[PATCH] drm/panel: Avoid warnings w/ panel-simple/panel-edp at shutdown

[Doug Anderson]

Hi,

On Wed, Jun 12, 2024 at 8:13 AM Daniel Vetter <daniel at ffwll.ch> wrote:
>
> > > I ran the coccinelle script we started with, and here are the results:
> > >
> > > ./drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:1640:25-39: ERROR: KMS driver vmw_pci_driver is missing shutdown implementation
> > > ./drivers/gpu/drm/kmb/kmb_drv.c:621:30-49: ERROR: KMS driver kmb_platform_driver is missing shutdown implementation
> > > ./drivers/gpu/drm/tiny/arcpgu.c:422:30-52: ERROR: KMS driver arcpgu_platform_driver is missing shutdown implementation
> >
> > Sure, although I think we agreed even back when we talked about this
> > last that your coccinelle script wasn't guaranteed to catch every
> > driver. ...so I guess the question is: are we willing to accept that
> > we'll stop disabling panels at shutdown for any drivers that might
> > were missed. For instance, looking at it by hand (which also could
> > miss things), I previously thought that we also might need:
> >
> > * nouveau
> > * tegra
> > * amdgpu
> > * sprd
> > * gma500
> > * radeon
> >
> > I sent patches for those drivers but they don't go through drm-misc
> > and some of the drivers had a lot of abstraction layers and were hard
> > to reason about. I'm also not 100% confident that all of those drivers
> > really are affected--they'd have to be used with panel-simple or
> > panel-edp...
>
> Aside from amdgpu and radeon they're all in -misc now, and Alex is
> generally fairly responsive.

Ah, nice! They weren't when I sent the patches ages ago. I guess I
should go ahead and repost things and maybe they'll get some traction.


> > In any case, having some sort of warning that would give us a
> > definitive answer would be nice. My proposed patch would give us that
> > warning. I could even jump to a WARN_ON right from the start.
>
> Yeah we defo want some warning to at least check this at runtime.

Yeah, my patch today currently just has a "dev_warn", but the question
is whether it would get more attention with a full on WARN_ON(). I
know WARN_ON() can be pretty controversial.

-Doug