diff --git a/drivers/gpu/drm/tegra/drm.h b/drivers/gpu/drm/tegra/drm.h
index 87df251c1fcf..0cb868065348 100644
--- a/drivers/gpu/drm/tegra/drm.h
+++ b/drivers/gpu/drm/tegra/drm.h
@@ -25,7 +25,7 @@
 #include "trace.h"
 
 /* XXX move to include/uapi/drm/drm_fourcc.h? */
-#define DRM_FORMAT_MOD_NVIDIA_SECTOR_LAYOUT BIT(22)
+#define DRM_FORMAT_MOD_NVIDIA_SECTOR_LAYOUT BIT_ULL(22)
 
 struct reset_control;
 
diff --git a/drivers/gpu/drm/tegra/hub.c b/drivers/gpu/drm/tegra/hub.c
index 79bff8b48271..bfae8a02f55b 100644
--- a/drivers/gpu/drm/tegra/hub.c
+++ b/drivers/gpu/drm/tegra/hub.c
@@ -510,7 +510,7 @@ static void tegra_shared_plane_atomic_update(struct drm_plane *plane,
 	 * dGPU sector layout.
 	 */
 	if (tegra_plane_state->tiling.sector_layout == TEGRA_BO_SECTOR_LAYOUT_GPU)
-		base |= BIT(39);
+		base |= BIT_ULL(39);
 #endif
 
 	tegra_plane_writel(p, tegra_plane_state->format, DC_WIN_COLOR_DEPTH);
diff --git a/drivers/gpu/drm/tegra/sor.c b/drivers/gpu/drm/tegra/sor.c
index 7b88261f57bb..0ea320c1092b 100644
--- a/drivers/gpu/drm/tegra/sor.c
+++ b/drivers/gpu/drm/tegra/sor.c
@@ -3125,21 +3125,21 @@ static int tegra_sor_init(struct host1x_client *client)
 		if (err < 0) {
 			dev_err(sor->dev, "failed to acquire SOR reset: %d\n",
 				err);
-			return err;
+			goto rpm_put;
 		}
 
 		err = reset_control_assert(sor->rst);
 		if (err < 0) {
 			dev_err(sor->dev, "failed to assert SOR reset: %d\n",
 				err);
-			return err;
+			goto rpm_put;
 		}
 	}
 
 	err = clk_prepare_enable(sor->clk);
 	if (err < 0) {
 		dev_err(sor->dev, "failed to enable clock: %d\n", err);
-		return err;
+		goto rpm_put;
 	}
 
 	usleep_range(1000, 3000);
@@ -3150,7 +3150,7 @@ static int tegra_sor_init(struct host1x_client *client)
 			dev_err(sor->dev, "failed to deassert SOR reset: %d\n",
 				err);
 			clk_disable_unprepare(sor->clk);
-			return err;
+			goto rpm_put;
 		}
 
 		reset_control_release(sor->rst);
@@ -3171,6 +3171,12 @@ static int tegra_sor_init(struct host1x_client *client)
 	}
 
 	return 0;
+
+rpm_put:
+	if (sor->rst)
+		pm_runtime_put(sor->dev);
+
+	return err;
 }
 
 static int tegra_sor_exit(struct host1x_client *client)
@@ -3739,12 +3745,8 @@ static int tegra_sor_probe(struct platform_device *pdev)
 		if (!sor->aux)
 			return -EPROBE_DEFER;
 
-		if (get_device(&sor->aux->ddc.dev)) {
-			if (try_module_get(sor->aux->ddc.owner))
-				sor->output.ddc = &sor->aux->ddc;
-			else
-				put_device(&sor->aux->ddc.dev);
-		}
+		if (get_device(sor->aux->dev))
+			sor->output.ddc = &sor->aux->ddc;
 	}
 
 	if (!sor->aux) {
@@ -3772,12 +3774,13 @@ static int tegra_sor_probe(struct platform_device *pdev)
 
 	err = tegra_sor_parse_dt(sor);
 	if (err < 0)
-		return err;
+		goto put_aux;
 
 	err = tegra_output_probe(&sor->output);
-	if (err < 0)
-		return dev_err_probe(&pdev->dev, err,
-				     "failed to probe output\n");
+	if (err < 0) {
+		dev_err_probe(&pdev->dev, err, "failed to probe output\n");
+		goto put_aux;
+	}
 
 	if (sor->ops && sor->ops->probe) {
 		err = sor->ops->probe(sor);
@@ -3916,17 +3919,10 @@ static int tegra_sor_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, sor);
 	pm_runtime_enable(&pdev->dev);
 
-	INIT_LIST_HEAD(&sor->client.list);
+	host1x_client_init(&sor->client);
 	sor->client.ops = &sor_client_ops;
 	sor->client.dev = &pdev->dev;
 
-	err = host1x_client_register(&sor->client);
-	if (err < 0) {
-		dev_err(&pdev->dev, "failed to register host1x client: %d\n",
-			err);
-		goto rpm_disable;
-	}
-
 	/*
 	 * On Tegra210 and earlier, provide our own implementation for the
 	 * pad output clock.
@@ -3938,13 +3934,13 @@ static int tegra_sor_probe(struct platform_device *pdev)
 				      sor->index);
 		if (!name) {
 			err = -ENOMEM;
-			goto unregister;
+			goto uninit;
 		}
 
 		err = host1x_client_resume(&sor->client);
 		if (err < 0) {
 			dev_err(sor->dev, "failed to resume: %d\n", err);
-			goto unregister;
+			goto uninit;
 		}
 
 		sor->clk_pad = tegra_clk_sor_pad_register(sor, name);
@@ -3955,17 +3951,30 @@ static int tegra_sor_probe(struct platform_device *pdev)
 		err = PTR_ERR(sor->clk_pad);
 		dev_err(sor->dev, "failed to register SOR pad clock: %d\n",
 			err);
-		goto unregister;
+		goto uninit;
+	}
+
+	err = __host1x_client_register(&sor->client);
+	if (err < 0) {
+		dev_err(&pdev->dev, "failed to register host1x client: %d\n",
+			err);
+		goto uninit;
 	}
 
 	return 0;
 
-unregister:
-	host1x_client_unregister(&sor->client);
-rpm_disable:
+uninit:
+	host1x_client_exit(&sor->client);
 	pm_runtime_disable(&pdev->dev);
 remove:
+	if (sor->aux)
+		sor->output.ddc = NULL;
+
 	tegra_output_remove(&sor->output);
+put_aux:
+	if (sor->aux)
+		put_device(sor->aux->dev);
+
 	return err;
 }
 
@@ -3983,6 +3992,11 @@ static int tegra_sor_remove(struct platform_device *pdev)
 
 	pm_runtime_disable(&pdev->dev);
 
+	if (sor->aux) {
+		put_device(sor->aux->dev);
+		sor->output.ddc = NULL;
+	}
+
 	tegra_output_remove(&sor->output);
 
 	return 0;
diff --git a/drivers/gpu/host1x/bus.c b/drivers/gpu/host1x/bus.c
index 46f69c532b6b..218e3718fd68 100644
--- a/drivers/gpu/host1x/bus.c
+++ b/drivers/gpu/host1x/bus.c
@@ -735,6 +735,29 @@ void host1x_driver_unregister(struct host1x_driver *driver)
 }
 EXPORT_SYMBOL(host1x_driver_unregister);
 
+/**
+ * __host1x_client_init() - initialize a host1x client
+ * @client: host1x client
+ * @key: lock class key for the client-specific mutex
+ */
+void __host1x_client_init(struct host1x_client *client, struct lock_class_key *key)
+{
+	INIT_LIST_HEAD(&client->list);
+	__mutex_init(&client->lock, "host1x client lock", key);
+	client->usecount = 0;
+}
+EXPORT_SYMBOL(__host1x_client_init);
+
+/**
+ * host1x_client_exit() - uninitialize a host1x client
+ * @client: host1x client
+ */
+void host1x_client_exit(struct host1x_client *client)
+{
+	mutex_destroy(&client->lock);
+}
+EXPORT_SYMBOL(host1x_client_exit);
+
 /**
  * __host1x_client_register() - register a host1x client
  * @client: host1x client
@@ -747,16 +770,11 @@ EXPORT_SYMBOL(host1x_driver_unregister);
  * device and call host1x_device_init(), which will in turn call each client's
  * &host1x_client_ops.init implementation.
  */
-int __host1x_client_register(struct host1x_client *client,
-			     struct lock_class_key *key)
+int __host1x_client_register(struct host1x_client *client)
 {
 	struct host1x *host1x;
 	int err;
 
-	INIT_LIST_HEAD(&client->list);
-	__mutex_init(&client->lock, "host1x client lock", key);
-	client->usecount = 0;
-
 	mutex_lock(&devices_lock);
 
 	list_for_each_entry(host1x, &devices, list) {
diff --git a/include/linux/host1x.h b/include/linux/host1x.h
index 232e1bd507a7..9b0487c88571 100644
--- a/include/linux/host1x.h
+++ b/include/linux/host1x.h
@@ -332,12 +332,30 @@ static inline struct host1x_device *to_host1x_device(struct device *dev)
 int host1x_device_init(struct host1x_device *device);
 int host1x_device_exit(struct host1x_device *device);
 
-int __host1x_client_register(struct host1x_client *client,
-			     struct lock_class_key *key);
-#define host1x_client_register(class) \
-	({ \
-		static struct lock_class_key __key; \
-		__host1x_client_register(class, &__key); \
+void __host1x_client_init(struct host1x_client *client, struct lock_class_key *key);
+void host1x_client_exit(struct host1x_client *client);
+
+#define host1x_client_init(client)			\
+	({						\
+		static struct lock_class_key __key;	\
+		__host1x_client_init(client, &__key);	\
+	})
+
+int __host1x_client_register(struct host1x_client *client);
+
+/*
+ * Note that this wrapper calls __host1x_client_init() for compatibility
+ * with existing callers. Callers that want to separately initialize and
+ * register a host1x client must first initialize using either of the
+ * __host1x_client_init() or host1x_client_init() functions and then use
+ * the low-level __host1x_client_register() function to avoid the client
+ * getting reinitialized.
+ */
+#define host1x_client_register(client)			\
+	({						\
+		static struct lock_class_key __key;	\
+		__host1x_client_init(client, &__key);	\
+		__host1x_client_register(client);	\
 	})
 
 int host1x_client_unregister(struct host1x_client *client);