ext3 umount oops in 2.5.2-pre10

ext3 umount oops in 2.5.2-pre10

[Steve Lord]

It looks like ext3 does not work if you do not use an external
journal device - the journal_bdev field is not initialized and
ext3_put_super goes belly up:


At the very least it needs this:

===========================================================================
Index: linux/fs/ext3/super.c
===========================================================================

--- /usr/tmp/TmpDir.13226-0/linux/fs/ext3/super.c_1.6	Wed Jan  9 12:38:48 2002
+++ linux/fs/ext3/super.c	Wed Jan  9 12:26:00 2002
@@ -429,7 +429,7 @@
 	J_ASSERT(list_empty(&sbi->s_orphan));
 
 	invalidate_bdev(sb->s_bdev, 0);
-	if (sbi->journal_bdev != sb->s_bdev) {
+	if (sbi->journal_bdev && (sbi->journal_bdev != sb->s_bdev)) {
 		/*
 		 * Invalidate the journal device's buffers.  We don't want them
 		 * floating about in memory - the physical journal device may



-- 

Steve

Re: ext3 umount oops in 2.5.2-pre10

[Alessandro Suardi]

Steve Lord wrote:
> 
> It looks like ext3 does not work if you do not use an external
> journal device - the journal_bdev field is not initialized and
> ext3_put_super goes belly up:

I have seen the umount oops -but- not 100% of the time. No time
 to copy the oops text since my laptop powers off, and as the
 issue occurred two times out of five or six, I haven't yet had
 a very strong need to hunt this further, confident someone more
 clueful than me (possibly the vast majority of l-k :) would do
 very soon. It looks like it happened. Heh. As always.

> At the very least it needs this:

[snipped patch]

OK, going to the usual patch/build/reboot/test sequence now.

Will let you know in 2 days at most. Thanks !

--alessandro

 "this machine will, will not communicate
   these thoughts and the strain I am under
  be a world child, form a circle before we all go under"
                         (Radiohead, "Street Spirit [fade out]")

Re: ext3 umount oops in 2.5.2-pre10

[Alessandro Suardi]

Alessandro Suardi wrote:
> 
> Steve Lord wrote:
> >
> > It looks like ext3 does not work if you do not use an external
> > journal device - the journal_bdev field is not initialized and
> > ext3_put_super goes belly up:
> 
> I have seen the umount oops -but- not 100% of the time. No time
>  to copy the oops text since my laptop powers off, and as the
>  issue occurred two times out of five or six, I haven't yet had
>  a very strong need to hunt this further, confident someone more
>  clueful than me (possibly the vast majority of l-k :) would do
>  very soon. It looks like it happened. Heh. As always.
> 
> > At the very least it needs this:
> 
> [snipped patch]
> 
> OK, going to the usual patch/build/reboot/test sequence now.

Of course the last 252p10 shutdown saw the oops, leaving me
 with the doubt that I didn't attend all the shutdowns :/

And of course the patch makes the oops go away (four reboots
 in a row without any problem). Thanks !

Ciao,

--alessandro

 "this machine will, will not communicate
   these thoughts and the strain I am under
  be a world child, form a circle before we all go under"
                         (Radiohead, "Street Spirit [fade out]")