All of lore.kernel.org
 help / color / mirror / Atom feed
From: Theodore Ts'o <tytso@mit.edu>
To: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>, Jan Kara <jack@suse.cz>,
	LKML <linux-kernel@vger.kernel.org>,
	xfs@oss.sgi.com, Andy Lutomirski <luto@amacapital.net>,
	linux-fsdevel@vger.kernel.org, linux-ext4@vger.kernel.org,
	Tim Chen <tim.c.chen@linux.intel.com>
Subject: Re: page fault scalability (ext3, ext4, xfs)
Date: Thu, 15 Aug 2013 15:31:22 -0400	[thread overview]
Message-ID: <20130815193122.GA19536@thunk.org> (raw)
In-Reply-To: <520D13A5.2070808@linux.intel.com>

On Thu, Aug 15, 2013 at 10:45:09AM -0700, Dave Hansen wrote:
> 
> I _believe_ this is because the block allocation is occurring during the
> warmup, even in those numbers I posted previously.  will-it-scale forks
> things off early and the tests spend most of their time in those while
> loops.  Each "page fault handled" (the y-axis) is a trip through the
> while loop, *not* a call to testcase().

Ah, OK.  Sorry, I misinterpreted what was going on.

So basically, what we have going on in the test is (a) we're bumping
i_version and/or mtime, and (b) the munmap() implies an msync(), so
writeback is happening in the background concurrently with the write
page faults, and we may be (actually, almost certainly) seeing some
interference between the writeback and the page_mkwrite operations.

That implies that if you redid the test using a ramdisk, which will
significantly speed up the writeback and overhead caused by the
journal transactions for the metadata updates, the results might very
well be different.

Cheers,

						- Ted

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

WARNING: multiple messages have this Message-ID (diff)
From: "Theodore Ts'o" <tytso@mit.edu>
To: Dave Hansen <dave.hansen@linux.intel.com>
Cc: linux-fsdevel@vger.kernel.org, xfs@oss.sgi.com,
	linux-ext4@vger.kernel.org, Jan Kara <jack@suse.cz>,
	LKML <linux-kernel@vger.kernel.org>,
	david@fromorbit.com, Tim Chen <tim.c.chen@linux.intel.com>,
	Andi Kleen <ak@linux.intel.com>,
	Andy Lutomirski <luto@amacapital.net>
Subject: Re: page fault scalability (ext3, ext4, xfs)
Date: Thu, 15 Aug 2013 15:31:22 -0400	[thread overview]
Message-ID: <20130815193122.GA19536@thunk.org> (raw)
In-Reply-To: <520D13A5.2070808@linux.intel.com>

On Thu, Aug 15, 2013 at 10:45:09AM -0700, Dave Hansen wrote:
> 
> I _believe_ this is because the block allocation is occurring during the
> warmup, even in those numbers I posted previously.  will-it-scale forks
> things off early and the tests spend most of their time in those while
> loops.  Each "page fault handled" (the y-axis) is a trip through the
> while loop, *not* a call to testcase().

Ah, OK.  Sorry, I misinterpreted what was going on.

So basically, what we have going on in the test is (a) we're bumping
i_version and/or mtime, and (b) the munmap() implies an msync(), so
writeback is happening in the background concurrently with the write
page faults, and we may be (actually, almost certainly) seeing some
interference between the writeback and the page_mkwrite operations.

That implies that if you redid the test using a ramdisk, which will
significantly speed up the writeback and overhead caused by the
journal transactions for the metadata updates, the results might very
well be different.

Cheers,

						- Ted

  reply	other threads:[~2013-08-15 19:31 UTC|newest]

Thread overview: 67+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-08-14 17:10 page fault scalability (ext3, ext4, xfs) Dave Hansen
2013-08-14 17:10 ` Dave Hansen
2013-08-14 19:43 ` Theodore Ts'o
2013-08-14 19:43   ` Theodore Ts'o
2013-08-14 20:50   ` Dave Hansen
2013-08-14 20:50     ` Dave Hansen
2013-08-14 23:06     ` Theodore Ts'o
2013-08-14 23:06       ` Theodore Ts'o
2013-08-14 23:38       ` Andy Lutomirski
2013-08-15  1:11         ` Theodore Ts'o
2013-08-15  2:10           ` Dave Chinner
2013-08-15  4:32             ` Andy Lutomirski
2013-08-15  4:32               ` Andy Lutomirski
2013-08-15  6:01               ` Dave Chinner
2013-08-15  6:14                 ` Andy Lutomirski
2013-08-15  6:14                   ` Andy Lutomirski
2013-08-15  6:18                   ` David Lang
2013-08-15  6:18                     ` David Lang
2013-08-15  6:28                     ` Andy Lutomirski
2013-08-15  6:28                       ` Andy Lutomirski
2013-08-15  7:11                   ` Dave Chinner
2013-08-15  7:11                     ` Dave Chinner
2013-08-15  7:45                     ` Jan Kara
2013-08-15 21:28                       ` Dave Chinner
2013-08-15 21:28                         ` Dave Chinner
2013-08-15 21:31                         ` Andy Lutomirski
2013-08-15 21:39                           ` Dave Chinner
2013-08-19 23:23                         ` David Lang
2013-08-19 23:23                           ` David Lang
2013-08-19 23:31                           ` Andy Lutomirski
2013-08-15 15:17                     ` Andy Lutomirski
2013-08-15 15:17                       ` Andy Lutomirski
2013-08-15 21:37                       ` Dave Chinner
2013-08-15 21:37                         ` Dave Chinner
2013-08-15 21:43                         ` Andy Lutomirski
2013-08-15 21:43                           ` Andy Lutomirski
2013-08-15 22:18                           ` Dave Chinner
2013-08-15 22:18                             ` Dave Chinner
2013-08-15 22:26                             ` Andy Lutomirski
2013-08-16  0:14                               ` Dave Chinner
2013-08-16  0:21                                 ` Andy Lutomirski
2013-08-16 22:02                         ` J. Bruce Fields
2013-08-16 22:02                           ` J. Bruce Fields
2013-08-16 23:18                           ` Andy Lutomirski
2013-08-16 23:18                             ` Andy Lutomirski
2013-08-18 20:17                             ` J. Bruce Fields
2013-08-18 20:17                               ` J. Bruce Fields
2013-08-19 22:17                 ` J. Bruce Fields
2013-08-19 22:17                   ` J. Bruce Fields
2013-08-19 22:29                   ` Andy Lutomirski
2013-08-19 22:29                     ` Andy Lutomirski
2013-08-15 15:14           ` Dave Hansen
2013-08-15 15:14             ` Dave Hansen
2013-08-15  0:24 ` Dave Chinner
2013-08-15  0:24   ` Dave Chinner
2013-08-15  2:24   ` Andi Kleen
2013-08-15  2:24     ` Andi Kleen
2013-08-15  4:29     ` Dave Chinner
2013-08-15  4:29       ` Dave Chinner
2013-08-15 15:36       ` Dave Hansen
2013-08-15 15:36         ` Dave Hansen
2013-08-15 15:09   ` Dave Hansen
2013-08-15 15:05 ` Theodore Ts'o
2013-08-15 17:45   ` Dave Hansen
2013-08-15 17:45     ` Dave Hansen
2013-08-15 19:31     ` Theodore Ts'o [this message]
2013-08-15 19:31       ` Theodore Ts'o

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=20130815193122.GA19536@thunk.org \
    --to=tytso@mit.edu \
    --cc=ak@linux.intel.com \
    --cc=dave.hansen@linux.intel.com \
    --cc=jack@suse.cz \
    --cc=linux-ext4@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=luto@amacapital.net \
    --cc=tim.c.chen@linux.intel.com \
    --cc=xfs@oss.sgi.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.