ice1712, snd_pcm_mmap_readi(), POLLERR in poll_revents
Hi all,
I am still trying to get the ice1712-based soundcard to work on an embedded platform, using alsa driver/lib 0.9.3
While capturing works for hw:0,0 (multitrack mode) using snd_pcm_readi(), I get an -EIO when capturing with snd_pcm_mmap_readi(), immediatly after the buffer is filled up with 4 periods. I tracked the problem down to the snd_read_areas() function, where snd_pcm_wait(pcm,-1) returns -EIO (detection of POLLERR in revents), waiting for the soundcard to become ready.
Do you have any idea where to start debugging in the driver from this point ?
Thanks for your help,
Peter
-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara The only event dedicated to issues related to Linux enterprise solutions www.enterpriselinuxforum.com
At Fri, 16 May 2003 16:03:48 +0200,
Quoted 11 lines · Peter Vollmer
Hi all,
I am still trying to get the ice1712-based soundcard to work on an embedded platform, using alsa driver/lib 0.9.3
While capturing works for hw:0,0 (multitrack mode) usingsnd_pcm_readi(), I get an-EIOwhen capturing withsnd_pcm_mmap_readi(), immediatly after the buffer is filled up with 4 periods.
I tracked the problem down to thesnd_read_areas()function, wheresnd_pcm_wait(pcm,-1)returns-EIO(detection of POLLERR in revents), waiting for the soundcard to become ready.
at first you should check in which state the pcm stream is when poll() returns the error. alsa-lib code checks the normal errors (XRUN and SUSPENDED) but the other status are handled as it is.
ciao,
Takashi
-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara The only event dedicated to issues related to Linux enterprise solutions www.enterpriselinuxforum.com
Quoted 16 lines · Takashi Iwai
At Fri, 16 May 2003 16:03:48 +0200,
Quoted 4 lines · Peter Vollmer
Hi all,
I am still trying to get the ice1712-based soundcard to work on an
embedded
Quoted 2 lines
platform, using alsa driver/lib 0.9.3
While capturing works for hw:0,0 (multitrack mode) using
snd_pcm_readi(), IQuoted 5 lines
get an
-EIOwhen capturing withsnd_pcm_mmap_readi(), immediatly after the buffer is filled up with 4 periods.
I tracked the problem down to thesnd_read_areas()function, wheresnd_pcm_wait(pcm,-1)returns-EIO(detection of POLLERR in revents), waiting for the soundcard to become ready.
Quoted 3 lines
at first you should check in which state the pcm stream is when
poll()returns the error. alsa-lib code checks the normal errors (XRUN and SUSPENDED) but the other status are handled as it is.
I checked this in the driver. Somehow the stream returned to OPEN state. In snd_pcm_capture_poll() the default case is applied. But the driver seems not to cause this, I checked all occurences of SNDRV_PCM_STATE_OPEN (in snd_pcm_open_substream(), snd_pcm_hw_params() and snd_pcm_hw_free(). If the state is only set using the enums SNDDRV_PCM_STATE_... , I guess I have a problem somewhere else.
Peter
-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara The only event dedicated to issues related to Linux enterprise solutions www.enterpriselinuxforum.com