1 files changed, 225 insertions, 199 deletions

diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
index 6fb7d74ff553..f455ce71e85d 100644
--- a/drivers/gpu/drm/tegra/drm.c
+++ b/drivers/gpu/drm/tegra/drm.c
@@ -20,10 +20,6 @@
 #include <drm/drm_prime.h>
 #include <drm/drm_vblank.h>
 
-#if IS_ENABLED(CONFIG_ARM_DMA_USE_IOMMU)
-#include <asm/dma-iommu.h>
-#endif
-
 #include "drm.h"
 #include "gem.h"
 
@@ -86,168 +82,6 @@ tegra_drm_mode_config_helpers = {
 	.atomic_commit_tail = tegra_atomic_commit_tail,
 };
 
-static int tegra_drm_load(struct drm_device *drm, unsigned long flags)
-{
-	struct host1x_device *device = to_host1x_device(drm->dev);
-	struct tegra_drm *tegra;
-	int err;
-
-	tegra = kzalloc(sizeof(*tegra), GFP_KERNEL);
-	if (!tegra)
-		return -ENOMEM;
-
-	if (iommu_present(&platform_bus_type)) {
-		tegra->domain = iommu_domain_alloc(&platform_bus_type);
-		if (!tegra->domain) {
-			err = -ENOMEM;
-			goto free;
-		}
-
-		err = iova_cache_get();
-		if (err < 0)
-			goto domain;
-	}
-
-	mutex_init(&tegra->clients_lock);
-	INIT_LIST_HEAD(&tegra->clients);
-
-	drm->dev_private = tegra;
-	tegra->drm = drm;
-
-	drm_mode_config_init(drm);
-
-	drm->mode_config.min_width = 0;
-	drm->mode_config.min_height = 0;
-
-	drm->mode_config.max_width = 4096;
-	drm->mode_config.max_height = 4096;
-
-	drm->mode_config.allow_fb_modifiers = true;
-
-	drm->mode_config.normalize_zpos = true;
-
-	drm->mode_config.funcs = &tegra_drm_mode_config_funcs;
-	drm->mode_config.helper_private = &tegra_drm_mode_config_helpers;
-
-	err = tegra_drm_fb_prepare(drm);
-	if (err < 0)
-		goto config;
-
-	drm_kms_helper_poll_init(drm);
-
-	err = host1x_device_init(device);
-	if (err < 0)
-		goto fbdev;
-
-	if (tegra->domain) {
-		u64 carveout_start, carveout_end, gem_start, gem_end;
-		u64 dma_mask = dma_get_mask(&device->dev);
-		dma_addr_t start, end;
-		unsigned long order;
-
-		start = tegra->domain->geometry.aperture_start & dma_mask;
-		end = tegra->domain->geometry.aperture_end & dma_mask;
-
-		gem_start = start;
-		gem_end = end - CARVEOUT_SZ;
-		carveout_start = gem_end + 1;
-		carveout_end = end;
-
-		order = __ffs(tegra->domain->pgsize_bitmap);
-		init_iova_domain(&tegra->carveout.domain, 1UL << order,
-				 carveout_start >> order);
-
-		tegra->carveout.shift = iova_shift(&tegra->carveout.domain);
-		tegra->carveout.limit = carveout_end >> tegra->carveout.shift;
-
-		drm_mm_init(&tegra->mm, gem_start, gem_end - gem_start + 1);
-		mutex_init(&tegra->mm_lock);
-
-		DRM_DEBUG("IOMMU apertures:\n");
-		DRM_DEBUG("  GEM: %#llx-%#llx\n", gem_start, gem_end);
-		DRM_DEBUG("  Carveout: %#llx-%#llx\n", carveout_start,
-			  carveout_end);
-	}
-
-	if (tegra->hub) {
-		err = tegra_display_hub_prepare(tegra->hub);
-		if (err < 0)
-			goto device;
-	}
-
-	/*
-	 * We don't use the drm_irq_install() helpers provided by the DRM
-	 * core, so we need to set this manually in order to allow the
-	 * DRM_IOCTL_WAIT_VBLANK to operate correctly.
-	 */
-	drm->irq_enabled = true;
-
-	/* syncpoints are used for full 32-bit hardware VBLANK counters */
-	drm->max_vblank_count = 0xffffffff;
-
-	err = drm_vblank_init(drm, drm->mode_config.num_crtc);
-	if (err < 0)
-		goto hub;
-
-	drm_mode_config_reset(drm);
-
-	err = tegra_drm_fb_init(drm);
-	if (err < 0)
-		goto hub;
-
-	return 0;
-
-hub:
-	if (tegra->hub)
-		tegra_display_hub_cleanup(tegra->hub);
-device:
-	host1x_device_exit(device);
-fbdev:
-	drm_kms_helper_poll_fini(drm);
-	tegra_drm_fb_free(drm);
-config:
-	drm_mode_config_cleanup(drm);
-
-	if (tegra->domain) {
-		mutex_destroy(&tegra->mm_lock);
-		drm_mm_takedown(&tegra->mm);
-		put_iova_domain(&tegra->carveout.domain);
-		iova_cache_put();
-	}
-domain:
-	if (tegra->domain)
-		iommu_domain_free(tegra->domain);
-free:
-	kfree(tegra);
-	return err;
-}
-
-static void tegra_drm_unload(struct drm_device *drm)
-{
-	struct host1x_device *device = to_host1x_device(drm->dev);
-	struct tegra_drm *tegra = drm->dev_private;
-	int err;
-
-	drm_kms_helper_poll_fini(drm);
-	tegra_drm_fb_exit(drm);
-	drm_atomic_helper_shutdown(drm);
-	drm_mode_config_cleanup(drm);
-
-	err = host1x_device_exit(device);
-	if (err < 0)
-		return;
-
-	if (tegra->domain) {
-		mutex_destroy(&tegra->mm_lock);
-		drm_mm_takedown(&tegra->mm);
-		put_iova_domain(&tegra->carveout.domain);
-		iova_cache_put();
-		iommu_domain_free(tegra->domain);
-	}
-
-	kfree(tegra);
-}
-
 static int tegra_drm_open(struct drm_device *drm, struct drm_file *filp)
 {
 	struct tegra_drm_file *fpriv;
@@ -311,6 +145,8 @@ static int host1x_reloc_copy_from_user(struct host1x_reloc *dest,
 	if (err < 0)
 		return err;
 
+	dest->flags = HOST1X_RELOC_READ | HOST1X_RELOC_WRITE;
+
 	dest->cmdbuf.bo = host1x_bo_lookup(file, cmdbuf);
 	if (!dest->cmdbuf.bo)
 		return -ENOENT;
@@ -1014,8 +850,6 @@ static int tegra_debugfs_init(struct drm_minor *minor)
 static struct drm_driver tegra_drm_driver = {
 	.driver_features = DRIVER_MODESET | DRIVER_GEM |
 			   DRIVER_ATOMIC | DRIVER_RENDER,
-	.load = tegra_drm_load,
-	.unload = tegra_drm_unload,
 	.open = tegra_drm_open,
 	.postclose = tegra_drm_postclose,
 	.lastclose = drm_fb_helper_lastclose,
@@ -1068,57 +902,61 @@ int tegra_drm_unregister_client(struct tegra_drm *tegra,
 	return 0;
 }
 
-struct iommu_group *host1x_client_iommu_attach(struct host1x_client *client,
-					       bool shared)
+int host1x_client_iommu_attach(struct host1x_client *client)
 {
+	struct iommu_domain *domain = iommu_get_domain_for_dev(client->dev);
 	struct drm_device *drm = dev_get_drvdata(client->parent);
 	struct tegra_drm *tegra = drm->dev_private;
 	struct iommu_group *group = NULL;
 	int err;
 
+	/*
+	 * If the host1x client is already attached to an IOMMU domain that is
+	 * not the shared IOMMU domain, don't try to attach it to a different
+	 * domain. This allows using the IOMMU-backed DMA API.
+	 */
+	if (domain && domain != tegra->domain)
+		return 0;
+
 	if (tegra->domain) {
 		group = iommu_group_get(client->dev);
-		if (!group) {
-			dev_err(client->dev, "failed to get IOMMU group\n");
-			return ERR_PTR(-ENODEV);
-		}
+		if (!group)
+			return -ENODEV;
 
-		if (!shared || (shared && (group != tegra->group))) {
-#if IS_ENABLED(CONFIG_ARM_DMA_USE_IOMMU)
-			if (client->dev->archdata.mapping) {
-				struct dma_iommu_mapping *mapping =
-					to_dma_iommu_mapping(client->dev);
-				arm_iommu_detach_device(client->dev);
-				arm_iommu_release_mapping(mapping);
-			}
-#endif
+		if (domain != tegra->domain) {
 			err = iommu_attach_group(tegra->domain, group);
 			if (err < 0) {
 				iommu_group_put(group);
-				return ERR_PTR(err);
+				return err;
 			}
-
-			if (shared && !tegra->group)
-				tegra->group = group;
 		}
+
+		tegra->use_explicit_iommu = true;
 	}
 
-	return group;
+	client->group = group;
+
+	return 0;
 }
 
-void host1x_client_iommu_detach(struct host1x_client *client,
-				struct iommu_group *group)
+void host1x_client_iommu_detach(struct host1x_client *client)
 {
 	struct drm_device *drm = dev_get_drvdata(client->parent);
 	struct tegra_drm *tegra = drm->dev_private;
+	struct iommu_domain *domain;
 
-	if (group) {
-		if (group == tegra->group) {
-			iommu_detach_group(tegra->domain, group);
-			tegra->group = NULL;
-		}
+	if (client->group) {
+		/*
+		 * Devices that are part of the same group may no longer be
+		 * attached to a domain at this point because their group may
+		 * have been detached by an earlier client.
+		 */
+		domain = iommu_get_domain_for_dev(client->dev);
+		if (domain)
+			iommu_detach_group(tegra->domain, client->group);
 
-		iommu_group_put(group);
+		iommu_group_put(client->group);
+		client->group = NULL;
 	}
 }
 
@@ -1202,6 +1040,8 @@ void tegra_drm_free(struct tegra_drm *tegra, size_t size, void *virt,
 static int host1x_drm_probe(struct host1x_device *dev)
 {
 	struct drm_driver *driver = &tegra_drm_driver;
+	struct iommu_domain *domain;
+	struct tegra_drm *tegra;
 	struct drm_device *drm;
 	int err;
 
@@ -1209,18 +1049,180 @@ static int host1x_drm_probe(struct host1x_device *dev)
 	if (IS_ERR(drm))
 		return PTR_ERR(drm);
 
+	tegra = kzalloc(sizeof(*tegra), GFP_KERNEL);
+	if (!tegra) {
+		err = -ENOMEM;
+		goto put;
+	}
+
+	/*
+	 * If the Tegra DRM clients are backed by an IOMMU, push buffers are
+	 * likely to be allocated beyond the 32-bit boundary if sufficient
+	 * system memory is available. This is problematic on earlier Tegra
+	 * generations where host1x supports a maximum of 32 address bits in
+	 * the GATHER opcode. In this case, unless host1x is behind an IOMMU
+	 * as well it won't be able to process buffers allocated beyond the
+	 * 32-bit boundary.
+	 *
+	 * The DMA API will use bounce buffers in this case, so that could
+	 * perhaps still be made to work, even if less efficient, but there
+	 * is another catch: in order to perform cache maintenance on pages
+	 * allocated for discontiguous buffers we need to map and unmap the
+	 * SG table representing these buffers. This is fine for something
+	 * small like a push buffer, but it exhausts the bounce buffer pool
+	 * (typically on the order of a few MiB) for framebuffers (many MiB
+	 * for any modern resolution).
+	 *
+	 * Work around this by making sure that Tegra DRM clients only use
+	 * an IOMMU if the parent host1x also uses an IOMMU.
+	 *
+	 * Note that there's still a small gap here that we don't cover: if
+	 * the DMA API is backed by an IOMMU there's no way to control which
+	 * device is attached to an IOMMU and which isn't, except via wiring
+	 * up the device tree appropriately. This is considered an problem
+	 * of integration, so care must be taken for the DT to be consistent.
+	 */
+	domain = iommu_get_domain_for_dev(drm->dev->parent);
+
+	if (domain && iommu_present(&platform_bus_type)) {
+		tegra->domain = iommu_domain_alloc(&platform_bus_type);
+		if (!tegra->domain) {
+			err = -ENOMEM;
+			goto free;
+		}
+
+		err = iova_cache_get();
+		if (err < 0)
+			goto domain;
+	}
+
+	mutex_init(&tegra->clients_lock);
+	INIT_LIST_HEAD(&tegra->clients);
+
 	dev_set_drvdata(&dev->dev, drm);
+	drm->dev_private = tegra;
+	tegra->drm = drm;
+
+	drm_mode_config_init(drm);
+
+	drm->mode_config.min_width = 0;
+	drm->mode_config.min_height = 0;
+
+	drm->mode_config.max_width = 4096;
+	drm->mode_config.max_height = 4096;
+
+	drm->mode_config.allow_fb_modifiers = true;
+
+	drm->mode_config.normalize_zpos = true;
 
-	err = drm_fb_helper_remove_conflicting_framebuffers(NULL, "tegradrmfb", false);
+	drm->mode_config.funcs = &tegra_drm_mode_config_funcs;
+	drm->mode_config.helper_private = &tegra_drm_mode_config_helpers;
+
+	err = tegra_drm_fb_prepare(drm);
 	if (err < 0)
-		goto put;
+		goto config;
+
+	drm_kms_helper_poll_init(drm);
+
+	err = host1x_device_init(dev);
+	if (err < 0)
+		goto fbdev;
+
+	if (tegra->use_explicit_iommu) {
+		u64 carveout_start, carveout_end, gem_start, gem_end;
+		u64 dma_mask = dma_get_mask(&dev->dev);
+		dma_addr_t start, end;
+		unsigned long order;
+
+		start = tegra->domain->geometry.aperture_start & dma_mask;
+		end = tegra->domain->geometry.aperture_end & dma_mask;
+
+		gem_start = start;
+		gem_end = end - CARVEOUT_SZ;
+		carveout_start = gem_end + 1;
+		carveout_end = end;
+
+		order = __ffs(tegra->domain->pgsize_bitmap);
+		init_iova_domain(&tegra->carveout.domain, 1UL << order,
+				 carveout_start >> order);
+
+		tegra->carveout.shift = iova_shift(&tegra->carveout.domain);
+		tegra->carveout.limit = carveout_end >> tegra->carveout.shift;
+
+		drm_mm_init(&tegra->mm, gem_start, gem_end - gem_start + 1);
+		mutex_init(&tegra->mm_lock);
+
+		DRM_DEBUG_DRIVER("IOMMU apertures:\n");
+		DRM_DEBUG_DRIVER("  GEM: %#llx-%#llx\n", gem_start, gem_end);
+		DRM_DEBUG_DRIVER("  Carveout: %#llx-%#llx\n", carveout_start,
+				 carveout_end);
+	} else if (tegra->domain) {
+		iommu_domain_free(tegra->domain);
+		tegra->domain = NULL;
+		iova_cache_put();
+	}
+
+	if (tegra->hub) {
+		err = tegra_display_hub_prepare(tegra->hub);
+		if (err < 0)
+			goto device;
+	}
+
+	/*
+	 * We don't use the drm_irq_install() helpers provided by the DRM
+	 * core, so we need to set this manually in order to allow the
+	 * DRM_IOCTL_WAIT_VBLANK to operate correctly.
+	 */
+	drm->irq_enabled = true;
+
+	/* syncpoints are used for full 32-bit hardware VBLANK counters */
+	drm->max_vblank_count = 0xffffffff;
+
+	err = drm_vblank_init(drm, drm->mode_config.num_crtc);
+	if (err < 0)
+		goto hub;
+
+	drm_mode_config_reset(drm);
+
+	err = drm_fb_helper_remove_conflicting_framebuffers(NULL, "tegradrmfb",
+							    false);
+	if (err < 0)
+		goto hub;
+
+	err = tegra_drm_fb_init(drm);
+	if (err < 0)
+		goto hub;
 
 	err = drm_dev_register(drm, 0);
 	if (err < 0)
-		goto put;
+		goto fb;
 
 	return 0;
 
+fb:
+	tegra_drm_fb_exit(drm);
+hub:
+	if (tegra->hub)
+		tegra_display_hub_cleanup(tegra->hub);
+device:
+	if (tegra->domain) {
+		mutex_destroy(&tegra->mm_lock);
+		drm_mm_takedown(&tegra->mm);
+		put_iova_domain(&tegra->carveout.domain);
+		iova_cache_put();
+	}
+
+	host1x_device_exit(dev);
+fbdev:
+	drm_kms_helper_poll_fini(drm);
+	tegra_drm_fb_free(drm);
+config:
+	drm_mode_config_cleanup(drm);
+domain:
+	if (tegra->domain)
+		iommu_domain_free(tegra->domain);
+free:
+	kfree(tegra);
 put:
 	drm_dev_put(drm);
 	return err;
@@ -1229,8 +1231,32 @@ put:
 static int host1x_drm_remove(struct host1x_device *dev)
 {
 	struct drm_device *drm = dev_get_drvdata(&dev->dev);
+	struct tegra_drm *tegra = drm->dev_private;
+	int err;
 
 	drm_dev_unregister(drm);
+
+	drm_kms_helper_poll_fini(drm);
+	tegra_drm_fb_exit(drm);
+	drm_atomic_helper_shutdown(drm);
+	drm_mode_config_cleanup(drm);
+
+	if (tegra->hub)
+		tegra_display_hub_cleanup(tegra->hub);
+
+	err = host1x_device_exit(dev);
+	if (err < 0)
+		dev_err(&dev->dev, "host1x device cleanup failed: %d\n", err);
+
+	if (tegra->domain) {
+		mutex_destroy(&tegra->mm_lock);
+		drm_mm_takedown(&tegra->mm);
+		put_iova_domain(&tegra->carveout.domain);
+		iova_cache_put();
+		iommu_domain_free(tegra->domain);
+	}
+
+	kfree(tegra);
 	drm_dev_put(drm);
 
 	return 0;