[Intel-gfx] [PATCH 3/3] drm/i915/mtl/UAPI: Disable SET_CACHING IOCTL for MTL+

[Lucas De Marchi]

On Mon, Nov 28, 2022 at 03:43:52PM +0530, Aravind Iddamsetty wrote:
>From: Pallavi Mishra <pallavi.mishra at intel.com>
>
>Caching mode for an object shall be selected via upcoming VM_BIND
>interface.

last I've heard there was no plan to support this through VM_BIND. Did
anything change?  Otherwise this needs a better explanation recorded in
the cover letter.

According to e7737b67ab46 ("drm/i915/uapi: reject caching ioctls for discrete")
it seems it was already planned to extend this to all platforms.

+Daniel, +Matt Auld

>
>Cc: Lucas De Marchi <lucas.demarchi at intel.com>
>Cc: Matt Roper <matthew.d.roper at intel.com>
>Cc: Joonas Lahtinen <joonas.lahtinen at linux.intel.com>
>
>Signed-off-by: Pallavi Mishra <pallavi.mishra at intel.com>
>Signed-off-by: Aravind Iddamsetty <aravind.iddamsetty at intel.com>
>---
> drivers/gpu/drm/i915/gem/i915_gem_domain.c | 3 +++
> 1 file changed, 3 insertions(+)
>
>diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c
>index d44a152ce680..aebbfe186143 100644
>--- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c
>+++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c
>@@ -332,6 +332,9 @@ int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data,
> 	if (IS_DGFX(i915))
> 		return -ENODEV;
>
>+	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 70))
>+		return -EOPNOTSUPP;

Why a different return? Should this be treated similar to the IS_DGFX()
case above? It seems we are also missing an equivalent change in
i915_gem_get_caching_ioctl().

include/uapi/drm/i915_drm.h also needs to be updated with documentation
about this behavior. See the commit mentioned above.

Lucas De Marchi



>+
> 	switch (args->caching) {
> 	case I915_CACHING_NONE:
> 		level = I915_CACHE_NONE;
>-- 
>2.25.1
>