[pulseaudio-tickets] [Bug 84878] protocol-native: I get endless underruns when I playback when alsa buffer size is small

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=84878

--- Comment #16 from Alban Browaeys <prahal at yahoo.com> ---
(In reply to Raymond from comment #14)
> (In reply to Alban Browaeys from comment #12)
> > Created attachment 112915 [details]
> > pcm_avail.c results : results from BURST
> > 
> > DESCRIPTOR results are same for setup output, only diff is in Playing silence
> 
> how do the hardware support both modes ?

The hardware is burst in that it has register to get the progress (destination
and source address registers).
Only that current kernel code does not get their values thus behaves the
descriptor way.



> it is strange that the avail increment by one in brust mode and not by
> period size in descriptor mode

Sorry an artefact : I switched only mode from BURST to DESCRIPTOR mode. When I
revert the whole changeset avail increment by 1024. 

Playing silence
now_us-last_us=1987690780        now_us-start_us=32      time_us-start_us=20   
 time_us-last_time=1987690768    time_us-start_time=9    Available: 0    loop
iteration: 0
now_us-last_us=27        now_us-start_us=20064   time_us-start_us=20040 
time_us-last_time=8     time_us-start_time=20029        Available: 1024        
loop iteration: 2117
now_us-last_us=26        now_us-start_us=41395   time_us-start_us=41372 
time_us-last_time=8     time_us-start_time=41361        Available: 2048        
loop iteration: 4844
now_us-last_us=26        now_us-start_us=62725   time_us-start_us=62702 
time_us-last_time=8     time_us-start_time=62691        Available: 3072        
loop iteration: 7575


still a bit dubious as to why there is a different behaviour then (after all
the only diff is that I compute the residue which should not matter in
descriptor mode as not used).
But mind that be it with the full revert or the partial revert both exhibit the
same issue this report is about : 
"the client always request a prebuf equal or greater than minreq. Only when
heuristics in fix__playback_buffer_attr increase minreq they do not change
prebuf)."

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/pulseaudio-bugs/attachments/20150205/a37e4b0f/attachment.html>