Re: How how latent should non-preemptive scheduling be?

Previous thread: [PATCH] bnx2: Promote vector field in bnx2_irq structure from u16 to unsigned int by Benjamin Li on Wednesday, September 17, 2008 - 11:52 am. (1 message)

Next thread: [PATCH] Audit: Log TIOCSTI by Miloslav on Wednesday, September 17, 2008 - 2:24 pm. (1 message)
From: Sitsofe Wheeler
Date: Wednesday, September 17, 2008 - 2:18 pm

Hi,

I have an EeePC 900 (Intel Celeron 900Mhz) and it seems to be skipping 
while playing sound through various desktop apps with a 2.6.27rc6 
kernel. It is running off an SD card which really shows up slow writes 
but the sound is seemingly skipping even when ext3 is not being used. 

top seems to show that there is CPU time still available. The kernel is 
compiled with voluntary preemption, dynticks and a HZ of 1000.

Is the scheduler waiting for CPU anything to do with the skips? If so 
what sort of maximum should I be expecting? If you are listing to music 
is it expected that you have preemption on? Would preemption even help?

--

From: Sitsofe Wheeler
Date: Wednesday, September 17, 2008 - 2:28 pm

One other small datapoint. When the skipping starts to occur regularly 
it seems to be exactly 30 seconds apart. ALSA doesn't report any xruns 
either...
--

From: Arjan van de Ven
Date: Wednesday, September 17, 2008 - 2:34 pm

On Wed, 17 Sep 2008 22:18:18 +0100


this says you haven't done "make install" on the latencytop directory
so it's not translating things for you.. can you do that please?
--

From: Matt Keenan
Date: Sunday, September 21, 2008 - 1:56 pm

[snip snip]

If you use ALSA you could write a custom .asoundrc file and increase
ALSA's default buffer size, this gives the hardware a longer buffer to
play which may work around the problem. This used to do the trick for me
with the old scheduler with slow machines. Not a fix but if nothing else
works...

Matt

--

From: Sitsofe Wheeler
Date: Sunday, September 21, 2008 - 11:50 pm

In all fairness turning on preemption also helps (I don't hear any drops 
or see any stalls once it's on) but I'm trying to work out if this is an 
underlying problem that can be fixed (and then there is the case of the 
default desktop distros).

-- 
Sitsofe | http://sucs.org/~sits/
--

Previous thread: [PATCH] bnx2: Promote vector field in bnx2_irq structure from u16 to unsigned int by Benjamin Li on Wednesday, September 17, 2008 - 11:52 am. (1 message)

Next thread: [PATCH] Audit: Log TIOCSTI by Miloslav on Wednesday, September 17, 2008 - 2:24 pm. (1 message)