linux-kernel - Re: Extreme slowness with xfs [WAS: Re: Slowness with new pc]

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]

Date:	Tue, 25 Nov 2008 21:39:38 +0300
From:	Michael Tokarev <mjt@....msk.ru>
To:	Justin Piszcz <jpiszcz@...idpixels.com>
CC:	Eric Sandeen <sandeen@...deen.net>,
	Stian Jordet <liste@...det.net>, linux-kernel@...r.kernel.org,
	xfs@....sgi.com, Alan Piszcz <ap@...arrain.com>
Subject: Re: Extreme slowness with xfs [WAS: Re: Slowness with new pc]

Justin Piszcz wrote:
[]
> barriers enabled:
> real    2m40.643s
> barriers disabled:
> real    0m27.612s

Barriers enabled:

$ time sh -c "tar xf linux-2.6.27.tar.bz2; sync"
real	2m3.317s
user	0m33.990s
sys	0m3.980s

$ time sh -c "rm -rf linux-2.6.27; sync"
real	1m4.033s
user	0m0.080s
sys	0m2.860s


Barriers disabled:

$ time sh -c "tar xf linux-2.6.27.tar.bz2; sync"
real	0m36.279s
user	0m25.610s
sys	0m2.800s

$ time sh -c "rm -rf linux-2.6.27; sync"
real	0m3.694s
user	0m0.010s
sys	0m2.230s


During unpack, with barriers=on, the cpu usage stays
hardy noticeable, while with barriers=off, the thing
becomes CPU-bound (needed for bzip2).

For comparison, here are results for jfs on the
same drive:

$ time sh -c "tar xf /stage/build/kernel/linux-2.6.27.tar.bz2; sync"
real	0m36.062s
user	0m25.370s
sys	0m2.860s

$ time sh -c "rm -rf linux-2.6.27; sync"
real	0m3.024s
user	0m0.040s
sys	0m0.750s

This jfs partition is located a bit further on the same disk,
so raw speed of it is a bit lower.  Yet the numbers are pretty
similar.  In any case, xfs with barriers is MUCH worse...

The disk is a 500Gig Hitachi HUA72105 one ("raid edition"),
on an AMD 780g/SB700 chipset in ahci mode.

/mjt
--
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/