With crossfade+alsa, xmms-1.2.8 would constantly segfault and *sometimes* hang on exit with *some* of the threads having finished and some more still alive, like this: ~> /sbin/pidof xmms 16337 16336 16293 16292 Only one thread does something interesting then: ~> strace -p 16336 Process 16336 attached - interrupt to quit close(11) = 0 close(12) = 0 shmget(1025, 65536, IPC_CREAT|0666) = 2010873869 shmat(2010873869, 0, 0) = 0x40012000 mlock(0x40012000, 65536) = -1 EPERM (Operation not permitted) shmdt(0x40012000) = 0 shmctl(2010873869, IPC_64|IPC_STAT, 0xbf3fef4c) = 0 unlink("/tmp/alsa-dmix-16333-1073234389-218308") = 0 shmdt(0x4107a000) = 0 shmctl(2010808332, IPC_64|IPC_STAT, 0xbf3fef4c) = 0 semctl(393217, 0, IPC_RMID, 0xbf3fef58) = 0 _exit(0) = ? Process 16336 detached Tested with xmms-1.2.8-alt3 + xmms-out-alsa-1.2.8-alt3 + xmms-out-crossfade-0.3.4-alt2; there's dmix setup in ~/.asound but removing the file (with xmms restart) doesn't change the usual situation much. Maybe it's more racey though... Plain OSS, plain ALSA, CF + builtin OSS, CF + OSS plugin all work just fine, but CF + ALSA plugin misbehaves.
2.4.2x / 2.6.0 -- all the same, btw.
Hah! corssfade's author fixed that with a one-liner :-) 1.2.8-alt4 on its way to Sisyphus.
Broken again with xmms-1.2.10-alt2, xmms-out-crossfade-0.3.4-alt2, xmms-out-alsa-1.2.10-alt2. With crossfade+alsa, when exiting xmms without stopping the playback, xmms hangs on exit: $ pidof xmms 19231 19223 19222 $ strace -p 19223 Process 19223 attached - interrupt to quit getppid() = 19222 poll([{fd=10, events=POLLIN}], 1, 2000) = 0 getppid() = 19222 poll([{fd=10, events=POLLIN}], 1, 2000) = 0 getppid() = 19222 poll([{fd=10, events=POLLIN}], 1, 2000) = 0 getppid() = 19222 poll([{fd=10, events=POLLIN}], 1, 2000) = 0 getppid() = 19222 poll([{fd=10, events=POLLIN}], 1, 2000) = 0 getppid() = 19222 poll( <unfinished ...> Process 19223 detached With both CF + OSS plugin and CF + builtin OSS in the same situation xmms segfaults.
what's with 1.2.10-alt7?
То ли было от чего другого, то ли полечилось -- уже не наблюдаю. Заодно сейчас займусь 0.3.10.
