[Intel-gfx] [PATCH] drm/i915/gem: Hold the obj->vma.lock while walking the vma.list

Chris Wilson chris at chris-wilson.co.uk
Wed Dec 4 16:45:27 UTC 2019


Remember to take the lock before walking the obj->vma.list so that the
nodes do not change beneath us! E.g.,

i915_gem_object_bump_inactive_ggtt: i915_gem_object_bump_inactive_ggtt:387 GEM_BUG_ON(vma->vm != &i915->ggtt.vm)

Closes: https://gitlab.freedesktop.org/drm/intel/issues/691
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld at intel.com>
---
 drivers/gpu/drm/i915/gem/i915_gem_domain.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c
index f2064950622b..808eb327a29b 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c
@@ -379,8 +379,11 @@ static void i915_gem_object_bump_inactive_ggtt(struct drm_i915_gem_object *obj)
 	struct i915_vma *vma;
 
 	GEM_BUG_ON(!i915_gem_object_has_pinned_pages(obj));
+	if (!atomic_read(&obj->bind_count))
+		return;
 
 	mutex_lock(&i915->ggtt.vm.mutex);
+	spin_lock(&obj->vma.lock);
 	for_each_ggtt_vma(vma, obj) {
 		if (!drm_mm_node_allocated(&vma->node))
 			continue;
@@ -388,6 +391,7 @@ static void i915_gem_object_bump_inactive_ggtt(struct drm_i915_gem_object *obj)
 		GEM_BUG_ON(vma->vm != &i915->ggtt.vm);
 		list_move_tail(&vma->vm_link, &vma->vm->bound_list);
 	}
+	spin_unlock(&obj->vma.lock);
 	mutex_unlock(&i915->ggtt.vm.mutex);
 
 	if (i915_gem_object_is_shrinkable(obj)) {
-- 
2.24.0



More information about the Intel-gfx mailing list