On Wed 27-08-25 13:20:56, Keith Busch wrote:

On Wed, Aug 27, 2025 at 05:20:53PM +0200, Jan Kara wrote:

quoted

Now both the old and new behavior make some sense so I won't argue that the
new iomap_iter() behavior is wrong. But I think we should change ext4 back
to the old behavior of failing unaligned dio writes instead of them falling
back to buffered IO. I think something like the attached patch should do
the trick - it makes unaligned dio writes fail again while writes to holes
of indirect-block mapped files still correctly fall back to buffered IO.
Once fstests run completes, I'll do a proper submission...

Your suggestion looks all well and good, but I have a general question
about fstests. I've written up some to test this series, and I have
filesystem specific expectations for what should error or succeed. If
you modify ext4 to fail direct-io as described, my test will have to be
kernel version specific too. Is there a best practice in fstests for
handling such scenarios?

Well, I'd just expect EINVAL for ext4 in the test. Certain kernel versions
(since February or so) will fail but that's just an indication you should
backport the fix if you care...

								Honza
-- 
Jan Kara [off-list ref]
SUSE Labs, CR