| To: | Christoph Hellwig <hch@xxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH 8/9] xfs: catch bad block numbers freeing extents. |
| From: | Dave Chinner <david@xxxxxxxxxxxxx> |
| Date: | Thu, 7 Apr 2011 10:18:22 +1000 |
| Cc: | xfs@xxxxxxxxxxx |
| In-reply-to: | <20110406135951.GA24618@xxxxxxxxxxxxx> |
| References: | <1302070758-17312-1-git-send-email-david@xxxxxxxxxxxxx> <1302070758-17312-9-git-send-email-david@xxxxxxxxxxxxx> <20110406135951.GA24618@xxxxxxxxxxxxx> |
| User-agent: | Mutt/1.5.20 (2009-06-14) |
On Wed, Apr 06, 2011 at 09:59:51AM -0400, Christoph Hellwig wrote:
> > + /* validate the extent size is legal now we have the agf locked */
> > + if (!args.agbp) {
> > + error = EFSCORRUPTED;
> > + goto error0;
> > + }
>
> This is not a corruption case, but would require someone changing
> xfs_fix_freelist and messing up. I think the check can be removed
> given that the potentially added bug would trivially trip up as
> a NULL pointer dereference.
Well, I figured the only way we wouldn't get the agbp is if it was
corrupted somehow (e.g. it gets re-read from disk), but that should
return an error from xfs_alloc_fix_freelist() anyway. So i think you
are right - the check can be dropped.
Cheers,
Dave.
--
Dave Chinner
david@xxxxxxxxxxxxx
|
| Previous by Date: | Re: [PATCH 7/9] xfs: push the AIL from memory reclaim and periodic sync, Dave Chinner |
|---|---|
| Next by Date: | Re: [PATCH 9/9] xfs: convert log tail checking to a warning, Dave Chinner |
| Previous by Thread: | Re: [PATCH 8/9] xfs: catch bad block numbers freeing extents., Christoph Hellwig |
| Next by Thread: | [PATCH 4/9] xfs: introduce background inode reclaim work, Dave Chinner |
| Indexes: | [Date] [Thread] [Top] [All Lists] |