Turns out we made a silly mistake when coming up with OR inheritance on
nouveau. On pre-DCB 4.1, iors are statically routed to output paths via the
DCB. On later generations iors are only routed to an output path if they're
actually being used. Unfortunately, it appears with NVIF_OUTP_INHERIT_V0
Nevermind - I don't think I'll need the logs, I stared at the code for long
enough and I think I realized what's happening.
I will have a patch for you to test in just a moment, just waiting for it to
compile so I can verify nothing else breaks
On Wed, 2023-12-13 at 18:48 -0500, Lyude Paul
Hopefully you're still on at this point - if you are, could you try starting
the machine up with the following kernel module arguments passed to nouveau?
debug=disp=trace
Then see if you can find any lines that mention INHERIT? I have a feeling I'm
just going to have to add a workaround for the
agh - thank you for repeatedly poking on this, I've been busy enough with GSP
work I totally missed this. Yes - I'm quite surprised that this is blowing up,
but considering that looks to be a GT218 I guess display state readback must
just work a bit differently there since that's really early on
On Wed, Dec 13, 2023 at 12:39:36PM +0100, Borislav Petkov wrote:
> We're getting close to releasing so I guess we either debug this or shut
> up the WARN.
Not only that - panic_on_warn turns this into an explosion so you don't
want that in a released kernel.
--
Regards/Gruss,
Boris.
On Tue, Dec 12, 2023 at 10:35:51PM -0500, Paul Dufresne wrote:
> https://gitlab.freedesktop.org/drm/nouveau/-/issues/282
Let's add more folks who were involved in
1b477f42285e ("drm/nouveau/kms: Add INHERIT ioctl to nvkm/nvif for reading IOR
state")
Apparently, someone wants to know that the