On Thu, Jan 29, 2026 at 04:29:22PM +0100, Sebastian Andrzej Siewior wrote:

On 2026-01-26 10:28:18 [+0100], Felix Maurer wrote:

quoted

I'll wait for feedback from Sebastian, but I'm also fine with merging
the fixes for the KUnit test into the other patches and reposting.

Overall the new algorithm is very nice. Feel free to add
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Thank you very much for taking the time to review this series! I'll
address the parts as I wrote in the other replies.

As Jakub wrote, there is the transient build failure due to the KUnit
test needing changes. I discussed that with Simon as well (off-list). In
the v3, I'll add the respective fixes to the KUnit test to the patches
where I change the algorithm so the build passes all the time. As the
patches 4 and 6 will look quite different with that, I'll only add your
Reviewed by tag to the remaining ones and leave 4 and 6 subject to a
quick re-review by you.

I made a few comments where I think it should be possible to avoid
tracking the sequence number for interface A and B but since you
preserve what is done now, I don't mind to keep it as-it and try to
remove it later on (unless I'm wrong and it is not doable).

I replied to that directly, I don't think that's doable. But we can take
a look at the HW-offload case in the future.

Thanks,
   Felix