@@ -238,6 +238,7 @@ struct create_ext {
struct drm_i915_private *i915;
struct intel_memory_region *placements[INTEL_REGION_UNKNOWN];
unsigned int n_placements;
+ unsigned int placement_mask;
unsigned long flags;
};
@@ -334,6 +335,7 @@ static int set_placements(struct drm_i915_gem_create_ext_memory_regions *args,
for (i = 0; i < args->num_regions; i++)
ext_data->placements[i] = placements[i];
+ ext_data->placement_mask = mask;
return 0;
out_dump:
@@ -408,7 +410,7 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data,
struct drm_i915_gem_object *obj;
int ret;
- if (args->flags)
+ if (args->flags & ~I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS)
return -EINVAL;
ret = i915_user_extensions(u64_to_user_ptr(args->extensions),
@@ -424,13 +426,21 @@ i915_gem_create_ext_ioctl(struct drm_device *dev, void *data,
ext_data.n_placements = 1;
}
- /*
- * TODO: add a userspace hint to force CPU_ACCESS for the object, which
- * can override this.
- */
- if (ext_data.n_placements > 1 ||
- ext_data.placements[0]->type != INTEL_MEMORY_SYSTEM)
- ext_data.flags |= I915_BO_ALLOC_GPU_ONLY;
+ if (args->flags & I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS) {
+ if (ext_data.n_placements == 1)
+ return -EINVAL;
+
+ /*
+ * We always need to be able to spill to system memory, if we
+ * can't place in the mappable part of LMEM.
+ */
+ if (!(ext_data.placement_mask & BIT(INTEL_REGION_SMEM)))
+ return -EINVAL;
+ } else {
+ if (ext_data.n_placements > 1 ||
+ ext_data.placements[0]->type != INTEL_MEMORY_SYSTEM)
+ ext_data.flags |= I915_BO_ALLOC_GPU_ONLY;
+ }
obj = __i915_gem_object_create_user_ext(i915, args->size,
ext_data.placements,
@@ -3157,7 +3157,36 @@ struct drm_i915_gem_create_ext {
* Object handles are nonzero.
*/
__u32 handle;
- /** @flags: MBZ */
+ /**
+ * @flags: Optional flags.
+ *
+ * Supported values:
+ *
+ * I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS - Signal to the kernel that
+ * the object will need to be accessed via the CPU.
+ *
+ * Only valid when placing objects in I915_MEMORY_CLASS_DEVICE, and
+ * only strictly required on platforms where only some of the device
+ * memory is directly visible or mappable through the CPU, like on DG2+.
+ *
+ * One of the placements MUST also be I915_MEMORY_CLASS_SYSTEM, to
+ * ensure we can always spill the allocation to system memory, if we
+ * can't place the object in the mappable part of
+ * I915_MEMORY_CLASS_DEVICE.
+ *
+ * Note that buffers that need to be captured with EXEC_OBJECT_CAPTURE,
+ * will need to enable this hint, if the object can also be placed in
+ * I915_MEMORY_CLASS_DEVICE, starting from DG2+. The execbuf call will
+ * throw an error otherwise. This also means that such objects will need
+ * I915_MEMORY_CLASS_SYSTEM set as a possible placement.
+ *
+ * Without this hint, the kernel will assume that non-mappable
+ * I915_MEMORY_CLASS_DEVICE is preferred for this object. Note that the
+ * kernel can still migrate the object to the mappable part, as a last
+ * resort, if userspace ever CPU faults this object, but this might be
+ * expensive, and so ideally should be avoided.
+ */
+#define I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS (1 << 0)
__u32 flags;
/**
* @extensions: The chain of extensions to apply to this object.
If set, force the allocation to be placed in the mappable portion of LMEM. One big restriction here is that system memory must be given as a potential placement for the object, that way we can always spill the object into system memory if we can't make space. XXX: Still needs IGTs. Including now just for the sake of having more complete picture. Signed-off-by: Matthew Auld <matthew.auld@intel.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Jon Bloomfield <jon.bloomfield@intel.com> --- drivers/gpu/drm/i915/gem/i915_gem_create.c | 26 ++++++++++++------ include/uapi/drm/i915_drm.h | 31 +++++++++++++++++++++- 2 files changed, 48 insertions(+), 9 deletions(-)