Re: [PATCH 2/2] gpu: drm: tyr: use IoMem directly instead of Devres

[Deborah Brouwer <deborah.brouwer@collabora.com>]

On Thu, May 28, 2026 at 01:58:35PM +0100, Gary Guo wrote:
> On Thu May 28, 2026 at 12:31 AM BST, Danilo Krummrich wrote:
> > On Thu May 28, 2026 at 1:03 AM CEST, Deborah Brouwer wrote:
> >> On Tue, May 26, 2026 at 11:43:38AM +0900, Eliot Courtney wrote:
> >>> On Tue May 26, 2026 at 8:01 AM JST, Danilo Krummrich wrote:
> >>> > @@ -40,9 +39,7 @@
> >>> >  pub(crate) struct GpuInfo(pub(crate) uapi::drm_panthor_gpu_info);
> >>> >  
> >>> >  impl GpuInfo {
> >>> > -    pub(crate) fn new(dev: &Device<Bound>, iomem: &Devres<IoMem>) -> Result<Self> {
> >>> > -        let io = (*iomem).access(dev)?;
> >>> > -
> >>> > +    pub(crate) fn new(io: &IoMem<'_>) -> Result<Self> {
> >>> 
> >>> Perhaps a future patch can remove this now unnecessary looking Result in
> >>> the return type.
> >>
> >> Oh that's right now that we don't have: let io = (*iomem).access(dev)?
> >> we don't need the Result anymore.
> >>
> >> Danilo, I can add this fix to our fw-boot series which I'm preparing to sit
> >> on top of these changes.
> >
> > I can send a v2, fix it up on apply when I get an ACK, have Alice fix it up on
> > apply, or have it done in a follow-up as you offer -- just let me know. :)
> >
> 
> I think it makes sense to have a v2 with this, as getting rid of impossible
> failure paths is the reason we go for lifetime in the first place, so this is a
> good demonstration of that benefit.

yeah v2 is probably the cleanest/easiest way to fix.

And while here it would be nice to keep the subject prefix consistent,
we've been using "drm/tyr:"

> 
> Best,
> Gary