Re: [patch] voluntary-preempt-2.6.8.1-P1

From: Ingo Molnar
Date: Mon Aug 16 2004 - 05:48:02 EST



* Lee Revell <rlrevell@xxxxxxxxxxx> wrote:

> With this, and the extract_entropy hack, the biggest common latency I
> am now seeing (the copy_page_one is bigger but rarer) is the XFree86
> unmap_vmas issue. This one actually occurs so often that I can't tell
> what #2 is.

found the unmap_vmas() latency: pages get queued up for TLB flush (and
subsequent freeing) and the lock-break in unmap_vmas() didnt account for
this. When there's a preemption request and we do the lock-break it's
already too late: we first have to free possibly thousands of pages,
resulting in the latency. The solution is to make the lock-break (and
hence, the queue-flush) periodically, regardless of preemption requests.
Will release -P2 soon.

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