From: Lee Revell <rlrevell@joe-job.com>
To: Linus Torvalds <torvalds@osdl.org>
Cc: Ingo Molnar <mingo@elte.hu>, Dave Jones <davej@redhat.com>,
Hugh Dickins <hugh@veritas.com>,
linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [patch] latency tracer, 2.6.15-rc7
Date: Fri, 30 Dec 2005 20:15:31 -0500 [thread overview]
Message-ID: <1135991732.31111.57.camel@mindpipe> (raw)
In-Reply-To: <Pine.LNX.4.64.0512301701320.3249@g5.osdl.org>
On Fri, 2005-12-30 at 17:02 -0800, Linus Torvalds wrote:
>
> On Fri, 30 Dec 2005, Lee Revell wrote:
> >
> > It seems that the networking code's use of RCU can cause 10ms+
> > latencies:
>
> Hmm. Is there a big jump at the 10ms mark? Do you have a 100Hz timer
> source?
>
> A latency jump at 10ms would tend to indicate a missed wakeup that
> was "picked up" by the next timer tick.
No there are no large jumps, it really seems that this was the network
code causing an RCU callback to drop ~2K routes at once. Specifically
RCU invokes dst_rcu_free 2085 times in a single batch
(call_rcu_bh(&rt->u.dst.rcu_head, dst_rcu_free) is only called from
rt_free() and rt_drop()).
I have found that many of the paths in the network stack that can cause
high latencies can be tuned via sysctls (for example
net.ipv4.route.gc_thresh); this one may be the same.
Lee
next prev parent reply other threads:[~2005-12-31 1:15 UTC|newest]
Thread overview: 52+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-12-27 23:31 2.6.15-rc5: latency regression vs 2.6.14 in exit_mmap->free_pgtables Lee Revell
2005-12-28 2:46 ` Lee Revell
2005-12-28 22:59 ` Hugh Dickins
2005-12-29 0:00 ` Lee Revell
2005-12-29 8:22 ` [patch] latency tracer, 2.6.15-rc7 Ingo Molnar
2005-12-29 10:02 ` Dave Jones
2005-12-29 10:17 ` Ingo Molnar
2005-12-29 20:11 ` Lee Revell
2005-12-29 20:26 ` Lee Revell
2005-12-29 20:28 ` Ingo Molnar
2005-12-29 22:18 ` Lee Revell
2005-12-30 0:08 ` Grant Coady
2005-12-30 0:18 ` Lee Revell
2005-12-30 0:42 ` Grant Coady
2005-12-30 8:09 ` Ingo Molnar
2005-12-30 16:51 ` Linus Torvalds
2006-01-02 20:14 ` Ingo Molnar
2006-01-19 1:49 ` Lee Revell
2006-01-19 6:17 ` Lee Revell
2005-12-31 0:20 ` Grant Coady
2005-12-30 2:16 ` Lee Revell
2005-12-30 8:00 ` Ingo Molnar
2005-12-30 18:51 ` Lee Revell
2005-12-31 0:51 ` Lee Revell
2005-12-31 0:59 ` Mark Knecht
2005-12-31 1:16 ` Lee Revell
2005-12-31 1:30 ` Lee Revell
2005-12-31 1:02 ` Linus Torvalds
2005-12-31 1:15 ` Lee Revell [this message]
2005-12-31 1:39 ` Linus Torvalds
2005-12-31 3:54 ` Lee Revell
2005-12-31 4:00 ` Lee Revell
2005-12-31 4:29 ` Paul E. McKenney
2005-12-31 4:54 ` Lee Revell
2005-12-31 20:14 ` Paul E. McKenney
2006-01-01 5:46 ` Lee Revell
2006-01-01 18:56 ` Linus Torvalds
2006-01-01 19:02 ` Lee Revell
2006-01-01 19:06 ` Lee Revell
2006-01-03 11:12 ` Dipankar Sarma
2006-01-03 13:28 ` David Lang
2006-01-03 14:11 ` Dipankar Sarma
2006-01-03 14:09 ` Dipankar Sarma
2006-01-03 15:55 ` Paul E. McKenney
2005-12-31 4:48 ` Linus Torvalds
2006-01-01 8:32 ` Lee Revell
2006-01-06 20:28 ` Lee Revell
2005-12-29 0:54 ` 2.6.15-rc5: latency regression vs 2.6.14 in exit_mmap->free_pgtables Lee Revell
2006-01-19 1:42 ` Lee Revell
2006-01-19 7:29 ` Hugh Dickins
2006-01-19 7:32 ` Lee Revell
2006-01-19 7:35 ` Lee Revell
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1135991732.31111.57.camel@mindpipe \
--to=rlrevell@joe-job.com \
--cc=davej@redhat.com \
--cc=hugh@veritas.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=torvalds@osdl.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox