As we allow for parallel threads to create vma instances in parallel, and we only filter out the duplicates upon reacquiring the spinlock for the rbtree, we have to free the loser of the constructors' race. When freeing, we should also drop any resource references acquired for the redundant vma.
Fixes: 2850748ef876 ("drm/i915: Pull i915_vma_pin under the vm->mutex") Signed-off-by: Chris Wilson chris@chris-wilson.co.uk Cc: Tvrtko Ursulin tvrtko.ursulin@intel.com Cc: stable@vger.kernel.org # v5.5+ --- drivers/gpu/drm/i915/i915_vma.c | 1 + 1 file changed, 1 insertion(+)
diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index 1f63c4a1f055..7fe1f317cd2b 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -198,6 +198,7 @@ vma_create(struct drm_i915_gem_object *obj, cmp = i915_vma_compare(pos, vm, view); if (cmp == 0) { spin_unlock(&obj->vma.lock); + i915_vm_put(vm); i915_vma_free(vma); return pos; }