drm/i915/gen9: Allow skl_allocate_pipe_ddb() to operate on in-flight state (v3)

We eventually want to calculate watermark values at atomic 'check' time
instead of atomic 'commit' time so that any requested configurations
that result in impossible watermark requirements are properly rejected.
The first step along this path is to allocate the DDB at atomic 'check'
time.  As we perform this transition, allow the main allocation function
to operate successfully on either an in-flight state or an
already-commited state.  Once we complete the transition in a future
patch, we'll come back and remove the unnecessary logic for the
already-committed case.

v2: Rebase/refactor; we should no longer need to grab extra plane states
    while allocating the DDB since we can pull cached data rates and
    minimum block counts from the CRTC state for any planes that aren't
    being modified by this transaction.

v3:
 - Simplify memsets to clear DDB plane entries.  (Maarten)
 - Drop a redundant memset of plane[pipe][PLANE_CURSOR] that was added
   by an earlier Coccinelle patch.  (Maarten)
 - Assign *num_active at the top of skl_ddb_get_pipe_allocation_limits()
   so that no code paths return without setting it.  (kbuild robot)

Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1463061971-19638-8-git-send-email-matthew.d.roper@intel.com

drivers/gpu/drm/i915/i915_drv.h

@@ -324,6 +324,12 @@ struct i915_hotplug {
 			    &dev->mode_config.plane_list,	\
 			    base.head)
 
+#define for_each_intel_plane_mask(dev, intel_plane, plane_mask)		\
+	list_for_each_entry(intel_plane, &dev->mode_config.plane_list,	\
+			    base.head)					\
+		for_each_if ((plane_mask) &				\
+			     (1 << drm_plane_index(&intel_plane->base)))
+
 #define for_each_intel_plane_on_crtc(dev, intel_crtc, intel_plane)	\
 	list_for_each_entry(intel_plane,				\
 			    &(dev)->mode_config.plane_list,		\

drivers/gpu/drm/i915/intel_pm.c

@@ -2849,13 +2849,25 @@ skl_wm_plane_id(const struct intel_plane *plane)
 static void
 skl_ddb_get_pipe_allocation_limits(struct drm_device *dev,
 				   const struct intel_crtc_state *cstate,
-				   const struct intel_wm_config *config,
-				   struct skl_ddb_entry *alloc /* out */)
+				   struct intel_wm_config *config,
+				   struct skl_ddb_entry *alloc, /* out */
+				   int *num_active /* out */)
 {
+	struct drm_atomic_state *state = cstate->base.state;
+	struct intel_atomic_state *intel_state = to_intel_atomic_state(state);
+	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct drm_crtc *for_crtc = cstate->base.crtc;
 	struct drm_crtc *crtc;
 	unsigned int pipe_size, ddb_size;
 	int nth_active_pipe;
+	int pipe = to_intel_crtc(for_crtc)->pipe;
+
+	if (intel_state && intel_state->active_pipe_changes)
+		*num_active = hweight32(intel_state->active_crtcs);
+	else if (intel_state)
+		*num_active = hweight32(dev_priv->active_crtcs);
+	else
+		*num_active = config->num_pipes_active;
 
 	if (!cstate->base.active) {
 		alloc->start = 0;
@@ -2870,25 +2882,56 @@ skl_ddb_get_pipe_allocation_limits(struct drm_device *dev,
 
 	ddb_size -= 4; /* 4 blocks for bypass path allocation */
 
-	nth_active_pipe = 0;
-	for_each_crtc(dev, crtc) {
-		if (!to_intel_crtc(crtc)->active)
-			continue;
+	/*
+	 * FIXME: At the moment we may be called on either in-flight or fully
+	 * committed cstate's.  Once we fully move DDB allocation in the check
+	 * phase, we'll only be called on in-flight states and the 'else'
+	 * branch here will go away.
+	 *
+	 * The 'else' branch is slightly racy here, but it was racy to begin
+	 * with; since it's going away soon, no effort is made to address that.
+	 */
+	if (state) {
+		/*
+		 * If the state doesn't change the active CRTC's, then there's
+		 * no need to recalculate; the existing pipe allocation limits
+		 * should remain unchanged.  Note that we're safe from racing
+		 * commits since any racing commit that changes the active CRTC
+		 * list would need to grab _all_ crtc locks, including the one
+		 * we currently hold.
+		 */
+		if (!intel_state->active_pipe_changes) {
+			*alloc = dev_priv->wm.skl_hw.ddb.pipe[pipe];
+			return;
+		}
 
-		if (crtc == for_crtc)
-			break;
+		nth_active_pipe = hweight32(intel_state->active_crtcs &
+					    (drm_crtc_mask(for_crtc) - 1));
+		pipe_size = ddb_size / hweight32(intel_state->active_crtcs);
+		alloc->start = nth_active_pipe * ddb_size / *num_active;
+		alloc->end = alloc->start + pipe_size;
+	} else {
+		nth_active_pipe = 0;
+		for_each_crtc(dev, crtc) {
+			if (!to_intel_crtc(crtc)->active)
+				continue;
 
-		nth_active_pipe++;
-	}
+			if (crtc == for_crtc)
+				break;
 
-	pipe_size = ddb_size / config->num_pipes_active;
-	alloc->start = nth_active_pipe * ddb_size / config->num_pipes_active;
-	alloc->end = alloc->start + pipe_size;
+			nth_active_pipe++;
+		}
+
+		pipe_size = ddb_size / config->num_pipes_active;
+		alloc->start = nth_active_pipe * ddb_size /
+			config->num_pipes_active;
+		alloc->end = alloc->start + pipe_size;
+	}
 }
 
-static unsigned int skl_cursor_allocation(const struct intel_wm_config *config)
+static unsigned int skl_cursor_allocation(int num_active)
 {
-	if (config->num_pipes_active == 1)
+	if (num_active == 1)
 		return 32;
 
 	return 8;
@@ -3054,33 +3097,44 @@ skl_get_total_relative_data_rate(struct intel_crtc_state *intel_cstate)
 	return total_data_rate;
 }
 
-static void
+static int
 skl_allocate_pipe_ddb(struct intel_crtc_state *cstate,
 		      struct skl_ddb_allocation *ddb /* out */)
 {
+	struct drm_atomic_state *state = cstate->base.state;
 	struct drm_crtc *crtc = cstate->base.crtc;
 	struct drm_device *dev = crtc->dev;
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	struct intel_wm_config *config = &dev_priv->wm.config;
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
 	struct intel_plane *intel_plane;
+	struct drm_plane *plane;
+	struct drm_plane_state *pstate;
 	enum pipe pipe = intel_crtc->pipe;
 	struct skl_ddb_entry *alloc = &ddb->pipe[pipe];
 	uint16_t alloc_size, start, cursor_blocks;
 	uint16_t *minimum = cstate->wm.skl.minimum_blocks;
 	uint16_t *y_minimum = cstate->wm.skl.minimum_y_blocks;
 	unsigned int total_data_rate;
+	int num_active;
+	int id, i;
 
-	skl_ddb_get_pipe_allocation_limits(dev, cstate, config, alloc);
+	if (!cstate->base.active) {
+		ddb->pipe[pipe].start = ddb->pipe[pipe].end = 0;
+		memset(ddb->plane[pipe], 0, sizeof(ddb->plane[pipe]));
+		memset(ddb->y_plane[pipe], 0, sizeof(ddb->y_plane[pipe]));
+		return 0;
+	}
+
+	skl_ddb_get_pipe_allocation_limits(dev, cstate, config, alloc,
+					   &num_active);
 	alloc_size = skl_ddb_entry_size(alloc);
 	if (alloc_size == 0) {
 		memset(ddb->plane[pipe], 0, sizeof(ddb->plane[pipe]));
-		memset(&ddb->plane[pipe][PLANE_CURSOR], 0,
-		       sizeof(ddb->plane[pipe][PLANE_CURSOR]));
-		return;
+		return 0;
 	}
 
-	cursor_blocks = skl_cursor_allocation(config);
+	cursor_blocks = skl_cursor_allocation(num_active);
 	ddb->plane[pipe][PLANE_CURSOR].start = alloc->end - cursor_blocks;
 	ddb->plane[pipe][PLANE_CURSOR].end = alloc->end;
 
@@ -3088,21 +3142,55 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *cstate,
 	alloc->end -= cursor_blocks;
 
 	/* 1. Allocate the mininum required blocks for each active plane */
-	for_each_intel_plane_on_crtc(dev, intel_crtc, intel_plane) {
-		struct drm_plane *plane = &intel_plane->base;
-		struct drm_framebuffer *fb = plane->state->fb;
-		int id = skl_wm_plane_id(intel_plane);
+	/*
+	 * TODO: Remove support for already-committed state once we
+	 * only allocate DDB on in-flight states.
+	 */
+	if (state) {
+		for_each_plane_in_state(state, plane, pstate, i) {
+			intel_plane = to_intel_plane(plane);
+			id = skl_wm_plane_id(intel_plane);
 
-		if (!to_intel_plane_state(plane->state)->visible)
-			continue;
+			if (intel_plane->pipe != pipe)
+				continue;
 
-		if (plane->type == DRM_PLANE_TYPE_CURSOR)
-			continue;
+			if (!to_intel_plane_state(pstate)->visible) {
+				minimum[id] = 0;
+				y_minimum[id] = 0;
+				continue;
+			}
+			if (plane->type == DRM_PLANE_TYPE_CURSOR) {
+				minimum[id] = 0;
+				y_minimum[id] = 0;
+				continue;
+			}
+
+			minimum[id] = 8;
+			if (pstate->fb->pixel_format == DRM_FORMAT_NV12)
+				y_minimum[id] = 8;
+			else
+				y_minimum[id] = 0;
+		}
+	} else {
+		for_each_intel_plane_on_crtc(dev, intel_crtc, intel_plane) {
+			struct drm_plane *plane = &intel_plane->base;
+			struct drm_framebuffer *fb = plane->state->fb;
+			int id = skl_wm_plane_id(intel_plane);
+
+			if (!to_intel_plane_state(plane->state)->visible)
+				continue;
+
+			if (plane->type == DRM_PLANE_TYPE_CURSOR)
+				continue;
+
+			minimum[id] = 8;
+			y_minimum[id] = (fb->pixel_format == DRM_FORMAT_NV12) ? 8 : 0;
+		}
+	}
 
-		minimum[id] = 8;
-		alloc_size -= minimum[id];
-		y_minimum[id] = (fb->pixel_format == DRM_FORMAT_NV12) ? 8 : 0;
-		alloc_size -= y_minimum[id];
+	for (i = 0; i < PLANE_CURSOR; i++) {
+		alloc_size -= minimum[i];
+		alloc_size -= y_minimum[i];
 	}
 
 	/*
@@ -3113,21 +3201,14 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *cstate,
 	 */
 	total_data_rate = skl_get_total_relative_data_rate(cstate);
 	if (total_data_rate == 0)
-		return;
+		return 0;
 
 	start = alloc->start;
 	for_each_intel_plane_on_crtc(dev, intel_crtc, intel_plane) {
-		struct drm_plane *plane = &intel_plane->base;
-		struct drm_plane_state *pstate = intel_plane->base.state;
 		unsigned int data_rate, y_data_rate;
 		uint16_t plane_blocks, y_plane_blocks = 0;
 		int id = skl_wm_plane_id(intel_plane);
 
-		if (!to_intel_plane_state(pstate)->visible)
-			continue;
-		if (plane->type == DRM_PLANE_TYPE_CURSOR)
-			continue;
-
 		data_rate = cstate->wm.skl.plane_data_rate[id];
 
 		/*
@@ -3139,8 +3220,11 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *cstate,
 		plane_blocks += div_u64((uint64_t)alloc_size * data_rate,
 					total_data_rate);
 
-		ddb->plane[pipe][id].start = start;
-		ddb->plane[pipe][id].end = start + plane_blocks;
+		/* Leave disabled planes at (0,0) */
+		if (data_rate) {
+			ddb->plane[pipe][id].start = start;
+			ddb->plane[pipe][id].end = start + plane_blocks;
+		}
 
 		start += plane_blocks;
 
@@ -3153,12 +3237,15 @@ skl_allocate_pipe_ddb(struct intel_crtc_state *cstate,
 		y_plane_blocks += div_u64((uint64_t)alloc_size * y_data_rate,
 					total_data_rate);
 
-		ddb->y_plane[pipe][id].start = start;
-		ddb->y_plane[pipe][id].end = start + y_plane_blocks;
+		if (y_data_rate) {
+			ddb->y_plane[pipe][id].start = start;
+			ddb->y_plane[pipe][id].end = start + y_plane_blocks;
+		}
 
 		start += y_plane_blocks;
 	}
 
+	return 0;
 }
 
 static uint32_t skl_pipe_pixel_rate(const struct intel_crtc_state *config)
@@ -3649,7 +3736,7 @@ static bool skl_update_pipe_wm(struct drm_crtc *crtc,
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
 	struct intel_crtc_state *cstate = to_intel_crtc_state(crtc->state);
 
-	skl_allocate_pipe_ddb(cstate, ddb);
+	WARN_ON(skl_allocate_pipe_ddb(cstate, ddb) != 0);
 	skl_build_pipe_wm(cstate, ddb, pipe_wm);
 
 	if (!memcmp(&intel_crtc->wm.active.skl, pipe_wm, sizeof(*pipe_wm)))