Re: [PATCH] Preemption Latency Measurement Tool

Robert Love (rml@tech9.net)
22 Sep 2001 22:50:56 -0400


On Sat, 2001-09-22 at 19:40, safemode wrote:
> ok. The preemption patch helps realtime applications in linux be a little
> more close to realtime. I understand that. But your mp3 player shouldn't
> need root permission or renicing or realtime priority flags to play mp3s.

It doesn't, it needs them to play with a dbench 32 running in the
background. This isn't nessecarily acceptable, either, but it is a
difference.

Note one thing the preemption patch does is really make `realtime' apps
accel. Without it, regardless of the priority of the application, the
app can be starved due to something in kernel mode. Now it can't, and
since said application is high priority, it will get the CPU when it
wants it.

This is not to say the preemption patch is no good if you don't run
stuff at realtime -- I don't (who uses nice, anyhow? :>), and I notice
a difference.

> To
> test how well the latency patches are working you should be running things
> all at the same priority. The main issue people are having with skipping
> mp3s is not in the decoding of the mp3 or in the retrieving of the file, it's
> in the playing in the soundcard. That's being affected by dbench flooding
> the system with irq requests. I'm inclined to believe it's irq requests
> because the _only_ time i have problems with mp3s (and i dont change priority
> levels) is when A. i do a cdparanoia -Z -B "1-" or dbench 32. I bet if
> someone did these tests on scsi hardware with the latency patch, they'd find
> much better results than us users of ide devices.

The skips are really big to be irq requests, although perhaps you are
right in that the handling of the irq (we disable preemption during
irq_off, of course, but also during bottom half execution) is the
problem.

However, I am more inclined to believe it is something else. All these
long held locks can indeed be the problem.

I am on an all UW2 SCSI system, and I have no major blips playing during
a `dbench 16' (never ran 32). However, many other users (Dieter, I
believe) are on a SCSI system too.

> even i dont get any skips when i run the player at nice -n -20. That
> doesn't tell you much about the preemption patch though. And it doesn't tell
> you about performance when you dont give linux the chance to do what it does,
> multitask. That's where the latency patch is directed at improving, i
> think.

Agreed.

-- 
Robert M. Love
rml at ufl.edu
rml at tech9.net

- 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/