Subject: Re: file corruption during emacs build on XFS logical volume
From: Steve Lord <lord@sgi.com>
To: Sean Neakums <sneakums@zork.net>
Cc: Linux XFS <linux-xfs@oss.sgi.com>
In-Reply-To: <1010429513.7120.35.camel@jen.americas.sgi.com>
References: <1010174871.30053.6.camel@jen.americas.sgi.com>
	<1010176193.2938.14.camel@UberGeek>
	<1010176393.30037.9.camel@jen.americas.sgi.com>
	<1010179700.30053.13.camel@jen.americas.sgi.com>
	<1010187371.30053.32.camel@jen.americas.sgi.com>
	<6uy9jdhdsr.fsf@zork.zork.net> <6upu4ph8c5.fsf@zork.zork.net>
	<3C37511F.6050809@sgi.com>  <6uheq0fsu6.fsf@zork.zork.net> 
	<1010429513.7120.35.camel@jen.americas.sgi.com>
Content-Type: multipart/mixed; boundary="=-rqdlBHSCOowOLYlhgOu5"
Date: 07 Jan 2002 14:35:24 -0600
Message-Id: <1010435724.15632.5.camel@jen.americas.sgi.com>
Mime-Version: 1.0
Sender: owner-linux-xfs@oss.sgi.com
Precedence: bulk
Status: O
Content-Length: 2160
Lines: 63

--=-rqdlBHSCOowOLYlhgOu5
Content-Type: text/plain
Content-Transfer-Encoding: 7bit

On Mon, 2002-01-07 at 12:51, Steve Lord wrote:
> On Sat, 2002-01-05 at 14:25, Sean Neakums wrote:
> > begin  Stephen Lord quotation:
> > 
> > > OK, thanks, this narrows it down even more than I had - I was
> > > running a kernel from Dec 22nd and recreating the problem. I had
> > > tried the individual patches to the I/O path and failed to recreate
> > > it - but maybe I should try that again.
> > 
> > I just found a kernel from December 8 (I had removed it from
> > lilo.conf, but forgotten to delete the kernel itself and the modules)
> > and I cannot recreate on that.
> > 
> 
> Sean, can you see if the attached patch against the current cvs tree
> fixes the problem for you? So far I have not been able to reproduce
> with this change. 
> 

And here is a patch I prefer you test - since I think this is more like
a real fix instead of a go back and mask the original problem again

Steve

-- 

Steve Lord                                      voice: +1-651-683-3511
Principal Engineer, Filesystem Software         email: lord@sgi.com

--=-rqdlBHSCOowOLYlhgOu5
Content-Disposition: attachment; filename=emacs.patch
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset=ISO-8859-1


=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Index: linux/fs/pagebuf/page_buf_io.c
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

--- /usr/tmp/TmpDir.15706-0/linux/fs/pagebuf/page_buf_io.c_1.104	Mon Jan  7=
 14:32:15 2002
+++ linux/fs/pagebuf/page_buf_io.c	Mon Jan  7 13:56:13 2002
@@ -121,7 +121,9 @@
 		PAGE_BUG(page);
 	if (!bh || !buffer_delay(bh))
 		BUG();
+	lock_buffer(bh);
 	clear_bit(BH_Delay, &bh->b_state);
+	unlock_buffer(bh);
 }
=20
 static inline int

--=-rqdlBHSCOowOLYlhgOu5--