Drastically reduce the number of spin lock / unlock operations by performing
unreserving and fencing under global locks.
Signed-off-by: Thomas Hellstrom <thellstrom at vmware.com>
---
drivers/gpu/drm/ttm/ttm_bo.c | 11 +++++++--
drivers/gpu/drm/ttm/ttm_execbuf_util.c | 36 ++++++++++++++++++++++---------
include/drm/ttm/ttm_bo_driver.h | 10 ++++++++
include/drm/ttm/ttm_execbuf_util.h | 2 +
4 files changed, 45 insertions(+), 14 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index d93c73b..551a5d3 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -299,14 +299,19 @@ int ttm_bo_reserve(struct ttm_buffer_object *bo,
return ret;
}
+void ttm_bo_unreserve_locked(struct ttm_buffer_object *bo)
+{
+ ttm_bo_add_to_lru(bo);
+ atomic_set(&bo->reserved, 0);
+ wake_up_all(&bo->event_queue);
+}
+
void ttm_bo_unreserve(struct ttm_buffer_object *bo)
{
struct ttm_bo_global *glob = bo->glob;
spin_lock(&glob->lru_lock);
- ttm_bo_add_to_lru(bo);
- atomic_set(&bo->reserved, 0);
- wake_up_all(&bo->event_queue);
+ ttm_bo_unreserve_locked(bo);
spin_unlock(&glob->lru_lock);
}
EXPORT_SYMBOL(ttm_bo_unreserve);
diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c
b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
index a3ae712..7646fb5 100644
--- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c
+++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
@@ -195,22 +195,36 @@ EXPORT_SYMBOL(ttm_eu_reserve_buffers);
void ttm_eu_fence_buffer_objects(struct list_head *list, void *sync_obj)
{
struct ttm_validate_buffer *entry;
+ struct ttm_buffer_object *bo;
+ struct ttm_bo_global *glob;
+ struct ttm_bo_device *bdev;
+ struct ttm_bo_driver *driver;
- list_for_each_entry(entry, list, head) {
- struct ttm_buffer_object *bo = entry->bo;
- struct ttm_bo_device *bdev = bo->bdev;
- struct ttm_bo_driver *driver = bdev->driver;
- void *old_sync_obj;
+ if (list_empty(list))
+ return;
+
+ bo = list_first_entry(list, struct ttm_validate_buffer, head)->bo;
+ bdev = bo->bdev;
+ driver = bdev->driver;
+ glob = bo->glob;
- spin_lock(&bdev->fence_lock);
- old_sync_obj = bo->sync_obj;
+ spin_lock(&bdev->fence_lock);
+ spin_lock(&glob->lru_lock);
+
+ list_for_each_entry(entry, list, head) {
+ bo = entry->bo;
+ entry->old_sync_obj = bo->sync_obj;
bo->sync_obj = driver->sync_obj_ref(sync_obj);
bo->sync_obj_arg = entry->new_sync_obj_arg;
- spin_unlock(&bdev->fence_lock);
- ttm_bo_unreserve(bo);
+ ttm_bo_unreserve_locked(bo);
entry->reserved = false;
- if (old_sync_obj)
- driver->sync_obj_unref(&old_sync_obj);
+ }
+ spin_unlock(&glob->lru_lock);
+ spin_unlock(&bdev->fence_lock);
+
+ list_for_each_entry(entry, list, head) {
+ if (entry->old_sync_obj)
+ driver->sync_obj_unref(&entry->old_sync_obj);
}
}
EXPORT_SYMBOL(ttm_eu_fence_buffer_objects);
diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h
index ca8131e..cfb9ca4 100644
--- a/include/drm/ttm/ttm_bo_driver.h
+++ b/include/drm/ttm/ttm_bo_driver.h
@@ -910,6 +910,16 @@ extern int ttm_bo_reserve_locked(struct ttm_buffer_object
*bo,
extern void ttm_bo_unreserve(struct ttm_buffer_object *bo);
/**
+ * ttm_bo_unreserve_locked
+ *
+ * @bo: A pointer to a struct ttm_buffer_object.
+ *
+ * Unreserve a previous reservation of @bo.
+ * Needs to be called with struct ttm_bo_global::lru_lock held.
+ */
+extern void ttm_bo_unreserve_locked(struct ttm_buffer_object *bo);
+
+/**
* ttm_bo_wait_unreserved
*
* @bo: A pointer to a struct ttm_buffer_object.
diff --git a/include/drm/ttm/ttm_execbuf_util.h
b/include/drm/ttm/ttm_execbuf_util.h
index fd09b84..535ab00 100644
--- a/include/drm/ttm/ttm_execbuf_util.h
+++ b/include/drm/ttm/ttm_execbuf_util.h
@@ -44,6 +44,7 @@
* @reserved: Indicates whether @bo has been reserved for validation.
* @removed: Indicates whether @bo has been removed from lru lists.
* @put_count: Number of outstanding references on bo::list_kref.
+ * @old_sync_obj: Pointer to a sync object about to be unreferenced
*/
struct ttm_validate_buffer {
@@ -53,6 +54,7 @@ struct ttm_validate_buffer {
bool reserved;
bool removed;
int put_count;
+ void *old_sync_obj;
};
/**
--
1.6.2.5
