On 10/05/2022 21:33, Andi Shyti wrote:
During object cleanup we invalidate the TLBs but we do it only
for gt0. Invalidate the caches for all the tiles.
Reported-by: Chris Wilson <ch...@chris-wilson.co.uk>
Signed-off-by: Andi Shyti <andi.sh...@linux.intel.com>
---
drivers/gpu/drm/i915/gem/i915_gem_pages.c | 11 ++++++++---
1 file changed, 8 insertions(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_pages.c
b/drivers/gpu/drm/i915/gem/i915_gem_pages.c
index 97c820eee115a..444b9f96ba77c 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_pages.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_pages.c
@@ -217,10 +217,15 @@ __i915_gem_object_unset_pages(struct drm_i915_gem_object
*obj)
if (test_and_clear_bit(I915_BO_WAS_BOUND_BIT, &obj->flags)) {
struct drm_i915_private *i915 = to_i915(obj->base.dev);
- intel_wakeref_t wakeref;
+ struct intel_gt *gt;
+ int i;
- with_intel_runtime_pm_if_active(&i915->runtime_pm, wakeref)
- intel_gt_invalidate_tlbs(to_gt(i915));
+ for_each_gt(gt, i915, i) {
+ intel_wakeref_t w;
+
+ with_intel_runtime_pm_if_active(gt->uncore->rpm, w)
+ intel_gt_invalidate_tlbs(gt);
+ }
LGTM.
Reviewed-by: Tvrtko Ursulin <tvrtko.ursu...@intel.com>
How about an improvement, either immediately or as followup, to define per tile
bound flags and so be able to avoid trashing innocent tiles?
Like:
i915_vma_bind:
set_bit(I915_BO_WAS_BOUND_BIT(vma->vm->gt->info.id), &vma->obj->flags);
__i915_gem_object_unset_pages:
for_each_gt(gt, i915, i) {
if (!test_and_clear_bit(I915_BO_WAS_BOUND_BIT(i), &obj->flags))
continue;
...
}
Regards,
Tvrtko
}
return pages;
