Skip to contentSkip to conversation

audio: don't use alsa by default

PATCH In review Gerd Hoffmann 3 participants 8 messages Feb 19, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 audio: don't use alsa by default

In case no sound hardware is present the driver initializes successfully and throws errors later on, i.e. effectively the automatic probing doesn't work. So the driver should not be used by default (until that is fixed). Drop the can_be_default flag.

Explicitly picking the alsa driver (using QEMU_AUDIO_DRV=alsa environment variable) will continue to work.

---
 audio/alsaaudio.c | 1 -
 1 file changed, 1 deletion(-)
audio/alsaaudio.c +0 -1
index 635be73bf4..1cc570e30b 100644
@@ -1218,7 +1218,6 @@ static struct audio_driver alsa_audio_driver = {
     .init           = alsa_audio_init,
     .fini           = alsa_audio_fini,
     .pcm_ops        = &alsa_pcm_ops,
-    .can_be_default = 1,
     .max_voices_out = INT_MAX,
     .max_voices_in  = INT_MAX,
     .voice_size_out = sizeof (ALSAVoiceOut),
Signature · 1 line
2.9.3
#1 lore raw
Quoted 8 lines · Gerd Hoffmann

ardware is present the driver initializes successfully and throws errors later on, i.e. effectively the automatic probing doesn't work. So the driver should not be used by default (until that is fixed). Drop the can_be_default flag.

Explicitly picking the alsa driver (using QEMU_AUDIO_DRV=alsa environment variable) will continue to work.

Just pointing out that with this patch, audio initialization still hinders me from starting guests on my Fedora 29 when running under root (I don't think this patch was also intended to fix that, just pointing out for me the problem still exists).

ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused
audio: Failed to create voice `pcspk'
qemu-system-x86_64: Initialization of device isa-pcspk failed:
Initializing audio voice failed

For whatever reason, on my Fedora29 root cannot access alsa devices. Maybe it is a Fedora 29 problem, maybe only my setup is messed up (e.g. because of an upgrade). The workaround for me is to not run QEMU as root. (I don't have time to look into the details)

Signature · 3 lines
Thanks,

David / dhildenb
#2 lore raw
Quoted 31 lines · David Hildenbrand
Quoted 8 lines · Gerd Hoffmann

ardware is present the driver initializes successfully and throws errors later on, i.e. effectively the automatic probing doesn't work. So the driver should not be used by default (until that is fixed). Drop the can_be_default flag.

Explicitly picking the alsa driver (using QEMU_AUDIO_DRV=alsa environment variable) will continue to work.

Just pointing out that with this patch, audio initialization still hinders me from starting guests on my Fedora 29 when running under root (I don't think this patch was also intended to fix that, just pointing out for me the problem still exists).

ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused
audio: Failed to create voice `pcspk'
qemu-system-x86_64: Initialization of device isa-pcspk failed:
Initializing audio voice failed

Quoted 4 lines

For whatever reason, on my Fedora29 root cannot access alsa devices. Maybe it is a Fedora 29 problem, maybe only my setup is messed up (e.g. because of an upgrade). The workaround for me is to not run QEMU as root. (I don't have time to look into the details)

Does your machine actually have any sound card in it ? ie does /dev/snd/ report any pcm* devices ?

Based on your logs here, I suspect it has no sound card. The reason it will work as non-root, is that QEMU will activate the pulseaudio backend that works fine [1] even if no sound card is present.

Regards,
Daniel

[1] by "fine" I mean it'll happily accept the audio stream from the app and send it to /dev/null :-)

Signature · 3 lines
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
#3 lore raw
Quoted 32 lines · Daniel P. Berrangé
Quoted 31 lines · David Hildenbrand
Quoted 8 lines · Gerd Hoffmann

ardware is present the driver initializes successfully and throws errors later on, i.e. effectively the automatic probing doesn't work. So the driver should not be used by default (until that is fixed). Drop the can_be_default flag.

Explicitly picking the alsa driver (using QEMU_AUDIO_DRV=alsa environment variable) will continue to work.

Just pointing out that with this patch, audio initialization still hinders me from starting guests on my Fedora 29 when running under root (I don't think this patch was also intended to fix that, just pointing out for me the problem still exists).

ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused
audio: Failed to create voice `pcspk'
qemu-system-x86_64: Initialization of device isa-pcspk failed:
Initializing audio voice failed

Quoted 7 lines
Quoted 4 lines

For whatever reason, on my Fedora29 root cannot access alsa devices. Maybe it is a Fedora 29 problem, maybe only my setup is messed up (e.g. because of an upgrade). The workaround for me is to not run QEMU as root. (I don't have time to look into the details)

Does your machine actually have any sound card in it ? ie does /dev/snd/ report any pcm* devices ?

Well, I can listen to audio, so I assume yes ;) Let's check

t460s: ~ $ ls -la /dev/snd/
insgesamt 0
drwxr-xr-x 4 root root 360 19. Feb 08:39 .
drwxr-xr-x 22 root root 4400 19. Feb 08:40 ..

drwxr-xr-x   2 root root       60 19. Feb 08:39 by-id
drwxr-xr-x   2 root root       80 19. Feb 08:39 by-path
crw-rw----+  1 root audio 116,  3 19. Feb 08:39 controlC0
crw-rw----+  1 root audio 116, 13 19. Feb 08:39 controlC1
crw-rw----+  1 root audio 116, 11 19. Feb 08:39 hwC1D0
crw-rw----+  1 root audio 116, 12 19. Feb 08:39 hwC1D2
crw-rw----+  1 root audio 116,  2 19. Feb 08:40 pcmC0D0c
crw-rw----+  1 root audio 116,  5 19. Feb 08:40 pcmC1D0c
crw-rw----+  1 root audio 116,  4 19. Feb 11:40 pcmC1D0p
crw-rw----+  1 root audio 116, 10 19. Feb 08:40 pcmC1D10p
crw-rw----+  1 root audio 116,  6 19. Feb 08:40 pcmC1D3p
crw-rw----+  1 root audio 116,  7 19. Feb 08:40 pcmC1D7p
crw-rw----+  1 root audio 116,  8 19. Feb 08:40 pcmC1D8p
crw-rw----+  1 root audio 116,  9 19. Feb 08:40 pcmC1D9p
Quoted 4 lines

Based on your logs here, I suspect it has no sound card. The reason it will work as non-root, is that QEMU will activate the pulseaudio backend that works fine [1] even if no sound card is present.

The thing is, I can play audio as normal user, but not as root

t460s: ~ $ LANG=C speaker-test -d default

speaker-test 1.1.6

Playback device is default
Stream parameters are 48000Hz, S16_LE, 1 channels
Using 16 octaves of pink noise
Rate set to 48000Hz (requested 48000Hz)
Buffer size range from 192 to 2097152
Period size range from 64 to 699051
Using max buffer size 2097152
Periods = 4
was set period_size = 524288
was set buffer_size = 2097152
ALSA <-> PulseAudio PCM I/O Plugin

t460s: ~ $ LANG=C sudo speaker-test -d default

speaker-test 1.1.6

Playback device is default
Stream parameters are 48000Hz, S16_LE, 1 channels
Using 16 octaves of pink noise
ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

Again, this might be an issue on my system. But as I never touched any of the configuration it either is broken on all Fedora 29 installations or only on mine due to system upgrades. I already tried downgrading pulse audio without luck.

Signature · 3 lines
Thanks,

David / dhildenb
#4 lore raw
In reply to David Hildenbrand
Quoted 24 lines · David Hildenbrand
Quoted 8 lines · Gerd Hoffmann

ardware is present the driver initializes successfully and throws errors later on, i.e. effectively the automatic probing doesn't work. So the driver should not be used by default (until that is fixed). Drop the can_be_default flag.

Explicitly picking the alsa driver (using QEMU_AUDIO_DRV=alsa environment variable) will continue to work.

Just pointing out that with this patch, audio initialization still hinders me from starting guests on my Fedora 29 when running under root (I don't think this patch was also intended to fix that, just pointing out for me the problem still exists).

ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

Quoted 5 lines

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused
audio: Failed to create voice `pcspk'
qemu-system-x86_64: Initialization of device isa-pcspk failed:
Initializing audio voice failed

So SDL has the same problem (possibly only when using alsa).

Hmm. Dropping can_be_default here will too break sound on BSD ...

#5 lore raw
Quoted 31 lines · Gerd Hoffmann
Quoted 24 lines · David Hildenbrand
Quoted 8 lines · Gerd Hoffmann

ardware is present the driver initializes successfully and throws errors later on, i.e. effectively the automatic probing doesn't work. So the driver should not be used by default (until that is fixed). Drop the can_be_default flag.

Explicitly picking the alsa driver (using QEMU_AUDIO_DRV=alsa environment variable) will continue to work.

Just pointing out that with this patch, audio initialization still hinders me from starting guests on my Fedora 29 when running under root (I don't think this patch was also intended to fix that, just pointing out for me the problem still exists).

ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect:
Connection refused

Quoted 5 lines

sdl: SDL_OpenAudio failed
sdl: Reason: ALSA: Couldn't open audio device: Connection refused
audio: Failed to create voice `pcspk'
qemu-system-x86_64: Initialization of device isa-pcspk failed:
Initializing audio voice failed

Quoted 3 lines

So SDL has the same problem (possibly only when using alsa).

Hmm. Dropping can_be_default here will too break sound on BSD ...

We could temporarily put the Linux default ordering back to have "oss" before sdl/alsa until the initialization code in sdl/alsa can be fixed to correctly detect working audio.

Regards,
Daniel

Signature · 3 lines
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
#6 lore raw
Quoted 7 lines
Quoted 3 lines

So SDL has the same problem (possibly only when using alsa).

Hmm. Dropping can_be_default here will too break sound on BSD ...

We could temporarily put the Linux default ordering back to have "oss" before sdl/alsa until the initialization code in sdl/alsa can be fixed to correctly detect working audio.

Reordering doesn't help, they would still probe successfully and fail later, only after oss probe failed ...

pa is ok, when the daemon isn't running init() fails properly.

cheers,
Gerd

#7 lore raw
In reply to Gerd Hoffmann
Quoted 20 lines · Gerd Hoffmann

In case no sound hardware is present the driver initializes successfully and throws errors later on, i.e. effectively the automatic probing doesn't work. So the driver should not be used by default (until that is fixed). Drop the can_be_default flag.

Explicitly picking the alsa driver (using QEMU_AUDIO_DRV=alsa environment variable) will continue to work.

---
 audio/alsaaudio.c | 1 -
 1 file changed, 1 deletion(-)
audio/alsaaudio.c +0 -0
index 635be73bf4..1cc570e30b 100644

audio/alsaaudio.c static struct audio_driver alsa_audio_driver = { · line 1223

Quoted 7 lines
audio/alsaaudio.c +0 -1
@@ -1218,7 +1218,6 @@ static struct audio_driver alsa_audio_driver = {
     .init           = alsa_audio_init,
     .fini           = alsa_audio_fini,
     .pcm_ops        = &alsa_pcm_ops,
-    .can_be_default = 1,
     .max_voices_out = INT_MAX,
     .max_voices_in  = INT_MAX,
     .voice_size_out = sizeof (ALSAVoiceOut),

This fixes the problem with alsa, but then we get exactly the same problem with sdl audio iniitalizing but then not working.

Regards,
Daniel

Signature · 3 lines
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|