Hi Simon, On 1/13/23 14:06, Simon Ser wrote:
Hm, unfortunately I think we need to keep the check in amdgpu for the same reason as i915: amdgpu will pick a modifier if user-space didn't supply one on GFX9+. I wonder if that also applies to vmwgfx? Maybe that would be a reason to have the check in framebuffer_init()? (Not sure!)
I tried to move the check to framebuffer_init(), but it ended up causing problems in the i915 driver (the kernel was emitting warnings when running the IGT tests). I was thinking of going back to the drm_gem_fb_create() approach [1], as it would make the other drivers return EINVAL in the case of a bad modifier and it wouldn't change the current behavior of i915 and amdgpu. [1] https://lore.kernel.org/dri-devel/20230103125322.855089-1-mcanal@xxxxxxxxxx/T/ Best Regards, - Maíra Canal
