ci: Travis: use gcc9 with gcov job #10480

blueyed · 2019-07-12T13:14:46Z

it is useful to test with the latest gcc in general
it might help with gcov issues (merge mismatch)

blueyed · 2019-07-12T14:12:48Z

Fails:

error: self-comparison always evaluates to true [-Werror=tautological-compare]

Could not reproduce this locally, so that might be something Arch's gcc has a patch for already.

Currently using gcc-9 amd64 9.1.0-2ubuntu2~16.04.

[139/262] Building C object src/nvim/CMakeFiles/nvim.dir/window.c.o
FAILED: /usr/lib/ccache/gcc-9 -DINCLUDE_GENERATED_DECLARATIONS -DMIN_LOG_LEVEL=3 -DNVIM_MSGPACK_HAS_FLOAT32 -DNVIM_UNIBI_HAS_VAR_FROM -DUSE_GCOV -D_GNU_SOURCE -Iconfig -I../src -Isrc/nvim/auto -Iinclude -I/home/travis/nvim-deps/usr/include/luajit-2.0 -isystem /home/travis/nvim-deps/usr/include --coverage -g   -Wall -Wextra -pedantic -Wno-unused-parameter -Wstrict-prototypes -std=gnu99 -Wshadow -Wconversion -Wmissing-prototypes -Wimplicit-fallthrough -Wvla -fstack-protector-strong -fdiagnostics-color=always -Wno-array-bounds -Werror  -Wno-conversion -MD -MT src/nvim/CMakeFiles/nvim.dir/window.c.o -MF src/nvim/CMakeFiles/nvim.dir/window.c.o.d -o src/nvim/CMakeFiles/nvim.dir/window.c.o   -c ../src/nvim/window.c
../src/nvim/window.c: In function ‘do_window’:
../src/nvim/window.c:183:33: error: self-comparison always evaluates to true [-Werror=tautological-compare]
  183 |     FOR_ALL_WINDOWS_IN_TAB(wp2, curtab) {
      |                                 ^~
../src/nvim/window.c: In function ‘win_valid’:
../src/nvim/window.c:1504:30: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 1504 |   FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                              ^~
../src/nvim/window.c: In function ‘win_count’:
../src/nvim/window.c:1536:30: Kerror: self-comparison always evaluates to true [-Werror=tautological-compare]
 1536 |   FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                              ^~
../src/nvim/window.c: In function ‘one_window’:
../src/nvim/window.c:2244:30: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 2244 |   FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                              ^~
../src/nvim/window.c: In function ‘buf_jump_open_win’:
../src/nvim/window.c:4411:32: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 4411 |     FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                                ^~
../src/nvim/window.c: In function ‘win_size_save’:
../src/nvim/window.c:4749:30: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 4749 |   FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                              ^~
../src/nvim/window.c: In function ‘win_size_restore’:
../src/nvim/window.c:4768:34: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 4768 |       FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                                  ^~
../src/nvim/window.c: In function ‘win_setminheight’:
../src/nvim/window.c:5223:32: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 5223 |     FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                                ^~
../src/nvim/window.c: In function ‘only_one_window’:
../src/nvim/window.c:5995:30: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 5995 |   FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                              ^~
../src/nvim/window.c: In function ‘win_id2win’:
../src/nvim/window.c:6691:30: error: self-comparison always evaluates to true [-Werror=tautological-compare]
 6691 |   FOR_ALL_WINDOWS_IN_TAB(wp, curtab) {
      |                              ^~
cc1: all warnings being treated as errors
[139/262] Building C object src/nvim/C...o/msgpack_lua_c_bindings.generated.c.o
ninja: build stopped: subcommand failed.

blueyed · 2019-07-12T15:24:37Z

Not sure though why I am not seeing this myself - appears to be legit after all, isn't it?

blueyed · 2019-07-12T16:21:31Z

Now:

In file included from ../src/nvim/eval.c:6112:
src/nvim/auto/funcs.generated.h: In function ‘hash_internal_func_gperf’:
src/nvim/auto/funcs.generated.h:84:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
   84 |         hval += asso_values[(unsigned char)str[13]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:86:7: note: here
   86 |       case 13:
      |       ^~~~
src/nvim/auto/funcs.generated.h:89:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
   89 |         hval += asso_values[(unsigned char)str[10]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:91:7: note: here
   91 |       case 10:
      |       ^~~~
src/nvim/auto/funcs.generated.h:92:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
   92 |         hval += asso_values[(unsigned char)str[9]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:94:7: note: here
   94 |       case 9:
      |       ^~~~
src/nvim/auto/funcs.generated.h:97:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
   97 |         hval += asso_values[(unsigned char)str[6]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:99:7: note: here
   99 |       case 6:
      |       ^~~~
src/nvim/auto/funcs.generated.h:100:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
  100 |         hval += asso_values[(unsigned char)str[5]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:102:7: note: here
  102 |       case 5:
      |       ^~~~
src/nvim/auto/funcs.generated.h:104:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
  104 |         hval += asso_values[(unsigned char)str[3]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:106:7: note: here
  106 |       case 3:
      |       ^~~~
src/nvim/auto/funcs.generated.h:107:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
  107 |         hval += asso_values[(unsigned char)str[2]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:109:7: note: here
  109 |       case 2:
      |       ^~~~
src/nvim/auto/funcs.generated.h:110:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
  110 |         hval += asso_values[(unsigned char)str[1]];
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/nvim/auto/funcs.generated.h:112:7: note: here
  112 |       case 1:
      |       ^~~~
../src/nvim/eval.c: At top level:
cc1: error: unrecognized command line option ‘-Wno-static-in-inline’ [-Werror]
cc1: all warnings being treated as errors
[179/262] Building C object src/nvim/CMakeFiles/nvim.dir/ex_cmds.c.o
ninja: build stopped: subcommand failed.
+exit 1
The command "ci/script.sh" exited with 1.

blueyed · 2019-07-12T17:56:43Z

The above should be handled through the PASSTHROUGH comment in the file.. looks like -Wimplicit-fallthrough=5 is used (requiring special attribute, not regexp comment).

blueyed · 2019-07-12T22:02:33Z

Looks like ccache is causing this?!

Temporary measure to avoid QuickBuild CI failure: 8:42:54,702 INFO - Executing post-execute action... 18:42:54,702 ERROR - Step 'master>buildall>build-node?testNode=freebsd-64>build-and-run-tests>build-and-run-tests-parameterized?buildType=Release>configure-neovim-and-build-nvim' is failed: Failed to run command: mkdir -p build/Release && cd build/Release && cmake -G "Unix Makefiles" -DBUSTED_OUTPUT_TYPE=TAP -DMIN_LOG_LEVEL=3 -DCMAKE_BUILD_TYPE=Release -DTRAVIS_CI_BUILD=ON ../.. && gmake VERBOSE=1 nvim unittest-prereqs functionaltest-prereqs Command return code: 2 Command error output: /usr/home/quickbuild/buildagent/workspace/root/neovim/pull-requests-automated/src/nvim/profile.c:70:27: error: implicit conversion changes signedness: 'proftime_T' (aka 'long') to 'unsigned long' [-Werror,-Wsign-conversion] STRICT_ADD(os_hrtime(), nsec, &rv, int64_t); ~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~ /usr/home/quickbuild/buildagent/workspace/root/neovim/pull-requests-automated/src/nvim/assert.h:150:26: note: expanded from macro 'STRICT_ADD' do { *(c) = (t)((a) + (b)); } while (0) ^ 1 error generated. gma...

This is the default since ccache 3.3, but Travis has 3.2.4. This fixes compiler warnings from macros with later clang/gcc. Unblocks neovim#10480, and neovim#10487. Ref: ccache/ccache@6d9cb3dfdd9

blueyed · 2019-07-17T16:48:10Z

Should be good/better after #10533.

This is the default since ccache 3.3, but Travis has 3.2.4. This fixes compiler warnings from macros with later clang/gcc. Using CCACHE_HASHDIR=1 fixes ccache v3.2.4, but CCACHE_NOHASHDIR=1 does not break v3.7.1. The real issue/fix appears to be ccache/ccache@284e3a0, and using the hashdir option seems to only work around this. Unblocks #10480, and #10487. Ref: ccache/ccache@6d9cb3dfdd9 Closes #10533.

blueyed · 2019-07-17T17:50:18Z

Might be good to test latest ccache in one job (this one) still?
(similar to testing oldest CMake)

Pulled that out, but happy to create a separate PR if you find it useful.

This is required to avoid warnings with newer gcc/clang. It is the correct fix - I must have messed up when bisecting. Follow-up to: neovim#10533

blueyed · 2019-07-17T22:52:19Z

.travis.yml

+    packages:
+      - ccache
+      - ninja
+


This is only moved to allow using its new anchor below.

justinmk

Ok, but I hope we don't do more of this type of thing: It forces us to spend too much time managing compiler versions. Travis should be doing that for us.

Also, adding extra apt sources was a source of pain in the past.

blueyed · 2019-07-17T23:26:04Z

It forces us to spend too much time managing compiler versions. Travis should be doing that for us.

Yes, and it still does.
It's just that it does not use the latest (nor oldest as with CMake), and it is good to have specific jobs for this then.

Also, adding extra apt sources was a source of pain in the past.

I can imagine. Therefore I think it is good to use YAML's features in this regard. But I agree with your other comment that it should not become too complex.

This is the default since ccache 3.3, but Travis has 3.2.4. This fixes compiler warnings from macros with later clang/gcc. Using CCACHE_HASHDIR=1 fixes ccache v3.2.4, but CCACHE_NOHASHDIR=1 does not break v3.7.1. The real issue/fix appears to be ccache/ccache@284e3a0, and using the hashdir option seems to only work around this. Unblocks neovim#10480, and neovim#10487. Ref: ccache/ccache@6d9cb3dfdd9 Closes neovim#10533.

* ci: Travis: use gcc-9 in gcov job * ci: Travis: use CCACHE_CPP2=1 This is required to avoid warnings with newer gcc/clang. Follow-up to: neovim#10533

blueyed changed the title ~~ci: Travis: use gcc9 with gcov job~~ [WIP/blocked] ci: Travis: use gcc9 with gcov job Jul 12, 2019

This comment has been minimized.

Sign in to view

blueyed force-pushed the gcc9 branch from 3a12f46 to a06439c Compare July 12, 2019 18:27

blueyed mentioned this pull request Jul 12, 2019

Revert "Downgrade to clang-4.0 to avoid false-positive warnings from clang" [skip appveyor] #10487

Merged

blueyed force-pushed the gcc9 branch from cb74e9c to 00bef5d Compare July 12, 2019 22:42

blueyed mentioned this pull request Jul 16, 2019

[WIP] Travis: debug GCOV_ERROR_FILE #10455

Closed

blueyed force-pushed the gcc9 branch 2 times, most recently from a4c722b to b38152e Compare July 17, 2019 00:26

blueyed mentioned this pull request Jul 17, 2019

ci: Travis: ccache: use CCACHE_HASHDIR [skip appveyor] #10533

Closed

blueyed force-pushed the gcc9 branch from 6424891 to ea352de Compare July 17, 2019 17:48

blueyed changed the title ~~[WIP/blocked] ci: Travis: use gcc9 with gcov job~~ [RFC] ci: Travis: use gcc9 with gcov job Jul 17, 2019

marvim added the RFC label Jul 17, 2019

blueyed force-pushed the gcc9 branch from b9f7af8 to 1285040 Compare July 17, 2019 20:46

blueyed requested a review from justinmk July 17, 2019 22:48

ci: Travis: use gcc-9 in gcov job

eb2f6da

blueyed force-pushed the gcc9 branch from 1285040 to 13e80db Compare July 17, 2019 22:49

blueyed closed this Jul 17, 2019

ci: Travis: use CCACHE_CPP2=1 [skip appveyor]

0914e34

This is required to avoid warnings with newer gcc/clang. It is the correct fix - I must have messed up when bisecting. Follow-up to: neovim#10533

blueyed reopened this Jul 17, 2019

blueyed force-pushed the gcc9 branch from 13e80db to 0914e34 Compare July 17, 2019 22:51

blueyed changed the title ~~[RFC] ci: Travis: use gcc9 with gcov job~~ ci: Travis: use gcc9 with gcov job Jul 17, 2019

blueyed commented Jul 17, 2019

View reviewed changes

.travis.yml

packages:

- ccache

- ninja

Copy link

Contributor Author

blueyed Jul 17, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is only moved to allow using its new anchor below.

justinmk approved these changes Jul 17, 2019

View reviewed changes

blueyed merged commit 3d356c1 into neovim:master Jul 17, 2019

blueyed deleted the gcc9 branch July 17, 2019 23:27

blueyed mentioned this pull request Jul 17, 2019

ci: gcov: "merge mismatch" errors #10332

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ci: Travis: use gcc9 with gcov job #10480

ci: Travis: use gcc9 with gcov job #10480

blueyed commented Jul 12, 2019

blueyed commented Jul 12, 2019

blueyed commented Jul 12, 2019 •

edited

blueyed commented Jul 12, 2019

blueyed commented Jul 12, 2019

This comment has been minimized.

blueyed commented Jul 12, 2019

blueyed commented Jul 17, 2019

blueyed commented Jul 17, 2019

blueyed Jul 17, 2019

justinmk left a comment •

edited

blueyed commented Jul 17, 2019

ci: Travis: use gcc9 with gcov job #10480

ci: Travis: use gcc9 with gcov job #10480

Conversation

blueyed commented Jul 12, 2019

blueyed commented Jul 12, 2019

blueyed commented Jul 12, 2019 • edited

blueyed commented Jul 12, 2019

blueyed commented Jul 12, 2019

This comment has been minimized.

blueyed commented Jul 12, 2019

blueyed commented Jul 17, 2019

blueyed commented Jul 17, 2019

blueyed Jul 17, 2019

Choose a reason for hiding this comment

justinmk left a comment • edited

Choose a reason for hiding this comment

blueyed commented Jul 17, 2019

blueyed commented Jul 12, 2019 •

edited

justinmk left a comment •

edited