On 12/14/2021 10:57 PM, zestypucker wrote:
> Hello everyone,
>
> Installed OpenBSD 7.0 (amd64 GENERIC.MP) and most things worked out of the box which is a very pleasant surprise ^_^
>
> There is still a small issue with the mpv-0.33.1p2 package installed using pkg_add that I'm hoping to get some help on:
>
> It always seg faults on exit, does not seem to matter what was played back (a single picture will do), as long as it's not a -h to print usage, mpv seg faults on exit.
>
> Got the following stack trace by running it under gdb but there's no symbols:
>
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to thread 592846]
> 0x000005f84812db50 in ?? ()
> (gdb) thread apply all backtrace
>
> Thread 5 (thread 249681):
> #0 futex () at /tmp/-:3
> #1 0x000005f899c62275 in _rthread_cond_timedwait (cond=0x5f8a64fcd80, mutexp=0x5f8a6502958, abs=0x0)
> at /usr/src/lib/libc/thread/rthread_cond.c:102
> #2 0x000005f5da17b3dc in worker_thread () from /usr/local/bin/mpv
> #3 0x000005f81d43b1c1 in _rthread_start (v=Unhandled dwarf expression opcode 0xa3
> ) at /usr/src/lib/librthread/rthread.c:96
> #4 0x000005f899bde18a in __tfork_thread () at /usr/src/lib/libc/arch/amd64/sys/tfork_thread.S:84
> #5 0x000005f899bde18a in __tfork_thread () at /usr/src/lib/libc/arch/amd64/sys/tfork_thread.S:84
> Previous frame identical to this frame (corrupt stack?)
> #0 0x000005f84812db50 in ?? ()
>
> Some more info:
>
> 1. Happens in a VM (AMD64 CPU, vmware GPU) as well
> 2. Happens without the X server as well, Running mpv directly in a vterm causes the same seg fault on exit
> 3. Sometimes a warning similar to "pthread_mutex_destroy() called while there are waiters on the mutex" is printed before the seg fault
>
> Would appreciate any pointers on how to prevent this from happening or properly report the issue.
>
> Also wondering would it be possible to downgrade to an earlier version of mpv just in case it 'fixes' the issue? Looked at the manual and FAQs for pkg_add but wasn't able to figure out how to list available versions of a package or pick a specific one.
>
> Thanks in advance!
> ZP
This is long-standing known issue that appears to be fixed in -current
No comments:
Post a Comment