lkml.org 
[lkml]   [2004]   [Mar]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: CONFIG_PREEMPT and server workloads
From
Date
On Thu, 2004-03-18 at 10:37, Andrea Arcangeli wrote:
> On Thu, Mar 18, 2004 at 10:20:23AM -0500, Tom Sightler wrote:
> > Well, I reported an issue on my laptop several weeks ago where network
> > activity via my aironet wireless adapter would use 60-70% of the CPU but
> > only when PREEMPT was enabled. Looking back over the list I see other
>
> sounds like a preempt bug triggering in a softirq, or maybe an SMP bug
> triggering in UP.
>
> I certainly agree with Andrew that preempt cannot cause a 60/70%
> slowdown with a network card at least (if you do nothing but spinlocking
> in a loop then it's possible a 60/70% slowdown instead but the system
> time that a wifi card should take is nothing compared to the idle/user
> time).

Actually, I managed to get two bugs mixed together, my system had a
problem with PREEMPT and ACPI but this certainly seems to be fixed in
the current kernels, at least in 2.6.3-mm2 (my current everyday kernel)
and 2.6.5-rc1-mm2 (just a quick test tonight). Somehow the combination
of PREEMPT and any little applet that used ACPI (battery monitor) would
use a lot of CPU, then using the aironet adapter caused the CPU to
spike. Turning off either PREEMPT or ACPI would get rid of the
problem. I forgot about the ACPI part being in the mix.

I still think the aironet driver acts strange, but I'm not really sure
how to describe it. In 2.6.3-mm2 which I transfer a file top will show
60% of the cpu time in 'irq' whatever that means. It seems vmstat still
calls this idle. Perhaps I'm interpreting those numbers wrong.

When I enable preempt on the same system my network performance with the
aironet driver drops tremendously, probably by 75%, and the rest of the
systems seems to crawl. Somehow without preempt this isn't a problem,
even though it shows 60% of the time in 'irq' the rest of the system
feels responsive. What would be the proper way to track down this type
of problem?

I'm having more problems with 2.6.5-rc1-mm2 and the aironet card. It
seems to preform poorly no matter if preempt is on or off. I'm still
looking into that. I don't think the drivers has changed significantly
between the two version so I'm hoping I just made a mistake.

Later,
Tom


-
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/

\
 
 \ /
  Last update: 2005-03-22 14:01    [W:0.934 / U:0.096 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site