Re: [PATCH v6 3/3] xfs: add support for FALLOC_FL_WRITE_ZEROES

On Thu, Jun 18, 2026 at 06:37:54AM -0700, Christoph Hellwig wrote:
> On Thu, Jun 18, 2026 at 03:26:14PM +0200, Pankaj Raghav (Samsung) wrote:
> > existing code **does not** overwrite any user data. Here is why:
> > 
> > - xfs_free_file_space (punch hole) punches inward (offset_ru -> end_rd)
> >   and zeroes out from (offset -> offset_ru) and (end_rd -> end) with
> >   xfs_zero_range
> 
> Ah, right.
> 
> > - Luckily, even though we consider offset_rd to end_ru in
> >   alloc_file_space, XFS_BMAPI_ZERO will skip zeroing already written
> >   edge blocks and only offset_ru -> end_rd are zeroed using unmap zero range.
> >   ( (offset -> offset_ru) -> EXT_NORM, (offset_ru -> end_rd) -> HOLE,
> >   (end_rd -> end) -> EXT_NORM)
> 
> Oh.  So this was lucky because of the corner cases.  I still think doing
> the proper alignment in the higher level code and only calling down for
> the area that we really want to zero would be benefitial here, if only
> to make the code easier to follow.

I agree. I will do the changes and document some of the nuances in the
code.

Thanks!

--
Pankaj