Re: realtime-preempt-2.6.12-final-V0.7.51-01 compile error and more problems

[Lee Revell <rlrevell@joe-job.com>]

On Tue, 2005-07-05 at 18:13 -0700, William Weston wrote:
> Audio without xruns is an RT requirement, IMHO  ;-}
> 

This isn't even an opinion, it's a fact.  If you are capturing data and
your audio handling thread does not get scheduled in (periods_per_buffer
- 1) * period_time time units, you lose that chunk of sound *forever*.

Some people on this list make the ridiculous argument that this is
different from more conventional RT applications because "they can
always record again".  Try telling that to someone who is paying for
studio time; you won't be running a studio for long.  Pro audio is
exactly like controlling a satellite or monitoring a power plant, if you
don't get scheduled in time, it's a fatal error.

In practice you only have to be more reliable than the alternatives,
which are pretty reliable these days.

> > 4) Xruns occur every 10-60 minutes even when the system is
> practically 
> > idle (no playback
> > or recording). When copying large files (between sata disks on two 
> > sil3112 controllers)
> > xruns frequency is much higher.  When sound is used xruns occur
> every 2 
> > or 20 minutes.
> 
> Do these xruns coincide with the RTC 'Read missed before next
> interrupt'
> messages?
> 
> Have you tried running JACK with a larger buffer period size?  Some
> cards

Disable CONFIG_RTC_HISTOGRAM.

Then before you try anything else, check the /proc/latency trace output.
What's the max reported latency?  Is it significantly less than the
xruns?

Lee

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/