Before performing the domain transition for pread/pwrite, and so a locked
completion if still required, wait for any pending rendering locklessly.
This should reduce contention when reading back results from the GPU, for
example.
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
---
drivers/gpu/drm/i915/i915_gem.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index a831641..447ad9d 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -599,6 +599,10 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data,
trace_i915_gem_object_pread(obj, args->offset, args->size);
+ ret = i915_gem_object_wait_rendering__nonblocking(obj, true);
+ if (ret)
+ goto out;
+
ret = i915_gem_shmem_pread(dev, obj, args, file);
out:
@@ -1001,6 +1005,10 @@ i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
goto out;
}
+ ret = i915_gem_object_wait_rendering__nonblocking(obj, false);
+ if (ret)
+ goto out;
+
if (obj->cache_level == I915_CACHE_NONE &&
obj->tiling_mode == I915_TILING_NONE &&
obj->base.write_domain != I915_GEM_DOMAIN_CPU) {
--
1.7.10.4
