Merge tag 'drm-vc4-next-2016-10-06' of https://github.com/anholt/linux into drm-next

This pull request brings in several fixes for drm-next, mostly for
HDMI.

* tag 'drm-vc4-next-2016-10-06' of https://github.com/anholt/linux:
  drm/vc4: Add support for double-clocked modes.
  drm/vc4: Set up the AVI and SPD infoframes.
  drm/vc4: Fix support for interlaced modes on HDMI.
  drm/vc4: Increase timeout for HDMI_SCHEDULER_CONTROL changes.
  drm/vc4: Fall back to using an EDID probe in the absence of a GPIO.
  drm/vc4: Enable limited range RGB output on HDMI with CEA modes.
  drm/vc4: Fix races when the CS reads from render targets.
  drm/vc4: cleanup with list_first_entry_or_null()

7 files changed, 306 insertions, 89 deletions

diff --git a/drivers/gpu/drm/vc4/vc4_crtc.c b/drivers/gpu/drm/vc4/vc4_crtc.c
index 2682f07d8f1e..7f08d681a74b 100644
--- a/drivers/gpu/drm/vc4/vc4_crtc.c
+++ b/drivers/gpu/drm/vc4/vc4_crtc.c
@@ -229,7 +229,7 @@ int vc4_crtc_get_scanoutpos(struct drm_device *dev, unsigned int crtc_id,
 	 * and need to make things up in a approximative but consistent way.
 	 */
 	ret |= DRM_SCANOUTPOS_IN_VBLANK;
-	vblank_lines = mode->crtc_vtotal - mode->crtc_vdisplay;
+	vblank_lines = mode->vtotal - mode->vdisplay;
 
 	if (flags & DRM_CALLED_FROM_VBLIRQ) {
 		/*
@@ -378,7 +378,7 @@ static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc)
 	struct drm_crtc_state *state = crtc->state;
 	struct drm_display_mode *mode = &state->adjusted_mode;
 	bool interlace = mode->flags & DRM_MODE_FLAG_INTERLACE;
-	u32 vactive = (mode->vdisplay >> (interlace ? 1 : 0));
+	u32 pixel_rep = (mode->flags & DRM_MODE_FLAG_DBLCLK) ? 2 : 1;
 	u32 format = PV_CONTROL_FORMAT_24;
 	bool debug_dump_regs = false;
 	int clock_select = vc4_get_clock_select(crtc);
@@ -394,47 +394,65 @@ static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc)
 	CRTC_WRITE(PV_CONTROL, 0);
 
 	CRTC_WRITE(PV_HORZA,
-		   VC4_SET_FIELD(mode->htotal - mode->hsync_end,
+		   VC4_SET_FIELD((mode->htotal -
+				  mode->hsync_end) * pixel_rep,
 				 PV_HORZA_HBP) |
-		   VC4_SET_FIELD(mode->hsync_end - mode->hsync_start,
+		   VC4_SET_FIELD((mode->hsync_end -
+				  mode->hsync_start) * pixel_rep,
 				 PV_HORZA_HSYNC));
 	CRTC_WRITE(PV_HORZB,
-		   VC4_SET_FIELD(mode->hsync_start - mode->hdisplay,
+		   VC4_SET_FIELD((mode->hsync_start -
+				  mode->hdisplay) * pixel_rep,
 				 PV_HORZB_HFP) |
-		   VC4_SET_FIELD(mode->hdisplay, PV_HORZB_HACTIVE));
+		   VC4_SET_FIELD(mode->hdisplay * pixel_rep, PV_HORZB_HACTIVE));
 
 	CRTC_WRITE(PV_VERTA,
-		   VC4_SET_FIELD(mode->vtotal - mode->vsync_end,
+		   VC4_SET_FIELD(mode->crtc_vtotal - mode->crtc_vsync_end,
 				 PV_VERTA_VBP) |
-		   VC4_SET_FIELD(mode->vsync_end - mode->vsync_start,
+		   VC4_SET_FIELD(mode->crtc_vsync_end - mode->crtc_vsync_start,
 				 PV_VERTA_VSYNC));
 	CRTC_WRITE(PV_VERTB,
-		   VC4_SET_FIELD(mode->vsync_start - mode->vdisplay,
+		   VC4_SET_FIELD(mode->crtc_vsync_start - mode->crtc_vdisplay,
 				 PV_VERTB_VFP) |
-		   VC4_SET_FIELD(vactive, PV_VERTB_VACTIVE));
+		   VC4_SET_FIELD(mode->crtc_vdisplay, PV_VERTB_VACTIVE));
 
 	if (interlace) {
 		CRTC_WRITE(PV_VERTA_EVEN,
-			   VC4_SET_FIELD(mode->vtotal - mode->vsync_end - 1,
+			   VC4_SET_FIELD(mode->crtc_vtotal -
+					 mode->crtc_vsync_end - 1,
 					 PV_VERTA_VBP) |
-			   VC4_SET_FIELD(mode->vsync_end - mode->vsync_start,
+			   VC4_SET_FIELD(mode->crtc_vsync_end -
+					 mode->crtc_vsync_start,
 					 PV_VERTA_VSYNC));
 		CRTC_WRITE(PV_VERTB_EVEN,
-			   VC4_SET_FIELD(mode->vsync_start - mode->vdisplay,
+			   VC4_SET_FIELD(mode->crtc_vsync_start -
+					 mode->crtc_vdisplay,
 					 PV_VERTB_VFP) |
-			   VC4_SET_FIELD(vactive, PV_VERTB_VACTIVE));
+			   VC4_SET_FIELD(mode->crtc_vdisplay, PV_VERTB_VACTIVE));
+
+		/* We set up first field even mode for HDMI.  VEC's
+		 * NTSC mode would want first field odd instead, once
+		 * we support it (to do so, set ODD_FIRST and put the
+		 * delay in VSYNCD_EVEN instead).
+		 */
+		CRTC_WRITE(PV_V_CONTROL,
+			   PV_VCONTROL_CONTINUOUS |
+			   PV_VCONTROL_INTERLACE |
+			   VC4_SET_FIELD(mode->htotal * pixel_rep / 2,
+					 PV_VCONTROL_ODD_DELAY));
+		CRTC_WRITE(PV_VSYNCD_EVEN, 0);
+	} else {
+		CRTC_WRITE(PV_V_CONTROL, PV_VCONTROL_CONTINUOUS);
 	}
 
-	CRTC_WRITE(PV_HACT_ACT, mode->hdisplay);
+	CRTC_WRITE(PV_HACT_ACT, mode->hdisplay * pixel_rep);
 
-	CRTC_WRITE(PV_V_CONTROL,
-		   PV_VCONTROL_CONTINUOUS |
-		   (interlace ? PV_VCONTROL_INTERLACE : 0));
 
 	CRTC_WRITE(PV_CONTROL,
 		   VC4_SET_FIELD(format, PV_CONTROL_FORMAT) |
 		   VC4_SET_FIELD(vc4_get_fifo_full_level(format),
 				 PV_CONTROL_FIFO_LEVEL) |
+		   VC4_SET_FIELD(pixel_rep - 1, PV_CONTROL_PIXEL_REP) |
 		   PV_CONTROL_CLR_AT_START |
 		   PV_CONTROL_TRIGGER_UNDERFLOW |
 		   PV_CONTROL_WAIT_HSTART |
@@ -544,16 +562,6 @@ static bool vc4_crtc_mode_fixup(struct drm_crtc *crtc,
 		return false;
 	}
 
-	/*
-	 * Interlaced video modes got CRTC_INTERLACE_HALVE_V applied when
-	 * coming from user space. We don't want this, as it screws up
-	 * vblank timestamping, so fix it up.
-	 */
-	drm_mode_set_crtcinfo(adjusted_mode, 0);
-
-	DRM_DEBUG_KMS("[CRTC:%d] adjusted_mode :\n", crtc->base.id);
-	drm_mode_debug_printmodeline(adjusted_mode);
-
 	return true;
 }
 
diff --git a/drivers/gpu/drm/vc4/vc4_drv.h b/drivers/gpu/drm/vc4/vc4_drv.h
index 428e24919ef1..7c1e4d97486f 100644
--- a/drivers/gpu/drm/vc4/vc4_drv.h
+++ b/drivers/gpu/drm/vc4/vc4_drv.h
@@ -122,9 +122,16 @@ to_vc4_dev(struct drm_device *dev)
 struct vc4_bo {
 	struct drm_gem_cma_object base;
 
-	/* seqno of the last job to render to this BO. */
+	/* seqno of the last job to render using this BO. */
 	uint64_t seqno;
 
+	/* seqno of the last job to use the RCL to write to this BO.
+	 *
+	 * Note that this doesn't include binner overflow memory
+	 * writes.
+	 */
+	uint64_t write_seqno;
+
 	/* List entry for the BO's position in either
 	 * vc4_exec_info->unref_list or vc4_dev->bo_cache.time_list
 	 */
@@ -216,6 +223,9 @@ struct vc4_exec_info {
 	/* Sequence number for this bin/render job. */
 	uint64_t seqno;
 
+	/* Latest write_seqno of any BO that binning depends on. */
+	uint64_t bin_dep_seqno;
+
 	/* Last current addresses the hardware was processing when the
 	 * hangcheck timer checked on us.
 	 */
@@ -230,6 +240,13 @@ struct vc4_exec_info {
 	struct drm_gem_cma_object **bo;
 	uint32_t bo_count;
 
+	/* List of BOs that are being written by the RCL.  Other than
+	 * the binner temporary storage, this is all the BOs written
+	 * by the job.
+	 */
+	struct drm_gem_cma_object *rcl_write_bo[4];
+	uint32_t rcl_write_bo_count;
+
 	/* Pointers for our position in vc4->job_list */
 	struct list_head head;
 
@@ -307,18 +324,15 @@ struct vc4_exec_info {
 static inline struct vc4_exec_info *
 vc4_first_bin_job(struct vc4_dev *vc4)
 {
-	if (list_empty(&vc4->bin_job_list))
-		return NULL;
-	return list_first_entry(&vc4->bin_job_list, struct vc4_exec_info, head);
+	return list_first_entry_or_null(&vc4->bin_job_list,
+					struct vc4_exec_info, head);
 }
 
 static inline struct vc4_exec_info *
 vc4_first_render_job(struct vc4_dev *vc4)
 {
-	if (list_empty(&vc4->render_job_list))
-		return NULL;
-	return list_first_entry(&vc4->render_job_list,
-				struct vc4_exec_info, head);
+	return list_first_entry_or_null(&vc4->render_job_list,
+					struct vc4_exec_info, head);
 }
 
 static inline struct vc4_exec_info *
diff --git a/drivers/gpu/drm/vc4/vc4_gem.c b/drivers/gpu/drm/vc4/vc4_gem.c
index 77daea6cb866..47a095f392f8 100644
--- a/drivers/gpu/drm/vc4/vc4_gem.c
+++ b/drivers/gpu/drm/vc4/vc4_gem.c
@@ -467,6 +467,11 @@ vc4_update_bo_seqnos(struct vc4_exec_info *exec, uint64_t seqno)
 	list_for_each_entry(bo, &exec->unref_list, unref_head) {
 		bo->seqno = seqno;
 	}
+
+	for (i = 0; i < exec->rcl_write_bo_count; i++) {
+		bo = to_vc4_bo(&exec->rcl_write_bo[i]->base);
+		bo->write_seqno = seqno;
+	}
 }
 
 /* Queues a struct vc4_exec_info for execution.  If no job is
@@ -669,6 +674,14 @@ vc4_get_bcl(struct drm_device *dev, struct vc4_exec_info *exec)
 		goto fail;
 
 	ret = vc4_validate_shader_recs(dev, exec);
+	if (ret)
+		goto fail;
+
+	/* Block waiting on any previous rendering into the CS's VBO,
+	 * IB, or textures, so that pixels are actually written by the
+	 * time we try to read them.
+	 */
+	ret = vc4_wait_for_seqno(dev, exec->bin_dep_seqno, ~0ull, true);
 
 fail:
 	drm_free_large(temp);
diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c
index 68ad10634b29..c4cb2e26de32 100644
--- a/drivers/gpu/drm/vc4/vc4_hdmi.c
+++ b/drivers/gpu/drm/vc4/vc4_hdmi.c
@@ -62,6 +62,8 @@ struct vc4_hdmi {
 struct vc4_hdmi_encoder {
 	struct vc4_encoder base;
 	bool hdmi_monitor;
+	bool limited_rgb_range;
+	bool rgb_range_selectable;
 };
 
 static inline struct vc4_hdmi_encoder *
@@ -174,6 +176,9 @@ vc4_hdmi_connector_detect(struct drm_connector *connector, bool force)
 			return connector_status_disconnected;
 	}
 
+	if (drm_probe_ddc(vc4->hdmi->ddc))
+		return connector_status_connected;
+
 	if (HDMI_READ(VC4_HDMI_HOTPLUG) & VC4_HDMI_HOTPLUG_CONNECTED)
 		return connector_status_connected;
 	else
@@ -202,41 +207,22 @@ static int vc4_hdmi_connector_get_modes(struct drm_connector *connector)
 		return -ENODEV;
 
 	vc4_encoder->hdmi_monitor = drm_detect_hdmi_monitor(edid);
+
+	if (edid && edid->input & DRM_EDID_INPUT_DIGITAL) {
+		vc4_encoder->rgb_range_selectable =
+			drm_rgb_quant_range_selectable(edid);
+	}
+
 	drm_mode_connector_update_edid_property(connector, edid);
 	ret = drm_add_edid_modes(connector, edid);
 
 	return ret;
 }
 
-/*
- * drm_helper_probe_single_connector_modes() applies drm_mode_set_crtcinfo to
- * all modes with flag CRTC_INTERLACE_HALVE_V. We don't want this, as it
- * screws up vblank timestamping for interlaced modes, so fix it up.
- */
-static int vc4_hdmi_connector_probe_modes(struct drm_connector *connector,
-					  uint32_t maxX, uint32_t maxY)
-{
-	struct drm_display_mode *mode;
-	int count;
-
-	count = drm_helper_probe_single_connector_modes(connector, maxX, maxY);
-	if (count == 0)
-		return 0;
-
-	DRM_DEBUG_KMS("[CONNECTOR:%d:%s] probed adapted modes :\n",
-		      connector->base.id, connector->name);
-	list_for_each_entry(mode, &connector->modes, head) {
-		drm_mode_set_crtcinfo(mode, 0);
-		drm_mode_debug_printmodeline(mode);
-	}
-
-	return count;
-}
-
 static const struct drm_connector_funcs vc4_hdmi_connector_funcs = {
 	.dpms = drm_atomic_helper_connector_dpms,
 	.detect = vc4_hdmi_connector_detect,
-	.fill_modes = vc4_hdmi_connector_probe_modes,
+	.fill_modes = drm_helper_probe_single_connector_modes,
 	.destroy = vc4_hdmi_connector_destroy,
 	.reset = drm_atomic_helper_connector_reset,
 	.atomic_duplicate_state = drm_atomic_helper_connector_duplicate_state,
@@ -294,25 +280,143 @@ static const struct drm_encoder_funcs vc4_hdmi_encoder_funcs = {
 	.destroy = vc4_hdmi_encoder_destroy,
 };
 
+static int vc4_hdmi_stop_packet(struct drm_encoder *encoder,
+				enum hdmi_infoframe_type type)
+{
+	struct drm_device *dev = encoder->dev;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	u32 packet_id = type - 0x80;
+
+	HDMI_WRITE(VC4_HDMI_RAM_PACKET_CONFIG,
+		   HDMI_READ(VC4_HDMI_RAM_PACKET_CONFIG) & ~BIT(packet_id));
+
+	return wait_for(!(HDMI_READ(VC4_HDMI_RAM_PACKET_STATUS) &
+			  BIT(packet_id)), 100);
+}
+
+static void vc4_hdmi_write_infoframe(struct drm_encoder *encoder,
+				     union hdmi_infoframe *frame)
+{
+	struct drm_device *dev = encoder->dev;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	u32 packet_id = frame->any.type - 0x80;
+	u32 packet_reg = VC4_HDMI_GCP_0 + VC4_HDMI_PACKET_STRIDE * packet_id;
+	uint8_t buffer[VC4_HDMI_PACKET_STRIDE];
+	ssize_t len, i;
+	int ret;
+
+	WARN_ONCE(!(HDMI_READ(VC4_HDMI_RAM_PACKET_CONFIG) &
+		    VC4_HDMI_RAM_PACKET_ENABLE),
+		  "Packet RAM has to be on to store the packet.");
+
+	len = hdmi_infoframe_pack(frame, buffer, sizeof(buffer));
+	if (len < 0)
+		return;
+
+	ret = vc4_hdmi_stop_packet(encoder, frame->any.type);
+	if (ret) {
+		DRM_ERROR("Failed to wait for infoframe to go idle: %d\n", ret);
+		return;
+	}
+
+	for (i = 0; i < len; i += 7) {
+		HDMI_WRITE(packet_reg,
+			   buffer[i + 0] << 0 |
+			   buffer[i + 1] << 8 |
+			   buffer[i + 2] << 16);
+		packet_reg += 4;
+
+		HDMI_WRITE(packet_reg,
+			   buffer[i + 3] << 0 |
+			   buffer[i + 4] << 8 |
+			   buffer[i + 5] << 16 |
+			   buffer[i + 6] << 24);
+		packet_reg += 4;
+	}
+
+	HDMI_WRITE(VC4_HDMI_RAM_PACKET_CONFIG,
+		   HDMI_READ(VC4_HDMI_RAM_PACKET_CONFIG) | BIT(packet_id));
+	ret = wait_for((HDMI_READ(VC4_HDMI_RAM_PACKET_STATUS) &
+			BIT(packet_id)), 100);
+	if (ret)
+		DRM_ERROR("Failed to wait for infoframe to start: %d\n", ret);
+}
+
+static void vc4_hdmi_set_avi_infoframe(struct drm_encoder *encoder)
+{
+	struct vc4_hdmi_encoder *vc4_encoder = to_vc4_hdmi_encoder(encoder);
+	struct drm_crtc *crtc = encoder->crtc;
+	const struct drm_display_mode *mode = &crtc->state->adjusted_mode;
+	union hdmi_infoframe frame;
+	int ret;
+
+	ret = drm_hdmi_avi_infoframe_from_display_mode(&frame.avi, mode);
+	if (ret < 0) {
+		DRM_ERROR("couldn't fill AVI infoframe\n");
+		return;
+	}
+
+	if (vc4_encoder->rgb_range_selectable) {
+		if (vc4_encoder->limited_rgb_range) {
+			frame.avi.quantization_range =
+				HDMI_QUANTIZATION_RANGE_LIMITED;
+		} else {
+			frame.avi.quantization_range =
+				HDMI_QUANTIZATION_RANGE_FULL;
+		}
+	}
+
+	vc4_hdmi_write_infoframe(encoder, &frame);
+}
+
+static void vc4_hdmi_set_spd_infoframe(struct drm_encoder *encoder)
+{
+	union hdmi_infoframe frame;
+	int ret;
+
+	ret = hdmi_spd_infoframe_init(&frame.spd, "Broadcom", "Videocore");
+	if (ret < 0) {
+		DRM_ERROR("couldn't fill SPD infoframe\n");
+		return;
+	}
+
+	frame.spd.sdi = HDMI_SPD_SDI_PC;
+
+	vc4_hdmi_write_infoframe(encoder, &frame);
+}
+
+static void vc4_hdmi_set_infoframes(struct drm_encoder *encoder)
+{
+	vc4_hdmi_set_avi_infoframe(encoder);
+	vc4_hdmi_set_spd_infoframe(encoder);
+}
+
 static void vc4_hdmi_encoder_mode_set(struct drm_encoder *encoder,
 				      struct drm_display_mode *unadjusted_mode,
 				      struct drm_display_mode *mode)
 {
+	struct vc4_hdmi_encoder *vc4_encoder = to_vc4_hdmi_encoder(encoder);
 	struct drm_device *dev = encoder->dev;
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	bool debug_dump_regs = false;
 	bool hsync_pos = mode->flags & DRM_MODE_FLAG_PHSYNC;
 	bool vsync_pos = mode->flags & DRM_MODE_FLAG_PVSYNC;
-	u32 vactive = (mode->vdisplay >>
-		       ((mode->flags & DRM_MODE_FLAG_INTERLACE) ? 1 : 0));
-	u32 verta = (VC4_SET_FIELD(mode->vsync_end - mode->vsync_start,
+	bool interlaced = mode->flags & DRM_MODE_FLAG_INTERLACE;
+	u32 pixel_rep = (mode->flags & DRM_MODE_FLAG_DBLCLK) ? 2 : 1;
+	u32 verta = (VC4_SET_FIELD(mode->crtc_vsync_end - mode->crtc_vsync_start,
 				   VC4_HDMI_VERTA_VSP) |
-		     VC4_SET_FIELD(mode->vsync_start - mode->vdisplay,
+		     VC4_SET_FIELD(mode->crtc_vsync_start - mode->crtc_vdisplay,
 				   VC4_HDMI_VERTA_VFP) |
-		     VC4_SET_FIELD(vactive, VC4_HDMI_VERTA_VAL));
+		     VC4_SET_FIELD(mode->crtc_vdisplay, VC4_HDMI_VERTA_VAL));
 	u32 vertb = (VC4_SET_FIELD(0, VC4_HDMI_VERTB_VSPO) |
-		     VC4_SET_FIELD(mode->vtotal - mode->vsync_end,
+		     VC4_SET_FIELD(mode->crtc_vtotal - mode->crtc_vsync_end,
 				   VC4_HDMI_VERTB_VBP));
+	u32 vertb_even = (VC4_SET_FIELD(0, VC4_HDMI_VERTB_VSPO) |
+			  VC4_SET_FIELD(mode->crtc_vtotal -
+					mode->crtc_vsync_end -
+					interlaced,
+					VC4_HDMI_VERTB_VBP));
+	u32 csc_ctl;
 
 	if (debug_dump_regs) {
 		DRM_INFO("HDMI regs before:\n");
@@ -321,7 +425,8 @@ static void vc4_hdmi_encoder_mode_set(struct drm_encoder *encoder,
 
 	HD_WRITE(VC4_HD_VID_CTL, 0);
 
-	clk_set_rate(vc4->hdmi->pixel_clock, mode->clock * 1000);
+	clk_set_rate(vc4->hdmi->pixel_clock, mode->clock * 1000 *
+		     ((mode->flags & DRM_MODE_FLAG_DBLCLK) ? 2 : 1));
 
 	HDMI_WRITE(VC4_HDMI_SCHEDULER_CONTROL,
 		   HDMI_READ(VC4_HDMI_SCHEDULER_CONTROL) |
@@ -331,29 +436,62 @@ static void vc4_hdmi_encoder_mode_set(struct drm_encoder *encoder,
 	HDMI_WRITE(VC4_HDMI_HORZA,
 		   (vsync_pos ? VC4_HDMI_HORZA_VPOS : 0) |
 		   (hsync_pos ? VC4_HDMI_HORZA_HPOS : 0) |
-		   VC4_SET_FIELD(mode->hdisplay, VC4_HDMI_HORZA_HAP));
+		   VC4_SET_FIELD(mode->hdisplay * pixel_rep,
+				 VC4_HDMI_HORZA_HAP));
 
 	HDMI_WRITE(VC4_HDMI_HORZB,
-		   VC4_SET_FIELD(mode->htotal - mode->hsync_end,
+		   VC4_SET_FIELD((mode->htotal -
+				  mode->hsync_end) * pixel_rep,
 				 VC4_HDMI_HORZB_HBP) |
-		   VC4_SET_FIELD(mode->hsync_end - mode->hsync_start,
+		   VC4_SET_FIELD((mode->hsync_end -
+				  mode->hsync_start) * pixel_rep,
 				 VC4_HDMI_HORZB_HSP) |
-		   VC4_SET_FIELD(mode->hsync_start - mode->hdisplay,
+		   VC4_SET_FIELD((mode->hsync_start -
+				  mode->hdisplay) * pixel_rep,
 				 VC4_HDMI_HORZB_HFP));
 
 	HDMI_WRITE(VC4_HDMI_VERTA0, verta);
 	HDMI_WRITE(VC4_HDMI_VERTA1, verta);
 
-	HDMI_WRITE(VC4_HDMI_VERTB0, vertb);
+	HDMI_WRITE(VC4_HDMI_VERTB0, vertb_even);
 	HDMI_WRITE(VC4_HDMI_VERTB1, vertb);
 
 	HD_WRITE(VC4_HD_VID_CTL,
 		 (vsync_pos ? 0 : VC4_HD_VID_CTL_VSYNC_LOW) |
 		 (hsync_pos ? 0 : VC4_HD_VID_CTL_HSYNC_LOW));
 
+	csc_ctl = VC4_SET_FIELD(VC4_HD_CSC_CTL_ORDER_BGR,
+				VC4_HD_CSC_CTL_ORDER);
+
+	if (vc4_encoder->hdmi_monitor && drm_match_cea_mode(mode) > 1) {
+		/* CEA VICs other than #1 requre limited range RGB
+		 * output unless overridden by an AVI infoframe.
+		 * Apply a colorspace conversion to squash 0-255 down
+		 * to 16-235.  The matrix here is:
+		 *
+		 * [ 0      0      0.8594 16]
+		 * [ 0      0.8594 0      16]
+		 * [ 0.8594 0      0      16]
+		 * [ 0      0      0       1]
+		 */
+		csc_ctl |= VC4_HD_CSC_CTL_ENABLE;
+		csc_ctl |= VC4_HD_CSC_CTL_RGB2YCC;
+		csc_ctl |= VC4_SET_FIELD(VC4_HD_CSC_CTL_MODE_CUSTOM,
+					 VC4_HD_CSC_CTL_MODE);
+
+		HD_WRITE(VC4_HD_CSC_12_11, (0x000 << 16) | 0x000);
+		HD_WRITE(VC4_HD_CSC_14_13, (0x100 << 16) | 0x6e0);
+		HD_WRITE(VC4_HD_CSC_22_21, (0x6e0 << 16) | 0x000);
+		HD_WRITE(VC4_HD_CSC_24_23, (0x100 << 16) | 0x000);
+		HD_WRITE(VC4_HD_CSC_32_31, (0x000 << 16) | 0x6e0);
+		HD_WRITE(VC4_HD_CSC_34_33, (0x100 << 16) | 0x000);
+		vc4_encoder->limited_rgb_range = true;
+	} else {
+		vc4_encoder->limited_rgb_range = false;
+	}
+
 	/* The RGB order applies even when CSC is disabled. */
-	HD_WRITE(VC4_HD_CSC_CTL, VC4_SET_FIELD(VC4_HD_CSC_CTL_ORDER_BGR,
-					       VC4_HD_CSC_CTL_ORDER));
+	HD_WRITE(VC4_HD_CSC_CTL, csc_ctl);
 
 	HDMI_WRITE(VC4_HDMI_FIFO_CTL, VC4_HDMI_FIFO_CTL_MASTER_SLAVE_N);
 
@@ -368,6 +506,8 @@ static void vc4_hdmi_encoder_disable(struct drm_encoder *encoder)
 	struct drm_device *dev = encoder->dev;
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 
+	HDMI_WRITE(VC4_HDMI_RAM_PACKET_CONFIG, 0);
+
 	HDMI_WRITE(VC4_HDMI_TX_PHY_RESET_CTL, 0xf << 16);
 	HD_WRITE(VC4_HD_VID_CTL,
 		 HD_READ(VC4_HD_VID_CTL) & ~VC4_HD_VID_CTL_ENABLE);
@@ -394,7 +534,7 @@ static void vc4_hdmi_encoder_enable(struct drm_encoder *encoder)
 			   VC4_HDMI_SCHEDULER_CONTROL_MODE_HDMI);
 
 		ret = wait_for(HDMI_READ(VC4_HDMI_SCHEDULER_CONTROL) &
-			       VC4_HDMI_SCHEDULER_CONTROL_HDMI_ACTIVE, 1);
+			       VC4_HDMI_SCHEDULER_CONTROL_HDMI_ACTIVE, 1000);
 		WARN_ONCE(ret, "Timeout waiting for "
 			  "VC4_HDMI_SCHEDULER_CONTROL_HDMI_ACTIVE\n");
 	} else {
@@ -406,7 +546,7 @@ static void vc4_hdmi_encoder_enable(struct drm_encoder *encoder)
 			   ~VC4_HDMI_SCHEDULER_CONTROL_MODE_HDMI);
 
 		ret = wait_for(!(HDMI_READ(VC4_HDMI_SCHEDULER_CONTROL) &
-				 VC4_HDMI_SCHEDULER_CONTROL_HDMI_ACTIVE), 1);
+				 VC4_HDMI_SCHEDULER_CONTROL_HDMI_ACTIVE), 1000);
 		WARN_ONCE(ret, "Timeout waiting for "
 			  "!VC4_HDMI_SCHEDULER_CONTROL_HDMI_ACTIVE\n");
 	}
@@ -420,9 +560,10 @@ static void vc4_hdmi_encoder_enable(struct drm_encoder *encoder)
 			   HDMI_READ(VC4_HDMI_SCHEDULER_CONTROL) |
 			   VC4_HDMI_SCHEDULER_CONTROL_VERT_ALWAYS_KEEPOUT);
 
-		/* XXX: Set HDMI_RAM_PACKET_CONFIG (1 << 16) and set
-		 * up the infoframe.
-		 */
+		HDMI_WRITE(VC4_HDMI_RAM_PACKET_CONFIG,
+			   VC4_HDMI_RAM_PACKET_ENABLE);
+
+		vc4_hdmi_set_infoframes(encoder);
 
 		drift = HDMI_READ(VC4_HDMI_FIFO_CTL);
 		drift &= VC4_HDMI_FIFO_VALID_WRITE_MASK;
diff --git a/drivers/gpu/drm/vc4/vc4_regs.h b/drivers/gpu/drm/vc4/vc4_regs.h
index 160942a9180e..1aa44c2db556 100644
--- a/drivers/gpu/drm/vc4/vc4_regs.h
+++ b/drivers/gpu/drm/vc4/vc4_regs.h
@@ -175,6 +175,8 @@
 # define PV_CONTROL_CLR_AT_START		BIT(14)
 # define PV_CONTROL_TRIGGER_UNDERFLOW		BIT(13)
 # define PV_CONTROL_WAIT_HSTART			BIT(12)
+# define PV_CONTROL_PIXEL_REP_MASK		VC4_MASK(5, 4)
+# define PV_CONTROL_PIXEL_REP_SHIFT		4
 # define PV_CONTROL_CLK_SELECT_DSI_VEC		0
 # define PV_CONTROL_CLK_SELECT_DPI_SMI_HDMI	1
 # define PV_CONTROL_CLK_SELECT_MASK		VC4_MASK(3, 2)
@@ -183,6 +185,9 @@
 # define PV_CONTROL_EN				BIT(0)
 
 #define PV_V_CONTROL				0x04
+# define PV_VCONTROL_ODD_DELAY_MASK		VC4_MASK(22, 6)
+# define PV_VCONTROL_ODD_DELAY_SHIFT		6
+# define PV_VCONTROL_ODD_FIRST			BIT(5)
 # define PV_VCONTROL_INTERLACE			BIT(4)
 # define PV_VCONTROL_CONTINUOUS			BIT(1)
 # define PV_VCONTROL_VIDEN			BIT(0)
@@ -438,6 +443,8 @@
 #define VC4_HDMI_RAM_PACKET_CONFIG		0x0a0
 # define VC4_HDMI_RAM_PACKET_ENABLE		BIT(16)
 
+#define VC4_HDMI_RAM_PACKET_STATUS		0x0a4
+
 #define VC4_HDMI_HORZA				0x0c4
 # define VC4_HDMI_HORZA_VPOS			BIT(14)
 # define VC4_HDMI_HORZA_HPOS			BIT(13)
@@ -499,6 +506,9 @@
 
 #define VC4_HDMI_TX_PHY_RESET_CTL		0x2c0
 
+#define VC4_HDMI_GCP_0				0x400
+#define VC4_HDMI_PACKET_STRIDE			0x24
+
 #define VC4_HD_M_CTL				0x00c
 # define VC4_HD_M_REGISTER_FILE_STANDBY		(3 << 6)
 # define VC4_HD_M_RAM_STANDBY			(3 << 4)
@@ -528,10 +538,17 @@
 # define VC4_HD_CSC_CTL_MODE_SHIFT		2
 # define VC4_HD_CSC_CTL_MODE_RGB_TO_SD_YPRPB	0
 # define VC4_HD_CSC_CTL_MODE_RGB_TO_HD_YPRPB	1
-# define VC4_HD_CSC_CTL_MODE_CUSTOM		2
+# define VC4_HD_CSC_CTL_MODE_CUSTOM		3
 # define VC4_HD_CSC_CTL_RGB2YCC			BIT(1)
 # define VC4_HD_CSC_CTL_ENABLE			BIT(0)
 
+#define VC4_HD_CSC_12_11			0x044
+#define VC4_HD_CSC_14_13			0x048
+#define VC4_HD_CSC_22_21			0x04c
+#define VC4_HD_CSC_24_23			0x050
+#define VC4_HD_CSC_32_31			0x054
+#define VC4_HD_CSC_34_33			0x058
+
 #define VC4_HD_FRAME_COUNT			0x068
 
 /* HVS display list information. */
diff --git a/drivers/gpu/drm/vc4/vc4_render_cl.c b/drivers/gpu/drm/vc4/vc4_render_cl.c
index 0f12418725e5..08886a309757 100644
--- a/drivers/gpu/drm/vc4/vc4_render_cl.c
+++ b/drivers/gpu/drm/vc4/vc4_render_cl.c
@@ -45,6 +45,8 @@ struct vc4_rcl_setup {
 
 	struct drm_gem_cma_object *rcl;
 	u32 next_offset;
+
+	u32 next_write_bo_index;
 };
 
 static inline void rcl_u8(struct vc4_rcl_setup *setup, u8 val)
@@ -407,6 +409,8 @@ static int vc4_rcl_msaa_surface_setup(struct vc4_exec_info *exec,
 	if (!*obj)
 		return -EINVAL;
 
+	exec->rcl_write_bo[exec->rcl_write_bo_count++] = *obj;
+
 	if (surf->offset & 0xf) {
 		DRM_ERROR("MSAA write must be 16b aligned.\n");
 		return -EINVAL;
@@ -417,7 +421,8 @@ static int vc4_rcl_msaa_surface_setup(struct vc4_exec_info *exec,
 
 static int vc4_rcl_surface_setup(struct vc4_exec_info *exec,
 				 struct drm_gem_cma_object **obj,
-				 struct drm_vc4_submit_rcl_surface *surf)
+				 struct drm_vc4_submit_rcl_surface *surf,
+				 bool is_write)
 {
 	uint8_t tiling = VC4_GET_FIELD(surf->bits,
 				       VC4_LOADSTORE_TILE_BUFFER_TILING);
@@ -440,6 +445,9 @@ static int vc4_rcl_surface_setup(struct vc4_exec_info *exec,
 	if (!*obj)
 		return -EINVAL;
 
+	if (is_write)
+		exec->rcl_write_bo[exec->rcl_write_bo_count++] = *obj;
+
 	if (surf->flags & VC4_SUBMIT_RCL_SURFACE_READ_IS_FULL_RES) {
 		if (surf == &exec->args->zs_write) {
 			DRM_ERROR("general zs write may not be a full-res.\n");
@@ -542,6 +550,8 @@ vc4_rcl_render_config_surface_setup(struct vc4_exec_info *exec,
 	if (!*obj)
 		return -EINVAL;
 
+	exec->rcl_write_bo[exec->rcl_write_bo_count++] = *obj;
+
 	if (tiling > VC4_TILING_FORMAT_LT) {
 		DRM_ERROR("Bad tiling format\n");
 		return -EINVAL;
@@ -599,15 +609,18 @@ int vc4_get_rcl(struct drm_device *dev, struct vc4_exec_info *exec)
 	if (ret)
 		return ret;
 
-	ret = vc4_rcl_surface_setup(exec, &setup.color_read, &args->color_read);
+	ret = vc4_rcl_surface_setup(exec, &setup.color_read, &args->color_read,
+				    false);
 	if (ret)
 		return ret;
 
-	ret = vc4_rcl_surface_setup(exec, &setup.zs_read, &args->zs_read);
+	ret = vc4_rcl_surface_setup(exec, &setup.zs_read, &args->zs_read,
+				    false);
 	if (ret)
 		return ret;
 
-	ret = vc4_rcl_surface_setup(exec, &setup.zs_write, &args->zs_write);
+	ret = vc4_rcl_surface_setup(exec, &setup.zs_write, &args->zs_write,
+				    true);
 	if (ret)
 		return ret;
 
diff --git a/drivers/gpu/drm/vc4/vc4_validate.c b/drivers/gpu/drm/vc4/vc4_validate.c
index 9ce1d0adf882..26503e307438 100644
--- a/drivers/gpu/drm/vc4/vc4_validate.c
+++ b/drivers/gpu/drm/vc4/vc4_validate.c
@@ -267,6 +267,9 @@ validate_indexed_prim_list(VALIDATE_ARGS)
 	if (!ib)
 		return -EINVAL;
 
+	exec->bin_dep_seqno = max(exec->bin_dep_seqno,
+				  to_vc4_bo(&ib->base)->write_seqno);
+
 	if (offset > ib->base.size ||
 	    (ib->base.size - offset) / index_size < length) {
 		DRM_ERROR("IB access overflow (%d + %d*%d > %zd)\n",
@@ -555,8 +558,7 @@ static bool
 reloc_tex(struct vc4_exec_info *exec,
 	  void *uniform_data_u,
 	  struct vc4_texture_sample_info *sample,
-	  uint32_t texture_handle_index)
-
+	  uint32_t texture_handle_index, bool is_cs)
 {
 	struct drm_gem_cma_object *tex;
 	uint32_t p0 = *(uint32_t *)(uniform_data_u + sample->p_offset[0]);
@@ -714,6 +716,11 @@ reloc_tex(struct vc4_exec_info *exec,
 
 	*validated_p0 = tex->paddr + p0;
 
+	if (is_cs) {
+		exec->bin_dep_seqno = max(exec->bin_dep_seqno,
+					  to_vc4_bo(&tex->base)->write_seqno);
+	}
+
 	return true;
  fail:
 	DRM_INFO("Texture p0 at %d: 0x%08x\n", sample->p_offset[0], p0);
@@ -835,7 +842,8 @@ validate_gl_shader_rec(struct drm_device *dev,
 			if (!reloc_tex(exec,
 				       uniform_data_u,
 				       &validated_shader->texture_samples[tex],
-				       texture_handles_u[tex])) {
+				       texture_handles_u[tex],
+				       i == 2)) {
 				return -EINVAL;
 			}
 		}
@@ -867,6 +875,9 @@ validate_gl_shader_rec(struct drm_device *dev,
 		uint32_t stride = *(uint8_t *)(pkt_u + o + 5);
 		uint32_t max_index;
 
+		exec->bin_dep_seqno = max(exec->bin_dep_seqno,
+					  to_vc4_bo(&vbo->base)->write_seqno);
+
 		if (state->addr & 0x8)
 			stride |= (*(uint32_t *)(pkt_u + 100 + i * 4)) & ~0xff;