Re: [PATCH v2 07/13] xfs: Introduce FORCEALIGN inode flag

[Christoph Hellwig]

On Wed, Jul 10, 2024 at 07:59:58PM -0700, Darrick J. Wong wrote:
> Hmm.  If we don't support reflink + forcealign ATM, then shouldn't the
> superblock verifier or xfs_fs_fill_super fail the mount so that old
> kernels won't abruptly emit EFSCORRUPTED errors if a future kernel adds
> support for forcealign'd cow and starts writing out files with both
> iflags set?

Yes.

> That said, if the bs>ps patchset lands, then I think forcealign cow is
> a simple matter of setting the min folio order to the forcealign size
> and making sure that we always write out entire folios if any of the
> blocks cached by the folio is shared.  Direct writes to forcealigned
> shared files probably has to be aligned to the forcealign size or fall
> back to buffered writes for cow.

It has all the same problems as rtexsize > 1 + reflink, and suppoting
it will require raiding your patch stack.  Or better just wait until
we've got all that in now that we're actively working on it.