drm/i915/guc: Move GuC wq_check_space to alloc_request_extras

Split GuC work queue space checking from submission and move it to
ring_alloc_request_extras. The reason is that failure in later
i915_add_request() won't be handled. In the case timeout happens,
driver can return early in order to handle the error.

v1: Move wq_reserve_space to ring_reserve_space
v2: Move wq_reserve_space to alloc_request_extras (Chris Wilson)
v3: The work queue head pointer is cached by driver now. So we can
    quickly return if space is available.
    s/reserve/check/g (Dave Gordon)
v4: Update cached wq head after ring doorbell; check wq space before
    ring doorbell in case unexpected error happens; call wq space
    check only when GuC submission is enabled. (Dave Gordon)

Signed-off-by: Alex Dai <yu.dai@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450295155-10050-1-git-send-email-yu.dai@intel.com
Reviewed-by: Dave Gordon <david.s.gordon@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c b/drivers/gpu/drm/i915/i915_guc_submission.c
index 9cc3b84..8b00f461 100644
--- a/drivers/gpu/drm/i915/i915_guc_submission.c
+++ b/drivers/gpu/drm/i915/i915_guc_submission.c
@@ -244,6 +244,9 @@
 			db_exc.cookie = 1;
 	}
 
+	/* Finally, update the cached copy of the GuC's WQ head */
+	gc->wq_head = desc->head;
+
 	kunmap_atomic(base);
 	return ret;
 }
@@ -469,28 +472,30 @@
 			     sizeof(desc) * client->ctx_index);
 }
 
-/* Get valid workqueue item and return it back to offset */
-static int guc_get_workqueue_space(struct i915_guc_client *gc, u32 *offset)
+int i915_guc_wq_check_space(struct i915_guc_client *gc)
 {
 	struct guc_process_desc *desc;
 	void *base;
 	u32 size = sizeof(struct guc_wq_item);
 	int ret = -ETIMEDOUT, timeout_counter = 200;
 
+	if (!gc)
+		return 0;
+
+	/* Quickly return if wq space is available since last time we cache the
+	 * head position. */
+	if (CIRC_SPACE(gc->wq_tail, gc->wq_head, gc->wq_size) >= size)
+		return 0;
+
 	base = kmap_atomic(i915_gem_object_get_page(gc->client_obj, 0));
 	desc = base + gc->proc_desc_offset;
 
 	while (timeout_counter-- > 0) {
-		if (CIRC_SPACE(gc->wq_tail, desc->head, gc->wq_size) >= size) {
-			*offset = gc->wq_tail;
+		gc->wq_head = desc->head;
 
-			/* advance the tail for next workqueue item */
-			gc->wq_tail += size;
-			gc->wq_tail &= gc->wq_size - 1;
-
-			/* this will break the loop */
-			timeout_counter = 0;
+		if (CIRC_SPACE(gc->wq_tail, gc->wq_head, gc->wq_size) >= size) {
 			ret = 0;
+			break;
 		}
 
 		if (timeout_counter)
@@ -508,12 +513,16 @@
 	enum intel_ring_id ring_id = rq->ring->id;
 	struct guc_wq_item *wqi;
 	void *base;
-	u32 tail, wq_len, wq_off = 0;
-	int ret;
+	u32 tail, wq_len, wq_off, space;
 
-	ret = guc_get_workqueue_space(gc, &wq_off);
-	if (ret)
-		return ret;
+	space = CIRC_SPACE(gc->wq_tail, gc->wq_head, gc->wq_size);
+	if (WARN_ON(space < sizeof(struct guc_wq_item)))
+		return -ENOSPC; /* shouldn't happen */
+
+	/* postincrement WQ tail for next time */
+	wq_off = gc->wq_tail;
+	gc->wq_tail += sizeof(struct guc_wq_item);
+	gc->wq_tail &= gc->wq_size - 1;
 
 	/* For now workqueue item is 4 DWs; workqueue buffer is 2 pages. So we
 	 * should not have the case where structure wqi is across page, neither
diff --git a/drivers/gpu/drm/i915/intel_guc.h b/drivers/gpu/drm/i915/intel_guc.h
index 8229522..e8612c9 100644
--- a/drivers/gpu/drm/i915/intel_guc.h
+++ b/drivers/gpu/drm/i915/intel_guc.h
@@ -43,6 +43,7 @@
 	uint32_t wq_offset;
 	uint32_t wq_size;
 	uint32_t wq_tail;
+	uint32_t wq_head;
 
 	/* GuC submission statistics & status */
 	uint64_t submissions[I915_NUM_RINGS];
@@ -122,5 +123,6 @@
 		    struct drm_i915_gem_request *rq);
 void i915_guc_submission_disable(struct drm_device *dev);
 void i915_guc_submission_fini(struct drm_device *dev);
+int i915_guc_wq_check_space(struct i915_guc_client *client);
 
 #endif
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 973487a..e095058 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -670,6 +670,19 @@
 			return ret;
 	}
 
+	if (i915.enable_guc_submission) {
+		/*
+		 * Check that the GuC has space for the request before
+		 * going any further, as the i915_add_request() call
+		 * later on mustn't fail ...
+		 */
+		struct intel_guc *guc = &request->i915->guc;
+
+		ret = i915_guc_wq_check_space(guc->execbuf_client);
+		if (ret)
+			return ret;
+	}
+
 	return 0;
 }