Re: [PATCH] clean_bdev_aliases: Prevent cleaning blocks that are not in block range

From: Jan Kara
Date: Mon Jan 02 2017 - 11:48:49 EST


On Tue 03-01-17 00:21:45, Eryu Guan wrote:
> On Mon, Jan 02, 2017 at 04:58:03PM +0100, Jan Kara wrote:
> > On Sun 25-12-16 19:01:03, Chandan Rajendra wrote:
> > > The first block to be cleaned may start at a non-zero page offset. In
> > > such a scenario clean_bdev_aliases() will end up cleaning blocks that
> > > do not fall in the range of blocks to be cleaned. This commit fixes the
> > > issue by skipping blocks that do not fall in valid block range.
> > >
> > > Signed-off-by: Chandan Rajendra <chandan@xxxxxxxxxxxxxxxxxx>
> >
> > Ah, very good catch! How did you spot this?
>
> I failed to notice this patch, and I came up with a same patch today
> myself, and I'm still testing it.
>
> I found this by xfstests, many tests (tens of tests) failed fsck after
> test when testing extN if blocksize < pagesize. E.g. generic/013 could
> reproduce the fs corruption quite reliablely for me.
>
> Reviewed-by: Eryu Guan <eguan@xxxxxxxxxx>

Thanks! Yeah, I had run xfstests with those patches but not with blocksize
< pagesize :-| Shit happens... need to be more careful next time.

Honza

> > The patch looks good. You can add:
> >
> > Reviewed-by: Jan Kara <jack@xxxxxxx>
> >
> > Jens, please merge this fix quickly as we may end up discarding changes to
> > innocent metadata blocks due to this... Thanks!
> >
> > Honza
> > > ---
> > > fs/buffer.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/fs/buffer.c b/fs/buffer.c
> > > index 1df2bd5..28484b3 100644
> > > --- a/fs/buffer.c
> > > +++ b/fs/buffer.c
> > > @@ -1660,7 +1660,7 @@ void clean_bdev_aliases(struct block_device *bdev, sector_t block, sector_t len)
> > > head = page_buffers(page);
> > > bh = head;
> > > do {
> > > - if (!buffer_mapped(bh))
> > > + if (!buffer_mapped(bh) || (bh->b_blocknr < block))
> > > goto next;
> > > if (bh->b_blocknr >= block + len)
> > > break;
> > > --
> > > 2.5.5
> > >
> > --
> > Jan Kara <jack@xxxxxxxx>
> > SUSE Labs, CR
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-block" in
> > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
--
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR