Re: [PATCH v2 1/2] drm/radeon: process the VCE interrupt on SI and Trinity

[Christian König]

On 10/27/26 01:15, Matheus Tagliari wrote:
> Hi Timur, Christian,
> 
> Thanks for the input, Timur.
> 
> To be clear, I'm not opposed to a backport. Fixing VCE1 for people stuck
> on the radeon driver with an LTS kernel is exactly the case that made me
> look into this in the first place. But my goal is only to help, and as I
> said before, I do not want to push a change you don't want.
> 
> Christian, if you'd rather keep this out of stable, I'll drop the
> "Cc: stable" tag in v3 and leave the decision to you. If there is a
> specific concern behind it (risk of regressions on other SI/Trinity
> parts, say), I'd be glad to hear it. I can also try to address it, for
> example with more testing, so that a backport can be reconsidered later.

I was part of the Trinity bringup back (I still have the engineering sample here :) and I filled a full swear jar bringing VCE up on that HW.

Interrupts were definitely not working correctly back then.

I have no idea when they were fixed but I want to avoid that people start streaming that this change made their system not boot any more and we are forced to revert a patch on stable kernels everywhere.

> On that note, the SI part was tested on real hardware together with
> patch 2, but the Trinity/Richland part only mirrors the existing CIK and
> SI handling and has not been tested on hardware. If anyone has a
> Trinity or Richland machine and can run a VCE encode with the series
> applied, a Tested-by would help for both the review and any backport
> discussion.

I'm pretty sure my engineering sample doesn't work any more. That was an A0 stepping and those boards are only stitched together to work for a few weeks/month during initial HW bringup.

Regards,
Christian.

> Either way, I'll respin v3 with whatever you both prefer.
> 
> Thanks,
> Matheus