From: Ingo Molnar <mingo@elte.hu>
To: Lee Revell <rlrevell@joe-job.com>
Cc: linux-kernel <linux-kernel@vger.kernel.org>,
Andrew Morton <akpm@osdl.org>
Subject: Re: ext3 allocate-with-reservation latencies
Date: Tue, 5 Apr 2005 06:13:59 +0200 [thread overview]
Message-ID: <20050405041359.GA17265@elte.hu> (raw)
In-Reply-To: <1112673094.14322.10.camel@mindpipe>
* Lee Revell <rlrevell@joe-job.com> wrote:
> I can trigger latencies up to ~1.1 ms with a CVS checkout. It looks
> like inside ext3_try_to_allocate_with_rsv, we spend a long time in this
> loop:
>
> ext3_test_allocatable (bitmap_search_next_usable_block)
> find_next_zero_bit (bitmap_search_next_usable_block)
> find_next_zero_bit (bitmap_search_next_usable_block)
>
> ext3_test_allocatable (bitmap_search_next_usable_block)
> find_next_zero_bit (bitmap_search_next_usable_block)
> find_next_zero_bit (bitmap_search_next_usable_block)
Breaking the lock is not really possible at that point, and it doesnt
look too easy to make that path preemptable either. (To make it
preemptable rsv_lock would need to become a semaphore (this could be
fine, as it's only used when a new reservation window is created).)
The hard part is the seqlock - the read side is performance-critical,
maybe it could be solved via a preemptable but still scalable seqlock
variant that uses a semaphore for the write side? It all depends on what
the scalability impact of using a semaphore for the new-window code
would be.
the best longterm solution for these types of tradeoffs seems to be to
add a locking primitive that is a spinlock on !PREEMPT kernels and a
semaphore on PREEMPT kernels. I.e. not as drastic as a full PREEMPT_RT
kernel, but good enough to make latency-critical codepaths of ext3
preemptable, without having to hurt scalability on !PREEMPT. The
PREEMPT_RT kernel has all the 'compile-time type-switching'
infrastructure for such tricks, all that needs to be changed to switch a
lock's type is to change the spinlock definition - all the
spin_lock(&lock) uses can remain unchanged. (The same method is used on
PREEMPT_RT to have 'dual-type' spinlocks.)
the same thing could then also be used for things like the mm lock, and
other longer-held locks that PREEMPT would like to see preemptable. It
would also be a good first step towards merging the PREEMPT_RT
infrastructure ;-) I'll cook up something.
Ingo
next prev parent reply other threads:[~2005-04-05 4:14 UTC|newest]
Thread overview: 68+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-04-05 3:51 ext3 allocate-with-reservation latencies Lee Revell
2005-04-05 4:13 ` Ingo Molnar [this message]
2005-04-05 6:10 ` Mingming Cao
2005-04-05 16:38 ` Lee Revell
2005-04-06 5:35 ` Mingming Cao
2005-04-06 9:51 ` Stephen C. Tweedie
2005-04-06 16:53 ` Mingming Cao
2005-04-06 18:22 ` Stephen C. Tweedie
2005-04-06 19:03 ` Mingming Cao
2005-04-07 8:14 ` Ingo Molnar
2005-04-07 13:08 ` Stephen C. Tweedie
2005-04-07 19:16 ` Mingming Cao
2005-04-07 23:37 ` Mingming Cao
2005-04-08 14:40 ` Stephen C. Tweedie
2005-04-08 16:06 ` Arjan van de Ven
2005-04-08 18:10 ` Mingming Cao
2005-04-08 18:12 ` Lee Revell
2005-04-11 11:48 ` Stephen C. Tweedie
2005-04-11 18:38 ` Mingming Cao
2005-04-11 19:12 ` Lee Revell
2005-04-11 19:22 ` Lee Revell
2005-04-11 19:57 ` Stephen C. Tweedie
2005-04-12 6:41 ` Mingming Cao
2005-04-12 11:18 ` Stephen C. Tweedie
2005-04-12 23:27 ` Mingming Cao
2005-04-13 10:29 ` Stephen C. Tweedie
[not found] ` <1113597161.3899.80.camel@localhost.localdomain>
2005-04-18 18:00 ` Stephen C. Tweedie
2005-04-18 21:56 ` [Ext2-devel] " Mingming Cao
2005-04-22 22:10 ` [RFC][PATCH] Reduce ext3 allocate-with-reservation lock latencies Mingming Cao
2005-04-28 3:45 ` Lee Revell
2005-04-28 7:37 ` Mingming Cao
2005-04-28 16:12 ` Lee Revell
2005-04-28 18:34 ` Mingming Cao
2005-04-29 6:18 ` Mingming Cao
2005-04-28 19:14 ` [RFC] Adding multiple block allocation to current ext3 Mingming Cao
2005-04-29 13:52 ` [Ext2-devel] [RFC] Adding multiple block allocation Suparna Bhattacharya
2005-04-29 17:10 ` Mingming Cao
2005-04-29 19:42 ` Mingming Cao
2005-04-29 20:57 ` Andrew Morton
2005-04-29 21:12 ` Mingming Cao
2005-04-29 21:34 ` Andrew Morton
2005-04-30 16:00 ` Suparna Bhattacharya
2005-04-29 18:45 ` Badari Pulavarty
2005-04-29 23:22 ` Mingming Cao
2005-04-30 16:10 ` Suparna Bhattacharya
2005-04-30 17:11 ` Suparna Bhattacharya
2005-04-30 18:07 ` Mingming Cao
2005-05-02 4:46 ` Suparna Bhattacharya
2005-04-30 16:52 ` Suparna Bhattacharya
2005-04-30 0:33 ` Mingming Cao
2005-04-30 0:44 ` Mingming Cao
2005-04-30 17:03 ` Suparna Bhattacharya
2006-01-10 23:26 ` [PATCH 0/5] multiple block allocation to current ext3 Mingming Cao
2006-01-11 5:25 ` Andrew Morton
2006-01-11 19:17 ` Mingming Cao
2006-01-11 19:43 ` Andrew Morton
2006-01-11 21:31 ` Mingming Cao
2006-01-14 1:12 ` Fall back io scheduler for 2.6.15? Mingming Cao
2006-01-14 1:49 ` Andrew Morton
2006-01-14 5:22 ` Dave Jones
2006-01-16 8:43 ` Jens Axboe
2006-01-16 19:45 ` [PATCH] Fall back io scheduler ( Re: [Ext2-devel] Re: Fall back io scheduler for 2.6.15?) Mingming Cao
2006-01-16 19:49 ` Jens Axboe
2006-01-16 19:57 ` Mingming Cao
2006-01-19 19:37 ` Fall back io scheduler for 2.6.15? Nate Diller
2006-01-20 8:10 ` Jens Axboe
2006-01-16 19:38 ` [Ext2-devel] " Mingming Cao
2005-04-29 6:28 ` [PATCH] Reduce ext3 allocate-with-reservation lock latencies Mingming Cao
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=20050405041359.GA17265@elte.hu \
--to=mingo@elte.hu \
--cc=akpm@osdl.org \
--cc=linux-kernel@vger.kernel.org \
--cc=rlrevell@joe-job.com \
/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;
as well as URLs for NNTP newsgroup(s).