Red Hat Bugzilla – Bug 966566
[abrt] pulseaudio-3.0-7.el7: snd_pcm_area_copy: Process /usr/bin/pulseaudio was killed by signal 6 (SIGABRT)
Last modified: 2014-03-04 11:19:19 EST
Description of problem:
Happened right after logging in.
Version-Release number of selected component:
cmdline: /usr/bin/pulseaudio --start --log-target=syslog
runlevel: N 5
Thread no. 1 (6 frames)
#4 snd_pcm_area_copy at /usr/lib64/libasound.so.2
#5 snd_pcm_areas_copy at /usr/lib64/libasound.so.2
#6 snd_pcm_softvol_write_areas at /usr/lib64/libasound.so.2
#7 snd_pcm_plugin_mmap_commit at /usr/lib64/libasound.so.2
#8 thread_func at /usr/lib64/pulse-3.0/modules/libalsa-util.so
#9 internal_thread_func at /usr/lib64/pulseaudio/libpulsecommon-3.0.so
Created attachment 752212 [details]
Created attachment 752213 [details]
Created attachment 752214 [details]
Created attachment 752215 [details]
Created attachment 752216 [details]
Created attachment 752217 [details]
Created attachment 752218 [details]
Created attachment 752219 [details]
Created attachment 752220 [details]
Created attachment 752221 [details]
This assertion looks like it's triggered because of overlapping areas when doing a memcpy. It does not look like the source memory and the slave memory should ever use the same memory regions unless some memory corruption is at play.
Can this be reproduced? Did it ever happen again?
I didn't really try when I reported the bug and now it doesn't happen at all.