drm/tegra: dc: Support memory bandwidth management

Display controller (DC) performs isochronous memory transfers, and thus,
has a requirement for a minimum memory bandwidth that shall be fulfilled,
otherwise framebuffer data can't be fetched fast enough and this results
in a DC's data-FIFO underflow that follows by a visual corruption.

The Memory Controller drivers provide facility for memory bandwidth
management via interconnect API. Let's wire up the interconnect API
support to the DC driver in order to fix the distorted display output
on T30 Ouya, T124 TK1 and other Tegra devices.

Tested-by: Peter Geis <pgwipeout@gmail.com> # Ouya T30
Tested-by: Matt Merhar <mattmerhar@protonmail.com> # Ouya T30
Tested-by: Nicolas Chauvet <kwizart@gmail.com> # PAZ00 T20 and TK1 T124
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
[treding@nvidia.com: unbreak Tegra186+ display support]
Signed-off-by: Thierry Reding <treding@nvidia.com>

1 files changed, 346 insertions, 2 deletions

diff --git a/drivers/gpu/drm/tegra/dc.c b/drivers/gpu/drm/tegra/dc.c
index 51bbbc42a144..865191216efc 100644
--- a/drivers/gpu/drm/tegra/dc.c
+++ b/drivers/gpu/drm/tegra/dc.c
@@ -8,6 +8,7 @@
 #include <linux/debugfs.h>
 #include <linux/delay.h>
 #include <linux/iommu.h>
+#include <linux/interconnect.h>
 #include <linux/module.h>
 #include <linux/of_device.h>
 #include <linux/pm_runtime.h>
@@ -618,9 +619,14 @@ static int tegra_plane_atomic_check(struct drm_plane *plane,
 	struct tegra_dc *dc = to_tegra_dc(new_plane_state->crtc);
 	int err;
 
+	plane_state->peak_memory_bandwidth = 0;
+	plane_state->avg_memory_bandwidth = 0;
+
 	/* no need for further checks if the plane is being disabled */
-	if (!new_plane_state->crtc)
+	if (!new_plane_state->crtc) {
+		plane_state->total_peak_memory_bandwidth = 0;
 		return 0;
+	}
 
 	err = tegra_plane_format(new_plane_state->fb->format->format,
 				 &plane_state->format,
@@ -808,6 +814,12 @@ static struct drm_plane *tegra_primary_plane_create(struct drm_device *drm,
 	formats = dc->soc->primary_formats;
 	modifiers = dc->soc->modifiers;
 
+	err = tegra_plane_interconnect_init(plane);
+	if (err) {
+		kfree(plane);
+		return ERR_PTR(err);
+	}
+
 	err = drm_universal_plane_init(drm, &plane->base, possible_crtcs,
 				       &tegra_plane_funcs, formats,
 				       num_formats, modifiers, type, NULL);
@@ -845,12 +857,18 @@ static int tegra_cursor_atomic_check(struct drm_plane *plane,
 {
 	struct drm_plane_state *new_plane_state = drm_atomic_get_new_plane_state(state,
 										 plane);
+	struct tegra_plane_state *plane_state = to_tegra_plane_state(new_plane_state);
 	struct tegra_plane *tegra = to_tegra_plane(plane);
 	int err;
 
+	plane_state->peak_memory_bandwidth = 0;
+	plane_state->avg_memory_bandwidth = 0;
+
 	/* no need for further checks if the plane is being disabled */
-	if (!new_plane_state->crtc)
+	if (!new_plane_state->crtc) {
+		plane_state->total_peak_memory_bandwidth = 0;
 		return 0;
+	}
 
 	/* scaling not supported for cursor */
 	if ((new_plane_state->src_w >> 16 != new_plane_state->crtc_w) ||
@@ -1030,6 +1048,12 @@ static struct drm_plane *tegra_dc_cursor_plane_create(struct drm_device *drm,
 	if (!dc->soc->has_nvdisplay) {
 		num_formats = ARRAY_SIZE(tegra_legacy_cursor_plane_formats);
 		formats = tegra_legacy_cursor_plane_formats;
+
+		err = tegra_plane_interconnect_init(plane);
+		if (err) {
+			kfree(plane);
+			return ERR_PTR(err);
+		}
 	} else {
 		num_formats = ARRAY_SIZE(tegra_cursor_plane_formats);
 		formats = tegra_cursor_plane_formats;
@@ -1149,6 +1173,12 @@ static struct drm_plane *tegra_dc_overlay_plane_create(struct drm_device *drm,
 	num_formats = dc->soc->num_overlay_formats;
 	formats = dc->soc->overlay_formats;
 
+	err = tegra_plane_interconnect_init(plane);
+	if (err) {
+		kfree(plane);
+		return ERR_PTR(err);
+	}
+
 	if (!cursor)
 		type = DRM_PLANE_TYPE_OVERLAY;
 	else
@@ -1804,6 +1834,106 @@ static int tegra_dc_wait_idle(struct tegra_dc *dc, unsigned long timeout)
 	return -ETIMEDOUT;
 }
 
+static void
+tegra_crtc_update_memory_bandwidth(struct drm_crtc *crtc,
+				   struct drm_atomic_state *state,
+				   bool prepare_bandwidth_transition)
+{
+	const struct tegra_plane_state *old_tegra_state, *new_tegra_state;
+	const struct tegra_dc_state *old_dc_state, *new_dc_state;
+	u32 i, new_avg_bw, old_avg_bw, new_peak_bw, old_peak_bw;
+	const struct drm_plane_state *old_plane_state;
+	const struct drm_crtc_state *old_crtc_state;
+	struct tegra_dc_window window, old_window;
+	struct tegra_dc *dc = to_tegra_dc(crtc);
+	struct tegra_plane *tegra;
+	struct drm_plane *plane;
+
+	if (dc->soc->has_nvdisplay)
+		return;
+
+	old_crtc_state = drm_atomic_get_old_crtc_state(state, crtc);
+	old_dc_state = to_const_dc_state(old_crtc_state);
+	new_dc_state = to_const_dc_state(crtc->state);
+
+	if (!crtc->state->active) {
+		if (!old_crtc_state->active)
+			return;
+
+		/*
+		 * When CRTC is disabled on DPMS, the state of attached planes
+		 * is kept unchanged. Hence we need to enforce removal of the
+		 * bandwidths from the ICC paths.
+		 */
+		drm_atomic_crtc_for_each_plane(plane, crtc) {
+			tegra = to_tegra_plane(plane);
+
+			icc_set_bw(tegra->icc_mem, 0, 0);
+			icc_set_bw(tegra->icc_mem_vfilter, 0, 0);
+		}
+
+		return;
+	}
+
+	for_each_old_plane_in_state(old_crtc_state->state, plane,
+				    old_plane_state, i) {
+		old_tegra_state = to_const_tegra_plane_state(old_plane_state);
+		new_tegra_state = to_const_tegra_plane_state(plane->state);
+		tegra = to_tegra_plane(plane);
+
+		/*
+		 * We're iterating over the global atomic state and it contains
+		 * planes from another CRTC, hence we need to filter out the
+		 * planes unrelated to this CRTC.
+		 */
+		if (tegra->dc != dc)
+			continue;
+
+		new_avg_bw = new_tegra_state->avg_memory_bandwidth;
+		old_avg_bw = old_tegra_state->avg_memory_bandwidth;
+
+		new_peak_bw = new_tegra_state->total_peak_memory_bandwidth;
+		old_peak_bw = old_tegra_state->total_peak_memory_bandwidth;
+
+		/*
+		 * See the comment related to !crtc->state->active above,
+		 * which explains why bandwidths need to be updated when
+		 * CRTC is turning ON.
+		 */
+		if (new_avg_bw == old_avg_bw && new_peak_bw == old_peak_bw &&
+		    old_crtc_state->active)
+			continue;
+
+		window.src.h = drm_rect_height(&plane->state->src) >> 16;
+		window.dst.h = drm_rect_height(&plane->state->dst);
+
+		old_window.src.h = drm_rect_height(&old_plane_state->src) >> 16;
+		old_window.dst.h = drm_rect_height(&old_plane_state->dst);
+
+		/*
+		 * During the preparation phase (atomic_begin), the memory
+		 * freq should go high before the DC changes are committed
+		 * if bandwidth requirement goes up, otherwise memory freq
+		 * should to stay high if BW requirement goes down.  The
+		 * opposite applies to the completion phase (post_commit).
+		 */
+		if (prepare_bandwidth_transition) {
+			new_avg_bw = max(old_avg_bw, new_avg_bw);
+			new_peak_bw = max(old_peak_bw, new_peak_bw);
+
+			if (tegra_plane_use_vertical_filtering(tegra, &old_window))
+				window = old_window;
+		}
+
+		icc_set_bw(tegra->icc_mem, new_avg_bw, new_peak_bw);
+
+		if (tegra_plane_use_vertical_filtering(tegra, &window))
+			icc_set_bw(tegra->icc_mem_vfilter, new_avg_bw, new_peak_bw);
+		else
+			icc_set_bw(tegra->icc_mem_vfilter, 0, 0);
+	}
+}
+
 static void tegra_crtc_atomic_disable(struct drm_crtc *crtc,
 				      struct drm_atomic_state *state)
 {
@@ -1985,6 +2115,8 @@ static void tegra_crtc_atomic_begin(struct drm_crtc *crtc,
 {
 	unsigned long flags;
 
+	tegra_crtc_update_memory_bandwidth(crtc, state, true);
+
 	if (crtc->state->event) {
 		spin_lock_irqsave(&crtc->dev->event_lock, flags);
 
@@ -2017,7 +2149,207 @@ static void tegra_crtc_atomic_flush(struct drm_crtc *crtc,
 	value = tegra_dc_readl(dc, DC_CMD_STATE_CONTROL);
 }
 
+static bool tegra_plane_is_cursor(const struct drm_plane_state *state)
+{
+	const struct tegra_dc_soc_info *soc = to_tegra_dc(state->crtc)->soc;
+	const struct drm_format_info *fmt = state->fb->format;
+	unsigned int src_w = drm_rect_width(&state->src) >> 16;
+	unsigned int dst_w = drm_rect_width(&state->dst);
+
+	if (state->plane->type != DRM_PLANE_TYPE_CURSOR)
+		return false;
+
+	if (soc->supports_cursor)
+		return true;
+
+	if (src_w != dst_w || fmt->num_planes != 1 || src_w * fmt->cpp[0] > 256)
+		return false;
+
+	return true;
+}
+
+static unsigned long
+tegra_plane_overlap_mask(struct drm_crtc_state *state,
+			 const struct drm_plane_state *plane_state)
+{
+	const struct drm_plane_state *other_state;
+	const struct tegra_plane *tegra;
+	unsigned long overlap_mask = 0;
+	struct drm_plane *plane;
+	struct drm_rect rect;
+
+	if (!plane_state->visible || !plane_state->fb)
+		return 0;
+
+	/*
+	 * Data-prefetch FIFO will easily help to overcome temporal memory
+	 * pressure if other plane overlaps with the cursor plane.
+	 */
+	if (tegra_plane_is_cursor(plane_state))
+		return 0;
+
+	drm_atomic_crtc_state_for_each_plane_state(plane, other_state, state) {
+		rect = plane_state->dst;
+
+		tegra = to_tegra_plane(other_state->plane);
+
+		if (!other_state->visible || !other_state->fb)
+			continue;
+
+		/*
+		 * Ignore cursor plane overlaps because it's not practical to
+		 * assume that it contributes to the bandwidth in overlapping
+		 * area if window width is small.
+		 */
+		if (tegra_plane_is_cursor(other_state))
+			continue;
+
+		if (drm_rect_intersect(&rect, &other_state->dst))
+			overlap_mask |= BIT(tegra->index);
+	}
+
+	return overlap_mask;
+}
+
+static int tegra_crtc_calculate_memory_bandwidth(struct drm_crtc *crtc,
+						 struct drm_atomic_state *state)
+{
+	ulong overlap_mask[TEGRA_DC_LEGACY_PLANES_NUM] = {}, mask;
+	u32 plane_peak_bw[TEGRA_DC_LEGACY_PLANES_NUM] = {};
+	bool all_planes_overlap_simultaneously = true;
+	const struct tegra_plane_state *tegra_state;
+	const struct drm_plane_state *plane_state;
+	struct tegra_dc *dc = to_tegra_dc(crtc);
+	const struct drm_crtc_state *old_state;
+	struct drm_crtc_state *new_state;
+	struct tegra_plane *tegra;
+	struct drm_plane *plane;
+
+	/*
+	 * The nv-display uses shared planes.  The algorithm below assumes
+	 * maximum 3 planes per-CRTC, this assumption isn't applicable to
+	 * the nv-display.  Note that T124 support has additional windows,
+	 * but currently they aren't supported by the driver.
+	 */
+	if (dc->soc->has_nvdisplay)
+		return 0;
+
+	new_state = drm_atomic_get_new_crtc_state(state, crtc);
+	old_state = drm_atomic_get_old_crtc_state(state, crtc);
+
+	/*
+	 * For overlapping planes pixel's data is fetched for each plane at
+	 * the same time, hence bandwidths are accumulated in this case.
+	 * This needs to be taken into account for calculating total bandwidth
+	 * consumed by all planes.
+	 *
+	 * Here we get the overlapping state of each plane, which is a
+	 * bitmask of plane indices telling with what planes there is an
+	 * overlap. Note that bitmask[plane] includes BIT(plane) in order
+	 * to make further code nicer and simpler.
+	 */
+	drm_atomic_crtc_state_for_each_plane_state(plane, plane_state, new_state) {
+		tegra_state = to_const_tegra_plane_state(plane_state);
+		tegra = to_tegra_plane(plane);
+
+		if (WARN_ON_ONCE(tegra->index >= TEGRA_DC_LEGACY_PLANES_NUM))
+			return -EINVAL;
+
+		plane_peak_bw[tegra->index] = tegra_state->peak_memory_bandwidth;
+		mask = tegra_plane_overlap_mask(new_state, plane_state);
+		overlap_mask[tegra->index] = mask;
+
+		if (hweight_long(mask) != 3)
+			all_planes_overlap_simultaneously = false;
+	}
+
+	/*
+	 * Then we calculate maximum bandwidth of each plane state.
+	 * The bandwidth includes the plane BW + BW of the "simultaneously"
+	 * overlapping planes, where "simultaneously" means areas where DC
+	 * fetches from the planes simultaneously during of scan-out process.
+	 *
+	 * For example, if plane A overlaps with planes B and C, but B and C
+	 * don't overlap, then the peak bandwidth will be either in area where
+	 * A-and-B or A-and-C planes overlap.
+	 *
+	 * The plane_peak_bw[] contains peak memory bandwidth values of
+	 * each plane, this information is needed by interconnect provider
+	 * in order to set up latency allowance based on the peak BW, see
+	 * tegra_crtc_update_memory_bandwidth().
+	 */
+	drm_atomic_crtc_state_for_each_plane_state(plane, plane_state, new_state) {
+		u32 i, old_peak_bw, new_peak_bw, overlap_bw = 0;
+
+		/*
+		 * Note that plane's atomic check doesn't touch the
+		 * total_peak_memory_bandwidth of enabled plane, hence the
+		 * current state contains the old bandwidth state from the
+		 * previous CRTC commit.
+		 */
+		tegra_state = to_const_tegra_plane_state(plane_state);
+		tegra = to_tegra_plane(plane);
+
+		for_each_set_bit(i, &overlap_mask[tegra->index], 3) {
+			if (i == tegra->index)
+				continue;
+
+			if (all_planes_overlap_simultaneously)
+				overlap_bw += plane_peak_bw[i];
+			else
+				overlap_bw = max(overlap_bw, plane_peak_bw[i]);
+		}
+
+		new_peak_bw = plane_peak_bw[tegra->index] + overlap_bw;
+		old_peak_bw = tegra_state->total_peak_memory_bandwidth;
+
+		/*
+		 * If plane's peak bandwidth changed (for example plane isn't
+		 * overlapped anymore) and plane isn't in the atomic state,
+		 * then add plane to the state in order to have the bandwidth
+		 * updated.
+		 */
+		if (old_peak_bw != new_peak_bw) {
+			struct tegra_plane_state *new_tegra_state;
+			struct drm_plane_state *new_plane_state;
+
+			new_plane_state = drm_atomic_get_plane_state(state, plane);
+			if (IS_ERR(new_plane_state))
+				return PTR_ERR(new_plane_state);
+
+			new_tegra_state = to_tegra_plane_state(new_plane_state);
+			new_tegra_state->total_peak_memory_bandwidth = new_peak_bw;
+		}
+	}
+
+	return 0;
+}
+
+static int tegra_crtc_atomic_check(struct drm_crtc *crtc,
+				   struct drm_atomic_state *state)
+{
+	int err;
+
+	err = tegra_crtc_calculate_memory_bandwidth(crtc, state);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+void tegra_crtc_atomic_post_commit(struct drm_crtc *crtc,
+				   struct drm_atomic_state *state)
+{
+	/*
+	 * Display bandwidth is allowed to go down only once hardware state
+	 * is known to be armed, i.e. state was committed and VBLANK event
+	 * received.
+	 */
+	tegra_crtc_update_memory_bandwidth(crtc, state, false);
+}
+
 static const struct drm_crtc_helper_funcs tegra_crtc_helper_funcs = {
+	.atomic_check = tegra_crtc_atomic_check,
 	.atomic_begin = tegra_crtc_atomic_begin,
 	.atomic_flush = tegra_crtc_atomic_flush,
 	.atomic_enable = tegra_crtc_atomic_enable,
@@ -2343,7 +2675,9 @@ static const struct tegra_dc_soc_info tegra20_dc_soc_info = {
 	.overlay_formats = tegra20_overlay_formats,
 	.modifiers = tegra20_modifiers,
 	.has_win_a_without_filters = true,
+	.has_win_b_vfilter_mem_client = true,
 	.has_win_c_without_vert_filter = true,
+	.plane_tiled_memory_bandwidth_x2 = false,
 };
 
 static const struct tegra_dc_soc_info tegra30_dc_soc_info = {
@@ -2363,7 +2697,9 @@ static const struct tegra_dc_soc_info tegra30_dc_soc_info = {
 	.overlay_formats = tegra20_overlay_formats,
 	.modifiers = tegra20_modifiers,
 	.has_win_a_without_filters = false,
+	.has_win_b_vfilter_mem_client = true,
 	.has_win_c_without_vert_filter = false,
+	.plane_tiled_memory_bandwidth_x2 = true,
 };
 
 static const struct tegra_dc_soc_info tegra114_dc_soc_info = {
@@ -2383,7 +2719,9 @@ static const struct tegra_dc_soc_info tegra114_dc_soc_info = {
 	.overlay_formats = tegra114_overlay_formats,
 	.modifiers = tegra20_modifiers,
 	.has_win_a_without_filters = false,
+	.has_win_b_vfilter_mem_client = false,
 	.has_win_c_without_vert_filter = false,
+	.plane_tiled_memory_bandwidth_x2 = true,
 };
 
 static const struct tegra_dc_soc_info tegra124_dc_soc_info = {
@@ -2403,7 +2741,9 @@ static const struct tegra_dc_soc_info tegra124_dc_soc_info = {
 	.overlay_formats = tegra124_overlay_formats,
 	.modifiers = tegra124_modifiers,
 	.has_win_a_without_filters = false,
+	.has_win_b_vfilter_mem_client = false,
 	.has_win_c_without_vert_filter = false,
+	.plane_tiled_memory_bandwidth_x2 = false,
 };
 
 static const struct tegra_dc_soc_info tegra210_dc_soc_info = {
@@ -2423,7 +2763,9 @@ static const struct tegra_dc_soc_info tegra210_dc_soc_info = {
 	.overlay_formats = tegra114_overlay_formats,
 	.modifiers = tegra124_modifiers,
 	.has_win_a_without_filters = false,
+	.has_win_b_vfilter_mem_client = false,
 	.has_win_c_without_vert_filter = false,
+	.plane_tiled_memory_bandwidth_x2 = false,
 };
 
 static const struct tegra_windowgroup_soc tegra186_dc_wgrps[] = {
@@ -2473,6 +2815,7 @@ static const struct tegra_dc_soc_info tegra186_dc_soc_info = {
 	.has_nvdisplay = true,
 	.wgrps = tegra186_dc_wgrps,
 	.num_wgrps = ARRAY_SIZE(tegra186_dc_wgrps),
+	.plane_tiled_memory_bandwidth_x2 = false,
 };
 
 static const struct tegra_windowgroup_soc tegra194_dc_wgrps[] = {
@@ -2522,6 +2865,7 @@ static const struct tegra_dc_soc_info tegra194_dc_soc_info = {
 	.has_nvdisplay = true,
 	.wgrps = tegra194_dc_wgrps,
 	.num_wgrps = ARRAY_SIZE(tegra194_dc_wgrps),
+	.plane_tiled_memory_bandwidth_x2 = false,
 };
 
 static const struct of_device_id tegra_dc_of_match[] = {