On Mon, Jul 28, 2025 at 5:25 AM Imre Deak <imre.d...@intel.com> wrote: > > Ok, thanks for testing it. I realized that it's better to fix this > aligning with the conversion for the rest of the drivers in the original > [1] patchset, could you still give a try if the corresponding patch [2] > also fixes the issue for you?
Looks like Nouveau is broken the same way: > diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c > b/drivers/gpu/drm/nouveau/nouveau_display.c > index 1ddd92901526..e1e542126310 100644 > --- a/drivers/gpu/drm/nouveau/nouveau_display.c > +++ b/drivers/gpu/drm/nouveau/nouveau_display.c > @@ -321,7 +321,7 @@ nouveau_framebuffer_new(struct drm_device *dev, > if (!(fb = *pfb = kzalloc(sizeof(*fb), GFP_KERNEL))) > return -ENOMEM; > > - drm_helper_mode_fill_fb_struct(dev, fb, mode_cmd); > + drm_helper_mode_fill_fb_struct(dev, fb, NULL, mode_cmd); > fb->obj[0] = gem; > > ret = drm_framebuffer_init(dev, fb, &nouveau_framebuffer_funcs); fb->format is NULL when drm_framebuffer_init() is called. This is in the drm-misc-next branch.
