Ok, this one's a bit indirect. Apparently, pipewire gets started when pygame tries to open an audio device in the test environment, and it causes a segfault on the Python executable itself. However, the backtrace pretty clearly indicates they're missing error handling somewhere: Core was generated by `/tmp/portage/dev-python/pyscreenshot-3.1/work/pyscreenshot-3.1-python3_11/insta'. Program terminated with signal SIGSEGV, Segmentation fault. #0 0x00007fd17e0f80c1 in pw_rtkit_make_realtime (impl=0x55ca0b1fcd60, thread=146, priority=1) at ../pipewire-0.3.75/src/modules/module-rt.c:426 426 if (!dbus_connection_send(connection->bus, m, &serial)) { [Current thread is 1 (Thread 0x7fd17bfff6c0 (LWP 144))] (gdb) p connection $1 = (struct pw_rtkit_bus *) 0x0 (gdb) bt #0 0x00007fd17e0f80c1 in pw_rtkit_make_realtime (impl=0x55ca0b1fcd60, thread=146, priority=1) at ../pipewire-0.3.75/src/modules/module-rt.c:426 #1 0x00007fd17e0f973b in do_make_realtime (loop=0x55ca0b240cf0, async=true, seq=0, data=0x55ca0b240e48, size=8, user_data=0x55ca0b1fcd60) at ../pipewire-0.3.75/src/modules/module-rt.c:813 #2 0x00007fd17c2e1f78 in flush_items (impl=0x55ca0b240cd8) at ../pipewire-0.3.75/spa/plugins/support/loop.c:171 #3 0x00007fd17c2e30fe in wakeup_func (data=0x55ca0b240cd8, count=1) at ../pipewire-0.3.75/spa/plugins/support/loop.c:306 #4 0x00007fd17c2e515a in source_event_func (source=0x55ca0b1fd200) at ../pipewire-0.3.75/spa/plugins/support/loop.c:650 #5 0x00007fd17c2e3ddd in loop_iterate (object=0x55ca0b240cd8, timeout=-1) at ../pipewire-0.3.75/spa/plugins/support/loop.c:483 #6 0x00007fd17c115552 in do_loop (user_data=0x55ca0b1fcff0) at ../pipewire-0.3.75/src/pipewire/thread-loop.c:288 #7 0x00007fd981ea2279 in start_thread (arg=<optimized out>) at pthread_create.c:444 #8 0x00007fd981f25b8c in clone3 () at ../sysdeps/unix/sysv/linux/x86_64/clone3.S:81
Does https://gitlab.freedesktop.org/pipewire/pipewire/-/commit/55812195ce3b77317e7a2dc642b78271f3a45c8e help at all?
(In reply to Sam James from comment #1) > Does > https://gitlab.freedesktop.org/pipewire/pipewire/-/commit/ > 55812195ce3b77317e7a2dc642b78271f3a45c8e help at all? Indeed, with that the segfault is gone and pyscreenshot tests pass.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=e29598d80e5b6f0e45eccdcab2f920304f2ed0b7 commit e29598d80e5b6f0e45eccdcab2f920304f2ed0b7 Author: Sam James <sam@gentoo.org> AuthorDate: 2023-07-23 17:39:52 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2023-07-23 17:39:52 +0000 media-video/pipewire: backport two fixes to 0.3.75 1. Crash fix for when can't connect to dbus (bug #910714) 2. Help out software misusing API, like older mpv (new fixed mpv will be in tree shortly anyway) (https://gitlab.freedesktop.org/pipewire/pipewire/-/issues/3374) Closes: https://bugs.gentoo.org/910714 Signed-off-by: Sam James <sam@gentoo.org> .../0001-module-rt-error-out-on-load-no-bus.patch | 83 ++++++++++++++++++++++ ...2-thread-loop-only-signal-when-option-set.patch | 68 ++++++++++++++++++ ...ire-0.3.75.ebuild => pipewire-0.3.75-r1.ebuild} | 0 3 files changed, 151 insertions(+)
