Re: [PATCH v2 1/5] iomap, xfs: lift zero range hole mapping flush into xfs

[Brian Foster <bfoster@redhat.com>]

On Thu, Mar 05, 2026 at 06:11:24AM -0800, Christoph Hellwig wrote:
> On Wed, Mar 04, 2026 at 12:04:04PM -0500, Brian Foster wrote:
> > This patch seems to work on a quick test. It's basically the two patches
> > squashed together (I'd post them as independent patches), so nothing too
> > different, but if we fix up the zero logic first that helps clean up the
> > indentation as a bonus.
> 
> Cool, thanks a lot for putting in this extra effort!
> 
> Cosmetic comment on the comment:
> 
> > +	/*
> > +	 * We don't allocate blocks for zeroing a hole, but we only report a
> > +	 * hole in zoned mode if one exists in both the COW and data forks.
> 
> I'd reword this a bit as:
> 
> 	/*
> 	 * When zeroing, don't allocate blocks for holes as they already
> 	 * zeroes, but we need to ensure that no extents exist in both
> 	 * the data and COW fork to ensure this really is a hole.
> > +	 *
> > +	 * There is currently a corner case where writeback removes the COW fork
> > +	 * mapping and unlocks the inode, leaving a transient state where a hole
> > +	 * exists in both forks until write completion maps blocks into the data
> > +	 * fork. Until we can avoid this transient hole state, detect and avoid
> > +	 * this with a flush of any such range that appears dirty in pagecache.
> > +	 */
> 
> 

Sure. I had reworked the latter part of the comment as well. With both
the changes it currently looks like:

        /*
         * When zeroing, don't allocate blocks for holes as they are already
         * zeroes, but we need to ensure that no extents exist in both the data
         * and COW fork to ensure this really is a hole.
         *
         * A window exists where we might observe a hole in both forks with
         * valid data in cache. Writeback removes the COW fork blocks on
         * submission but doesn't remap into the data fork until completion. If
         * the data fork was previously a hole, we'll fail to zero. Until we
         * find a way to avoid this transient state, check for dirty pagecache
         * and flush to wait on blocks to land in the data fork.
         */

Brian