Merge branch 'drm-next' of git://people.freedesktop.org/~airlied/linux

Pull drm merge from Dave Airlie:
 "Highlights:

   - TI LCD controller KMS driver

   - TI OMAP KMS driver merged from staging

   - drop gma500 stub driver

   - the fbcon locking fixes

   - the vgacon dirty like zebra fix.

   - open firmware videomode and hdmi common code helpers

   - major locking rework for kms object handling - pageflip/cursor
     won't block on polling anymore!

   - fbcon helper and prime helper cleanups

   - i915: all over the map, haswell power well enhancements, valleyview
     macro horrors cleaned up, killing lots of legacy GTT code,

   - radeon: CS ioctl unification, deprecated UMS support, gpu reset
     rework, VM fixes

   - nouveau: reworked thermal code, external dp/tmds encoder support
     (anx9805), fences sleep instead of polling,

   - exynos: all over the driver fixes."

Lovely conflict in radeon/evergreen_cs.c between commit de0babd60d8d
("drm/radeon: enforce use of radeon_get_ib_value when reading user cmd")
and the new changes that modified that evergreen_dma_cs_parse()
function.

* 'drm-next' of git://people.freedesktop.org/~airlied/linux: (508 commits)
  drm/tilcdc: only build on arm
  drm/i915: Revert hdmi HDP pin checks
  drm/tegra: Add list of framebuffers to debugfs
  drm/tegra: Fix color expansion
  drm/tegra: Split DC_CMD_STATE_CONTROL register write
  drm/tegra: Implement page-flipping support
  drm/tegra: Implement VBLANK support
  drm/tegra: Implement .mode_set_base()
  drm/tegra: Add plane support
  drm/tegra: Remove bogus tegra_framebuffer structure
  drm: Add consistency check for page-flipping
  drm/radeon: Use generic HDMI infoframe helpers
  drm/tegra: Use generic HDMI infoframe helpers
  drm: Add EDID helper documentation
  drm: Add HDMI infoframe helpers
  video: Add generic HDMI infoframe helpers
  drm: Add some missing forward declarations
  drm: Move mode tables to drm_edid.c
  drm: Remove duplicate drm_mode_cea_vic()
  gma500: Fix n, m1 and m2 clock limits for sdvo and lvds
  ...

2 files changed, 126 insertions, 55 deletions

diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index 52b20b12c83a..9b07b7d44a58 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -158,7 +158,8 @@ static void ttm_bo_release_list(struct kref *list_kref)
 	ttm_mem_global_free(bdev->glob->mem_glob, acc_size);
 }
 
-int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo, bool interruptible)
+static int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo,
+				  bool interruptible)
 {
 	if (interruptible) {
 		return wait_event_interruptible(bo->event_queue,
@@ -168,7 +169,6 @@ int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo, bool interruptible)
 		return 0;
 	}
 }
-EXPORT_SYMBOL(ttm_bo_wait_unreserved);
 
 void ttm_bo_add_to_lru(struct ttm_buffer_object *bo)
 {
@@ -213,14 +213,13 @@ int ttm_bo_del_from_lru(struct ttm_buffer_object *bo)
 	return put_count;
 }
 
-int ttm_bo_reserve_locked(struct ttm_buffer_object *bo,
+int ttm_bo_reserve_nolru(struct ttm_buffer_object *bo,
 			  bool interruptible,
 			  bool no_wait, bool use_sequence, uint32_t sequence)
 {
-	struct ttm_bo_global *glob = bo->glob;
 	int ret;
 
-	while (unlikely(atomic_read(&bo->reserved) != 0)) {
+	while (unlikely(atomic_xchg(&bo->reserved, 1) != 0)) {
 		/**
 		 * Deadlock avoidance for multi-bo reserving.
 		 */
@@ -241,26 +240,36 @@ int ttm_bo_reserve_locked(struct ttm_buffer_object *bo,
 		if (no_wait)
 			return -EBUSY;
 
-		spin_unlock(&glob->lru_lock);
 		ret = ttm_bo_wait_unreserved(bo, interruptible);
-		spin_lock(&glob->lru_lock);
 
 		if (unlikely(ret))
 			return ret;
 	}
 
-	atomic_set(&bo->reserved, 1);
 	if (use_sequence) {
+		bool wake_up = false;
 		/**
 		 * Wake up waiters that may need to recheck for deadlock,
 		 * if we decreased the sequence number.
 		 */
 		if (unlikely((bo->val_seq - sequence < (1 << 31))
 			     || !bo->seq_valid))
-			wake_up_all(&bo->event_queue);
+			wake_up = true;
 
+		/*
+		 * In the worst case with memory ordering these values can be
+		 * seen in the wrong order. However since we call wake_up_all
+		 * in that case, this will hopefully not pose a problem,
+		 * and the worst case would only cause someone to accidentally
+		 * hit -EAGAIN in ttm_bo_reserve when they see old value of
+		 * val_seq. However this would only happen if seq_valid was
+		 * written before val_seq was, and just means some slightly
+		 * increased cpu usage
+		 */
 		bo->val_seq = sequence;
 		bo->seq_valid = true;
+		if (wake_up)
+			wake_up_all(&bo->event_queue);
 	} else {
 		bo->seq_valid = false;
 	}
@@ -289,17 +298,64 @@ int ttm_bo_reserve(struct ttm_buffer_object *bo,
 	int put_count = 0;
 	int ret;
 
-	spin_lock(&glob->lru_lock);
-	ret = ttm_bo_reserve_locked(bo, interruptible, no_wait, use_sequence,
-				    sequence);
-	if (likely(ret == 0))
+	ret = ttm_bo_reserve_nolru(bo, interruptible, no_wait, use_sequence,
+				   sequence);
+	if (likely(ret == 0)) {
+		spin_lock(&glob->lru_lock);
 		put_count = ttm_bo_del_from_lru(bo);
-	spin_unlock(&glob->lru_lock);
+		spin_unlock(&glob->lru_lock);
+		ttm_bo_list_ref_sub(bo, put_count, true);
+	}
 
-	ttm_bo_list_ref_sub(bo, put_count, true);
+	return ret;
+}
+
+int ttm_bo_reserve_slowpath_nolru(struct ttm_buffer_object *bo,
+				  bool interruptible, uint32_t sequence)
+{
+	bool wake_up = false;
+	int ret;
+
+	while (unlikely(atomic_xchg(&bo->reserved, 1) != 0)) {
+		WARN_ON(bo->seq_valid && sequence == bo->val_seq);
+
+		ret = ttm_bo_wait_unreserved(bo, interruptible);
 
+		if (unlikely(ret))
+			return ret;
+	}
+
+	if ((bo->val_seq - sequence < (1 << 31)) || !bo->seq_valid)
+		wake_up = true;
+
+	/**
+	 * Wake up waiters that may need to recheck for deadlock,
+	 * if we decreased the sequence number.
+	 */
+	bo->val_seq = sequence;
+	bo->seq_valid = true;
+	if (wake_up)
+		wake_up_all(&bo->event_queue);
+
+	return 0;
+}
+
+int ttm_bo_reserve_slowpath(struct ttm_buffer_object *bo,
+			    bool interruptible, uint32_t sequence)
+{
+	struct ttm_bo_global *glob = bo->glob;
+	int put_count, ret;
+
+	ret = ttm_bo_reserve_slowpath_nolru(bo, interruptible, sequence);
+	if (likely(!ret)) {
+		spin_lock(&glob->lru_lock);
+		put_count = ttm_bo_del_from_lru(bo);
+		spin_unlock(&glob->lru_lock);
+		ttm_bo_list_ref_sub(bo, put_count, true);
+	}
 	return ret;
 }
+EXPORT_SYMBOL(ttm_bo_reserve_slowpath);
 
 void ttm_bo_unreserve_locked(struct ttm_buffer_object *bo)
 {
@@ -511,7 +567,7 @@ static void ttm_bo_cleanup_refs_or_queue(struct ttm_buffer_object *bo)
 	int ret;
 
 	spin_lock(&glob->lru_lock);
-	ret = ttm_bo_reserve_locked(bo, false, true, false, 0);
+	ret = ttm_bo_reserve_nolru(bo, false, true, false, 0);
 
 	spin_lock(&bdev->fence_lock);
 	(void) ttm_bo_wait(bo, false, false, true);
@@ -604,7 +660,7 @@ static int ttm_bo_cleanup_refs_and_unlock(struct ttm_buffer_object *bo,
 			return ret;
 
 		spin_lock(&glob->lru_lock);
-		ret = ttm_bo_reserve_locked(bo, false, true, false, 0);
+		ret = ttm_bo_reserve_nolru(bo, false, true, false, 0);
 
 		/*
 		 * We raced, and lost, someone else holds the reservation now,
@@ -668,7 +724,14 @@ static int ttm_bo_delayed_delete(struct ttm_bo_device *bdev, bool remove_all)
 			kref_get(&nentry->list_kref);
 		}
 
-		ret = ttm_bo_reserve_locked(entry, false, !remove_all, false, 0);
+		ret = ttm_bo_reserve_nolru(entry, false, true, false, 0);
+		if (remove_all && ret) {
+			spin_unlock(&glob->lru_lock);
+			ret = ttm_bo_reserve_nolru(entry, false, false,
+						   false, 0);
+			spin_lock(&glob->lru_lock);
+		}
+
 		if (!ret)
 			ret = ttm_bo_cleanup_refs_and_unlock(entry, false,
 							     !remove_all);
@@ -816,7 +879,7 @@ static int ttm_mem_evict_first(struct ttm_bo_device *bdev,
 
 	spin_lock(&glob->lru_lock);
 	list_for_each_entry(bo, &man->lru, lru) {
-		ret = ttm_bo_reserve_locked(bo, false, true, false, 0);
+		ret = ttm_bo_reserve_nolru(bo, false, true, false, 0);
 		if (!ret)
 			break;
 	}
@@ -1797,7 +1860,7 @@ static int ttm_bo_swapout(struct ttm_mem_shrink *shrink)
 
 	spin_lock(&glob->lru_lock);
 	list_for_each_entry(bo, &glob->swap_lru, swap) {
-		ret = ttm_bo_reserve_locked(bo, false, true, false, 0);
+		ret = ttm_bo_reserve_nolru(bo, false, true, false, 0);
 		if (!ret)
 			break;
 	}
diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
index cd9e4523dc56..7b90def15674 100644
--- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c
+++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
@@ -82,22 +82,6 @@ static void ttm_eu_list_ref_sub(struct list_head *list)
 	}
 }
 
-static int ttm_eu_wait_unreserved_locked(struct list_head *list,
-					 struct ttm_buffer_object *bo)
-{
-	struct ttm_bo_global *glob = bo->glob;
-	int ret;
-
-	ttm_eu_del_from_lru_locked(list);
-	spin_unlock(&glob->lru_lock);
-	ret = ttm_bo_wait_unreserved(bo, true);
-	spin_lock(&glob->lru_lock);
-	if (unlikely(ret != 0))
-		ttm_eu_backoff_reservation_locked(list);
-	return ret;
-}
-
-
 void ttm_eu_backoff_reservation(struct list_head *list)
 {
 	struct ttm_validate_buffer *entry;
@@ -145,47 +129,65 @@ int ttm_eu_reserve_buffers(struct list_head *list)
 	entry = list_first_entry(list, struct ttm_validate_buffer, head);
 	glob = entry->bo->glob;
 
-retry:
 	spin_lock(&glob->lru_lock);
 	val_seq = entry->bo->bdev->val_seq++;
 
+retry:
 	list_for_each_entry(entry, list, head) {
 		struct ttm_buffer_object *bo = entry->bo;
 
-retry_this_bo:
-		ret = ttm_bo_reserve_locked(bo, true, true, true, val_seq);
+		/* already slowpath reserved? */
+		if (entry->reserved)
+			continue;
+
+		ret = ttm_bo_reserve_nolru(bo, true, true, true, val_seq);
 		switch (ret) {
 		case 0:
 			break;
 		case -EBUSY:
-			ret = ttm_eu_wait_unreserved_locked(list, bo);
-			if (unlikely(ret != 0)) {
-				spin_unlock(&glob->lru_lock);
-				ttm_eu_list_ref_sub(list);
-				return ret;
-			}
-			goto retry_this_bo;
+			ttm_eu_del_from_lru_locked(list);
+			spin_unlock(&glob->lru_lock);
+			ret = ttm_bo_reserve_nolru(bo, true, false,
+						   true, val_seq);
+			spin_lock(&glob->lru_lock);
+			if (!ret)
+				break;
+
+			if (unlikely(ret != -EAGAIN))
+				goto err;
+
+			/* fallthrough */
 		case -EAGAIN:
 			ttm_eu_backoff_reservation_locked(list);
+
+			/*
+			 * temporarily increase sequence number every retry,
+			 * to prevent us from seeing our old reservation
+			 * sequence when someone else reserved the buffer,
+			 * but hasn't updated the seq_valid/seqno members yet.
+			 */
+			val_seq = entry->bo->bdev->val_seq++;
+
 			spin_unlock(&glob->lru_lock);
 			ttm_eu_list_ref_sub(list);
-			ret = ttm_bo_wait_unreserved(bo, true);
+			ret = ttm_bo_reserve_slowpath_nolru(bo, true, val_seq);
 			if (unlikely(ret != 0))
 				return ret;
+			spin_lock(&glob->lru_lock);
+			entry->reserved = true;
+			if (unlikely(atomic_read(&bo->cpu_writers) > 0)) {
+				ret = -EBUSY;
+				goto err;
+			}
 			goto retry;
 		default:
-			ttm_eu_backoff_reservation_locked(list);
-			spin_unlock(&glob->lru_lock);
-			ttm_eu_list_ref_sub(list);
-			return ret;
+			goto err;
 		}
 
 		entry->reserved = true;
 		if (unlikely(atomic_read(&bo->cpu_writers) > 0)) {
-			ttm_eu_backoff_reservation_locked(list);
-			spin_unlock(&glob->lru_lock);
-			ttm_eu_list_ref_sub(list);
-			return -EBUSY;
+			ret = -EBUSY;
+			goto err;
 		}
 	}
 
@@ -194,6 +196,12 @@ retry_this_bo:
 	ttm_eu_list_ref_sub(list);
 
 	return 0;
+
+err:
+	ttm_eu_backoff_reservation_locked(list);
+	spin_unlock(&glob->lru_lock);
+	ttm_eu_list_ref_sub(list);
+	return ret;
 }
 EXPORT_SYMBOL(ttm_eu_reserve_buffers);