Resguardo inicial

linux-7.2.3/drivers/gpu/drm/vc4/Kconfig

@@ -0,0 +1,57 @@
+# SPDX-License-Identifier: GPL-2.0-only
+config DRM_VC4
+	tristate "Broadcom VC4 Graphics"
+	depends on ARCH_BCM || ARCH_BCM2835 || COMPILE_TEST
+	# Make sure not 'y' when RASPBERRYPI_FIRMWARE is 'm'. This can only
+	# happen when COMPILE_TEST=y, hence the added !RASPBERRYPI_FIRMWARE.
+	depends on RASPBERRYPI_FIRMWARE || (COMPILE_TEST && !RASPBERRYPI_FIRMWARE)
+	depends on DRM
+	depends on SND && SND_SOC
+	depends on COMMON_CLK
+	depends on PM
+	select DRM_CLIENT_SELECTION
+	select DRM_DISPLAY_HDMI_AUDIO_HELPER
+	select DRM_DISPLAY_HDMI_HELPER
+	select DRM_DISPLAY_HDMI_STATE_HELPER
+	select DRM_DISPLAY_HELPER
+	select DRM_EXEC
+	select DRM_KMS_HELPER
+	select DRM_GEM_DMA_HELPER
+	select DRM_PANEL_BRIDGE
+	select SND_PCM
+	select SND_PCM_ELD
+	select SND_SOC_GENERIC_DMAENGINE_PCM
+	select SND_SOC_HDMI_CODEC
+	select DRM_MIPI_DSI
+	help
+	  Choose this option if you have a system that has a Broadcom
+	  VC4 GPU, such as the Raspberry Pi or other BCM2708/BCM2835.
+
+	  This driver requires that "avoid_warnings=2" be present in
+	  the config.txt for the firmware, to keep it from smashing
+	  our display setup.
+
+config DRM_VC4_HDMI_CEC
+	bool "Broadcom VC4 HDMI CEC Support"
+	depends on DRM_VC4
+	select CEC_CORE
+	select DRM_DISPLAY_HDMI_CEC_HELPER
+	help
+	  Choose this option if you have a Broadcom VC4 GPU
+	  and want to use CEC.
+
+config DRM_VC4_KUNIT_TEST
+	tristate "KUnit tests for VC4" if !KUNIT_ALL_TESTS
+	depends on DRM_VC4 && KUNIT
+	select DRM_KUNIT_TEST_HELPERS
+	default KUNIT_ALL_TESTS
+	help
+	  This builds unit tests for the VC4 DRM/KMS driver. This option is
+	  not useful for distributions or general kernels, but only for kernel
+	  developers working on the VC4 driver.
+
+	  For more information on KUnit and unit tests in general,
+	  please refer to the KUnit documentation in
+	  Documentation/dev-tools/kunit/.
+
+	  If in doubt, say "N".

linux-7.2.3/drivers/gpu/drm/vc4/Makefile

@@ -0,0 +1,37 @@
+# SPDX-License-Identifier: GPL-2.0
+# Please keep these build lists sorted!
+
+# core driver code
+vc4-y := \
+	vc4_bo.o \
+	vc4_crtc.o \
+	vc4_drv.o \
+	vc4_dpi.o \
+	vc4_dsi.o \
+	vc4_fence.o \
+	vc4_kms.o \
+	vc4_gem.o \
+	vc4_hdmi.o \
+	vc4_hdmi_phy.o \
+	vc4_vec.o \
+	vc4_hvs.o \
+	vc4_irq.o \
+	vc4_perfmon.o \
+	vc4_plane.o \
+	vc4_render_cl.o \
+	vc4_trace_points.o \
+	vc4_txp.o \
+	vc4_v3d.o \
+	vc4_validate.o \
+	vc4_validate_shaders.o
+
+vc4-$(CONFIG_DRM_VC4_KUNIT_TEST) += \
+	tests/vc4_mock.o \
+	tests/vc4_mock_crtc.o \
+	tests/vc4_mock_output.o \
+	tests/vc4_mock_plane.o \
+	tests/vc4_test_pv_muxing.o
+
+vc4-$(CONFIG_DEBUG_FS) += vc4_debugfs.o
+
+obj-$(CONFIG_DRM_VC4)  += vc4.o

linux-7.2.3/drivers/gpu/drm/vc4/tests/.kunitconfig

@@ -0,0 +1,13 @@
+CONFIG_ARCH_BCM=y
+CONFIG_ARCH_BCM2835=y
+CONFIG_BCM2835_MBOX=y
+CONFIG_KUNIT=y
+CONFIG_DRM=y
+CONFIG_DRM_VC4=y
+CONFIG_DRM_VC4_KUNIT_TEST=y
+CONFIG_MAILBOX=y
+CONFIG_RASPBERRYPI_FIRMWARE=y
+CONFIG_SND=y
+CONFIG_SND_SOC=y
+CONFIG_SOUND=y
+CONFIG_COMMON_CLK=y

linux-7.2.3/drivers/gpu/drm/vc4/tests/vc4_mock.c

@@ -0,0 +1,207 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <drm/drm_drv.h>
+#include <drm/drm_kunit_helpers.h>
+
+#include <kunit/test.h>
+
+#include "vc4_mock.h"
+
+struct vc4_mock_output_desc {
+	enum vc4_encoder_type	vc4_encoder_type;
+	unsigned int		encoder_type;
+	unsigned int		connector_type;
+};
+
+#define VC4_MOCK_OUTPUT_DESC(_vc4_type, _etype, _ctype)					\
+	{										\
+		.vc4_encoder_type = _vc4_type,						\
+		.encoder_type = _etype,							\
+		.connector_type = _ctype,						\
+	}
+
+struct vc4_mock_pipe_desc {
+	const struct vc4_crtc_data *data;
+	const struct vc4_mock_output_desc *outputs;
+	unsigned int noutputs;
+};
+
+#define VC4_MOCK_CRTC_DESC(_data, ...)							\
+	{										\
+		.data = _data,								\
+		.outputs = (struct vc4_mock_output_desc[]) { __VA_ARGS__ },		\
+		.noutputs = sizeof((struct vc4_mock_output_desc[]) { __VA_ARGS__ }) /	\
+			     sizeof(struct vc4_mock_output_desc),			\
+	}
+
+#define VC4_MOCK_PIXELVALVE_DESC(_data, ...)						\
+	VC4_MOCK_CRTC_DESC(&(_data)->base, __VA_ARGS__)
+
+struct vc4_mock_desc {
+	const struct vc4_mock_pipe_desc *pipes;
+	unsigned int npipes;
+};
+
+#define VC4_MOCK_DESC(...)								\
+	{										\
+		.pipes = (struct vc4_mock_pipe_desc[]) { __VA_ARGS__ },			\
+		.npipes = sizeof((struct vc4_mock_pipe_desc[]) { __VA_ARGS__ }) /	\
+			     sizeof(struct vc4_mock_pipe_desc),				\
+	}
+
+static const struct vc4_mock_desc vc4_mock =
+	VC4_MOCK_DESC(
+		VC4_MOCK_CRTC_DESC(&bcm2835_txp_data.base,
+				   VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_TXP0,
+							DRM_MODE_ENCODER_VIRTUAL,
+							DRM_MODE_CONNECTOR_WRITEBACK)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2835_pv0_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_DSI0,
+							      DRM_MODE_ENCODER_DSI,
+							      DRM_MODE_CONNECTOR_DSI),
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_DPI,
+							      DRM_MODE_ENCODER_DPI,
+							      DRM_MODE_CONNECTOR_DPI)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2835_pv1_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_DSI1,
+							      DRM_MODE_ENCODER_DSI,
+							      DRM_MODE_CONNECTOR_DSI)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2835_pv2_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_HDMI0,
+							      DRM_MODE_ENCODER_TMDS,
+							      DRM_MODE_CONNECTOR_HDMIA),
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_VEC,
+							      DRM_MODE_ENCODER_TVDAC,
+							      DRM_MODE_CONNECTOR_Composite)),
+);
+
+static const struct vc4_mock_desc vc5_mock =
+	VC4_MOCK_DESC(
+		VC4_MOCK_CRTC_DESC(&bcm2835_txp_data.base,
+				   VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_TXP0,
+							DRM_MODE_ENCODER_VIRTUAL,
+							DRM_MODE_CONNECTOR_WRITEBACK)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2711_pv0_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_DSI0,
+							      DRM_MODE_ENCODER_DSI,
+							      DRM_MODE_CONNECTOR_DSI),
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_DPI,
+							      DRM_MODE_ENCODER_DPI,
+							      DRM_MODE_CONNECTOR_DPI)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2711_pv1_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_DSI1,
+							      DRM_MODE_ENCODER_DSI,
+							      DRM_MODE_CONNECTOR_DSI)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2711_pv2_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_HDMI0,
+							      DRM_MODE_ENCODER_TMDS,
+							      DRM_MODE_CONNECTOR_HDMIA)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2711_pv3_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_VEC,
+							      DRM_MODE_ENCODER_TVDAC,
+							      DRM_MODE_CONNECTOR_Composite)),
+		VC4_MOCK_PIXELVALVE_DESC(&bcm2711_pv4_data,
+					 VC4_MOCK_OUTPUT_DESC(VC4_ENCODER_TYPE_HDMI1,
+							      DRM_MODE_ENCODER_TMDS,
+							      DRM_MODE_CONNECTOR_HDMIA)),
+);
+
+static int __build_one_pipe(struct kunit *test, struct drm_device *drm,
+			    const struct vc4_mock_pipe_desc *pipe)
+{
+	struct drm_plane *plane;
+	struct vc4_dummy_crtc *dummy_crtc;
+	struct drm_crtc *crtc;
+	unsigned int i;
+
+	plane = vc4_dummy_plane(test, drm, DRM_PLANE_TYPE_PRIMARY);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, plane);
+
+	dummy_crtc = vc4_mock_pv(test, drm, plane, pipe->data);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, dummy_crtc);
+
+	crtc = &dummy_crtc->crtc.base;
+	for (i = 0; i < pipe->noutputs; i++) {
+		const struct vc4_mock_output_desc *mock_output = &pipe->outputs[i];
+		struct vc4_dummy_output *dummy_output;
+
+		dummy_output = vc4_dummy_output(test, drm, crtc,
+						mock_output->vc4_encoder_type,
+						mock_output->encoder_type,
+						mock_output->connector_type);
+		KUNIT_ASSERT_NOT_ERR_OR_NULL(test, dummy_output);
+	}
+
+	return 0;
+}
+
+static int __build_mock(struct kunit *test, struct drm_device *drm,
+			const struct vc4_mock_desc *mock)
+{
+	unsigned int i;
+
+	for (i = 0; i < mock->npipes; i++) {
+		const struct vc4_mock_pipe_desc *pipe = &mock->pipes[i];
+		int ret;
+
+		ret = __build_one_pipe(test, drm, pipe);
+		KUNIT_ASSERT_EQ(test, ret, 0);
+	}
+
+	return 0;
+}
+
+KUNIT_DEFINE_ACTION_WRAPPER(kunit_action_drm_dev_unregister,
+			    drm_dev_unregister,
+			    struct drm_device *);
+
+static struct vc4_dev *__mock_device(struct kunit *test, enum vc4_gen gen)
+{
+	struct drm_device *drm;
+	const struct drm_driver *drv = (gen == VC4_GEN_5) ? &vc5_drm_driver : &vc4_drm_driver;
+	const struct vc4_mock_desc *desc = (gen == VC4_GEN_5) ? &vc5_mock : &vc4_mock;
+	struct vc4_dev *vc4;
+	struct device *dev;
+	int ret;
+
+	dev = drm_kunit_helper_alloc_device(test);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, dev);
+
+	vc4 = drm_kunit_helper_alloc_drm_device_with_driver(test, dev,
+							    struct vc4_dev, base,
+							    drv);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, vc4);
+
+	vc4->dev = dev;
+	vc4->gen = gen;
+
+	vc4->hvs = __vc4_hvs_alloc(vc4, NULL, NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, vc4->hvs);
+
+	drm = &vc4->base;
+	ret = __build_mock(test, drm, desc);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = vc4_kms_load(drm);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = drm_dev_register(drm, 0);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ret = kunit_add_action_or_reset(test,
+					kunit_action_drm_dev_unregister,
+					drm);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	return vc4;
+}
+
+struct vc4_dev *vc4_mock_device(struct kunit *test)
+{
+	return __mock_device(test, VC4_GEN_4);
+}
+
+struct vc4_dev *vc5_mock_device(struct kunit *test)
+{
+	return __mock_device(test, VC4_GEN_5);
+}

linux-7.2.3/drivers/gpu/drm/vc4/tests/vc4_mock.h

@@ -0,0 +1,61 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef VC4_MOCK_H_
+#define VC4_MOCK_H_
+
+#include "../vc4_drv.h"
+
+static inline
+struct drm_crtc *vc4_find_crtc_for_encoder(struct kunit *test,
+					   struct drm_device *drm,
+					   struct drm_encoder *encoder)
+{
+	struct drm_crtc *crtc;
+
+	KUNIT_ASSERT_EQ(test, hweight32(encoder->possible_crtcs), 1);
+
+	drm_for_each_crtc(crtc, drm)
+		if (encoder->possible_crtcs & drm_crtc_mask(crtc))
+			return crtc;
+
+	return NULL;
+}
+
+struct drm_plane *vc4_dummy_plane(struct kunit *test, struct drm_device *drm,
+				  enum drm_plane_type type);
+
+struct vc4_dummy_crtc {
+	struct vc4_crtc crtc;
+};
+
+struct vc4_dummy_crtc *vc4_mock_pv(struct kunit *test,
+				   struct drm_device *drm,
+				   struct drm_plane *plane,
+				   const struct vc4_crtc_data *data);
+
+struct vc4_dummy_output {
+	struct vc4_encoder encoder;
+	struct drm_connector connector;
+};
+
+#define encoder_to_vc4_dummy_output(_enc)				\
+	container_of_const(_enc, struct vc4_dummy_output, encoder.base)
+
+struct vc4_dummy_output *vc4_dummy_output(struct kunit *test,
+					  struct drm_device *drm,
+					  struct drm_crtc *crtc,
+					  enum vc4_encoder_type vc4_encoder_type,
+					  unsigned int kms_encoder_type,
+					  unsigned int connector_type);
+
+struct vc4_dev *vc4_mock_device(struct kunit *test);
+struct vc4_dev *vc5_mock_device(struct kunit *test);
+
+int vc4_mock_atomic_add_output(struct kunit *test,
+			       struct drm_atomic_commit *state,
+			       enum vc4_encoder_type type);
+int vc4_mock_atomic_del_output(struct kunit *test,
+			       struct drm_atomic_commit *state,
+			       enum vc4_encoder_type type);
+
+#endif // VC4_MOCK_H_

linux-7.2.3/drivers/gpu/drm/vc4/tests/vc4_mock_crtc.c

@@ -0,0 +1,41 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <drm/drm_atomic_state_helper.h>
+#include <drm/drm_modeset_helper_vtables.h>
+
+#include <kunit/test.h>
+
+#include "vc4_mock.h"
+
+static const struct drm_crtc_helper_funcs vc4_dummy_crtc_helper_funcs = {
+	.atomic_check	= vc4_crtc_atomic_check,
+};
+
+static const struct drm_crtc_funcs vc4_dummy_crtc_funcs = {
+	.atomic_destroy_state	= vc4_crtc_destroy_state,
+	.atomic_duplicate_state	= vc4_crtc_duplicate_state,
+	.reset			= vc4_crtc_reset,
+};
+
+struct vc4_dummy_crtc *vc4_mock_pv(struct kunit *test,
+				   struct drm_device *drm,
+				   struct drm_plane *plane,
+				   const struct vc4_crtc_data *data)
+{
+	struct vc4_dummy_crtc *dummy_crtc;
+	struct vc4_crtc *vc4_crtc;
+	int ret;
+
+	dummy_crtc = drmm_kzalloc(drm, sizeof(*dummy_crtc), GFP_KERNEL);
+	KUNIT_ASSERT_NOT_NULL(test, dummy_crtc);
+
+	vc4_crtc = &dummy_crtc->crtc;
+	ret = __vc4_crtc_init(drm, NULL,
+			      vc4_crtc, data, plane,
+			      &vc4_dummy_crtc_funcs,
+			      &vc4_dummy_crtc_helper_funcs,
+			      false);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	return dummy_crtc;
+}

linux-7.2.3/drivers/gpu/drm/vc4/tests/vc4_mock_output.c

@@ -0,0 +1,176 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <drm/drm_atomic_state_helper.h>
+#include <drm/drm_atomic_uapi.h>
+#include <drm/drm_connector.h>
+#include <drm/drm_crtc.h>
+#include <drm/drm_encoder.h>
+#include <drm/drm_modeset_helper_vtables.h>
+
+#include <kunit/test.h>
+
+#include "vc4_mock.h"
+
+static const struct drm_connector_helper_funcs vc4_dummy_connector_helper_funcs = {
+};
+
+static const struct drm_connector_funcs vc4_dummy_connector_funcs = {
+	.atomic_destroy_state	= drm_atomic_helper_connector_destroy_state,
+	.atomic_duplicate_state	= drm_atomic_helper_connector_duplicate_state,
+	.reset			= drm_atomic_helper_connector_reset,
+};
+
+struct vc4_dummy_output *vc4_dummy_output(struct kunit *test,
+					  struct drm_device *drm,
+					  struct drm_crtc *crtc,
+					  enum vc4_encoder_type vc4_encoder_type,
+					  unsigned int kms_encoder_type,
+					  unsigned int connector_type)
+{
+	struct vc4_dummy_output *dummy_output;
+	struct drm_connector *conn;
+	struct drm_encoder *enc;
+	int ret;
+
+	dummy_output = drmm_kzalloc(drm, sizeof(*dummy_output), GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, dummy_output);
+	dummy_output->encoder.type = vc4_encoder_type;
+
+	enc = &dummy_output->encoder.base;
+	ret = drmm_encoder_init(drm, enc,
+				NULL,
+				kms_encoder_type,
+				NULL);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+	enc->possible_crtcs = drm_crtc_mask(crtc);
+
+	conn = &dummy_output->connector;
+	ret = drmm_connector_init(drm, conn,
+				  &vc4_dummy_connector_funcs,
+				  connector_type,
+				  NULL);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	drm_connector_helper_add(conn, &vc4_dummy_connector_helper_funcs);
+	drm_connector_attach_encoder(conn, enc);
+
+	return dummy_output;
+}
+
+static const struct drm_display_mode default_mode = {
+	DRM_SIMPLE_MODE(640, 480, 64, 48)
+};
+
+/**
+ * vc4_mock_atomic_add_output() - Enables an output in a state
+ * @test: The test context object
+ * @state: Atomic state to enable the output in.
+ * @type: Type of the output encoder
+ *
+ * Adds an output CRTC and connector to a state, and enables them.
+ *
+ * Returns:
+ * 0 on success, a negative error code on failure. If the error is
+ * EDEADLK, the entire atomic sequence must be restarted. All other
+ * errors are fatal.
+ */
+int vc4_mock_atomic_add_output(struct kunit *test,
+			       struct drm_atomic_commit *state,
+			       enum vc4_encoder_type type)
+{
+	struct drm_device *drm = state->dev;
+	struct drm_connector_state *conn_state;
+	struct drm_crtc_state *crtc_state;
+	struct vc4_dummy_output *output;
+	struct drm_connector *conn;
+	struct drm_encoder *encoder;
+	struct drm_crtc *crtc;
+	int ret;
+
+	encoder = vc4_find_encoder_by_type(drm, type);
+	if (!encoder)
+		return -ENODEV;
+
+	crtc = vc4_find_crtc_for_encoder(test, drm, encoder);
+	if (!crtc)
+		return -ENODEV;
+
+	output = encoder_to_vc4_dummy_output(encoder);
+	conn = &output->connector;
+	conn_state = drm_atomic_get_connector_state(state, conn);
+	if (IS_ERR(conn_state))
+		return PTR_ERR(conn_state);
+
+	ret = drm_atomic_set_crtc_for_connector(conn_state, crtc);
+	if (ret)
+		return ret;
+
+	crtc_state = drm_atomic_get_crtc_state(state, crtc);
+	if (IS_ERR(crtc_state))
+		return PTR_ERR(crtc_state);
+
+	ret = drm_atomic_set_mode_for_crtc(crtc_state, &default_mode);
+	if (ret)
+		return ret;
+
+	crtc_state->active = true;
+
+	return 0;
+}
+
+/**
+ * vc4_mock_atomic_del_output() - Disables an output in a state
+ * @test: The test context object
+ * @state: Atomic state to disable the output in.
+ * @type: Type of the output encoder
+ *
+ * Adds an output CRTC and connector to a state, and disables them.
+ *
+ * Returns:
+ * 0 on success, a negative error code on failure. If the error is
+ * EDEADLK, the entire atomic sequence must be restarted. All other
+ * errors are fatal.
+ */
+int vc4_mock_atomic_del_output(struct kunit *test,
+			       struct drm_atomic_commit *state,
+			       enum vc4_encoder_type type)
+{
+	struct drm_device *drm = state->dev;
+	struct drm_connector_state *conn_state;
+	struct drm_crtc_state *crtc_state;
+	struct vc4_dummy_output *output;
+	struct drm_connector *conn;
+	struct drm_encoder *encoder;
+	struct drm_crtc *crtc;
+	int ret;
+
+	encoder = vc4_find_encoder_by_type(drm, type);
+	if (!encoder)
+		return -ENODEV;
+
+	crtc = vc4_find_crtc_for_encoder(test, drm, encoder);
+	if (!crtc)
+		return -ENODEV;
+
+	crtc_state = drm_atomic_get_crtc_state(state, crtc);
+	if (IS_ERR(crtc_state))
+		return PTR_ERR(crtc_state);
+
+	crtc_state->active = false;
+
+	ret = drm_atomic_set_mode_for_crtc(crtc_state, NULL);
+	if (ret)
+		return ret;
+
+	output = encoder_to_vc4_dummy_output(encoder);
+	conn = &output->connector;
+	conn_state = drm_atomic_get_connector_state(state, conn);
+	if (IS_ERR(conn_state))
+		return PTR_ERR(conn_state);
+
+	ret = drm_atomic_set_crtc_for_connector(conn_state, NULL);
+	if (ret)
+		return ret;
+
+	return 0;
+}

linux-7.2.3/drivers/gpu/drm/vc4/tests/vc4_mock_plane.c

@@ -0,0 +1,25 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <drm/drm_kunit_helpers.h>
+#include <drm/drm_plane.h>
+
+#include <kunit/test.h>
+
+#include "vc4_mock.h"
+
+struct drm_plane *vc4_dummy_plane(struct kunit *test, struct drm_device *drm,
+				  enum drm_plane_type type)
+{
+	struct drm_plane *plane;
+
+	KUNIT_ASSERT_EQ(test, type, DRM_PLANE_TYPE_PRIMARY);
+
+	plane = drm_kunit_helper_create_primary_plane(test, drm,
+						      NULL,
+						      NULL,
+						      NULL, 0,
+						      NULL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, plane);
+
+	return plane;
+}

linux-7.2.3/drivers/gpu/drm/vc4/vc4_debugfs.c

@@ -0,0 +1,59 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ *  Copyright © 2014 Broadcom
+ */
+
+#include <drm/drm_drv.h>
+#include <drm/drm_print.h>
+
+#include <linux/seq_file.h>
+#include <linux/circ_buf.h>
+#include <linux/ctype.h>
+#include <linux/debugfs.h>
+#include <linux/platform_device.h>
+
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+
+/*
+ * Called at drm_dev_register() time on each of the minors registered
+ * by the DRM device, to attach the debugfs files.
+ */
+void
+vc4_debugfs_init(struct drm_minor *minor)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(minor->dev);
+	struct drm_device *drm = &vc4->base;
+
+	drm_WARN_ON(drm, vc4_hvs_debugfs_init(minor));
+
+	if (vc4->v3d) {
+		drm_WARN_ON(drm, vc4_bo_debugfs_init(minor));
+		drm_WARN_ON(drm, vc4_v3d_debugfs_init(minor));
+	}
+}
+
+static int vc4_debugfs_regset32(struct seq_file *m, void *unused)
+{
+	struct drm_debugfs_entry *entry = m->private;
+	struct drm_device *drm = entry->dev;
+	struct debugfs_regset32 *regset = entry->file.data;
+	struct drm_printer p = drm_seq_file_printer(m);
+	int idx;
+
+	if (!drm_dev_enter(drm, &idx))
+		return -ENODEV;
+
+	drm_print_regset32(&p, regset);
+
+	drm_dev_exit(idx);
+
+	return 0;
+}
+
+void vc4_debugfs_add_regset32(struct drm_device *drm,
+			      const char *name,
+			      struct debugfs_regset32 *regset)
+{
+	drm_debugfs_add_file(drm, name, vc4_debugfs_regset32, regset);
+}

linux-7.2.3/drivers/gpu/drm/vc4/vc4_dpi.c

@@ -0,0 +1,403 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2016 Broadcom Limited
+ */
+
+/**
+ * DOC: VC4 DPI module
+ *
+ * The VC4 DPI hardware supports MIPI DPI type 4 and Nokia ViSSI
+ * signals.  On BCM2835, these can be routed out to GPIO0-27 with the
+ * ALT2 function.
+ */
+
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_bridge.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_of.h>
+#include <drm/drm_panel.h>
+#include <drm/drm_print.h>
+#include <drm/drm_probe_helper.h>
+#include <drm/drm_simple_kms_helper.h>
+#include <linux/clk.h>
+#include <linux/component.h>
+#include <linux/media-bus-format.h>
+#include <linux/platform_device.h>
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+
+#define DPI_C			0x00
+# define DPI_OUTPUT_ENABLE_MODE		BIT(16)
+
+/* The order field takes the incoming 24 bit RGB from the pixel valve
+ * and shuffles the 3 channels.
+ */
+# define DPI_ORDER_MASK			VC4_MASK(15, 14)
+# define DPI_ORDER_SHIFT		14
+# define DPI_ORDER_RGB			0
+# define DPI_ORDER_BGR			1
+# define DPI_ORDER_GRB			2
+# define DPI_ORDER_BRG			3
+
+/* The format field takes the ORDER-shuffled pixel valve data and
+ * formats it onto the output lines.
+ */
+# define DPI_FORMAT_MASK		VC4_MASK(13, 11)
+# define DPI_FORMAT_SHIFT		11
+/* This define is named in the hardware, but actually just outputs 0. */
+# define DPI_FORMAT_9BIT_666_RGB	0
+/* Outputs 00000000rrrrrggggggbbbbb */
+# define DPI_FORMAT_16BIT_565_RGB_1	1
+/* Outputs 000rrrrr00gggggg000bbbbb */
+# define DPI_FORMAT_16BIT_565_RGB_2	2
+/* Outputs 00rrrrr000gggggg00bbbbb0 */
+# define DPI_FORMAT_16BIT_565_RGB_3	3
+/* Outputs 000000rrrrrrggggggbbbbbb */
+# define DPI_FORMAT_18BIT_666_RGB_1	4
+/* Outputs 00rrrrrr00gggggg00bbbbbb */
+# define DPI_FORMAT_18BIT_666_RGB_2	5
+/* Outputs rrrrrrrrggggggggbbbbbbbb */
+# define DPI_FORMAT_24BIT_888_RGB	6
+
+/* Reverses the polarity of the corresponding signal */
+# define DPI_PIXEL_CLK_INVERT		BIT(10)
+# define DPI_HSYNC_INVERT		BIT(9)
+# define DPI_VSYNC_INVERT		BIT(8)
+# define DPI_OUTPUT_ENABLE_INVERT	BIT(7)
+
+/* Outputs the signal the falling clock edge instead of rising. */
+# define DPI_HSYNC_NEGATE		BIT(6)
+# define DPI_VSYNC_NEGATE		BIT(5)
+# define DPI_OUTPUT_ENABLE_NEGATE	BIT(4)
+
+/* Disables the signal */
+# define DPI_HSYNC_DISABLE		BIT(3)
+# define DPI_VSYNC_DISABLE		BIT(2)
+# define DPI_OUTPUT_ENABLE_DISABLE	BIT(1)
+
+/* Power gate to the device, full reset at 0 -> 1 transition */
+# define DPI_ENABLE			BIT(0)
+
+/* All other registers besides DPI_C return the ID */
+#define DPI_ID			0x04
+# define DPI_ID_VALUE		0x00647069
+
+/* General DPI hardware state. */
+struct vc4_dpi {
+	struct vc4_encoder encoder;
+
+	struct platform_device *pdev;
+
+	void __iomem *regs;
+
+	struct clk *pixel_clock;
+	struct clk *core_clock;
+
+	struct debugfs_regset32 regset;
+};
+
+#define to_vc4_dpi(_encoder)						\
+	container_of_const(_encoder, struct vc4_dpi, encoder.base)
+
+#define DPI_READ(offset)								\
+	({										\
+		kunit_fail_current_test("Accessing a register in a unit test!\n");	\
+		readl(dpi->regs + (offset));						\
+	})
+
+#define DPI_WRITE(offset, val)								\
+	do {										\
+		kunit_fail_current_test("Accessing a register in a unit test!\n");	\
+		writel(val, dpi->regs + (offset));					\
+	} while (0)
+
+static const struct debugfs_reg32 dpi_regs[] = {
+	VC4_REG32(DPI_C),
+	VC4_REG32(DPI_ID),
+};
+
+static void vc4_dpi_encoder_disable(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct vc4_dpi *dpi = to_vc4_dpi(encoder);
+	int idx;
+
+	if (!drm_dev_enter(dev, &idx))
+		return;
+
+	clk_disable_unprepare(dpi->pixel_clock);
+
+	drm_dev_exit(idx);
+}
+
+static void vc4_dpi_encoder_enable(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct drm_display_mode *mode = &encoder->crtc->mode;
+	struct vc4_dpi *dpi = to_vc4_dpi(encoder);
+	struct drm_connector_list_iter conn_iter;
+	struct drm_connector *connector = NULL, *connector_scan;
+	u32 dpi_c = DPI_ENABLE;
+	int idx;
+	int ret;
+
+	/* Look up the connector attached to DPI so we can get the
+	 * bus_format.  Ideally the bridge would tell us the
+	 * bus_format we want, but it doesn't yet, so assume that it's
+	 * uniform throughout the bridge chain.
+	 */
+	drm_connector_list_iter_begin(dev, &conn_iter);
+	drm_for_each_connector_iter(connector_scan, &conn_iter) {
+		if (connector_scan->encoder == encoder) {
+			connector = connector_scan;
+			break;
+		}
+	}
+	drm_connector_list_iter_end(&conn_iter);
+
+	/* Default to 18bit if no connector or format found. */
+	dpi_c |= VC4_SET_FIELD(DPI_FORMAT_18BIT_666_RGB_1, DPI_FORMAT);
+
+	if (connector) {
+		if (connector->display_info.num_bus_formats) {
+			u32 bus_format = connector->display_info.bus_formats[0];
+
+			dpi_c &= ~DPI_FORMAT_MASK;
+
+			switch (bus_format) {
+			case MEDIA_BUS_FMT_RGB888_1X24:
+				dpi_c |= VC4_SET_FIELD(DPI_FORMAT_24BIT_888_RGB,
+						       DPI_FORMAT);
+				break;
+			case MEDIA_BUS_FMT_BGR888_1X24:
+				dpi_c |= VC4_SET_FIELD(DPI_FORMAT_24BIT_888_RGB,
+						       DPI_FORMAT);
+				dpi_c |= VC4_SET_FIELD(DPI_ORDER_BGR,
+						       DPI_ORDER);
+				break;
+			case MEDIA_BUS_FMT_BGR666_1X24_CPADHI:
+				dpi_c |= VC4_SET_FIELD(DPI_ORDER_BGR, DPI_ORDER);
+				fallthrough;
+			case MEDIA_BUS_FMT_RGB666_1X24_CPADHI:
+				dpi_c |= VC4_SET_FIELD(DPI_FORMAT_18BIT_666_RGB_2,
+						       DPI_FORMAT);
+				break;
+			case MEDIA_BUS_FMT_BGR666_1X18:
+				dpi_c |= VC4_SET_FIELD(DPI_ORDER_BGR, DPI_ORDER);
+				fallthrough;
+			case MEDIA_BUS_FMT_RGB666_1X18:
+				dpi_c |= VC4_SET_FIELD(DPI_FORMAT_18BIT_666_RGB_1,
+						       DPI_FORMAT);
+				break;
+			case MEDIA_BUS_FMT_RGB565_1X16:
+				dpi_c |= VC4_SET_FIELD(DPI_FORMAT_16BIT_565_RGB_1,
+						       DPI_FORMAT);
+				break;
+			case MEDIA_BUS_FMT_RGB565_1X24_CPADHI:
+				dpi_c |= VC4_SET_FIELD(DPI_FORMAT_16BIT_565_RGB_2,
+						       DPI_FORMAT);
+				break;
+			default:
+				drm_err(dev, "Unknown media bus format %d\n",
+					bus_format);
+				break;
+			}
+		}
+
+		if (connector->display_info.bus_flags & DRM_BUS_FLAG_PIXDATA_DRIVE_NEGEDGE)
+			dpi_c |= DPI_PIXEL_CLK_INVERT;
+
+		if (connector->display_info.bus_flags & DRM_BUS_FLAG_DE_LOW)
+			dpi_c |= DPI_OUTPUT_ENABLE_INVERT;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_CSYNC) {
+		if (mode->flags & DRM_MODE_FLAG_NCSYNC)
+			dpi_c |= DPI_OUTPUT_ENABLE_INVERT;
+	} else {
+		dpi_c |= DPI_OUTPUT_ENABLE_MODE;
+
+		if (mode->flags & DRM_MODE_FLAG_NHSYNC)
+			dpi_c |= DPI_HSYNC_INVERT;
+		else if (!(mode->flags & DRM_MODE_FLAG_PHSYNC))
+			dpi_c |= DPI_HSYNC_DISABLE;
+
+		if (mode->flags & DRM_MODE_FLAG_NVSYNC)
+			dpi_c |= DPI_VSYNC_INVERT;
+		else if (!(mode->flags & DRM_MODE_FLAG_PVSYNC))
+			dpi_c |= DPI_VSYNC_DISABLE;
+	}
+
+	if (!drm_dev_enter(dev, &idx))
+		return;
+
+	DPI_WRITE(DPI_C, dpi_c);
+
+	ret = clk_set_rate(dpi->pixel_clock, mode->clock * 1000);
+	if (ret)
+		drm_err(dev, "Failed to set clock rate: %d\n", ret);
+
+	ret = clk_prepare_enable(dpi->pixel_clock);
+	if (ret)
+		drm_err(dev, "Failed to set clock rate: %d\n", ret);
+
+	drm_dev_exit(idx);
+}
+
+static enum drm_mode_status vc4_dpi_encoder_mode_valid(struct drm_encoder *encoder,
+						       const struct drm_display_mode *mode)
+{
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		return MODE_NO_INTERLACE;
+
+	return MODE_OK;
+}
+
+static const struct drm_encoder_helper_funcs vc4_dpi_encoder_helper_funcs = {
+	.disable = vc4_dpi_encoder_disable,
+	.enable = vc4_dpi_encoder_enable,
+	.mode_valid = vc4_dpi_encoder_mode_valid,
+};
+
+static int vc4_dpi_late_register(struct drm_encoder *encoder)
+{
+	struct drm_device *drm = encoder->dev;
+	struct vc4_dpi *dpi = to_vc4_dpi(encoder);
+
+	vc4_debugfs_add_regset32(drm, "dpi_regs", &dpi->regset);
+
+	return 0;
+}
+
+static const struct drm_encoder_funcs vc4_dpi_encoder_funcs = {
+	.late_register = vc4_dpi_late_register,
+};
+
+static const struct of_device_id vc4_dpi_dt_match[] = {
+	{ .compatible = "brcm,bcm2835-dpi", .data = NULL },
+	{}
+};
+
+/* Sets up the next link in the display chain, whether it's a panel or
+ * a bridge.
+ */
+static int vc4_dpi_init_bridge(struct vc4_dpi *dpi)
+{
+	struct drm_device *drm = dpi->encoder.base.dev;
+	struct device *dev = &dpi->pdev->dev;
+	struct drm_bridge *bridge;
+
+	bridge = drmm_of_get_bridge(drm, dev->of_node, 0, 0);
+	if (IS_ERR(bridge)) {
+		/* If nothing was connected in the DT, that's not an
+		 * error.
+		 */
+		if (PTR_ERR(bridge) == -ENODEV)
+			return 0;
+		else
+			return PTR_ERR(bridge);
+	}
+
+	return drm_bridge_attach(&dpi->encoder.base, bridge, NULL, 0);
+}
+
+static void vc4_dpi_disable_clock(void *ptr)
+{
+	struct vc4_dpi *dpi = ptr;
+
+	clk_disable_unprepare(dpi->core_clock);
+}
+
+static int vc4_dpi_bind(struct device *dev, struct device *master, void *data)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct drm_device *drm = dev_get_drvdata(master);
+	struct vc4_dpi *dpi;
+	int ret;
+
+	dpi = drmm_kzalloc(drm, sizeof(*dpi), GFP_KERNEL);
+	if (!dpi)
+		return -ENOMEM;
+
+	dpi->encoder.type = VC4_ENCODER_TYPE_DPI;
+	dpi->pdev = pdev;
+	dpi->regs = vc4_ioremap_regs(pdev, 0);
+	if (IS_ERR(dpi->regs))
+		return PTR_ERR(dpi->regs);
+	dpi->regset.base = dpi->regs;
+	dpi->regset.regs = dpi_regs;
+	dpi->regset.nregs = ARRAY_SIZE(dpi_regs);
+
+	if (DPI_READ(DPI_ID) != DPI_ID_VALUE) {
+		dev_err(dev, "Port returned 0x%08x for ID instead of 0x%08x\n",
+			DPI_READ(DPI_ID), DPI_ID_VALUE);
+		return -ENODEV;
+	}
+
+	dpi->core_clock = devm_clk_get(dev, "core");
+	if (IS_ERR(dpi->core_clock)) {
+		ret = PTR_ERR(dpi->core_clock);
+		if (ret != -EPROBE_DEFER)
+			drm_err(drm, "Failed to get core clock: %d\n", ret);
+		return ret;
+	}
+
+	dpi->pixel_clock = devm_clk_get(dev, "pixel");
+	if (IS_ERR(dpi->pixel_clock)) {
+		ret = PTR_ERR(dpi->pixel_clock);
+		if (ret != -EPROBE_DEFER)
+			drm_err(drm, "Failed to get pixel clock: %d\n", ret);
+		return ret;
+	}
+
+	ret = clk_prepare_enable(dpi->core_clock);
+	if (ret) {
+		drm_err(drm, "Failed to turn on core clock: %d\n", ret);
+		return ret;
+	}
+
+	ret = devm_add_action_or_reset(dev, vc4_dpi_disable_clock, dpi);
+	if (ret)
+		return ret;
+
+	ret = drmm_encoder_init(drm, &dpi->encoder.base,
+				&vc4_dpi_encoder_funcs,
+				DRM_MODE_ENCODER_DPI,
+				NULL);
+	if (ret)
+		return ret;
+
+	drm_encoder_helper_add(&dpi->encoder.base, &vc4_dpi_encoder_helper_funcs);
+
+	ret = vc4_dpi_init_bridge(dpi);
+	if (ret)
+		return ret;
+
+	dev_set_drvdata(dev, dpi);
+
+	return 0;
+}
+
+static const struct component_ops vc4_dpi_ops = {
+	.bind   = vc4_dpi_bind,
+};
+
+static int vc4_dpi_dev_probe(struct platform_device *pdev)
+{
+	return component_add(&pdev->dev, &vc4_dpi_ops);
+}
+
+static void vc4_dpi_dev_remove(struct platform_device *pdev)
+{
+	component_del(&pdev->dev, &vc4_dpi_ops);
+}
+
+struct platform_driver vc4_dpi_driver = {
+	.probe = vc4_dpi_dev_probe,
+	.remove = vc4_dpi_dev_remove,
+	.driver = {
+		.name = "vc4_dpi",
+		.of_match_table = vc4_dpi_dt_match,
+	},
+};

linux-7.2.3/drivers/gpu/drm/vc4/vc4_drv.c

@@ -0,0 +1,517 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2014-2015 Broadcom
+ * Copyright (C) 2013 Red Hat
+ */
+
+/**
+ * DOC: Broadcom VC4 Graphics Driver
+ *
+ * The Broadcom VideoCore 4 (present in the Raspberry Pi) contains a
+ * OpenGL ES 2.0-compatible 3D engine called V3D, and a highly
+ * configurable display output pipeline that supports HDMI, DSI, DPI,
+ * and Composite TV output.
+ *
+ * The 3D engine also has an interface for submitting arbitrary
+ * compute shader-style jobs using the same shader processor as is
+ * used for vertex and fragment shaders in GLES 2.0.  However, given
+ * that the hardware isn't able to expose any standard interfaces like
+ * OpenGL compute shaders or OpenCL, it isn't supported by this
+ * driver.
+ */
+
+#include <linux/aperture.h>
+#include <linux/clk.h>
+#include <linux/component.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/clients/drm_client_setup.h>
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_fbdev_dma.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_print.h>
+#include <drm/drm_vblank.h>
+
+#include <soc/bcm2835/raspberrypi-firmware.h>
+
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+
+#define DRIVER_NAME "vc4"
+#define DRIVER_DESC "Broadcom VC4 graphics"
+#define DRIVER_MAJOR 0
+#define DRIVER_MINOR 0
+#define DRIVER_PATCHLEVEL 0
+
+/* Helper function for mapping the regs on a platform device. */
+void __iomem *vc4_ioremap_regs(struct platform_device *pdev, int index)
+{
+	void __iomem *map;
+
+	map = devm_platform_ioremap_resource(pdev, index);
+	if (IS_ERR(map))
+		return map;
+
+	return map;
+}
+
+int vc4_dumb_fixup_args(struct drm_mode_create_dumb *args)
+{
+	int min_pitch = DIV_ROUND_UP(args->width * args->bpp, 8);
+
+	if (args->pitch < min_pitch)
+		args->pitch = min_pitch;
+
+	if (args->size < args->pitch * args->height)
+		args->size = args->pitch * args->height;
+
+	return 0;
+}
+
+static int vc5_dumb_create(struct drm_file *file_priv,
+			   struct drm_device *dev,
+			   struct drm_mode_create_dumb *args)
+{
+	int ret;
+
+	ret = vc4_dumb_fixup_args(args);
+	if (ret)
+		return ret;
+
+	return drm_gem_dma_dumb_create_internal(file_priv, dev, args);
+}
+
+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;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	if (!vc4->v3d)
+		return -ENODEV;
+
+	switch (args->param) {
+	case DRM_VC4_PARAM_V3D_IDENT0:
+		ret = vc4_v3d_pm_get(vc4);
+		if (ret)
+			return ret;
+		args->value = V3D_READ(V3D_IDENT0);
+		vc4_v3d_pm_put(vc4);
+		break;
+	case DRM_VC4_PARAM_V3D_IDENT1:
+		ret = vc4_v3d_pm_get(vc4);
+		if (ret)
+			return ret;
+		args->value = V3D_READ(V3D_IDENT1);
+		vc4_v3d_pm_put(vc4);
+		break;
+	case DRM_VC4_PARAM_V3D_IDENT2:
+		ret = vc4_v3d_pm_get(vc4);
+		if (ret)
+			return ret;
+		args->value = V3D_READ(V3D_IDENT2);
+		vc4_v3d_pm_put(vc4);
+		break;
+	case DRM_VC4_PARAM_SUPPORTS_BRANCHES:
+	case DRM_VC4_PARAM_SUPPORTS_ETC1:
+	case DRM_VC4_PARAM_SUPPORTS_THREADED_FS:
+	case DRM_VC4_PARAM_SUPPORTS_FIXED_RCL_ORDER:
+	case DRM_VC4_PARAM_SUPPORTS_MADVISE:
+	case DRM_VC4_PARAM_SUPPORTS_PERFMON:
+		args->value = true;
+		break;
+	default:
+		DRM_DEBUG("Unknown parameter %d\n", args->param);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int vc4_open(struct drm_device *dev, struct drm_file *file)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_file *vc4file;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	vc4file = kzalloc_obj(*vc4file);
+	if (!vc4file)
+		return -ENOMEM;
+	vc4file->dev = vc4;
+
+	vc4_perfmon_open_file(vc4file);
+	file->driver_priv = vc4file;
+	return 0;
+}
+
+static void vc4_close(struct drm_device *dev, struct drm_file *file)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_file *vc4file = file->driver_priv;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	if (vc4file->bin_bo_used)
+		vc4_v3d_bin_bo_put(vc4);
+
+	vc4_perfmon_close_file(vc4file);
+	kfree(vc4file);
+}
+
+DEFINE_DRM_GEM_FOPS(vc4_drm_fops);
+
+static const struct drm_ioctl_desc vc4_drm_ioctls[] = {
+	DRM_IOCTL_DEF_DRV(VC4_SUBMIT_CL, vc4_submit_cl_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_WAIT_SEQNO, vc4_wait_seqno_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_WAIT_BO, vc4_wait_bo_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_CREATE_BO, vc4_create_bo_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_MMAP_BO, vc4_mmap_bo_ioctl, DRM_RENDER_ALLOW),
+	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),
+	DRM_IOCTL_DEF_DRV(VC4_SET_TILING, vc4_set_tiling_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_GET_TILING, vc4_get_tiling_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_LABEL_BO, vc4_label_bo_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_GEM_MADVISE, vc4_gem_madvise_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_PERFMON_CREATE, vc4_perfmon_create_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_PERFMON_DESTROY, vc4_perfmon_destroy_ioctl, DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(VC4_PERFMON_GET_VALUES, vc4_perfmon_get_values_ioctl, DRM_RENDER_ALLOW),
+};
+
+const struct drm_driver vc4_drm_driver = {
+	.driver_features = (DRIVER_MODESET |
+			    DRIVER_ATOMIC |
+			    DRIVER_GEM |
+			    DRIVER_RENDER |
+			    DRIVER_SYNCOBJ),
+	.open = vc4_open,
+	.postclose = vc4_close,
+
+#if defined(CONFIG_DEBUG_FS)
+	.debugfs_init = vc4_debugfs_init,
+#endif
+
+	.gem_create_object = vc4_create_object,
+
+	DRM_GEM_DMA_DRIVER_OPS_WITH_DUMB_CREATE(vc4_bo_dumb_create),
+	DRM_FBDEV_DMA_DRIVER_OPS,
+
+	.ioctls = vc4_drm_ioctls,
+	.num_ioctls = ARRAY_SIZE(vc4_drm_ioctls),
+	.fops = &vc4_drm_fops,
+
+	.name = DRIVER_NAME,
+	.desc = DRIVER_DESC,
+	.major = DRIVER_MAJOR,
+	.minor = DRIVER_MINOR,
+	.patchlevel = DRIVER_PATCHLEVEL,
+};
+
+const struct drm_driver vc5_drm_driver = {
+	.driver_features = (DRIVER_MODESET |
+			    DRIVER_ATOMIC |
+			    DRIVER_GEM),
+
+#if defined(CONFIG_DEBUG_FS)
+	.debugfs_init = vc4_debugfs_init,
+#endif
+
+	DRM_GEM_DMA_DRIVER_OPS_WITH_DUMB_CREATE(vc5_dumb_create),
+	DRM_FBDEV_DMA_DRIVER_OPS,
+
+	.fops = &vc4_drm_fops,
+
+	.name = DRIVER_NAME,
+	.desc = DRIVER_DESC,
+	.major = DRIVER_MAJOR,
+	.minor = DRIVER_MINOR,
+	.patchlevel = DRIVER_PATCHLEVEL,
+};
+
+static void vc4_match_add_drivers(struct device *dev,
+				  struct component_match **match,
+				  struct platform_driver *const *drivers,
+				  int count)
+{
+	int i;
+
+	for (i = 0; i < count; i++) {
+		struct device_driver *drv = &drivers[i]->driver;
+		struct device *p = NULL, *d;
+
+		while ((d = platform_find_device_by_driver(p, drv))) {
+			put_device(p);
+			component_match_add(dev, match, component_compare_dev, d);
+			p = d;
+		}
+		put_device(p);
+	}
+}
+
+static void vc4_component_unbind_all(void *ptr)
+{
+	struct vc4_dev *vc4 = ptr;
+
+	component_unbind_all(vc4->dev, &vc4->base);
+}
+
+static const struct of_device_id vc4_dma_range_matches[] = {
+	{ .compatible = "brcm,bcm2711-hvs" },
+	{ .compatible = "brcm,bcm2712-hvs" },
+	{ .compatible = "brcm,bcm2835-hvs" },
+	{ .compatible = "brcm,bcm2835-v3d" },
+	{ .compatible = "brcm,cygnus-v3d" },
+	{ .compatible = "brcm,vc4-v3d" },
+	{}
+};
+
+static int vc4_drm_bind(struct device *dev)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	const struct drm_driver *driver;
+	struct rpi_firmware *firmware = NULL;
+	struct drm_device *drm;
+	struct vc4_dev *vc4;
+	struct device_node *node;
+	struct drm_crtc *crtc;
+	enum vc4_gen gen;
+	int ret = 0;
+
+	dev->coherent_dma_mask = DMA_BIT_MASK(32);
+
+	gen = (enum vc4_gen)of_device_get_match_data(dev);
+
+	if (gen > VC4_GEN_4)
+		driver = &vc5_drm_driver;
+	else
+		driver = &vc4_drm_driver;
+
+	if (gen >= VC4_GEN_6_C)
+		dma_set_mask_and_coherent(dev, DMA_BIT_MASK(36));
+	else
+		dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));
+
+	node = of_find_matching_node_and_match(NULL, vc4_dma_range_matches,
+					       NULL);
+	if (node) {
+		ret = of_dma_configure(dev, node, true);
+		of_node_put(node);
+
+		if (ret)
+			return ret;
+	}
+
+	vc4 = devm_drm_dev_alloc(dev, driver, struct vc4_dev, base);
+	if (IS_ERR(vc4))
+		return PTR_ERR(vc4);
+	vc4->gen = gen;
+	vc4->dev = dev;
+
+	drm = &vc4->base;
+	platform_set_drvdata(pdev, drm);
+
+	if (gen == VC4_GEN_4) {
+		ret = drmm_mutex_init(drm, &vc4->bin_bo_lock);
+		if (ret)
+			goto err;
+
+		ret = vc4_bo_cache_init(drm);
+		if (ret)
+			goto err;
+	}
+
+	ret = drmm_mode_config_init(drm);
+	if (ret)
+		goto err;
+
+	if (gen == VC4_GEN_4) {
+		ret = vc4_gem_init(drm);
+		if (ret)
+			goto err;
+	}
+
+	node = of_find_compatible_node(NULL, NULL, "raspberrypi,bcm2835-firmware");
+	if (node) {
+		firmware = rpi_firmware_get(node);
+		of_node_put(node);
+
+		if (!firmware) {
+			ret = -EPROBE_DEFER;
+			goto err;
+		}
+	}
+
+	ret = aperture_remove_all_conflicting_devices(driver->name);
+	if (ret)
+		goto err;
+
+	if (firmware) {
+		ret = rpi_firmware_property(firmware,
+					    RPI_FIRMWARE_NOTIFY_DISPLAY_DONE,
+					    NULL, 0);
+		if (ret)
+			drm_warn(drm, "Couldn't stop firmware display driver: %d\n", ret);
+
+		rpi_firmware_put(firmware);
+	}
+
+	ret = component_bind_all(dev, drm);
+	if (ret)
+		goto err;
+
+	ret = devm_add_action_or_reset(dev, vc4_component_unbind_all, vc4);
+	if (ret)
+		goto err;
+
+	ret = vc4_plane_create_additional_planes(drm);
+	if (ret)
+		goto err;
+
+	ret = vc4_kms_load(drm);
+	if (ret < 0)
+		goto err;
+
+	drm_for_each_crtc(crtc, drm)
+		vc4_crtc_disable_at_boot(crtc);
+
+	ret = drm_dev_register(drm, 0);
+	if (ret < 0)
+		goto err;
+
+	drm_client_setup_with_fourcc(drm, DRM_FORMAT_RGB565);
+
+	return 0;
+
+err:
+	platform_set_drvdata(pdev, NULL);
+	return ret;
+}
+
+static void vc4_drm_unbind(struct device *dev)
+{
+	struct drm_device *drm = dev_get_drvdata(dev);
+
+	drm_dev_unplug(drm);
+	drm_atomic_helper_shutdown(drm);
+	dev_set_drvdata(dev, NULL);
+}
+
+static const struct component_master_ops vc4_drm_ops = {
+	.bind = vc4_drm_bind,
+	.unbind = vc4_drm_unbind,
+};
+
+/*
+ * This list determines the binding order of our components, and we have
+ * a few constraints:
+ *   - The TXP driver needs to be bound before the PixelValves (CRTC)
+ *     but after the HVS to set the possible_crtc field properly
+ *   - The HDMI driver needs to be bound after the HVS so that we can
+ *     lookup the HVS maximum core clock rate and figure out if we
+ *     support 4kp60 or not.
+ */
+static struct platform_driver *const component_drivers[] = {
+	&vc4_hvs_driver,
+	&vc4_hdmi_driver,
+	&vc4_vec_driver,
+	&vc4_dpi_driver,
+	&vc4_dsi_driver,
+	&vc4_txp_driver,
+	&vc4_crtc_driver,
+	&vc4_v3d_driver,
+};
+
+static int vc4_platform_drm_probe(struct platform_device *pdev)
+{
+	struct component_match *match = NULL;
+	struct device *dev = &pdev->dev;
+
+	vc4_match_add_drivers(dev, &match,
+			      component_drivers, ARRAY_SIZE(component_drivers));
+
+	return component_master_add_with_match(dev, &vc4_drm_ops, match);
+}
+
+static void vc4_platform_drm_remove(struct platform_device *pdev)
+{
+	component_master_del(&pdev->dev, &vc4_drm_ops);
+}
+
+static void vc4_platform_drm_shutdown(struct platform_device *pdev)
+{
+	drm_atomic_helper_shutdown(platform_get_drvdata(pdev));
+}
+
+static const struct of_device_id vc4_of_match[] = {
+	{ .compatible = "brcm,bcm2711-vc5", .data = (void *)VC4_GEN_5 },
+	/* NB GEN_6_C will be corrected on D0 hw to GEN_6_D via vc4_hvs_bind */
+	{ .compatible = "brcm,bcm2712-vc6", .data = (void *)VC4_GEN_6_C },
+	{ .compatible = "brcm,bcm2835-vc4", .data = (void *)VC4_GEN_4 },
+	{ .compatible = "brcm,cygnus-vc4", .data = (void *)VC4_GEN_4 },
+	{},
+};
+MODULE_DEVICE_TABLE(of, vc4_of_match);
+
+static struct platform_driver vc4_platform_driver = {
+	.probe		= vc4_platform_drm_probe,
+	.remove		= vc4_platform_drm_remove,
+	.shutdown	= vc4_platform_drm_shutdown,
+	.driver		= {
+		.name	= "vc4-drm",
+		.of_match_table = vc4_of_match,
+	},
+};
+
+static int __init vc4_drm_register(void)
+{
+	int ret;
+
+	if (drm_firmware_drivers_only())
+		return -ENODEV;
+
+	ret = platform_register_drivers(component_drivers,
+					ARRAY_SIZE(component_drivers));
+	if (ret)
+		return ret;
+
+	ret = platform_driver_register(&vc4_platform_driver);
+	if (ret)
+		platform_unregister_drivers(component_drivers,
+					    ARRAY_SIZE(component_drivers));
+
+	return ret;
+}
+
+static void __exit vc4_drm_unregister(void)
+{
+	platform_unregister_drivers(component_drivers,
+				    ARRAY_SIZE(component_drivers));
+	platform_driver_unregister(&vc4_platform_driver);
+}
+
+module_init(vc4_drm_register);
+module_exit(vc4_drm_unregister);
+
+MODULE_ALIAS("platform:vc4-drm");
+MODULE_SOFTDEP("pre: snd-soc-hdmi-codec");
+MODULE_DESCRIPTION("Broadcom VC4 DRM Driver");
+MODULE_AUTHOR("Eric Anholt <eric@anholt.net>");
+MODULE_LICENSE("GPL v2");

linux-7.2.3/drivers/gpu/drm/vc4/vc4_fence.c

@@ -0,0 +1,48 @@
+/*
+ * Copyright © 2017 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "vc4_drv.h"
+
+static const char *vc4_fence_get_driver_name(struct dma_fence *fence)
+{
+	return "vc4";
+}
+
+static const char *vc4_fence_get_timeline_name(struct dma_fence *fence)
+{
+	return "vc4-v3d";
+}
+
+static bool vc4_fence_signaled(struct dma_fence *fence)
+{
+	struct vc4_fence *f = to_vc4_fence(fence);
+	struct vc4_dev *vc4 = to_vc4_dev(f->dev);
+
+	return vc4->finished_seqno >= f->seqno;
+}
+
+const struct dma_fence_ops vc4_fence_ops = {
+	.get_driver_name = vc4_fence_get_driver_name,
+	.get_timeline_name = vc4_fence_get_timeline_name,
+	.signaled = vc4_fence_signaled,
+};

linux-7.2.3/drivers/gpu/drm/vc4/vc4_hdmi.h

@@ -0,0 +1,248 @@
+#ifndef _VC4_HDMI_H_
+#define _VC4_HDMI_H_
+
+#include <drm/drm_connector.h>
+#include <media/cec.h>
+#include <sound/dmaengine_pcm.h>
+#include <sound/hdmi-codec.h>
+#include <sound/soc.h>
+
+#include "vc4_drv.h"
+
+struct vc4_hdmi;
+struct vc4_hdmi_register;
+
+enum vc4_hdmi_phy_channel {
+	PHY_LANE_0 = 0,
+	PHY_LANE_1,
+	PHY_LANE_2,
+	PHY_LANE_CK,
+};
+
+struct vc4_hdmi_variant {
+	/* Encoder Type for that controller */
+	enum vc4_encoder_type encoder_type;
+
+	/* ALSA card name */
+	const char *card_name;
+
+	/* Filename to expose the registers in debugfs */
+	const char *debugfs_name;
+
+	/* Maximum pixel clock supported by the controller (in Hz) */
+	unsigned long long max_pixel_clock;
+
+	/* List of the registers available on that variant */
+	const struct vc4_hdmi_register *registers;
+
+	/* Number of registers on that variant */
+	unsigned int num_registers;
+
+	/* BCM2711 Only.
+	 * The variants don't map the lane in the same order in the
+	 * PHY, so this is an array mapping the HDMI channel (index)
+	 * to the PHY lane (value).
+	 */
+	enum vc4_hdmi_phy_channel phy_lane_mapping[4];
+
+	/* The BCM2711 cannot deal with odd horizontal pixel timings */
+	bool unsupported_odd_h_timings;
+
+	/*
+	 * The BCM2711 CEC/hotplug IRQ controller is shared between the
+	 * two HDMI controllers, and we have a proper irqchip driver for
+	 * it.
+	 */
+	bool external_irq_controller;
+
+	/* Callback to get the resources (memory region, interrupts,
+	 * clocks, etc) for that variant.
+	 */
+	int (*init_resources)(struct drm_device *drm,
+			      struct vc4_hdmi *vc4_hdmi);
+
+	/* Callback to reset the HDMI block */
+	void (*reset)(struct vc4_hdmi *vc4_hdmi);
+
+	/* Callback to enable / disable the CSC */
+	void (*csc_setup)(struct vc4_hdmi *vc4_hdmi,
+			  struct drm_connector_state *state,
+			  const struct drm_display_mode *mode);
+
+	/* Callback to configure the video timings in the HDMI block */
+	void (*set_timings)(struct vc4_hdmi *vc4_hdmi,
+			    struct drm_connector_state *state,
+			    const struct drm_display_mode *mode);
+
+	/* Callback to initialize the PHY according to the connector state */
+	void (*phy_init)(struct vc4_hdmi *vc4_hdmi,
+			 struct drm_connector_state *conn_state);
+
+	/* Callback to disable the PHY */
+	void (*phy_disable)(struct vc4_hdmi *vc4_hdmi);
+
+	/* Callback to enable the RNG in the PHY */
+	void (*phy_rng_enable)(struct vc4_hdmi *vc4_hdmi);
+
+	/* Callback to disable the RNG in the PHY */
+	void (*phy_rng_disable)(struct vc4_hdmi *vc4_hdmi);
+
+	/* Callback to get channel map */
+	u32 (*channel_map)(struct vc4_hdmi *vc4_hdmi, u32 channel_mask);
+
+	/* Enables HDR metadata */
+	bool supports_hdr;
+
+	/* Callback for hardware specific hotplug detect */
+	bool (*hp_detect)(struct vc4_hdmi *vc4_hdmi);
+};
+
+/* HDMI audio information */
+struct vc4_hdmi_audio {
+	struct snd_soc_card card;
+	struct snd_soc_dai_link link;
+	struct snd_soc_dai_link_component cpu;
+	struct snd_soc_dai_link_component codec;
+	struct snd_soc_dai_link_component platform;
+	struct snd_dmaengine_dai_dma_data dma_data;
+	bool streaming;
+};
+
+/* General HDMI hardware state. */
+struct vc4_hdmi {
+	struct vc4_hdmi_audio audio;
+
+	struct platform_device *pdev;
+	const struct vc4_hdmi_variant *variant;
+
+	struct vc4_encoder encoder;
+	struct drm_connector connector;
+
+	struct delayed_work scrambling_work;
+
+	struct i2c_adapter *ddc;
+	void __iomem *hdmicore_regs;
+	void __iomem *hd_regs;
+
+	/* VC5 Only */
+	void __iomem *cec_regs;
+	/* VC5 Only */
+	void __iomem *csc_regs;
+	/* VC5 Only */
+	void __iomem *dvp_regs;
+	/* VC5 Only */
+	void __iomem *phy_regs;
+	/* VC5 Only */
+	void __iomem *ram_regs;
+	/* VC5 Only */
+	void __iomem *rm_regs;
+
+	struct gpio_desc *hpd_gpio;
+
+	/*
+	 * On some systems (like the RPi4), some modes are in the same
+	 * frequency range than the WiFi channels (1440p@60Hz for
+	 * example). Should we take evasive actions because that system
+	 * has a wifi adapter?
+	 */
+	bool disable_wifi_frequencies;
+
+	struct cec_msg cec_rx_msg;
+	bool cec_tx_ok;
+	bool cec_irq_was_rx;
+
+	struct clk *cec_clock;
+	struct clk *pixel_clock;
+	struct clk *hsm_clock;
+	struct clk *audio_clock;
+	struct clk *pixel_bvb_clock;
+
+	struct reset_control *reset;
+
+	struct debugfs_regset32 hdmi_regset;
+	struct debugfs_regset32 hd_regset;
+
+	/* VC5 only */
+	struct debugfs_regset32 cec_regset;
+	struct debugfs_regset32 csc_regset;
+	struct debugfs_regset32 dvp_regset;
+	struct debugfs_regset32 phy_regset;
+	struct debugfs_regset32 ram_regset;
+	struct debugfs_regset32 rm_regset;
+
+	/**
+	 * @hw_lock: Spinlock protecting device register access.
+	 */
+	spinlock_t hw_lock;
+
+	/**
+	 * @mutex: Mutex protecting the driver access across multiple
+	 * frameworks (KMS, ALSA, CEC).
+	 */
+	struct mutex mutex;
+
+	/**
+	 * @saved_adjusted_mode: Copy of @drm_crtc_state.adjusted_mode
+	 * for use by ALSA hooks and interrupt handlers. Protected by @mutex.
+	 */
+	struct drm_display_mode saved_adjusted_mode;
+
+	/**
+	 * @packet_ram_enabled: Is the HDMI controller packet RAM currently
+	 * on? Protected by @mutex.
+	 */
+	bool packet_ram_enabled;
+
+	/**
+	 * @scdc_enabled: Is the HDMI controller currently running with
+	 * the scrambler on? Protected by @mutex.
+	 */
+	bool scdc_enabled;
+
+	/**
+	 * @output_bpc: Copy of @drm_connector_state.hdmi.output_bpc for
+	 * use outside of KMS hooks. Protected by @mutex.
+	 */
+	unsigned int output_bpc;
+
+	/**
+	 * @output_format: Copy of
+	 * @drm_connector_state.hdmi.output_format for use outside of
+	 * KMS hooks. Protected by @mutex.
+	 */
+	enum drm_output_color_format output_format;
+
+	/**
+	 * @hdmi_jack: Represents the connection state of the HDMI plug, for
+	 * ALSA jack detection.
+	 */
+	struct snd_soc_jack hdmi_jack;
+};
+
+#define connector_to_vc4_hdmi(_connector)				\
+	container_of_const(_connector, struct vc4_hdmi, connector)
+
+static inline struct vc4_hdmi *
+encoder_to_vc4_hdmi(struct drm_encoder *encoder)
+{
+	struct vc4_encoder *_encoder = to_vc4_encoder(encoder);
+	return container_of_const(_encoder, struct vc4_hdmi, encoder);
+}
+
+void vc4_hdmi_phy_init(struct vc4_hdmi *vc4_hdmi,
+		       struct drm_connector_state *conn_state);
+void vc4_hdmi_phy_disable(struct vc4_hdmi *vc4_hdmi);
+void vc4_hdmi_phy_rng_enable(struct vc4_hdmi *vc4_hdmi);
+void vc4_hdmi_phy_rng_disable(struct vc4_hdmi *vc4_hdmi);
+
+void vc5_hdmi_phy_init(struct vc4_hdmi *vc4_hdmi,
+		       struct drm_connector_state *conn_state);
+void vc5_hdmi_phy_disable(struct vc4_hdmi *vc4_hdmi);
+void vc5_hdmi_phy_rng_enable(struct vc4_hdmi *vc4_hdmi);
+void vc5_hdmi_phy_rng_disable(struct vc4_hdmi *vc4_hdmi);
+
+void vc6_hdmi_phy_init(struct vc4_hdmi *vc4_hdmi,
+		       struct drm_connector_state *conn_state);
+void vc6_hdmi_phy_disable(struct vc4_hdmi *vc4_hdmi);
+
+#endif /* _VC4_HDMI_H_ */

linux-7.2.3/drivers/gpu/drm/vc4/vc4_hdmi_regs.h

@@ -0,0 +1,728 @@
+#ifndef _VC4_HDMI_REGS_H_
+#define _VC4_HDMI_REGS_H_
+
+#include <linux/pm_runtime.h>
+
+#include "vc4_hdmi.h"
+
+#define VC4_HDMI_PACKET_STRIDE			0x24
+
+enum vc4_hdmi_regs {
+	VC4_INVALID = 0,
+	VC4_HDMI,
+	VC4_HD,
+	VC5_CEC,
+	VC5_CSC,
+	VC5_DVP,
+	VC5_PHY,
+	VC5_RAM,
+	VC5_RM,
+};
+
+enum vc4_hdmi_field {
+	HDMI_AUDIO_PACKET_CONFIG,
+	HDMI_CEC_CNTRL_1,
+	HDMI_CEC_CNTRL_2,
+	HDMI_CEC_CNTRL_3,
+	HDMI_CEC_CNTRL_4,
+	HDMI_CEC_CNTRL_5,
+	HDMI_CEC_CPU_CLEAR,
+	HDMI_CEC_CPU_MASK_CLEAR,
+	HDMI_CEC_CPU_MASK_SET,
+	HDMI_CEC_CPU_MASK_STATUS,
+	HDMI_CEC_CPU_STATUS,
+	HDMI_CEC_CPU_SET,
+
+	/*
+	 * Transmit data, first byte is low byte of the 32-bit reg.
+	 * MSB of each byte transmitted first.
+	 */
+	HDMI_CEC_RX_DATA_1,
+	HDMI_CEC_RX_DATA_2,
+	HDMI_CEC_RX_DATA_3,
+	HDMI_CEC_RX_DATA_4,
+	HDMI_CEC_TX_DATA_1,
+	HDMI_CEC_TX_DATA_2,
+	HDMI_CEC_TX_DATA_3,
+	HDMI_CEC_TX_DATA_4,
+	HDMI_CLOCK_STOP,
+	HDMI_CORE_REV,
+	HDMI_CRP_CFG,
+	HDMI_CSC_12_11,
+	HDMI_CSC_14_13,
+	HDMI_CSC_22_21,
+	HDMI_CSC_24_23,
+	HDMI_CSC_32_31,
+	HDMI_CSC_34_33,
+	HDMI_CSC_CHANNEL_CTL,
+	HDMI_CSC_CTL,
+
+	/*
+	 * 20-bit fields containing CTS values to be transmitted if
+	 * !EXTERNAL_CTS_EN
+	 */
+	HDMI_CTS_0,
+	HDMI_CTS_1,
+	HDMI_DEEP_COLOR_CONFIG_1,
+	HDMI_DVP_CTL,
+	HDMI_FIFO_CTL,
+	HDMI_FRAME_COUNT,
+	HDMI_GCP_CONFIG,
+	HDMI_GCP_WORD_1,
+	HDMI_HORZA,
+	HDMI_HORZB,
+	HDMI_HOTPLUG,
+	HDMI_HOTPLUG_INT,
+
+	/*
+	 * 3 bits per field, where each field maps from that
+	 * corresponding MAI bus channel to the given HDMI channel.
+	 */
+	HDMI_MAI_CHANNEL_MAP,
+	HDMI_MAI_CONFIG,
+	HDMI_MAI_CTL,
+
+	/*
+	 * Register for DMAing in audio data to be transported over
+	 * the MAI bus to the Falcon core.
+	 */
+	HDMI_MAI_DATA,
+
+	/* Format header to be placed on the MAI data. Unused. */
+	HDMI_MAI_FMT,
+
+	/* Last received format word on the MAI bus. */
+	HDMI_MAI_FORMAT,
+	HDMI_MAI_SMP,
+	HDMI_MAI_THR,
+	HDMI_M_CTL,
+	HDMI_RAM_PACKET_CONFIG,
+	HDMI_RAM_PACKET_START,
+	HDMI_RAM_PACKET_STATUS,
+	HDMI_RM_CONTROL,
+	HDMI_RM_FORMAT,
+	HDMI_RM_OFFSET,
+	HDMI_SCHEDULER_CONTROL,
+	HDMI_SCRAMBLER_CTL,
+	HDMI_SW_RESET_CONTROL,
+	HDMI_TX_PHY_CHANNEL_SWAP,
+	HDMI_TX_PHY_CLK_DIV,
+	HDMI_TX_PHY_CTL_0,
+	HDMI_TX_PHY_CTL_1,
+	HDMI_TX_PHY_CTL_2,
+	HDMI_TX_PHY_CTL_3,
+	HDMI_TX_PHY_CTL_CK,
+	HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_1,
+	HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_2,
+	HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_4,
+	HDMI_TX_PHY_PLL_CFG,
+	HDMI_TX_PHY_PLL_CFG_PDIV,
+	HDMI_TX_PHY_PLL_CTL_0,
+	HDMI_TX_PHY_PLL_CTL_1,
+	HDMI_TX_PHY_PLL_MISC_0,
+	HDMI_TX_PHY_PLL_MISC_1,
+	HDMI_TX_PHY_PLL_MISC_2,
+	HDMI_TX_PHY_PLL_MISC_3,
+	HDMI_TX_PHY_PLL_MISC_4,
+	HDMI_TX_PHY_PLL_MISC_5,
+	HDMI_TX_PHY_PLL_MISC_6,
+	HDMI_TX_PHY_PLL_MISC_7,
+	HDMI_TX_PHY_PLL_MISC_8,
+	HDMI_TX_PHY_PLL_POST_KDIV,
+	HDMI_TX_PHY_PLL_POWERUP_CTL,
+	HDMI_TX_PHY_PLL_REFCLK,
+	HDMI_TX_PHY_PLL_RESET_CTL,
+	HDMI_TX_PHY_PLL_VCOCLK_DIV,
+	HDMI_TX_PHY_POWERDOWN_CTL,
+	HDMI_TX_PHY_POWERUP_CTL,
+	HDMI_TX_PHY_RESET_CTL,
+	HDMI_TX_PHY_TMDS_CLK_WORD_SEL,
+	HDMI_VEC_INTERFACE_CFG,
+	HDMI_VEC_INTERFACE_XBAR,
+	HDMI_VERTA0,
+	HDMI_VERTA1,
+	HDMI_VERTB0,
+	HDMI_VERTB1,
+	HDMI_VID_CTL,
+	HDMI_MISC_CONTROL,
+	HDMI_FORMAT_DET_1,
+	HDMI_FORMAT_DET_2,
+	HDMI_FORMAT_DET_3,
+	HDMI_FORMAT_DET_4,
+	HDMI_FORMAT_DET_5,
+	HDMI_FORMAT_DET_6,
+	HDMI_FORMAT_DET_7,
+	HDMI_FORMAT_DET_8,
+	HDMI_FORMAT_DET_9,
+	HDMI_FORMAT_DET_10,
+};
+
+struct vc4_hdmi_register {
+	char *name;
+	enum vc4_hdmi_regs reg;
+	unsigned int offset;
+};
+
+#define _VC4_REG(_base, _reg, _offset)	\
+	[_reg] = {				\
+		.name = #_reg,			\
+		.reg = _base,			\
+		.offset = _offset,		\
+	}
+
+#define VC4_HD_REG(reg, offset)		_VC4_REG(VC4_HD, reg, offset)
+#define VC4_HDMI_REG(reg, offset)	_VC4_REG(VC4_HDMI, reg, offset)
+#define VC5_CEC_REG(reg, offset)	_VC4_REG(VC5_CEC, reg, offset)
+#define VC5_CSC_REG(reg, offset)	_VC4_REG(VC5_CSC, reg, offset)
+#define VC5_DVP_REG(reg, offset)	_VC4_REG(VC5_DVP, reg, offset)
+#define VC5_PHY_REG(reg, offset)	_VC4_REG(VC5_PHY, reg, offset)
+#define VC5_RAM_REG(reg, offset)	_VC4_REG(VC5_RAM, reg, offset)
+#define VC5_RM_REG(reg, offset)		_VC4_REG(VC5_RM, reg, offset)
+
+static const struct vc4_hdmi_register __maybe_unused vc4_hdmi_fields[] = {
+	VC4_HD_REG(HDMI_M_CTL, 0x000c),
+	VC4_HD_REG(HDMI_MAI_CTL, 0x0014),
+	VC4_HD_REG(HDMI_MAI_THR, 0x0018),
+	VC4_HD_REG(HDMI_MAI_FMT, 0x001c),
+	VC4_HD_REG(HDMI_MAI_DATA, 0x0020),
+	VC4_HD_REG(HDMI_MAI_SMP, 0x002c),
+	VC4_HD_REG(HDMI_VID_CTL, 0x0038),
+	VC4_HD_REG(HDMI_CSC_CTL, 0x0040),
+	VC4_HD_REG(HDMI_CSC_12_11, 0x0044),
+	VC4_HD_REG(HDMI_CSC_14_13, 0x0048),
+	VC4_HD_REG(HDMI_CSC_22_21, 0x004c),
+	VC4_HD_REG(HDMI_CSC_24_23, 0x0050),
+	VC4_HD_REG(HDMI_CSC_32_31, 0x0054),
+	VC4_HD_REG(HDMI_CSC_34_33, 0x0058),
+	VC4_HD_REG(HDMI_FRAME_COUNT, 0x0068),
+
+	VC4_HDMI_REG(HDMI_CORE_REV, 0x0000),
+	VC4_HDMI_REG(HDMI_SW_RESET_CONTROL, 0x0004),
+	VC4_HDMI_REG(HDMI_HOTPLUG_INT, 0x0008),
+	VC4_HDMI_REG(HDMI_HOTPLUG, 0x000c),
+	VC4_HDMI_REG(HDMI_FIFO_CTL, 0x005c),
+	VC4_HDMI_REG(HDMI_MAI_CHANNEL_MAP, 0x0090),
+	VC4_HDMI_REG(HDMI_MAI_CONFIG, 0x0094),
+	VC4_HDMI_REG(HDMI_MAI_FORMAT, 0x0098),
+	VC4_HDMI_REG(HDMI_AUDIO_PACKET_CONFIG, 0x009c),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_CONFIG, 0x00a0),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_STATUS, 0x00a4),
+	VC4_HDMI_REG(HDMI_CRP_CFG, 0x00a8),
+	VC4_HDMI_REG(HDMI_CTS_0, 0x00ac),
+	VC4_HDMI_REG(HDMI_CTS_1, 0x00b0),
+	VC4_HDMI_REG(HDMI_SCHEDULER_CONTROL, 0x00c0),
+	VC4_HDMI_REG(HDMI_HORZA, 0x00c4),
+	VC4_HDMI_REG(HDMI_HORZB, 0x00c8),
+	VC4_HDMI_REG(HDMI_VERTA0, 0x00cc),
+	VC4_HDMI_REG(HDMI_VERTB0, 0x00d0),
+	VC4_HDMI_REG(HDMI_VERTA1, 0x00d4),
+	VC4_HDMI_REG(HDMI_VERTB1, 0x00d8),
+	VC4_HDMI_REG(HDMI_MISC_CONTROL, 0x00e4),
+	VC4_HDMI_REG(HDMI_CEC_CNTRL_1, 0x00e8),
+	VC4_HDMI_REG(HDMI_CEC_CNTRL_2, 0x00ec),
+	VC4_HDMI_REG(HDMI_CEC_CNTRL_3, 0x00f0),
+	VC4_HDMI_REG(HDMI_CEC_CNTRL_4, 0x00f4),
+	VC4_HDMI_REG(HDMI_CEC_CNTRL_5, 0x00f8),
+	VC4_HDMI_REG(HDMI_CEC_TX_DATA_1, 0x00fc),
+	VC4_HDMI_REG(HDMI_CEC_TX_DATA_2, 0x0100),
+	VC4_HDMI_REG(HDMI_CEC_TX_DATA_3, 0x0104),
+	VC4_HDMI_REG(HDMI_CEC_TX_DATA_4, 0x0108),
+	VC4_HDMI_REG(HDMI_CEC_RX_DATA_1, 0x010c),
+	VC4_HDMI_REG(HDMI_CEC_RX_DATA_2, 0x0110),
+	VC4_HDMI_REG(HDMI_CEC_RX_DATA_3, 0x0114),
+	VC4_HDMI_REG(HDMI_CEC_RX_DATA_4, 0x0118),
+	VC4_HDMI_REG(HDMI_TX_PHY_RESET_CTL, 0x02c0),
+	VC4_HDMI_REG(HDMI_TX_PHY_CTL_0, 0x02c4),
+	VC4_HDMI_REG(HDMI_CEC_CPU_STATUS, 0x0340),
+	VC4_HDMI_REG(HDMI_CEC_CPU_SET, 0x0344),
+	VC4_HDMI_REG(HDMI_CEC_CPU_CLEAR, 0x0348),
+	VC4_HDMI_REG(HDMI_CEC_CPU_MASK_STATUS, 0x034c),
+	VC4_HDMI_REG(HDMI_CEC_CPU_MASK_SET, 0x0350),
+	VC4_HDMI_REG(HDMI_CEC_CPU_MASK_CLEAR, 0x0354),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_START, 0x0400),
+};
+
+static const struct vc4_hdmi_register __maybe_unused vc5_hdmi_hdmi0_fields[] = {
+	VC4_HD_REG(HDMI_DVP_CTL, 0x0000),
+	VC4_HD_REG(HDMI_MAI_CTL, 0x0010),
+	VC4_HD_REG(HDMI_MAI_THR, 0x0014),
+	VC4_HD_REG(HDMI_MAI_FMT, 0x0018),
+	VC4_HD_REG(HDMI_MAI_DATA, 0x001c),
+	VC4_HD_REG(HDMI_MAI_SMP, 0x0020),
+	VC4_HD_REG(HDMI_VID_CTL, 0x0044),
+	VC4_HD_REG(HDMI_FRAME_COUNT, 0x0060),
+
+	VC4_HDMI_REG(HDMI_FIFO_CTL, 0x074),
+	VC4_HDMI_REG(HDMI_AUDIO_PACKET_CONFIG, 0x0b8),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_CONFIG, 0x0bc),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_STATUS, 0x0c4),
+	VC4_HDMI_REG(HDMI_CRP_CFG, 0x0c8),
+	VC4_HDMI_REG(HDMI_CTS_0, 0x0cc),
+	VC4_HDMI_REG(HDMI_CTS_1, 0x0d0),
+	VC4_HDMI_REG(HDMI_SCHEDULER_CONTROL, 0x0e0),
+	VC4_HDMI_REG(HDMI_HORZA, 0x0e4),
+	VC4_HDMI_REG(HDMI_HORZB, 0x0e8),
+	VC4_HDMI_REG(HDMI_VERTA0, 0x0ec),
+	VC4_HDMI_REG(HDMI_VERTB0, 0x0f0),
+	VC4_HDMI_REG(HDMI_VERTA1, 0x0f4),
+	VC4_HDMI_REG(HDMI_VERTB1, 0x0f8),
+	VC4_HDMI_REG(HDMI_MISC_CONTROL, 0x100),
+	VC4_HDMI_REG(HDMI_MAI_CHANNEL_MAP, 0x09c),
+	VC4_HDMI_REG(HDMI_MAI_CONFIG, 0x0a0),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_1, 0x134),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_2, 0x138),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_3, 0x13c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_4, 0x140),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_5, 0x144),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_6, 0x148),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_7, 0x14c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_8, 0x150),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_9, 0x154),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_10, 0x158),
+	VC4_HDMI_REG(HDMI_DEEP_COLOR_CONFIG_1, 0x170),
+	VC4_HDMI_REG(HDMI_GCP_CONFIG, 0x178),
+	VC4_HDMI_REG(HDMI_GCP_WORD_1, 0x17c),
+	VC4_HDMI_REG(HDMI_HOTPLUG, 0x1a8),
+	VC4_HDMI_REG(HDMI_SCRAMBLER_CTL, 0x1c4),
+
+	VC5_DVP_REG(HDMI_CLOCK_STOP, 0x0bc),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_CFG, 0x0ec),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_XBAR, 0x0f0),
+
+	VC5_PHY_REG(HDMI_TX_PHY_RESET_CTL, 0x000),
+	VC5_PHY_REG(HDMI_TX_PHY_POWERDOWN_CTL, 0x004),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_0, 0x008),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_1, 0x00c),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_2, 0x010),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_3, 0x014),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CTL_0, 0x01c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CTL_1, 0x020),
+	VC5_PHY_REG(HDMI_TX_PHY_CLK_DIV, 0x028),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CFG, 0x034),
+	VC5_PHY_REG(HDMI_TX_PHY_TMDS_CLK_WORD_SEL, 0x044),
+	VC5_PHY_REG(HDMI_TX_PHY_CHANNEL_SWAP, 0x04c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_1, 0x050),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_2, 0x054),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_4, 0x05c),
+
+	VC5_RM_REG(HDMI_RM_CONTROL, 0x000),
+	VC5_RM_REG(HDMI_RM_OFFSET, 0x018),
+	VC5_RM_REG(HDMI_RM_FORMAT, 0x01c),
+
+	VC5_RAM_REG(HDMI_RAM_PACKET_START, 0x000),
+
+	VC5_CEC_REG(HDMI_CEC_CNTRL_1, 0x010),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_2, 0x014),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_3, 0x018),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_4, 0x01c),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_5, 0x020),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_1, 0x028),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_2, 0x02c),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_3, 0x030),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_4, 0x034),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_1, 0x038),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_2, 0x03c),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_3, 0x040),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_4, 0x044),
+
+	VC5_CSC_REG(HDMI_CSC_CTL, 0x000),
+	VC5_CSC_REG(HDMI_CSC_12_11, 0x004),
+	VC5_CSC_REG(HDMI_CSC_14_13, 0x008),
+	VC5_CSC_REG(HDMI_CSC_22_21, 0x00c),
+	VC5_CSC_REG(HDMI_CSC_24_23, 0x010),
+	VC5_CSC_REG(HDMI_CSC_32_31, 0x014),
+	VC5_CSC_REG(HDMI_CSC_34_33, 0x018),
+	VC5_CSC_REG(HDMI_CSC_CHANNEL_CTL, 0x02c),
+};
+
+static const struct vc4_hdmi_register __maybe_unused vc5_hdmi_hdmi1_fields[] = {
+	VC4_HD_REG(HDMI_DVP_CTL, 0x0000),
+	VC4_HD_REG(HDMI_MAI_CTL, 0x0030),
+	VC4_HD_REG(HDMI_MAI_THR, 0x0034),
+	VC4_HD_REG(HDMI_MAI_FMT, 0x0038),
+	VC4_HD_REG(HDMI_MAI_DATA, 0x003c),
+	VC4_HD_REG(HDMI_MAI_SMP, 0x0040),
+	VC4_HD_REG(HDMI_VID_CTL, 0x0048),
+	VC4_HD_REG(HDMI_FRAME_COUNT, 0x0064),
+
+	VC4_HDMI_REG(HDMI_FIFO_CTL, 0x074),
+	VC4_HDMI_REG(HDMI_AUDIO_PACKET_CONFIG, 0x0b8),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_CONFIG, 0x0bc),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_STATUS, 0x0c4),
+	VC4_HDMI_REG(HDMI_CRP_CFG, 0x0c8),
+	VC4_HDMI_REG(HDMI_CTS_0, 0x0cc),
+	VC4_HDMI_REG(HDMI_CTS_1, 0x0d0),
+	VC4_HDMI_REG(HDMI_SCHEDULER_CONTROL, 0x0e0),
+	VC4_HDMI_REG(HDMI_HORZA, 0x0e4),
+	VC4_HDMI_REG(HDMI_HORZB, 0x0e8),
+	VC4_HDMI_REG(HDMI_VERTA0, 0x0ec),
+	VC4_HDMI_REG(HDMI_VERTB0, 0x0f0),
+	VC4_HDMI_REG(HDMI_VERTA1, 0x0f4),
+	VC4_HDMI_REG(HDMI_VERTB1, 0x0f8),
+	VC4_HDMI_REG(HDMI_MISC_CONTROL, 0x100),
+	VC4_HDMI_REG(HDMI_MAI_CHANNEL_MAP, 0x09c),
+	VC4_HDMI_REG(HDMI_MAI_CONFIG, 0x0a0),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_1, 0x134),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_2, 0x138),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_3, 0x13c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_4, 0x140),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_5, 0x144),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_6, 0x148),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_7, 0x14c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_8, 0x150),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_9, 0x154),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_10, 0x158),
+	VC4_HDMI_REG(HDMI_DEEP_COLOR_CONFIG_1, 0x170),
+	VC4_HDMI_REG(HDMI_GCP_CONFIG, 0x178),
+	VC4_HDMI_REG(HDMI_GCP_WORD_1, 0x17c),
+	VC4_HDMI_REG(HDMI_HOTPLUG, 0x1a8),
+	VC4_HDMI_REG(HDMI_SCRAMBLER_CTL, 0x1c4),
+
+	VC5_DVP_REG(HDMI_CLOCK_STOP, 0x0bc),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_CFG, 0x0ec),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_XBAR, 0x0f0),
+
+	VC5_PHY_REG(HDMI_TX_PHY_RESET_CTL, 0x000),
+	VC5_PHY_REG(HDMI_TX_PHY_POWERDOWN_CTL, 0x004),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_0, 0x008),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_1, 0x00c),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_2, 0x010),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_3, 0x014),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CTL_0, 0x01c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CTL_1, 0x020),
+	VC5_PHY_REG(HDMI_TX_PHY_CLK_DIV, 0x028),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CFG, 0x034),
+	VC5_PHY_REG(HDMI_TX_PHY_CHANNEL_SWAP, 0x04c),
+	VC5_PHY_REG(HDMI_TX_PHY_TMDS_CLK_WORD_SEL, 0x044),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_1, 0x050),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_2, 0x054),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CALIBRATION_CONFIG_4, 0x05c),
+
+	VC5_RM_REG(HDMI_RM_CONTROL, 0x000),
+	VC5_RM_REG(HDMI_RM_OFFSET, 0x018),
+	VC5_RM_REG(HDMI_RM_FORMAT, 0x01c),
+
+	VC5_RAM_REG(HDMI_RAM_PACKET_START, 0x000),
+
+	VC5_CEC_REG(HDMI_CEC_CNTRL_1, 0x010),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_2, 0x014),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_3, 0x018),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_4, 0x01c),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_5, 0x020),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_1, 0x028),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_2, 0x02c),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_3, 0x030),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_4, 0x034),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_1, 0x038),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_2, 0x03c),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_3, 0x040),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_4, 0x044),
+
+	VC5_CSC_REG(HDMI_CSC_CTL, 0x000),
+	VC5_CSC_REG(HDMI_CSC_12_11, 0x004),
+	VC5_CSC_REG(HDMI_CSC_14_13, 0x008),
+	VC5_CSC_REG(HDMI_CSC_22_21, 0x00c),
+	VC5_CSC_REG(HDMI_CSC_24_23, 0x010),
+	VC5_CSC_REG(HDMI_CSC_32_31, 0x014),
+	VC5_CSC_REG(HDMI_CSC_34_33, 0x018),
+	VC5_CSC_REG(HDMI_CSC_CHANNEL_CTL, 0x02c),
+};
+
+static const struct vc4_hdmi_register __maybe_unused vc6_hdmi_hdmi0_fields[] = {
+	VC4_HD_REG(HDMI_DVP_CTL, 0x0000),
+	VC4_HD_REG(HDMI_MAI_CTL, 0x0010),
+	VC4_HD_REG(HDMI_MAI_THR, 0x0014),
+	VC4_HD_REG(HDMI_MAI_FMT, 0x0018),
+	VC4_HD_REG(HDMI_MAI_DATA, 0x001c),
+	VC4_HD_REG(HDMI_MAI_SMP, 0x0020),
+	VC4_HD_REG(HDMI_VID_CTL, 0x0044),
+	VC4_HD_REG(HDMI_FRAME_COUNT, 0x0060),
+
+	VC4_HDMI_REG(HDMI_FIFO_CTL, 0x07c),
+	VC4_HDMI_REG(HDMI_AUDIO_PACKET_CONFIG, 0x0c0),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_CONFIG, 0x0c4),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_STATUS, 0x0cc),
+	VC4_HDMI_REG(HDMI_CRP_CFG, 0x0d0),
+	VC4_HDMI_REG(HDMI_CTS_0, 0x0d4),
+	VC4_HDMI_REG(HDMI_CTS_1, 0x0d8),
+	VC4_HDMI_REG(HDMI_SCHEDULER_CONTROL, 0x0e8),
+	VC4_HDMI_REG(HDMI_HORZA, 0x0ec),
+	VC4_HDMI_REG(HDMI_HORZB, 0x0f0),
+	VC4_HDMI_REG(HDMI_VERTA0, 0x0f4),
+	VC4_HDMI_REG(HDMI_VERTB0, 0x0f8),
+	VC4_HDMI_REG(HDMI_VERTA1, 0x100),
+	VC4_HDMI_REG(HDMI_VERTB1, 0x104),
+	VC4_HDMI_REG(HDMI_MISC_CONTROL, 0x114),
+	VC4_HDMI_REG(HDMI_MAI_CHANNEL_MAP, 0x0a4),
+	VC4_HDMI_REG(HDMI_MAI_CONFIG, 0x0a8),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_1, 0x148),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_2, 0x14c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_3, 0x150),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_4, 0x158),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_5, 0x15c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_6, 0x160),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_7, 0x164),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_8, 0x168),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_9, 0x16c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_10, 0x170),
+	VC4_HDMI_REG(HDMI_DEEP_COLOR_CONFIG_1, 0x18c),
+	VC4_HDMI_REG(HDMI_GCP_CONFIG, 0x194),
+	VC4_HDMI_REG(HDMI_GCP_WORD_1, 0x198),
+	VC4_HDMI_REG(HDMI_HOTPLUG, 0x1c8),
+	VC4_HDMI_REG(HDMI_SCRAMBLER_CTL, 0x1e4),
+
+	VC5_DVP_REG(HDMI_CLOCK_STOP, 0x0bc),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_CFG, 0x0f0),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_XBAR, 0x0f4),
+
+	VC5_PHY_REG(HDMI_TX_PHY_RESET_CTL, 0x000),
+	VC5_PHY_REG(HDMI_TX_PHY_POWERUP_CTL, 0x004),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_0, 0x008),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_1, 0x00c),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_2, 0x010),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_CK, 0x014),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_REFCLK, 0x01c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_POST_KDIV, 0x028),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_VCOCLK_DIV, 0x02c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CFG, 0x044),
+	VC5_PHY_REG(HDMI_TX_PHY_TMDS_CLK_WORD_SEL, 0x054),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_0, 0x060),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_1, 0x064),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_2, 0x068),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_3, 0x06c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_4, 0x070),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_5, 0x074),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_6, 0x078),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_7, 0x07c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_8, 0x080),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_RESET_CTL, 0x190),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_POWERUP_CTL, 0x194),
+
+	VC5_RM_REG(HDMI_RM_CONTROL, 0x000),
+	VC5_RM_REG(HDMI_RM_OFFSET, 0x018),
+	VC5_RM_REG(HDMI_RM_FORMAT, 0x01c),
+
+	VC5_RAM_REG(HDMI_RAM_PACKET_START, 0x000),
+
+	VC5_CEC_REG(HDMI_CEC_CNTRL_1, 0x010),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_2, 0x014),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_3, 0x018),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_4, 0x01c),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_5, 0x020),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_1, 0x028),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_2, 0x02c),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_3, 0x030),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_4, 0x034),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_1, 0x038),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_2, 0x03c),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_3, 0x040),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_4, 0x044),
+
+	VC5_CSC_REG(HDMI_CSC_CTL, 0x000),
+	VC5_CSC_REG(HDMI_CSC_12_11, 0x004),
+	VC5_CSC_REG(HDMI_CSC_14_13, 0x008),
+	VC5_CSC_REG(HDMI_CSC_22_21, 0x00c),
+	VC5_CSC_REG(HDMI_CSC_24_23, 0x010),
+	VC5_CSC_REG(HDMI_CSC_32_31, 0x014),
+	VC5_CSC_REG(HDMI_CSC_34_33, 0x018),
+	VC5_CSC_REG(HDMI_CSC_CHANNEL_CTL, 0x02c),
+};
+
+static const struct vc4_hdmi_register __maybe_unused vc6_hdmi_hdmi1_fields[] = {
+	VC4_HD_REG(HDMI_DVP_CTL, 0x0000),
+	VC4_HD_REG(HDMI_MAI_CTL, 0x0030),
+	VC4_HD_REG(HDMI_MAI_THR, 0x0034),
+	VC4_HD_REG(HDMI_MAI_FMT, 0x0038),
+	VC4_HD_REG(HDMI_MAI_DATA, 0x003c),
+	VC4_HD_REG(HDMI_MAI_SMP, 0x0040),
+	VC4_HD_REG(HDMI_VID_CTL, 0x0048),
+	VC4_HD_REG(HDMI_FRAME_COUNT, 0x0064),
+
+	VC4_HDMI_REG(HDMI_FIFO_CTL, 0x07c),
+	VC4_HDMI_REG(HDMI_AUDIO_PACKET_CONFIG, 0x0c0),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_CONFIG, 0x0c4),
+	VC4_HDMI_REG(HDMI_RAM_PACKET_STATUS, 0x0cc),
+	VC4_HDMI_REG(HDMI_CRP_CFG, 0x0d0),
+	VC4_HDMI_REG(HDMI_CTS_0, 0x0d4),
+	VC4_HDMI_REG(HDMI_CTS_1, 0x0d8),
+	VC4_HDMI_REG(HDMI_SCHEDULER_CONTROL, 0x0e8),
+	VC4_HDMI_REG(HDMI_HORZA, 0x0ec),
+	VC4_HDMI_REG(HDMI_HORZB, 0x0f0),
+	VC4_HDMI_REG(HDMI_VERTA0, 0x0f4),
+	VC4_HDMI_REG(HDMI_VERTB0, 0x0f8),
+	VC4_HDMI_REG(HDMI_VERTA1, 0x100),
+	VC4_HDMI_REG(HDMI_VERTB1, 0x104),
+	VC4_HDMI_REG(HDMI_MISC_CONTROL, 0x114),
+	VC4_HDMI_REG(HDMI_MAI_CHANNEL_MAP, 0x0a4),
+	VC4_HDMI_REG(HDMI_MAI_CONFIG, 0x0a8),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_1, 0x148),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_2, 0x14c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_3, 0x150),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_4, 0x158),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_5, 0x15c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_6, 0x160),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_7, 0x164),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_8, 0x168),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_9, 0x16c),
+	VC4_HDMI_REG(HDMI_FORMAT_DET_10, 0x170),
+	VC4_HDMI_REG(HDMI_DEEP_COLOR_CONFIG_1, 0x18c),
+	VC4_HDMI_REG(HDMI_GCP_CONFIG, 0x194),
+	VC4_HDMI_REG(HDMI_GCP_WORD_1, 0x198),
+	VC4_HDMI_REG(HDMI_HOTPLUG, 0x1c8),
+	VC4_HDMI_REG(HDMI_SCRAMBLER_CTL, 0x1e4),
+
+	VC5_DVP_REG(HDMI_CLOCK_STOP, 0x0bc),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_CFG, 0x0f0),
+	VC5_DVP_REG(HDMI_VEC_INTERFACE_XBAR, 0x0f4),
+
+	VC5_PHY_REG(HDMI_TX_PHY_RESET_CTL, 0x000),
+	VC5_PHY_REG(HDMI_TX_PHY_POWERUP_CTL, 0x004),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_0, 0x008),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_1, 0x00c),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_2, 0x010),
+	VC5_PHY_REG(HDMI_TX_PHY_CTL_CK, 0x014),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_REFCLK, 0x01c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_POST_KDIV, 0x028),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_VCOCLK_DIV, 0x02c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_CFG, 0x044),
+	VC5_PHY_REG(HDMI_TX_PHY_TMDS_CLK_WORD_SEL, 0x054),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_0, 0x060),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_1, 0x064),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_2, 0x068),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_3, 0x06c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_4, 0x070),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_5, 0x074),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_6, 0x078),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_7, 0x07c),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_MISC_8, 0x080),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_RESET_CTL, 0x190),
+	VC5_PHY_REG(HDMI_TX_PHY_PLL_POWERUP_CTL, 0x194),
+
+	VC5_RM_REG(HDMI_RM_CONTROL, 0x000),
+	VC5_RM_REG(HDMI_RM_OFFSET, 0x018),
+	VC5_RM_REG(HDMI_RM_FORMAT, 0x01c),
+
+	VC5_RAM_REG(HDMI_RAM_PACKET_START, 0x000),
+
+	VC5_CEC_REG(HDMI_CEC_CNTRL_1, 0x010),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_2, 0x014),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_3, 0x018),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_4, 0x01c),
+	VC5_CEC_REG(HDMI_CEC_CNTRL_5, 0x020),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_1, 0x028),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_2, 0x02c),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_3, 0x030),
+	VC5_CEC_REG(HDMI_CEC_TX_DATA_4, 0x034),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_1, 0x038),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_2, 0x03c),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_3, 0x040),
+	VC5_CEC_REG(HDMI_CEC_RX_DATA_4, 0x044),
+
+	VC5_CSC_REG(HDMI_CSC_CTL, 0x000),
+	VC5_CSC_REG(HDMI_CSC_12_11, 0x004),
+	VC5_CSC_REG(HDMI_CSC_14_13, 0x008),
+	VC5_CSC_REG(HDMI_CSC_22_21, 0x00c),
+	VC5_CSC_REG(HDMI_CSC_24_23, 0x010),
+	VC5_CSC_REG(HDMI_CSC_32_31, 0x014),
+	VC5_CSC_REG(HDMI_CSC_34_33, 0x018),
+	VC5_CSC_REG(HDMI_CSC_CHANNEL_CTL, 0x02c),
+};
+
+static inline
+void __iomem *__vc4_hdmi_get_field_base(struct vc4_hdmi *hdmi,
+					enum vc4_hdmi_regs reg)
+{
+	switch (reg) {
+	case VC4_HD:
+		return hdmi->hd_regs;
+
+	case VC4_HDMI:
+		return hdmi->hdmicore_regs;
+
+	case VC5_CSC:
+		return hdmi->csc_regs;
+
+	case VC5_CEC:
+		return hdmi->cec_regs;
+
+	case VC5_DVP:
+		return hdmi->dvp_regs;
+
+	case VC5_PHY:
+		return hdmi->phy_regs;
+
+	case VC5_RAM:
+		return hdmi->ram_regs;
+
+	case VC5_RM:
+		return hdmi->rm_regs;
+
+	default:
+		return NULL;
+	}
+
+	return NULL;
+}
+
+static inline u32 vc4_hdmi_read(struct vc4_hdmi *hdmi,
+				enum vc4_hdmi_field reg)
+{
+	const struct vc4_hdmi_register *field;
+	const struct vc4_hdmi_variant *variant = hdmi->variant;
+	void __iomem *base;
+
+	WARN_ON(pm_runtime_status_suspended(&hdmi->pdev->dev));
+
+	kunit_fail_current_test("Accessing an HDMI register in a unit test!\n");
+
+	if (reg >= variant->num_registers) {
+		dev_warn(&hdmi->pdev->dev,
+			 "Invalid register ID %u\n", reg);
+		return 0;
+	}
+
+	field = &variant->registers[reg];
+	base = __vc4_hdmi_get_field_base(hdmi, field->reg);
+	if (!base) {
+		dev_warn(&hdmi->pdev->dev,
+			 "Unknown register ID %u\n", reg);
+		return 0;
+	}
+
+	return readl(base + field->offset);
+}
+#define HDMI_READ(reg)		vc4_hdmi_read(vc4_hdmi, reg)
+
+static inline void vc4_hdmi_write(struct vc4_hdmi *hdmi,
+				  enum vc4_hdmi_field reg,
+				  u32 value)
+{
+	const struct vc4_hdmi_register *field;
+	const struct vc4_hdmi_variant *variant = hdmi->variant;
+	void __iomem *base;
+
+	lockdep_assert_held(&hdmi->hw_lock);
+
+	WARN_ON(pm_runtime_status_suspended(&hdmi->pdev->dev));
+
+	kunit_fail_current_test("Accessing an HDMI register in a unit test!\n");
+
+	if (reg >= variant->num_registers) {
+		dev_warn(&hdmi->pdev->dev,
+			 "Invalid register ID %u\n", reg);
+		return;
+	}
+
+	field = &variant->registers[reg];
+	base = __vc4_hdmi_get_field_base(hdmi, field->reg);
+	if (!base) {
+		dev_warn(&hdmi->pdev->dev,
+			 "Unknown register ID %u\n", reg);
+		return;
+	}
+
+	writel(value, base + field->offset);
+}
+#define HDMI_WRITE(reg, val)	vc4_hdmi_write(vc4_hdmi, reg, val)
+
+#endif /* _VC4_HDMI_REGS_H_ */

linux-7.2.3/drivers/gpu/drm/vc4/vc4_irq.c

@@ -0,0 +1,350 @@
+/*
+ * Copyright © 2014 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/**
+ * DOC: Interrupt management for the V3D engine
+ *
+ * We have an interrupt status register (V3D_INTCTL) which reports
+ * interrupts, and where writing 1 bits clears those interrupts.
+ * There are also a pair of interrupt registers
+ * (V3D_INTENA/V3D_INTDIS) where writing a 1 to their bits enables or
+ * disables that specific interrupt, and 0s written are ignored
+ * (reading either one returns the set of enabled interrupts).
+ *
+ * When we take a binning flush done interrupt, we need to submit the
+ * next frame for binning and move the finished frame to the render
+ * thread.
+ *
+ * When we take a render frame interrupt, we need to wake the
+ * processes waiting for some frame to be done, and get the next frame
+ * submitted ASAP (so the hardware doesn't sit idle when there's work
+ * to do).
+ *
+ * When we take the binner out of memory interrupt, we need to
+ * allocate some new memory and pass it to the binner so that the
+ * current job can make progress.
+ */
+
+#include <linux/platform_device.h>
+
+#include <drm/drm_print.h>
+
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+#include "vc4_trace.h"
+
+#define V3D_DRIVER_IRQS (V3D_INT_OUTOMEM | \
+			 V3D_INT_FLDONE | \
+			 V3D_INT_FRDONE)
+
+static void
+vc4_overflow_mem_work(struct work_struct *work)
+{
+	struct vc4_dev *vc4 =
+		container_of(work, struct vc4_dev, overflow_mem_work);
+	struct vc4_bo *bo;
+	int bin_bo_slot;
+	struct vc4_exec_info *exec;
+	unsigned long irqflags;
+
+	mutex_lock(&vc4->bin_bo_lock);
+
+	if (!vc4->bin_bo)
+		goto complete;
+
+	bo = vc4->bin_bo;
+
+	bin_bo_slot = vc4_v3d_get_bin_slot(vc4);
+	if (bin_bo_slot < 0) {
+		drm_err(&vc4->base, "Couldn't allocate binner overflow mem\n");
+		goto complete;
+	}
+
+	spin_lock_irqsave(&vc4->job_lock, irqflags);
+
+	if (vc4->bin_alloc_overflow) {
+		/* If we had overflow memory allocated previously,
+		 * then that chunk will free when the current bin job
+		 * is done.  If we don't have a bin job running, then
+		 * the chunk will be done whenever the list of render
+		 * jobs has drained.
+		 */
+		exec = vc4_first_bin_job(vc4);
+		if (!exec)
+			exec = vc4_last_render_job(vc4);
+		if (exec) {
+			exec->bin_slots |= vc4->bin_alloc_overflow;
+		} else {
+			/* There's nothing queued in the hardware, so
+			 * the old slot is free immediately.
+			 */
+			vc4->bin_alloc_used &= ~vc4->bin_alloc_overflow;
+		}
+	}
+	vc4->bin_alloc_overflow = BIT(bin_bo_slot);
+
+	V3D_WRITE(V3D_BPOA, bo->base.dma_addr + bin_bo_slot * vc4->bin_alloc_size);
+	V3D_WRITE(V3D_BPOS, vc4->bin_alloc_size);
+	V3D_WRITE(V3D_INTCTL, V3D_INT_OUTOMEM);
+	V3D_WRITE(V3D_INTENA, V3D_INT_OUTOMEM);
+	spin_unlock_irqrestore(&vc4->job_lock, irqflags);
+
+complete:
+	mutex_unlock(&vc4->bin_bo_lock);
+}
+
+static void
+vc4_irq_finish_bin_job(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_exec_info *next, *exec = vc4_first_bin_job(vc4);
+
+	if (!exec)
+		return;
+
+	trace_vc4_bcl_end_irq(dev, exec->seqno);
+
+	vc4_move_job_to_render(dev, exec);
+	next = vc4_first_bin_job(vc4);
+
+	/* Only submit the next job in the bin list if it matches the perfmon
+	 * attached to the one that just finished (or if both jobs don't have
+	 * perfmon attached to them).
+	 */
+	if (next && next->perfmon == exec->perfmon)
+		vc4_submit_next_bin_job(dev);
+}
+
+static void
+vc4_cancel_bin_job(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_exec_info *exec = vc4_first_bin_job(vc4);
+
+	if (!exec)
+		return;
+
+	/* Stop the perfmon so that the next bin job can be started. */
+	if (exec->perfmon)
+		vc4_perfmon_stop(vc4, exec->perfmon, false);
+
+	list_move_tail(&exec->head, &vc4->bin_job_list);
+	vc4_submit_next_bin_job(dev);
+}
+
+static void
+vc4_irq_finish_render_job(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_exec_info *exec = vc4_first_render_job(vc4);
+	struct vc4_exec_info *nextbin, *nextrender;
+
+	if (!exec)
+		return;
+
+	trace_vc4_rcl_end_irq(dev, exec->seqno);
+
+	vc4->finished_seqno++;
+	list_move_tail(&exec->head, &vc4->job_done_list);
+
+	nextbin = vc4_first_bin_job(vc4);
+	nextrender = vc4_first_render_job(vc4);
+
+	/* Only stop the perfmon if following jobs in the queue don't expect it
+	 * to be enabled.
+	 */
+	if (exec->perfmon && !nextrender &&
+	    (!nextbin || nextbin->perfmon != exec->perfmon))
+		vc4_perfmon_stop(vc4, exec->perfmon, true);
+
+	/* If there's a render job waiting, start it. If this is not the case
+	 * we may have to unblock the binner if it's been stalled because of
+	 * perfmon (this can be checked by comparing the perfmon attached to
+	 * the finished renderjob to the one attached to the next bin job: if
+	 * they don't match, this means the binner is stalled and should be
+	 * restarted).
+	 */
+	if (nextrender)
+		vc4_submit_next_render_job(dev);
+	else if (nextbin && nextbin->perfmon != exec->perfmon)
+		vc4_submit_next_bin_job(dev);
+
+	if (exec->fence) {
+		dma_fence_signal_locked(exec->fence);
+		dma_fence_put(exec->fence);
+		exec->fence = NULL;
+	}
+
+	wake_up_all(&vc4->job_wait_queue);
+	schedule_work(&vc4->job_done_work);
+}
+
+static irqreturn_t
+vc4_irq(int irq, void *arg)
+{
+	struct drm_device *dev = arg;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	uint32_t intctl;
+	irqreturn_t status = IRQ_NONE;
+
+	barrier();
+	intctl = V3D_READ(V3D_INTCTL);
+
+	/* Acknowledge the interrupts we're handling here. The binner
+	 * last flush / render frame done interrupt will be cleared,
+	 * while OUTOMEM will stay high until the underlying cause is
+	 * cleared.
+	 */
+	V3D_WRITE(V3D_INTCTL, intctl);
+
+	if (intctl & V3D_INT_OUTOMEM) {
+		/* Disable OUTOMEM until the work is done. */
+		V3D_WRITE(V3D_INTDIS, V3D_INT_OUTOMEM);
+		schedule_work(&vc4->overflow_mem_work);
+		status = IRQ_HANDLED;
+	}
+
+	if (intctl & V3D_INT_FLDONE) {
+		spin_lock(&vc4->job_lock);
+		vc4_irq_finish_bin_job(dev);
+		spin_unlock(&vc4->job_lock);
+		status = IRQ_HANDLED;
+	}
+
+	if (intctl & V3D_INT_FRDONE) {
+		spin_lock(&vc4->job_lock);
+		vc4_irq_finish_render_job(dev);
+		spin_unlock(&vc4->job_lock);
+		status = IRQ_HANDLED;
+	}
+
+	return status;
+}
+
+void
+vc4_irq_enable(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	if (!vc4->v3d)
+		return;
+
+	/* Enable the render done interrupts. The out-of-memory interrupt is
+	 * enabled as soon as we have a binner BO allocated.
+	 */
+	V3D_WRITE(V3D_INTENA, V3D_INT_FLDONE | V3D_INT_FRDONE);
+}
+
+void
+vc4_irq_disable(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	if (!vc4->v3d)
+		return;
+
+	/* Disable sending interrupts for our driver's IRQs. */
+	V3D_WRITE(V3D_INTDIS, V3D_DRIVER_IRQS);
+
+	/* Clear any pending interrupts we might have left. */
+	V3D_WRITE(V3D_INTCTL, V3D_DRIVER_IRQS);
+
+	/* Finish any interrupt handler still in flight. */
+	synchronize_irq(vc4->irq);
+
+	cancel_work_sync(&vc4->overflow_mem_work);
+}
+
+int vc4_irq_install(struct drm_device *dev, int irq)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	int ret;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	if (!vc4->v3d)
+		return -ENODEV;
+
+	if (irq == IRQ_NOTCONNECTED)
+		return -ENOTCONN;
+
+	init_waitqueue_head(&vc4->job_wait_queue);
+	INIT_WORK(&vc4->overflow_mem_work, vc4_overflow_mem_work);
+
+	/* Clear any pending interrupts someone might have left around
+	 * for us.
+	 */
+	V3D_WRITE(V3D_INTCTL, V3D_DRIVER_IRQS);
+
+	ret = devm_request_irq(dev->dev, irq, vc4_irq, 0,
+			       dev_name(dev->dev), dev);
+	if (ret)
+		return ret;
+
+	vc4_irq_enable(dev);
+
+	return 0;
+}
+
+void vc4_irq_uninstall(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	vc4_irq_disable(dev);
+}
+
+/** Reinitializes interrupt registers when a GPU reset is performed. */
+void vc4_irq_reset(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	unsigned long irqflags;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	/* Acknowledge any stale IRQs. */
+	V3D_WRITE(V3D_INTCTL, V3D_DRIVER_IRQS);
+
+	/*
+	 * Turn all our interrupts on.  Binner out of memory is the
+	 * only one we expect to trigger at this point, since we've
+	 * just come from poweron and haven't supplied any overflow
+	 * memory yet.
+	 */
+	V3D_WRITE(V3D_INTENA, V3D_DRIVER_IRQS);
+
+	spin_lock_irqsave(&vc4->job_lock, irqflags);
+	vc4_cancel_bin_job(dev);
+	vc4_irq_finish_render_job(dev);
+	spin_unlock_irqrestore(&vc4->job_lock, irqflags);
+}

linux-7.2.3/drivers/gpu/drm/vc4/vc4_packet.h

@@ -0,0 +1,399 @@
+/*
+ * Copyright © 2014 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef VC4_PACKET_H
+#define VC4_PACKET_H
+
+#include "vc4_regs.h" /* for VC4_MASK, VC4_GET_FIELD, VC4_SET_FIELD */
+
+enum vc4_packet {
+	VC4_PACKET_HALT = 0,
+	VC4_PACKET_NOP = 1,
+
+	VC4_PACKET_FLUSH = 4,
+	VC4_PACKET_FLUSH_ALL = 5,
+	VC4_PACKET_START_TILE_BINNING = 6,
+	VC4_PACKET_INCREMENT_SEMAPHORE = 7,
+	VC4_PACKET_WAIT_ON_SEMAPHORE = 8,
+
+	VC4_PACKET_BRANCH = 16,
+	VC4_PACKET_BRANCH_TO_SUB_LIST = 17,
+
+	VC4_PACKET_STORE_MS_TILE_BUFFER = 24,
+	VC4_PACKET_STORE_MS_TILE_BUFFER_AND_EOF = 25,
+	VC4_PACKET_STORE_FULL_RES_TILE_BUFFER = 26,
+	VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER = 27,
+	VC4_PACKET_STORE_TILE_BUFFER_GENERAL = 28,
+	VC4_PACKET_LOAD_TILE_BUFFER_GENERAL = 29,
+
+	VC4_PACKET_GL_INDEXED_PRIMITIVE = 32,
+	VC4_PACKET_GL_ARRAY_PRIMITIVE = 33,
+
+	VC4_PACKET_COMPRESSED_PRIMITIVE = 48,
+	VC4_PACKET_CLIPPED_COMPRESSED_PRIMITIVE = 49,
+
+	VC4_PACKET_PRIMITIVE_LIST_FORMAT = 56,
+
+	VC4_PACKET_GL_SHADER_STATE = 64,
+	VC4_PACKET_NV_SHADER_STATE = 65,
+	VC4_PACKET_VG_SHADER_STATE = 66,
+
+	VC4_PACKET_CONFIGURATION_BITS = 96,
+	VC4_PACKET_FLAT_SHADE_FLAGS = 97,
+	VC4_PACKET_POINT_SIZE = 98,
+	VC4_PACKET_LINE_WIDTH = 99,
+	VC4_PACKET_RHT_X_BOUNDARY = 100,
+	VC4_PACKET_DEPTH_OFFSET = 101,
+	VC4_PACKET_CLIP_WINDOW = 102,
+	VC4_PACKET_VIEWPORT_OFFSET = 103,
+	VC4_PACKET_Z_CLIPPING = 104,
+	VC4_PACKET_CLIPPER_XY_SCALING = 105,
+	VC4_PACKET_CLIPPER_Z_SCALING = 106,
+
+	VC4_PACKET_TILE_BINNING_MODE_CONFIG = 112,
+	VC4_PACKET_TILE_RENDERING_MODE_CONFIG = 113,
+	VC4_PACKET_CLEAR_COLORS = 114,
+	VC4_PACKET_TILE_COORDINATES = 115,
+
+	/* Not an actual hardware packet -- this is what we use to put
+	 * references to GEM bos in the command stream, since we need the u32
+	 * int the actual address packet in order to store the offset from the
+	 * start of the BO.
+	 */
+	VC4_PACKET_GEM_HANDLES = 254,
+} __attribute__ ((__packed__));
+
+#define VC4_PACKET_HALT_SIZE						1
+#define VC4_PACKET_NOP_SIZE						1
+#define VC4_PACKET_FLUSH_SIZE						1
+#define VC4_PACKET_FLUSH_ALL_SIZE					1
+#define VC4_PACKET_START_TILE_BINNING_SIZE				1
+#define VC4_PACKET_INCREMENT_SEMAPHORE_SIZE				1
+#define VC4_PACKET_WAIT_ON_SEMAPHORE_SIZE				1
+#define VC4_PACKET_BRANCH_SIZE						5
+#define VC4_PACKET_BRANCH_TO_SUB_LIST_SIZE				5
+#define VC4_PACKET_STORE_MS_TILE_BUFFER_SIZE				1
+#define VC4_PACKET_STORE_MS_TILE_BUFFER_AND_EOF_SIZE			1
+#define VC4_PACKET_STORE_FULL_RES_TILE_BUFFER_SIZE			5
+#define VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER_SIZE			5
+#define VC4_PACKET_STORE_TILE_BUFFER_GENERAL_SIZE			7
+#define VC4_PACKET_LOAD_TILE_BUFFER_GENERAL_SIZE			7
+#define VC4_PACKET_GL_INDEXED_PRIMITIVE_SIZE				14
+#define VC4_PACKET_GL_ARRAY_PRIMITIVE_SIZE				10
+#define VC4_PACKET_COMPRESSED_PRIMITIVE_SIZE				1
+#define VC4_PACKET_CLIPPED_COMPRESSED_PRIMITIVE_SIZE			1
+#define VC4_PACKET_PRIMITIVE_LIST_FORMAT_SIZE				2
+#define VC4_PACKET_GL_SHADER_STATE_SIZE					5
+#define VC4_PACKET_NV_SHADER_STATE_SIZE					5
+#define VC4_PACKET_VG_SHADER_STATE_SIZE					5
+#define VC4_PACKET_CONFIGURATION_BITS_SIZE				4
+#define VC4_PACKET_FLAT_SHADE_FLAGS_SIZE				5
+#define VC4_PACKET_POINT_SIZE_SIZE					5
+#define VC4_PACKET_LINE_WIDTH_SIZE					5
+#define VC4_PACKET_RHT_X_BOUNDARY_SIZE					3
+#define VC4_PACKET_DEPTH_OFFSET_SIZE					5
+#define VC4_PACKET_CLIP_WINDOW_SIZE					9
+#define VC4_PACKET_VIEWPORT_OFFSET_SIZE					5
+#define VC4_PACKET_Z_CLIPPING_SIZE					9
+#define VC4_PACKET_CLIPPER_XY_SCALING_SIZE				9
+#define VC4_PACKET_CLIPPER_Z_SCALING_SIZE				9
+#define VC4_PACKET_TILE_BINNING_MODE_CONFIG_SIZE			16
+#define VC4_PACKET_TILE_RENDERING_MODE_CONFIG_SIZE			11
+#define VC4_PACKET_CLEAR_COLORS_SIZE					14
+#define VC4_PACKET_TILE_COORDINATES_SIZE				3
+#define VC4_PACKET_GEM_HANDLES_SIZE					9
+
+/* Number of multisamples supported. */
+#define VC4_MAX_SAMPLES							4
+/* Size of a full resolution color or Z tile buffer load/store. */
+#define VC4_TILE_BUFFER_SIZE			(64 * 64 * 4)
+
+/** @{
+ * Bits used by packets like VC4_PACKET_STORE_TILE_BUFFER_GENERAL and
+ * VC4_PACKET_TILE_RENDERING_MODE_CONFIG.
+*/
+#define VC4_TILING_FORMAT_LINEAR    0
+#define VC4_TILING_FORMAT_T         1
+#define VC4_TILING_FORMAT_LT        2
+/** @} */
+
+/** @{
+ *
+ * low bits of VC4_PACKET_STORE_FULL_RES_TILE_BUFFER and
+ * VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER.
+ */
+#define VC4_LOADSTORE_FULL_RES_EOF                     BIT(3)
+#define VC4_LOADSTORE_FULL_RES_DISABLE_CLEAR_ALL       BIT(2)
+#define VC4_LOADSTORE_FULL_RES_DISABLE_ZS              BIT(1)
+#define VC4_LOADSTORE_FULL_RES_DISABLE_COLOR           BIT(0)
+
+/** @{
+ *
+ * low bits of VC4_PACKET_STORE_FULL_RES_TILE_BUFFER and
+ * VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER.
+ */
+#define VC4_LOADSTORE_FULL_RES_EOF                     BIT(3)
+#define VC4_LOADSTORE_FULL_RES_DISABLE_CLEAR_ALL       BIT(2)
+#define VC4_LOADSTORE_FULL_RES_DISABLE_ZS              BIT(1)
+#define VC4_LOADSTORE_FULL_RES_DISABLE_COLOR           BIT(0)
+
+/** @{
+ *
+ * byte 2 of VC4_PACKET_STORE_TILE_BUFFER_GENERAL and
+ * VC4_PACKET_LOAD_TILE_BUFFER_GENERAL (low bits of the address)
+ */
+
+#define VC4_LOADSTORE_TILE_BUFFER_EOF                  BIT(3)
+#define VC4_LOADSTORE_TILE_BUFFER_DISABLE_FULL_VG_MASK BIT(2)
+#define VC4_LOADSTORE_TILE_BUFFER_DISABLE_FULL_ZS      BIT(1)
+#define VC4_LOADSTORE_TILE_BUFFER_DISABLE_FULL_COLOR   BIT(0)
+
+/** @} */
+
+/** @{
+ *
+ * byte 0-1 of VC4_PACKET_STORE_TILE_BUFFER_GENERAL and
+ * VC4_PACKET_LOAD_TILE_BUFFER_GENERAL
+ */
+#define VC4_STORE_TILE_BUFFER_DISABLE_VG_MASK_CLEAR BIT(15)
+#define VC4_STORE_TILE_BUFFER_DISABLE_ZS_CLEAR     BIT(14)
+#define VC4_STORE_TILE_BUFFER_DISABLE_COLOR_CLEAR  BIT(13)
+#define VC4_STORE_TILE_BUFFER_DISABLE_SWAP         BIT(12)
+
+#define VC4_LOADSTORE_TILE_BUFFER_FORMAT_MASK      VC4_MASK(9, 8)
+#define VC4_LOADSTORE_TILE_BUFFER_FORMAT_SHIFT     8
+#define VC4_LOADSTORE_TILE_BUFFER_RGBA8888         0
+#define VC4_LOADSTORE_TILE_BUFFER_BGR565_DITHER    1
+#define VC4_LOADSTORE_TILE_BUFFER_BGR565           2
+/** @} */
+
+/** @{
+ *
+ * byte 0 of VC4_PACKET_STORE_TILE_BUFFER_GENERAL and
+ * VC4_PACKET_LOAD_TILE_BUFFER_GENERAL
+ */
+#define VC4_STORE_TILE_BUFFER_MODE_MASK            VC4_MASK(7, 6)
+#define VC4_STORE_TILE_BUFFER_MODE_SHIFT           6
+#define VC4_STORE_TILE_BUFFER_MODE_SAMPLE0         (0 << 6)
+#define VC4_STORE_TILE_BUFFER_MODE_DECIMATE_X4     (1 << 6)
+#define VC4_STORE_TILE_BUFFER_MODE_DECIMATE_X16    (2 << 6)
+
+/** The values of the field are VC4_TILING_FORMAT_* */
+#define VC4_LOADSTORE_TILE_BUFFER_TILING_MASK      VC4_MASK(5, 4)
+#define VC4_LOADSTORE_TILE_BUFFER_TILING_SHIFT     4
+
+#define VC4_LOADSTORE_TILE_BUFFER_BUFFER_MASK      VC4_MASK(2, 0)
+#define VC4_LOADSTORE_TILE_BUFFER_BUFFER_SHIFT     0
+#define VC4_LOADSTORE_TILE_BUFFER_NONE             0
+#define VC4_LOADSTORE_TILE_BUFFER_COLOR            1
+#define VC4_LOADSTORE_TILE_BUFFER_ZS               2
+#define VC4_LOADSTORE_TILE_BUFFER_Z                3
+#define VC4_LOADSTORE_TILE_BUFFER_VG_MASK          4
+#define VC4_LOADSTORE_TILE_BUFFER_FULL             5
+/** @} */
+
+#define VC4_INDEX_BUFFER_U8                        (0 << 4)
+#define VC4_INDEX_BUFFER_U16                       (1 << 4)
+
+/* This flag is only present in NV shader state. */
+#define VC4_SHADER_FLAG_SHADED_CLIP_COORDS         BIT(3)
+#define VC4_SHADER_FLAG_ENABLE_CLIPPING            BIT(2)
+#define VC4_SHADER_FLAG_VS_POINT_SIZE              BIT(1)
+#define VC4_SHADER_FLAG_FS_SINGLE_THREAD           BIT(0)
+
+/** @{ byte 2 of config bits. */
+#define VC4_CONFIG_BITS_EARLY_Z_UPDATE             BIT(1)
+#define VC4_CONFIG_BITS_EARLY_Z                    BIT(0)
+/** @} */
+
+/** @{ byte 1 of config bits. */
+#define VC4_CONFIG_BITS_Z_UPDATE                   BIT(7)
+/** same values in this 3-bit field as PIPE_FUNC_* */
+#define VC4_CONFIG_BITS_DEPTH_FUNC_SHIFT           4
+#define VC4_CONFIG_BITS_COVERAGE_READ_LEAVE        BIT(3)
+
+#define VC4_CONFIG_BITS_COVERAGE_UPDATE_NONZERO    (0 << 1)
+#define VC4_CONFIG_BITS_COVERAGE_UPDATE_ODD        (1 << 1)
+#define VC4_CONFIG_BITS_COVERAGE_UPDATE_OR         (2 << 1)
+#define VC4_CONFIG_BITS_COVERAGE_UPDATE_ZERO       (3 << 1)
+
+#define VC4_CONFIG_BITS_COVERAGE_PIPE_SELECT       BIT(0)
+/** @} */
+
+/** @{ byte 0 of config bits. */
+#define VC4_CONFIG_BITS_RASTERIZER_OVERSAMPLE_NONE (0 << 6)
+#define VC4_CONFIG_BITS_RASTERIZER_OVERSAMPLE_4X   (1 << 6)
+#define VC4_CONFIG_BITS_RASTERIZER_OVERSAMPLE_16X  (2 << 6)
+
+#define VC4_CONFIG_BITS_AA_POINTS_AND_LINES        BIT(4)
+#define VC4_CONFIG_BITS_ENABLE_DEPTH_OFFSET        BIT(3)
+#define VC4_CONFIG_BITS_CW_PRIMITIVES              BIT(2)
+#define VC4_CONFIG_BITS_ENABLE_PRIM_BACK           BIT(1)
+#define VC4_CONFIG_BITS_ENABLE_PRIM_FRONT          BIT(0)
+/** @} */
+
+/** @{ bits in the last u8 of VC4_PACKET_TILE_BINNING_MODE_CONFIG */
+#define VC4_BIN_CONFIG_DB_NON_MS                   BIT(7)
+
+#define VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_MASK       VC4_MASK(6, 5)
+#define VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_SHIFT      5
+#define VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_32         0
+#define VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_64         1
+#define VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_128        2
+#define VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_256        3
+
+#define VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_MASK  VC4_MASK(4, 3)
+#define VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_SHIFT 3
+#define VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_32    0
+#define VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_64    1
+#define VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_128   2
+#define VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_256   3
+
+#define VC4_BIN_CONFIG_AUTO_INIT_TSDA              BIT(2)
+#define VC4_BIN_CONFIG_TILE_BUFFER_64BIT           BIT(1)
+#define VC4_BIN_CONFIG_MS_MODE_4X                  BIT(0)
+/** @} */
+
+/** @{ bits in the last u16 of VC4_PACKET_TILE_RENDERING_MODE_CONFIG */
+#define VC4_RENDER_CONFIG_DB_NON_MS                BIT(12)
+#define VC4_RENDER_CONFIG_EARLY_Z_COVERAGE_DISABLE BIT(11)
+#define VC4_RENDER_CONFIG_EARLY_Z_DIRECTION_G      BIT(10)
+#define VC4_RENDER_CONFIG_COVERAGE_MODE            BIT(9)
+#define VC4_RENDER_CONFIG_ENABLE_VG_MASK           BIT(8)
+
+/** The values of the field are VC4_TILING_FORMAT_* */
+#define VC4_RENDER_CONFIG_MEMORY_FORMAT_MASK       VC4_MASK(7, 6)
+#define VC4_RENDER_CONFIG_MEMORY_FORMAT_SHIFT      6
+
+#define VC4_RENDER_CONFIG_DECIMATE_MODE_1X         (0 << 4)
+#define VC4_RENDER_CONFIG_DECIMATE_MODE_4X         (1 << 4)
+#define VC4_RENDER_CONFIG_DECIMATE_MODE_16X        (2 << 4)
+
+#define VC4_RENDER_CONFIG_FORMAT_MASK              VC4_MASK(3, 2)
+#define VC4_RENDER_CONFIG_FORMAT_SHIFT             2
+#define VC4_RENDER_CONFIG_FORMAT_BGR565_DITHERED   0
+#define VC4_RENDER_CONFIG_FORMAT_RGBA8888          1
+#define VC4_RENDER_CONFIG_FORMAT_BGR565            2
+
+#define VC4_RENDER_CONFIG_TILE_BUFFER_64BIT        BIT(1)
+#define VC4_RENDER_CONFIG_MS_MODE_4X               BIT(0)
+
+#define VC4_PRIMITIVE_LIST_FORMAT_16_INDEX         (1 << 4)
+#define VC4_PRIMITIVE_LIST_FORMAT_32_XY            (3 << 4)
+#define VC4_PRIMITIVE_LIST_FORMAT_TYPE_POINTS      (0 << 0)
+#define VC4_PRIMITIVE_LIST_FORMAT_TYPE_LINES       (1 << 0)
+#define VC4_PRIMITIVE_LIST_FORMAT_TYPE_TRIANGLES   (2 << 0)
+#define VC4_PRIMITIVE_LIST_FORMAT_TYPE_RHT         (3 << 0)
+
+enum vc4_texture_data_type {
+	VC4_TEXTURE_TYPE_RGBA8888 = 0,
+	VC4_TEXTURE_TYPE_RGBX8888 = 1,
+	VC4_TEXTURE_TYPE_RGBA4444 = 2,
+	VC4_TEXTURE_TYPE_RGBA5551 = 3,
+	VC4_TEXTURE_TYPE_RGB565 = 4,
+	VC4_TEXTURE_TYPE_LUMINANCE = 5,
+	VC4_TEXTURE_TYPE_ALPHA = 6,
+	VC4_TEXTURE_TYPE_LUMALPHA = 7,
+	VC4_TEXTURE_TYPE_ETC1 = 8,
+	VC4_TEXTURE_TYPE_S16F = 9,
+	VC4_TEXTURE_TYPE_S8 = 10,
+	VC4_TEXTURE_TYPE_S16 = 11,
+	VC4_TEXTURE_TYPE_BW1 = 12,
+	VC4_TEXTURE_TYPE_A4 = 13,
+	VC4_TEXTURE_TYPE_A1 = 14,
+	VC4_TEXTURE_TYPE_RGBA64 = 15,
+	VC4_TEXTURE_TYPE_RGBA32R = 16,
+	VC4_TEXTURE_TYPE_YUV422R = 17,
+};
+
+#define VC4_TEX_P0_OFFSET_MASK                     VC4_MASK(31, 12)
+#define VC4_TEX_P0_OFFSET_SHIFT                    12
+#define VC4_TEX_P0_CSWIZ_MASK                      VC4_MASK(11, 10)
+#define VC4_TEX_P0_CSWIZ_SHIFT                     10
+#define VC4_TEX_P0_CMMODE_MASK                     VC4_MASK(9, 9)
+#define VC4_TEX_P0_CMMODE_SHIFT                    9
+#define VC4_TEX_P0_FLIPY_MASK                      VC4_MASK(8, 8)
+#define VC4_TEX_P0_FLIPY_SHIFT                     8
+#define VC4_TEX_P0_TYPE_MASK                       VC4_MASK(7, 4)
+#define VC4_TEX_P0_TYPE_SHIFT                      4
+#define VC4_TEX_P0_MIPLVLS_MASK                    VC4_MASK(3, 0)
+#define VC4_TEX_P0_MIPLVLS_SHIFT                   0
+
+#define VC4_TEX_P1_TYPE4_MASK                      VC4_MASK(31, 31)
+#define VC4_TEX_P1_TYPE4_SHIFT                     31
+#define VC4_TEX_P1_HEIGHT_MASK                     VC4_MASK(30, 20)
+#define VC4_TEX_P1_HEIGHT_SHIFT                    20
+#define VC4_TEX_P1_ETCFLIP_MASK                    VC4_MASK(19, 19)
+#define VC4_TEX_P1_ETCFLIP_SHIFT                   19
+#define VC4_TEX_P1_WIDTH_MASK                      VC4_MASK(18, 8)
+#define VC4_TEX_P1_WIDTH_SHIFT                     8
+
+#define VC4_TEX_P1_MAGFILT_MASK                    VC4_MASK(7, 7)
+#define VC4_TEX_P1_MAGFILT_SHIFT                   7
+# define VC4_TEX_P1_MAGFILT_LINEAR                 0
+# define VC4_TEX_P1_MAGFILT_NEAREST                1
+
+#define VC4_TEX_P1_MINFILT_MASK                    VC4_MASK(6, 4)
+#define VC4_TEX_P1_MINFILT_SHIFT                   4
+# define VC4_TEX_P1_MINFILT_LINEAR                 0
+# define VC4_TEX_P1_MINFILT_NEAREST                1
+# define VC4_TEX_P1_MINFILT_NEAR_MIP_NEAR          2
+# define VC4_TEX_P1_MINFILT_NEAR_MIP_LIN           3
+# define VC4_TEX_P1_MINFILT_LIN_MIP_NEAR           4
+# define VC4_TEX_P1_MINFILT_LIN_MIP_LIN            5
+
+#define VC4_TEX_P1_WRAP_T_MASK                     VC4_MASK(3, 2)
+#define VC4_TEX_P1_WRAP_T_SHIFT                    2
+#define VC4_TEX_P1_WRAP_S_MASK                     VC4_MASK(1, 0)
+#define VC4_TEX_P1_WRAP_S_SHIFT                    0
+# define VC4_TEX_P1_WRAP_REPEAT                    0
+# define VC4_TEX_P1_WRAP_CLAMP                     1
+# define VC4_TEX_P1_WRAP_MIRROR                    2
+# define VC4_TEX_P1_WRAP_BORDER                    3
+
+#define VC4_TEX_P2_PTYPE_MASK                      VC4_MASK(31, 30)
+#define VC4_TEX_P2_PTYPE_SHIFT                     30
+# define VC4_TEX_P2_PTYPE_IGNORED                  0
+# define VC4_TEX_P2_PTYPE_CUBE_MAP_STRIDE          1
+# define VC4_TEX_P2_PTYPE_CHILD_IMAGE_DIMENSIONS   2
+# define VC4_TEX_P2_PTYPE_CHILD_IMAGE_OFFSETS      3
+
+/* VC4_TEX_P2_PTYPE_CUBE_MAP_STRIDE bits */
+#define VC4_TEX_P2_CMST_MASK                       VC4_MASK(29, 12)
+#define VC4_TEX_P2_CMST_SHIFT                      12
+#define VC4_TEX_P2_BSLOD_MASK                      VC4_MASK(0, 0)
+#define VC4_TEX_P2_BSLOD_SHIFT                     0
+
+/* VC4_TEX_P2_PTYPE_CHILD_IMAGE_DIMENSIONS */
+#define VC4_TEX_P2_CHEIGHT_MASK                    VC4_MASK(22, 12)
+#define VC4_TEX_P2_CHEIGHT_SHIFT                   12
+#define VC4_TEX_P2_CWIDTH_MASK                     VC4_MASK(10, 0)
+#define VC4_TEX_P2_CWIDTH_SHIFT                    0
+
+/* VC4_TEX_P2_PTYPE_CHILD_IMAGE_OFFSETS */
+#define VC4_TEX_P2_CYOFF_MASK                      VC4_MASK(22, 12)
+#define VC4_TEX_P2_CYOFF_SHIFT                     12
+#define VC4_TEX_P2_CXOFF_MASK                      VC4_MASK(10, 0)
+#define VC4_TEX_P2_CXOFF_SHIFT                     0
+
+#endif /* VC4_PACKET_H */

linux-7.2.3/drivers/gpu/drm/vc4/vc4_perfmon.c

@@ -0,0 +1,248 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2018 Broadcom
+ */
+
+/**
+ * DOC: VC4 V3D performance monitor module
+ *
+ * The V3D block provides 16 hardware counters which can count various events.
+ */
+
+#include <drm/drm_print.h>
+
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+
+void vc4_perfmon_get(struct vc4_perfmon *perfmon)
+{
+	struct vc4_dev *vc4;
+
+	if (!perfmon)
+		return;
+
+	vc4 = perfmon->dev;
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	refcount_inc(&perfmon->refcnt);
+}
+
+void vc4_perfmon_put(struct vc4_perfmon *perfmon)
+{
+	struct vc4_dev *vc4;
+
+	if (!perfmon)
+		return;
+
+	vc4 = perfmon->dev;
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	if (refcount_dec_and_test(&perfmon->refcnt))
+		kfree(perfmon);
+}
+
+void vc4_perfmon_start(struct vc4_dev *vc4, struct vc4_perfmon *perfmon)
+{
+	unsigned int i;
+	u32 mask;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	if (WARN_ON_ONCE(!perfmon || vc4->active_perfmon))
+		return;
+
+	for (i = 0; i < perfmon->ncounters; i++)
+		V3D_WRITE(V3D_PCTRS(i), perfmon->events[i]);
+
+	mask = GENMASK(perfmon->ncounters - 1, 0);
+	V3D_WRITE(V3D_PCTRC, mask);
+	V3D_WRITE(V3D_PCTRE, V3D_PCTRE_EN | mask);
+	vc4->active_perfmon = perfmon;
+}
+
+void vc4_perfmon_stop(struct vc4_dev *vc4, struct vc4_perfmon *perfmon,
+		      bool capture)
+{
+	unsigned int i;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	if (WARN_ON_ONCE(!vc4->active_perfmon ||
+			 perfmon != vc4->active_perfmon))
+		return;
+
+	if (capture) {
+		for (i = 0; i < perfmon->ncounters; i++)
+			perfmon->counters[i] += V3D_READ(V3D_PCTR(i));
+	}
+
+	V3D_WRITE(V3D_PCTRE, 0);
+	vc4->active_perfmon = NULL;
+}
+
+struct vc4_perfmon *vc4_perfmon_find(struct vc4_file *vc4file, int id)
+{
+	struct vc4_dev *vc4 = vc4file->dev;
+	struct vc4_perfmon *perfmon;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return NULL;
+
+	xa_lock(&vc4file->perfmons);
+	perfmon = xa_load(&vc4file->perfmons, id);
+	vc4_perfmon_get(perfmon);
+	xa_unlock(&vc4file->perfmons);
+
+	return perfmon;
+}
+
+void vc4_perfmon_open_file(struct vc4_file *vc4file)
+{
+	struct vc4_dev *vc4 = vc4file->dev;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	xa_init_flags(&vc4file->perfmons, XA_FLAGS_ALLOC1);
+}
+
+static void vc4_perfmon_delete(struct vc4_file *vc4file,
+			       struct vc4_perfmon *perfmon)
+{
+	struct vc4_dev *vc4 = vc4file->dev;
+
+	/* If the active perfmon is being destroyed, stop it first */
+	if (perfmon == vc4->active_perfmon)
+		vc4_perfmon_stop(vc4, perfmon, false);
+
+	vc4_perfmon_put(perfmon);
+}
+
+void vc4_perfmon_close_file(struct vc4_file *vc4file)
+{
+	struct vc4_dev *vc4 = vc4file->dev;
+	struct vc4_perfmon *perfmon;
+	unsigned long id;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	xa_for_each(&vc4file->perfmons, id, perfmon)
+		vc4_perfmon_delete(vc4file, perfmon);
+
+	xa_destroy(&vc4file->perfmons);
+}
+
+int vc4_perfmon_create_ioctl(struct drm_device *dev, void *data,
+			     struct drm_file *file_priv)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_file *vc4file = file_priv->driver_priv;
+	struct drm_vc4_perfmon_create *req = data;
+	struct vc4_perfmon *perfmon;
+	unsigned int i;
+	int ret;
+	u32 id;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	if (!vc4->v3d) {
+		DRM_DEBUG("Creating perfmon no VC4 V3D probed\n");
+		return -ENODEV;
+	}
+
+	/* Number of monitored counters cannot exceed HW limits. */
+	if (req->ncounters > DRM_VC4_MAX_PERF_COUNTERS ||
+	    !req->ncounters)
+		return -EINVAL;
+
+	/* Make sure all events are valid. */
+	for (i = 0; i < req->ncounters; i++) {
+		if (req->events[i] >= VC4_PERFCNT_NUM_EVENTS)
+			return -EINVAL;
+	}
+
+	perfmon = kzalloc_flex(*perfmon, counters, req->ncounters);
+	if (!perfmon)
+		return -ENOMEM;
+	perfmon->dev = vc4;
+
+	for (i = 0; i < req->ncounters; i++)
+		perfmon->events[i] = req->events[i];
+
+	perfmon->ncounters = req->ncounters;
+
+	refcount_set(&perfmon->refcnt, 1);
+
+	ret = xa_alloc(&vc4file->perfmons, &id, perfmon, xa_limit_32b,
+		       GFP_KERNEL);
+	if (ret < 0) {
+		kfree(perfmon);
+		return ret;
+	}
+
+	req->id = id;
+
+	return 0;
+}
+
+int vc4_perfmon_destroy_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *file_priv)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_file *vc4file = file_priv->driver_priv;
+	struct drm_vc4_perfmon_destroy *req = data;
+	struct vc4_perfmon *perfmon;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	if (!vc4->v3d) {
+		DRM_DEBUG("Destroying perfmon no VC4 V3D probed\n");
+		return -ENODEV;
+	}
+
+	perfmon = xa_erase(&vc4file->perfmons, req->id);
+	if (!perfmon)
+		return -EINVAL;
+
+	vc4_perfmon_delete(vc4file, perfmon);
+
+	return 0;
+}
+
+int vc4_perfmon_get_values_ioctl(struct drm_device *dev, void *data,
+				 struct drm_file *file_priv)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_file *vc4file = file_priv->driver_priv;
+	struct drm_vc4_perfmon_get_values *req = data;
+	struct vc4_perfmon *perfmon;
+	int ret;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	if (!vc4->v3d) {
+		DRM_DEBUG("Getting perfmon no VC4 V3D probed\n");
+		return -ENODEV;
+	}
+
+	perfmon = vc4_perfmon_find(vc4file, req->id);
+	if (!perfmon)
+		return -EINVAL;
+
+	if (copy_to_user(u64_to_user_ptr(req->values_ptr), perfmon->counters,
+			 perfmon->ncounters * sizeof(u64)))
+		ret = -EFAULT;
+	else
+		ret = 0;
+
+	vc4_perfmon_put(perfmon);
+	return ret;
+}

linux-7.2.3/drivers/gpu/drm/vc4/vc4_qpu_defines.h

@@ -0,0 +1,279 @@
+/*
+ * Copyright © 2014 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef VC4_QPU_DEFINES_H
+#define VC4_QPU_DEFINES_H
+
+enum qpu_op_add {
+	QPU_A_NOP,
+	QPU_A_FADD,
+	QPU_A_FSUB,
+	QPU_A_FMIN,
+	QPU_A_FMAX,
+	QPU_A_FMINABS,
+	QPU_A_FMAXABS,
+	QPU_A_FTOI,
+	QPU_A_ITOF,
+	QPU_A_ADD = 12,
+	QPU_A_SUB,
+	QPU_A_SHR,
+	QPU_A_ASR,
+	QPU_A_ROR,
+	QPU_A_SHL,
+	QPU_A_MIN,
+	QPU_A_MAX,
+	QPU_A_AND,
+	QPU_A_OR,
+	QPU_A_XOR,
+	QPU_A_NOT,
+	QPU_A_CLZ,
+	QPU_A_V8ADDS = 30,
+	QPU_A_V8SUBS = 31,
+};
+
+enum qpu_op_mul {
+	QPU_M_NOP,
+	QPU_M_FMUL,
+	QPU_M_MUL24,
+	QPU_M_V8MULD,
+	QPU_M_V8MIN,
+	QPU_M_V8MAX,
+	QPU_M_V8ADDS,
+	QPU_M_V8SUBS,
+};
+
+enum qpu_raddr {
+	QPU_R_FRAG_PAYLOAD_ZW = 15, /* W for A file, Z for B file */
+	/* 0-31 are the plain regfile a or b fields */
+	QPU_R_UNIF = 32,
+	QPU_R_VARY = 35,
+	QPU_R_ELEM_QPU = 38,
+	QPU_R_NOP,
+	QPU_R_XY_PIXEL_COORD = 41,
+	QPU_R_MS_REV_FLAGS = 42,
+	QPU_R_VPM = 48,
+	QPU_R_VPM_LD_BUSY,
+	QPU_R_VPM_LD_WAIT,
+	QPU_R_MUTEX_ACQUIRE,
+};
+
+enum qpu_waddr {
+	/* 0-31 are the plain regfile a or b fields */
+	QPU_W_ACC0 = 32, /* aka r0 */
+	QPU_W_ACC1,
+	QPU_W_ACC2,
+	QPU_W_ACC3,
+	QPU_W_TMU_NOSWAP,
+	QPU_W_ACC5,
+	QPU_W_HOST_INT,
+	QPU_W_NOP,
+	QPU_W_UNIFORMS_ADDRESS,
+	QPU_W_QUAD_XY, /* X for regfile a, Y for regfile b */
+	QPU_W_MS_FLAGS = 42,
+	QPU_W_REV_FLAG = 42,
+	QPU_W_TLB_STENCIL_SETUP = 43,
+	QPU_W_TLB_Z,
+	QPU_W_TLB_COLOR_MS,
+	QPU_W_TLB_COLOR_ALL,
+	QPU_W_TLB_ALPHA_MASK,
+	QPU_W_VPM,
+	QPU_W_VPMVCD_SETUP, /* LD for regfile a, ST for regfile b */
+	QPU_W_VPM_ADDR, /* LD for regfile a, ST for regfile b */
+	QPU_W_MUTEX_RELEASE,
+	QPU_W_SFU_RECIP,
+	QPU_W_SFU_RECIPSQRT,
+	QPU_W_SFU_EXP,
+	QPU_W_SFU_LOG,
+	QPU_W_TMU0_S,
+	QPU_W_TMU0_T,
+	QPU_W_TMU0_R,
+	QPU_W_TMU0_B,
+	QPU_W_TMU1_S,
+	QPU_W_TMU1_T,
+	QPU_W_TMU1_R,
+	QPU_W_TMU1_B,
+};
+
+enum qpu_sig_bits {
+	QPU_SIG_SW_BREAKPOINT,
+	QPU_SIG_NONE,
+	QPU_SIG_THREAD_SWITCH,
+	QPU_SIG_PROG_END,
+	QPU_SIG_WAIT_FOR_SCOREBOARD,
+	QPU_SIG_SCOREBOARD_UNLOCK,
+	QPU_SIG_LAST_THREAD_SWITCH,
+	QPU_SIG_COVERAGE_LOAD,
+	QPU_SIG_COLOR_LOAD,
+	QPU_SIG_COLOR_LOAD_END,
+	QPU_SIG_LOAD_TMU0,
+	QPU_SIG_LOAD_TMU1,
+	QPU_SIG_ALPHA_MASK_LOAD,
+	QPU_SIG_SMALL_IMM,
+	QPU_SIG_LOAD_IMM,
+	QPU_SIG_BRANCH
+};
+
+enum qpu_mux {
+	/* hardware mux values */
+	QPU_MUX_R0,
+	QPU_MUX_R1,
+	QPU_MUX_R2,
+	QPU_MUX_R3,
+	QPU_MUX_R4,
+	QPU_MUX_R5,
+	QPU_MUX_A,
+	QPU_MUX_B,
+
+	/* non-hardware mux values */
+	QPU_MUX_IMM,
+};
+
+enum qpu_cond {
+	QPU_COND_NEVER,
+	QPU_COND_ALWAYS,
+	QPU_COND_ZS,
+	QPU_COND_ZC,
+	QPU_COND_NS,
+	QPU_COND_NC,
+	QPU_COND_CS,
+	QPU_COND_CC,
+};
+
+enum qpu_pack_mul {
+	QPU_PACK_MUL_NOP,
+	/* replicated to each 8 bits of the 32-bit dst. */
+	QPU_PACK_MUL_8888 = 3,
+	QPU_PACK_MUL_8A,
+	QPU_PACK_MUL_8B,
+	QPU_PACK_MUL_8C,
+	QPU_PACK_MUL_8D,
+};
+
+enum qpu_pack_a {
+	QPU_PACK_A_NOP,
+	/* convert to 16 bit float if float input, or to int16. */
+	QPU_PACK_A_16A,
+	QPU_PACK_A_16B,
+	/* replicated to each 8 bits of the 32-bit dst. */
+	QPU_PACK_A_8888,
+	/* Convert to 8-bit unsigned int. */
+	QPU_PACK_A_8A,
+	QPU_PACK_A_8B,
+	QPU_PACK_A_8C,
+	QPU_PACK_A_8D,
+
+	/* Saturating variants of the previous instructions. */
+	QPU_PACK_A_32_SAT, /* int-only */
+	QPU_PACK_A_16A_SAT, /* int or float */
+	QPU_PACK_A_16B_SAT,
+	QPU_PACK_A_8888_SAT,
+	QPU_PACK_A_8A_SAT,
+	QPU_PACK_A_8B_SAT,
+	QPU_PACK_A_8C_SAT,
+	QPU_PACK_A_8D_SAT,
+};
+
+enum qpu_unpack_r4 {
+	QPU_UNPACK_R4_NOP,
+	QPU_UNPACK_R4_F16A_TO_F32,
+	QPU_UNPACK_R4_F16B_TO_F32,
+	QPU_UNPACK_R4_8D_REP,
+	QPU_UNPACK_R4_8A,
+	QPU_UNPACK_R4_8B,
+	QPU_UNPACK_R4_8C,
+	QPU_UNPACK_R4_8D,
+};
+
+#define QPU_MASK(high, low) \
+	((((uint64_t)1 << ((high) - (low) + 1)) - 1) << (low))
+
+#define QPU_GET_FIELD(word, field) \
+	((uint32_t)(((word)  & field ## _MASK) >> field ## _SHIFT))
+
+#define QPU_SIG_SHIFT                   60
+#define QPU_SIG_MASK                    QPU_MASK(63, 60)
+
+#define QPU_UNPACK_SHIFT                57
+#define QPU_UNPACK_MASK                 QPU_MASK(59, 57)
+
+/**
+ * If set, the pack field means PACK_MUL or R4 packing, instead of normal
+ * regfile a packing.
+ */
+#define QPU_PM                          ((uint64_t)1 << 56)
+
+#define QPU_PACK_SHIFT                  52
+#define QPU_PACK_MASK                   QPU_MASK(55, 52)
+
+#define QPU_COND_ADD_SHIFT              49
+#define QPU_COND_ADD_MASK               QPU_MASK(51, 49)
+#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
+#define QPU_WADDR_ADD_MASK              QPU_MASK(43, 38)
+#define QPU_WADDR_MUL_SHIFT             32
+#define QPU_WADDR_MUL_MASK              QPU_MASK(37, 32)
+
+#define QPU_OP_MUL_SHIFT                29
+#define QPU_OP_MUL_MASK                 QPU_MASK(31, 29)
+
+#define QPU_RADDR_A_SHIFT               18
+#define QPU_RADDR_A_MASK                QPU_MASK(23, 18)
+#define QPU_RADDR_B_SHIFT               12
+#define QPU_RADDR_B_MASK                QPU_MASK(17, 12)
+#define QPU_SMALL_IMM_SHIFT             12
+#define QPU_SMALL_IMM_MASK              QPU_MASK(17, 12)
+
+#define QPU_ADD_A_SHIFT                 9
+#define QPU_ADD_A_MASK                  QPU_MASK(11, 9)
+#define QPU_ADD_B_SHIFT                 6
+#define QPU_ADD_B_MASK                  QPU_MASK(8, 6)
+#define QPU_MUL_A_SHIFT                 3
+#define QPU_MUL_A_MASK                  QPU_MASK(5, 3)
+#define QPU_MUL_B_SHIFT                 0
+#define QPU_MUL_B_MASK                  QPU_MASK(2, 0)
+
+#define QPU_WS                          ((uint64_t)1 << 44)
+
+#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 */

linux-7.2.3/drivers/gpu/drm/vc4/vc4_render_cl.c

@@ -0,0 +1,665 @@
+/*
+ * Copyright © 2014-2015 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/**
+ * DOC: Render command list generation
+ *
+ * In the V3D hardware, render command lists are what load and store
+ * tiles of a framebuffer and optionally call out to binner-generated
+ * command lists to do the 3D drawing for that tile.
+ *
+ * In the VC4 driver, render command list generation is performed by the
+ * kernel instead of userspace.  We do this because validating a
+ * user-submitted command list is hard to get right and has high CPU overhead,
+ * while the number of valid configurations for render command lists is
+ * actually fairly low.
+ */
+
+#include <drm/drm_print.h>
+
+#include "vc4_drv.h"
+#include "vc4_packet.h"
+
+struct vc4_rcl_setup {
+	struct drm_gem_dma_object *color_read;
+	struct drm_gem_dma_object *color_write;
+	struct drm_gem_dma_object *zs_read;
+	struct drm_gem_dma_object *zs_write;
+	struct drm_gem_dma_object *msaa_color_write;
+	struct drm_gem_dma_object *msaa_zs_write;
+
+	struct drm_gem_dma_object *rcl;
+	u32 next_offset;
+
+	u32 next_write_bo_index;
+};
+
+static inline void rcl_u8(struct vc4_rcl_setup *setup, u8 val)
+{
+	*(u8 *)(setup->rcl->vaddr + setup->next_offset) = val;
+	setup->next_offset += 1;
+}
+
+static inline void rcl_u16(struct vc4_rcl_setup *setup, u16 val)
+{
+	*(u16 *)(setup->rcl->vaddr + setup->next_offset) = val;
+	setup->next_offset += 2;
+}
+
+static inline void rcl_u32(struct vc4_rcl_setup *setup, u32 val)
+{
+	*(u32 *)(setup->rcl->vaddr + setup->next_offset) = val;
+	setup->next_offset += 4;
+}
+
+/*
+ * Emits a no-op STORE_TILE_BUFFER_GENERAL.
+ *
+ * If we emit a PACKET_TILE_COORDINATES, it must be followed by a store of
+ * some sort before another load is triggered.
+ */
+static void vc4_store_before_load(struct vc4_rcl_setup *setup)
+{
+	rcl_u8(setup, VC4_PACKET_STORE_TILE_BUFFER_GENERAL);
+	rcl_u16(setup,
+		VC4_SET_FIELD(VC4_LOADSTORE_TILE_BUFFER_NONE,
+			      VC4_LOADSTORE_TILE_BUFFER_BUFFER) |
+		VC4_STORE_TILE_BUFFER_DISABLE_COLOR_CLEAR |
+		VC4_STORE_TILE_BUFFER_DISABLE_ZS_CLEAR |
+		VC4_STORE_TILE_BUFFER_DISABLE_VG_MASK_CLEAR);
+	rcl_u32(setup, 0); /* no address, since we're in None mode */
+}
+
+/*
+ * Calculates the physical address of the start of a tile in a RCL surface.
+ *
+ * Unlike the other load/store packets,
+ * VC4_PACKET_LOAD/STORE_FULL_RES_TILE_BUFFER don't look at the tile
+ * coordinates packet, and instead just store to the address given.
+ */
+static uint32_t vc4_full_res_offset(struct vc4_exec_info *exec,
+				    struct drm_gem_dma_object *bo,
+				    struct drm_vc4_submit_rcl_surface *surf,
+				    uint8_t x, uint8_t y)
+{
+	return bo->dma_addr + surf->offset + VC4_TILE_BUFFER_SIZE *
+		(DIV_ROUND_UP(exec->args->width, 32) * y + x);
+}
+
+/*
+ * Emits a PACKET_TILE_COORDINATES if one isn't already pending.
+ *
+ * The tile coordinates packet triggers a pending load if there is one, are
+ * used for clipping during rendering, and determine where loads/stores happen
+ * relative to their base address.
+ */
+static void vc4_tile_coordinates(struct vc4_rcl_setup *setup,
+				 uint32_t x, uint32_t y)
+{
+	rcl_u8(setup, VC4_PACKET_TILE_COORDINATES);
+	rcl_u8(setup, x);
+	rcl_u8(setup, y);
+}
+
+static void emit_tile(struct vc4_exec_info *exec,
+		      struct vc4_rcl_setup *setup,
+		      uint8_t x, uint8_t y, bool first, bool last)
+{
+	struct drm_vc4_submit_cl *args = exec->args;
+	bool has_bin = args->bin_cl_size != 0;
+
+	/* Note that the load doesn't actually occur until the
+	 * tile coords packet is processed, and only one load
+	 * may be outstanding at a time.
+	 */
+	if (setup->color_read) {
+		if (args->color_read.flags &
+		    VC4_SUBMIT_RCL_SURFACE_READ_IS_FULL_RES) {
+			rcl_u8(setup, VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER);
+			rcl_u32(setup,
+				vc4_full_res_offset(exec, setup->color_read,
+						    &args->color_read, x, y) |
+				VC4_LOADSTORE_FULL_RES_DISABLE_ZS);
+		} else {
+			rcl_u8(setup, VC4_PACKET_LOAD_TILE_BUFFER_GENERAL);
+			rcl_u16(setup, args->color_read.bits);
+			rcl_u32(setup, setup->color_read->dma_addr +
+				args->color_read.offset);
+		}
+	}
+
+	if (setup->zs_read) {
+		if (setup->color_read) {
+			/* Exec previous load. */
+			vc4_tile_coordinates(setup, x, y);
+			vc4_store_before_load(setup);
+		}
+
+		if (args->zs_read.flags &
+		    VC4_SUBMIT_RCL_SURFACE_READ_IS_FULL_RES) {
+			rcl_u8(setup, VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER);
+			rcl_u32(setup,
+				vc4_full_res_offset(exec, setup->zs_read,
+						    &args->zs_read, x, y) |
+				VC4_LOADSTORE_FULL_RES_DISABLE_COLOR);
+		} else {
+			rcl_u8(setup, VC4_PACKET_LOAD_TILE_BUFFER_GENERAL);
+			rcl_u16(setup, args->zs_read.bits);
+			rcl_u32(setup, setup->zs_read->dma_addr +
+				args->zs_read.offset);
+		}
+	}
+
+	/* Clipping depends on tile coordinates having been
+	 * emitted, so we always need one here.
+	 */
+	vc4_tile_coordinates(setup, x, y);
+
+	/* Wait for the binner before jumping to the first
+	 * tile's lists.
+	 */
+	if (first && has_bin)
+		rcl_u8(setup, VC4_PACKET_WAIT_ON_SEMAPHORE);
+
+	if (has_bin) {
+		rcl_u8(setup, VC4_PACKET_BRANCH_TO_SUB_LIST);
+		rcl_u32(setup, (exec->tile_alloc_offset +
+				(y * exec->bin_tiles_x + x) * 32));
+	}
+
+	if (setup->msaa_color_write) {
+		bool last_tile_write = (!setup->msaa_zs_write &&
+					!setup->zs_write &&
+					!setup->color_write);
+		uint32_t bits = VC4_LOADSTORE_FULL_RES_DISABLE_ZS;
+
+		if (!last_tile_write)
+			bits |= VC4_LOADSTORE_FULL_RES_DISABLE_CLEAR_ALL;
+		else if (last)
+			bits |= VC4_LOADSTORE_FULL_RES_EOF;
+		rcl_u8(setup, VC4_PACKET_STORE_FULL_RES_TILE_BUFFER);
+		rcl_u32(setup,
+			vc4_full_res_offset(exec, setup->msaa_color_write,
+					    &args->msaa_color_write, x, y) |
+			bits);
+	}
+
+	if (setup->msaa_zs_write) {
+		bool last_tile_write = (!setup->zs_write &&
+					!setup->color_write);
+		uint32_t bits = VC4_LOADSTORE_FULL_RES_DISABLE_COLOR;
+
+		if (setup->msaa_color_write)
+			vc4_tile_coordinates(setup, x, y);
+		if (!last_tile_write)
+			bits |= VC4_LOADSTORE_FULL_RES_DISABLE_CLEAR_ALL;
+		else if (last)
+			bits |= VC4_LOADSTORE_FULL_RES_EOF;
+		rcl_u8(setup, VC4_PACKET_STORE_FULL_RES_TILE_BUFFER);
+		rcl_u32(setup,
+			vc4_full_res_offset(exec, setup->msaa_zs_write,
+					    &args->msaa_zs_write, x, y) |
+			bits);
+	}
+
+	if (setup->zs_write) {
+		bool last_tile_write = !setup->color_write;
+
+		if (setup->msaa_color_write || setup->msaa_zs_write)
+			vc4_tile_coordinates(setup, x, y);
+
+		rcl_u8(setup, VC4_PACKET_STORE_TILE_BUFFER_GENERAL);
+		rcl_u16(setup, args->zs_write.bits |
+			(last_tile_write ?
+			 0 : VC4_STORE_TILE_BUFFER_DISABLE_COLOR_CLEAR));
+		rcl_u32(setup,
+			(setup->zs_write->dma_addr + args->zs_write.offset) |
+			((last && last_tile_write) ?
+			 VC4_LOADSTORE_TILE_BUFFER_EOF : 0));
+	}
+
+	if (setup->color_write) {
+		if (setup->msaa_color_write || setup->msaa_zs_write ||
+		    setup->zs_write) {
+			vc4_tile_coordinates(setup, x, y);
+		}
+
+		if (last)
+			rcl_u8(setup, VC4_PACKET_STORE_MS_TILE_BUFFER_AND_EOF);
+		else
+			rcl_u8(setup, VC4_PACKET_STORE_MS_TILE_BUFFER);
+	}
+}
+
+static int vc4_create_rcl_bo(struct drm_device *dev, struct vc4_exec_info *exec,
+			     struct vc4_rcl_setup *setup)
+{
+	struct drm_vc4_submit_cl *args = exec->args;
+	bool has_bin = args->bin_cl_size != 0;
+	uint8_t min_x_tile = args->min_x_tile;
+	uint8_t min_y_tile = args->min_y_tile;
+	uint8_t max_x_tile = args->max_x_tile;
+	uint8_t max_y_tile = args->max_y_tile;
+	uint8_t xtiles = max_x_tile - min_x_tile + 1;
+	uint8_t ytiles = max_y_tile - min_y_tile + 1;
+	uint8_t xi, yi;
+	uint32_t size, loop_body_size;
+	bool positive_x = true;
+	bool positive_y = true;
+
+	if (args->flags & VC4_SUBMIT_CL_FIXED_RCL_ORDER) {
+		if (!(args->flags & VC4_SUBMIT_CL_RCL_ORDER_INCREASING_X))
+			positive_x = false;
+		if (!(args->flags & VC4_SUBMIT_CL_RCL_ORDER_INCREASING_Y))
+			positive_y = false;
+	}
+
+	size = VC4_PACKET_TILE_RENDERING_MODE_CONFIG_SIZE;
+	loop_body_size = VC4_PACKET_TILE_COORDINATES_SIZE;
+
+	if (args->flags & VC4_SUBMIT_CL_USE_CLEAR_COLOR) {
+		size += VC4_PACKET_CLEAR_COLORS_SIZE +
+			VC4_PACKET_TILE_COORDINATES_SIZE +
+			VC4_PACKET_STORE_TILE_BUFFER_GENERAL_SIZE;
+	}
+
+	if (setup->color_read) {
+		if (args->color_read.flags &
+		    VC4_SUBMIT_RCL_SURFACE_READ_IS_FULL_RES) {
+			loop_body_size += VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER_SIZE;
+		} else {
+			loop_body_size += VC4_PACKET_LOAD_TILE_BUFFER_GENERAL_SIZE;
+		}
+	}
+	if (setup->zs_read) {
+		if (setup->color_read) {
+			loop_body_size += VC4_PACKET_TILE_COORDINATES_SIZE;
+			loop_body_size += VC4_PACKET_STORE_TILE_BUFFER_GENERAL_SIZE;
+		}
+
+		if (args->zs_read.flags &
+		    VC4_SUBMIT_RCL_SURFACE_READ_IS_FULL_RES) {
+			loop_body_size += VC4_PACKET_LOAD_FULL_RES_TILE_BUFFER_SIZE;
+		} else {
+			loop_body_size += VC4_PACKET_LOAD_TILE_BUFFER_GENERAL_SIZE;
+		}
+	}
+
+	if (has_bin) {
+		size += VC4_PACKET_WAIT_ON_SEMAPHORE_SIZE;
+		loop_body_size += VC4_PACKET_BRANCH_TO_SUB_LIST_SIZE;
+	}
+
+	if (setup->msaa_color_write)
+		loop_body_size += VC4_PACKET_STORE_FULL_RES_TILE_BUFFER_SIZE;
+	if (setup->msaa_zs_write)
+		loop_body_size += VC4_PACKET_STORE_FULL_RES_TILE_BUFFER_SIZE;
+
+	if (setup->zs_write)
+		loop_body_size += VC4_PACKET_STORE_TILE_BUFFER_GENERAL_SIZE;
+	if (setup->color_write)
+		loop_body_size += VC4_PACKET_STORE_MS_TILE_BUFFER_SIZE;
+
+	/* We need a VC4_PACKET_TILE_COORDINATES in between each store. */
+	loop_body_size += VC4_PACKET_TILE_COORDINATES_SIZE *
+		((setup->msaa_color_write != NULL) +
+		 (setup->msaa_zs_write != NULL) +
+		 (setup->color_write != NULL) +
+		 (setup->zs_write != NULL) - 1);
+
+	size += xtiles * ytiles * loop_body_size;
+
+	setup->rcl = &vc4_bo_create(dev, size, true, VC4_BO_TYPE_RCL)->base;
+	if (IS_ERR(setup->rcl))
+		return PTR_ERR(setup->rcl);
+	list_add_tail(&to_vc4_bo(&setup->rcl->base)->unref_head,
+		      &exec->unref_list);
+
+	/* The tile buffer gets cleared when the previous tile is stored.  If
+	 * the clear values changed between frames, then the tile buffer has
+	 * stale clear values in it, so we have to do a store in None mode (no
+	 * writes) so that we trigger the tile buffer clear.
+	 */
+	if (args->flags & VC4_SUBMIT_CL_USE_CLEAR_COLOR) {
+		rcl_u8(setup, VC4_PACKET_CLEAR_COLORS);
+		rcl_u32(setup, args->clear_color[0]);
+		rcl_u32(setup, args->clear_color[1]);
+		rcl_u32(setup, args->clear_z);
+		rcl_u8(setup, args->clear_s);
+
+		vc4_tile_coordinates(setup, 0, 0);
+
+		rcl_u8(setup, VC4_PACKET_STORE_TILE_BUFFER_GENERAL);
+		rcl_u16(setup, VC4_LOADSTORE_TILE_BUFFER_NONE);
+		rcl_u32(setup, 0); /* no address, since we're in None mode */
+	}
+
+	rcl_u8(setup, VC4_PACKET_TILE_RENDERING_MODE_CONFIG);
+	rcl_u32(setup,
+		(setup->color_write ? (setup->color_write->dma_addr +
+				       args->color_write.offset) :
+		 0));
+	rcl_u16(setup, args->width);
+	rcl_u16(setup, args->height);
+	rcl_u16(setup, args->color_write.bits);
+
+	for (yi = 0; yi < ytiles; yi++) {
+		int y = positive_y ? min_y_tile + yi : max_y_tile - yi;
+		for (xi = 0; xi < xtiles; xi++) {
+			int x = positive_x ? min_x_tile + xi : max_x_tile - xi;
+			bool first = (xi == 0 && yi == 0);
+			bool last = (xi == xtiles - 1 && yi == ytiles - 1);
+
+			emit_tile(exec, setup, x, y, first, last);
+		}
+	}
+
+	BUG_ON(setup->next_offset != size);
+	exec->ct1ca = setup->rcl->dma_addr;
+	exec->ct1ea = setup->rcl->dma_addr + setup->next_offset;
+
+	return 0;
+}
+
+static int vc4_full_res_bounds_check(struct vc4_exec_info *exec,
+				     struct drm_gem_dma_object *obj,
+				     struct drm_vc4_submit_rcl_surface *surf)
+{
+	struct drm_vc4_submit_cl *args = exec->args;
+	u32 render_tiles_stride = DIV_ROUND_UP(exec->args->width, 32);
+
+	if (surf->offset > obj->base.size) {
+		DRM_DEBUG("surface offset %d > BO size %zd\n",
+			  surf->offset, obj->base.size);
+		return -EINVAL;
+	}
+
+	if ((obj->base.size - surf->offset) / VC4_TILE_BUFFER_SIZE <
+	    render_tiles_stride * args->max_y_tile + args->max_x_tile) {
+		DRM_DEBUG("MSAA tile %d, %d out of bounds "
+			  "(bo size %zd, offset %d).\n",
+			  args->max_x_tile, args->max_y_tile,
+			  obj->base.size,
+			  surf->offset);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int vc4_rcl_msaa_surface_setup(struct vc4_exec_info *exec,
+				      struct drm_gem_dma_object **obj,
+				      struct drm_vc4_submit_rcl_surface *surf)
+{
+	if (surf->flags != 0 || surf->bits != 0) {
+		DRM_DEBUG("MSAA surface had nonzero flags/bits\n");
+		return -EINVAL;
+	}
+
+	if (surf->hindex == ~0)
+		return 0;
+
+	*obj = vc4_use_bo(exec, surf->hindex);
+	if (!*obj)
+		return -EINVAL;
+
+	exec->rcl_write_bo[exec->rcl_write_bo_count++] = *obj;
+
+	if (surf->offset & 0xf) {
+		DRM_DEBUG("MSAA write must be 16b aligned.\n");
+		return -EINVAL;
+	}
+
+	return vc4_full_res_bounds_check(exec, *obj, surf);
+}
+
+static int vc4_rcl_surface_setup(struct vc4_exec_info *exec,
+				 struct drm_gem_dma_object **obj,
+				 struct drm_vc4_submit_rcl_surface *surf,
+				 bool is_write)
+{
+	uint8_t tiling = VC4_GET_FIELD(surf->bits,
+				       VC4_LOADSTORE_TILE_BUFFER_TILING);
+	uint8_t buffer = VC4_GET_FIELD(surf->bits,
+				       VC4_LOADSTORE_TILE_BUFFER_BUFFER);
+	uint8_t format = VC4_GET_FIELD(surf->bits,
+				       VC4_LOADSTORE_TILE_BUFFER_FORMAT);
+	int cpp;
+	int ret;
+
+	if (surf->flags & ~VC4_SUBMIT_RCL_SURFACE_READ_IS_FULL_RES) {
+		DRM_DEBUG("Extra flags set\n");
+		return -EINVAL;
+	}
+
+	if (surf->hindex == ~0)
+		return 0;
+
+	*obj = vc4_use_bo(exec, surf->hindex);
+	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_DEBUG("general zs write may not be a full-res.\n");
+			return -EINVAL;
+		}
+
+		if (surf->bits != 0) {
+			DRM_DEBUG("load/store general bits set with "
+				  "full res load/store.\n");
+			return -EINVAL;
+		}
+
+		ret = vc4_full_res_bounds_check(exec, *obj, surf);
+		if (ret)
+			return ret;
+
+		return 0;
+	}
+
+	if (surf->bits & ~(VC4_LOADSTORE_TILE_BUFFER_TILING_MASK |
+			   VC4_LOADSTORE_TILE_BUFFER_BUFFER_MASK |
+			   VC4_LOADSTORE_TILE_BUFFER_FORMAT_MASK)) {
+		DRM_DEBUG("Unknown bits in load/store: 0x%04x\n",
+			  surf->bits);
+		return -EINVAL;
+	}
+
+	if (tiling > VC4_TILING_FORMAT_LT) {
+		DRM_DEBUG("Bad tiling format\n");
+		return -EINVAL;
+	}
+
+	if (buffer == VC4_LOADSTORE_TILE_BUFFER_ZS) {
+		if (format != 0) {
+			DRM_DEBUG("No color format should be set for ZS\n");
+			return -EINVAL;
+		}
+		cpp = 4;
+	} else if (buffer == VC4_LOADSTORE_TILE_BUFFER_COLOR) {
+		switch (format) {
+		case VC4_LOADSTORE_TILE_BUFFER_BGR565:
+		case VC4_LOADSTORE_TILE_BUFFER_BGR565_DITHER:
+			cpp = 2;
+			break;
+		case VC4_LOADSTORE_TILE_BUFFER_RGBA8888:
+			cpp = 4;
+			break;
+		default:
+			DRM_DEBUG("Bad tile buffer format\n");
+			return -EINVAL;
+		}
+	} else {
+		DRM_DEBUG("Bad load/store buffer %d.\n", buffer);
+		return -EINVAL;
+	}
+
+	if (surf->offset & 0xf) {
+		DRM_DEBUG("load/store buffer must be 16b aligned.\n");
+		return -EINVAL;
+	}
+
+	if (!vc4_check_tex_size(exec, *obj, surf->offset, tiling,
+				exec->args->width, exec->args->height, cpp)) {
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+vc4_rcl_render_config_surface_setup(struct vc4_exec_info *exec,
+				    struct vc4_rcl_setup *setup,
+				    struct drm_gem_dma_object **obj,
+				    struct drm_vc4_submit_rcl_surface *surf)
+{
+	uint8_t tiling = VC4_GET_FIELD(surf->bits,
+				       VC4_RENDER_CONFIG_MEMORY_FORMAT);
+	uint8_t format = VC4_GET_FIELD(surf->bits,
+				       VC4_RENDER_CONFIG_FORMAT);
+	int cpp;
+
+	if (surf->flags != 0) {
+		DRM_DEBUG("No flags supported on render config.\n");
+		return -EINVAL;
+	}
+
+	if (surf->bits & ~(VC4_RENDER_CONFIG_MEMORY_FORMAT_MASK |
+			   VC4_RENDER_CONFIG_FORMAT_MASK |
+			   VC4_RENDER_CONFIG_MS_MODE_4X |
+			   VC4_RENDER_CONFIG_DECIMATE_MODE_4X)) {
+		DRM_DEBUG("Unknown bits in render config: 0x%04x\n",
+			  surf->bits);
+		return -EINVAL;
+	}
+
+	if (surf->hindex == ~0)
+		return 0;
+
+	*obj = vc4_use_bo(exec, surf->hindex);
+	if (!*obj)
+		return -EINVAL;
+
+	exec->rcl_write_bo[exec->rcl_write_bo_count++] = *obj;
+
+	if (tiling > VC4_TILING_FORMAT_LT) {
+		DRM_DEBUG("Bad tiling format\n");
+		return -EINVAL;
+	}
+
+	switch (format) {
+	case VC4_RENDER_CONFIG_FORMAT_BGR565_DITHERED:
+	case VC4_RENDER_CONFIG_FORMAT_BGR565:
+		cpp = 2;
+		break;
+	case VC4_RENDER_CONFIG_FORMAT_RGBA8888:
+		cpp = 4;
+		break;
+	default:
+		DRM_DEBUG("Bad tile buffer format\n");
+		return -EINVAL;
+	}
+
+	if (!vc4_check_tex_size(exec, *obj, surf->offset, tiling,
+				exec->args->width, exec->args->height, cpp)) {
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int vc4_get_rcl(struct drm_device *dev, struct vc4_exec_info *exec)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_rcl_setup setup = {0};
+	struct drm_vc4_submit_cl *args = exec->args;
+	bool has_bin = args->bin_cl_size != 0;
+	int ret;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	if (args->min_x_tile > args->max_x_tile ||
+	    args->min_y_tile > args->max_y_tile) {
+		DRM_DEBUG("Bad render tile set (%d,%d)-(%d,%d)\n",
+			  args->min_x_tile, args->min_y_tile,
+			  args->max_x_tile, args->max_y_tile);
+		return -EINVAL;
+	}
+
+	if (has_bin &&
+	    (args->max_x_tile > exec->bin_tiles_x ||
+	     args->max_y_tile > exec->bin_tiles_y)) {
+		DRM_DEBUG("Render tiles (%d,%d) outside of bin config "
+			  "(%d,%d)\n",
+			  args->max_x_tile, args->max_y_tile,
+			  exec->bin_tiles_x, exec->bin_tiles_y);
+		return -EINVAL;
+	}
+
+	ret = vc4_rcl_render_config_surface_setup(exec, &setup,
+						  &setup.color_write,
+						  &args->color_write);
+	if (ret)
+		return ret;
+
+	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,
+				    false);
+	if (ret)
+		return ret;
+
+	ret = vc4_rcl_surface_setup(exec, &setup.zs_write, &args->zs_write,
+				    true);
+	if (ret)
+		return ret;
+
+	ret = vc4_rcl_msaa_surface_setup(exec, &setup.msaa_color_write,
+					 &args->msaa_color_write);
+	if (ret)
+		return ret;
+
+	ret = vc4_rcl_msaa_surface_setup(exec, &setup.msaa_zs_write,
+					 &args->msaa_zs_write);
+	if (ret)
+		return ret;
+
+	/* We shouldn't even have the job submitted to us if there's no
+	 * surface to write out.
+	 */
+	if (!setup.color_write && !setup.zs_write &&
+	    !setup.msaa_color_write && !setup.msaa_zs_write) {
+		DRM_DEBUG("RCL requires color or Z/S write\n");
+		return -EINVAL;
+	}
+
+	return vc4_create_rcl_bo(dev, exec, &setup);
+}

linux-7.2.3/drivers/gpu/drm/vc4/vc4_trace.h

@@ -0,0 +1,155 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2015 Broadcom
+ */
+
+#if !defined(_VC4_TRACE_H_) || defined(TRACE_HEADER_MULTI_READ)
+#define _VC4_TRACE_H_
+
+#include <linux/stringify.h>
+#include <linux/types.h>
+#include <linux/tracepoint.h>
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM vc4
+#define TRACE_INCLUDE_FILE vc4_trace
+
+TRACE_EVENT(vc4_wait_for_seqno_begin,
+	    TP_PROTO(struct drm_device *dev, uint64_t seqno, uint64_t timeout),
+	    TP_ARGS(dev, seqno, timeout),
+
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(u64, seqno)
+			     __field(u64, timeout)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->seqno = seqno;
+			   __entry->timeout = timeout;
+			   ),
+
+	    TP_printk("dev=%u, seqno=%llu, timeout=%llu",
+		      __entry->dev, __entry->seqno, __entry->timeout)
+);
+
+TRACE_EVENT(vc4_wait_for_seqno_end,
+	    TP_PROTO(struct drm_device *dev, uint64_t seqno),
+	    TP_ARGS(dev, seqno),
+
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(u64, seqno)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->seqno = seqno;
+			   ),
+
+	    TP_printk("dev=%u, seqno=%llu",
+		      __entry->dev, __entry->seqno)
+);
+
+TRACE_EVENT(vc4_submit_cl_ioctl,
+	    TP_PROTO(struct drm_device *dev, u32 bin_cl_size, u32 shader_rec_size, u32 bo_count),
+	    TP_ARGS(dev, bin_cl_size, shader_rec_size, bo_count),
+
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(u32, bin_cl_size)
+			     __field(u32, shader_rec_size)
+			     __field(u32, bo_count)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->bin_cl_size = bin_cl_size;
+			   __entry->shader_rec_size = shader_rec_size;
+			   __entry->bo_count = bo_count;
+			   ),
+
+	    TP_printk("dev=%u, bin_cl_size=%u, shader_rec_size=%u, bo_count=%u",
+		      __entry->dev,
+		      __entry->bin_cl_size,
+		      __entry->shader_rec_size,
+		      __entry->bo_count)
+);
+
+TRACE_EVENT(vc4_submit_cl,
+	    TP_PROTO(struct drm_device *dev, bool is_render,
+		     uint64_t seqno,
+		     u32 ctnqba, u32 ctnqea),
+	    TP_ARGS(dev, is_render, seqno, ctnqba, ctnqea),
+
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(bool, is_render)
+			     __field(u64, seqno)
+			     __field(u32, ctnqba)
+			     __field(u32, ctnqea)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->is_render = is_render;
+			   __entry->seqno = seqno;
+			   __entry->ctnqba = ctnqba;
+			   __entry->ctnqea = ctnqea;
+			   ),
+
+	    TP_printk("dev=%u, %s, seqno=%llu, 0x%08x..0x%08x",
+		      __entry->dev,
+		      __entry->is_render ? "RCL" : "BCL",
+		      __entry->seqno,
+		      __entry->ctnqba,
+		      __entry->ctnqea)
+);
+
+TRACE_EVENT(vc4_bcl_end_irq,
+	    TP_PROTO(struct drm_device *dev,
+		     uint64_t seqno),
+	    TP_ARGS(dev, seqno),
+
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(u64, seqno)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->seqno = seqno;
+			   ),
+
+	    TP_printk("dev=%u, seqno=%llu",
+		      __entry->dev,
+		      __entry->seqno)
+);
+
+TRACE_EVENT(vc4_rcl_end_irq,
+	    TP_PROTO(struct drm_device *dev,
+		     uint64_t seqno),
+	    TP_ARGS(dev, seqno),
+
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(u64, seqno)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->seqno = seqno;
+			   ),
+
+	    TP_printk("dev=%u, seqno=%llu",
+		      __entry->dev,
+		      __entry->seqno)
+);
+
+#endif /* _VC4_TRACE_H_ */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH ../../drivers/gpu/drm/vc4
+#include <trace/define_trace.h>

linux-7.2.3/drivers/gpu/drm/vc4/vc4_trace_points.c

@@ -0,0 +1,11 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2015 Broadcom
+ */
+
+#include "vc4_drv.h"
+
+#ifndef __CHECKER__
+#define CREATE_TRACE_POINTS
+#include "vc4_trace.h"
+#endif

linux-7.2.3/drivers/gpu/drm/vc4/vc4_txp.c

@@ -0,0 +1,658 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright © 2018 Broadcom
+ *
+ * Authors:
+ *	Eric Anholt <eric@anholt.net>
+ *	Boris Brezillon <boris.brezillon@bootlin.com>
+ */
+
+#include <linux/clk.h>
+#include <linux/component.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/drm_atomic.h>
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_fb_dma_helper.h>
+#include <drm/drm_fourcc.h>
+#include <drm/drm_framebuffer.h>
+#include <drm/drm_panel.h>
+#include <drm/drm_print.h>
+#include <drm/drm_probe_helper.h>
+#include <drm/drm_vblank.h>
+#include <drm/drm_writeback.h>
+
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+
+/* Base address of the output.  Raster formats must be 4-byte aligned,
+ * T and LT must be 16-byte aligned or maybe utile-aligned (docs are
+ * inconsistent, but probably utile).
+ */
+#define TXP_DST_PTR		0x00
+
+/* Pitch in bytes for raster images, 16-byte aligned.  For tiled, it's
+ * the width in tiles.
+ */
+#define TXP_DST_PITCH		0x04
+/* For T-tiled imgaes, DST_PITCH should be the number of tiles wide,
+ * shifted up.
+ */
+# define TXP_T_TILE_WIDTH_SHIFT		7
+/* For LT-tiled images, DST_PITCH should be the number of utiles wide,
+ * shifted up.
+ */
+# define TXP_LT_TILE_WIDTH_SHIFT	4
+
+/* Pre-rotation width/height of the image.  Must match HVS config.
+ *
+ * If TFORMAT and 32-bit, limit is 1920 for 32-bit and 3840 to 16-bit
+ * and width/height must be tile or utile-aligned as appropriate.  If
+ * transposing (rotating), width is limited to 1920.
+ *
+ * Height is limited to various numbers between 4088 and 4095.  I'd
+ * just use 4088 to be safe.
+ */
+#define TXP_DIM			0x08
+# define TXP_HEIGHT_SHIFT		16
+# define TXP_HEIGHT_MASK		GENMASK(31, 16)
+# define TXP_WIDTH_SHIFT		0
+# define TXP_WIDTH_MASK			GENMASK(15, 0)
+
+#define TXP_DST_CTRL		0x0c
+/* These bits are set to 0x54 */
+#define TXP_PILOT_SHIFT			24
+#define TXP_PILOT_MASK			GENMASK(31, 24)
+/* Bits 22-23 are set to 0x01 */
+#define TXP_VERSION_SHIFT		22
+#define TXP_VERSION_MASK		GENMASK(23, 22)
+
+/* Powers down the internal memory. */
+# define TXP_POWERDOWN			BIT(21)
+
+/* Enables storing the alpha component in 8888/4444, instead of
+ * filling with ~ALPHA_INVERT.
+ */
+# define TXP_ALPHA_ENABLE		BIT(20)
+
+/* 4 bits, each enables stores for a channel in each set of 4 bytes.
+ * Set to 0xf for normal operation.
+ */
+# define TXP_BYTE_ENABLE_SHIFT		16
+# define TXP_BYTE_ENABLE_MASK		GENMASK(19, 16)
+
+/* Debug: Generate VSTART again at EOF. */
+# define TXP_VSTART_AT_EOF		BIT(15)
+
+/* Debug: Terminate the current frame immediately.  Stops AXI
+ * writes.
+ */
+# define TXP_ABORT			BIT(14)
+
+# define TXP_DITHER			BIT(13)
+
+/* Inverts alpha if TXP_ALPHA_ENABLE, chooses fill value for
+ * !TXP_ALPHA_ENABLE.
+ */
+# define TXP_ALPHA_INVERT		BIT(12)
+
+/* Note: I've listed the channels here in high bit (in byte 3/2/1) to
+ * low bit (in byte 0) order.
+ */
+# define TXP_FORMAT_SHIFT		8
+# define TXP_FORMAT_MASK		GENMASK(11, 8)
+# define TXP_FORMAT_ABGR4444		0
+# define TXP_FORMAT_ARGB4444		1
+# define TXP_FORMAT_BGRA4444		2
+# define TXP_FORMAT_RGBA4444		3
+# define TXP_FORMAT_BGR565		6
+# define TXP_FORMAT_RGB565		7
+/* 888s are non-rotated, raster-only */
+# define TXP_FORMAT_BGR888		8
+# define TXP_FORMAT_RGB888		9
+# define TXP_FORMAT_ABGR8888		12
+# define TXP_FORMAT_ARGB8888		13
+# define TXP_FORMAT_BGRA8888		14
+# define TXP_FORMAT_RGBA8888		15
+
+/* If TFORMAT is set, generates LT instead of T format. */
+# define TXP_LINEAR_UTILE		BIT(7)
+
+/* Rotate output by 90 degrees. */
+# define TXP_TRANSPOSE			BIT(6)
+
+/* Generate a tiled format for V3D. */
+# define TXP_TFORMAT			BIT(5)
+
+/* Generates some undefined test mode output. */
+# define TXP_TEST_MODE			BIT(4)
+
+/* Request odd field from HVS. */
+# define TXP_FIELD			BIT(3)
+
+/* Raise interrupt when idle. */
+# define TXP_EI				BIT(2)
+
+/* Set when generating a frame, clears when idle. */
+# define TXP_BUSY			BIT(1)
+
+/* Starts a frame.  Self-clearing. */
+# define TXP_GO				BIT(0)
+
+/* Number of lines received and committed to memory. */
+#define TXP_PROGRESS		0x10
+
+#define TXP_DST_PTR_HIGH_MOPLET	0x1c
+#define TXP_DST_PTR_HIGH_MOP	0x24
+
+#define TXP_READ(offset)								\
+	({										\
+		kunit_fail_current_test("Accessing a register in a unit test!\n");	\
+		readl(txp->regs + (offset));						\
+	})
+
+#define TXP_WRITE(offset, val)								\
+	do {										\
+		kunit_fail_current_test("Accessing a register in a unit test!\n");	\
+		writel(val, txp->regs + (offset));					\
+	} while (0)
+
+struct vc4_txp {
+	struct vc4_crtc	base;
+	const struct vc4_txp_data *data;
+
+	struct platform_device *pdev;
+
+	struct vc4_encoder encoder;
+	struct drm_writeback_connector connector;
+
+	void __iomem *regs;
+};
+
+#define encoder_to_vc4_txp(_encoder)					\
+	container_of_const(_encoder, struct vc4_txp, encoder.base)
+
+#define connector_to_vc4_txp(_connector)				\
+	container_of_const(_connector, struct vc4_txp, connector.base)
+
+static const struct debugfs_reg32 txp_regs[] = {
+	VC4_REG32(TXP_DST_PTR),
+	VC4_REG32(TXP_DST_PITCH),
+	VC4_REG32(TXP_DIM),
+	VC4_REG32(TXP_DST_CTRL),
+	VC4_REG32(TXP_PROGRESS),
+};
+
+static int vc4_txp_connector_get_modes(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+
+	return drm_add_modes_noedid(connector, dev->mode_config.max_width,
+				    dev->mode_config.max_height);
+}
+
+static enum drm_mode_status
+vc4_txp_connector_mode_valid(struct drm_connector *connector,
+			     const struct drm_display_mode *mode)
+{
+	struct drm_device *dev = connector->dev;
+	struct drm_mode_config *mode_config = &dev->mode_config;
+	int w = mode->hdisplay, h = mode->vdisplay;
+
+	if (w < mode_config->min_width || w > mode_config->max_width)
+		return MODE_BAD_HVALUE;
+
+	if (h < mode_config->min_height || h > mode_config->max_height)
+		return MODE_BAD_VVALUE;
+
+	return MODE_OK;
+}
+
+static const u32 drm_fmts[] = {
+	DRM_FORMAT_RGB888,
+	DRM_FORMAT_BGR888,
+	DRM_FORMAT_XRGB8888,
+	DRM_FORMAT_XBGR8888,
+	DRM_FORMAT_ARGB8888,
+	DRM_FORMAT_ABGR8888,
+	DRM_FORMAT_RGBX8888,
+	DRM_FORMAT_BGRX8888,
+	DRM_FORMAT_RGBA8888,
+	DRM_FORMAT_BGRA8888,
+};
+
+static const u32 txp_fmts[] = {
+	TXP_FORMAT_RGB888,
+	TXP_FORMAT_BGR888,
+	TXP_FORMAT_ARGB8888,
+	TXP_FORMAT_ABGR8888,
+	TXP_FORMAT_ARGB8888,
+	TXP_FORMAT_ABGR8888,
+	TXP_FORMAT_RGBA8888,
+	TXP_FORMAT_BGRA8888,
+	TXP_FORMAT_RGBA8888,
+	TXP_FORMAT_BGRA8888,
+};
+
+static void vc4_txp_armed(struct drm_crtc_state *state)
+{
+	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(state);
+
+	vc4_state->txp_armed = true;
+}
+
+static int vc4_txp_connector_atomic_check(struct drm_connector *conn,
+					  struct drm_atomic_commit *state)
+{
+	struct drm_connector_state *conn_state;
+	struct drm_crtc_state *crtc_state;
+	struct drm_framebuffer *fb;
+	int i;
+
+	conn_state = drm_atomic_get_new_connector_state(state, conn);
+	if (!conn_state->writeback_job)
+		return 0;
+
+	crtc_state = drm_atomic_get_new_crtc_state(state, conn_state->crtc);
+
+	fb = conn_state->writeback_job->fb;
+	if (fb->width != crtc_state->mode.hdisplay ||
+	    fb->height != crtc_state->mode.vdisplay) {
+		DRM_DEBUG_KMS("Invalid framebuffer size %ux%u\n",
+			      fb->width, fb->height);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(drm_fmts); i++) {
+		if (fb->format->format == drm_fmts[i])
+			break;
+	}
+
+	if (i == ARRAY_SIZE(drm_fmts))
+		return -EINVAL;
+
+	/* Pitch must be aligned on 16 bytes. */
+	if (fb->pitches[0] & GENMASK(3, 0))
+		return -EINVAL;
+
+	vc4_txp_armed(crtc_state);
+
+	return 0;
+}
+
+static void vc4_txp_connector_atomic_commit(struct drm_connector *conn,
+					struct drm_atomic_commit *state)
+{
+	struct drm_device *drm = conn->dev;
+	struct drm_connector_state *conn_state = drm_atomic_get_new_connector_state(state,
+										    conn);
+	struct vc4_txp *txp = connector_to_vc4_txp(conn);
+	const struct vc4_txp_data *txp_data = txp->data;
+	struct drm_gem_dma_object *gem;
+	struct drm_display_mode *mode;
+	struct drm_framebuffer *fb;
+	unsigned int hdisplay;
+	unsigned int vdisplay;
+	dma_addr_t addr;
+	u32 ctrl;
+	int idx;
+	int i;
+
+	if (WARN_ON(!conn_state->writeback_job))
+		return;
+
+	mode = &conn_state->crtc->state->adjusted_mode;
+	fb = conn_state->writeback_job->fb;
+
+	for (i = 0; i < ARRAY_SIZE(drm_fmts); i++) {
+		if (fb->format->format == drm_fmts[i])
+			break;
+	}
+
+	if (WARN_ON(i == ARRAY_SIZE(drm_fmts)))
+		return;
+
+	ctrl = TXP_GO | TXP_EI |
+	       VC4_SET_FIELD(txp_fmts[i], TXP_FORMAT);
+
+	if (txp_data->has_byte_enable)
+		ctrl |= VC4_SET_FIELD(0xf, TXP_BYTE_ENABLE);
+
+	if (fb->format->has_alpha)
+		ctrl |= TXP_ALPHA_ENABLE;
+	else
+		/*
+		 * If TXP_ALPHA_ENABLE isn't set and TXP_ALPHA_INVERT is, the
+		 * hardware will force the output padding to be 0xff.
+		 */
+		ctrl |= TXP_ALPHA_INVERT;
+
+	if (!drm_dev_enter(drm, &idx))
+		return;
+
+	gem = drm_fb_dma_get_gem_obj(fb, 0);
+	addr = gem->dma_addr + fb->offsets[0];
+
+	TXP_WRITE(TXP_DST_PTR, lower_32_bits(addr));
+
+	if (txp_data->supports_40bit_addresses)
+		TXP_WRITE(txp_data->high_addr_ptr_reg, upper_32_bits(addr) & 0xff);
+
+	TXP_WRITE(TXP_DST_PITCH, fb->pitches[0]);
+
+	hdisplay = mode->hdisplay ?: 1;
+	vdisplay = mode->vdisplay ?: 1;
+	if (txp_data->size_minus_one) {
+		hdisplay -= 1;
+		vdisplay -= 1;
+	}
+
+	TXP_WRITE(TXP_DIM,
+		  VC4_SET_FIELD(hdisplay, TXP_WIDTH) |
+		  VC4_SET_FIELD(vdisplay, TXP_HEIGHT));
+
+	TXP_WRITE(TXP_DST_CTRL, ctrl);
+
+	drm_writeback_queue_job(&txp->connector, conn_state);
+
+	drm_dev_exit(idx);
+}
+
+static const struct drm_connector_helper_funcs vc4_txp_connector_helper_funcs = {
+	.get_modes = vc4_txp_connector_get_modes,
+	.mode_valid = vc4_txp_connector_mode_valid,
+	.atomic_check = vc4_txp_connector_atomic_check,
+	.atomic_commit = vc4_txp_connector_atomic_commit,
+};
+
+static enum drm_connector_status
+vc4_txp_connector_detect(struct drm_connector *connector, bool force)
+{
+	return connector_status_connected;
+}
+
+static const struct drm_connector_funcs vc4_txp_connector_funcs = {
+	.detect = vc4_txp_connector_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.destroy = drm_connector_cleanup,
+	.reset = drm_atomic_helper_connector_reset,
+	.atomic_duplicate_state = drm_atomic_helper_connector_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_connector_destroy_state,
+};
+
+static void vc4_txp_encoder_disable(struct drm_encoder *encoder)
+{
+	struct drm_device *drm = encoder->dev;
+	struct vc4_dev *vc4 = to_vc4_dev(drm);
+	struct vc4_txp *txp = encoder_to_vc4_txp(encoder);
+	int idx;
+
+	if (!drm_dev_enter(drm, &idx))
+		return;
+
+	if (TXP_READ(TXP_DST_CTRL) & TXP_BUSY) {
+		unsigned long timeout = jiffies + msecs_to_jiffies(1000);
+
+		TXP_WRITE(TXP_DST_CTRL, TXP_ABORT);
+
+		while (TXP_READ(TXP_DST_CTRL) & TXP_BUSY &&
+		       time_before(jiffies, timeout))
+			;
+
+		WARN_ON(TXP_READ(TXP_DST_CTRL) & TXP_BUSY);
+	}
+
+	if (vc4->gen < VC4_GEN_6_C)
+		TXP_WRITE(TXP_DST_CTRL, TXP_POWERDOWN);
+
+	drm_dev_exit(idx);
+}
+
+static const struct drm_encoder_helper_funcs vc4_txp_encoder_helper_funcs = {
+	.disable = vc4_txp_encoder_disable,
+};
+
+static int vc4_txp_enable_vblank(struct drm_crtc *crtc)
+{
+	return 0;
+}
+
+static void vc4_txp_disable_vblank(struct drm_crtc *crtc) {}
+
+static const struct drm_crtc_funcs vc4_txp_crtc_funcs = {
+	.set_config		= drm_atomic_helper_set_config,
+	.page_flip		= vc4_page_flip,
+	.reset			= vc4_crtc_reset,
+	.atomic_duplicate_state	= vc4_crtc_duplicate_state,
+	.atomic_destroy_state	= vc4_crtc_destroy_state,
+	.enable_vblank		= vc4_txp_enable_vblank,
+	.disable_vblank		= vc4_txp_disable_vblank,
+	.late_register		= vc4_crtc_late_register,
+};
+
+static int vc4_txp_atomic_check(struct drm_crtc *crtc,
+				struct drm_atomic_commit *state)
+{
+	struct drm_crtc_state *crtc_state = drm_atomic_get_new_crtc_state(state,
+									  crtc);
+	int ret;
+
+	ret = vc4_hvs_atomic_check(crtc, state);
+	if (ret)
+		return ret;
+
+	crtc_state->no_vblank = true;
+
+	return 0;
+}
+
+static void vc4_txp_atomic_enable(struct drm_crtc *crtc,
+				  struct drm_atomic_commit *state)
+{
+	drm_crtc_vblank_on(crtc);
+	vc4_hvs_atomic_enable(crtc, state);
+}
+
+static void vc4_txp_atomic_disable(struct drm_crtc *crtc,
+				   struct drm_atomic_commit *state)
+{
+	struct drm_device *dev = crtc->dev;
+
+	/* Disable vblank irq handling before crtc is disabled. */
+	drm_crtc_vblank_off(crtc);
+
+	vc4_hvs_atomic_disable(crtc, state);
+
+	/*
+	 * Make sure we issue a vblank event after disabling the CRTC if
+	 * someone was waiting it.
+	 */
+	if (crtc->state->event) {
+		unsigned long flags;
+
+		spin_lock_irqsave(&dev->event_lock, flags);
+		drm_crtc_send_vblank_event(crtc, crtc->state->event);
+		crtc->state->event = NULL;
+		spin_unlock_irqrestore(&dev->event_lock, flags);
+	}
+}
+
+static const struct drm_crtc_helper_funcs vc4_txp_crtc_helper_funcs = {
+	.atomic_check	= vc4_txp_atomic_check,
+	.atomic_begin	= vc4_hvs_atomic_begin,
+	.atomic_flush	= vc4_hvs_atomic_flush,
+	.atomic_enable	= vc4_txp_atomic_enable,
+	.atomic_disable	= vc4_txp_atomic_disable,
+};
+
+static irqreturn_t vc4_txp_interrupt(int irq, void *data)
+{
+	struct vc4_txp *txp = data;
+	struct vc4_crtc *vc4_crtc = &txp->base;
+
+	/*
+	 * We don't need to protect the register access using
+	 * drm_dev_enter() there because the interrupt handler lifetime
+	 * is tied to the device itself, and not to the DRM device.
+	 *
+	 * So when the device will be gone, one of the first thing we
+	 * will be doing will be to unregister the interrupt handler,
+	 * and then unregister the DRM device. drm_dev_enter() would
+	 * thus always succeed if we are here.
+	 */
+	TXP_WRITE(TXP_DST_CTRL, TXP_READ(TXP_DST_CTRL) & ~TXP_EI);
+	vc4_crtc_handle_vblank(vc4_crtc);
+	drm_writeback_signal_completion(&txp->connector, 0);
+
+	return IRQ_HANDLED;
+}
+
+static const struct vc4_txp_data bcm2712_mop_data = {
+	.base = {
+		.name = "mop",
+		.debugfs_name = "mop_regs",
+		.hvs_available_channels = BIT(2),
+		.hvs_output = 2,
+	},
+	.encoder_type = VC4_ENCODER_TYPE_TXP0,
+	.high_addr_ptr_reg = TXP_DST_PTR_HIGH_MOP,
+	.has_byte_enable = true,
+	.size_minus_one = true,
+	.supports_40bit_addresses = true,
+};
+
+static const struct vc4_txp_data bcm2712_moplet_data = {
+	.base = {
+		.name = "moplet",
+		.debugfs_name = "moplet_regs",
+		.hvs_available_channels = BIT(1),
+		.hvs_output = 4,
+	},
+	.encoder_type = VC4_ENCODER_TYPE_TXP1,
+	.high_addr_ptr_reg = TXP_DST_PTR_HIGH_MOPLET,
+	.size_minus_one = true,
+	.supports_40bit_addresses = true,
+};
+
+const struct vc4_txp_data bcm2835_txp_data = {
+	.base = {
+		.name = "txp",
+		.debugfs_name = "txp_regs",
+		.hvs_available_channels = BIT(2),
+		.hvs_output = 2,
+	},
+	.encoder_type = VC4_ENCODER_TYPE_TXP0,
+	.has_byte_enable = true,
+};
+
+static int vc4_txp_bind(struct device *dev, struct device *master, void *data)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct drm_device *drm = dev_get_drvdata(master);
+	const struct vc4_txp_data *txp_data;
+	struct vc4_encoder *vc4_encoder;
+	struct drm_encoder *encoder;
+	struct vc4_crtc *vc4_crtc;
+	struct vc4_txp *txp;
+	int ret, irq;
+
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0)
+		return irq;
+
+	txp = drmm_kzalloc(drm, sizeof(*txp), GFP_KERNEL);
+	if (!txp)
+		return -ENOMEM;
+
+	txp_data = of_device_get_match_data(dev);
+	if (!txp_data)
+		return -ENODEV;
+
+	txp->data = txp_data;
+	txp->pdev = pdev;
+	txp->regs = vc4_ioremap_regs(pdev, 0);
+	if (IS_ERR(txp->regs))
+		return PTR_ERR(txp->regs);
+
+	vc4_crtc = &txp->base;
+	vc4_crtc->regset.base = txp->regs;
+	vc4_crtc->regset.regs = txp_regs;
+	vc4_crtc->regset.nregs = ARRAY_SIZE(txp_regs);
+
+	ret = vc4_crtc_init(drm, pdev, vc4_crtc, &txp_data->base,
+			    &vc4_txp_crtc_funcs, &vc4_txp_crtc_helper_funcs, true);
+	if (ret)
+		return ret;
+
+	vc4_encoder = &txp->encoder;
+	txp->encoder.type = txp_data->encoder_type;
+
+	encoder = &vc4_encoder->base;
+	encoder->possible_crtcs = drm_crtc_mask(&vc4_crtc->base);
+
+	drm_encoder_helper_add(encoder, &vc4_txp_encoder_helper_funcs);
+
+	ret = drmm_encoder_init(drm, encoder, NULL, DRM_MODE_ENCODER_VIRTUAL, NULL);
+	if (ret)
+		return ret;
+
+	drm_connector_helper_add(&txp->connector.base,
+				 &vc4_txp_connector_helper_funcs);
+	ret = drm_writeback_connector_init_with_encoder(drm, &txp->connector,
+							encoder,
+							&vc4_txp_connector_funcs,
+							drm_fmts, ARRAY_SIZE(drm_fmts));
+	if (ret)
+		return ret;
+
+	ret = devm_request_irq(dev, irq, vc4_txp_interrupt, 0,
+			       dev_name(dev), txp);
+	if (ret)
+		return ret;
+
+	dev_set_drvdata(dev, txp);
+
+	return 0;
+}
+
+static void vc4_txp_unbind(struct device *dev, struct device *master,
+			   void *data)
+{
+	struct vc4_txp *txp = dev_get_drvdata(dev);
+
+	drm_connector_cleanup(&txp->connector.base);
+}
+
+static const struct component_ops vc4_txp_ops = {
+	.bind   = vc4_txp_bind,
+	.unbind = vc4_txp_unbind,
+};
+
+static int vc4_txp_probe(struct platform_device *pdev)
+{
+	return component_add(&pdev->dev, &vc4_txp_ops);
+}
+
+static void vc4_txp_remove(struct platform_device *pdev)
+{
+	component_del(&pdev->dev, &vc4_txp_ops);
+}
+
+static const struct of_device_id vc4_txp_dt_match[] = {
+	{ .compatible = "brcm,bcm2712-mop", .data = &bcm2712_mop_data },
+	{ .compatible = "brcm,bcm2712-moplet", .data = &bcm2712_moplet_data },
+	{ .compatible = "brcm,bcm2835-txp", .data = &bcm2835_txp_data },
+	{ /* sentinel */ },
+};
+
+struct platform_driver vc4_txp_driver = {
+	.probe = vc4_txp_probe,
+	.remove = vc4_txp_remove,
+	.driver = {
+		.name = "vc4_txp",
+		.of_match_table = vc4_txp_dt_match,
+	},
+};

linux-7.2.3/drivers/gpu/drm/vc4/vc4_v3d.c

@@ -0,0 +1,546 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2014 The Linux Foundation. All rights reserved.
+ * Copyright (C) 2013 Red Hat
+ * Author: Rob Clark <robdclark@gmail.com>
+ */
+
+#include <linux/clk.h>
+#include <linux/component.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include <drm/drm_print.h>
+
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+
+static const struct debugfs_reg32 v3d_regs[] = {
+	VC4_REG32(V3D_IDENT0),
+	VC4_REG32(V3D_IDENT1),
+	VC4_REG32(V3D_IDENT2),
+	VC4_REG32(V3D_SCRATCH),
+	VC4_REG32(V3D_L2CACTL),
+	VC4_REG32(V3D_SLCACTL),
+	VC4_REG32(V3D_INTCTL),
+	VC4_REG32(V3D_INTENA),
+	VC4_REG32(V3D_INTDIS),
+	VC4_REG32(V3D_CT0CS),
+	VC4_REG32(V3D_CT1CS),
+	VC4_REG32(V3D_CT0EA),
+	VC4_REG32(V3D_CT1EA),
+	VC4_REG32(V3D_CT0CA),
+	VC4_REG32(V3D_CT1CA),
+	VC4_REG32(V3D_CT00RA0),
+	VC4_REG32(V3D_CT01RA0),
+	VC4_REG32(V3D_CT0LC),
+	VC4_REG32(V3D_CT1LC),
+	VC4_REG32(V3D_CT0PC),
+	VC4_REG32(V3D_CT1PC),
+	VC4_REG32(V3D_PCS),
+	VC4_REG32(V3D_BFC),
+	VC4_REG32(V3D_RFC),
+	VC4_REG32(V3D_BPCA),
+	VC4_REG32(V3D_BPCS),
+	VC4_REG32(V3D_BPOA),
+	VC4_REG32(V3D_BPOS),
+	VC4_REG32(V3D_BXCF),
+	VC4_REG32(V3D_SQRSV0),
+	VC4_REG32(V3D_SQRSV1),
+	VC4_REG32(V3D_SQCNTL),
+	VC4_REG32(V3D_SRQPC),
+	VC4_REG32(V3D_SRQUA),
+	VC4_REG32(V3D_SRQUL),
+	VC4_REG32(V3D_SRQCS),
+	VC4_REG32(V3D_VPACNTL),
+	VC4_REG32(V3D_VPMBASE),
+	VC4_REG32(V3D_PCTRC),
+	VC4_REG32(V3D_PCTRE),
+	VC4_REG32(V3D_PCTR(0)),
+	VC4_REG32(V3D_PCTRS(0)),
+	VC4_REG32(V3D_PCTR(1)),
+	VC4_REG32(V3D_PCTRS(1)),
+	VC4_REG32(V3D_PCTR(2)),
+	VC4_REG32(V3D_PCTRS(2)),
+	VC4_REG32(V3D_PCTR(3)),
+	VC4_REG32(V3D_PCTRS(3)),
+	VC4_REG32(V3D_PCTR(4)),
+	VC4_REG32(V3D_PCTRS(4)),
+	VC4_REG32(V3D_PCTR(5)),
+	VC4_REG32(V3D_PCTRS(5)),
+	VC4_REG32(V3D_PCTR(6)),
+	VC4_REG32(V3D_PCTRS(6)),
+	VC4_REG32(V3D_PCTR(7)),
+	VC4_REG32(V3D_PCTRS(7)),
+	VC4_REG32(V3D_PCTR(8)),
+	VC4_REG32(V3D_PCTRS(8)),
+	VC4_REG32(V3D_PCTR(9)),
+	VC4_REG32(V3D_PCTRS(9)),
+	VC4_REG32(V3D_PCTR(10)),
+	VC4_REG32(V3D_PCTRS(10)),
+	VC4_REG32(V3D_PCTR(11)),
+	VC4_REG32(V3D_PCTRS(11)),
+	VC4_REG32(V3D_PCTR(12)),
+	VC4_REG32(V3D_PCTRS(12)),
+	VC4_REG32(V3D_PCTR(13)),
+	VC4_REG32(V3D_PCTRS(13)),
+	VC4_REG32(V3D_PCTR(14)),
+	VC4_REG32(V3D_PCTRS(14)),
+	VC4_REG32(V3D_PCTR(15)),
+	VC4_REG32(V3D_PCTRS(15)),
+	VC4_REG32(V3D_DBGE),
+	VC4_REG32(V3D_FDBGO),
+	VC4_REG32(V3D_FDBGB),
+	VC4_REG32(V3D_FDBGR),
+	VC4_REG32(V3D_FDBGS),
+	VC4_REG32(V3D_ERRSTAT),
+};
+
+static int vc4_v3d_debugfs_ident(struct seq_file *m, void *unused)
+{
+	struct drm_debugfs_entry *entry = m->private;
+	struct drm_device *dev = entry->dev;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	int ret = vc4_v3d_pm_get(vc4);
+
+	if (ret == 0) {
+		uint32_t ident1 = V3D_READ(V3D_IDENT1);
+		uint32_t nslc = VC4_GET_FIELD(ident1, V3D_IDENT1_NSLC);
+		uint32_t tups = VC4_GET_FIELD(ident1, V3D_IDENT1_TUPS);
+		uint32_t qups = VC4_GET_FIELD(ident1, V3D_IDENT1_QUPS);
+
+		seq_printf(m, "Revision:   %d\n",
+			   VC4_GET_FIELD(ident1, V3D_IDENT1_REV));
+		seq_printf(m, "Slices:     %d\n", nslc);
+		seq_printf(m, "TMUs:       %d\n", nslc * tups);
+		seq_printf(m, "QPUs:       %d\n", nslc * qups);
+		seq_printf(m, "Semaphores: %d\n",
+			   VC4_GET_FIELD(ident1, V3D_IDENT1_NSEM));
+		vc4_v3d_pm_put(vc4);
+	}
+
+	return 0;
+}
+
+/*
+ * Wraps pm_runtime_get_sync() in a refcount, so that we can reliably
+ * get the pm_runtime refcount to 0 in vc4_reset().
+ */
+int
+vc4_v3d_pm_get(struct vc4_dev *vc4)
+{
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	mutex_lock(&vc4->power_lock);
+	if (vc4->power_refcount++ == 0) {
+		int ret = pm_runtime_get_sync(&vc4->v3d->pdev->dev);
+
+		if (ret < 0) {
+			vc4->power_refcount--;
+			mutex_unlock(&vc4->power_lock);
+			return ret;
+		}
+	}
+	mutex_unlock(&vc4->power_lock);
+
+	return 0;
+}
+
+void
+vc4_v3d_pm_put(struct vc4_dev *vc4)
+{
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	mutex_lock(&vc4->power_lock);
+	if (--vc4->power_refcount == 0) {
+		pm_runtime_mark_last_busy(&vc4->v3d->pdev->dev);
+		pm_runtime_put_autosuspend(&vc4->v3d->pdev->dev);
+	}
+	mutex_unlock(&vc4->power_lock);
+}
+
+static void vc4_v3d_init_hw(struct drm_device *dev)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+
+	/* Take all the memory that would have been reserved for user
+	 * QPU programs, since we don't have an interface for running
+	 * them, anyway.
+	 */
+	V3D_WRITE(V3D_VPMBASE, 0);
+}
+
+int vc4_v3d_get_bin_slot(struct vc4_dev *vc4)
+{
+	struct drm_device *dev = &vc4->base;
+	unsigned long irqflags;
+	int slot;
+	uint64_t seqno = 0;
+	struct vc4_exec_info *exec;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+try_again:
+	spin_lock_irqsave(&vc4->job_lock, irqflags);
+	slot = ffs(~vc4->bin_alloc_used);
+	if (slot != 0) {
+		/* Switch from ffs() bit index to a 0-based index. */
+		slot--;
+		vc4->bin_alloc_used |= BIT(slot);
+		spin_unlock_irqrestore(&vc4->job_lock, irqflags);
+		return slot;
+	}
+
+	/* Couldn't find an open slot.  Wait for render to complete
+	 * and try again.
+	 */
+	exec = vc4_last_render_job(vc4);
+	if (exec)
+		seqno = exec->seqno;
+	spin_unlock_irqrestore(&vc4->job_lock, irqflags);
+
+	if (seqno) {
+		int ret = vc4_wait_for_seqno(dev, seqno, ~0ull, true);
+
+		if (ret == 0)
+			goto try_again;
+
+		return ret;
+	}
+
+	return -ENOMEM;
+}
+
+/*
+ * bin_bo_alloc() - allocates the memory that will be used for
+ * tile binning.
+ *
+ * The binner has a limitation that the addresses in the tile state
+ * buffer that point into the tile alloc buffer or binner overflow
+ * memory only have 28 bits (256MB), and the top 4 on the bus for
+ * tile alloc references end up coming from the tile state buffer's
+ * address.
+ *
+ * To work around this, we allocate a single large buffer while V3D is
+ * in use, make sure that it has the top 4 bits constant across its
+ * entire extent, and then put the tile state, tile alloc, and binner
+ * overflow memory inside that buffer.
+ *
+ * This creates a limitation where we may not be able to execute a job
+ * if it doesn't fit within the buffer that we allocated up front.
+ * However, it turns out that 16MB is "enough for anybody", and
+ * real-world applications run into allocation failures from the
+ * overall DMA pool before they make scenes complicated enough to run
+ * out of bin space.
+ */
+static int bin_bo_alloc(struct vc4_dev *vc4)
+{
+	struct vc4_v3d *v3d = vc4->v3d;
+	uint32_t size = 16 * 1024 * 1024;
+	int ret = 0;
+	struct list_head list;
+
+	if (!v3d)
+		return -ENODEV;
+
+	/* We may need to try allocating more than once to get a BO
+	 * that doesn't cross 256MB.  Track the ones we've allocated
+	 * that failed so far, so that we can free them when we've got
+	 * one that succeeded (if we freed them right away, our next
+	 * allocation would probably be the same chunk of memory).
+	 */
+	INIT_LIST_HEAD(&list);
+
+	while (true) {
+		struct vc4_bo *bo = vc4_bo_create(&vc4->base, size, true,
+						  VC4_BO_TYPE_BIN);
+
+		if (IS_ERR(bo)) {
+			ret = PTR_ERR(bo);
+
+			dev_err(&v3d->pdev->dev,
+				"Failed to allocate memory for tile binning: "
+				"%d. You may need to enable DMA or give it "
+				"more memory.",
+				ret);
+			break;
+		}
+
+		/* Check if this BO won't trigger the addressing bug. */
+		if ((bo->base.dma_addr & 0xf0000000) ==
+		    ((bo->base.dma_addr + bo->base.base.size - 1) & 0xf0000000)) {
+			vc4->bin_bo = bo;
+
+			/* Set up for allocating 512KB chunks of
+			 * binner memory.  The biggest allocation we
+			 * need to do is for the initial tile alloc +
+			 * tile state buffer.  We can render to a
+			 * maximum of ((2048*2048) / (32*32) = 4096
+			 * tiles in a frame (until we do floating
+			 * point rendering, at which point it would be
+			 * 8192).  Tile state is 48b/tile (rounded to
+			 * a page), and tile alloc is 32b/tile
+			 * (rounded to a page), plus a page of extra,
+			 * for a total of 320kb for our worst-case.
+			 * We choose 512kb so that it divides evenly
+			 * into our 16MB, and the rest of the 512kb
+			 * will be used as storage for the overflow
+			 * from the initial 32b CL per bin.
+			 */
+			vc4->bin_alloc_size = 512 * 1024;
+			vc4->bin_alloc_used = 0;
+			vc4->bin_alloc_overflow = 0;
+			WARN_ON_ONCE(sizeof(vc4->bin_alloc_used) * 8 !=
+				     bo->base.base.size / vc4->bin_alloc_size);
+
+			kref_init(&vc4->bin_bo_kref);
+
+			/* Enable the out-of-memory interrupt to set our
+			 * newly-allocated binner BO, potentially from an
+			 * already-pending-but-masked interrupt.
+			 */
+			V3D_WRITE(V3D_INTENA, V3D_INT_OUTOMEM);
+
+			break;
+		}
+
+		/* Put it on the list to free later, and try again. */
+		list_add(&bo->unref_head, &list);
+	}
+
+	/* Free all the BOs we allocated but didn't choose. */
+	while (!list_empty(&list)) {
+		struct vc4_bo *bo = list_last_entry(&list,
+						    struct vc4_bo, unref_head);
+
+		list_del(&bo->unref_head);
+		drm_gem_object_put(&bo->base.base);
+	}
+
+	return ret;
+}
+
+int vc4_v3d_bin_bo_get(struct vc4_dev *vc4, bool *used)
+{
+	int ret = 0;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	mutex_lock(&vc4->bin_bo_lock);
+
+	if (used && *used)
+		goto complete;
+
+	if (vc4->bin_bo)
+		kref_get(&vc4->bin_bo_kref);
+	else
+		ret = bin_bo_alloc(vc4);
+
+	if (ret == 0 && used)
+		*used = true;
+
+complete:
+	mutex_unlock(&vc4->bin_bo_lock);
+
+	return ret;
+}
+
+static void bin_bo_release(struct kref *ref)
+{
+	struct vc4_dev *vc4 = container_of(ref, struct vc4_dev, bin_bo_kref);
+
+	if (WARN_ON_ONCE(!vc4->bin_bo))
+		return;
+
+	drm_gem_object_put(&vc4->bin_bo->base.base);
+	vc4->bin_bo = NULL;
+}
+
+void vc4_v3d_bin_bo_put(struct vc4_dev *vc4)
+{
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return;
+
+	mutex_lock(&vc4->bin_bo_lock);
+	kref_put(&vc4->bin_bo_kref, bin_bo_release);
+	mutex_unlock(&vc4->bin_bo_lock);
+}
+
+#ifdef CONFIG_PM
+static int vc4_v3d_runtime_suspend(struct device *dev)
+{
+	struct vc4_v3d *v3d = dev_get_drvdata(dev);
+	struct vc4_dev *vc4 = v3d->vc4;
+
+	vc4_irq_disable(&vc4->base);
+
+	clk_disable_unprepare(v3d->clk);
+
+	return 0;
+}
+
+static int vc4_v3d_runtime_resume(struct device *dev)
+{
+	struct vc4_v3d *v3d = dev_get_drvdata(dev);
+	struct vc4_dev *vc4 = v3d->vc4;
+	int ret;
+
+	ret = clk_prepare_enable(v3d->clk);
+	if (ret != 0)
+		return ret;
+
+	vc4_v3d_init_hw(&vc4->base);
+
+	vc4_irq_enable(&vc4->base);
+
+	return 0;
+}
+#endif
+
+int vc4_v3d_debugfs_init(struct drm_minor *minor)
+{
+	struct drm_device *drm = minor->dev;
+	struct vc4_dev *vc4 = to_vc4_dev(drm);
+	struct vc4_v3d *v3d = vc4->v3d;
+
+	if (!vc4->v3d)
+		return -ENODEV;
+
+	drm_debugfs_add_file(drm, "v3d_ident", vc4_v3d_debugfs_ident, NULL);
+
+	vc4_debugfs_add_regset32(drm, "v3d_regs", &v3d->regset);
+
+	return 0;
+}
+
+static int vc4_v3d_bind(struct device *dev, struct device *master, void *data)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct drm_device *drm = dev_get_drvdata(master);
+	struct vc4_dev *vc4 = to_vc4_dev(drm);
+	struct vc4_v3d *v3d = NULL;
+	int ret;
+
+	v3d = devm_kzalloc(&pdev->dev, sizeof(*v3d), GFP_KERNEL);
+	if (!v3d)
+		return -ENOMEM;
+
+	dev_set_drvdata(dev, v3d);
+
+	v3d->pdev = pdev;
+
+	v3d->regs = vc4_ioremap_regs(pdev, 0);
+	if (IS_ERR(v3d->regs))
+		return PTR_ERR(v3d->regs);
+	v3d->regset.base = v3d->regs;
+	v3d->regset.regs = v3d_regs;
+	v3d->regset.nregs = ARRAY_SIZE(v3d_regs);
+
+	vc4->v3d = v3d;
+	v3d->vc4 = vc4;
+
+	v3d->clk = devm_clk_get_optional(dev, NULL);
+	if (IS_ERR(v3d->clk))
+		return dev_err_probe(dev, PTR_ERR(v3d->clk), "Failed to get V3D clock\n");
+
+	ret = platform_get_irq(pdev, 0);
+	if (ret < 0)
+		return ret;
+	vc4->irq = ret;
+
+	ret = devm_pm_runtime_enable(dev);
+	if (ret)
+		return ret;
+
+	ret = pm_runtime_resume_and_get(dev);
+	if (ret)
+		return ret;
+
+	if (V3D_READ(V3D_IDENT0) != V3D_EXPECTED_IDENT0) {
+		drm_err(drm, "V3D_IDENT0 read 0x%08x instead of 0x%08x\n",
+			V3D_READ(V3D_IDENT0), V3D_EXPECTED_IDENT0);
+		ret = -EINVAL;
+		goto err_put_runtime_pm;
+	}
+
+	/* Reset the binner overflow address/size at setup, to be sure
+	 * we don't reuse an old one.
+	 */
+	V3D_WRITE(V3D_BPOA, 0);
+	V3D_WRITE(V3D_BPOS, 0);
+
+	ret = vc4_irq_install(drm, vc4->irq);
+	if (ret) {
+		drm_err(drm, "Failed to install IRQ handler\n");
+		goto err_put_runtime_pm;
+	}
+
+	pm_runtime_use_autosuspend(dev);
+	pm_runtime_set_autosuspend_delay(dev, 40); /* a little over 2 frames. */
+	pm_runtime_put_autosuspend(dev);
+
+	return 0;
+
+err_put_runtime_pm:
+	pm_runtime_put(dev);
+
+	return ret;
+}
+
+static void vc4_v3d_unbind(struct device *dev, struct device *master,
+			   void *data)
+{
+	struct drm_device *drm = data;
+	struct vc4_dev *vc4 = to_vc4_dev(drm);
+
+	vc4_irq_uninstall(drm);
+
+	/* Disable the binner's overflow memory address, so the next
+	 * driver probe (if any) doesn't try to reuse our old
+	 * allocation.
+	 */
+	V3D_WRITE(V3D_BPOA, 0);
+	V3D_WRITE(V3D_BPOS, 0);
+
+	vc4->v3d = NULL;
+}
+
+static const struct dev_pm_ops vc4_v3d_pm_ops = {
+	SET_RUNTIME_PM_OPS(vc4_v3d_runtime_suspend, vc4_v3d_runtime_resume, NULL)
+};
+
+static const struct component_ops vc4_v3d_ops = {
+	.bind   = vc4_v3d_bind,
+	.unbind = vc4_v3d_unbind,
+};
+
+static int vc4_v3d_dev_probe(struct platform_device *pdev)
+{
+	return component_add(&pdev->dev, &vc4_v3d_ops);
+}
+
+static void vc4_v3d_dev_remove(struct platform_device *pdev)
+{
+	component_del(&pdev->dev, &vc4_v3d_ops);
+}
+
+const struct of_device_id vc4_v3d_dt_match[] = {
+	{ .compatible = "brcm,bcm2835-v3d" },
+	{ .compatible = "brcm,cygnus-v3d" },
+	{ .compatible = "brcm,vc4-v3d" },
+	{}
+};
+
+struct platform_driver vc4_v3d_driver = {
+	.probe = vc4_v3d_dev_probe,
+	.remove = vc4_v3d_dev_remove,
+	.driver = {
+		.name = "vc4_v3d",
+		.of_match_table = vc4_v3d_dt_match,
+		.pm = &vc4_v3d_pm_ops,
+	},
+};

linux-7.2.3/drivers/gpu/drm/vc4/vc4_validate.c

@@ -0,0 +1,962 @@
+/*
+ * Copyright © 2014 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/**
+ * DOC: Command list validator for VC4.
+ *
+ * Since the VC4 has no IOMMU between it and system memory, a user
+ * with access to execute command lists could escalate privilege by
+ * overwriting system memory (drawing to it as a framebuffer) or
+ * reading system memory it shouldn't (reading it as a vertex buffer
+ * or index buffer)
+ *
+ * We validate binner command lists to ensure that all accesses are
+ * within the bounds of the GEM objects referenced by the submitted
+ * job.  It explicitly whitelists packets, and looks at the offsets in
+ * any address fields to make sure they're contained within the BOs
+ * they reference.
+ *
+ * Note that because CL validation is already reading the
+ * user-submitted CL and writing the validated copy out to the memory
+ * that the GPU will actually read, this is also where GEM relocation
+ * processing (turning BO references into actual addresses for the GPU
+ * to use) happens.
+ */
+
+#include <drm/drm_print.h>
+
+#include "vc4_drv.h"
+#include "vc4_packet.h"
+
+#define VALIDATE_ARGS \
+	struct vc4_exec_info *exec,			\
+	void *validated,				\
+	void *untrusted
+
+/** Return the width in pixels of a 64-byte microtile. */
+static uint32_t
+utile_width(int cpp)
+{
+	switch (cpp) {
+	case 1:
+	case 2:
+		return 8;
+	case 4:
+		return 4;
+	case 8:
+		return 2;
+	default:
+		pr_err("unknown cpp: %d\n", cpp);
+		return 1;
+	}
+}
+
+/** Return the height in pixels of a 64-byte microtile. */
+static uint32_t
+utile_height(int cpp)
+{
+	switch (cpp) {
+	case 1:
+		return 8;
+	case 2:
+	case 4:
+	case 8:
+		return 4;
+	default:
+		pr_err("unknown cpp: %d\n", cpp);
+		return 1;
+	}
+}
+
+/**
+ * size_is_lt() - Returns whether a miplevel of the given size will
+ * use the lineartile (LT) tiling layout rather than the normal T
+ * tiling layout.
+ * @width: Width in pixels of the miplevel
+ * @height: Height in pixels of the miplevel
+ * @cpp: Bytes per pixel of the pixel format
+ */
+static bool
+size_is_lt(uint32_t width, uint32_t height, int cpp)
+{
+	return (width <= 4 * utile_width(cpp) ||
+		height <= 4 * utile_height(cpp));
+}
+
+struct drm_gem_dma_object *
+vc4_use_bo(struct vc4_exec_info *exec, uint32_t hindex)
+{
+	struct vc4_dev *vc4 = exec->dev;
+	struct drm_gem_dma_object *obj;
+	struct vc4_bo *bo;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return NULL;
+
+	if (hindex >= exec->bo_count) {
+		DRM_DEBUG("BO index %d greater than BO count %d\n",
+			  hindex, exec->bo_count);
+		return NULL;
+	}
+	obj = to_drm_gem_dma_obj(exec->bo[hindex]);
+	bo = to_vc4_bo(&obj->base);
+
+	if (bo->validated_shader) {
+		DRM_DEBUG("Trying to use shader BO as something other than "
+			  "a shader\n");
+		return NULL;
+	}
+
+	return obj;
+}
+
+static struct drm_gem_dma_object *
+vc4_use_handle(struct vc4_exec_info *exec, uint32_t gem_handles_packet_index)
+{
+	return vc4_use_bo(exec, exec->bo_index[gem_handles_packet_index]);
+}
+
+static bool
+validate_bin_pos(struct vc4_exec_info *exec, void *untrusted, uint32_t pos)
+{
+	/* Note that the untrusted pointer passed to these functions is
+	 * incremented past the packet byte.
+	 */
+	return (untrusted - 1 == exec->bin_u + pos);
+}
+
+static uint32_t
+gl_shader_rec_size(uint32_t pointer_bits)
+{
+	uint32_t attribute_count = pointer_bits & 7;
+	bool extended = pointer_bits & 8;
+
+	if (attribute_count == 0)
+		attribute_count = 8;
+
+	if (extended)
+		return 100 + attribute_count * 4;
+	else
+		return 36 + attribute_count * 8;
+}
+
+bool
+vc4_check_tex_size(struct vc4_exec_info *exec, struct drm_gem_dma_object *fbo,
+		   uint32_t offset, uint8_t tiling_format,
+		   uint32_t width, uint32_t height, uint8_t cpp)
+{
+	struct vc4_dev *vc4 = exec->dev;
+	uint32_t aligned_width, aligned_height, stride, size;
+	uint32_t utile_w = utile_width(cpp);
+	uint32_t utile_h = utile_height(cpp);
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return false;
+
+	/* The shaded vertex format stores signed 12.4 fixed point
+	 * (-2048,2047) offsets from the viewport center, so we should
+	 * never have a render target larger than 4096.  The texture
+	 * unit can only sample from 2048x2048, so it's even more
+	 * restricted.  This lets us avoid worrying about overflow in
+	 * our math.
+	 */
+	if (width > 4096 || height > 4096) {
+		DRM_DEBUG("Surface dimensions (%d,%d) too large",
+			  width, height);
+		return false;
+	}
+
+	switch (tiling_format) {
+	case VC4_TILING_FORMAT_LINEAR:
+		aligned_width = round_up(width, utile_w);
+		aligned_height = height;
+		break;
+	case VC4_TILING_FORMAT_T:
+		aligned_width = round_up(width, utile_w * 8);
+		aligned_height = round_up(height, utile_h * 8);
+		break;
+	case VC4_TILING_FORMAT_LT:
+		aligned_width = round_up(width, utile_w);
+		aligned_height = round_up(height, utile_h);
+		break;
+	default:
+		DRM_DEBUG("buffer tiling %d unsupported\n", tiling_format);
+		return false;
+	}
+
+	stride = aligned_width * cpp;
+	size = stride * aligned_height;
+
+	if (size + offset < size ||
+	    size + offset > fbo->base.size) {
+		DRM_DEBUG("Overflow in %dx%d (%dx%d) fbo size (%d + %d > %zd)\n",
+			  width, height,
+			  aligned_width, aligned_height,
+			  size, offset, fbo->base.size);
+		return false;
+	}
+
+	return true;
+}
+
+static int
+validate_flush(VALIDATE_ARGS)
+{
+	if (!validate_bin_pos(exec, untrusted, exec->args->bin_cl_size - 1)) {
+		DRM_DEBUG("Bin CL must end with VC4_PACKET_FLUSH\n");
+		return -EINVAL;
+	}
+	exec->found_flush = true;
+
+	return 0;
+}
+
+static int
+validate_start_tile_binning(VALIDATE_ARGS)
+{
+	if (exec->found_start_tile_binning_packet) {
+		DRM_DEBUG("Duplicate VC4_PACKET_START_TILE_BINNING\n");
+		return -EINVAL;
+	}
+	exec->found_start_tile_binning_packet = true;
+
+	if (!exec->found_tile_binning_mode_config_packet) {
+		DRM_DEBUG("missing VC4_PACKET_TILE_BINNING_MODE_CONFIG\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+validate_increment_semaphore(VALIDATE_ARGS)
+{
+	if (!validate_bin_pos(exec, untrusted, exec->args->bin_cl_size - 2)) {
+		DRM_DEBUG("Bin CL must end with "
+			  "VC4_PACKET_INCREMENT_SEMAPHORE\n");
+		return -EINVAL;
+	}
+	exec->found_increment_semaphore_packet = true;
+
+	return 0;
+}
+
+static int
+validate_indexed_prim_list(VALIDATE_ARGS)
+{
+	struct drm_gem_dma_object *ib;
+	uint32_t length = *(uint32_t *)(untrusted + 1);
+	uint32_t offset = *(uint32_t *)(untrusted + 5);
+	uint32_t max_index = *(uint32_t *)(untrusted + 9);
+	uint32_t index_size = (*(uint8_t *)(untrusted + 0) >> 4) ? 2 : 1;
+	struct vc4_shader_state *shader_state;
+
+	/* Check overflow condition */
+	if (exec->shader_state_count == 0) {
+		DRM_DEBUG("shader state must precede primitives\n");
+		return -EINVAL;
+	}
+	shader_state = &exec->shader_state[exec->shader_state_count - 1];
+
+	if (max_index > shader_state->max_index)
+		shader_state->max_index = max_index;
+
+	ib = vc4_use_handle(exec, 0);
+	if (!ib)
+		return -EINVAL;
+
+	if (offset > ib->base.size ||
+	    (ib->base.size - offset) / index_size < length) {
+		DRM_DEBUG("IB access overflow (%d + %d*%d > %zd)\n",
+			  offset, length, index_size, ib->base.size);
+		return -EINVAL;
+	}
+
+	*(uint32_t *)(validated + 5) = ib->dma_addr + offset;
+
+	return 0;
+}
+
+static int
+validate_gl_array_primitive(VALIDATE_ARGS)
+{
+	uint32_t length = *(uint32_t *)(untrusted + 1);
+	uint32_t base_index = *(uint32_t *)(untrusted + 5);
+	uint32_t max_index;
+	struct vc4_shader_state *shader_state;
+
+	/* Check overflow condition */
+	if (exec->shader_state_count == 0) {
+		DRM_DEBUG("shader state must precede primitives\n");
+		return -EINVAL;
+	}
+	shader_state = &exec->shader_state[exec->shader_state_count - 1];
+
+	if (length + base_index < length) {
+		DRM_DEBUG("primitive vertex count overflow\n");
+		return -EINVAL;
+	}
+	max_index = length + base_index - 1;
+
+	if (max_index > shader_state->max_index)
+		shader_state->max_index = max_index;
+
+	return 0;
+}
+
+static int
+validate_gl_shader_state(VALIDATE_ARGS)
+{
+	uint32_t i = exec->shader_state_count++;
+
+	if (i >= exec->shader_state_size) {
+		DRM_DEBUG("More requests for shader states than declared\n");
+		return -EINVAL;
+	}
+
+	exec->shader_state[i].addr = *(uint32_t *)untrusted;
+	exec->shader_state[i].max_index = 0;
+
+	if (exec->shader_state[i].addr & ~0xf) {
+		DRM_DEBUG("high bits set in GL shader rec reference\n");
+		return -EINVAL;
+	}
+
+	*(uint32_t *)validated = (exec->shader_rec_p +
+				  exec->shader_state[i].addr);
+
+	exec->shader_rec_p +=
+		roundup(gl_shader_rec_size(exec->shader_state[i].addr), 16);
+
+	return 0;
+}
+
+static int
+validate_tile_binning_config(VALIDATE_ARGS)
+{
+	struct drm_device *dev = exec->exec_bo->base.dev;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	uint8_t flags;
+	uint32_t tile_state_size;
+	uint32_t tile_count, bin_addr;
+	int bin_slot;
+
+	if (exec->found_tile_binning_mode_config_packet) {
+		DRM_DEBUG("Duplicate VC4_PACKET_TILE_BINNING_MODE_CONFIG\n");
+		return -EINVAL;
+	}
+	exec->found_tile_binning_mode_config_packet = true;
+
+	exec->bin_tiles_x = *(uint8_t *)(untrusted + 12);
+	exec->bin_tiles_y = *(uint8_t *)(untrusted + 13);
+	tile_count = exec->bin_tiles_x * exec->bin_tiles_y;
+	flags = *(uint8_t *)(untrusted + 14);
+
+	if (exec->bin_tiles_x == 0 ||
+	    exec->bin_tiles_y == 0) {
+		DRM_DEBUG("Tile binning config of %dx%d too small\n",
+			  exec->bin_tiles_x, exec->bin_tiles_y);
+		return -EINVAL;
+	}
+
+	if (flags & (VC4_BIN_CONFIG_DB_NON_MS |
+		     VC4_BIN_CONFIG_TILE_BUFFER_64BIT)) {
+		DRM_DEBUG("unsupported binning config flags 0x%02x\n", flags);
+		return -EINVAL;
+	}
+
+	/* The tile state data array is 48 bytes per tile, and we put it at
+	 * the start of a BO containing both it and the tile alloc.
+	 */
+	tile_state_size = 48 * tile_count;
+
+	/* Since the tile alloc array will follow us, align. */
+	tile_state_size = roundup(tile_state_size, 4096);
+
+	/* Reject configurations whose tile state would leave no room for
+	 * the tile alloc pool that follows it in the slot.
+	 */
+	if (tile_state_size >= vc4->bin_alloc_size) {
+		DRM_DEBUG("Tile binning config of %dx%d too large\n",
+			  exec->bin_tiles_x, exec->bin_tiles_y);
+		return -EINVAL;
+	}
+
+	bin_slot = vc4_v3d_get_bin_slot(vc4);
+	if (bin_slot < 0) {
+		if (bin_slot != -EINTR && bin_slot != -ERESTARTSYS) {
+			drm_err(dev, "Failed to allocate binner memory: %d\n",
+				bin_slot);
+		}
+		return bin_slot;
+	}
+
+	/* The slot we allocated will only be used by this job, and is
+	 * free when the job completes rendering.
+	 */
+	exec->bin_slots |= BIT(bin_slot);
+	bin_addr = vc4->bin_bo->base.dma_addr + bin_slot * vc4->bin_alloc_size;
+
+	exec->tile_alloc_offset = bin_addr + tile_state_size;
+
+	/* The TSDA area must be zeroed out before use, otherwise the PTB might
+	 * consume a stale tile state.
+	 */
+	memset(vc4->bin_bo->base.vaddr + bin_slot * vc4->bin_alloc_size, 0,
+	       tile_state_size);
+
+	*(uint8_t *)(validated + 14) =
+		((flags & ~(VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_MASK |
+			    VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_MASK)) |
+		 VC4_BIN_CONFIG_AUTO_INIT_TSDA |
+		 VC4_SET_FIELD(VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE_32,
+			       VC4_BIN_CONFIG_ALLOC_INIT_BLOCK_SIZE) |
+		 VC4_SET_FIELD(VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE_128,
+			       VC4_BIN_CONFIG_ALLOC_BLOCK_SIZE));
+
+	/* tile alloc address. */
+	*(uint32_t *)(validated + 0) = exec->tile_alloc_offset;
+	/* tile alloc size. */
+	*(uint32_t *)(validated + 4) = (bin_addr + vc4->bin_alloc_size -
+					exec->tile_alloc_offset);
+	/* tile state address. */
+	*(uint32_t *)(validated + 8) = bin_addr;
+
+	return 0;
+}
+
+static int
+validate_gem_handles(VALIDATE_ARGS)
+{
+	memcpy(exec->bo_index, untrusted, sizeof(exec->bo_index));
+	return 0;
+}
+
+#define VC4_DEFINE_PACKET(packet, func) \
+	[packet] = { packet ## _SIZE, #packet, func }
+
+static const struct cmd_info {
+	uint16_t len;
+	const char *name;
+	int (*func)(struct vc4_exec_info *exec, void *validated,
+		    void *untrusted);
+} cmd_info[] = {
+	VC4_DEFINE_PACKET(VC4_PACKET_HALT, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_NOP, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH, validate_flush),
+	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH_ALL, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_START_TILE_BINNING,
+			  validate_start_tile_binning),
+	VC4_DEFINE_PACKET(VC4_PACKET_INCREMENT_SEMAPHORE,
+			  validate_increment_semaphore),
+
+	VC4_DEFINE_PACKET(VC4_PACKET_GL_INDEXED_PRIMITIVE,
+			  validate_indexed_prim_list),
+	VC4_DEFINE_PACKET(VC4_PACKET_GL_ARRAY_PRIMITIVE,
+			  validate_gl_array_primitive),
+
+	VC4_DEFINE_PACKET(VC4_PACKET_PRIMITIVE_LIST_FORMAT, NULL),
+
+	VC4_DEFINE_PACKET(VC4_PACKET_GL_SHADER_STATE, validate_gl_shader_state),
+
+	VC4_DEFINE_PACKET(VC4_PACKET_CONFIGURATION_BITS, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_FLAT_SHADE_FLAGS, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_POINT_SIZE, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_LINE_WIDTH, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_RHT_X_BOUNDARY, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_DEPTH_OFFSET, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_CLIP_WINDOW, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_VIEWPORT_OFFSET, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_CLIPPER_XY_SCALING, NULL),
+	/* Note: The docs say this was also 105, but it was 106 in the
+	 * initial userland code drop.
+	 */
+	VC4_DEFINE_PACKET(VC4_PACKET_CLIPPER_Z_SCALING, NULL),
+
+	VC4_DEFINE_PACKET(VC4_PACKET_TILE_BINNING_MODE_CONFIG,
+			  validate_tile_binning_config),
+
+	VC4_DEFINE_PACKET(VC4_PACKET_GEM_HANDLES, validate_gem_handles),
+};
+
+int
+vc4_validate_bin_cl(struct drm_device *dev,
+		    void *validated,
+		    void *unvalidated,
+		    struct vc4_exec_info *exec)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	uint32_t len = exec->args->bin_cl_size;
+	uint32_t dst_offset = 0;
+	uint32_t src_offset = 0;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	while (src_offset < len) {
+		void *dst_pkt = validated + dst_offset;
+		void *src_pkt = unvalidated + src_offset;
+		u8 cmd = *(uint8_t *)src_pkt;
+		const struct cmd_info *info;
+
+		if (cmd >= ARRAY_SIZE(cmd_info)) {
+			DRM_DEBUG("0x%08x: packet %d out of bounds\n",
+				  src_offset, cmd);
+			return -EINVAL;
+		}
+
+		info = &cmd_info[cmd];
+		if (!info->name) {
+			DRM_DEBUG("0x%08x: packet %d invalid\n",
+				  src_offset, cmd);
+			return -EINVAL;
+		}
+
+		if (src_offset + info->len > len) {
+			DRM_DEBUG("0x%08x: packet %d (%s) length 0x%08x "
+				  "exceeds bounds (0x%08x)\n",
+				  src_offset, cmd, info->name, info->len,
+				  src_offset + len);
+			return -EINVAL;
+		}
+
+		if (cmd != VC4_PACKET_GEM_HANDLES)
+			memcpy(dst_pkt, src_pkt, info->len);
+
+		if (info->func && info->func(exec,
+					     dst_pkt + 1,
+					     src_pkt + 1)) {
+			DRM_DEBUG("0x%08x: packet %d (%s) failed to validate\n",
+				  src_offset, cmd, info->name);
+			return -EINVAL;
+		}
+
+		src_offset += info->len;
+		/* GEM handle loading doesn't produce HW packets. */
+		if (cmd != VC4_PACKET_GEM_HANDLES)
+			dst_offset += info->len;
+
+		/* When the CL hits halt, it'll stop reading anything else. */
+		if (cmd == VC4_PACKET_HALT)
+			break;
+	}
+
+	exec->ct0ea = exec->ct0ca + dst_offset;
+
+	if (!exec->found_start_tile_binning_packet) {
+		DRM_DEBUG("Bin CL missing VC4_PACKET_START_TILE_BINNING\n");
+		return -EINVAL;
+	}
+
+	/* The bin CL must be ended with INCREMENT_SEMAPHORE and FLUSH.  The
+	 * semaphore is used to trigger the render CL to start up, and the
+	 * FLUSH is what caps the bin lists with
+	 * VC4_PACKET_RETURN_FROM_SUB_LIST (so they jump back to the main
+	 * render CL when they get called to) and actually triggers the queued
+	 * semaphore increment.
+	 */
+	if (!exec->found_increment_semaphore_packet || !exec->found_flush) {
+		DRM_DEBUG("Bin CL missing VC4_PACKET_INCREMENT_SEMAPHORE + "
+			  "VC4_PACKET_FLUSH\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static bool
+reloc_tex(struct vc4_exec_info *exec,
+	  void *uniform_data_u,
+	  struct vc4_texture_sample_info *sample,
+	  uint32_t texture_handle_index, bool is_cs)
+{
+	struct drm_gem_dma_object *tex;
+	uint32_t p0 = *(uint32_t *)(uniform_data_u + sample->p_offset[0]);
+	uint32_t p1 = *(uint32_t *)(uniform_data_u + sample->p_offset[1]);
+	uint32_t p2 = (sample->p_offset[2] != ~0 ?
+		       *(uint32_t *)(uniform_data_u + sample->p_offset[2]) : 0);
+	uint32_t p3 = (sample->p_offset[3] != ~0 ?
+		       *(uint32_t *)(uniform_data_u + sample->p_offset[3]) : 0);
+	uint32_t *validated_p0 = exec->uniforms_v + sample->p_offset[0];
+	uint32_t offset = p0 & VC4_TEX_P0_OFFSET_MASK;
+	uint32_t miplevels = VC4_GET_FIELD(p0, VC4_TEX_P0_MIPLVLS);
+	uint32_t width = VC4_GET_FIELD(p1, VC4_TEX_P1_WIDTH);
+	uint32_t height = VC4_GET_FIELD(p1, VC4_TEX_P1_HEIGHT);
+	uint32_t cpp, tiling_format, utile_w, utile_h;
+	uint32_t i;
+	uint32_t cube_map_stride = 0;
+	enum vc4_texture_data_type type;
+
+	tex = vc4_use_bo(exec, texture_handle_index);
+	if (!tex)
+		return false;
+
+	if (sample->is_direct) {
+		uint32_t remaining_size = tex->base.size - p0;
+
+		if (p0 > tex->base.size - 4) {
+			DRM_DEBUG("UBO offset greater than UBO size\n");
+			goto fail;
+		}
+		if (p1 > remaining_size - 4) {
+			DRM_DEBUG("UBO clamp would allow reads "
+				  "outside of UBO\n");
+			goto fail;
+		}
+		*validated_p0 = tex->dma_addr + p0;
+		return true;
+	}
+
+	if (width == 0)
+		width = 2048;
+	if (height == 0)
+		height = 2048;
+
+	if (p0 & VC4_TEX_P0_CMMODE_MASK) {
+		if (VC4_GET_FIELD(p2, VC4_TEX_P2_PTYPE) ==
+		    VC4_TEX_P2_PTYPE_CUBE_MAP_STRIDE)
+			cube_map_stride = p2 & VC4_TEX_P2_CMST_MASK;
+		if (VC4_GET_FIELD(p3, VC4_TEX_P2_PTYPE) ==
+		    VC4_TEX_P2_PTYPE_CUBE_MAP_STRIDE) {
+			if (cube_map_stride) {
+				DRM_DEBUG("Cube map stride set twice\n");
+				goto fail;
+			}
+
+			cube_map_stride = p3 & VC4_TEX_P2_CMST_MASK;
+		}
+		if (!cube_map_stride) {
+			DRM_DEBUG("Cube map stride not set\n");
+			goto fail;
+		}
+	}
+
+	type = (VC4_GET_FIELD(p0, VC4_TEX_P0_TYPE) |
+		(VC4_GET_FIELD(p1, VC4_TEX_P1_TYPE4) << 4));
+
+	switch (type) {
+	case VC4_TEXTURE_TYPE_RGBA8888:
+	case VC4_TEXTURE_TYPE_RGBX8888:
+	case VC4_TEXTURE_TYPE_RGBA32R:
+		cpp = 4;
+		break;
+	case VC4_TEXTURE_TYPE_RGBA4444:
+	case VC4_TEXTURE_TYPE_RGBA5551:
+	case VC4_TEXTURE_TYPE_RGB565:
+	case VC4_TEXTURE_TYPE_LUMALPHA:
+	case VC4_TEXTURE_TYPE_S16F:
+	case VC4_TEXTURE_TYPE_S16:
+		cpp = 2;
+		break;
+	case VC4_TEXTURE_TYPE_LUMINANCE:
+	case VC4_TEXTURE_TYPE_ALPHA:
+	case VC4_TEXTURE_TYPE_S8:
+		cpp = 1;
+		break;
+	case VC4_TEXTURE_TYPE_ETC1:
+		/* ETC1 is arranged as 64-bit blocks, where each block is 4x4
+		 * pixels.
+		 */
+		cpp = 8;
+		width = (width + 3) >> 2;
+		height = (height + 3) >> 2;
+		break;
+	case VC4_TEXTURE_TYPE_BW1:
+	case VC4_TEXTURE_TYPE_A4:
+	case VC4_TEXTURE_TYPE_A1:
+	case VC4_TEXTURE_TYPE_RGBA64:
+	case VC4_TEXTURE_TYPE_YUV422R:
+	default:
+		DRM_DEBUG("Texture format %d unsupported\n", type);
+		goto fail;
+	}
+	utile_w = utile_width(cpp);
+	utile_h = utile_height(cpp);
+
+	if (type == VC4_TEXTURE_TYPE_RGBA32R) {
+		tiling_format = VC4_TILING_FORMAT_LINEAR;
+	} else {
+		if (size_is_lt(width, height, cpp))
+			tiling_format = VC4_TILING_FORMAT_LT;
+		else
+			tiling_format = VC4_TILING_FORMAT_T;
+	}
+
+	if (!vc4_check_tex_size(exec, tex, offset + cube_map_stride * 5,
+				tiling_format, width, height, cpp)) {
+		goto fail;
+	}
+
+	/* The mipmap levels are stored before the base of the texture.  Make
+	 * sure there is actually space in the BO.
+	 */
+	for (i = 1; i <= miplevels; i++) {
+		uint32_t level_width = max(width >> i, 1u);
+		uint32_t level_height = max(height >> i, 1u);
+		uint32_t aligned_width, aligned_height;
+		uint32_t level_size;
+
+		/* Once the levels get small enough, they drop from T to LT. */
+		if (tiling_format == VC4_TILING_FORMAT_T &&
+		    size_is_lt(level_width, level_height, cpp)) {
+			tiling_format = VC4_TILING_FORMAT_LT;
+		}
+
+		switch (tiling_format) {
+		case VC4_TILING_FORMAT_T:
+			aligned_width = round_up(level_width, utile_w * 8);
+			aligned_height = round_up(level_height, utile_h * 8);
+			break;
+		case VC4_TILING_FORMAT_LT:
+			aligned_width = round_up(level_width, utile_w);
+			aligned_height = round_up(level_height, utile_h);
+			break;
+		default:
+			aligned_width = round_up(level_width, utile_w);
+			aligned_height = level_height;
+			break;
+		}
+
+		level_size = aligned_width * cpp * aligned_height;
+
+		if (offset < level_size) {
+			DRM_DEBUG("Level %d (%dx%d -> %dx%d) size %db "
+				  "overflowed buffer bounds (offset %d)\n",
+				  i, level_width, level_height,
+				  aligned_width, aligned_height,
+				  level_size, offset);
+			goto fail;
+		}
+
+		offset -= level_size;
+	}
+
+	*validated_p0 = tex->dma_addr + p0;
+
+	return true;
+ fail:
+	DRM_INFO("Texture p0 at %d: 0x%08x\n", sample->p_offset[0], p0);
+	DRM_INFO("Texture p1 at %d: 0x%08x\n", sample->p_offset[1], p1);
+	DRM_INFO("Texture p2 at %d: 0x%08x\n", sample->p_offset[2], p2);
+	DRM_INFO("Texture p3 at %d: 0x%08x\n", sample->p_offset[3], p3);
+	return false;
+}
+
+static int
+validate_gl_shader_rec(struct drm_device *dev,
+		       struct vc4_exec_info *exec,
+		       struct vc4_shader_state *state)
+{
+	uint32_t *src_handles;
+	void *pkt_u, *pkt_v;
+	static const uint32_t shader_reloc_offsets[] = {
+		4, /* fs */
+		16, /* vs */
+		28, /* cs */
+	};
+	uint32_t shader_reloc_count = ARRAY_SIZE(shader_reloc_offsets);
+	struct drm_gem_dma_object *bo[ARRAY_SIZE(shader_reloc_offsets) + 8];
+	uint32_t nr_attributes, nr_relocs, packet_size;
+	int i;
+
+	nr_attributes = state->addr & 0x7;
+	if (nr_attributes == 0)
+		nr_attributes = 8;
+	packet_size = gl_shader_rec_size(state->addr);
+
+	nr_relocs = ARRAY_SIZE(shader_reloc_offsets) + nr_attributes;
+	if (nr_relocs * 4 > exec->shader_rec_size) {
+		DRM_DEBUG("overflowed shader recs reading %d handles "
+			  "from %d bytes left\n",
+			  nr_relocs, exec->shader_rec_size);
+		return -EINVAL;
+	}
+	src_handles = exec->shader_rec_u;
+	exec->shader_rec_u += nr_relocs * 4;
+	exec->shader_rec_size -= nr_relocs * 4;
+
+	if (packet_size > exec->shader_rec_size) {
+		DRM_DEBUG("overflowed shader recs copying %db packet "
+			  "from %d bytes left\n",
+			  packet_size, exec->shader_rec_size);
+		return -EINVAL;
+	}
+	pkt_u = exec->shader_rec_u;
+	pkt_v = exec->shader_rec_v;
+	memcpy(pkt_v, pkt_u, packet_size);
+	exec->shader_rec_u += packet_size;
+	/* Shader recs have to be aligned to 16 bytes (due to the attribute
+	 * flags being in the low bytes), so round the next validated shader
+	 * rec address up.  This should be safe, since we've got so many
+	 * relocations in a shader rec packet.
+	 */
+	BUG_ON(roundup(packet_size, 16) - packet_size > nr_relocs * 4);
+	exec->shader_rec_v += roundup(packet_size, 16);
+	exec->shader_rec_size -= packet_size;
+
+	for (i = 0; i < shader_reloc_count; i++) {
+		if (src_handles[i] > exec->bo_count) {
+			DRM_DEBUG("Shader handle %d too big\n", src_handles[i]);
+			return -EINVAL;
+		}
+
+		bo[i] = to_drm_gem_dma_obj(exec->bo[src_handles[i]]);
+		if (!bo[i])
+			return -EINVAL;
+	}
+	for (i = shader_reloc_count; i < nr_relocs; i++) {
+		bo[i] = vc4_use_bo(exec, src_handles[i]);
+		if (!bo[i])
+			return -EINVAL;
+	}
+
+	if (((*(uint16_t *)pkt_u & VC4_SHADER_FLAG_FS_SINGLE_THREAD) == 0) !=
+	    to_vc4_bo(&bo[0]->base)->validated_shader->is_threaded) {
+		DRM_DEBUG("Thread mode of CL and FS do not match\n");
+		return -EINVAL;
+	}
+
+	if (to_vc4_bo(&bo[1]->base)->validated_shader->is_threaded ||
+	    to_vc4_bo(&bo[2]->base)->validated_shader->is_threaded) {
+		DRM_DEBUG("cs and vs cannot be threaded\n");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < shader_reloc_count; i++) {
+		struct vc4_validated_shader_info *validated_shader;
+		uint32_t o = shader_reloc_offsets[i];
+		uint32_t src_offset = *(uint32_t *)(pkt_u + o);
+		uint32_t *texture_handles_u;
+		void *uniform_data_u;
+		uint32_t tex, uni;
+
+		*(uint32_t *)(pkt_v + o) = bo[i]->dma_addr + src_offset;
+
+		if (src_offset != 0) {
+			DRM_DEBUG("Shaders must be at offset 0 of "
+				  "the BO.\n");
+			return -EINVAL;
+		}
+
+		validated_shader = to_vc4_bo(&bo[i]->base)->validated_shader;
+		if (!validated_shader)
+			return -EINVAL;
+
+		if (validated_shader->uniforms_src_size >
+		    exec->uniforms_size) {
+			DRM_DEBUG("Uniforms src buffer overflow\n");
+			return -EINVAL;
+		}
+
+		texture_handles_u = exec->uniforms_u;
+		uniform_data_u = (texture_handles_u +
+				  validated_shader->num_texture_samples);
+
+		memcpy(exec->uniforms_v, uniform_data_u,
+		       validated_shader->uniforms_size);
+
+		for (tex = 0;
+		     tex < validated_shader->num_texture_samples;
+		     tex++) {
+			if (!reloc_tex(exec,
+				       uniform_data_u,
+				       &validated_shader->texture_samples[tex],
+				       texture_handles_u[tex],
+				       i == 2)) {
+				return -EINVAL;
+			}
+		}
+
+		/* Fill in the uniform slots that need this shader's
+		 * start-of-uniforms address (used for resetting the uniform
+		 * stream in the presence of control flow).
+		 */
+		for (uni = 0;
+		     uni < validated_shader->num_uniform_addr_offsets;
+		     uni++) {
+			uint32_t o = validated_shader->uniform_addr_offsets[uni];
+			((uint32_t *)exec->uniforms_v)[o] = exec->uniforms_p;
+		}
+
+		*(uint32_t *)(pkt_v + o + 4) = exec->uniforms_p;
+
+		exec->uniforms_u += validated_shader->uniforms_src_size;
+		exec->uniforms_v += validated_shader->uniforms_size;
+		exec->uniforms_p += validated_shader->uniforms_size;
+	}
+
+	for (i = 0; i < nr_attributes; i++) {
+		struct drm_gem_dma_object *vbo =
+			bo[ARRAY_SIZE(shader_reloc_offsets) + i];
+		uint32_t o = 36 + i * 8;
+		uint32_t offset = *(uint32_t *)(pkt_u + o + 0);
+		uint32_t attr_size = *(uint8_t *)(pkt_u + o + 4) + 1;
+		uint32_t stride = *(uint8_t *)(pkt_u + o + 5);
+		uint32_t max_index;
+
+		if (state->addr & 0x8)
+			stride |= (*(uint32_t *)(pkt_u + 100 + i * 4)) & ~0xff;
+
+		if (vbo->base.size < offset ||
+		    vbo->base.size - offset < attr_size) {
+			DRM_DEBUG("BO offset overflow (%d + %d > %zu)\n",
+				  offset, attr_size, vbo->base.size);
+			return -EINVAL;
+		}
+
+		if (stride != 0) {
+			max_index = ((vbo->base.size - offset - attr_size) /
+				     stride);
+			if (state->max_index > max_index) {
+				DRM_DEBUG("primitives use index %d out of "
+					  "supplied %d\n",
+					  state->max_index, max_index);
+				return -EINVAL;
+			}
+		}
+
+		*(uint32_t *)(pkt_v + o) = vbo->dma_addr + offset;
+	}
+
+	return 0;
+}
+
+int
+vc4_validate_shader_recs(struct drm_device *dev,
+			 struct vc4_exec_info *exec)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	uint32_t i;
+	int ret = 0;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return -ENODEV;
+
+	for (i = 0; i < exec->shader_state_count; i++) {
+		ret = validate_gl_shader_rec(dev, exec, &exec->shader_state[i]);
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}

linux-7.2.3/drivers/gpu/drm/vc4/vc4_validate_shaders.c

@@ -0,0 +1,957 @@
+/*
+ * Copyright © 2014 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/**
+ * DOC: Shader validator for VC4.
+ *
+ * Since the VC4 has no IOMMU between it and system memory, a user
+ * with access to execute shaders could escalate privilege by
+ * overwriting system memory (using the VPM write address register in
+ * the general-purpose DMA mode) or reading system memory it shouldn't
+ * (reading it as a texture, uniform data, or direct-addressed TMU
+ * lookup).
+ *
+ * The shader validator walks over a shader's BO, ensuring that its
+ * accesses are appropriately bounded, and recording where texture
+ * accesses are made so that we can do relocations for them in the
+ * uniform stream.
+ *
+ * Shader BO are immutable for their lifetimes (enforced by not
+ * allowing mmaps, GEM prime export, or rendering to from a CL), so
+ * this validation is only performed at BO creation time.
+ */
+
+#include <drm/drm_print.h>
+
+#include "vc4_drv.h"
+#include "vc4_qpu_defines.h"
+
+#define LIVE_REG_COUNT (32 + 32 + 4)
+
+struct vc4_shader_validation_state {
+	/* Current IP being validated. */
+	uint32_t ip;
+
+	/* IP at the end of the BO, do not read shader[max_ip] */
+	uint32_t max_ip;
+
+	uint64_t *shader;
+
+	struct vc4_texture_sample_info tmu_setup[2];
+	int tmu_write_count[2];
+
+	/* For registers that were last written to by a MIN instruction with
+	 * one argument being a uniform, the address of the uniform.
+	 * Otherwise, ~0.
+	 *
+	 * This is used for the validation of direct address memory reads.
+	 */
+	uint32_t live_min_clamp_offsets[LIVE_REG_COUNT];
+	bool live_max_clamp_regs[LIVE_REG_COUNT];
+	uint32_t live_immediates[LIVE_REG_COUNT];
+
+	/* Bitfield of which IPs are used as branch targets.
+	 *
+	 * Used for validation that the uniform stream is updated at the right
+	 * points and clearing the texturing/clamping state.
+	 */
+	unsigned long *branch_targets;
+
+	/* Set when entering a basic block, and cleared when the uniform
+	 * address update is found.  This is used to make sure that we don't
+	 * read uniforms when the address is undefined.
+	 */
+	bool needs_uniform_address_update;
+
+	/* Set when we find a backwards branch.  If the branch is backwards,
+	 * the taraget is probably doing an address reset to read uniforms,
+	 * and so we need to be sure that a uniforms address is present in the
+	 * stream, even if the shader didn't need to read uniforms in later
+	 * basic blocks.
+	 */
+	bool needs_uniform_address_for_loop;
+
+	/* Set when we find an instruction writing the top half of the
+	 * register files.  If we allowed writing the unusable regs in
+	 * a threaded shader, then the other shader running on our
+	 * QPU's clamp validation would be invalid.
+	 */
+	bool all_registers_used;
+};
+
+static uint32_t
+waddr_to_live_reg_index(uint32_t waddr, bool is_b)
+{
+	if (waddr < 32) {
+		if (is_b)
+			return 32 + waddr;
+		else
+			return waddr;
+	} else if (waddr <= QPU_W_ACC3) {
+		return 64 + waddr - QPU_W_ACC0;
+	} else {
+		return ~0;
+	}
+}
+
+static uint32_t
+raddr_add_a_to_live_reg_index(uint64_t inst)
+{
+	uint32_t sig = QPU_GET_FIELD(inst, QPU_SIG);
+	uint32_t add_a = QPU_GET_FIELD(inst, QPU_ADD_A);
+	uint32_t raddr_a = QPU_GET_FIELD(inst, QPU_RADDR_A);
+	uint32_t raddr_b = QPU_GET_FIELD(inst, QPU_RADDR_B);
+
+	if (add_a == QPU_MUX_A)
+		return raddr_a;
+	else if (add_a == QPU_MUX_B && sig != QPU_SIG_SMALL_IMM)
+		return 32 + raddr_b;
+	else if (add_a <= QPU_MUX_R3)
+		return 64 + add_a;
+	else
+		return ~0;
+}
+
+static bool
+live_reg_is_upper_half(uint32_t lri)
+{
+	return	(lri >= 16 && lri < 32) ||
+		(lri >= 32 + 16 && lri < 32 + 32);
+}
+
+static bool
+is_tmu_submit(uint32_t waddr)
+{
+	return (waddr == QPU_W_TMU0_S ||
+		waddr == QPU_W_TMU1_S);
+}
+
+static bool
+is_tmu_write(uint32_t waddr)
+{
+	return (waddr >= QPU_W_TMU0_S &&
+		waddr <= QPU_W_TMU1_B);
+}
+
+static bool
+record_texture_sample(struct vc4_validated_shader_info *validated_shader,
+		      struct vc4_shader_validation_state *validation_state,
+		      int tmu)
+{
+	uint32_t s = validated_shader->num_texture_samples;
+	int i;
+	struct vc4_texture_sample_info *temp_samples;
+
+	temp_samples = krealloc(validated_shader->texture_samples,
+				(s + 1) * sizeof(*temp_samples),
+				GFP_KERNEL);
+	if (!temp_samples)
+		return false;
+
+	memcpy(&temp_samples[s],
+	       &validation_state->tmu_setup[tmu],
+	       sizeof(*temp_samples));
+
+	validated_shader->num_texture_samples = s + 1;
+	validated_shader->texture_samples = temp_samples;
+
+	for (i = 0; i < 4; i++)
+		validation_state->tmu_setup[tmu].p_offset[i] = ~0;
+
+	return true;
+}
+
+static bool
+check_tmu_write(struct vc4_validated_shader_info *validated_shader,
+		struct vc4_shader_validation_state *validation_state,
+		bool is_mul)
+{
+	uint64_t inst = validation_state->shader[validation_state->ip];
+	uint32_t waddr = (is_mul ?
+			  QPU_GET_FIELD(inst, QPU_WADDR_MUL) :
+			  QPU_GET_FIELD(inst, QPU_WADDR_ADD));
+	uint32_t raddr_a = QPU_GET_FIELD(inst, QPU_RADDR_A);
+	uint32_t raddr_b = QPU_GET_FIELD(inst, QPU_RADDR_B);
+	int tmu = waddr > QPU_W_TMU0_B;
+	bool submit = is_tmu_submit(waddr);
+	bool is_direct = submit && validation_state->tmu_write_count[tmu] == 0;
+	uint32_t sig = QPU_GET_FIELD(inst, QPU_SIG);
+
+	if (is_direct) {
+		uint32_t add_b = QPU_GET_FIELD(inst, QPU_ADD_B);
+		uint32_t clamp_reg, clamp_offset;
+
+		if (sig == QPU_SIG_SMALL_IMM) {
+			DRM_DEBUG("direct TMU read used small immediate\n");
+			return false;
+		}
+
+		/* Make sure that this texture load is an add of the base
+		 * address of the UBO to a clamped offset within the UBO.
+		 */
+		if (is_mul ||
+		    QPU_GET_FIELD(inst, QPU_OP_ADD) != QPU_A_ADD) {
+			DRM_DEBUG("direct TMU load wasn't an add\n");
+			return false;
+		}
+
+		/* We assert that the clamped address is the first
+		 * argument, and the UBO base address is the second argument.
+		 * This is arbitrary, but simpler than supporting flipping the
+		 * two either way.
+		 */
+		clamp_reg = raddr_add_a_to_live_reg_index(inst);
+		if (clamp_reg == ~0) {
+			DRM_DEBUG("direct TMU load wasn't clamped\n");
+			return false;
+		}
+
+		clamp_offset = validation_state->live_min_clamp_offsets[clamp_reg];
+		if (clamp_offset == ~0) {
+			DRM_DEBUG("direct TMU load wasn't clamped\n");
+			return false;
+		}
+
+		/* Store the clamp value's offset in p1 (see reloc_tex() in
+		 * vc4_validate.c).
+		 */
+		validation_state->tmu_setup[tmu].p_offset[1] =
+			clamp_offset;
+
+		if (!(add_b == QPU_MUX_A && raddr_a == QPU_R_UNIF) &&
+		    !(add_b == QPU_MUX_B && raddr_b == QPU_R_UNIF)) {
+			DRM_DEBUG("direct TMU load didn't add to a uniform\n");
+			return false;
+		}
+
+		validation_state->tmu_setup[tmu].is_direct = true;
+	} else {
+		if (raddr_a == QPU_R_UNIF || (sig != QPU_SIG_SMALL_IMM &&
+					      raddr_b == QPU_R_UNIF)) {
+			DRM_DEBUG("uniform read in the same instruction as "
+				  "texture setup.\n");
+			return false;
+		}
+	}
+
+	if (validation_state->tmu_write_count[tmu] >= 4) {
+		DRM_DEBUG("TMU%d got too many parameters before dispatch\n",
+			  tmu);
+		return false;
+	}
+	validation_state->tmu_setup[tmu].p_offset[validation_state->tmu_write_count[tmu]] =
+		validated_shader->uniforms_size;
+	validation_state->tmu_write_count[tmu]++;
+	/* Since direct uses a RADDR uniform reference, it will get counted in
+	 * check_instruction_reads()
+	 */
+	if (!is_direct) {
+		if (validation_state->needs_uniform_address_update) {
+			DRM_DEBUG("Texturing with undefined uniform address\n");
+			return false;
+		}
+
+		validated_shader->uniforms_size += 4;
+	}
+
+	if (submit) {
+		if (!record_texture_sample(validated_shader,
+					   validation_state, tmu)) {
+			return false;
+		}
+
+		validation_state->tmu_write_count[tmu] = 0;
+	}
+
+	return true;
+}
+
+static bool require_uniform_address_uniform(struct vc4_validated_shader_info *validated_shader)
+{
+	uint32_t o = validated_shader->num_uniform_addr_offsets;
+	uint32_t num_uniforms = validated_shader->uniforms_size / 4;
+	u32 *offsets;
+
+	offsets = krealloc_array(validated_shader->uniform_addr_offsets,
+				 o + 1,
+				 sizeof(*validated_shader->uniform_addr_offsets),
+				 GFP_KERNEL);
+	if (!offsets)
+		return false;
+
+	validated_shader->uniform_addr_offsets = offsets;
+	validated_shader->uniform_addr_offsets[o] = num_uniforms;
+	validated_shader->num_uniform_addr_offsets++;
+
+	return true;
+}
+
+static bool
+validate_uniform_address_write(struct vc4_validated_shader_info *validated_shader,
+			       struct vc4_shader_validation_state *validation_state,
+			       bool is_mul)
+{
+	uint64_t inst = validation_state->shader[validation_state->ip];
+	u32 add_b = QPU_GET_FIELD(inst, QPU_ADD_B);
+	u32 raddr_a = QPU_GET_FIELD(inst, QPU_RADDR_A);
+	u32 raddr_b = QPU_GET_FIELD(inst, QPU_RADDR_B);
+	u32 add_lri = raddr_add_a_to_live_reg_index(inst);
+	/* We want our reset to be pointing at whatever uniform follows the
+	 * uniforms base address.
+	 */
+	u32 expected_offset = validated_shader->uniforms_size + 4;
+
+	/* We only support absolute uniform address changes, and we
+	 * require that they be in the current basic block before any
+	 * of its uniform reads.
+	 *
+	 * One could potentially emit more efficient QPU code, by
+	 * noticing that (say) an if statement does uniform control
+	 * flow for all threads and that the if reads the same number
+	 * of uniforms on each side.  However, this scheme is easy to
+	 * validate so it's all we allow for now.
+	 */
+	switch (QPU_GET_FIELD(inst, QPU_SIG)) {
+	case QPU_SIG_NONE:
+	case QPU_SIG_SCOREBOARD_UNLOCK:
+	case QPU_SIG_COLOR_LOAD:
+	case QPU_SIG_LOAD_TMU0:
+	case QPU_SIG_LOAD_TMU1:
+		break;
+	default:
+		DRM_DEBUG("uniforms address change must be "
+			  "normal math\n");
+		return false;
+	}
+
+	if (is_mul || QPU_GET_FIELD(inst, QPU_OP_ADD) != QPU_A_ADD) {
+		DRM_DEBUG("Uniform address reset must be an ADD.\n");
+		return false;
+	}
+
+	if (QPU_GET_FIELD(inst, QPU_COND_ADD) != QPU_COND_ALWAYS) {
+		DRM_DEBUG("Uniform address reset must be unconditional.\n");
+		return false;
+	}
+
+	if (QPU_GET_FIELD(inst, QPU_PACK) != QPU_PACK_A_NOP &&
+	    !(inst & QPU_PM)) {
+		DRM_DEBUG("No packing allowed on uniforms reset\n");
+		return false;
+	}
+
+	if (add_lri == -1) {
+		DRM_DEBUG("First argument of uniform address write must be "
+			  "an immediate value.\n");
+		return false;
+	}
+
+	if (validation_state->live_immediates[add_lri] != expected_offset) {
+		DRM_DEBUG("Resetting uniforms with offset %db instead of %db\n",
+			  validation_state->live_immediates[add_lri],
+			  expected_offset);
+		return false;
+	}
+
+	if (!(add_b == QPU_MUX_A && raddr_a == QPU_R_UNIF) &&
+	    !(add_b == QPU_MUX_B && raddr_b == QPU_R_UNIF)) {
+		DRM_DEBUG("Second argument of uniform address write must be "
+			  "a uniform.\n");
+		return false;
+	}
+
+	validation_state->needs_uniform_address_update = false;
+	validation_state->needs_uniform_address_for_loop = false;
+	return require_uniform_address_uniform(validated_shader);
+}
+
+static bool
+check_reg_write(struct vc4_validated_shader_info *validated_shader,
+		struct vc4_shader_validation_state *validation_state,
+		bool is_mul)
+{
+	uint64_t inst = validation_state->shader[validation_state->ip];
+	uint32_t waddr = (is_mul ?
+			  QPU_GET_FIELD(inst, QPU_WADDR_MUL) :
+			  QPU_GET_FIELD(inst, QPU_WADDR_ADD));
+	uint32_t sig = QPU_GET_FIELD(inst, QPU_SIG);
+	bool ws = inst & QPU_WS;
+	bool is_b = is_mul ^ ws;
+	u32 lri = waddr_to_live_reg_index(waddr, is_b);
+
+	if (lri != -1) {
+		uint32_t cond_add = QPU_GET_FIELD(inst, QPU_COND_ADD);
+		uint32_t cond_mul = QPU_GET_FIELD(inst, QPU_COND_MUL);
+
+		if (sig == QPU_SIG_LOAD_IMM &&
+		    QPU_GET_FIELD(inst, QPU_PACK) == QPU_PACK_A_NOP &&
+		    ((is_mul && cond_mul == QPU_COND_ALWAYS) ||
+		     (!is_mul && cond_add == QPU_COND_ALWAYS))) {
+			validation_state->live_immediates[lri] =
+				QPU_GET_FIELD(inst, QPU_LOAD_IMM);
+		} else {
+			validation_state->live_immediates[lri] = ~0;
+		}
+
+		if (live_reg_is_upper_half(lri))
+			validation_state->all_registers_used = true;
+	}
+
+	switch (waddr) {
+	case QPU_W_UNIFORMS_ADDRESS:
+		if (is_b) {
+			DRM_DEBUG("relative uniforms address change "
+				  "unsupported\n");
+			return false;
+		}
+
+		return validate_uniform_address_write(validated_shader,
+						      validation_state,
+						      is_mul);
+
+	case QPU_W_TLB_COLOR_MS:
+	case QPU_W_TLB_COLOR_ALL:
+	case QPU_W_TLB_Z:
+		/* These only interact with the tile buffer, not main memory,
+		 * so they're safe.
+		 */
+		return true;
+
+	case QPU_W_TMU0_S:
+	case QPU_W_TMU0_T:
+	case QPU_W_TMU0_R:
+	case QPU_W_TMU0_B:
+	case QPU_W_TMU1_S:
+	case QPU_W_TMU1_T:
+	case QPU_W_TMU1_R:
+	case QPU_W_TMU1_B:
+		return check_tmu_write(validated_shader, validation_state,
+				       is_mul);
+
+	case QPU_W_HOST_INT:
+	case QPU_W_TMU_NOSWAP:
+	case QPU_W_TLB_ALPHA_MASK:
+	case QPU_W_MUTEX_RELEASE:
+		/* XXX: I haven't thought about these, so don't support them
+		 * for now.
+		 */
+		DRM_DEBUG("Unsupported waddr %d\n", waddr);
+		return false;
+
+	case QPU_W_VPM_ADDR:
+		DRM_DEBUG("General VPM DMA unsupported\n");
+		return false;
+
+	case QPU_W_VPM:
+	case QPU_W_VPMVCD_SETUP:
+		/* We allow VPM setup in general, even including VPM DMA
+		 * configuration setup, because the (unsafe) DMA can only be
+		 * triggered by QPU_W_VPM_ADDR writes.
+		 */
+		return true;
+
+	case QPU_W_TLB_STENCIL_SETUP:
+		return true;
+	}
+
+	return true;
+}
+
+static void
+track_live_clamps(struct vc4_validated_shader_info *validated_shader,
+		  struct vc4_shader_validation_state *validation_state)
+{
+	uint64_t inst = validation_state->shader[validation_state->ip];
+	uint32_t op_add = QPU_GET_FIELD(inst, QPU_OP_ADD);
+	uint32_t waddr_add = QPU_GET_FIELD(inst, QPU_WADDR_ADD);
+	uint32_t waddr_mul = QPU_GET_FIELD(inst, QPU_WADDR_MUL);
+	uint32_t cond_add = QPU_GET_FIELD(inst, QPU_COND_ADD);
+	uint32_t add_a = QPU_GET_FIELD(inst, QPU_ADD_A);
+	uint32_t add_b = QPU_GET_FIELD(inst, QPU_ADD_B);
+	uint32_t raddr_a = QPU_GET_FIELD(inst, QPU_RADDR_A);
+	uint32_t raddr_b = QPU_GET_FIELD(inst, QPU_RADDR_B);
+	uint32_t sig = QPU_GET_FIELD(inst, QPU_SIG);
+	bool ws = inst & QPU_WS;
+	uint32_t lri_add_a, lri_add, lri_mul;
+	bool add_a_is_min_0;
+
+	/* Check whether OP_ADD's A argumennt comes from a live MAX(x, 0),
+	 * before we clear previous live state.
+	 */
+	lri_add_a = raddr_add_a_to_live_reg_index(inst);
+	add_a_is_min_0 = (lri_add_a != ~0 &&
+			  validation_state->live_max_clamp_regs[lri_add_a]);
+
+	/* Clear live state for registers written by our instruction. */
+	lri_add = waddr_to_live_reg_index(waddr_add, ws);
+	lri_mul = waddr_to_live_reg_index(waddr_mul, !ws);
+	if (lri_mul != ~0) {
+		validation_state->live_max_clamp_regs[lri_mul] = false;
+		validation_state->live_min_clamp_offsets[lri_mul] = ~0;
+	}
+	if (lri_add != ~0) {
+		validation_state->live_max_clamp_regs[lri_add] = false;
+		validation_state->live_min_clamp_offsets[lri_add] = ~0;
+	} else {
+		/* Nothing further to do for live tracking, since only ADDs
+		 * generate new live clamp registers.
+		 */
+		return;
+	}
+
+	/* Now, handle remaining live clamp tracking for the ADD operation. */
+
+	if (cond_add != QPU_COND_ALWAYS)
+		return;
+
+	if (op_add == QPU_A_MAX) {
+		/* Track live clamps of a value to a minimum of 0 (in either
+		 * arg).
+		 */
+		if (sig != QPU_SIG_SMALL_IMM || raddr_b != 0 ||
+		    (add_a != QPU_MUX_B && add_b != QPU_MUX_B)) {
+			return;
+		}
+
+		validation_state->live_max_clamp_regs[lri_add] = true;
+	} else if (op_add == QPU_A_MIN) {
+		/* Track live clamps of a value clamped to a minimum of 0 and
+		 * a maximum of some uniform's offset.
+		 */
+		if (!add_a_is_min_0)
+			return;
+
+		if (!(add_b == QPU_MUX_A && raddr_a == QPU_R_UNIF) &&
+		    !(add_b == QPU_MUX_B && raddr_b == QPU_R_UNIF &&
+		      sig != QPU_SIG_SMALL_IMM)) {
+			return;
+		}
+
+		validation_state->live_min_clamp_offsets[lri_add] =
+			validated_shader->uniforms_size;
+	}
+}
+
+static bool
+check_instruction_writes(struct vc4_validated_shader_info *validated_shader,
+			 struct vc4_shader_validation_state *validation_state)
+{
+	uint64_t inst = validation_state->shader[validation_state->ip];
+	uint32_t waddr_add = QPU_GET_FIELD(inst, QPU_WADDR_ADD);
+	uint32_t waddr_mul = QPU_GET_FIELD(inst, QPU_WADDR_MUL);
+	bool ok;
+
+	if (is_tmu_write(waddr_add) && is_tmu_write(waddr_mul)) {
+		DRM_DEBUG("ADD and MUL both set up textures\n");
+		return false;
+	}
+
+	ok = (check_reg_write(validated_shader, validation_state, false) &&
+	      check_reg_write(validated_shader, validation_state, true));
+
+	track_live_clamps(validated_shader, validation_state);
+
+	return ok;
+}
+
+static bool
+check_branch(uint64_t inst,
+	     struct vc4_validated_shader_info *validated_shader,
+	     struct vc4_shader_validation_state *validation_state,
+	     int ip)
+{
+	int32_t branch_imm = QPU_GET_FIELD(inst, QPU_BRANCH_TARGET);
+	uint32_t waddr_add = QPU_GET_FIELD(inst, QPU_WADDR_ADD);
+	uint32_t waddr_mul = QPU_GET_FIELD(inst, QPU_WADDR_MUL);
+
+	if ((int)branch_imm < 0)
+		validation_state->needs_uniform_address_for_loop = true;
+
+	/* We don't want to have to worry about validation of this, and
+	 * there's no need for it.
+	 */
+	if (waddr_add != QPU_W_NOP || waddr_mul != QPU_W_NOP) {
+		DRM_DEBUG("branch instruction at %d wrote a register.\n",
+			  validation_state->ip);
+		return false;
+	}
+
+	return true;
+}
+
+static bool
+check_instruction_reads(struct vc4_validated_shader_info *validated_shader,
+			struct vc4_shader_validation_state *validation_state)
+{
+	uint64_t inst = validation_state->shader[validation_state->ip];
+	uint32_t raddr_a = QPU_GET_FIELD(inst, QPU_RADDR_A);
+	uint32_t raddr_b = QPU_GET_FIELD(inst, QPU_RADDR_B);
+	uint32_t sig = QPU_GET_FIELD(inst, QPU_SIG);
+
+	if (raddr_a == QPU_R_UNIF ||
+	    (raddr_b == QPU_R_UNIF && sig != QPU_SIG_SMALL_IMM)) {
+		/* This can't overflow the uint32_t, because we're reading 8
+		 * bytes of instruction to increment by 4 here, so we'd
+		 * already be OOM.
+		 */
+		validated_shader->uniforms_size += 4;
+
+		if (validation_state->needs_uniform_address_update) {
+			DRM_DEBUG("Uniform read with undefined uniform "
+				  "address\n");
+			return false;
+		}
+	}
+
+	if ((raddr_a >= 16 && raddr_a < 32) ||
+	    (raddr_b >= 16 && raddr_b < 32 && sig != QPU_SIG_SMALL_IMM)) {
+		validation_state->all_registers_used = true;
+	}
+
+	return true;
+}
+
+/* Make sure that all branches are absolute and point within the shader, and
+ * note their targets for later.
+ */
+static bool
+vc4_validate_branches(struct vc4_shader_validation_state *validation_state)
+{
+	uint32_t max_branch_target = 0;
+	int ip;
+	int last_branch = -2;
+
+	for (ip = 0; ip < validation_state->max_ip; ip++) {
+		uint64_t inst = validation_state->shader[ip];
+		int32_t branch_imm = QPU_GET_FIELD(inst, QPU_BRANCH_TARGET);
+		uint32_t sig = QPU_GET_FIELD(inst, QPU_SIG);
+		uint32_t after_delay_ip = ip + 4;
+		uint32_t branch_target_ip;
+
+		if (sig == QPU_SIG_PROG_END) {
+			/* There are two delay slots after program end is
+			 * signaled that are still executed, then we're
+			 * finished.  validation_state->max_ip is the
+			 * instruction after the last valid instruction in the
+			 * program.
+			 */
+			validation_state->max_ip = ip + 3;
+			continue;
+		}
+
+		if (sig != QPU_SIG_BRANCH)
+			continue;
+
+		if (ip - last_branch < 4) {
+			DRM_DEBUG("Branch at %d during delay slots\n", ip);
+			return false;
+		}
+		last_branch = ip;
+
+		if (inst & QPU_BRANCH_REG) {
+			DRM_DEBUG("branching from register relative "
+				  "not supported\n");
+			return false;
+		}
+
+		if (!(inst & QPU_BRANCH_REL)) {
+			DRM_DEBUG("relative branching required\n");
+			return false;
+		}
+
+		/* The actual branch target is the instruction after the delay
+		 * slots, plus whatever byte offset is in the low 32 bits of
+		 * the instruction.  Make sure we're not branching beyond the
+		 * end of the shader object.
+		 */
+		if (branch_imm % sizeof(inst) != 0) {
+			DRM_DEBUG("branch target not aligned\n");
+			return false;
+		}
+
+		branch_target_ip = after_delay_ip + (branch_imm >> 3);
+		if (branch_target_ip >= validation_state->max_ip) {
+			DRM_DEBUG("Branch at %d outside of shader (ip %d/%d)\n",
+				  ip, branch_target_ip,
+				  validation_state->max_ip);
+			return false;
+		}
+		set_bit(branch_target_ip, validation_state->branch_targets);
+
+		/* Make sure that the non-branching path is also not outside
+		 * the shader.
+		 */
+		if (after_delay_ip >= validation_state->max_ip) {
+			DRM_DEBUG("Branch at %d continues past shader end "
+				  "(%d/%d)\n",
+				  ip, after_delay_ip, validation_state->max_ip);
+			return false;
+		}
+		set_bit(after_delay_ip, validation_state->branch_targets);
+		max_branch_target = max(max_branch_target, after_delay_ip);
+	}
+
+	if (max_branch_target > validation_state->max_ip - 3) {
+		DRM_DEBUG("Branch landed after QPU_SIG_PROG_END");
+		return false;
+	}
+
+	return true;
+}
+
+/* Resets any known state for the shader, used when we may be branched to from
+ * multiple locations in the program (or at shader start).
+ */
+static void
+reset_validation_state(struct vc4_shader_validation_state *validation_state)
+{
+	int i;
+
+	for (i = 0; i < 8; i++)
+		validation_state->tmu_setup[i / 4].p_offset[i % 4] = ~0;
+
+	for (i = 0; i < LIVE_REG_COUNT; i++) {
+		validation_state->live_min_clamp_offsets[i] = ~0;
+		validation_state->live_max_clamp_regs[i] = false;
+		validation_state->live_immediates[i] = ~0;
+	}
+}
+
+static bool
+texturing_in_progress(struct vc4_shader_validation_state *validation_state)
+{
+	return (validation_state->tmu_write_count[0] != 0 ||
+		validation_state->tmu_write_count[1] != 0);
+}
+
+static bool
+vc4_handle_branch_target(struct vc4_shader_validation_state *validation_state)
+{
+	uint32_t ip = validation_state->ip;
+
+	if (!test_bit(ip, validation_state->branch_targets))
+		return true;
+
+	if (texturing_in_progress(validation_state)) {
+		DRM_DEBUG("Branch target landed during TMU setup\n");
+		return false;
+	}
+
+	/* Reset our live values tracking, since this instruction may have
+	 * multiple predecessors.
+	 *
+	 * One could potentially do analysis to determine that, for
+	 * example, all predecessors have a live max clamp in the same
+	 * register, but we don't bother with that.
+	 */
+	reset_validation_state(validation_state);
+
+	/* Since we've entered a basic block from potentially multiple
+	 * predecessors, we need the uniforms address to be updated before any
+	 * unforms are read.  We require that after any branch point, the next
+	 * uniform to be loaded is a uniform address offset.  That uniform's
+	 * offset will be marked by the uniform address register write
+	 * validation, or a one-off the end-of-program check.
+	 */
+	validation_state->needs_uniform_address_update = true;
+
+	return true;
+}
+
+struct vc4_validated_shader_info *
+vc4_validate_shader(struct drm_gem_dma_object *shader_obj)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(shader_obj->base.dev);
+	bool found_shader_end = false;
+	int shader_end_ip = 0;
+	uint32_t last_thread_switch_ip = -3;
+	uint32_t ip;
+	struct vc4_validated_shader_info *validated_shader = NULL;
+	struct vc4_shader_validation_state validation_state;
+
+	if (WARN_ON_ONCE(vc4->gen > VC4_GEN_4))
+		return NULL;
+
+	memset(&validation_state, 0, sizeof(validation_state));
+	validation_state.shader = shader_obj->vaddr;
+	validation_state.max_ip = shader_obj->base.size / sizeof(uint64_t);
+
+	reset_validation_state(&validation_state);
+
+	validation_state.branch_targets =
+		kcalloc(BITS_TO_LONGS(validation_state.max_ip),
+			sizeof(unsigned long), GFP_KERNEL);
+	if (!validation_state.branch_targets)
+		goto fail;
+
+	validated_shader = kzalloc_objs(*validated_shader, 1);
+	if (!validated_shader)
+		goto fail;
+
+	if (!vc4_validate_branches(&validation_state))
+		goto fail;
+
+	for (ip = 0; ip < validation_state.max_ip; ip++) {
+		uint64_t inst = validation_state.shader[ip];
+		uint32_t sig = QPU_GET_FIELD(inst, QPU_SIG);
+
+		validation_state.ip = ip;
+
+		if (!vc4_handle_branch_target(&validation_state))
+			goto fail;
+
+		if (ip == last_thread_switch_ip + 3) {
+			/* Reset r0-r3 live clamp data */
+			int i;
+
+			for (i = 64; i < LIVE_REG_COUNT; i++) {
+				validation_state.live_min_clamp_offsets[i] = ~0;
+				validation_state.live_max_clamp_regs[i] = false;
+				validation_state.live_immediates[i] = ~0;
+			}
+		}
+
+		switch (sig) {
+		case QPU_SIG_NONE:
+		case QPU_SIG_WAIT_FOR_SCOREBOARD:
+		case QPU_SIG_SCOREBOARD_UNLOCK:
+		case QPU_SIG_COLOR_LOAD:
+		case QPU_SIG_LOAD_TMU0:
+		case QPU_SIG_LOAD_TMU1:
+		case QPU_SIG_PROG_END:
+		case QPU_SIG_SMALL_IMM:
+		case QPU_SIG_THREAD_SWITCH:
+		case QPU_SIG_LAST_THREAD_SWITCH:
+			if (!check_instruction_writes(validated_shader,
+						      &validation_state)) {
+				DRM_DEBUG("Bad write at ip %d\n", ip);
+				goto fail;
+			}
+
+			if (!check_instruction_reads(validated_shader,
+						     &validation_state))
+				goto fail;
+
+			if (sig == QPU_SIG_PROG_END) {
+				found_shader_end = true;
+				shader_end_ip = ip;
+			}
+
+			if (sig == QPU_SIG_THREAD_SWITCH ||
+			    sig == QPU_SIG_LAST_THREAD_SWITCH) {
+				validated_shader->is_threaded = true;
+
+				if (ip < last_thread_switch_ip + 3) {
+					DRM_DEBUG("Thread switch too soon after "
+						  "last switch at ip %d\n", ip);
+					goto fail;
+				}
+				last_thread_switch_ip = ip;
+			}
+
+			break;
+
+		case QPU_SIG_LOAD_IMM:
+			if (!check_instruction_writes(validated_shader,
+						      &validation_state)) {
+				DRM_DEBUG("Bad LOAD_IMM write at ip %d\n", ip);
+				goto fail;
+			}
+			break;
+
+		case QPU_SIG_BRANCH:
+			if (!check_branch(inst, validated_shader,
+					  &validation_state, ip))
+				goto fail;
+
+			if (ip < last_thread_switch_ip + 3) {
+				DRM_DEBUG("Branch in thread switch at ip %d",
+					  ip);
+				goto fail;
+			}
+
+			break;
+		default:
+			DRM_DEBUG("Unsupported QPU signal %d at "
+				  "instruction %d\n", sig, ip);
+			goto fail;
+		}
+
+		/* There are two delay slots after program end is signaled
+		 * that are still executed, then we're finished.
+		 */
+		if (found_shader_end && ip == shader_end_ip + 2)
+			break;
+	}
+
+	if (ip == validation_state.max_ip) {
+		DRM_DEBUG("shader failed to terminate before "
+			  "shader BO end at %zd\n",
+			  shader_obj->base.size);
+		goto fail;
+	}
+
+	/* Might corrupt other thread */
+	if (validated_shader->is_threaded &&
+	    validation_state.all_registers_used) {
+		DRM_DEBUG("Shader uses threading, but uses the upper "
+			  "half of the registers, too\n");
+		goto fail;
+	}
+
+	/* If we did a backwards branch and we haven't emitted a uniforms
+	 * reset since then, we still need the uniforms stream to have the
+	 * uniforms address available so that the backwards branch can do its
+	 * uniforms reset.
+	 *
+	 * We could potentially prove that the backwards branch doesn't
+	 * contain any uses of uniforms until program exit, but that doesn't
+	 * seem to be worth the trouble.
+	 */
+	if (validation_state.needs_uniform_address_for_loop) {
+		if (!require_uniform_address_uniform(validated_shader))
+			goto fail;
+		validated_shader->uniforms_size += 4;
+	}
+
+	/* Again, no chance of integer overflow here because the worst case
+	 * scenario is 8 bytes of uniforms plus handles per 8-byte
+	 * instruction.
+	 */
+	validated_shader->uniforms_src_size =
+		(validated_shader->uniforms_size +
+		 4 * validated_shader->num_texture_samples);
+
+	kfree(validation_state.branch_targets);
+
+	return validated_shader;
+
+fail:
+	kfree(validation_state.branch_targets);
+	if (validated_shader) {
+		kfree(validated_shader->uniform_addr_offsets);
+		kfree(validated_shader->texture_samples);
+		kfree(validated_shader);
+	}
+	return NULL;
+}

linux-7.2.3/drivers/gpu/drm/vc4/vc4_vec.c

@@ -0,0 +1,849 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2016 Broadcom
+ */
+
+/**
+ * DOC: VC4 SDTV module
+ *
+ * The VEC encoder generates PAL or NTSC composite video output.
+ *
+ * TV mode selection is done by an atomic property on the encoder,
+ * because a drm_mode_modeinfo is insufficient to distinguish between
+ * PAL and PAL-M or NTSC and NTSC-J.
+ */
+
+#include <drm/drm_atomic_helper.h>
+#include <drm/drm_drv.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_panel.h>
+#include <drm/drm_print.h>
+#include <drm/drm_probe_helper.h>
+#include <drm/drm_simple_kms_helper.h>
+#include <linux/clk.h>
+#include <linux/component.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+
+#include "vc4_drv.h"
+#include "vc4_regs.h"
+
+/* WSE Registers */
+#define VEC_WSE_RESET			0xc0
+
+#define VEC_WSE_CONTROL			0xc4
+#define VEC_WSE_WSS_ENABLE		BIT(7)
+
+#define VEC_WSE_WSS_DATA		0xc8
+#define VEC_WSE_VPS_DATA1		0xcc
+#define VEC_WSE_VPS_CONTROL		0xd0
+
+/* VEC Registers */
+#define VEC_REVID			0x100
+
+#define VEC_CONFIG0			0x104
+#define VEC_CONFIG0_YDEL_MASK		GENMASK(28, 26)
+#define VEC_CONFIG0_YDEL(x)		((x) << 26)
+#define VEC_CONFIG0_CDEL_MASK		GENMASK(25, 24)
+#define VEC_CONFIG0_CDEL(x)		((x) << 24)
+#define VEC_CONFIG0_SECAM_STD		BIT(21)
+#define VEC_CONFIG0_PBPR_FIL		BIT(18)
+#define VEC_CONFIG0_CHROMA_GAIN_MASK	GENMASK(17, 16)
+#define VEC_CONFIG0_CHROMA_GAIN_UNITY	(0 << 16)
+#define VEC_CONFIG0_CHROMA_GAIN_1_32	(1 << 16)
+#define VEC_CONFIG0_CHROMA_GAIN_1_16	(2 << 16)
+#define VEC_CONFIG0_CHROMA_GAIN_1_8	(3 << 16)
+#define VEC_CONFIG0_CBURST_GAIN_MASK	GENMASK(14, 13)
+#define VEC_CONFIG0_CBURST_GAIN_UNITY	(0 << 13)
+#define VEC_CONFIG0_CBURST_GAIN_1_128	(1 << 13)
+#define VEC_CONFIG0_CBURST_GAIN_1_64	(2 << 13)
+#define VEC_CONFIG0_CBURST_GAIN_1_32	(3 << 13)
+#define VEC_CONFIG0_CHRBW1		BIT(11)
+#define VEC_CONFIG0_CHRBW0		BIT(10)
+#define VEC_CONFIG0_SYNCDIS		BIT(9)
+#define VEC_CONFIG0_BURDIS		BIT(8)
+#define VEC_CONFIG0_CHRDIS		BIT(7)
+#define VEC_CONFIG0_PDEN		BIT(6)
+#define VEC_CONFIG0_YCDELAY		BIT(4)
+#define VEC_CONFIG0_RAMPEN		BIT(2)
+#define VEC_CONFIG0_YCDIS		BIT(2)
+#define VEC_CONFIG0_STD_MASK		GENMASK(1, 0)
+#define VEC_CONFIG0_NTSC_STD		0
+#define VEC_CONFIG0_PAL_BDGHI_STD	1
+#define VEC_CONFIG0_PAL_M_STD		2
+#define VEC_CONFIG0_PAL_N_STD		3
+
+#define VEC_SCHPH			0x108
+#define VEC_SOFT_RESET			0x10c
+#define VEC_CLMP0_START			0x144
+#define VEC_CLMP0_END			0x148
+
+/*
+ * These set the color subcarrier frequency
+ * if VEC_CONFIG1_CUSTOM_FREQ is enabled.
+ *
+ * VEC_FREQ1_0 contains the most significant 16-bit half-word,
+ * VEC_FREQ3_2 contains the least significant 16-bit half-word.
+ * 0x80000000 seems to be equivalent to the pixel clock
+ * (which itself is the VEC clock divided by 8).
+ *
+ * Reference values (with the default pixel clock of 13.5 MHz):
+ *
+ * NTSC  (3579545.[45] Hz)     - 0x21F07C1F
+ * PAL   (4433618.75 Hz)       - 0x2A098ACB
+ * PAL-M (3575611.[888111] Hz) - 0x21E6EFE3
+ * PAL-N (3582056.25 Hz)       - 0x21F69446
+ *
+ * NOTE: For SECAM, it is used as the Dr center frequency,
+ * regardless of whether VEC_CONFIG1_CUSTOM_FREQ is enabled or not;
+ * that is specified as 4406250 Hz, which corresponds to 0x29C71C72.
+ */
+#define VEC_FREQ3_2			0x180
+#define VEC_FREQ1_0			0x184
+
+#define VEC_CONFIG1			0x188
+#define VEC_CONFIG_VEC_RESYNC_OFF	BIT(18)
+#define VEC_CONFIG_RGB219		BIT(17)
+#define VEC_CONFIG_CBAR_EN		BIT(16)
+#define VEC_CONFIG_TC_OBB		BIT(15)
+#define VEC_CONFIG1_OUTPUT_MODE_MASK	GENMASK(12, 10)
+#define VEC_CONFIG1_C_Y_CVBS		(0 << 10)
+#define VEC_CONFIG1_CVBS_Y_C		(1 << 10)
+#define VEC_CONFIG1_PR_Y_PB		(2 << 10)
+#define VEC_CONFIG1_RGB			(4 << 10)
+#define VEC_CONFIG1_Y_C_CVBS		(5 << 10)
+#define VEC_CONFIG1_C_CVBS_Y		(6 << 10)
+#define VEC_CONFIG1_C_CVBS_CVBS		(7 << 10)
+#define VEC_CONFIG1_DIS_CHR		BIT(9)
+#define VEC_CONFIG1_DIS_LUMA		BIT(8)
+#define VEC_CONFIG1_YCBCR_IN		BIT(6)
+#define VEC_CONFIG1_DITHER_TYPE_LFSR	0
+#define VEC_CONFIG1_DITHER_TYPE_COUNTER	BIT(5)
+#define VEC_CONFIG1_DITHER_EN		BIT(4)
+#define VEC_CONFIG1_CYDELAY		BIT(3)
+#define VEC_CONFIG1_LUMADIS		BIT(2)
+#define VEC_CONFIG1_COMPDIS		BIT(1)
+#define VEC_CONFIG1_CUSTOM_FREQ		BIT(0)
+
+#define VEC_CONFIG2			0x18c
+#define VEC_CONFIG2_PROG_SCAN		BIT(15)
+#define VEC_CONFIG2_SYNC_ADJ_MASK	GENMASK(14, 12)
+#define VEC_CONFIG2_SYNC_ADJ(x)		(((x) / 2) << 12)
+#define VEC_CONFIG2_PBPR_EN		BIT(10)
+#define VEC_CONFIG2_UV_DIG_DIS		BIT(6)
+#define VEC_CONFIG2_RGB_DIG_DIS		BIT(5)
+#define VEC_CONFIG2_TMUX_MASK		GENMASK(3, 2)
+#define VEC_CONFIG2_TMUX_DRIVE0		(0 << 2)
+#define VEC_CONFIG2_TMUX_RG_COMP	(1 << 2)
+#define VEC_CONFIG2_TMUX_UV_YC		(2 << 2)
+#define VEC_CONFIG2_TMUX_SYNC_YC	(3 << 2)
+
+#define VEC_INTERRUPT_CONTROL		0x190
+#define VEC_INTERRUPT_STATUS		0x194
+
+/*
+ * Db center frequency for SECAM; the clock for this is the same as for
+ * VEC_FREQ3_2/VEC_FREQ1_0, which is used for Dr center frequency.
+ *
+ * This is specified as 4250000 Hz, which corresponds to 0x284BDA13.
+ * That is also the default value, so no need to set it explicitly.
+ */
+#define VEC_FCW_SECAM_B			0x198
+#define VEC_SECAM_GAIN_VAL		0x19c
+
+#define VEC_CONFIG3			0x1a0
+#define VEC_CONFIG3_HORIZ_LEN_STD	(0 << 0)
+#define VEC_CONFIG3_HORIZ_LEN_MPEG1_SIF	(1 << 0)
+#define VEC_CONFIG3_SHAPE_NON_LINEAR	BIT(1)
+
+#define VEC_STATUS0			0x200
+#define VEC_MASK0			0x204
+
+#define VEC_CFG				0x208
+#define VEC_CFG_SG_MODE_MASK		GENMASK(6, 5)
+#define VEC_CFG_SG_MODE(x)		((x) << 5)
+#define VEC_CFG_SG_EN			BIT(4)
+#define VEC_CFG_VEC_EN			BIT(3)
+#define VEC_CFG_MB_EN			BIT(2)
+#define VEC_CFG_ENABLE			BIT(1)
+#define VEC_CFG_TB_EN			BIT(0)
+
+#define VEC_DAC_TEST			0x20c
+
+#define VEC_DAC_CONFIG			0x210
+#define VEC_DAC_CONFIG_LDO_BIAS_CTRL(x)	((x) << 24)
+#define VEC_DAC_CONFIG_DRIVER_CTRL(x)	((x) << 16)
+#define VEC_DAC_CONFIG_DAC_CTRL(x)	(x)
+
+#define VEC_DAC_MISC			0x214
+#define VEC_DAC_MISC_VCD_CTRL_MASK	GENMASK(31, 16)
+#define VEC_DAC_MISC_VCD_CTRL(x)	((x) << 16)
+#define VEC_DAC_MISC_VID_ACT		BIT(8)
+#define VEC_DAC_MISC_VCD_PWRDN		BIT(6)
+#define VEC_DAC_MISC_BIAS_PWRDN		BIT(5)
+#define VEC_DAC_MISC_DAC_PWRDN		BIT(2)
+#define VEC_DAC_MISC_LDO_PWRDN		BIT(1)
+#define VEC_DAC_MISC_DAC_RST_N		BIT(0)
+
+
+struct vc4_vec_variant {
+	u32 dac_config;
+};
+
+/* General VEC hardware state. */
+struct vc4_vec {
+	struct vc4_encoder encoder;
+	struct drm_connector connector;
+
+	struct platform_device *pdev;
+	const struct vc4_vec_variant *variant;
+
+	void __iomem *regs;
+
+	struct clk *clock;
+
+	struct drm_property *legacy_tv_mode_property;
+
+	struct debugfs_regset32 regset;
+};
+
+#define VEC_READ(offset)								\
+	({										\
+		kunit_fail_current_test("Accessing a register in a unit test!\n");	\
+		readl(vec->regs + (offset));						\
+	})
+
+#define VEC_WRITE(offset, val)								\
+	do {										\
+		kunit_fail_current_test("Accessing a register in a unit test!\n");	\
+		writel(val, vec->regs + (offset));					\
+	} while (0)
+
+#define encoder_to_vc4_vec(_encoder)					\
+	container_of_const(_encoder, struct vc4_vec, encoder.base)
+
+#define connector_to_vc4_vec(_connector)				\
+	container_of_const(_connector, struct vc4_vec, connector)
+
+enum vc4_vec_tv_mode_id {
+	VC4_VEC_TV_MODE_NTSC,
+	VC4_VEC_TV_MODE_NTSC_J,
+	VC4_VEC_TV_MODE_PAL,
+	VC4_VEC_TV_MODE_PAL_M,
+	VC4_VEC_TV_MODE_NTSC_443,
+	VC4_VEC_TV_MODE_PAL_60,
+	VC4_VEC_TV_MODE_PAL_N,
+	VC4_VEC_TV_MODE_SECAM,
+	VC4_VEC_TV_MODE_MONOCHROME,
+};
+
+struct vc4_vec_tv_mode {
+	unsigned int mode;
+	u16 expected_htotal;
+	u32 config0;
+	u32 config1;
+	u32 custom_freq;
+};
+
+static const struct debugfs_reg32 vec_regs[] = {
+	VC4_REG32(VEC_WSE_CONTROL),
+	VC4_REG32(VEC_WSE_WSS_DATA),
+	VC4_REG32(VEC_WSE_VPS_DATA1),
+	VC4_REG32(VEC_WSE_VPS_CONTROL),
+	VC4_REG32(VEC_REVID),
+	VC4_REG32(VEC_CONFIG0),
+	VC4_REG32(VEC_SCHPH),
+	VC4_REG32(VEC_CLMP0_START),
+	VC4_REG32(VEC_CLMP0_END),
+	VC4_REG32(VEC_FREQ3_2),
+	VC4_REG32(VEC_FREQ1_0),
+	VC4_REG32(VEC_CONFIG1),
+	VC4_REG32(VEC_CONFIG2),
+	VC4_REG32(VEC_INTERRUPT_CONTROL),
+	VC4_REG32(VEC_INTERRUPT_STATUS),
+	VC4_REG32(VEC_FCW_SECAM_B),
+	VC4_REG32(VEC_SECAM_GAIN_VAL),
+	VC4_REG32(VEC_CONFIG3),
+	VC4_REG32(VEC_STATUS0),
+	VC4_REG32(VEC_MASK0),
+	VC4_REG32(VEC_CFG),
+	VC4_REG32(VEC_DAC_TEST),
+	VC4_REG32(VEC_DAC_CONFIG),
+	VC4_REG32(VEC_DAC_MISC),
+};
+
+static const struct vc4_vec_tv_mode vc4_vec_tv_modes[] = {
+	{
+		.mode = DRM_MODE_TV_MODE_NTSC,
+		.expected_htotal = 858,
+		.config0 = VEC_CONFIG0_NTSC_STD | VEC_CONFIG0_PDEN,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+	},
+	{
+		.mode = DRM_MODE_TV_MODE_NTSC_443,
+		.expected_htotal = 858,
+		.config0 = VEC_CONFIG0_NTSC_STD,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS | VEC_CONFIG1_CUSTOM_FREQ,
+		.custom_freq = 0x2a098acb,
+	},
+	{
+		.mode = DRM_MODE_TV_MODE_NTSC_J,
+		.expected_htotal = 858,
+		.config0 = VEC_CONFIG0_NTSC_STD,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+	},
+	{
+		.mode = DRM_MODE_TV_MODE_PAL,
+		.expected_htotal = 864,
+		.config0 = VEC_CONFIG0_PAL_BDGHI_STD,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+	},
+	{
+		/* PAL-60 */
+		.mode = DRM_MODE_TV_MODE_PAL,
+		.expected_htotal = 858,
+		.config0 = VEC_CONFIG0_PAL_M_STD,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS | VEC_CONFIG1_CUSTOM_FREQ,
+		.custom_freq = 0x2a098acb,
+	},
+	{
+		.mode = DRM_MODE_TV_MODE_PAL_M,
+		.expected_htotal = 858,
+		.config0 = VEC_CONFIG0_PAL_M_STD,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+	},
+	{
+		.mode = DRM_MODE_TV_MODE_PAL_N,
+		.expected_htotal = 864,
+		.config0 = VEC_CONFIG0_PAL_N_STD,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+	},
+	{
+		.mode = DRM_MODE_TV_MODE_SECAM,
+		.expected_htotal = 864,
+		.config0 = VEC_CONFIG0_SECAM_STD,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+		.custom_freq = 0x29c71c72,
+	},
+	{
+		/* 50Hz mono */
+		.mode = DRM_MODE_TV_MODE_MONOCHROME,
+		.expected_htotal = 864,
+		.config0 = VEC_CONFIG0_PAL_BDGHI_STD | VEC_CONFIG0_BURDIS |
+			   VEC_CONFIG0_CHRDIS,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+	},
+	{
+		/* 60Hz mono */
+		.mode = DRM_MODE_TV_MODE_MONOCHROME,
+		.expected_htotal = 858,
+		.config0 = VEC_CONFIG0_PAL_M_STD | VEC_CONFIG0_BURDIS |
+			   VEC_CONFIG0_CHRDIS,
+		.config1 = VEC_CONFIG1_C_CVBS_CVBS,
+	},
+};
+
+static inline const struct vc4_vec_tv_mode *
+vc4_vec_tv_mode_lookup(unsigned int mode, u16 htotal)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(vc4_vec_tv_modes); i++) {
+		const struct vc4_vec_tv_mode *tv_mode = &vc4_vec_tv_modes[i];
+
+		if (tv_mode->mode == mode &&
+		    tv_mode->expected_htotal == htotal)
+			return tv_mode;
+	}
+
+	return NULL;
+}
+
+static const struct drm_prop_enum_list legacy_tv_mode_names[] = {
+	{ VC4_VEC_TV_MODE_NTSC, "NTSC", },
+	{ VC4_VEC_TV_MODE_NTSC_443, "NTSC-443", },
+	{ VC4_VEC_TV_MODE_NTSC_J, "NTSC-J", },
+	{ VC4_VEC_TV_MODE_PAL, "PAL", },
+	{ VC4_VEC_TV_MODE_PAL_60, "PAL-60", },
+	{ VC4_VEC_TV_MODE_PAL_M, "PAL-M", },
+	{ VC4_VEC_TV_MODE_PAL_N, "PAL-N", },
+	{ VC4_VEC_TV_MODE_SECAM, "SECAM", },
+	{ VC4_VEC_TV_MODE_MONOCHROME, "Mono", },
+};
+
+static enum drm_connector_status
+vc4_vec_connector_detect(struct drm_connector *connector, bool force)
+{
+	return connector_status_unknown;
+}
+
+static void vc4_vec_connector_reset(struct drm_connector *connector)
+{
+	drm_atomic_helper_connector_reset(connector);
+	drm_atomic_helper_connector_tv_reset(connector);
+}
+
+static int
+vc4_vec_connector_set_property(struct drm_connector *connector,
+			       struct drm_connector_state *state,
+			       struct drm_property *property,
+			       uint64_t val)
+{
+	struct vc4_vec *vec = connector_to_vc4_vec(connector);
+
+	if (property != vec->legacy_tv_mode_property)
+		return -EINVAL;
+
+	switch (val) {
+	case VC4_VEC_TV_MODE_NTSC:
+		state->tv.mode = DRM_MODE_TV_MODE_NTSC;
+		break;
+
+	case VC4_VEC_TV_MODE_NTSC_443:
+		state->tv.mode = DRM_MODE_TV_MODE_NTSC_443;
+		break;
+
+	case VC4_VEC_TV_MODE_NTSC_J:
+		state->tv.mode = DRM_MODE_TV_MODE_NTSC_J;
+		break;
+
+	case VC4_VEC_TV_MODE_PAL:
+	case VC4_VEC_TV_MODE_PAL_60:
+		state->tv.mode = DRM_MODE_TV_MODE_PAL;
+		break;
+
+	case VC4_VEC_TV_MODE_PAL_M:
+		state->tv.mode = DRM_MODE_TV_MODE_PAL_M;
+		break;
+
+	case VC4_VEC_TV_MODE_PAL_N:
+		state->tv.mode = DRM_MODE_TV_MODE_PAL_N;
+		break;
+
+	case VC4_VEC_TV_MODE_SECAM:
+		state->tv.mode = DRM_MODE_TV_MODE_SECAM;
+		break;
+
+	case VC4_VEC_TV_MODE_MONOCHROME:
+		state->tv.mode = DRM_MODE_TV_MODE_MONOCHROME;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+vc4_vec_connector_get_property(struct drm_connector *connector,
+			       const struct drm_connector_state *state,
+			       struct drm_property *property,
+			       uint64_t *val)
+{
+	struct vc4_vec *vec = connector_to_vc4_vec(connector);
+
+	if (property != vec->legacy_tv_mode_property)
+		return -EINVAL;
+
+	switch (state->tv.mode) {
+	case DRM_MODE_TV_MODE_NTSC:
+		*val = VC4_VEC_TV_MODE_NTSC;
+		break;
+
+	case DRM_MODE_TV_MODE_NTSC_443:
+		*val = VC4_VEC_TV_MODE_NTSC_443;
+		break;
+
+	case DRM_MODE_TV_MODE_NTSC_J:
+		*val = VC4_VEC_TV_MODE_NTSC_J;
+		break;
+
+	case DRM_MODE_TV_MODE_PAL:
+		*val = VC4_VEC_TV_MODE_PAL;
+		break;
+
+	case DRM_MODE_TV_MODE_PAL_M:
+		*val = VC4_VEC_TV_MODE_PAL_M;
+		break;
+
+	case DRM_MODE_TV_MODE_PAL_N:
+		*val = VC4_VEC_TV_MODE_PAL_N;
+		break;
+
+	case DRM_MODE_TV_MODE_SECAM:
+		*val = VC4_VEC_TV_MODE_SECAM;
+		break;
+
+	case DRM_MODE_TV_MODE_MONOCHROME:
+		*val = VC4_VEC_TV_MODE_MONOCHROME;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const struct drm_connector_funcs vc4_vec_connector_funcs = {
+	.detect = vc4_vec_connector_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.reset = vc4_vec_connector_reset,
+	.atomic_duplicate_state = drm_atomic_helper_connector_duplicate_state,
+	.atomic_destroy_state = drm_atomic_helper_connector_destroy_state,
+	.atomic_get_property = vc4_vec_connector_get_property,
+	.atomic_set_property = vc4_vec_connector_set_property,
+};
+
+static const struct drm_connector_helper_funcs vc4_vec_connector_helper_funcs = {
+	.atomic_check = drm_atomic_helper_connector_tv_check,
+	.get_modes = drm_connector_helper_tv_get_modes,
+};
+
+static int vc4_vec_connector_init(struct drm_device *dev, struct vc4_vec *vec)
+{
+	struct drm_connector *connector = &vec->connector;
+	struct drm_property *prop;
+	int ret;
+
+	connector->interlace_allowed = true;
+
+	ret = drmm_connector_init(dev, connector, &vc4_vec_connector_funcs,
+				 DRM_MODE_CONNECTOR_Composite, NULL);
+	if (ret)
+		return ret;
+
+	drm_connector_helper_add(connector, &vc4_vec_connector_helper_funcs);
+
+	drm_object_attach_property(&connector->base,
+				   dev->mode_config.tv_mode_property,
+				   DRM_MODE_TV_MODE_NTSC);
+
+	prop = drm_property_create_enum(dev, 0, "mode",
+					legacy_tv_mode_names,
+					ARRAY_SIZE(legacy_tv_mode_names));
+	if (!prop)
+		return -ENOMEM;
+	vec->legacy_tv_mode_property = prop;
+
+	drm_object_attach_property(&connector->base, prop, VC4_VEC_TV_MODE_NTSC);
+
+	drm_connector_attach_tv_margin_properties(connector);
+
+	drm_connector_attach_encoder(connector, &vec->encoder.base);
+
+	return 0;
+}
+
+static void vc4_vec_encoder_disable(struct drm_encoder *encoder,
+				    struct drm_atomic_commit *state)
+{
+	struct drm_device *drm = encoder->dev;
+	struct vc4_vec *vec = encoder_to_vc4_vec(encoder);
+	int idx;
+
+	if (!drm_dev_enter(drm, &idx))
+		return;
+
+	VEC_WRITE(VEC_CFG, 0);
+	VEC_WRITE(VEC_DAC_MISC,
+		  VEC_DAC_MISC_VCD_PWRDN |
+		  VEC_DAC_MISC_BIAS_PWRDN |
+		  VEC_DAC_MISC_DAC_PWRDN |
+		  VEC_DAC_MISC_LDO_PWRDN);
+
+	clk_disable_unprepare(vec->clock);
+
+	pm_runtime_put(&vec->pdev->dev);
+
+	drm_dev_exit(idx);
+}
+
+static void vc4_vec_encoder_enable(struct drm_encoder *encoder,
+				   struct drm_atomic_commit *state)
+{
+	struct drm_device *drm = encoder->dev;
+	struct vc4_vec *vec = encoder_to_vc4_vec(encoder);
+	struct drm_connector *connector = &vec->connector;
+	struct drm_connector_state *conn_state =
+		drm_atomic_get_new_connector_state(state, connector);
+	struct drm_display_mode *adjusted_mode =
+		&encoder->crtc->state->adjusted_mode;
+	const struct vc4_vec_tv_mode *tv_mode;
+	int idx, ret;
+
+	if (!drm_dev_enter(drm, &idx))
+		return;
+
+	tv_mode = vc4_vec_tv_mode_lookup(conn_state->tv.mode,
+					 adjusted_mode->htotal);
+	if (!tv_mode)
+		goto err_dev_exit;
+
+	ret = pm_runtime_resume_and_get(&vec->pdev->dev);
+	if (ret < 0) {
+		drm_err(drm, "Failed to retain power domain: %d\n", ret);
+		goto err_dev_exit;
+	}
+
+	/*
+	 * We need to set the clock rate each time we enable the encoder
+	 * because there's a chance we share the same parent with the HDMI
+	 * clock, and both drivers are requesting different rates.
+	 * The good news is, these 2 encoders cannot be enabled at the same
+	 * time, thus preventing incompatible rate requests.
+	 */
+	ret = clk_set_rate(vec->clock, 108000000);
+	if (ret) {
+		drm_err(drm, "Failed to set clock rate: %d\n", ret);
+		goto err_put_runtime_pm;
+	}
+
+	ret = clk_prepare_enable(vec->clock);
+	if (ret) {
+		drm_err(drm, "Failed to turn on core clock: %d\n", ret);
+		goto err_put_runtime_pm;
+	}
+
+	/* Reset the different blocks */
+	VEC_WRITE(VEC_WSE_RESET, 1);
+	VEC_WRITE(VEC_SOFT_RESET, 1);
+
+	/* Disable the CGSM-A and WSE blocks */
+	VEC_WRITE(VEC_WSE_CONTROL, 0);
+
+	/* Write config common to all modes. */
+
+	/*
+	 * Color subcarrier phase: phase = 360 * SCHPH / 256.
+	 * 0x28 <=> 39.375 deg.
+	 */
+	VEC_WRITE(VEC_SCHPH, 0x28);
+
+	/*
+	 * Reset to default values.
+	 */
+	VEC_WRITE(VEC_CLMP0_START, 0xac);
+	VEC_WRITE(VEC_CLMP0_END, 0xec);
+	VEC_WRITE(VEC_CONFIG2,
+		  VEC_CONFIG2_UV_DIG_DIS |
+		  VEC_CONFIG2_RGB_DIG_DIS |
+		  ((adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) ? 0 : VEC_CONFIG2_PROG_SCAN));
+	VEC_WRITE(VEC_CONFIG3, VEC_CONFIG3_HORIZ_LEN_STD);
+	VEC_WRITE(VEC_DAC_CONFIG, vec->variant->dac_config);
+
+	/* Mask all interrupts. */
+	VEC_WRITE(VEC_MASK0, 0);
+
+	VEC_WRITE(VEC_CONFIG0, tv_mode->config0);
+	VEC_WRITE(VEC_CONFIG1, tv_mode->config1);
+
+	if (tv_mode->custom_freq) {
+		VEC_WRITE(VEC_FREQ3_2,
+			  (tv_mode->custom_freq >> 16) & 0xffff);
+		VEC_WRITE(VEC_FREQ1_0,
+			  tv_mode->custom_freq & 0xffff);
+	}
+
+	VEC_WRITE(VEC_DAC_MISC,
+		  VEC_DAC_MISC_VID_ACT | VEC_DAC_MISC_DAC_RST_N);
+	VEC_WRITE(VEC_CFG, VEC_CFG_VEC_EN);
+
+	drm_dev_exit(idx);
+	return;
+
+err_put_runtime_pm:
+	pm_runtime_put(&vec->pdev->dev);
+err_dev_exit:
+	drm_dev_exit(idx);
+}
+
+static int vc4_vec_encoder_atomic_check(struct drm_encoder *encoder,
+					struct drm_crtc_state *crtc_state,
+					struct drm_connector_state *conn_state)
+{
+	const struct drm_display_mode *mode = &crtc_state->adjusted_mode;
+	const struct vc4_vec_tv_mode *tv_mode;
+
+	tv_mode = vc4_vec_tv_mode_lookup(conn_state->tv.mode, mode->htotal);
+	if (!tv_mode)
+		return -EINVAL;
+
+	if (mode->crtc_hdisplay % 4)
+		return -EINVAL;
+
+	if (!(mode->crtc_hsync_end - mode->crtc_hsync_start))
+		return -EINVAL;
+
+	switch (mode->htotal) {
+	/* NTSC */
+	case 858:
+		if (mode->crtc_vtotal > 262)
+			return -EINVAL;
+
+		if (mode->crtc_vdisplay < 1 || mode->crtc_vdisplay > 253)
+			return -EINVAL;
+
+		if (!(mode->crtc_vsync_start - mode->crtc_vdisplay))
+			return -EINVAL;
+
+		if ((mode->crtc_vsync_end - mode->crtc_vsync_start) != 3)
+			return -EINVAL;
+
+		if ((mode->crtc_vtotal - mode->crtc_vsync_end) < 4)
+			return -EINVAL;
+
+		break;
+
+	/* PAL/SECAM */
+	case 864:
+		if (mode->crtc_vtotal > 312)
+			return -EINVAL;
+
+		if (mode->crtc_vdisplay < 1 || mode->crtc_vdisplay > 305)
+			return -EINVAL;
+
+		if (!(mode->crtc_vsync_start - mode->crtc_vdisplay))
+			return -EINVAL;
+
+		if ((mode->crtc_vsync_end - mode->crtc_vsync_start) != 3)
+			return -EINVAL;
+
+		if ((mode->crtc_vtotal - mode->crtc_vsync_end) < 2)
+			return -EINVAL;
+
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const struct drm_encoder_helper_funcs vc4_vec_encoder_helper_funcs = {
+	.atomic_check = vc4_vec_encoder_atomic_check,
+	.atomic_disable = vc4_vec_encoder_disable,
+	.atomic_enable = vc4_vec_encoder_enable,
+};
+
+static int vc4_vec_late_register(struct drm_encoder *encoder)
+{
+	struct drm_device *drm = encoder->dev;
+	struct vc4_vec *vec = encoder_to_vc4_vec(encoder);
+
+	vc4_debugfs_add_regset32(drm, "vec_regs", &vec->regset);
+
+	return 0;
+}
+
+static const struct drm_encoder_funcs vc4_vec_encoder_funcs = {
+	.late_register = vc4_vec_late_register,
+};
+
+static const struct vc4_vec_variant bcm2835_vec_variant = {
+	.dac_config = VEC_DAC_CONFIG_DAC_CTRL(0xc) |
+		      VEC_DAC_CONFIG_DRIVER_CTRL(0xc) |
+		      VEC_DAC_CONFIG_LDO_BIAS_CTRL(0x46)
+};
+
+static const struct vc4_vec_variant bcm2711_vec_variant = {
+	.dac_config = VEC_DAC_CONFIG_DAC_CTRL(0x0) |
+		      VEC_DAC_CONFIG_DRIVER_CTRL(0x80) |
+		      VEC_DAC_CONFIG_LDO_BIAS_CTRL(0x61)
+};
+
+static const struct of_device_id vc4_vec_dt_match[] = {
+	{ .compatible = "brcm,bcm2835-vec", .data = &bcm2835_vec_variant },
+	{ .compatible = "brcm,bcm2711-vec", .data = &bcm2711_vec_variant },
+	{ /* sentinel */ },
+};
+
+static int vc4_vec_bind(struct device *dev, struct device *master, void *data)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct drm_device *drm = dev_get_drvdata(master);
+	struct vc4_vec *vec;
+	int ret;
+
+	ret = drm_mode_create_tv_properties(drm,
+					    BIT(DRM_MODE_TV_MODE_NTSC) |
+					    BIT(DRM_MODE_TV_MODE_NTSC_443) |
+					    BIT(DRM_MODE_TV_MODE_NTSC_J) |
+					    BIT(DRM_MODE_TV_MODE_PAL) |
+					    BIT(DRM_MODE_TV_MODE_PAL_M) |
+					    BIT(DRM_MODE_TV_MODE_PAL_N) |
+					    BIT(DRM_MODE_TV_MODE_SECAM) |
+					    BIT(DRM_MODE_TV_MODE_MONOCHROME));
+	if (ret)
+		return ret;
+
+	vec = drmm_kzalloc(drm, sizeof(*vec), GFP_KERNEL);
+	if (!vec)
+		return -ENOMEM;
+
+	vec->encoder.type = VC4_ENCODER_TYPE_VEC;
+	vec->pdev = pdev;
+	vec->variant = (const struct vc4_vec_variant *)
+		of_device_get_match_data(dev);
+	vec->regs = vc4_ioremap_regs(pdev, 0);
+	if (IS_ERR(vec->regs))
+		return PTR_ERR(vec->regs);
+	vec->regset.base = vec->regs;
+	vec->regset.regs = vec_regs;
+	vec->regset.nregs = ARRAY_SIZE(vec_regs);
+
+	vec->clock = devm_clk_get(dev, NULL);
+	if (IS_ERR(vec->clock)) {
+		ret = PTR_ERR(vec->clock);
+		if (ret != -EPROBE_DEFER)
+			drm_err(drm, "Failed to get clock: %d\n", ret);
+		return ret;
+	}
+
+	ret = devm_pm_runtime_enable(dev);
+	if (ret)
+		return ret;
+
+	ret = drmm_encoder_init(drm, &vec->encoder.base,
+				&vc4_vec_encoder_funcs,
+				DRM_MODE_ENCODER_TVDAC,
+				NULL);
+	if (ret)
+		return ret;
+
+	drm_encoder_helper_add(&vec->encoder.base, &vc4_vec_encoder_helper_funcs);
+
+	ret = vc4_vec_connector_init(drm, vec);
+	if (ret)
+		return ret;
+
+	dev_set_drvdata(dev, vec);
+
+	return 0;
+}
+
+static const struct component_ops vc4_vec_ops = {
+	.bind   = vc4_vec_bind,
+};
+
+static int vc4_vec_dev_probe(struct platform_device *pdev)
+{
+	return component_add(&pdev->dev, &vc4_vec_ops);
+}
+
+static void vc4_vec_dev_remove(struct platform_device *pdev)
+{
+	component_del(&pdev->dev, &vc4_vec_ops);
+}
+
+struct platform_driver vc4_vec_driver = {
+	.probe = vc4_vec_dev_probe,
+	.remove = vc4_vec_dev_remove,
+	.driver = {
+		.name = "vc4_vec",
+		.of_match_table = vc4_vec_dt_match,
+	},
+};