2010/7/16 Jaroslav Kysela perex@perex.cz
On Thu, 15 Jul 2010, Chris wrote:
On Thu, 2010-07-15 at 16:07 +0200, Jaroslav Kysela wrote:
On Thu, 15 Jul 2010, Colin Guthrie wrote:
'Twas brillig, and Chris at 15/07/10 02:38 did gyre and gimble:
On Thu, 2010-07-15 at 08:08 +0800, Raymond Yau wrote:
> I posted both of those bug reports. Honestly, I just want it to work > without all the error output to syslog and thrashing of my HD for > minutes at a time. As I said, I've got the output of pulseaudio
-vvvvv
> if anyone would like to look at it and tell me what I need to do to
fix
> the problem. > > Chris > > > Just upload your current pulseaudio.log and output of alsa-info.sh to mandriva 's bugzilla
Cannot obtain any conclusion based on the log which you have posted
last
year ( xmms)
as Jaroslav Kysela said , need the real system time between I/O
operations
and you'll see if it's issue in the ALSA driver or a task scheduling problem.
Here is the bug report with the outputs uploaded.
Well seeing as my process when getting such a bug is to report it here on behalf of the user, I suspect that this has reached a stalemate....
Please, provide also /var/log/messages output for xrun_debug=11 - see http://www.alsa-project.org/main/index.php/XRUN_Debug for more details.
We
can detect the ALSA ring buffer issues in this way, too.
JaroslavSee pastebin file http://pastebin.com/MdENbSNd Hopefully I did this correctly.
Thanks. The hardware ring buffer pointers looks OK so it does not appear like a problem in our ALSA code. I would really consult these problems with PulseAudio developers. Return back, if they say that it's really ALSA issue, not the system scheduler issue.
Note that for your configuration, you have the ring buffer containing only 0.1 second (it means that standard underruns can happen more frequently
usuall symptom is for example moving a window in X GUI).
Jaroslav
Refer to the log , the values of pos and hwptr seem multiple of 8
5.5 PCI Data transfer
Only brust read/write transfers are allowed. All data transfer are 8 Long Word brust transfers
This mean that the driver may need to limit the period bytes to be multiple of 8 Long Word since interrupt occur when a cache fill/transfer between the PCI memory and the internal memory.
snd_pcm_hw_constraint_step(runtime, 0, SNDRV_PCM_HW_PARAM_PERIOD_BYTES, 32);
and you may also need to set "rewind_safeguard" in PA too.
http://git.0pointer.de/?p=pulseaudio.git;a=commit;h=4df443bbe682055a41e7c224...