linux-kernel - Re: [PATCH] readahead:add blk_run_backing

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: <20090527020730.GA17658@localhost>
Date:	Wed, 27 May 2009 10:07:30 +0800
From:	Wu Fengguang <fengguang.wu@...el.com>
To:	Andrew Morton <akpm@...ux-foundation.org>
Cc:	"hifumi.hisashi@....ntt.co.jp" <hifumi.hisashi@....ntt.co.jp>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"linux-fsdevel@...r.kernel.org" <linux-fsdevel@...r.kernel.org>,
	"kosaki.motohiro@...fujitsu.com" <kosaki.motohiro@...fujitsu.com>,
	"linux-mm@...ck.org" <linux-mm@...ck.org>,
	"jens.axboe@...cle.com" <jens.axboe@...cle.com>
Subject: Re: [PATCH] readahead:add blk_run_backing_dev

On Wed, May 27, 2009 at 07:42:52AM +0800, Andrew Morton wrote:
> On Fri, 22 May 2009 10:33:23 +0800
> Wu Fengguang <fengguang.wu@...el.com> wrote:
> 
> > > I tested above patch, and I got same performance number.
> > > I wonder why if (PageUptodate(page)) check is there...
> > 
> > Thanks!  This is an interesting micro timing behavior that
> > demands some research work.  The above check is to confirm if it's
> > the PageUptodate() case that makes the difference. So why that case
> > happens so frequently so as to impact the performance? Will it also
> > happen in NFS?
> > 
> > The problem is readahead IO pipeline is not running smoothly, which is
> > undesirable and not well understood for now.
> 
> The patch causes a remarkably large performance increase.  A 9%
> reduction in time for a linear read?  I'd be surprised if the workload
> even consumed 9% of a CPU, so where on earth has the kernel gone to?
> 
> Have you been able to reproduce this in your testing?

No I cannot reproduce it on raw partition and ext4fs.

The commands I run:

        # echo 1 > /proc/sys/vm/drop_caches
        # dd if=/dev/sda1 of=/dev/null bs=16384 count=100000 # sda1 is not mounted

The results are almost identical:

before:
        1638400000 bytes (1.6 GB) copied, 31.3073 s, 52.3 MB/s
        1638400000 bytes (1.6 GB) copied, 31.3393 s, 52.3 MB/s
after:
        1638400000 bytes (1.6 GB) copied, 31.3216 s, 52.3 MB/s
        1638400000 bytes (1.6 GB) copied, 31.3762 s, 52.2 MB/s

My kernel is
        Linux hp 2.6.30-rc6 #281 SMP Wed May 27 09:32:37 CST 2009 x86_64 GNU/Linux

The readahead size is the default one:
        # blockdev --getra  /dev/sda    
        256

I tried another ext4 directory with many ~100MB files(vmlinux-2.6.*) in it:

        # time tar cf - /hp/boot | cat > /dev/null

before:
        tar cf - /hp/boot  0.22s user 5.63s system 21% cpu 26.750 total
        tar cf - /hp/boot  0.26s user 5.53s system 21% cpu 26.620 total
after:
        tar cf - /hp/boot  0.18s user 5.57s system 21% cpu 26.719 total
        tar cf - /hp/boot  0.22s user 5.32s system 21% cpu 26.321 total

Another round with 1MB readahead size:

before:
        tar cf - /hp/boot  0.24s user 4.70s system 19% cpu 25.689 total
        tar cf - /hp/boot  0.22s user 4.99s system 20% cpu 25.634 total
after:
        tar cf - /hp/boot  0.18s user 4.89s system 19% cpu 25.599 total
        tar cf - /hp/boot  0.18s user 4.97s system 20% cpu 25.645 total

Thanks,
Fengguang
--
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/