Small objects that only occupy a single page are naturally contiguous, so mark them as such and allow them the special abilities that come with it.
A more thorough treatment would extend i915_gem_object_pin_map() to support discontiguous lmem objects, following the example of ioremap_prot() and use get_vm_area() + remap_io_sg(). Signed-off-by: Chris Wilson <[email protected]> Cc: Matthew Auld <[email protected]> --- drivers/gpu/drm/i915/gem/i915_gem_region.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_region.c b/drivers/gpu/drm/i915/gem/i915_gem_region.c index d50adac12249..1515384d7e0e 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_region.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_region.c @@ -107,7 +107,10 @@ void i915_gem_object_init_memory_region(struct drm_i915_gem_object *obj, { INIT_LIST_HEAD(&obj->mm.blocks); obj->mm.region = intel_memory_region_get(mem); + obj->flags |= flags; + if (obj->base.size <= mem->min_page_size) + obj->flags |= I915_BO_ALLOC_CONTIGUOUS; mutex_lock(&mem->objects.lock); -- 2.25.0.rc0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
