Bug 1766519

Description of problem:

Assertion fails while building snapd 2.42 on ppc64le when running unit tests under valgrind.

Relevant build log: https://kojipkgs.fedoraproject.org//work/tasks/3761/38393761/build.log

Extracted part of the log:

# GLib-DEBUG: posix_spawn avoided (workdir specified) (fd close requested) 
==12500== Warning: invalid file descriptor 1024 in syscall close()
==12500== Warning: invalid file descriptor 1025 in syscall close()
==12500== Warning: invalid file descriptor 1026 in syscall close()
==12500== Warning: invalid file descriptor 1027 in syscall close()
==12500==    Use --log-fd=<number> to select an alternative log fd.
==12500== Warning: invalid file descriptor 1028 in syscall close()
==12500== Warning: invalid file descriptor 1029 in syscall close()
valgrind: m_sigframe/sigframe-ppc64-linux.c:155 (vgPlain_sigframe_create): Assertion 'VG_IS_16_ALIGNED(sizeof(struct rt_sigframe))' failed.
host stacktrace:
==12497==    at 0x5805AD74: ??? (in /usr/libexec/valgrind/memcheck-ppc64le-linux)
==12497==    by 0x5805AF0F: ??? (in /usr/libexec/valgrind/memcheck-ppc64le-linux)
==12497==    by 0x5805B09B: ??? (in /usr/libexec/valgrind/memcheck-ppc64le-linux)
==12497==    by 0x580D937F: ??? (in /usr/libexec/valgrind/memcheck-ppc64le-linux)
==12497==    by 0x58073177: ??? (in /usr/libexec/valgrind/memcheck-ppc64le-linux)
==12497==    by 0x580740A7: ??? (in /usr/libexec/valgrind/memcheck-ppc64le-linux)
==12497==    by 0x7FFFA8C204D7: ???
sched status:
  running_tid=1
Thread 1: status = VgTs_Runnable (lwpid 12497)
==12497==    at 0x4BBFBD8: waitpid (in /usr/lib64/libc-2.30.9000.so)
==12497==    by 0x493DDB3: g_child_watch_source_new (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x497320F: ??? (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x497977B: g_test_trap_subprocess (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x18588F: test_sc_error_code__NULL (error-test.c:125)
==12497==    by 0x497896F: ??? (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x49786C7: ??? (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x49786C7: ??? (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x4978FAB: g_test_run_suite (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x4979003: g_test_run (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x19BA2B: sc_run_unit_tests (unit-tests.c:28)
==12497==    by 0x1843CB: main (unit-tests-main.c:21)
client stack range: [0x1FFEFF0000 0x1FFF00FFFF] client SP: 0x1FFF00E910
valgrind stack range: [0x1002E90000 0x1002F8FFFF] top usage: 9920 of 1048576
Thread 2: status = VgTs_WaitSys syscall 167 (lwpid 12499)
==12497==    at 0x4BF343C: poll (in /usr/lib64/libc-2.30.9000.so)
==12497==    by 0x49537E7: g_poll (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x493BBE7: ??? (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x493BD97: g_main_context_iteration (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x493BE3B: ??? (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x497A45B: ??? (in /usr/lib64/libglib-2.0.so.0.6300.0)
==12497==    by 0x4D89767: start_thread (in /usr/lib64/libpthread-2.30.9000.so)
==12497==    by 0x4C03697: clone (in /usr/lib64/libc-2.30.9000.so)
client stack range: [0x51E0000 0x59DFFFF] client SP: 0x59DE5E0
valgrind stack range: [0x1005D80000 0x1005E7FFFF] top usage: 5600 of 1048576
Note: see also the FAQ in the source distribution.
It contains workarounds to several common problems.
In particular, if Valgrind aborted or crashed after
identifying problems in your program, there's a good chance
that fixing those problems will prevent Valgrind aborting or
crashing, especially if it happened in m_mallocfree.c.
If that doesn't help, please report this bug to: www.valgrind.org
In the bug report, send all the above text, the valgrind
version, and what OS and version you are using.  Thanks.
make: *** [Makefile:4137: check-unit-tests] Error 1

Version-Release number of selected component (if applicable):
valgrind      ppc64le1:3.15.0-12.fc32  

Steps to reproduce:

I don't have access to ppc64le hardware, but scratch builds of snapd fail the same way. I suspect that manual rpmbuild would reproduce that too.