Bug 839413

Description of problem:


usermod -G audio dummy

That works without changing permissions on the /dev/snd/* devices.

[root@seaking ~]# ls -l /dev/snd/*
crw-rw----. 1 root audio 116, 12 Jun 27 18:20 /dev/snd/controlC0
crw-rw----. 1 root audio 116, 11 Jun 27 18:20 /dev/snd/hwC0D2
crw-rw----. 1 root audio 116, 10 Jun 27 18:20 /dev/snd/hwC0D3
crw-rw----. 1 root audio 116,  9 Jun 27 18:20 /dev/snd/pcmC0D0c
crw-rw----. 1 root audio 116,  8 Jun 27 18:23 /dev/snd/pcmC0D0p
crw-rw----. 1 root audio 116,  7 Jun 27 18:20 /dev/snd/pcmC0D1c
crw-rw----. 1 root audio 116,  6 Jun 27 18:20 /dev/snd/pcmC0D1p
crw-rw----. 1 root audio 116,  5 Jun 27 18:20 /dev/snd/pcmC0D2c
crw-rw----. 1 root audio 116,  4 Jun 27 18:20 /dev/snd/pcmC0D3p
crw-rw----. 1 root audio 116,  3 Jun 27 18:20 /dev/snd/seq
crw-rw----. 1 root audio 116,  2 Jun 27 18:20 /dev/snd/timer

I went a little bit further with this and created another user dummy2.

Either way that I get this to work, the way above or the way from Comment #19.

If I log in as two consoles.

console1
su - dummy

console2
su - dummy2

console1
speaker-test
#it works, so ctrl-c

console2
speaker-test
#it doesn't work

[dummy2@seaking ~]$ speaker-test 

speaker-test 1.0.21

Playback device is default
Stream parameters are 48000Hz, S16_LE, 1 channels
Using 16 octaves of pink noise
ALSA lib pcm_dmix.c:985:(snd_pcm_dmix_open) unable to create IPC semaphore
Playback open error: -13,Permission denied
ALSA lib pcm_dmix.c:985:(snd_pcm_dmix_open) unable to create IPC semaphore
Playback open error: -13,Permission denied

but if I go back to root in console2 I can simultaneously have both root and dummy running speaker-test at the same time, I can hear both working.

so if I do an ipcs -a I see some sem/shm left around for dummy after I do a ctrl-c, I suspect this is a bug with speaker-test.

because as soon as I remove the sem and shared memory, I can then run speaker-test as  dummy2

let me try to find another way to test the sound, nevermind

#-l option let's speaker-test stop after so many iterations, and doing it this way cleans up the shared memory/semaphores
speaker-test -c 2 -t wav -l 1

once that is cleared up I can then run as the other user dummy2

Note the adding of dummy to the audio group was only to allow users to run speaker-test from a root owned 777 directory

See following KCS article.
https://access.redhat.com/knowledge/solutions/128963

This case is just to ensure that speaker-test on ctrl-c clears out shared memory.