Red Hat Bugzilla – Bug 433495
snd_hda_intel, doesn't recognise device
Last modified: 2009-04-03 10:54:39 EDT
Description of problem:
snd_hda_intel, doesn;t find correct sound outputs on sound card
Version-Release number of selected component (if applicable):
kernel-126.96.36.199-31.fc9.x86_64 or greater fails
kernel-188.8.131.52-137.fc8.x86_64 (installed on rawhide, works fine)
Steps to Reproduce:
1. install kernel-184.108.40.206-31.fc9.x86_64 or greater
Sound pops only, no other audio
System audio system working.
ALSA sound/pci/hda/hda_intel.c:1792: chipset global capabilities = 0x4401
ALSA sound/pci/hda/hda_intel.c:744: codec_mask = 0x3
hda_codec: Unknown model for ALC883, trying auto-probe from BIOS...
ALSA sound/pci/hda/hda_codec.c:2854: autoconfig: line_outs=4 (0x14/0x15/0x16/0x1
ALSA sound/pci/hda/hda_codec.c:2858: speaker_outs=0 (0x0/0x0/0x0/0x0/0x0)
ALSA sound/pci/hda/hda_codec.c:2862: hp_outs=1 (0x1b/0x0/0x0/0x0/0x0)
ALSA sound/pci/hda/hda_codec.c:2863: mono: mono_out=0x0
ALSA sound/pci/hda/hda_codec.c:2871: inputs: mic=0x18, fmic=0x19, line=0x1a,
fline=0x0, cd=0x0, aux=0x0
ALSA sound/pci/hda/hda_generic.c:678: hda_generic: no proper input path found
ALSA sound/pci/hda/hda_generic.c:426: hda_generic: no proper output path found
ALSA sound/pci/hda/hda_generic.c:1021: hda_generic: no PCM found
ALSA sound/pci/hda/hda_codec.c:1070: Cannot find slave Headphone Playback Volume
ALSA sound/pci/hda/hda_codec.c:1070: Cannot find slave Speaker Playback Volume,
ALSA sound/pci/hda/hda_codec.c:1070: Cannot find slave Mono Playback Volume, ski
ALSA sound/pci/hda/hda_codec.c:1070: Cannot find slave Line-Out Playback Volume,
ALSA sound/pci/hda/hda_codec.c:1070: Cannot find slave Speaker Playback Switch,
ALSA sound/pci/hda/hda_codec.c:1070: Cannot find slave Mono Playback Switch, ski
00:1b.0 Audio device: Intel Corporation 82801I (ICH9 Family) HD Audio Controller
Subsystem: ASUSTeK Computer Inc. Unknown device 829f
Flags: bus master, fast devsel, latency 0, IRQ 22
Memory at fe7f4000 (64-bit, non-prefetchable) [size=16K]
Capabilities: <access denied>
Kernel driver in use: HDA Intel
Kernel modules: snd-hda-intel
setting model=asus in modprobe.conf does not help.
Problem is still present in kernel-2.6.25-0.54.rc2.fc9.x86_64
Tried to compile Alsa 1.0.16 drivers and replace the ones in the kernel, but the
drivers won't compile on rawhide, compiler is more picky.
kernel-2.6.25-0.64.rc2.git5.fc9.x86_64 exhibits same problem
kernel-2.6.25-0.65.rc2.git7.fc9.x86_64 also exhibits the same problem. However,
I found that if I upgraded alsa-drv from 1.0.16rc2 to 1.0.16 the problem with
the audio was fixed. I did have to make some simple patches to alsa-drv 1.0.16
(commenting out comparisons to current->fs->root)
So please upgrade alsa-drv in kernel to 1.0.16
I did have to set model=asus-dig2 in modprobe.conf to make sure alsa was using
the right model driver for my card.
kernel-2.6.25-0.69.rc3.git1.fc9.x86_64 exhibits same problem.
Again upgrading Alsa to 1.0.16 final fixes the problem of the sound card not
working. Is anyone even reading this report?
yes, there are just a lot more bugs than there are people to work on them
unfortunately. I'll forward this on to the upstream alsa developers.
thanks for taking the time to debug this as far as you have, it's very helpful.
Just an FYI, but 2.6.25-0.82.rc3.git2.fc9 still has broken sound still. Again
upgrading alsa to 1.0.16 final fixes it.
rc3-git2 had the latest round of ALSA updates applied, too:
Merge: 99642e2... 0d9ac27...
Author: Linus Torvalds <email@example.com>
Date: Fri Feb 29 08:44:00 2008 -0800
Chuck, is there any additional info you would like me to provide?
I found this tool to be quite helpful
Might want to see if that is still reporting alsa-drv version 1.0.16rc2 or 1.0.16
This is what alsa-info looks like when the driver is working correctly.
I just booted into the new kernel (2.6.25-0.87.rc3.git4.fc9) and since the sound
didn't work in it either. I thought I would run alsa-info.sh before I installed
the known working drivers and after. The main thing I found in this is that the
broken kernel is still reporting an alsa version of 1.0.16rc2 for the alsa-drv
portion. Alsa-drver final was released on 2-06 and rc2 was released on 1-29.
Anyway we can get the final release of the driver pushed into the mainline
kernel. At least before Fedora freezes start to happen.
Created attachment 296733 [details]
Run of alsa-info.sh when sound doesn't work
Created attachment 296734 [details]
Output of alsa-info.sh after upgrading alsa drivers to 1.0.16
Status update: Still broken in kernel-2.6.25-0.105.rc5.fc9.x86_64, same fix
Apparently the fix is in the -mm tree. Don't know why it hasn't been merged in.
Status update: Still broken in kernel-2.6.25-0.150.rc6.git7.fc9.i686, same fix
Had to switch my machine from 64bit to 32bit for other reasons
Based on a tip from the Fedora Test mailing list I set
options snd-hda-intel index=0 model=asus-dig2 power_save=0
in /etc/modprobe.conf prior to this, I did not have power_save=0 in the file
after adding that the sound card worked properly. So I also recommend that this
be a default setting for the snd-hda-intel driver.
Changed the default to 0 (disabled.)
(In reply to comment #16)
> Changed the default to 0 (disabled.)
FWIW, I strongly dislike this fix. Alsa's power_save mode can save quite an
amount of power and thus my notebook's battery can last a bit longer. So I think
there is more then one problem report needed to justify disabling power_save for
everyone. Especially if a real fix seems to be in -mm and alsa 1.0.16
Kevin, when you built the ALSA modules yourself, what values did you set for
CONFIG_SND_HDA_POWER_SAVE and CONFIG_SND_HDA_POWER_SAVE_DEFAULT?
Whatever the defaults are...
This is the command I used to compile...
./configure --prefix=/usr --with-cards=hda-intel --with-sequencer=yes
So maybe off is default?
(In reply to comment #17)
> (In reply to comment #16)
> > Changed the default to 0 (disabled.)
> FWIW, I strongly dislike this fix. Alsa's power_save mode can save quite an
> amount of power and thus my notebook's battery can last a bit longer. So I think
> there is more then one problem report needed to justify disabling power_save for
> everyone. Especially if a real fix seems to be in -mm and alsa 1.0.16
It's not disabled, just off by default. You can enable it by adding:
to the module options for snd_hda_intel.
*** Bug 405741 has been marked as a duplicate of this bug. ***
Just a note - I've re-enabled the HDA powersave (at 5s) in rawhide, it'd be great to get some feedback on how this is working at this point.
I'm currently running Fedora 10, can I just install the kernel from rawhide? Or is there other stuff I would need to install. I just don't want to pull in too much of rawhide.
You could just test it on your current kernel by:
echo 5 > /sys/module/snd_hda_intel/parameters/power_save
FYI, I'm running 2.6.29-0.6.rc3.fc10.x86_64 since a few days with "options snd-hda-intel power_save=1" in modprobe.d/audio and it works really well (Dell Latitude D630, GM965)
Since this bug was originally about device recognition, and is now closed, if people could put information regarding pop/click & powersave problems in this driver over in bug 493972.