[alsa-devel] Intel 82801I (ICH9 Family) HD Audio Controller on a HP dv4 computer

[Jean-Pierre ANDRE]

Hi,

I am getting difficulty setting up ALSA for my new HP portable computer.
lspci displays the audio chip as :
Audio device: Intel Corporation 82801I (ICH9 Family) HD Audio Controller (rev 03)
and /proc/asound displays the codec as IDT 92HD71B7X

I have uploaded the detailed hardware and configuration parameters to
http://pagesperso-orange.fr/b.andre/visits.html/92HD71B7X

I am using Fedora 10 x86_64 without pulseaudio and with the latest alsa-driver :
alsa-driver-1.0.18a.16.g4012f.139.g6e583.tar.bz2

I have googled for a similar problem, but did not get to an actual fix,
the following thread appears to deal with a most similar situation :
http://forums.opensuse.org/pre-release-beta/399731-beta-5-no-sound.html

After several tries, I have set modprobe.conf as
alias snd-card-0 snd-hda-intel
options sound slots=snd-hda-intel
options snd-hda-intel model=hp-m4 single_cmd=1 enable_msi=1

Results :

aplay freezes, with some sound output which is the continual repeat  
of the initial expected sound (some fraction of a second repeated).
However if I boot with pci=noacpi the output is correct (I do not
understand the relation of power management to sound, there must be
one as I see :
HDA Intel 0000:00:1b.0: power state changed by ACPI to D0
when I unload/reload alsa)

WITHOUT the pci=noacpi option :

- mplayer outputs the sound correctly (does mplayer bypass alsa ?)
- xine outputs a chopped sound (some 100ms OK, some 100ms silent,
  some 100ms OK, etc, with a period which could be 200..500ms)
- most players (eg Real player) just freeze.

I would say that when some sound buffer gets empty, the driver does
not get the information, so it does not fill the buffer again.   

I have done some digging into the code. I see that this 92HD71B7X chip
is supported for HP dv5 and dv7 computers (subsystem ID 103c30f2 and
103C30f4), but mine is a HP dv4 showing a subsystem ID 103c30f7,
so it must be somewhat different and not supported yet.

Can anybody suggest some fix ?

If needed I can apply patches to the code and do some testing so 
that my configuration gets supported in subsequent versions.  

Regards

Jean-Pierre