linux-kernel - Re: [GIT PULL] Ext3 latency fixes

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]

Message-ID: <alpine.LFD.2.00.0904051227120.4023@localhost.localdomain>
Date:	Sun, 5 Apr 2009 12:34:32 -0700 (PDT)
From:	Linus Torvalds <torvalds@...ux-foundation.org>
To:	Arjan van de Ven <arjan@...radead.org>
cc:	Theodore Tso <tytso@....edu>, Jens Axboe <jens.axboe@...cle.com>,
	Linux Kernel Developers List <linux-kernel@...r.kernel.org>,
	Ext4 Developers List <linux-ext4@...r.kernel.org>
Subject: Re: [GIT PULL] Ext3 latency fixes

On Sun, 5 Apr 2009, Arjan van de Ven wrote:
>
> > See get_request():
> 
> our default number of requests is so low that we very regularly hit the
> limit. In addition to setting kjournald to higher priority, I tend to
> set the number of requests to 4096 or so to improve interactive
> performance on my own systems. That way at least the elevator has a
> chance to see the requests ;-)

That's insane. Long queues make the problem harder to hit, yes. But it 
also tends to make the problem them a million times worse when you _do_ 
hit it.

I would suggest looking instead at trying to have separate allocation 
pools for bulk and "sync" IO. Instead of having just one rq->rq_pool, we 
could easily have a rq->rq_bulk_pool and rq->rq_sync_pool.

We might even _save_ memory by having two pools simply because that may 
make it much less important to have a big pool. Most subsystems don't 
really need that many requests in flight anyway, and the advantage to the 
elevator of huge pools is rather dubious.

So you obviously need more requests than the hardware can have in flight 
(since you want to be able to feed the hardware new requests and overlap 
the refill with the ones executing), but 4096 sounds excessive if you're 
doing something like SATA that can only have 32 actual outstanding 
requests at the hardware.

But yes, if a synchronous request gets blocked just because we've already 
used all the requests, latency will be suffer. 

			Linus
--
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/