Re: Another -pre

Neil Brown (neilb@cse.unsw.edu.au)
Tue, 4 Jun 2002 12:46:50 +1000 (EST)


On Monday June 3, adilger@clusterfs.com wrote:
> On Jun 03, 2002 20:27 -0300, Arnaldo Carvalho de Melo wrote:
> > > 1 Weird corruption report with AMD chipset in PIO mode
> >
> > Oh, I'm not alone ;) Well, up to now it _seems_ that ext3 is saving my day,
> > but it only happened two time after I upgraded to 2.4.19-pre8-ac5, none after
> > I upgraded to 2.4.19-pre9-ac3, but I can't manage to make 'hdparm -X68 /dev/hdd'
> > to work :( I have already sent detailed information to Andre and discussed
> > and tried several things sugested in a irc chat.
> >
> > Short description: I use ext3 over raid0, using /dev/hda4 and /dev/hdd1,
> > /dev/hdc has a CDRW drive, mostly unused, /dev/hdb has nothing, two times
> > /dev/hda stopped responding, not reproducible AFAIT.
>
> Well, there was some corruption in ext3 if you used it over MD RAID with
> data=journal mode that was discussed recently on ext3-users. There was
> a patch posted by Neil Brown which I resend here (full thread archived
> at https://listman.redhat.com/pipermail/ext3-users/).

It turns out that I missed an important bit (literally) in that patch
(and another few files got corrupted....)
Just after (or before)
+ clear_bit(BH_Freed, &bh->b_state);
We need
+ clear_bit(BH_JBDDirty, &bh->b_state);

because __journal_unfile_buffer will convert JBDdirty to Dirty just
like __journal_refile_buffer does.

NeilBrown
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/