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

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Mon, 29 Sep 2008 10:37:09 +0200
From:	Ingo Molnar <mingo@...e.hu>
To:	Sitsofe Wheeler <sitsofe@...oo.com>
Cc:	linux-kernel@...r.kernel.org,
	Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Arjan van de Ven <arjan@...radead.org>,
	Steven Rostedt <rostedt@...dmis.org>
Subject: Re: How how latent should non-preemptive scheduling be?


* Sitsofe Wheeler <sitsofe@...oo.com> wrote:

> Ingo Molnar wrote:
>> does it get better if you have CONFIG_PREEMPT_VOLUNTARY=y enabled? That 
>> 
>
> CONFIG_PREEMPT_VOLUNTARY=y has always been enabled in these results.

okay - then this definitely needs fixed.

>> _should_ break up this section neatly. If it doesnt then please add a  
>> might_sleep() check to kernel/kernel/semaphore.c's down_timeout()  
>> function - that is called a number of times in this trace.
>
> I added might_sleep() to the start of down_timeout() but it neither 
> printed anything to dmesg nor changed the latency issue...
>
> Strangely stalling only seems to turn up in linux-tip kernels with 
> very little debugging options set within them. I have a linux-tip that 
> has lots of extra debugging options set and this problem doesn't show 
> up...
>
> I've also asked about this on the ACPI mailing list and had a huge 
> amount of help from finding a good point for a cond_resched - 
> http://marc.info/?l=linux-acpi&m=122236450105747&w=2 . Certain things 
> (like games) still stutter when the battery is read but it's enough to 
> play back sound in rhythmbox without stuttering.
>
> Any ideas why the issue would go away with a debug kernel though?

hm, that's weird indeed. You could try to trace the full battery readout 
itself, via a script like this:

  cat /debug/tracing/trace > /dev/null        # drain trace
  cat /proc/acpi/whatever  
  cat /debug/tracing/trace > trace.txt

this way you should be seeing a full trace of the whole thing. (with the 
'ftrace' tracer plugin set in current_tracer)

then, assuming the trace.txt is complete, you can check where it 
reschedules and why. Search for 'schedule' calls. Compare the debug 
versus non-debug traces.

You can even turn on stack backtrace tracing feature, via:

  echo stacktrace > /debug/tracing/iter_ctrl

this adds extra trace entries that show the call path of every 
reschedule. [this attribute is default-disabled.]

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