https://bugs.kde.org/show_bug.cgi?id=327548
Samuel Thibault <samuel.thiba...@ens-lyon.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Resolution|WORKSFORME |--- Ever confirmed|0 |1 Status|RESOLVED |REOPENED --- Comment #7 from Samuel Thibault <samuel.thiba...@ens-lyon.org> --- To provide various datapoints, I have tried it valgrind --tool=helgrind ./test on: - Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz - Intel(R) Xeon(R) Gold 6140 CPU @ 2.30GHz - Intel(R) Xeon(R) CPU E5-2650L v4 @ 1.70GHz - AMD EPYC 7352 24-Core Processor - AMD EPYC 7502 32-Core Processor with glibc 2.35 I got the hit ==420467== Helgrind, a thread error detector ==420467== Copyright (C) 2007-2017, and GNU GPL'd, by OpenWorks LLP et al. ==420467== Using Valgrind-3.19.0 and LibVEX; rerun with -h for copyright info ==420467== Command: ./test ==420467== ==420467== ---Thread-Announcement------------------------------------------ ==420467== ==420467== Thread #1 is the program's root thread ==420467== ==420467== ---Thread-Announcement------------------------------------------ ==420467== ==420467== Thread #2 was created ==420467== at 0x49792FF: clone (clone.S:76) ==420467== by 0x497A146: __clone_internal (clone-internal.c:83) ==420467== by 0x48F6484: create_thread (pthread_create.c:295) ==420467== by 0x48F6F78: pthread_create@@GLIBC_2.34 (pthread_create.c:828) ==420467== by 0x484E5D7: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so) ==420467== by 0x1092DA: main (in /net/inria/home/sthibault/test) ==420467== ==420467== ---------------------------------------------------------------- ==420467== ==420467== Possible data race during read of size 1 at 0x10C128 by thread #1 ==420467== Locks held: none ==420467== at 0x484B225: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so) ==420467== by 0x484B4CA: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so) ==420467== by 0x10937C: main (in /net/inria/home/sthibault/test) ==420467== ==420467== This conflicts with a previous write of size 4 by thread #2 ==420467== Locks held: none ==420467== at 0x48FB4D8: __pthread_mutex_unlock_usercnt (pthread_mutex_unlock.c:62) ==420467== by 0x484BCE8: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so) ==420467== by 0x10929A: f (in /net/inria/home/sthibault/test) ==420467== by 0x484E7D6: ??? (in /usr/libexec/valgrind/vgpreload_helgrind-amd64-linux.so) ==420467== by 0x48F6849: start_thread (pthread_create.c:442) ==420467== by 0x497930F: clone (clone.S:100) ==420467== Address 0x10c128 is 8 bytes inside data symbol "mutex" ==420467== immediately on almost all of them, only AMD EPYC 7352 24-Core Processor took 1m to reproduce. On an AMD Opteron(tm) Processor 6174 with glibc 2.28 I couldn't reproduce it within an hour. On a Intel(R) Core(TM) i5-10210U CPU @ 1.60GHz, I tried with various glibcs: 2.19, 2.24, 2.28, 2.31, 2.36, with all versions it happens within half an hour, sometimes within a minute. On an arm64 with glibc 2.36 it appeared immediately On a mips64el Cavium Octeon III V0.2 FPU V0.0 with glibc 2.36 it appeared immediately On a ppc64el POWER8 with glibc 2.36 it appeared within a minute. On an s390x it appeared in 8 minutes. -- You are receiving this mail because: You are watching all bug changes.