Hi Hans,
On Sat, Feb 18, 2023 at 11:45:04AM +0100, Hans Verkuil wrote:
> > diff --git a/drivers/gpu/drm/vc4/vc4_bo.c b/drivers/gpu/drm/vc4/vc4_bo.c
> > index 86d629e45307..d0a00ed42cb0 100644
> > --- a/drivers/gpu/drm/vc4/vc4_bo.c
> > +++ b/drivers/gpu/drm/vc4/vc4_bo.c
> > @@ -609,7 +609,7 @@ static void vc4_free_object(struct drm_gem_object *gem_bo)
> > static void vc4_bo_cache_time_work(struct work_struct *work)
> > {
> > struct vc4_dev *vc4 =
> > - container_of(work, struct vc4_dev, bo_cache.time_work);
> > + container_of_const(work, struct vc4_dev, bo_cache.time_work);
>
> ...I think this is misleading. It's definitely not const, so switching to
> container_of_const suggests that there is some 'constness' involved, which
> isn't the case. I'd leave this just as 'container_of'. This also reduces the
> size of the patch, since this is done in quite a few places.
The name threw me off too, but it's supposed to keep the argument
pointer constness, not always take and return a const pointer. I still
believe that it's beneficial since, if the work pointer was ever to
change constness, we would have that additional check.
Maxime
