From: Herbert Xu <herbert@gondor.apana.org.au>
To: Jens Axboe <jens.axboe@oracle.com>
Cc: Evgeniy Polyakov <zbr@ioremap.net>, Willy Tarreau <w@1wt.eu>,
linux-kernel@vger.kernel.org, netdev@vger.kernel.org
Subject: Re: Data corruption issue with splice() on 2.6.27.10
Date: Wed, 7 Jan 2009 15:42:32 +1100 [thread overview]
Message-ID: <20090107044232.GA22218@gondor.apana.org.au> (raw)
In-Reply-To: <20090106183704.GC32491@kernel.dk>
On Tue, Jan 06, 2009 at 06:37:05PM +0000, Jens Axboe wrote:
>
> I'll give this a spin tomorrow as well. A hunch tells me that this is
> likely a page reuse issue, that splice is getting the reference to the
> buffer dropped before the data has really been transmitted. IOW, the
> page is likely fine reaching the ->sendpage() bit, but will be reused
> before the data has actually been transmitted. So once you get that far,
> other random data from that page is going out.
I see the problem.
The socket pipes in net/core/skbuff.c use references on the skb
to hold down the memory in skb->head as well as the pages in the
skb.
Unfortunately, once the pipe is fed into sendpage we only use
page reference counting to pin down the memory. So as soon as
sendpage returns we drop the ref count on the skb, thus freeing
the memory in skb->head, which is yet to be transmitted.
Moral: Using page reference counts on skb->head is wrong.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
next prev parent reply other threads:[~2009-01-07 4:43 UTC|newest]
Thread overview: 70+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-12-24 15:28 Willy Tarreau
2009-01-06 8:54 ` Jarek Poplawski
2009-01-06 9:41 ` Willy Tarreau
2009-01-06 10:01 ` Jarek Poplawski
2009-01-06 10:04 ` Willy Tarreau
2009-01-06 15:57 ` Willy Tarreau
2009-01-07 9:39 ` Jarek Poplawski
2009-01-07 12:22 ` Willy Tarreau
2009-01-07 12:24 ` Herbert Xu
2009-01-07 12:38 ` Jarek Poplawski
2009-01-07 12:31 ` Jarek Poplawski
2009-01-07 12:35 ` Jens Axboe
2009-01-07 12:40 ` Evgeniy Polyakov
2009-01-07 12:52 ` Willy Tarreau
2009-01-07 12:53 ` Herbert Xu
2009-01-07 12:57 ` Evgeniy Polyakov
2009-01-07 13:08 ` Willy Tarreau
2009-01-07 12:49 ` Jarek Poplawski
2009-01-07 12:52 ` Herbert Xu
2009-01-07 13:00 ` Willy Tarreau
2009-01-07 13:01 ` Herbert Xu
2009-01-07 13:02 ` Jarek Poplawski
2009-01-12 12:02 ` Herbert Xu
2009-01-12 12:45 ` Evgeniy Polyakov
2009-01-12 12:56 ` Herbert Xu
2009-01-12 12:59 ` Evgeniy Polyakov
2009-01-12 21:11 ` Herbert Xu
2009-01-12 13:15 ` Jarek Poplawski
2009-01-12 21:12 ` Herbert Xu
2009-01-19 7:32 ` Jarek Poplawski
2009-01-07 12:39 ` Willy Tarreau
2009-01-07 12:56 ` Jarek Poplawski
2009-01-07 12:44 ` Herbert Xu
2009-01-06 17:42 ` Ben Mansell
2009-01-06 18:15 ` Willy Tarreau
2009-01-08 7:16 ` Jarek Poplawski
2009-01-08 8:05 ` Willy Tarreau
2009-01-08 14:53 ` Ingo Molnar
2009-01-08 15:16 ` Ben Mansell
2009-01-08 17:14 ` Willy Tarreau
2009-01-06 18:32 ` Evgeniy Polyakov
2009-01-06 18:37 ` Jens Axboe
2009-01-06 18:55 ` Willy Tarreau
2009-01-07 4:42 ` Herbert Xu [this message]
2009-01-07 6:38 ` Willy Tarreau
2009-01-07 9:52 ` Herbert Xu
2009-01-07 9:54 ` Willy Tarreau
2009-01-07 11:52 ` Herbert Xu
2009-01-07 8:17 ` Jens Axboe
2009-01-07 11:29 ` Evgeniy Polyakov
2009-01-07 11:50 ` Herbert Xu
2009-01-07 11:56 ` Evgeniy Polyakov
2009-01-07 11:59 ` Herbert Xu
2009-01-07 12:15 ` Evgeniy Polyakov
2009-01-07 12:22 ` Herbert Xu
2009-01-07 12:27 ` Herbert Xu
2009-01-07 12:30 ` Herbert Xu
2009-01-07 12:37 ` Evgeniy Polyakov
2009-01-07 12:42 ` Herbert Xu
2009-01-07 12:46 ` Evgeniy Polyakov
2009-01-07 12:55 ` Willy Tarreau
2009-01-07 12:57 ` Herbert Xu
2009-01-07 13:02 ` Evgeniy Polyakov
2009-01-07 13:10 ` Jarek Poplawski
2009-01-07 13:15 ` Willy Tarreau
2009-01-07 13:22 ` Jarek Poplawski
2009-01-07 14:01 ` Jarek Poplawski
2009-01-06 18:50 ` Willy Tarreau
2009-01-19 8:39 ` Lennert Buytenhek
2009-01-19 9:53 ` Willy Tarreau
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=20090107044232.GA22218@gondor.apana.org.au \
--to=herbert@gondor.apana.org.au \
--cc=jens.axboe@oracle.com \
--cc=linux-kernel@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=w@1wt.eu \
--cc=zbr@ioremap.net \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®