linux-kernel - Re: [PATCH] mm: fix page_mkclean_one (was: 2.6.19 file content corruption on ext3)

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20061220161158.acb77ce6.akpm@osdl.org>
Date:	Wed, 20 Dec 2006 16:11:58 -0800
From:	Andrew Morton <akpm@...l.org>
To:	Linus Torvalds <torvalds@...l.org>
Cc:	Martin Michlmayr <tbm@...ius.com>,
	Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Hugh Dickins <hugh@...itas.com>,
	Nick Piggin <nickpiggin@...oo.com.au>,
	Arjan van de Ven <arjan@...radead.org>,
	Andrei Popa <andrei.popa@...eo.ro>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	Florian Weimer <fw@...eb.enyo.de>,
	Marc Haber <mh+linux-kernel@...schlus.de>,
	Martin Schwidefsky <schwidefsky@...ibm.com>,
	Heiko Carstens <heiko.carstens@...ibm.com>,
	Arnd Bergmann <arnd.bergmann@...ibm.com>,
	gordonfarquharson@...il.com,
	"Chen, Kenneth W" <kenneth.w.chen@...el.com>
Subject: Re: [PATCH] mm: fix page_mkclean_one (was: 2.6.19 file content
 corruption on ext3)

On Wed, 20 Dec 2006 15:55:14 -0800 (PST)
Linus Torvalds <torvalds@...l.org> wrote:

> > > @@ -386,12 +399,8 @@ int invalidate_inode_pages2_range(struct address_space *mapping,
> > 
> > invalidate_complete_page2() is pretty gruesome.  We're handling the case
> > where someone went and redirtied the page (and hence its buffers) after the
> > invalidate_inode_pages2() caller (generic_file_direct_IO) synced it to
> > disk.
> > 
> > I'd prefer to just fail the direct-io if someone did that, but then
> > people's tests fail and they whine.
> 
> So with my change, afaik, we will just return EIO to the invalidate, and 
> do the write.

The write's already been done by this stage.

> Which should be ok. In fact, it appears to be the only 
> possibly valid thing to do.
> 
> It really boils down to that same thing: if you remove the dirty bit, 
> there is NO CONCEIVABLE GOOD THING YOU CAN DO EXCEPT FOR:
>  - do the damn IO already ("clear_page_dirty_for_io()")
>  - truncate the page (unmap and destroy it both from page cache AND from 
>    any user-visible filesystem cases)

There's also redirty_page_for_writepage().

-
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/