Hi Esaki-san,
On Tue, 30 Nov 2021 at 08:44, Esaki Tomohito etom@igel.co.jp wrote:
On 2021/11/18 23:05, Laurent Pinchart wrote:
On Thu, Nov 18, 2021 at 01:02:11PM +0000, Daniel Stone wrote:
Laurent's concern is that the DRM core should handle this rather than open-coding in every driver, which I agree with. Some drivers (e.g. radeon, maybe legacy NV?) do not support modifiers, and _also_ do magic inference of the actual layout of the underlying buffer. However, these drivers are legacy and we do not accept any new addition of inferring layout without modifiers.
I think the best way forward here is:
- add a new mode_config.cannot_support_modifiers flag, and enable
this in radeon (plus any other drivers in the same boat)
Is there an easy way to identify the drivers that need this ?
Should I find a driver that has not use drm_plane_funcs?
I don't think there's a good way to systematically audit it. The only two I know are radeon (i.e. pre-amdgpu) and nouveau (pre-nv50), both of which pass no modifiers to drm_universal_plane_init(), but do have magic back channels to communicate tiling information. If anyone knows of any others, well, I guess we'll find out. :)
- change drm_universal_plane_init() to advertise the LINEAR modifier
when NULL is passed as the modifier list (including installing a default .format_mod_supported hook)
... except when cannot_support_modifiers is set.
- remove the mode_config.allow_fb_modifiers hook and always
advertise modifier support, unless mode_config.cannot_support_modifiers is set
Looks good to me.
I agree with this way, I'll try to create a patches.
Thanks a lot! :)
Cheers, Daniel