Hi, I have to tell at the start - I'm not sure whether it is issue of glibc, but make/xargs binaries (where segfaults seem to origin) are compiled with glibc, so maybe just they need to be fixed... but I rather file it on glibc first, because it was the only updated package between successful build and the failed one (at least in Vim) which IMO can cause the similar segfaults. The several builds failed only on i686/armv7hl: Vim - segfaults on make: https://kojipkgs.fedoraproject.org//work/tasks/8185/43148185/build.log flocq - segfaults on remake: https://kojipkgs.fedoraproject.org//work/tasks/5608/43145608/build.log nip2 - segfaults on configure: https://kojipkgs.fedoraproject.org//work/tasks/2414/43142414/build.log rust-proc-macro-error - segfaults on xargs: https://kojipkgs.fedoraproject.org//work/tasks/2854/43142854/build.log podman - segfaults on xargs: https://kojipkgs.fedoraproject.org//work/tasks/2661/43142661/build.log lmod - segfaults on xargs: https://kojipkgs.fedoraproject.org//work/tasks/4155/43144155/build.log Would you mind looking into the issue?
In the meantime, I'm running: $ koji tag f33-pending glibc-2.31.9000-6.fc33 Created task 43153881 Watching tasks (this may be safely interrupted)... 43153881 tagBuild (noarch): free 43153881 tagBuild (noarch): free -> closed 0 free 0 open 1 done 0 failed 43153881 tagBuild (noarch) completed successfully However I'm never sure if this is enough.
Releng ticket to untag: https://pagure.io/releng/issue/9395
With glibc-2.31.9000-6.fc33 back in the buildroot, the problem seem to be gone (for now).
The make failure looks like this (only happens with -jN if more than one target is built): ==46== Invalid write of size 4 ==46== at 0x4A8B4F9: valid_timeval64_to_timeval (time.h:367) ==46== by 0x4A8B4F9: rusage64_to_rusage (resource.h:97) ==46== by 0x4A8B4F9: wait4 (wait4.c:125) ==46== by 0x4A8B2C9: waitpid (waitpid.c:38) ==46== by 0x12104A: reap_children (in /usr/bin/make) ==46== by 0x121D94: new_job (in /usr/bin/make) ==46== by 0x1133C8: execute_file_commands (in /usr/bin/make) ==46== by 0x12E33C: ??? (in /usr/bin/make) ==46== by 0x12EE12: update_goal_chain (in /usr/bin/make) ==46== by 0x11074A: main (in /usr/bin/make) ==46== Address 0x0 is not stack'd, malloc'd or (recently) free'd This was reported on libc-alpha: https://sourceware.org/pipermail/libc-alpha/2020-April/112547.html A fix is under review: https://sourceware.org/pipermail/libc-alpha/2020-April/112574.html https://sourceware.org/pipermail/libc-alpha/2020-April/112596.html The patch needs a v3.
Fix was pushed upstream: https://sourceware.org/git/?p=glibc.git;a=commitdiff;h=00515ea3a15703a3d196c1d1bd372214abc990ad So we a good to rebase glibc past this bug.
https://koji.fedoraproject.org/koji/taskinfo?taskID=43439516 https://koji.fedoraproject.org/koji/taskinfo?taskID=43439614 ??
I'm seeing this again as well.
https://koji.fedoraproject.org/koji/taskinfo?taskID=43439082 https://koji.fedoraproject.org/koji/taskinfo?taskID=43439297 https://koji.fedoraproject.org/koji/taskinfo?taskID=43439518
Untag request: https://pagure.io/releng/issue/9405 I'm deeply sorry about this, I had a momentary lapse of judgement.
*** Bug 1824307 has been marked as a duplicate of this bug. ***