From ac832d51da00be28cf70a2c33cfb07ef06a7e94c Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Fri, 4 Dec 2015 11:35:34 -0800
Subject: [PATCH 114/304] drm/vc4: Update a bunch of code to match upstream
 submission.

This gets almost everything matching, except for the MSAA support and
using generic PM domains.

Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/drm_gem_cma_helper.c       |  13 +-
 drivers/gpu/drm/vc4/vc4_bo.c               | 322 +++++++++++++++++------------
 drivers/gpu/drm/vc4/vc4_crtc.c             |   7 +-
 drivers/gpu/drm/vc4/vc4_drv.c              |   6 +-
 drivers/gpu/drm/vc4/vc4_drv.h              |  20 +-
 drivers/gpu/drm/vc4/vc4_gem.c              |  24 ++-
 drivers/gpu/drm/vc4/vc4_irq.c              |   5 +-
 drivers/gpu/drm/vc4/vc4_kms.c              |   1 +
 drivers/gpu/drm/vc4/vc4_packet.h           | 210 +++++++++----------
 drivers/gpu/drm/vc4/vc4_qpu_defines.h      | 308 ++++++++++++++-------------
 drivers/gpu/drm/vc4/vc4_render_cl.c        |   4 +-
 drivers/gpu/drm/vc4/vc4_v3d.c              |  10 +-
 drivers/gpu/drm/vc4/vc4_validate.c         | 130 ++++++------
 drivers/gpu/drm/vc4/vc4_validate_shaders.c |  66 +++---
 include/drm/drmP.h                         |   8 +-
 15 files changed, 598 insertions(+), 536 deletions(-)

--- a/drivers/gpu/drm/drm_gem_cma_helper.c
+++ b/drivers/gpu/drm/drm_gem_cma_helper.c
@@ -58,15 +58,14 @@ __drm_gem_cma_create(struct drm_device *
 	struct drm_gem_cma_object *cma_obj;
 	struct drm_gem_object *gem_obj;
 	int ret;
-	size_t obj_size = (drm->driver->gem_obj_size ?
-			   drm->driver->gem_obj_size :
-			   sizeof(*cma_obj));
 
-	cma_obj = kzalloc(obj_size, GFP_KERNEL);
-	if (!cma_obj)
+	if (drm->driver->gem_create_object)
+		gem_obj = drm->driver->gem_create_object(drm, size);
+	else
+		gem_obj = kzalloc(sizeof(*cma_obj), GFP_KERNEL);
+	if (!gem_obj)
 		return ERR_PTR(-ENOMEM);
-
-	gem_obj = &cma_obj->base;
+	cma_obj = container_of(gem_obj, struct drm_gem_cma_object, base);
 
 	ret = drm_gem_object_init(drm, gem_obj, size);
 	if (ret)
--- a/drivers/gpu/drm/vc4/vc4_bo.c
+++ b/drivers/gpu/drm/vc4/vc4_bo.c
@@ -12,6 +12,10 @@
  * access to system memory with no MMU in between.  To support it, we
  * use the GEM CMA helper functions to allocate contiguous ranges of
  * physical memory for our BOs.
+ *
+ * Since the CMA allocator is very slow, we keep a cache of recently
+ * freed BOs around so that the kernel's allocation of objects for 3D
+ * rendering can return quickly.
  */
 
 #include "vc4_drv.h"
@@ -34,6 +38,36 @@ static void vc4_bo_stats_dump(struct vc4
 		 vc4->bo_stats.size_cached / 1024);
 }
 
+#ifdef CONFIG_DEBUG_FS
+int vc4_bo_stats_debugfs(struct seq_file *m, void *unused)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_bo_stats stats;
+
+	/* Take a snapshot of the current stats with the lock held. */
+	mutex_lock(&vc4->bo_lock);
+	stats = vc4->bo_stats;
+	mutex_unlock(&vc4->bo_lock);
+
+	seq_printf(m, "num bos allocated: %d\n",
+		   stats.num_allocated);
+	seq_printf(m, "size bos allocated: %dkb\n",
+		   stats.size_allocated / 1024);
+	seq_printf(m, "num bos used: %d\n",
+		   stats.num_allocated - stats.num_cached);
+	seq_printf(m, "size bos used: %dkb\n",
+		   (stats.size_allocated - stats.size_cached) / 1024);
+	seq_printf(m, "num bos cached: %d\n",
+		   stats.num_cached);
+	seq_printf(m, "size bos cached: %dkb\n",
+		   stats.size_cached / 1024);
+
+	return 0;
+}
+#endif
+
 static uint32_t bo_page_index(size_t size)
 {
 	return (size / PAGE_SIZE) - 1;
@@ -81,8 +115,8 @@ static struct list_head *vc4_get_cache_l
 		struct list_head *new_list;
 		uint32_t i;
 
-		new_list = kmalloc(new_size * sizeof(struct list_head),
-				   GFP_KERNEL);
+		new_list = kmalloc_array(new_size, sizeof(struct list_head),
+					 GFP_KERNEL);
 		if (!new_list)
 			return NULL;
 
@@ -90,7 +124,9 @@ static struct list_head *vc4_get_cache_l
 		 * head locations.
 		 */
 		for (i = 0; i < vc4->bo_cache.size_list_size; i++) {
-			struct list_head *old_list = &vc4->bo_cache.size_list[i];
+			struct list_head *old_list =
+				&vc4->bo_cache.size_list[i];
+
 			if (list_empty(old_list))
 				INIT_LIST_HEAD(&new_list[i]);
 			else
@@ -122,11 +158,60 @@ void vc4_bo_cache_purge(struct drm_devic
 	mutex_unlock(&vc4->bo_lock);
 }
 
-struct vc4_bo *vc4_bo_create(struct drm_device *dev, size_t unaligned_size)
+static struct vc4_bo *vc4_bo_get_from_cache(struct drm_device *dev,
+					    uint32_t size)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
-	uint32_t size = roundup(unaligned_size, PAGE_SIZE);
 	uint32_t page_index = bo_page_index(size);
+	struct vc4_bo *bo = NULL;
+
+	size = roundup(size, PAGE_SIZE);
+
+	mutex_lock(&vc4->bo_lock);
+	if (page_index >= vc4->bo_cache.size_list_size)
+		goto out;
+
+	if (list_empty(&vc4->bo_cache.size_list[page_index]))
+		goto out;
+
+	bo = list_first_entry(&vc4->bo_cache.size_list[page_index],
+			      struct vc4_bo, size_head);
+	vc4_bo_remove_from_cache(bo);
+	kref_init(&bo->base.base.refcount);
+
+out:
+	mutex_unlock(&vc4->bo_lock);
+	return bo;
+}
+
+/**
+ * vc4_gem_create_object - Implementation of driver->gem_create_object.
+ *
+ * This lets the CMA helpers allocate object structs for us, and keep
+ * our BO stats correct.
+ */
+struct drm_gem_object *vc4_create_object(struct drm_device *dev, size_t size)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+	struct vc4_bo *bo;
+
+	bo = kzalloc(sizeof(*bo), GFP_KERNEL);
+	if (!bo)
+		return ERR_PTR(-ENOMEM);
+
+	mutex_lock(&vc4->bo_lock);
+	vc4->bo_stats.num_allocated++;
+	vc4->bo_stats.size_allocated += size;
+	mutex_unlock(&vc4->bo_lock);
+
+	return &bo->base.base;
+}
+
+struct vc4_bo *vc4_bo_create(struct drm_device *dev, size_t unaligned_size,
+			     bool from_cache)
+{
+	size_t size = roundup(unaligned_size, PAGE_SIZE);
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	struct drm_gem_cma_object *cma_obj;
 	int pass;
 
@@ -134,18 +219,12 @@ struct vc4_bo *vc4_bo_create(struct drm_
 		return NULL;
 
 	/* First, try to get a vc4_bo from the kernel BO cache. */
-	mutex_lock(&vc4->bo_lock);
-	if (page_index < vc4->bo_cache.size_list_size &&
-	    !list_empty(&vc4->bo_cache.size_list[page_index])) {
-		struct vc4_bo *bo =
-			list_first_entry(&vc4->bo_cache.size_list[page_index],
-					 struct vc4_bo, size_head);
-		vc4_bo_remove_from_cache(bo);
-		mutex_unlock(&vc4->bo_lock);
-		kref_init(&bo->base.base.refcount);
-		return bo;
+	if (from_cache) {
+		struct vc4_bo *bo = vc4_bo_get_from_cache(dev, size);
+
+		if (bo)
+			return bo;
 	}
-	mutex_unlock(&vc4->bo_lock);
 
 	/* Otherwise, make a new BO. */
 	for (pass = 0; ; pass++) {
@@ -179,9 +258,6 @@ struct vc4_bo *vc4_bo_create(struct drm_
 		}
 	}
 
-	vc4->bo_stats.num_allocated++;
-	vc4->bo_stats.size_allocated += size;
-
 	return to_vc4_bo(&cma_obj->base);
 }
 
@@ -199,7 +275,7 @@ int vc4_dumb_create(struct drm_file *fil
 	if (args->size < args->pitch * args->height)
 		args->size = args->pitch * args->height;
 
-	bo = vc4_bo_create(dev, args->size);
+	bo = vc4_bo_create(dev, args->size, false);
 	if (!bo)
 		return -ENOMEM;
 
@@ -209,8 +285,8 @@ int vc4_dumb_create(struct drm_file *fil
 	return ret;
 }
 
-static void
-vc4_bo_cache_free_old(struct drm_device *dev)
+/* Must be called with bo_lock held. */
+static void vc4_bo_cache_free_old(struct drm_device *dev)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	unsigned long expire_time = jiffies - msecs_to_jiffies(1000);
@@ -313,15 +389,77 @@ vc4_prime_export(struct drm_device *dev,
 	return drm_gem_prime_export(dev, obj, flags);
 }
 
-int
-vc4_create_bo_ioctl(struct drm_device *dev, void *data,
-		    struct drm_file *file_priv)
+int vc4_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct drm_gem_object *gem_obj;
+	struct vc4_bo *bo;
+	int ret;
+
+	ret = drm_gem_mmap(filp, vma);
+	if (ret)
+		return ret;
+
+	gem_obj = vma->vm_private_data;
+	bo = to_vc4_bo(gem_obj);
+
+	if (bo->validated_shader && (vma->vm_flags & VM_WRITE)) {
+		DRM_ERROR("mmaping of shader BOs for writing not allowed.\n");
+		return -EINVAL;
+	}
+
+	/*
+	 * Clear the VM_PFNMAP flag that was set by drm_gem_mmap(), and set the
+	 * vm_pgoff (used as a fake buffer offset by DRM) to 0 as we want to map
+	 * the whole buffer.
+	 */
+	vma->vm_flags &= ~VM_PFNMAP;
+	vma->vm_pgoff = 0;
+
+	ret = dma_mmap_writecombine(bo->base.base.dev->dev, vma,
+				    bo->base.vaddr, bo->base.paddr,
+				    vma->vm_end - vma->vm_start);
+	if (ret)
+		drm_gem_vm_close(vma);
+
+	return ret;
+}
+
+int vc4_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma)
+{
+	struct vc4_bo *bo = to_vc4_bo(obj);
+
+	if (bo->validated_shader && (vma->vm_flags & VM_WRITE)) {
+		DRM_ERROR("mmaping of shader BOs for writing not allowed.\n");
+		return -EINVAL;
+	}
+
+	return drm_gem_cma_prime_mmap(obj, vma);
+}
+
+void *vc4_prime_vmap(struct drm_gem_object *obj)
+{
+	struct vc4_bo *bo = to_vc4_bo(obj);
+
+	if (bo->validated_shader) {
+		DRM_ERROR("mmaping of shader BOs not allowed.\n");
+		return ERR_PTR(-EINVAL);
+	}
+
+	return drm_gem_cma_prime_vmap(obj);
+}
+
+int vc4_create_bo_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *file_priv)
 {
 	struct drm_vc4_create_bo *args = data;
 	struct vc4_bo *bo = NULL;
 	int ret;
 
-	bo = vc4_bo_create(dev, args->size);
+	/*
+	 * We can't allocate from the BO cache, because the BOs don't
+	 * get zeroed, and that might leak data between users.
+	 */
+	bo = vc4_bo_create(dev, args->size, false);
 	if (!bo)
 		return -ENOMEM;
 
@@ -331,6 +469,25 @@ vc4_create_bo_ioctl(struct drm_device *d
 	return ret;
 }
 
+int vc4_mmap_bo_ioctl(struct drm_device *dev, void *data,
+		      struct drm_file *file_priv)
+{
+	struct drm_vc4_mmap_bo *args = data;
+	struct drm_gem_object *gem_obj;
+
+	gem_obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+	if (!gem_obj) {
+		DRM_ERROR("Failed to look up GEM BO %d\n", args->handle);
+		return -EINVAL;
+	}
+
+	/* The mmap offset was set up at BO allocation time. */
+	args->offset = drm_vma_node_offset_addr(&gem_obj->vma_node);
+
+	drm_gem_object_unreference_unlocked(gem_obj);
+	return 0;
+}
+
 int
 vc4_create_shader_bo_ioctl(struct drm_device *dev, void *data,
 			   struct drm_file *file_priv)
@@ -355,7 +512,7 @@ vc4_create_shader_bo_ioctl(struct drm_de
 		return -EINVAL;
 	}
 
-	bo = vc4_bo_create(dev, args->size);
+	bo = vc4_bo_create(dev, args->size, true);
 	if (!bo)
 		return -ENOMEM;
 
@@ -364,6 +521,11 @@ vc4_create_shader_bo_ioctl(struct drm_de
 			     args->size);
 	if (ret != 0)
 		goto fail;
+	/* Clear the rest of the memory from allocating from the BO
+	 * cache.
+	 */
+	memset(bo->base.vaddr + args->size, 0,
+	       bo->base.base.size - args->size);
 
 	bo->validated_shader = vc4_validate_shader(&bo->base);
 	if (!bo->validated_shader) {
@@ -382,85 +544,6 @@ vc4_create_shader_bo_ioctl(struct drm_de
 	return ret;
 }
 
-int
-vc4_mmap_bo_ioctl(struct drm_device *dev, void *data,
-		  struct drm_file *file_priv)
-{
-	struct drm_vc4_mmap_bo *args = data;
-	struct drm_gem_object *gem_obj;
-
-	gem_obj = drm_gem_object_lookup(dev, file_priv, args->handle);
-	if (!gem_obj) {
-		DRM_ERROR("Failed to look up GEM BO %d\n", args->handle);
-		return -EINVAL;
-	}
-
-	/* The mmap offset was set up at BO allocation time. */
-	args->offset = drm_vma_node_offset_addr(&gem_obj->vma_node);
-
-	drm_gem_object_unreference(gem_obj);
-	return 0;
-}
-
-int vc4_mmap(struct file *filp, struct vm_area_struct *vma)
-{
-	struct drm_gem_object *gem_obj;
-	struct vc4_bo *bo;
-	int ret;
-
-	ret = drm_gem_mmap(filp, vma);
-	if (ret)
-		return ret;
-
-	gem_obj = vma->vm_private_data;
-	bo = to_vc4_bo(gem_obj);
-
-	if (bo->validated_shader && (vma->vm_flags & VM_WRITE)) {
-		DRM_ERROR("mmaping of shader BOs for writing not allowed.\n");
-		return -EINVAL;
-	}
-
-	/*
-	 * Clear the VM_PFNMAP flag that was set by drm_gem_mmap(), and set the
-	 * vm_pgoff (used as a fake buffer offset by DRM) to 0 as we want to map
-	 * the whole buffer.
-	 */
-	vma->vm_flags &= ~VM_PFNMAP;
-	vma->vm_pgoff = 0;
-
-	ret = dma_mmap_writecombine(bo->base.base.dev->dev, vma,
-				    bo->base.vaddr, bo->base.paddr,
-				    vma->vm_end - vma->vm_start);
-	if (ret)
-		drm_gem_vm_close(vma);
-
-	return ret;
-}
-
-int vc4_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma)
-{
-	struct vc4_bo *bo = to_vc4_bo(obj);
-
-	if (bo->validated_shader) {
-		DRM_ERROR("mmaping of shader BOs not allowed.\n");
-		return -EINVAL;
-	}
-
-	return drm_gem_cma_prime_mmap(obj, vma);
-}
-
-void *vc4_prime_vmap(struct drm_gem_object *obj)
-{
-	struct vc4_bo *bo = to_vc4_bo(obj);
-
-	if (bo->validated_shader) {
-		DRM_ERROR("mmaping of shader BOs not allowed.\n");
-		return ERR_PTR(-EINVAL);
-	}
-
-	return drm_gem_cma_prime_vmap(obj);
-}
-
 void vc4_bo_cache_init(struct drm_device *dev)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
@@ -472,7 +555,7 @@ void vc4_bo_cache_init(struct drm_device
 	INIT_WORK(&vc4->bo_cache.time_work, vc4_bo_cache_time_work);
 	setup_timer(&vc4->bo_cache.time_timer,
 		    vc4_bo_cache_time_timer,
-		    (unsigned long) dev);
+		    (unsigned long)dev);
 }
 
 void vc4_bo_cache_destroy(struct drm_device *dev)
@@ -489,28 +572,3 @@ void vc4_bo_cache_destroy(struct drm_dev
 		vc4_bo_stats_dump(vc4);
 	}
 }
-
-#ifdef CONFIG_DEBUG_FS
-int vc4_bo_stats_debugfs(struct seq_file *m, void *unused)
-{
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct vc4_dev *vc4 = to_vc4_dev(dev);
-	struct vc4_bo_stats stats;
-
-	mutex_lock(&vc4->bo_lock);
-	stats = vc4->bo_stats;
-	mutex_unlock(&vc4->bo_lock);
-
-	seq_printf(m, "num bos allocated: %d\n", stats.num_allocated);
-	seq_printf(m, "size bos allocated: %dkb\n", stats.size_allocated / 1024);
-	seq_printf(m, "num bos used: %d\n", (stats.num_allocated -
-					     stats.num_cached));
-	seq_printf(m, "size bos used: %dkb\n", (stats.size_allocated -
-						stats.size_cached) / 1024);
-	seq_printf(m, "num bos cached: %d\n", stats.num_cached);
-	seq_printf(m, "size bos cached: %dkb\n", stats.size_cached / 1024);
-
-	return 0;
-}
-#endif
--- a/drivers/gpu/drm/vc4/vc4_crtc.c
+++ b/drivers/gpu/drm/vc4/vc4_crtc.c
@@ -501,6 +501,7 @@ vc4_async_page_flip_complete(struct vc4_
 	vc4_plane_async_set_fb(plane, flip_state->fb);
 	if (flip_state->event) {
 		unsigned long flags;
+
 		spin_lock_irqsave(&dev->event_lock, flags);
 		drm_crtc_send_vblank_event(crtc, flip_state->event);
 		spin_unlock_irqrestore(&dev->event_lock, flags);
@@ -562,9 +563,9 @@ static int vc4_async_page_flip(struct dr
 }
 
 static int vc4_page_flip(struct drm_crtc *crtc,
-		  struct drm_framebuffer *fb,
-		  struct drm_pending_vblank_event *event,
-		  uint32_t flags)
+			 struct drm_framebuffer *fb,
+			 struct drm_pending_vblank_event *event,
+			 uint32_t flags)
 {
 	if (flags & DRM_MODE_PAGE_FLIP_ASYNC)
 		return vc4_async_page_flip(crtc, fb, event, flags);
--- a/drivers/gpu/drm/vc4/vc4_drv.c
+++ b/drivers/gpu/drm/vc4/vc4_drv.c
@@ -81,7 +81,8 @@ static const struct drm_ioctl_desc vc4_d
 	DRM_IOCTL_DEF_DRV(VC4_CREATE_BO, vc4_create_bo_ioctl, 0),
 	DRM_IOCTL_DEF_DRV(VC4_MMAP_BO, vc4_mmap_bo_ioctl, 0),
 	DRM_IOCTL_DEF_DRV(VC4_CREATE_SHADER_BO, vc4_create_shader_bo_ioctl, 0),
-	DRM_IOCTL_DEF_DRV(VC4_GET_HANG_STATE, vc4_get_hang_state_ioctl, DRM_ROOT_ONLY),
+	DRM_IOCTL_DEF_DRV(VC4_GET_HANG_STATE, vc4_get_hang_state_ioctl,
+			  DRM_ROOT_ONLY),
 };
 
 static struct drm_driver vc4_drm_driver = {
@@ -107,6 +108,7 @@ static struct drm_driver vc4_drm_driver
 	.debugfs_cleanup = vc4_debugfs_cleanup,
 #endif
 
+	.gem_create_object = vc4_create_object,
 	.gem_free_object = vc4_free_object,
 	.gem_vm_ops = &drm_gem_cma_vm_ops,
 
@@ -128,8 +130,6 @@ static struct drm_driver vc4_drm_driver
 	.num_ioctls = ARRAY_SIZE(vc4_drm_ioctls),
 	.fops = &vc4_drm_fops,
 
-	//.gem_obj_size = sizeof(struct vc4_bo),
-
 	.name = DRIVER_NAME,
 	.desc = DRIVER_DESC,
 	.date = DRIVER_DATE,
--- a/drivers/gpu/drm/vc4/vc4_drv.h
+++ b/drivers/gpu/drm/vc4/vc4_drv.h
@@ -72,6 +72,9 @@ struct vc4_dev {
 	 * job_done_work.
 	 */
 	struct list_head job_done_list;
+	/* Spinlock used to synchronize the job_list and seqno
+	 * accesses between the IRQ handler and GEM ioctls.
+	 */
 	spinlock_t job_lock;
 	wait_queue_head_t job_wait_queue;
 	struct work_struct job_done_work;
@@ -318,8 +321,7 @@ struct vc4_texture_sample_info {
  * and validate the shader state record's uniforms that define the texture
  * samples.
  */
-struct vc4_validated_shader_info
-{
+struct vc4_validated_shader_info {
 	uint32_t uniforms_size;
 	uint32_t uniforms_src_size;
 	uint32_t num_texture_samples;
@@ -355,8 +357,10 @@ struct vc4_validated_shader_info
 #define wait_for(COND, MS) _wait_for(COND, MS, 1)
 
 /* vc4_bo.c */
+struct drm_gem_object *vc4_create_object(struct drm_device *dev, size_t size);
 void vc4_free_object(struct drm_gem_object *gem_obj);
-struct vc4_bo *vc4_bo_create(struct drm_device *dev, size_t size);
+struct vc4_bo *vc4_bo_create(struct drm_device *dev, size_t size,
+			     bool from_cache);
 int vc4_dumb_create(struct drm_file *file_priv,
 		    struct drm_device *dev,
 		    struct drm_mode_create_dumb *args);
@@ -432,7 +436,8 @@ struct drm_plane *vc4_plane_init(struct
 				 enum drm_plane_type type);
 u32 vc4_plane_write_dlist(struct drm_plane *plane, u32 __iomem *dlist);
 u32 vc4_plane_dlist_size(struct drm_plane_state *state);
-void vc4_plane_async_set_fb(struct drm_plane *plane, struct drm_framebuffer *fb);
+void vc4_plane_async_set_fb(struct drm_plane *plane,
+			    struct drm_framebuffer *fb);
 
 /* vc4_v3d.c */
 extern struct platform_driver vc4_v3d_driver;
@@ -450,9 +455,6 @@ vc4_validate_bin_cl(struct drm_device *d
 int
 vc4_validate_shader_recs(struct drm_device *dev, struct vc4_exec_info *exec);
 
-struct vc4_validated_shader_info *
-vc4_validate_shader(struct drm_gem_cma_object *shader_obj);
-
 bool vc4_use_bo(struct vc4_exec_info *exec,
 		uint32_t hindex,
 		enum vc4_bo_mode mode,
@@ -464,3 +466,7 @@ bool vc4_check_tex_size(struct vc4_exec_
 			struct drm_gem_cma_object *fbo,
 			uint32_t offset, uint8_t tiling_format,
 			uint32_t width, uint32_t height, uint8_t cpp);
+
+/* vc4_validate_shader.c */
+struct vc4_validated_shader_info *
+vc4_validate_shader(struct drm_gem_cma_object *shader_obj);
--- a/drivers/gpu/drm/vc4/vc4_gem.c
+++ b/drivers/gpu/drm/vc4/vc4_gem.c
@@ -53,9 +53,8 @@ vc4_free_hang_state(struct drm_device *d
 	unsigned int i;
 
 	mutex_lock(&dev->struct_mutex);
-	for (i = 0; i < state->user_state.bo_count; i++) {
+	for (i = 0; i < state->user_state.bo_count; i++)
 		drm_gem_object_unreference(state->bo[i]);
-	}
 	mutex_unlock(&dev->struct_mutex);
 
 	kfree(state);
@@ -65,10 +64,10 @@ int
 vc4_get_hang_state_ioctl(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv)
 {
- 	struct drm_vc4_get_hang_state *get_state = data;
+	struct drm_vc4_get_hang_state *get_state = data;
 	struct drm_vc4_get_hang_state_bo *bo_state;
 	struct vc4_hang_state *kernel_state;
- 	struct drm_vc4_get_hang_state *state;
+	struct drm_vc4_get_hang_state *state;
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	unsigned long irqflags;
 	u32 i;
@@ -107,6 +106,7 @@ vc4_get_hang_state_ioctl(struct drm_devi
 	for (i = 0; i < state->bo_count; i++) {
 		struct vc4_bo *vc4_bo = to_vc4_bo(kernel_state->bo[i]);
 		u32 handle;
+
 		ret = drm_gem_handle_create(file_priv, kernel_state->bo[i],
 					    &handle);
 
@@ -124,7 +124,7 @@ vc4_get_hang_state_ioctl(struct drm_devi
 			   state->bo_count * sizeof(*bo_state));
 	kfree(bo_state);
 
- err_free:
+err_free:
 
 	vc4_free_hang_state(dev, kernel_state);
 
@@ -578,7 +578,7 @@ vc4_get_bcl(struct drm_device *dev, stru
 		goto fail;
 	}
 
-	bo = vc4_bo_create(dev, exec_size);
+	bo = vc4_bo_create(dev, exec_size, true);
 	if (!bo) {
 		DRM_ERROR("Couldn't allocate BO for binning\n");
 		ret = PTR_ERR(exec->exec_bo);
@@ -668,6 +668,7 @@ vc4_job_handle_completed(struct vc4_dev
 static void vc4_seqno_cb_work(struct work_struct *work)
 {
 	struct vc4_seqno_cb *cb = container_of(work, struct vc4_seqno_cb, work);
+
 	cb->func(cb);
 }
 
@@ -717,6 +718,7 @@ vc4_wait_for_seqno_ioctl_helper(struct d
 
 	if ((ret == -EINTR || ret == -ERESTARTSYS) && *timeout_ns != ~0ull) {
 		uint64_t delta = jiffies_to_nsecs(jiffies - start);
+
 		if (*timeout_ns >= delta)
 			*timeout_ns -= delta;
 	}
@@ -750,9 +752,10 @@ vc4_wait_bo_ioctl(struct drm_device *dev
 	}
 	bo = to_vc4_bo(gem_obj);
 
-	ret = vc4_wait_for_seqno_ioctl_helper(dev, bo->seqno, &args->timeout_ns);
+	ret = vc4_wait_for_seqno_ioctl_helper(dev, bo->seqno,
+					      &args->timeout_ns);
 
-	drm_gem_object_unreference(gem_obj);
+	drm_gem_object_unreference_unlocked(gem_obj);
 	return ret;
 }
 
@@ -793,7 +796,8 @@ vc4_submit_cl_ioctl(struct drm_device *d
 		if (ret)
 			goto fail;
 	} else {
-		exec->ct0ca = exec->ct0ea = 0;
+		exec->ct0ca = 0;
+		exec->ct0ea = 0;
 	}
 
 	ret = vc4_get_rcl(dev, exec);
@@ -831,7 +835,7 @@ vc4_gem_init(struct drm_device *dev)
 	INIT_WORK(&vc4->hangcheck.reset_work, vc4_reset_work);
 	setup_timer(&vc4->hangcheck.timer,
 		    vc4_hangcheck_elapsed,
-		    (unsigned long) dev);
+		    (unsigned long)dev);
 
 	INIT_WORK(&vc4->job_done_work, vc4_job_done_work);
 }
--- a/drivers/gpu/drm/vc4/vc4_irq.c
+++ b/drivers/gpu/drm/vc4/vc4_irq.c
@@ -56,7 +56,7 @@ vc4_overflow_mem_work(struct work_struct
 	struct drm_device *dev = vc4->dev;
 	struct vc4_bo *bo;
 
-	bo = vc4_bo_create(dev, 256 * 1024);
+	bo = vc4_bo_create(dev, 256 * 1024, true);
 	if (!bo) {
 		DRM_ERROR("Couldn't allocate binner overflow mem\n");
 		return;
@@ -87,9 +87,8 @@ vc4_overflow_mem_work(struct work_struct
 		spin_unlock_irqrestore(&vc4->job_lock, irqflags);
 	}
 
-	if (vc4->overflow_mem) {
+	if (vc4->overflow_mem)
 		drm_gem_object_unreference_unlocked(&vc4->overflow_mem->base.base);
-	}
 	vc4->overflow_mem = bo;
 
 	V3D_WRITE(V3D_BPOA, bo->base.paddr);
--- a/drivers/gpu/drm/vc4/vc4_kms.c
+++ b/drivers/gpu/drm/vc4/vc4_kms.c
@@ -132,6 +132,7 @@ static int vc4_atomic_commit(struct drm_
 			struct drm_gem_cma_object *cma_bo =
 				drm_fb_cma_get_gem_obj(new_state->fb, 0);
 			struct vc4_bo *bo = to_vc4_bo(&cma_bo->base);
+
 			wait_seqno = max(bo->seqno, wait_seqno);
 		}
 	}
--- a/drivers/gpu/drm/vc4/vc4_packet.h
+++ b/drivers/gpu/drm/vc4/vc4_packet.h
@@ -27,60 +27,60 @@
 #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_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,
+	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
@@ -148,10 +148,10 @@ enum vc4_packet {
  * VC4_PACKET_LOAD_TILE_BUFFER_GENERAL (low bits of the address)
  */
 
-#define VC4_LOADSTORE_TILE_BUFFER_EOF                  (1 << 3)
-#define VC4_LOADSTORE_TILE_BUFFER_DISABLE_FULL_VG_MASK (1 << 2)
-#define VC4_LOADSTORE_TILE_BUFFER_DISABLE_FULL_ZS      (1 << 1)
-#define VC4_LOADSTORE_TILE_BUFFER_DISABLE_FULL_COLOR   (1 << 0)
+#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)
 
 /** @} */
 
@@ -160,10 +160,10 @@ enum vc4_packet {
  * 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 (1 << 15)
-#define VC4_STORE_TILE_BUFFER_DISABLE_ZS_CLEAR     (1 << 14)
-#define VC4_STORE_TILE_BUFFER_DISABLE_COLOR_CLEAR  (1 << 13)
-#define VC4_STORE_TILE_BUFFER_DISABLE_SWAP         (1 << 12)
+#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
@@ -201,28 +201,28 @@ enum vc4_packet {
 #define VC4_INDEX_BUFFER_U16                       (1 << 4)
 
 /* This flag is only present in NV shader state. */
-#define VC4_SHADER_FLAG_SHADED_CLIP_COORDS         (1 << 3)
-#define VC4_SHADER_FLAG_ENABLE_CLIPPING            (1 << 2)
-#define VC4_SHADER_FLAG_VS_POINT_SIZE              (1 << 1)
-#define VC4_SHADER_FLAG_FS_SINGLE_THREAD           (1 << 0)
+#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             (1 << 1)
-#define VC4_CONFIG_BITS_EARLY_Z                    (1 << 0)
+#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                   (1 << 7)
+#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        (1 << 3)
+#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       (1 << 0)
+#define VC4_CONFIG_BITS_COVERAGE_PIPE_SELECT       BIT(0)
 /** @} */
 
 /** @{ byte 0 of config bits. */
@@ -230,15 +230,15 @@ enum vc4_packet {
 #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        (1 << 4)
-#define VC4_CONFIG_BITS_ENABLE_DEPTH_OFFSET        (1 << 3)
-#define VC4_CONFIG_BITS_CW_PRIMITIVES              (1 << 2)
-#define VC4_CONFIG_BITS_ENABLE_PRIM_BACK           (1 << 1)
-#define VC4_CONFIG_BITS_ENABLE_PRIM_FRONT          (1 << 0)
+#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                   (1 << 7)
+#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
@@ -254,17 +254,17 @@ enum vc4_packet {
 #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              (1 << 2)
-#define VC4_BIN_CONFIG_TILE_BUFFER_64BIT           (1 << 1)
-#define VC4_BIN_CONFIG_MS_MODE_4X                  (1 << 0)
+#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                (1 << 12)
-#define VC4_RENDER_CONFIG_EARLY_Z_COVERAGE_DISABLE (1 << 11)
-#define VC4_RENDER_CONFIG_EARLY_Z_DIRECTION_G      (1 << 10)
-#define VC4_RENDER_CONFIG_COVERAGE_MODE            (1 << 9)
-#define VC4_RENDER_CONFIG_ENABLE_VG_MASK           (1 << 8)
+#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)
@@ -280,8 +280,8 @@ enum vc4_packet {
 #define VC4_RENDER_CONFIG_FORMAT_RGBA8888          1
 #define VC4_RENDER_CONFIG_FORMAT_BGR565            2
 
-#define VC4_RENDER_CONFIG_TILE_BUFFER_64BIT        (1 << 1)
-#define VC4_RENDER_CONFIG_MS_MODE_4X               (1 << 0)
+#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)
@@ -291,24 +291,24 @@ enum vc4_packet {
 #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,
+	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)
--- a/drivers/gpu/drm/vc4/vc4_qpu_defines.h
+++ b/drivers/gpu/drm/vc4/vc4_qpu_defines.h
@@ -25,194 +25,190 @@
 #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,
+	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,
+	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 = 41,
-        QPU_R_VPM = 48,
-        QPU_R_VPM_LD_BUSY,
-        QPU_R_VPM_LD_WAIT,
-        QPU_R_MUTEX_ACQUIRE,
+	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 = 41,
+	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,
+	/* 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
+	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,
+	/* 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,
+	/* 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,
+	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,
-        QPU_PACK_MUL_8888 = 3, /* replicated to each 8 bits of the 32-bit dst. */
-        QPU_PACK_MUL_8A,
-        QPU_PACK_MUL_8B,
-        QPU_PACK_MUL_8C,
-        QPU_PACK_MUL_8D,
+	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,
+	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))
-/* Using the GNU statement expression extension */
-#define QPU_SET_FIELD(value, field)                                       \
-        ({                                                                \
-                uint64_t fieldval = (uint64_t)(value) << field ## _SHIFT; \
-                assert((fieldval & ~ field ## _MASK) == 0);               \
-                fieldval & field ## _MASK;                                \
-         })
+	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_GET_FIELD(word, field) \
+	((uint32_t)(((word)  & field ## _MASK) >> field ## _SHIFT))
 
 #define QPU_SIG_SHIFT                   60
 #define QPU_SIG_MASK                    QPU_MASK(63, 60)
--- a/drivers/gpu/drm/vc4/vc4_render_cl.c
+++ b/drivers/gpu/drm/vc4/vc4_render_cl.c
@@ -63,7 +63,6 @@ static inline void rcl_u32(struct vc4_rc
 	setup->next_offset += 4;
 }
 
-
 /*
  * Emits a no-op STORE_TILE_BUFFER_GENERAL.
  *
@@ -217,7 +216,7 @@ static int vc4_create_rcl_bo(struct drm_
 	}
 	size += xtiles * ytiles * loop_body_size;
 
-	setup->rcl = &vc4_bo_create(dev, size)->base;
+	setup->rcl = &vc4_bo_create(dev, size, true)->base;
 	if (!setup->rcl)
 		return -ENOMEM;
 	list_add_tail(&to_vc4_bo(&setup->rcl->base)->unref_head,
@@ -256,6 +255,7 @@ static int vc4_create_rcl_bo(struct drm_
 		for (x = min_x_tile; x <= max_x_tile; x++) {
 			bool first = (x == min_x_tile && y == min_y_tile);
 			bool last = (x == max_x_tile && y == max_y_tile);
+
 			emit_tile(exec, setup, x, y, first, last);
 		}
 	}
--- a/drivers/gpu/drm/vc4/vc4_v3d.c
+++ b/drivers/gpu/drm/vc4/vc4_v3d.c
@@ -125,7 +125,7 @@ int vc4_v3d_debugfs_regs(struct seq_file
 
 int vc4_v3d_debugfs_ident(struct seq_file *m, void *unused)
 {
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
 	struct drm_device *dev = node->minor->dev;
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	uint32_t ident1 = V3D_READ(V3D_IDENT1);
@@ -133,11 +133,13 @@ int vc4_v3d_debugfs_ident(struct seq_fil
 	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, "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));
+	seq_printf(m, "Semaphores: %d\n",
+		   VC4_GET_FIELD(ident1, V3D_IDENT1_NSEM));
 
 	return 0;
 }
@@ -218,7 +220,7 @@ static int vc4_v3d_bind(struct device *d
 }
 
 static void vc4_v3d_unbind(struct device *dev, struct device *master,
-			    void *data)
+			   void *data)
 {
 	struct drm_device *drm = dev_get_drvdata(master);
 	struct vc4_dev *vc4 = to_vc4_dev(drm);
--- a/drivers/gpu/drm/vc4/vc4_validate.c
+++ b/drivers/gpu/drm/vc4/vc4_validate.c
@@ -48,7 +48,6 @@
 	void *validated,				\
 	void *untrusted
 
-
 /** Return the width in pixels of a 64-byte microtile. */
 static uint32_t
 utile_width(int cpp)
@@ -192,7 +191,7 @@ vc4_check_tex_size(struct vc4_exec_info
 
 	if (size + offset < size ||
 	    size + offset > fbo->base.size) {
-		DRM_ERROR("Overflow in %dx%d (%dx%d) fbo size (%d + %d > %d)\n",
+		DRM_ERROR("Overflow in %dx%d (%dx%d) fbo size (%d + %d > %zd)\n",
 			  width, height,
 			  aligned_width, aligned_height,
 			  size, offset, fbo->base.size);
@@ -278,7 +277,7 @@ validate_indexed_prim_list(VALIDATE_ARGS
 
 	if (offset > ib->base.size ||
 	    (ib->base.size - offset) / index_size < length) {
-		DRM_ERROR("IB access overflow (%d + %d*%d > %d)\n",
+		DRM_ERROR("IB access overflow (%d + %d*%d > %zd)\n",
 			  offset, length, index_size, ib->base.size);
 		return -EINVAL;
 	}
@@ -377,6 +376,7 @@ static int
 validate_tile_binning_config(VALIDATE_ARGS)
 {
 	struct drm_device *dev = exec->exec_bo->base.dev;
+	struct vc4_bo *tile_bo;
 	uint8_t flags;
 	uint32_t tile_state_size, tile_alloc_size;
 	uint32_t tile_count;
@@ -438,12 +438,12 @@ validate_tile_binning_config(VALIDATE_AR
 	 */
 	tile_alloc_size += 1024 * 1024;
 
-	exec->tile_bo = &vc4_bo_create(dev, exec->tile_alloc_offset +
-				       tile_alloc_size)->base;
+	tile_bo = vc4_bo_create(dev, exec->tile_alloc_offset + tile_alloc_size,
+				true);
+	exec->tile_bo = &tile_bo->base;
 	if (!exec->tile_bo)
 		return -ENOMEM;
-	list_add_tail(&to_vc4_bo(&exec->tile_bo->base)->unref_head,
-		     &exec->unref_list);
+	list_add_tail(&tile_bo->unref_head, &exec->unref_list);
 
 	/* tile alloc address. */
 	*(uint32_t *)(validated + 0) = (exec->tile_bo->paddr +
@@ -463,8 +463,8 @@ validate_gem_handles(VALIDATE_ARGS)
 	return 0;
 }
 
-#define VC4_DEFINE_PACKET(packet, name, func) \
-	[packet] = { packet ## _SIZE, name, func }
+#define VC4_DEFINE_PACKET(packet, func) \
+	[packet] = { packet ## _SIZE, #packet, func }
 
 static const struct cmd_info {
 	uint16_t len;
@@ -472,42 +472,43 @@ static const struct cmd_info {
 	int (*func)(struct vc4_exec_info *exec, void *validated,
 		    void *untrusted);
 } cmd_info[] = {
-	VC4_DEFINE_PACKET(VC4_PACKET_HALT, "halt", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_NOP, "nop", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH, "flush", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH_ALL, "flush all state", validate_flush_all),
-	VC4_DEFINE_PACKET(VC4_PACKET_START_TILE_BINNING, "start tile binning", validate_start_tile_binning),
-	VC4_DEFINE_PACKET(VC4_PACKET_INCREMENT_SEMAPHORE, "increment semaphore", validate_increment_semaphore),
-
-	VC4_DEFINE_PACKET(VC4_PACKET_GL_INDEXED_PRIMITIVE, "Indexed Primitive List", validate_indexed_prim_list),
-
-	VC4_DEFINE_PACKET(VC4_PACKET_GL_ARRAY_PRIMITIVE, "Vertex Array Primitives", validate_gl_array_primitive),
-
-	/* This is only used by clipped primitives (packets 48 and 49), which
-	 * we don't support parsing yet.
-	 */
-	VC4_DEFINE_PACKET(VC4_PACKET_PRIMITIVE_LIST_FORMAT, "primitive list format", NULL),
-
-	VC4_DEFINE_PACKET(VC4_PACKET_GL_SHADER_STATE, "GL Shader State", validate_gl_shader_state),
-	VC4_DEFINE_PACKET(VC4_PACKET_NV_SHADER_STATE, "NV Shader State", validate_nv_shader_state),
-
-	VC4_DEFINE_PACKET(VC4_PACKET_CONFIGURATION_BITS, "configuration bits", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_FLAT_SHADE_FLAGS, "flat shade flags", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_POINT_SIZE, "point size", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_LINE_WIDTH, "line width", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_RHT_X_BOUNDARY, "RHT X boundary", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_DEPTH_OFFSET, "Depth Offset", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_CLIP_WINDOW, "Clip Window", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_VIEWPORT_OFFSET, "Viewport Offset", NULL),
-	VC4_DEFINE_PACKET(VC4_PACKET_CLIPPER_XY_SCALING, "Clipper XY Scaling", NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_HALT, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_NOP, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH, NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_FLUSH_ALL, validate_flush_all),
+	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_NV_SHADER_STATE, validate_nv_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, "Clipper Z Scale and Offset", NULL),
+	VC4_DEFINE_PACKET(VC4_PACKET_CLIPPER_Z_SCALING, NULL),
 
-	VC4_DEFINE_PACKET(VC4_PACKET_TILE_BINNING_MODE_CONFIG, "tile binning configuration", validate_tile_binning_config),
+	VC4_DEFINE_PACKET(VC4_PACKET_TILE_BINNING_MODE_CONFIG,
+			  validate_tile_binning_config),
 
-	VC4_DEFINE_PACKET(VC4_PACKET_GEM_HANDLES, "GEM handles", validate_gem_handles),
+	VC4_DEFINE_PACKET(VC4_PACKET_GEM_HANDLES, validate_gem_handles),
 };
 
 int
@@ -526,7 +527,7 @@ vc4_validate_bin_cl(struct drm_device *d
 		u8 cmd = *(uint8_t *)src_pkt;
 		const struct cmd_info *info;
 
-		if (cmd > ARRAY_SIZE(cmd_info)) {
+		if (cmd >= ARRAY_SIZE(cmd_info)) {
 			DRM_ERROR("0x%08x: packet %d out of bounds\n",
 				  src_offset, cmd);
 			return -EINVAL;
@@ -539,11 +540,6 @@ vc4_validate_bin_cl(struct drm_device *d
 			return -EINVAL;
 		}
 
-#if 0
-		DRM_INFO("0x%08x: packet %d (%s) size %d processing...\n",
-			 src_offset, cmd, info->name, info->len);
-#endif
-
 		if (src_offset + info->len > len) {
 			DRM_ERROR("0x%08x: packet %d (%s) length 0x%08x "
 				  "exceeds bounds (0x%08x)\n",
@@ -558,8 +554,7 @@ vc4_validate_bin_cl(struct drm_device *d
 		if (info->func && info->func(exec,
 					     dst_pkt + 1,
 					     src_pkt + 1)) {
-			DRM_ERROR("0x%08x: packet %d (%s) failed to "
-				  "validate\n",
+			DRM_ERROR("0x%08x: packet %d (%s) failed to validate\n",
 				  src_offset, cmd, info->name);
 			return -EINVAL;
 		}
@@ -618,12 +613,14 @@ reloc_tex(struct vc4_exec_info *exec,
 
 	if (sample->is_direct) {
 		uint32_t remaining_size = tex->base.size - p0;
+
 		if (p0 > tex->base.size - 4) {
 			DRM_ERROR("UBO offset greater than UBO size\n");
 			goto fail;
 		}
 		if (p1 > remaining_size - 4) {
-			DRM_ERROR("UBO clamp would allow reads outside of UBO\n");
+			DRM_ERROR("UBO clamp would allow reads "
+				  "outside of UBO\n");
 			goto fail;
 		}
 		*validated_p0 = tex->paddr + p0;
@@ -786,7 +783,7 @@ validate_shader_rec(struct drm_device *d
 	struct drm_gem_cma_object *bo[ARRAY_SIZE(gl_relocs) + 8];
 	uint32_t nr_attributes = 0, nr_fixed_relocs, nr_relocs, packet_size;
 	int i;
-	struct vc4_validated_shader_info *validated_shader;
+	struct vc4_validated_shader_info *shader;
 
 	if (state->packet == VC4_PACKET_NV_SHADER_STATE) {
 		relocs = nv_relocs;
@@ -841,12 +838,12 @@ validate_shader_rec(struct drm_device *d
 		else
 			mode = VC4_MODE_RENDER;
 
-		if (!vc4_use_bo(exec, src_handles[i], mode, &bo[i])) {
+		if (!vc4_use_bo(exec, src_handles[i], mode, &bo[i]))
 			return false;
-		}
 	}
 
 	for (i = 0; i < nr_fixed_relocs; i++) {
+		struct vc4_bo *vc4_bo;
 		uint32_t o = relocs[i].offset;
 		uint32_t src_offset = *(uint32_t *)(pkt_u + o);
 		uint32_t *texture_handles_u;
@@ -858,34 +855,34 @@ validate_shader_rec(struct drm_device *d
 		switch (relocs[i].type) {
 		case RELOC_CODE:
 			if (src_offset != 0) {
-				DRM_ERROR("Shaders must be at offset 0 of "
-					  "the BO.\n");
+				DRM_ERROR("Shaders must be at offset 0 "
+					  "of the BO.\n");
 				goto fail;
 			}
 
-			validated_shader = to_vc4_bo(&bo[i]->base)->validated_shader;
-			if (!validated_shader)
+			vc4_bo = to_vc4_bo(&bo[i]->base);
+			shader = vc4_bo->validated_shader;
+			if (!shader)
 				goto fail;
 
-			if (validated_shader->uniforms_src_size >
-			    exec->uniforms_size) {
+			if (shader->uniforms_src_size > exec->uniforms_size) {
 				DRM_ERROR("Uniforms src buffer overflow\n");
 				goto fail;
 			}
 
 			texture_handles_u = exec->uniforms_u;
 			uniform_data_u = (texture_handles_u +
-					  validated_shader->num_texture_samples);
+					  shader->num_texture_samples);
 
 			memcpy(exec->uniforms_v, uniform_data_u,
-			       validated_shader->uniforms_size);
+			       shader->uniforms_size);
 
 			for (tex = 0;
-			     tex < validated_shader->num_texture_samples;
+			     tex < shader->num_texture_samples;
 			     tex++) {
 				if (!reloc_tex(exec,
 					       uniform_data_u,
-					       &validated_shader->texture_samples[tex],
+					       &shader->texture_samples[tex],
 					       texture_handles_u[tex])) {
 					goto fail;
 				}
@@ -893,9 +890,9 @@ validate_shader_rec(struct drm_device *d
 
 			*(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;
+			exec->uniforms_u += shader->uniforms_src_size;
+			exec->uniforms_v += shader->uniforms_size;
+			exec->uniforms_p += shader->uniforms_size;
 
 			break;
 
@@ -926,7 +923,8 @@ validate_shader_rec(struct drm_device *d
 			max_index = ((vbo->base.size - offset - attr_size) /
 				     stride);
 			if (state->max_index > max_index) {
-				DRM_ERROR("primitives use index %d out of supplied %d\n",
+				DRM_ERROR("primitives use index %d out of "
+					  "supplied %d\n",
 					  state->max_index, max_index);
 				return -EINVAL;
 			}
--- a/drivers/gpu/drm/vc4/vc4_validate_shaders.c
+++ b/drivers/gpu/drm/vc4/vc4_validate_shaders.c
@@ -24,24 +24,16 @@
 /**
  * DOC: Shader validator for VC4.
  *
- * The VC4 has no IOMMU between it and system memory.  So, 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, or uniform
- * data, or vertex data).
+ * The VC4 has no IOMMU between it and system memory, so 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, or uniform data, or vertex data).
  *
- * This walks over a shader starting from some offset within a BO, ensuring
- * that its accesses are appropriately bounded, and recording how many texture
- * accesses are made and where so that we can do relocations for them in the
+ * This walks over a shader BO, ensuring that its accesses are
+ * appropriately bounded, and recording how many texture accesses are
+ * made and where so that we can do relocations for them in the
  * uniform stream.
- *
- * The kernel API has shaders stored in user-mapped BOs.  The BOs will be
- * forcibly unmapped from the process before validation, and any cache of
- * validated state will be flushed if the mapping is faulted back in.
- *
- * Storing the shaders in BOs means that the validation process will be slow
- * due to uncached reads, but since shaders are long-lived and shader BOs are
- * never actually modified, this shouldn't be a problem.
  */
 
 #include "vc4_drv.h"
@@ -70,7 +62,6 @@ waddr_to_live_reg_index(uint32_t waddr,
 		else
 			return waddr;
 	} else if (waddr <= QPU_W_ACC3) {
-
 		return 64 + waddr - QPU_W_ACC0;
 	} else {
 		return ~0;
@@ -85,15 +76,14 @@ raddr_add_a_to_live_reg_index(uint64_t i
 	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) {
+	if (add_a == QPU_MUX_A)
 		return raddr_a;
-	} else if (add_a == QPU_MUX_B && sig != QPU_SIG_SMALL_IMM) {
+	else if (add_a == QPU_MUX_B && sig != QPU_SIG_SMALL_IMM)
 		return 32 + raddr_b;
-	} else if (add_a <= QPU_MUX_R3) {
+	else if (add_a <= QPU_MUX_R3)
 		return 64 + add_a;
-	} else {
+	else
 		return ~0;
-	}
 }
 
 static bool
@@ -111,9 +101,9 @@ is_tmu_write(uint32_t waddr)
 }
 
 static bool
-record_validated_texture_sample(struct vc4_validated_shader_info *validated_shader,
-				struct vc4_shader_validation_state *validation_state,
-				int tmu)
+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;
@@ -226,8 +216,8 @@ check_tmu_write(uint64_t inst,
 		validated_shader->uniforms_size += 4;
 
 	if (submit) {
-		if (!record_validated_texture_sample(validated_shader,
-						     validation_state, tmu)) {
+		if (!record_texture_sample(validated_shader,
+					   validation_state, tmu)) {
 			return false;
 		}
 
@@ -238,10 +228,10 @@ check_tmu_write(uint64_t inst,
 }
 
 static bool
-check_register_write(uint64_t inst,
-		     struct vc4_validated_shader_info *validated_shader,
-		     struct vc4_shader_validation_state *validation_state,
-		     bool is_mul)
+check_reg_write(uint64_t inst,
+		struct vc4_validated_shader_info *validated_shader,
+		struct vc4_shader_validation_state *validation_state,
+		bool is_mul)
 {
 	uint32_t waddr = (is_mul ?
 			  QPU_GET_FIELD(inst, QPU_WADDR_MUL) :
@@ -297,7 +287,7 @@ check_register_write(uint64_t inst,
 		return true;
 
 	case QPU_W_TLB_STENCIL_SETUP:
-                return true;
+		return true;
 	}
 
 	return true;
@@ -360,7 +350,7 @@ track_live_clamps(uint64_t inst,
 		}
 
 		validation_state->live_max_clamp_regs[lri_add] = true;
-	} if (op_add == QPU_A_MIN) {
+	} 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.
 		 */
@@ -392,8 +382,10 @@ check_instruction_writes(uint64_t inst,
 		return false;
 	}
 
-	ok = (check_register_write(inst, validated_shader, validation_state, false) &&
-	      check_register_write(inst, validated_shader, validation_state, true));
+	ok = (check_reg_write(inst, validated_shader, validation_state,
+			      false) &&
+	      check_reg_write(inst, validated_shader, validation_state,
+			      true));
 
 	track_live_clamps(inst, validated_shader, validation_state);
 
@@ -441,7 +433,7 @@ vc4_validate_shader(struct drm_gem_cma_o
 	shader = shader_obj->vaddr;
 	max_ip = shader_obj->base.size / sizeof(uint64_t);
 
-	validated_shader = kcalloc(sizeof(*validated_shader), 1, GFP_KERNEL);
+	validated_shader = kcalloc(1, sizeof(*validated_shader), GFP_KERNEL);
 	if (!validated_shader)
 		return NULL;
 
@@ -497,7 +489,7 @@ vc4_validate_shader(struct drm_gem_cma_o
 
 	if (ip == max_ip) {
 		DRM_ERROR("shader failed to terminate before "
-			  "shader BO end at %d\n",
+			  "shader BO end at %zd\n",
 			  shader_obj->base.size);
 		goto fail;
 	}
--- a/include/drm/drmP.h
+++ b/include/drm/drmP.h
@@ -585,6 +585,13 @@ struct drm_driver {
 	int (*gem_open_object) (struct drm_gem_object *, struct drm_file *);
 	void (*gem_close_object) (struct drm_gem_object *, struct drm_file *);
 
+	/**
+	 * Hook for allocating the GEM object struct, for use by core
+	 * helpers.
+	 */
+	struct drm_gem_object *(*gem_create_object)(struct drm_device *dev,
+						    size_t size);
+
 	/* prime: */
 	/* export handle -> fd (see drm_gem_prime_handle_to_fd() helper) */
 	int (*prime_handle_to_fd)(struct drm_device *dev, struct drm_file *file_priv,
@@ -639,7 +646,6 @@ struct drm_driver {
 
 	u32 driver_features;
 	int dev_priv_size;
-	size_t gem_obj_size;
 	const struct drm_ioctl_desc *ioctls;
 	int num_ioctls;
 	const struct file_operations *fops;