NVidia AC97 / CK804: arecord hangs in system call poll()

NVidia AC97 / CK804: arecord hangs in system call poll()

[Rainer Menzner]

Hi,

my soundcard is an onboard AC97 CK804 NVidia. I use snd-intel8x0 module. 
Everything works fine except recording sound. When calling

   arecord -f cd buffer.wav

the program infinitely hangs, and an strace shows that it is a poll() 
call that does not return. After interrupting, a wave file of 44 bytes 
which is the header only is left. My setup is Kernel 2.6.21.3, all 
ALSA-software is updated to the latest version, that is 1.0.14(a).

In a previous setup with an older ALSA version and kernel 2.6.16.11 
recording did work with the same hardware. Now that I updated all ALSA 
software, even with that older kernel version the same recording problem 
came up. From that I concluded that it is somehow related to the 
ALSA-lib version. I have tried to downgrade ALSA-lib down to 1.0.0 to no 
avail.

I have also made tests appropriate if this would be an interrupt problem 
(using pci=noroute et al) which also did not solve the problem.

I have scrolled though the archives of alsa-user and there have been a 
few posts regarding the same problem (but with other chip sets). 
However, in those threads no definitive solution of the problem has been 
discussed. So I tried it again in alsa-user but did not get any feedback ...

Now, I finally try this developer list. Your comments and help are much 
appreciated. Let me know if you need further info. I'm not a Linux 
system programmer or specialist, but I would be willing to do some 
debugging if somebody explains what to check.

Thanks for any support,

-Rainer

Re: NVidia AC97 / CK804: arecord hangs in system call poll()

[Takashi Iwai]

At Fri, 29 Jun 2007 13:36:15 +0200,
Rainer Menzner wrote:
> 
> Hi,
> 
> my soundcard is an onboard AC97 CK804 NVidia. I use snd-intel8x0 module. 
> Everything works fine except recording sound. When calling
> 
>    arecord -f cd buffer.wav
> 
> the program infinitely hangs, and an strace shows that it is a poll() 
> call that does not return. After interrupting, a wave file of 44 bytes 
> which is the header only is left. My setup is Kernel 2.6.21.3, all 
> ALSA-software is updated to the latest version, that is 1.0.14(a).
> 
> In a previous setup with an older ALSA version and kernel 2.6.16.11 
> recording did work with the same hardware. Now that I updated all ALSA 
> software, even with that older kernel version the same recording problem 
> came up. From that I concluded that it is somehow related to the 
> ALSA-lib version. I have tried to downgrade ALSA-lib down to 1.0.0 to no 
> avail.
> 
> I have also made tests appropriate if this would be an interrupt problem 
> (using pci=noroute et al) which also did not solve the problem.
> 
> I have scrolled though the archives of alsa-user and there have been a 
> few posts regarding the same problem (but with other chip sets). 
> However, in those threads no definitive solution of the problem has been 
> discussed. So I tried it again in alsa-user but did not get any feedback ...
> 
> Now, I finally try this developer list. Your comments and help are much 
> appreciated. Let me know if you need further info. I'm not a Linux 
> system programmer or specialist, but I would be willing to do some 
> debugging if somebody explains what to check.

Most likely a mixer configuration.  For example, it's now set up to
record from the SPDIF input or so.
You can check the detailed configuration from the generated file via
"alsactl -f somefile store".


Takashi

Re: NVidia AC97 / CK804: arecord hangs in system call poll()

[Rainer Menzner]

your guess was correct, I hadn't believed in such trivial solution.

Thanks,
-Rainer

Takashi Iwai wrote:
> At Fri, 29 Jun 2007 13:36:15 +0200,
> Rainer Menzner wrote:
>   
>> Hi,
>>
>> my soundcard is an onboard AC97 CK804 NVidia. I use snd-intel8x0 module. 
>> Everything works fine except recording sound. When calling
>>
>>    arecord -f cd buffer.wav
>>
>> the program infinitely hangs, and an strace shows that it is a poll() 
>> call that does not return. After interrupting, a wave file of 44 bytes 
>> which is the header only is left. My setup is Kernel 2.6.21.3, all 
>> ALSA-software is updated to the latest version, that is 1.0.14(a).
>>
>> In a previous setup with an older ALSA version and kernel 2.6.16.11 
>> recording did work with the same hardware. Now that I updated all ALSA 
>> software, even with that older kernel version the same recording problem 
>> came up. From that I concluded that it is somehow related to the 
>> ALSA-lib version. I have tried to downgrade ALSA-lib down to 1.0.0 to no 
>> avail.
>>
>> I have also made tests appropriate if this would be an interrupt problem 
>> (using pci=noroute et al) which also did not solve the problem.
>>
>> I have scrolled though the archives of alsa-user and there have been a 
>> few posts regarding the same problem (but with other chip sets). 
>> However, in those threads no definitive solution of the problem has been 
>> discussed. So I tried it again in alsa-user but did not get any feedback ...
>>
>> Now, I finally try this developer list. Your comments and help are much 
>> appreciated. Let me know if you need further info. I'm not a Linux 
>> system programmer or specialist, but I would be willing to do some 
>> debugging if somebody explains what to check.
>>     
>
> Most likely a mixer configuration.  For example, it's now set up to
> record from the SPDIF input or so.
> You can check the detailed configuration from the generated file via
> "alsactl -f somefile store".
>
>
> Takashi
>
>   


-- 

______________________________________________
 
*Dr. Rainer Menzner*
Principal Engineer, R&D
 
*L-1 Identity Solutions AG*
/Protecting and Securing Personal Identities and Assets/
 
Universitätsstr. 160
44801 Bochum, Germany
Telephone:  	+49-234-9787-41
Facsimile:  	+49-234-9787-77
Mobile:  	0160 / 700 99 07
email: 	rmenzner@L1id.com <mailto:rmenzner@L1id.com>

 
www.L1ID.com <http://www.l1id.com/>
 
This message is only for the use of the intended recipient and may 
contain information that is CONFIDENTIAL and PROPRIETARY to L-1 Identity 
Solutions. If you are not the intended recipient, please erase all 
copies of the message and its attachments and notify the sender immediately.
 
VORSTAND / EXECUTIVE BOARD Hartmuth von Maltzahn (Vors. / Chairman), Dr. 
Stefan Gehlen | AUFSICHTSRAT / SUPERVISORY BOARD James H. Moar (Vors. / 
Chairman) | GERICHTSSTAND / REGISTER COURT Amtsgericht Bochum, HRB 69 54 
| UST-ID / VAT-REG.-NO. DE 813124378