and load symbols from a different file (e.g. libc library has no symbols but has .gnu_debuglink pointing to somewhere in /usr/lib/debug/.build-id/*/*.debug to an ELF file with all the .debug_* sections). Maybe don't actually open the other file, that sounds sketchy, just throw an exception with the path of the file with symbols.
Parse locations (DW_AT_location/DW_AT_frame_base) into a column, maybe pretty-printing DWARF expressions.

robot-clickhouse added the pr-feature Pull request with new product feature label Oct 10, 2023

al13n321 force-pushed the dwarf branch from b03acd7 to c0b2864 Compare October 10, 2023 09:59

al13n321 added the force tests Force test ignoring fast test output. label Oct 10, 2023

al13n321 force-pushed the dwarf branch from c0b2864 to 8a96915 Compare October 10, 2023 10:02

al13n321 added the can be tested Allows running workflows for external contributors label Oct 10, 2023

al13n321 force-pushed the dwarf branch 2 times, most recently from ffaf0de to 782f24d Compare October 10, 2023 20:49

al13n321 removed the force tests Force test ignoring fast test output. label Oct 10, 2023

al13n321 marked this pull request as ready for review October 10, 2023 20:50

al13n321 force-pushed the dwarf branch 3 times, most recently from 150e193 to a4d4c1c Compare October 11, 2023 03:37

robot-clickhouse-ci-2 added the submodule changed At least one submodule changed in this PR. label Oct 11, 2023

al13n321 force-pushed the dwarf branch from a4d4c1c to 6c87277 Compare October 11, 2023 05:05

antonio2368 self-assigned this Oct 11, 2023

al13n321 force-pushed the dwarf branch 3 times, most recently from e405ad1 to 637822e Compare October 12, 2023 04:03

antonio2368 reviewed Oct 12, 2023

View reviewed changes

al13n321 force-pushed the dwarf branch from 637822e to ab036f4 Compare October 12, 2023 16:44

al13n321 commented Oct 12, 2023

View reviewed changes

al13n321 force-pushed the dwarf branch from 0ad07da to d52c9ef Compare October 14, 2023 08:03

antonio2368 approved these changes Oct 16, 2023

View reviewed changes

al13n321 added 3 commits October 16, 2023 20:01

Add ReadBufferFromFileBase::isRegularLocalFile()

a8149a9

DWARF input format

01ea732

Review comments

b0e90a9

al13n321 added 3 commits October 16, 2023 20:01

Changed things around ENABLE_EMBEDDED_COMPILER build setting

6222680

Added 'ranges' column

6fe0db5

no-msan no-ubsan

abd35e0

al13n321 force-pushed the dwarf branch from d52c9ef to abd35e0 Compare October 16, 2023 20:02

al13n321 merged commit ce7eca0 into master Oct 17, 2023
283 checks passed

al13n321 deleted the dwarf branch October 17, 2023 00:00

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

DWARF input format #55450

DWARF input format #55450

al13n321 commented Oct 10, 2023 •

edited

robot-clickhouse commented Oct 10, 2023 •

edited by robot-ch-test-poll

antonio2368 Oct 11, 2023

al13n321 Oct 12, 2023 •

edited

antonio2368 Oct 12, 2023

al13n321 Oct 12, 2023

antonio2368 Oct 12, 2023

al13n321 Oct 12, 2023

al13n321 commented Oct 12, 2023

al13n321 Oct 12, 2023

al13n321 Oct 12, 2023

al13n321 Oct 12, 2023 •

edited

al13n321 commented Oct 16, 2023

		@@ -1,6 +1,5 @@
		# Quick syntax check (2 minutes on 16-core server)
		# Relatively quick syntax check (20 minutes on 16-core server)

DWARF input format #55450

DWARF input format #55450

Conversation

al13n321 commented Oct 10, 2023 • edited

Changelog category (leave one):

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

Documentation entry for user-facing changes

robot-clickhouse commented Oct 10, 2023 • edited by robot-ch-test-poll

antonio2368 Oct 11, 2023

Choose a reason for hiding this comment

al13n321 Oct 12, 2023 • edited

Choose a reason for hiding this comment

antonio2368 Oct 12, 2023

Choose a reason for hiding this comment

al13n321 Oct 12, 2023

Choose a reason for hiding this comment

antonio2368 Oct 12, 2023

Choose a reason for hiding this comment

al13n321 Oct 12, 2023

Choose a reason for hiding this comment

al13n321 commented Oct 12, 2023

al13n321 Oct 12, 2023

Choose a reason for hiding this comment

al13n321 Oct 12, 2023

Choose a reason for hiding this comment

al13n321 Oct 12, 2023 • edited

Choose a reason for hiding this comment

al13n321 commented Oct 16, 2023

al13n321 commented Oct 10, 2023 •

edited

robot-clickhouse commented Oct 10, 2023 •

edited by robot-ch-test-poll

al13n321 Oct 12, 2023 •

edited

al13n321 Oct 12, 2023 •

edited