On Thu, Apr 10, 2014 at 03:40:01AM -0700, Christoph Hellwig wrote:
> On Thu, Apr 10, 2014 at 03:00:51PM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@xxxxxxxxxx>
> >
> > When we are zeroing space andit is covered by a delalloc range, we
> > need to punch the delalloc range out before we truncate the page
> > cache. Failing to do so leaves and inconsistency between the page
> > cache and the extent tree, which we later trip over when doing
> > direct IO over the same range.
>
> Looks good.
>
> Which test found this?
The fsstress tests - any of them will do. It showed after I merged
Lukas' FALLOC_FL_ZERO_RANGE support patches for fsstress and fsx
into xfstests, but it wasn't reliably triggering until I fixed the
__xfs_get_blocks DIO bug that Al found...
> > @@ -1432,9 +1434,18 @@ xfs_zero_file_space(
> > ASSERT(end_boundary <= offset + len);
> >
> > if (start_boundary < end_boundary - 1) {
> > - /* punch out the page cache over the conversion range */
> > + /*
> > + * punch out delayed allocation blocks and the page cache over
> > + * the conversion range
> > + */
> > + xfs_ilock(ip, XFS_ILOCK_EXCL);
> > + error = xfs_bmap_punch_delalloc_range(ip,
> > + XFS_B_TO_FSB(mp, start_boundary),
>
> Shouldn't this be XFS_B_TO_FSBT?
XFS_B_TO_FSB gives the same result given that we've already rounded
start_boundary to block granularity....
Cheers,
Dave.
--
Dave Chinner
david@xxxxxxxxxxxxx
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
