vc4 loops support

arch/arm/boot/dts/bcm2708_common.dtsi

@@ -347,7 +347,6 @@
 			reg = <0x7e902000 0x600>,
 			      <0x7e808000 0x100>;
 			ddc = <&i2c2>;
-			hpd-gpios = <&gpio 46 GPIO_ACTIVE_HIGH>;
 			clocks = <&cprman BCM2835_PLLH_PIX>,
 				 <&cprman BCM2835_CLOCK_HSM>;
 			clock-names = "pixel", "hdmi";
@@ -366,6 +365,7 @@
 		v3d: v3d@7ec00000 {
 			compatible = "brcm,vc4-v3d";
 			reg = <0x7ec00000 0x1000>;
+			power-domains = <&power RPI_POWER_DOMAIN_V3D>;
 			status = "disabled";
 		};
 

arch/arm/boot/dts/bcm2710-rpi-3-b.dts

@@ -166,10 +166,6 @@
 	};
 };
 
-&hdmi {
-	hpd-gpios = <&gpio 46 GPIO_ACTIVE_LOW>;
-};
-
 &audio {
 	pinctrl-names = "default";
 	pinctrl-0 = <&audio_pins>;

drivers/gpu/drm/panel/panel-raspberrypi-touchscreen.c

@@ -106,7 +106,6 @@ struct regdump {
 static int rpi_touchscreen_disable(struct drm_panel *panel)
 {
 	struct rpi_touchscreen *ts = panel_to_ts(panel);
-	pr_err("disable\n");
 
 	if (!ts->enabled)
 		return 0;

drivers/gpu/drm/vc4/vc4_drv.c

@@ -15,7 +15,7 @@
 #include <linux/module.h>
 #include <linux/of_platform.h>
 #include <linux/platform_device.h>
-#include <soc/bcm2835/raspberrypi-firmware.h>
+#include <linux/pm_runtime.h>
 #include "drm_fb_cma_helper.h"
 
 #include "uapi/drm/vc4_drm.h"
@@ -65,6 +65,49 @@ void vc4_dump_regs32(const struct debugfs_reg32 *regs, unsigned int num_regs,
 	}
 }
 
+static int vc4_get_param_ioctl(struct drm_device *dev, void *data,
+			       struct drm_file *file_priv)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct drm_vc4_get_param *args = data;
+	int ret;
+
+	if (args->pad != 0)
+		return -EINVAL;
+
+	switch (args->param) {
+	case DRM_VC4_PARAM_V3D_IDENT0:
+		ret = pm_runtime_get_sync(&vc4->v3d->pdev->dev);
+		if (ret < 0)
+			return ret;
+		args->value = V3D_READ(V3D_IDENT0);
+		pm_runtime_put(&vc4->v3d->pdev->dev);
+		break;
+	case DRM_VC4_PARAM_V3D_IDENT1:
+		ret = pm_runtime_get_sync(&vc4->v3d->pdev->dev);
+		if (ret < 0)
+			return ret;
+		args->value = V3D_READ(V3D_IDENT1);
+		pm_runtime_put(&vc4->v3d->pdev->dev);
+		break;
+	case DRM_VC4_PARAM_V3D_IDENT2:
+		ret = pm_runtime_get_sync(&vc4->v3d->pdev->dev);
+		if (ret < 0)
+			return ret;
+		args->value = V3D_READ(V3D_IDENT2);
+		pm_runtime_put(&vc4->v3d->pdev->dev);
+		break;
+	case DRM_VC4_PARAM_SUPPORTS_BRANCHES:
+		args->value = true;
+		break;
+	default:
+		DRM_DEBUG("Unknown parameter %d\n", args->param);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 static void vc4_lastclose(struct drm_device *dev)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
@@ -96,6 +139,7 @@ static const struct drm_ioctl_desc vc4_drm_ioctls[] = {
 	DRM_IOCTL_DEF_DRV(VC4_CREATE_SHADER_BO, vc4_create_shader_bo_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(VC4_GET_HANG_STATE, vc4_get_hang_state_ioctl,
 			  DRM_ROOT_ONLY),
+	DRM_IOCTL_DEF_DRV(VC4_GET_PARAM, vc4_get_param_ioctl, DRM_RENDER_ALLOW),
 };
 
 static struct drm_driver vc4_drm_driver = {
@@ -204,7 +248,6 @@ static int vc4_drm_bind(struct device *dev)
 	struct drm_device *drm;
 	struct drm_connector *connector;
 	struct vc4_dev *vc4;
-	struct device_node *firmware_node;
 	int ret = 0;
 
 	dev->coherent_dma_mask = DMA_BIT_MASK(32);
@@ -213,14 +256,6 @@ static int vc4_drm_bind(struct device *dev)
 	if (!vc4)
 		return -ENOMEM;
 
-	firmware_node = of_parse_phandle(dev->of_node, "firmware", 0);
-	vc4->firmware = rpi_firmware_get(firmware_node);
-	if (!vc4->firmware) {
-		DRM_DEBUG("Failed to get Raspberry Pi firmware reference.\n");
-		return -EPROBE_DEFER;
-	}
-	of_node_put(firmware_node);
-
 	drm = drm_dev_alloc(&vc4_drm_driver, dev);
 	if (!drm)
 		return -ENOMEM;

drivers/gpu/drm/vc4/vc4_drv.h

@@ -23,7 +23,6 @@ struct vc4_dev {
 	struct vc4_dsi *dsi1;
 
 	struct drm_fbdev_cma *fbdev;
-	struct rpi_firmware *firmware;
 
 	struct vc4_hang_state *hang_state;
 
@@ -105,6 +104,11 @@ struct vc4_dev {
 	struct vc4_bo *overflow_mem;
 	struct work_struct overflow_mem_work;
 
+	int power_refcount;
+
+	/* Mutex controlling the power refcount. */
+	struct mutex power_lock;
+
 	struct {
 		struct timer_list timer;
 		struct work_struct reset_work;
@@ -155,6 +159,7 @@ struct vc4_seqno_cb {
 };
 
 struct vc4_v3d {
+	struct vc4_dev *vc4;
 	struct platform_device *pdev;
 	void __iomem *regs;
 };
@@ -324,6 +329,15 @@ vc4_first_render_job(struct vc4_dev *vc4)
 				struct vc4_exec_info, head);
 }
 
+static inline struct vc4_exec_info *
+vc4_last_render_job(struct vc4_dev *vc4)
+{
+	if (list_empty(&vc4->render_job_list))
+		return NULL;
+	return list_last_entry(&vc4->render_job_list,
+			       struct vc4_exec_info, head);
+}
+
 /**
  * struct vc4_texture_sample_info - saves the offsets into the UBO for texture
  * setup parameters.
@@ -358,6 +372,9 @@ struct vc4_validated_shader_info {
 	uint32_t uniforms_src_size;
 	uint32_t num_texture_samples;
 	struct vc4_texture_sample_info *texture_samples;
+
+	uint32_t num_uniform_addr_offsets;
+	uint32_t *uniform_addr_offsets;
 };
 
 /**
@@ -495,7 +512,6 @@ void vc4_plane_async_set_fb(struct drm_plane *plane,
 extern struct platform_driver vc4_v3d_driver;
 int vc4_v3d_debugfs_ident(struct seq_file *m, void *unused);
 int vc4_v3d_debugfs_regs(struct seq_file *m, void *unused);
-int vc4_v3d_set_power(struct vc4_dev *vc4, bool on);
 
 /* vc4_validate.c */
 int

drivers/gpu/drm/vc4/vc4_gem.c

@@ -23,6 +23,7 @@
 
 #include <linux/module.h>
 #include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
 #include <linux/device.h>
 #include <linux/io.h>
 
@@ -260,8 +261,16 @@ vc4_reset(struct drm_device *dev)
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 
 	DRM_INFO("Resetting GPU.\n");
-	vc4_v3d_set_power(vc4, false);
-	vc4_v3d_set_power(vc4, true);
+
+	mutex_lock(&vc4->power_lock);
+	if (vc4->power_refcount) {
+		/* Power the device off and back on the by dropping the
+		 * reference on runtime PM.
+		 */
+		pm_runtime_put_sync_suspend(&vc4->v3d->pdev->dev);
+		pm_runtime_get_sync(&vc4->v3d->pdev->dev);
+	}
+	mutex_unlock(&vc4->power_lock);
 
 	vc4_irq_reset(dev);
 
@@ -426,10 +435,6 @@ vc4_submit_next_bin_job(struct drm_device *dev)
 
 	vc4_flush_caches(dev);
 
-	/* Disable the binner's pre-loaded overflow memory address */
-	V3D_WRITE(V3D_BPOA, 0);
-	V3D_WRITE(V3D_BPOS, 0);
-
 	/* Either put the job in the binner if it uses the binner, or
 	 * immediately move it to the to-be-rendered queue.
 	 */
@@ -579,8 +584,8 @@ vc4_cl_lookup_bos(struct drm_device *dev,
 	spin_unlock(&file_priv->table_lock);
 
 fail:
-	kfree(handles);
-	return 0;
+	drm_free_large(handles);
+	return ret;
 }
 
 static int
@@ -689,6 +694,7 @@ vc4_get_bcl(struct drm_device *dev, struct vc4_exec_info *exec)
 static void
 vc4_complete_exec(struct drm_device *dev, struct vc4_exec_info *exec)
 {
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	unsigned i;
 
 	/* Need the struct lock for drm_gem_object_unreference(). */
@@ -707,6 +713,11 @@ vc4_complete_exec(struct drm_device *dev, struct vc4_exec_info *exec)
 	}
 	mutex_unlock(&dev->struct_mutex);
 
+	mutex_lock(&vc4->power_lock);
+	if (--vc4->power_refcount == 0)
+		pm_runtime_put(&vc4->v3d->pdev->dev);
+	mutex_unlock(&vc4->power_lock);
+
 	kfree(exec);
 }
 
@@ -847,7 +858,7 @@ vc4_submit_cl_ioctl(struct drm_device *dev, void *data,
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	struct drm_vc4_submit_cl *args = data;
 	struct vc4_exec_info *exec;
-	int ret;
+	int ret = 0;
 
 	if ((args->flags & ~VC4_SUBMIT_CL_USE_CLEAR_COLOR) != 0) {
 		DRM_ERROR("Unknown flags: 0x%02x\n", args->flags);
@@ -860,6 +871,15 @@ vc4_submit_cl_ioctl(struct drm_device *dev, void *data,
 		return -ENOMEM;
 	}
 
+	mutex_lock(&vc4->power_lock);
+	if (vc4->power_refcount++ == 0)
+		ret = pm_runtime_get_sync(&vc4->v3d->pdev->dev);
+	mutex_unlock(&vc4->power_lock);
+	if (ret < 0) {
+		kfree(exec);
+		return ret;
+	}
+
 	exec->args = args;
 	INIT_LIST_HEAD(&exec->unref_list);
 
@@ -915,6 +935,8 @@ vc4_gem_init(struct drm_device *dev)
 		    (unsigned long)dev);
 
 	INIT_WORK(&vc4->job_done_work, vc4_job_done_work);
+
+	mutex_init(&vc4->power_lock);
 }
 
 void

drivers/gpu/drm/vc4/vc4_hdmi.c

@@ -166,8 +166,6 @@ vc4_hdmi_connector_detect(struct drm_connector *connector, bool force)
 	struct drm_device *dev = connector->dev;
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 
-	return connector_status_connected;
-
 	if (vc4->hdmi->hpd_gpio) {
 		if (gpio_get_value_cansleep(vc4->hdmi->hpd_gpio) ^
 		    vc4->hdmi->hpd_active_low)
@@ -176,6 +174,9 @@ 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

drivers/gpu/drm/vc4/vc4_irq.c

@@ -83,8 +83,10 @@ vc4_overflow_mem_work(struct work_struct *work)
 
 		spin_lock_irqsave(&vc4->job_lock, irqflags);
 		current_exec = vc4_first_bin_job(vc4);
+		if (!current_exec)
+			current_exec = vc4_last_render_job(vc4);
 		if (current_exec) {
-			vc4->overflow_mem->seqno = vc4->finished_seqno + 1;
+			vc4->overflow_mem->seqno = current_exec->seqno;
 			list_add_tail(&vc4->overflow_mem->unref_head,
 				      &current_exec->unref_list);
 			vc4->overflow_mem = NULL;

drivers/gpu/drm/vc4/vc4_qpu_defines.h

@@ -230,6 +230,15 @@ enum qpu_unpack_r4 {
 #define QPU_COND_MUL_SHIFT              46
 #define QPU_COND_MUL_MASK               QPU_MASK(48, 46)
 
+#define QPU_BRANCH_COND_SHIFT           52
+#define QPU_BRANCH_COND_MASK            QPU_MASK(55, 52)
+
+#define QPU_BRANCH_REL                  ((uint64_t)1 << 51)
+#define QPU_BRANCH_REG                  ((uint64_t)1 << 50)
+
+#define QPU_BRANCH_RADDR_A_SHIFT        45
+#define QPU_BRANCH_RADDR_A_MASK         QPU_MASK(49, 45)
+
 #define QPU_SF                          ((uint64_t)1 << 45)
 
 #define QPU_WADDR_ADD_SHIFT             38
@@ -261,4 +270,10 @@ enum qpu_unpack_r4 {
 #define QPU_OP_ADD_SHIFT                24
 #define QPU_OP_ADD_MASK                 QPU_MASK(28, 24)
 
+#define QPU_LOAD_IMM_SHIFT              0
+#define QPU_LOAD_IMM_MASK               QPU_MASK(31, 0)
+
+#define QPU_BRANCH_TARGET_SHIFT         0
+#define QPU_BRANCH_TARGET_MASK          QPU_MASK(31, 0)
+
 #endif /* VC4_QPU_DEFINES_H */