From 30bc06c0fa99a09d1132a2a5dec4bdf27426916f Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Wed, 20 Jul 2016 09:21:14 +0100 Subject: [PATCH] drm/i915: Mark imported dma-buf objects as being coherent A foreign dma-buf does not share our cache domain tracking, and we rely on the producer ensuring cache coherency. Marking them as being in the CPU domain is incorrect. v2: Add commentary about the GTT domain. This is not the best place for it, but pending an actual overhaul of our domain tracking and explaining each one, this comment should help the next reader... Signed-off-by: Chris Wilson Reviewed-by: Mika Kuoppala Reviewed-by: Joonas Lahtinen Link: http://patchwork.freedesktop.org/patch/msgid/1469002875-2335-7-git-send-email-chris@chris-wilson.co.uk --- drivers/gpu/drm/i915/i915_gem_dmabuf.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/drivers/gpu/drm/i915/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/i915_gem_dmabuf.c index 80bbe43a2e92..7d08ac08c401 100644 --- a/drivers/gpu/drm/i915/i915_gem_dmabuf.c +++ b/drivers/gpu/drm/i915/i915_gem_dmabuf.c @@ -300,6 +300,16 @@ struct drm_gem_object *i915_gem_prime_import(struct drm_device *dev, i915_gem_object_init(obj, &i915_gem_object_dmabuf_ops); obj->base.import_attach = attach; + /* We use GTT as shorthand for a coherent domain, one that is + * neither in the GPU cache nor in the CPU cache, where all + * writes are immediately visible in memory. (That's not strictly + * true, but it's close! There are internal buffers such as the + * write-combined buffer or a delay through the chipset for GTT + * writes that do require us to treat GTT as a separate cache domain.) + */ + obj->base.read_domains = I915_GEM_DOMAIN_GTT; + obj->base.write_domain = 0; + return &obj->base; fail_detach: -- 2.20.1