Re: [PATCH 0/2] drm/nouveau: fix GK107 VDPAU regressions

[lyude@redhat.com]

I am going to look at this patch when I have a moment, but I am in the
middle of moving right now so it's likely going to take me a bit,
however:

On Thu, 2026-10-22 at 21:07 +0530, Kush Chandak wrote:
> On Sun, 13 Sep 2026 23:12:23 +0300, Risto Pajula wrote:
> > This series fixes two independent Nouveau regressions affecting
> > VDPAU
> > hardware video decoding on NVIDIA GK107 (Kepler).
> 
> Tested-by: Kush Chandak <kushc451@gmail.com>
> 
> I used an AI assistant (Claude) to help debug and test this.

Please don't clog up the mailing list with this. I have seen multiple
people now going through and 'testing' things with LLMs, and the
reality is I'm not going to carry across the tested tags.

If you would like to test, there are plenty of resources available to
figure out how to build a kernel against a patch on a mailing list to
test changes. LLMs frequently hallucinate and lack an understanding of
the tasks they perform. It's already difficult enough keeping up with
code submissions, and given that you cannot use LLM output without both
vetting it, and having the knowledge required to properly vet it, there
really is nothing I can actually establish from "testing" like this. If
you weren't able to actually test without using an LLM and we have
established we can't trust the output of the LLM - how am I supposed to
know it was tested? I can't verify it from what you gave me, and even
if I could - that's just doing what is way too commonplace with these
tools and offloading the work of verifying anything that it spits to
others.