From 9e7756a2ef810709a1c38779c69eebdd768bfb0f Mon Sep 17 00:00:00 2001 From: Matthew Auld Date: Mon, 21 Feb 2022 14:16:15 +0000 Subject: lib/i915_drm_local: Add I915_GEM_CREATE_EXT_FLAG_NEEDS_CPU_ACCESS MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit For now dump into i915_drm_local.h. Once the uapi on the kernel side is merged, and is part of drm-next, we can sync the kernel headers and remove this. Signed-off-by: Matthew Auld Cc: Thomas Hellström Reviewed-by: Nirmoy Das --- lib/i915/i915_drm_local.h | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) diff --git a/lib/i915/i915_drm_local.h b/lib/i915/i915_drm_local.h index 9a2273c4..ac35abf6 100644 --- a/lib/i915/i915_drm_local.h +++ b/lib/i915/i915_drm_local.h @@ -23,6 +23,27 @@ extern "C" { #define DRM_I915_QUERY_GEOMETRY_SUBSLICES 6 +/* + * 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. + * + * 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) + #if defined(__cplusplus) } #endif -- cgit v1.2.3