Re: [Intel-gfx] [PATCH] drm/i915: Set correct domains values at _i915_vma_move_to_active

[Matthew Auld]

On 07/09/2022 18:26, Nirmoy Das wrote:

Fix regression introduced by commit:
"drm/i915: Individualize fences before adding to dma_resv obj"
which sets obj->read_domains to 0 for both read and write paths.
Also set obj->write_domain to 0 on read path which was removed by
the commit.

References: https://gitlab.freedesktop.org/drm/intel/-/issues/6639
Fixes: 842d9346b2fd ("drm/i915: Individualize fences before adding to dma_resv 
obj")
Signed-off-by: Nirmoy Das <nirmoy....@intel.com>
Cc: <sta...@vger.kernel.org> # v5.16+
Cc: Matthew Auld <matthew.a...@intel.com>
Cc: Andrzej Hajda <andrzej.ha...@intel.com>

Should I go ahead and push this?

---
  drivers/gpu/drm/i915/i915_vma.c | 3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index 260371716490..373582cfd8f3 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -1882,12 +1882,13 @@ int _i915_vma_move_to_active(struct i915_vma *vma,
                enum dma_resv_usage usage;
                int idx;

- obj->read_domains = 0;

                if (flags & EXEC_OBJECT_WRITE) {
                        usage = DMA_RESV_USAGE_WRITE;
                        obj->write_domain = I915_GEM_DOMAIN_RENDER;
+                       obj->read_domains = 0;
                } else {
                        usage = DMA_RESV_USAGE_READ;
+                       obj->write_domain = 0;
                }

dma_fence_array_for_each(curr, idx, fence)