From: Nick Piggin <nickpiggin@yahoo.com.au>
Subject: Re: [patch 12/17] vfs: pagecache usage optimization for pagesize!=blocksize
Date: Mon, 4 Aug 2008 17:19:42 +1000
Message-ID: <200808041719.43293.nickpiggin@yahoo.com.au>
References: <200807282246.m6SMkaHT032267@imap1.linux-foundation.org> <20080728230031.GA22218@infradead.org>
Mime-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Cc: akpm@linux-foundation.org, torvalds@linux-foundation.org,
	hifumi.hisashi@oss.ntt.co.jp, jack@ucw.cz,
	linux-ext4@vger.kernel.org
To: Christoph Hellwig <hch@infradead.org>
In-Reply-To: <20080728230031.GA22218@infradead.org>
Content-Disposition: inline
Sender: linux-ext4-owner@vger.kernel.org

On Tuesday 29 July 2008 09:00, Christoph Hellwig wrote:
> On Mon, Jul 28, 2008 at 03:46:36PM -0700, akpm@linux-foundation.org wrote:
> > From: Hisashi Hifumi <hifumi.hisashi@oss.ntt.co.jp>
> >
> > When we read some part of a file through pagecache, if there is a
> > pagecache of corresponding index but this page is not uptodate, read IO
> > is issued and this page will be uptodate.
>
> I was under the impression we wanted to do this in a nicer way than
> the hacky method?

This patch unfortunately appears like it may introduce an
uninitialized memory leak due to a data race between one
thread initializing a buffer then marking it uptodate, and
the other testing buffer uptodate then reading from the
buffer (buffer, read as: page memory covered by buffer head).

For reference, this is basically the same class of data race
that I fixed 0ed361dec36945f3116ee1338638ada9a8920905

I should have picked up on this before it was merged, but I
was kind of rushed to review other things before they got
merged.

I don't think this patch got quite enough justification to
warrant just blindly putting barriers in the buffer bitops.
The best-case numbers for it were reasonable enough when the
downside was only an extra branch or two in a relatively slow
path. I don't really know how best to go from here (maybe
someone can argue it is not a problem or come up with a better
fix?).

Thanks,
Nick