On Fri, Sep 23, 2022 at 11:16:00AM +0200, Javier Martinez Canillas wrote:
> Hello Ville,
>
> On 9/23/22 11:05, Ville Syrjälä wrote:
> > On Fri, Sep 23, 2022 at 10:34:47AM +0200, Javier Martinez Canillas wrote:
> >> The struct drm_plane .state shouldn't be accessed directly but instead the
> >> drm_atomic_get_new_plane_state() helper function should be used.
> >>
> >> This is based on a similar patch from Thomas Zimmermann for the simpledrm
> >> driver. No functional changes.
> >>
> >> Suggested-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> >> Signed-off-by: Javier Martinez Canillas <javierm@xxxxxxxxxx>
> >
> > Reviewed-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> >
>
> Thanks.
>
> > I wonder how many naked obj->state dereferences are still
> > left in places where they should be using the get_{new,old}()
> > stuff. Might have to write a bit of cocci to find out...
> >
> >
> > Btw on a somewhat related note, I've been thinking about bringing
> > for_each_crtc_in_state() & co. back (got removed in commit
> > 77ac3b00b131 ("drm/atomic: Remove deprecated accessor macros"))
> > but this time without any object state iterator variable. Now that
> > we're more often just plumbing the full atomic state through I
> > think there are bunch of places that don't need the object state(s)
> > within the loop at all, so having to have those variables around
> > makes the whole thing a bit noisy. Also IIRC we had to add some
> > (void) casts into the current macros to hide some compiler warnings
> > about unused variables. Could get rid of at least some of those extra
> > casts again.
> >
> > I don't suppose there's anyone interested in doing that so I don't
> > have to? ;)
> >
>
> Maybe you can add an entry in Documentation/gpu/todo.rst, explaining
> this and putting yourself as a contact?
Seems about as much work as just doing it.
--
Ville Syrjälä
Intel
