summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/i915/i915_vma.c
diff options
context:
space:
mode:
authorChris Wilson <chris@chris-wilson.co.uk>2019-06-21 19:37:59 +0100
committerChris Wilson <chris@chris-wilson.co.uk>2019-06-21 19:47:51 +0100
commita93615f900bd19b59e74e04f7d8d4663ee5ea68f (patch)
tree02e0dceeb8c404520f4d5aa10b2a106199c1b52d /drivers/gpu/drm/i915/i915_vma.c
parent5361db1a33c7e2d58af7df045d4d3ddd4c87ab56 (diff)
drm/i915: Throw away the active object retirement complexity
Remove the accumulated optimisations that we have for i915_vma_retire and reduce it to the bare essential of tracking the active object reference. This allows us to only use atomic operations, and so will be able to avoid the struct_mutex requirement. The principal loss here is the shrinker MRU bumping, so now if we have to shrink, we will do so in much more random order and more likely to try and shrink recently used objects. That is a nuisance, but shrinking active objects is a second step we try to avoid and will always be a system-wide performance issue. The other loss is here is in the automatic pruning of the reservation_object when idling. This is not as large an issue as upon reservation_object introduction as now adding new fences into the object replaces already signaled fences, keeping the array compact. But we do lose the auto-expiration of stale fences and unused arrays. That may be a noticeable problem for which we need to re-implement autopruning. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190621183801.23252-3-chris@chris-wilson.co.uk
Diffstat (limited to 'drivers/gpu/drm/i915/i915_vma.c')
-rw-r--r--drivers/gpu/drm/i915/i915_vma.c54
1 files changed, 11 insertions, 43 deletions
diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index 503f1180af12..c13b86e6ef1f 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -78,43 +78,11 @@ static void vma_print_allocator(struct i915_vma *vma, const char *reason)
#endif
-static void obj_bump_mru(struct drm_i915_gem_object *obj)
-{
- struct drm_i915_private *i915 = to_i915(obj->base.dev);
- unsigned long flags;
-
- spin_lock_irqsave(&i915->mm.obj_lock, flags);
- list_move_tail(&obj->mm.link, &i915->mm.shrink_list);
- spin_unlock_irqrestore(&i915->mm.obj_lock, flags);
-
- obj->mm.dirty = true; /* be paranoid */
-}
-
static void __i915_vma_retire(struct i915_active *ref)
{
struct i915_vma *vma = container_of(ref, typeof(*vma), active);
- struct drm_i915_gem_object *obj = vma->obj;
-
- GEM_BUG_ON(!i915_gem_object_is_active(obj));
- if (--obj->active_count)
- return;
-
- /* Prune the shared fence arrays iff completely idle (inc. external) */
- if (reservation_object_trylock(obj->base.resv)) {
- if (reservation_object_test_signaled_rcu(obj->base.resv, true))
- reservation_object_add_excl_fence(obj->base.resv, NULL);
- reservation_object_unlock(obj->base.resv);
- }
- /*
- * Bump our place on the bound list to keep it roughly in LRU order
- * so that we don't steal from recently used but inactive objects
- * (unless we are forced to ofc!)
- */
- if (i915_gem_object_is_shrinkable(obj))
- obj_bump_mru(obj);
-
- i915_gem_object_put(obj); /* and drop the active reference */
+ i915_vma_put(vma);
}
static struct i915_vma *
@@ -922,6 +890,7 @@ int i915_vma_move_to_active(struct i915_vma *vma,
unsigned int flags)
{
struct drm_i915_gem_object *obj = vma->obj;
+ int err;
assert_vma_held(vma);
assert_object_held(obj);
@@ -935,17 +904,13 @@ int i915_vma_move_to_active(struct i915_vma *vma,
* add the active reference first and queue for it to be dropped
* *last*.
*/
- if (!vma->active.count && !obj->active_count++)
- i915_gem_object_get(obj); /* once more for the active ref */
-
- if (unlikely(i915_active_ref(&vma->active, rq->fence.context, rq))) {
- if (!vma->active.count && !--obj->active_count)
- i915_gem_object_put(obj);
- return -ENOMEM;
- }
+ if (i915_active_acquire(&vma->active))
+ i915_vma_get(vma);
- GEM_BUG_ON(!i915_vma_is_active(vma));
- GEM_BUG_ON(!obj->active_count);
+ err = i915_active_ref(&vma->active, rq->fence.context, rq);
+ i915_active_release(&vma->active);
+ if (unlikely(err))
+ return err;
obj->write_domain = 0;
if (flags & EXEC_OBJECT_WRITE) {
@@ -957,11 +922,14 @@ int i915_vma_move_to_active(struct i915_vma *vma,
obj->read_domains = 0;
}
obj->read_domains |= I915_GEM_GPU_DOMAINS;
+ obj->mm.dirty = true;
if (flags & EXEC_OBJECT_NEEDS_FENCE)
__i915_active_request_set(&vma->last_fence, rq);
export_fence(vma, rq, flags);
+
+ GEM_BUG_ON(!i915_vma_is_active(vma));
return 0;
}