Created attachment 893847 [details] My emerge --info This is my first bug report, so let me know if I did anything missing or wrong here. Today I caught this message from dmesg while compiling Python 3.12.3 [ 414.592217] python[134969]: segfault at 10 ip 00007faea8799987 sp 00007ffec87dec80 error 4 in libpython3.12.so.1.0[7faea863b000+438000] likely on CPU 1 (core 1, socket 0) [ 450.669537] python[136538]: segfault at 10 ip 00007fadd1876987 sp 00007fadd0fffda0 error 4 in libpython3.12.so.1.0[7fadd1718000+438000] likely on CPU 7 (core 1, socket 0) My chosen USE flags for Python; dev-lang/python-3.12.3-r1:3.12::gentoo USE="bluetooth ensurepip gdbm ncurses pgo readline sqlite ssl -build -debug -examples -libedit -test -tk -valgrind -verify-sig" 0 KiB I attached my emerge --info to this bug report. Might it be hardware related problem, or something else, or not a problem at all?
(In reply to Kerem Caner Ozin from comment #0) > Created attachment 893847 [details] > My emerge --info > > This is my first bug report, so let me know if I did anything missing or > wrong here. > > Today I caught this message from dmesg while compiling Python 3.12.3 > > [ 414.592217] python[134969]: segfault at 10 ip 00007faea8799987 sp > 00007ffec87dec80 error 4 in libpython3.12.so.1.0[7faea863b000+438000] likely > on CPU 1 (core 1, socket 0) > [ 450.669537] python[136538]: segfault at 10 ip 00007fadd1876987 sp > 00007fadd0fffda0 error 4 in libpython3.12.so.1.0[7fadd1718000+438000] likely > on CPU 7 (core 1, socket 0) > > My chosen USE flags for Python; > dev-lang/python-3.12.3-r1:3.12::gentoo USE="bluetooth ensurepip gdbm > ncurses pgo readline sqlite ssl -build -debug -examples -libedit -test -tk > -valgrind -verify-sig" 0 KiB > > I attached my emerge --info to this bug report. > > Might it be hardware related problem, or something else, or not a problem at > all? Python builds and works fine, despite the segmentation fault message during compilation.
From the information given I get the understanding that is most likely python segfaulting during its pgo tests. As a bug this isn't actionable as with segfaults you would have to first make sure it its reproducible (maintainer cant do anything for bad ram for example). > This is my first bug report, so let me know if I did anything missing or wrong here. Build.log of course would be appreciated when possible. (You can configure portage to save successful logs as well https://wiki.gentoo.org/wiki/Portage_log#Always_save_build_logs ) As this isn't showing up as a build failure or runtime issues I think a bug report may be premature.
I also notice some of these segfaults with python3.12 and python3.13 now that I look (alongside expected test failures for conftest). [448004.265451] python[19887]: segfault at 10 ip 000071466d2d11c4 sp 00007fffc3c47b20 error 4 in libpython3.12.so.1.0[71466d0bc000+6e3000] likely on CPU 6 (core 0, socket 0) [448004.265574] Code: 00 48 89 02 48 89 05 ab 97 9d 00 e8 60 ff ff ff 48 89 45 f8 48 8b 05 93 97 9d 00 48 83 c0 01 48 89 05 88 97 9d 00 48 8b 45 f8 <48> 8b 40 10 c9 c3 f3 0f 1e fa 55 48 89 e5 48 83 ec 10 48 89 7d f8 [448039.714597] python[21467]: segfault at 10 ip 00007c11ce9601c4 sp 00007c11cddffd00 error 4 in libpython3.12.so.1.0[7c11ce74b000+6e3000] likely on CPU 11 (core 6, socket 0) [448039.714611] Code: 00 48 89 02 48 89 05 ab 97 9d 00 e8 60 ff ff ff 48 89 45 f8 48 8b 05 93 97 9d 00 48 83 c0 01 48 89 05 88 97 9d 00 48 8b 45 f8 <48> 8b 40 10 c9 c3 f3 0f 1e fa 55 48 89 e5 48 83 ec 10 48 89 7d f8 [448774.956981] python[3489]: segfault at 10 ip 00007f00da415148 sp 00007f00d9bffd90 error 4 in libpython3.13.so.1.0[7f00da3ad000+725000] likely on CPU 11 (core 6, socket 0) [448774.956997] Code: 00 00 48 ff 05 a1 1b b2 00 48 83 3d f1 26 b6 00 00 0f 84 fb 00 00 00 48 8d 3d b4 ba 8a 00 e8 af 88 f9 ff 48 8b 80 08 00 00 00 <48> 8b 40 10 48 8b a8 30 2a 00 00 48 85 ed 0f 84 10 07 00 00 48 ff [557766.575765] python[11199]: segfault at 10 ip 000076bafe2e2148 sp 000076bafd9ffd90 error 4 in libpython3.13.so.1.0[76bafe27a000+725000] likely on CPU 6 (core 0, socket 0) [557766.575779] Code: 00 00 48 ff 05 a1 1b b2 00 48 83 3d f1 26 b6 00 00 0f 84 fb 00 00 00 48 8d 3d b4 ba 8a 00 e8 af 88 f9 ff 48 8b 80 08 00 00 00 <48> 8b 40 10 48 8b a8 30 2a 00 00 48 85 ed 0f 84 10 07 00 00 48 ff Is it expected behavior like with conftest? I dont know for certain
Created attachment 893849 [details] dev-lang/python-3.12.3 build.log Found them in them in the logs. 0:00:02 load avg: 4.75 [ 21/435] test_ast worker non-zero exit code (Exit code -11 (SIGSEGV)) Fatal Python error: Segmentation fault 0:00:08 load avg: 5.41 [ 42/435] test_call worker non-zero exit code (Exit code -11 (SIGSEGV)) Fatal Python error: Segmentation fault
This bug report would be better if you included tests with the build. The pgo segfaults are from the tests so if you explicitly run the tests you will have an easier time tracking these failures. (FEATURES="test-fail-continue" lets you install despite them failing)
Created attachment 893876 [details] My last build log for Python 3.12.3
Roger that, I also added my build log for Python 3.12.3 just in case. Segmentation fault is occuring during CONFTEST when compiling Python with PGO use flag.
Atleast with python3.13 this issues are going to be tracked better with these failures becoming fatal https://bugs.gentoo.org/931888 > Segmentation fault is occuring during CONFTEST when compiling Python with PGO use flag. "Test failures due to segfaults during pgo tests in python." maybe better. The conftests I referred to are from different builds where its expected.
