1 files changed, 423 insertions, 446 deletions

diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c
index 44f84bfd0c0e..75d9f3574370 100644
--- a/drivers/gpu/drm/msm/msm_gem_submit.c
+++ b/drivers/gpu/drm/msm/msm_gem_submit.c
@@ -4,6 +4,7 @@
  * Author: Rob Clark <robdclark@gmail.com>
  */
 
+#include <linux/dma-fence-unwrap.h>
 #include <linux/file.h>
 #include <linux/sync_file.h>
 #include <linux/uaccess.h>
@@ -16,48 +17,67 @@
 #include "msm_gpu.h"
 #include "msm_gem.h"
 #include "msm_gpu_trace.h"
+#include "msm_syncobj.h"
+
+/* For userspace errors, use DRM_UT_DRIVER.. so that userspace can enable
+ * error msgs for debugging, but we don't spam dmesg by default
+ */
+#define SUBMIT_ERROR(err, submit, fmt, ...) \
+	UERR(err, (submit)->dev, fmt, ##__VA_ARGS__)
 
 /*
  * Cmdstream submission:
  */
 
-/* make sure these don't conflict w/ MSM_SUBMIT_BO_x */
-#define BO_VALID    0x8000   /* is current addr in cmdstream correct/valid? */
-#define BO_LOCKED   0x4000
-#define BO_PINNED   0x2000
-
 static struct msm_gem_submit *submit_create(struct drm_device *dev,
 		struct msm_gpu *gpu,
 		struct msm_gpu_submitqueue *queue, uint32_t nr_bos,
-		uint32_t nr_cmds)
+		uint32_t nr_cmds, u64 drm_client_id)
 {
+	static atomic_t ident = ATOMIC_INIT(0);
 	struct msm_gem_submit *submit;
-	uint64_t sz = struct_size(submit, bos, nr_bos) +
-				  ((u64)nr_cmds * sizeof(submit->cmd[0]));
+	uint64_t sz;
+	int ret;
+
+	sz = struct_size(submit, bos, nr_bos) +
+			((u64)nr_cmds * sizeof(submit->cmd[0]));
 
 	if (sz > SIZE_MAX)
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 
-	submit = kmalloc(sz, GFP_KERNEL | __GFP_NOWARN | __GFP_NORETRY);
+	submit = kzalloc(sz, GFP_KERNEL | __GFP_NOWARN);
 	if (!submit)
-		return NULL;
+		return ERR_PTR(-ENOMEM);
+
+	submit->hw_fence = msm_fence_alloc();
+	if (IS_ERR(submit->hw_fence)) {
+		ret = PTR_ERR(submit->hw_fence);
+		kfree(submit);
+		return ERR_PTR(ret);
+	}
+
+	ret = drm_sched_job_init(&submit->base, queue->entity, 1, queue,
+				 drm_client_id);
+	if (ret) {
+		kfree(submit->hw_fence);
+		kfree(submit);
+		return ERR_PTR(ret);
+	}
 
 	kref_init(&submit->ref);
 	submit->dev = dev;
-	submit->aspace = queue->ctx->aspace;
+	submit->vm = msm_context_vm(dev, queue->ctx);
 	submit->gpu = gpu;
-	submit->fence = NULL;
 	submit->cmd = (void *)&submit->bos[nr_bos];
 	submit->queue = queue;
-	submit->ring = gpu->rb[queue->prio];
+	submit->pid = get_pid(task_pid(current));
+	submit->ring = gpu->rb[queue->ring_nr];
 	submit->fault_dumped = false;
 
-	/* initially, until copy_from_user() and bo lookup succeeds: */
-	submit->nr_bos = 0;
-	submit->nr_cmds = 0;
+	/* Get a unique identifier for the submission for logging purposes */
+	submit->ident = atomic_inc_return(&ident) - 1;
 
 	INIT_LIST_HEAD(&submit->node);
-	INIT_LIST_HEAD(&submit->bo_list);
 
 	return submit;
 }
@@ -68,7 +88,36 @@ void __msm_gem_submit_destroy(struct kref *kref)
 			container_of(kref, struct msm_gem_submit, ref);
 	unsigned i;
 
-	dma_fence_put(submit->fence);
+	/*
+	 * In error paths, we could unref the submit without calling
+	 * drm_sched_entity_push_job(), so msm_job_free() will never
+	 * get called.  Since drm_sched_job_cleanup() will NULL out
+	 * s_fence, we can use that to detect this case.
+	 */
+	if (submit->base.s_fence)
+		drm_sched_job_cleanup(&submit->base);
+
+	if (submit->fence_id) {
+		spin_lock(&submit->queue->idr_lock);
+		idr_remove(&submit->queue->fence_idr, submit->fence_id);
+		spin_unlock(&submit->queue->idr_lock);
+	}
+
+	dma_fence_put(submit->user_fence);
+
+	/*
+	 * If the submit is freed before msm_job_run(), then hw_fence is
+	 * just some pre-allocated memory, not a reference counted fence.
+	 * Once the job runs and the hw_fence is initialized, it will
+	 * have a refcount of at least one, since the submit holds a ref
+	 * to the hw_fence.
+	 */
+	if (kref_read(&submit->hw_fence->refcount) == 0) {
+		kfree(submit->hw_fence);
+	} else {
+		dma_fence_put(submit->hw_fence);
+	}
+
 	put_pid(submit->pid);
 	msm_submitqueue_put(submit->queue);
 
@@ -105,48 +154,32 @@ static int submit_lookup_objects(struct msm_gem_submit *submit,
 
 		if ((submit_bo.flags & ~MSM_SUBMIT_BO_FLAGS) ||
 			!(submit_bo.flags & MANDATORY_FLAGS)) {
-			DRM_ERROR("invalid flags: %x\n", submit_bo.flags);
-			ret = -EINVAL;
+			ret = SUBMIT_ERROR(EINVAL, submit, "invalid flags: %x\n", submit_bo.flags);
 			i = 0;
 			goto out;
 		}
 
 		submit->bos[i].handle = submit_bo.handle;
 		submit->bos[i].flags = submit_bo.flags;
-		/* in validate_objects() we figure out if this is true: */
-		submit->bos[i].iova  = submit_bo.presumed;
 	}
 
 	spin_lock(&file->table_lock);
 
 	for (i = 0; i < args->nr_bos; i++) {
 		struct drm_gem_object *obj;
-		struct msm_gem_object *msm_obj;
 
 		/* normally use drm_gem_object_lookup(), but for bulk lookup
 		 * all under single table_lock just hit object_idr directly:
 		 */
 		obj = idr_find(&file->object_idr, submit->bos[i].handle);
 		if (!obj) {
-			DRM_ERROR("invalid handle %u at index %u\n", submit->bos[i].handle, i);
-			ret = -EINVAL;
-			goto out_unlock;
-		}
-
-		msm_obj = to_msm_bo(obj);
-
-		if (!list_empty(&msm_obj->submit_entry)) {
-			DRM_ERROR("handle %u at index %u already on submit list\n",
-					submit->bos[i].handle, i);
-			ret = -EINVAL;
+			ret = SUBMIT_ERROR(EINVAL, submit, "invalid handle %u at index %u\n", submit->bos[i].handle, i);
 			goto out_unlock;
 		}
 
 		drm_gem_object_get(obj);
 
-		submit->bos[i].obj = msm_obj;
-
-		list_add_tail(&msm_obj->submit_entry, &submit->bo_list);
+		submit->bos[i].obj = obj;
 	}
 
 out_unlock:
@@ -161,7 +194,9 @@ out:
 static int submit_lookup_cmds(struct msm_gem_submit *submit,
 		struct drm_msm_gem_submit *args, struct drm_file *file)
 {
-	unsigned i, sz;
+	struct msm_context *ctx = file->driver_priv;
+	unsigned i;
+	size_t sz;
 	int ret = 0;
 
 	for (i = 0; i < args->nr_cmds; i++) {
@@ -182,17 +217,29 @@ static int submit_lookup_cmds(struct msm_gem_submit *submit,
 		case MSM_SUBMIT_CMD_CTX_RESTORE_BUF:
 			break;
 		default:
-			DRM_ERROR("invalid type: %08x\n", submit_cmd.type);
-			return -EINVAL;
+			return SUBMIT_ERROR(EINVAL, submit, "invalid type: %08x\n", submit_cmd.type);
 		}
 
 		if (submit_cmd.size % 4) {
-			DRM_ERROR("non-aligned cmdstream buffer size: %u\n",
-					submit_cmd.size);
-			ret = -EINVAL;
+			ret = SUBMIT_ERROR(EINVAL, submit, "non-aligned cmdstream buffer size: %u\n",
+					   submit_cmd.size);
 			goto out;
 		}
 
+		if (msm_context_is_vmbind(ctx)) {
+			if (submit_cmd.nr_relocs) {
+				ret = SUBMIT_ERROR(EINVAL, submit, "nr_relocs must be zero");
+				goto out;
+			}
+
+			if (submit_cmd.submit_idx || submit_cmd.submit_offset) {
+				ret = SUBMIT_ERROR(EINVAL, submit, "submit_idx/offset must be zero");
+				goto out;
+			}
+
+			submit->cmd[i].iova = submit_cmd.iova;
+		}
+
 		submit->cmd[i].type = submit_cmd.type;
 		submit->cmd[i].size = submit_cmd.size / 4;
 		submit->cmd[i].offset = submit_cmd.submit_offset / 4;
@@ -208,7 +255,11 @@ static int submit_lookup_cmds(struct msm_gem_submit *submit,
 			ret = -ENOMEM;
 			goto out;
 		}
-		submit->cmd[i].relocs = kmalloc(sz, GFP_KERNEL);
+		submit->cmd[i].relocs = kmalloc(sz, GFP_KERNEL | __GFP_NOWARN);
+		if (!submit->cmd[i].relocs) {
+			ret = -ENOMEM;
+			goto out;
+		}
 		ret = copy_from_user(submit->cmd[i].relocs, userptr, sz);
 		if (ret) {
 			ret = -EFAULT;
@@ -220,97 +271,76 @@ out:
 	return ret;
 }
 
-static void submit_unlock_unpin_bo(struct msm_gem_submit *submit,
-		int i, bool backoff)
+static int submit_lock_objects_vmbind(struct msm_gem_submit *submit)
 {
-	struct msm_gem_object *msm_obj = submit->bos[i].obj;
+	unsigned flags = DRM_EXEC_INTERRUPTIBLE_WAIT | DRM_EXEC_IGNORE_DUPLICATES;
+	struct drm_exec *exec = &submit->exec;
+	int ret = 0;
 
-	if (submit->bos[i].flags & BO_PINNED)
-		msm_gem_unpin_iova_locked(&msm_obj->base, submit->aspace);
+	drm_exec_init(&submit->exec, flags, submit->nr_bos);
 
-	if (submit->bos[i].flags & BO_LOCKED)
-		dma_resv_unlock(msm_obj->base.resv);
+	drm_exec_until_all_locked (&submit->exec) {
+		ret = drm_gpuvm_prepare_vm(submit->vm, exec, 1);
+		drm_exec_retry_on_contention(exec);
+		if (ret)
+			break;
 
-	if (backoff && !(submit->bos[i].flags & BO_VALID))
-		submit->bos[i].iova = 0;
+		ret = drm_gpuvm_prepare_objects(submit->vm, exec, 1);
+		drm_exec_retry_on_contention(exec);
+		if (ret)
+			break;
+	}
 
-	submit->bos[i].flags &= ~(BO_LOCKED | BO_PINNED);
+	return ret;
 }
 
 /* This is where we make sure all the bo's are reserved and pin'd: */
 static int submit_lock_objects(struct msm_gem_submit *submit)
 {
-	int contended, slow_locked = -1, i, ret = 0;
-
-retry:
-	for (i = 0; i < submit->nr_bos; i++) {
-		struct msm_gem_object *msm_obj = submit->bos[i].obj;
+	unsigned flags = DRM_EXEC_INTERRUPTIBLE_WAIT;
+	int ret = 0;
 
-		if (slow_locked == i)
-			slow_locked = -1;
+	if (msm_context_is_vmbind(submit->queue->ctx))
+		return submit_lock_objects_vmbind(submit);
 
-		contended = i;
+	drm_exec_init(&submit->exec, flags, submit->nr_bos);
 
-		if (!(submit->bos[i].flags & BO_LOCKED)) {
-			ret = dma_resv_lock_interruptible(msm_obj->base.resv,
-							  &submit->ticket);
+	drm_exec_until_all_locked (&submit->exec) {
+		ret = drm_exec_lock_obj(&submit->exec,
+					drm_gpuvm_resv_obj(submit->vm));
+		drm_exec_retry_on_contention(&submit->exec);
+		if (ret)
+			break;
+		for (unsigned i = 0; i < submit->nr_bos; i++) {
+			struct drm_gem_object *obj = submit->bos[i].obj;
+			ret = drm_exec_prepare_obj(&submit->exec, obj, 1);
+			drm_exec_retry_on_contention(&submit->exec);
 			if (ret)
-				goto fail;
-			submit->bos[i].flags |= BO_LOCKED;
-		}
-	}
-
-	ww_acquire_done(&submit->ticket);
-
-	return 0;
-
-fail:
-	for (; i >= 0; i--)
-		submit_unlock_unpin_bo(submit, i, true);
-
-	if (slow_locked > 0)
-		submit_unlock_unpin_bo(submit, slow_locked, true);
-
-	if (ret == -EDEADLK) {
-		struct msm_gem_object *msm_obj = submit->bos[contended].obj;
-		/* we lost out in a seqno race, lock and retry.. */
-		ret = dma_resv_lock_slow_interruptible(msm_obj->base.resv,
-						       &submit->ticket);
-		if (!ret) {
-			submit->bos[contended].flags |= BO_LOCKED;
-			slow_locked = contended;
-			goto retry;
+				break;
 		}
 	}
 
 	return ret;
 }
 
-static int submit_fence_sync(struct msm_gem_submit *submit, bool no_implicit)
+static int submit_fence_sync(struct msm_gem_submit *submit)
 {
 	int i, ret = 0;
 
 	for (i = 0; i < submit->nr_bos; i++) {
-		struct msm_gem_object *msm_obj = submit->bos[i].obj;
+		struct drm_gem_object *obj = submit->bos[i].obj;
 		bool write = submit->bos[i].flags & MSM_SUBMIT_BO_WRITE;
 
-		if (!write) {
-			/* NOTE: _reserve_shared() must happen before
-			 * _add_shared_fence(), which makes this a slightly
-			 * strange place to call it.  OTOH this is a
-			 * convenient can-fail point to hook it in.
-			 */
-			ret = dma_resv_reserve_shared(msm_obj->base.resv,
-								1);
-			if (ret)
-				return ret;
-		}
-
-		if (no_implicit)
+		/* Otherwise userspace can ask for implicit sync to be
+		 * disabled on specific buffers.  This is useful for internal
+		 * usermode driver managed buffers, suballocation, etc.
+		 */
+		if (submit->bos[i].flags & MSM_SUBMIT_BO_NO_IMPLICIT)
 			continue;
 
-		ret = msm_gem_sync_object(&msm_obj->base, submit->ring->fctx,
-			write);
+		ret = drm_sched_job_add_implicit_dependencies(&submit->base,
+							      obj,
+							      write);
 		if (ret)
 			break;
 	}
@@ -320,75 +350,121 @@ static int submit_fence_sync(struct msm_gem_submit *submit, bool no_implicit)
 
 static int submit_pin_objects(struct msm_gem_submit *submit)
 {
+	struct msm_drm_private *priv = submit->dev->dev_private;
 	int i, ret = 0;
 
-	submit->valid = true;
-
 	for (i = 0; i < submit->nr_bos; i++) {
-		struct msm_gem_object *msm_obj = submit->bos[i].obj;
-		uint64_t iova;
+		struct drm_gem_object *obj = submit->bos[i].obj;
+		struct drm_gpuva *vma;
 
 		/* if locking succeeded, pin bo: */
-		ret = msm_gem_get_and_pin_iova_locked(&msm_obj->base,
-				submit->aspace, &iova);
+		vma = msm_gem_get_vma_locked(obj, submit->vm);
+		if (IS_ERR(vma)) {
+			ret = PTR_ERR(vma);
+			break;
+		}
 
+		ret = msm_gem_pin_vma_locked(obj, vma);
 		if (ret)
 			break;
 
-		submit->bos[i].flags |= BO_PINNED;
+		submit->bos[i].vm_bo = drm_gpuvm_bo_get(vma->vm_bo);
+		submit->bos[i].iova = vma->va.addr;
+	}
 
-		if (iova == submit->bos[i].iova) {
-			submit->bos[i].flags |= BO_VALID;
-		} else {
-			submit->bos[i].iova = iova;
-			/* iova changed, so address in cmdstream is not valid: */
-			submit->bos[i].flags &= ~BO_VALID;
-			submit->valid = false;
-		}
+	/*
+	 * A second loop while holding the LRU lock (a) avoids acquiring/dropping
+	 * the LRU lock for each individual bo, while (b) avoiding holding the
+	 * LRU lock while calling msm_gem_pin_vma_locked() (which could trigger
+	 * get_pages() which could trigger reclaim.. and if we held the LRU lock
+	 * could trigger deadlock with the shrinker).
+	 */
+	mutex_lock(&priv->lru.lock);
+	for (i = 0; i < submit->nr_bos; i++) {
+		msm_gem_pin_obj_locked(submit->bos[i].obj);
 	}
+	mutex_unlock(&priv->lru.lock);
+
+	submit->bos_pinned = true;
 
 	return ret;
 }
 
+static void submit_unpin_objects(struct msm_gem_submit *submit)
+{
+	if (!submit->bos_pinned)
+		return;
+
+	for (int i = 0; i < submit->nr_bos; i++) {
+		struct drm_gem_object *obj = submit->bos[i].obj;
+
+		msm_gem_unpin_locked(obj);
+	}
+
+	submit->bos_pinned = false;
+}
+
+static void submit_attach_object_fences(struct msm_gem_submit *submit)
+{
+	struct msm_gem_vm *vm = to_msm_vm(submit->vm);
+	struct dma_fence *last_fence;
+
+	if (msm_context_is_vmbind(submit->queue->ctx)) {
+		drm_gpuvm_resv_add_fence(submit->vm, &submit->exec,
+					 submit->user_fence,
+					 DMA_RESV_USAGE_BOOKKEEP,
+					 DMA_RESV_USAGE_BOOKKEEP);
+
+		last_fence = vm->last_fence;
+		vm->last_fence = dma_fence_unwrap_merge(submit->user_fence, last_fence);
+		dma_fence_put(last_fence);
+
+		return;
+	}
+
+	for (unsigned i = 0; i < submit->nr_bos; i++) {
+		struct drm_gem_object *obj = submit->bos[i].obj;
+
+		if (submit->bos[i].flags & MSM_SUBMIT_BO_WRITE)
+			dma_resv_add_fence(obj->resv, submit->user_fence,
+					   DMA_RESV_USAGE_WRITE);
+		else if (submit->bos[i].flags & MSM_SUBMIT_BO_READ)
+			dma_resv_add_fence(obj->resv, submit->user_fence,
+					   DMA_RESV_USAGE_READ);
+	}
+}
+
 static int submit_bo(struct msm_gem_submit *submit, uint32_t idx,
-		struct msm_gem_object **obj, uint64_t *iova, bool *valid)
+		struct drm_gem_object **obj, uint64_t *iova)
 {
 	if (idx >= submit->nr_bos) {
-		DRM_ERROR("invalid buffer index: %u (out of %u)\n",
-				idx, submit->nr_bos);
-		return -EINVAL;
+		return SUBMIT_ERROR(EINVAL, submit, "invalid buffer index: %u (out of %u)\n",
+				    idx, submit->nr_bos);
 	}
 
 	if (obj)
 		*obj = submit->bos[idx].obj;
 	if (iova)
 		*iova = submit->bos[idx].iova;
-	if (valid)
-		*valid = !!(submit->bos[idx].flags & BO_VALID);
 
 	return 0;
 }
 
 /* process the reloc's and patch up the cmdstream as needed: */
-static int submit_reloc(struct msm_gem_submit *submit, struct msm_gem_object *obj,
+static int submit_reloc(struct msm_gem_submit *submit, struct drm_gem_object *obj,
 		uint32_t offset, uint32_t nr_relocs, struct drm_msm_gem_submit_reloc *relocs)
 {
 	uint32_t i, last_offset = 0;
 	uint32_t *ptr;
 	int ret = 0;
 
-	if (!nr_relocs)
-		return 0;
-
-	if (offset % 4) {
-		DRM_ERROR("non-aligned cmdstream buffer: %u\n", offset);
-		return -EINVAL;
-	}
+	if (offset % 4)
+		return SUBMIT_ERROR(EINVAL, submit, "non-aligned cmdstream buffer: %u\n", offset);
 
 	/* For now, just map the entire thing.  Eventually we probably
 	 * to do it page-by-page, w/ kmap() if not vmap()d..
 	 */
-	ptr = msm_gem_get_vaddr_locked(&obj->base);
+	ptr = msm_gem_get_vaddr_locked(obj);
 
 	if (IS_ERR(ptr)) {
 		ret = PTR_ERR(ptr);
@@ -400,32 +476,26 @@ static int submit_reloc(struct msm_gem_submit *submit, struct msm_gem_object *ob
 		struct drm_msm_gem_submit_reloc submit_reloc = relocs[i];
 		uint32_t off;
 		uint64_t iova;
-		bool valid;
 
 		if (submit_reloc.submit_offset % 4) {
-			DRM_ERROR("non-aligned reloc offset: %u\n",
-					submit_reloc.submit_offset);
-			ret = -EINVAL;
+			ret = SUBMIT_ERROR(EINVAL, submit, "non-aligned reloc offset: %u\n",
+					   submit_reloc.submit_offset);
 			goto out;
 		}
 
 		/* offset in dwords: */
 		off = submit_reloc.submit_offset / 4;
 
-		if ((off >= (obj->base.size / 4)) ||
+		if ((off >= (obj->size / 4)) ||
 				(off < last_offset)) {
-			DRM_ERROR("invalid offset %u at reloc %u\n", off, i);
-			ret = -EINVAL;
+			ret = SUBMIT_ERROR(EINVAL, submit, "invalid offset %u at reloc %u\n", off, i);
 			goto out;
 		}
 
-		ret = submit_bo(submit, submit_reloc.reloc_idx, NULL, &iova, &valid);
+		ret = submit_bo(submit, submit_reloc.reloc_idx, NULL, &iova);
 		if (ret)
 			goto out;
 
-		if (valid)
-			continue;
-
 		iova += submit_reloc.reloc_offset;
 
 		if (submit_reloc.shift < 0)
@@ -439,236 +509,78 @@ static int submit_reloc(struct msm_gem_submit *submit, struct msm_gem_object *ob
 	}
 
 out:
-	msm_gem_put_vaddr_locked(&obj->base);
+	msm_gem_put_vaddr_locked(obj);
 
 	return ret;
 }
 
-static void submit_cleanup(struct msm_gem_submit *submit)
-{
-	unsigned i;
-
-	for (i = 0; i < submit->nr_bos; i++) {
-		struct msm_gem_object *msm_obj = submit->bos[i].obj;
-		submit_unlock_unpin_bo(submit, i, false);
-		list_del_init(&msm_obj->submit_entry);
-		drm_gem_object_put_locked(&msm_obj->base);
-	}
-}
-
-
-struct msm_submit_post_dep {
-	struct drm_syncobj *syncobj;
-	uint64_t point;
-	struct dma_fence_chain *chain;
-};
-
-static struct drm_syncobj **msm_wait_deps(struct drm_device *dev,
-                                          struct drm_file *file,
-                                          uint64_t in_syncobjs_addr,
-                                          uint32_t nr_in_syncobjs,
-                                          size_t syncobj_stride,
-                                          struct msm_ringbuffer *ring)
+/* Cleanup submit at end of ioctl.  In the error case, this also drops
+ * references, unpins, and drops active refcnt.  In the non-error case,
+ * this is done when the submit is retired.
+ */
+static void submit_cleanup(struct msm_gem_submit *submit, bool error)
 {
-	struct drm_syncobj **syncobjs = NULL;
-	struct drm_msm_gem_submit_syncobj syncobj_desc = {0};
-	int ret = 0;
-	uint32_t i, j;
+	if (error)
+		submit_unpin_objects(submit);
 
-	syncobjs = kcalloc(nr_in_syncobjs, sizeof(*syncobjs),
-	                   GFP_KERNEL | __GFP_NOWARN | __GFP_NORETRY);
-	if (!syncobjs)
-		return ERR_PTR(-ENOMEM);
-
-	for (i = 0; i < nr_in_syncobjs; ++i) {
-		uint64_t address = in_syncobjs_addr + i * syncobj_stride;
-		struct dma_fence *fence;
-
-		if (copy_from_user(&syncobj_desc,
-			           u64_to_user_ptr(address),
-			           min(syncobj_stride, sizeof(syncobj_desc)))) {
-			ret = -EFAULT;
-			break;
-		}
-
-		if (syncobj_desc.point &&
-		    !drm_core_check_feature(dev, DRIVER_SYNCOBJ_TIMELINE)) {
-			ret = -EOPNOTSUPP;
-			break;
-		}
-
-		if (syncobj_desc.flags & ~MSM_SUBMIT_SYNCOBJ_FLAGS) {
-			ret = -EINVAL;
-			break;
-		}
-
-		ret = drm_syncobj_find_fence(file, syncobj_desc.handle,
-		                             syncobj_desc.point, 0, &fence);
-		if (ret)
-			break;
-
-		if (!dma_fence_match_context(fence, ring->fctx->context))
-			ret = dma_fence_wait(fence, true);
-
-		dma_fence_put(fence);
-		if (ret)
-			break;
-
-		if (syncobj_desc.flags & MSM_SUBMIT_SYNCOBJ_RESET) {
-			syncobjs[i] =
-				drm_syncobj_find(file, syncobj_desc.handle);
-			if (!syncobjs[i]) {
-				ret = -EINVAL;
-				break;
-			}
-		}
-	}
-
-	if (ret) {
-		for (j = 0; j <= i; ++j) {
-			if (syncobjs[j])
-				drm_syncobj_put(syncobjs[j]);
-		}
-		kfree(syncobjs);
-		return ERR_PTR(ret);
-	}
-	return syncobjs;
-}
-
-static void msm_reset_syncobjs(struct drm_syncobj **syncobjs,
-                               uint32_t nr_syncobjs)
-{
-	uint32_t i;
+	if (submit->exec.objects)
+		drm_exec_fini(&submit->exec);
 
-	for (i = 0; syncobjs && i < nr_syncobjs; ++i) {
-		if (syncobjs[i])
-			drm_syncobj_replace_fence(syncobjs[i], NULL);
-	}
+	/* if job wasn't enqueued to scheduler, early retirement: */
+	if (error)
+		msm_submit_retire(submit);
 }
 
-static struct msm_submit_post_dep *msm_parse_post_deps(struct drm_device *dev,
-                                                       struct drm_file *file,
-                                                       uint64_t syncobjs_addr,
-                                                       uint32_t nr_syncobjs,
-                                                       size_t syncobj_stride)
+void msm_submit_retire(struct msm_gem_submit *submit)
 {
-	struct msm_submit_post_dep *post_deps;
-	struct drm_msm_gem_submit_syncobj syncobj_desc = {0};
-	int ret = 0;
-	uint32_t i, j;
-
-	post_deps = kmalloc_array(nr_syncobjs, sizeof(*post_deps),
-	                          GFP_KERNEL | __GFP_NOWARN | __GFP_NORETRY);
-	if (!post_deps)
-		return ERR_PTR(-ENOMEM);
+	int i;
 
-	for (i = 0; i < nr_syncobjs; ++i) {
-		uint64_t address = syncobjs_addr + i * syncobj_stride;
-
-		if (copy_from_user(&syncobj_desc,
-			           u64_to_user_ptr(address),
-			           min(syncobj_stride, sizeof(syncobj_desc)))) {
-			ret = -EFAULT;
-			break;
-		}
-
-		post_deps[i].point = syncobj_desc.point;
-		post_deps[i].chain = NULL;
-
-		if (syncobj_desc.flags) {
-			ret = -EINVAL;
-			break;
-		}
-
-		if (syncobj_desc.point) {
-			if (!drm_core_check_feature(dev,
-			                            DRIVER_SYNCOBJ_TIMELINE)) {
-				ret = -EOPNOTSUPP;
-				break;
-			}
-
-			post_deps[i].chain =
-				kmalloc(sizeof(*post_deps[i].chain),
-				        GFP_KERNEL);
-			if (!post_deps[i].chain) {
-				ret = -ENOMEM;
-				break;
-			}
-		}
-
-		post_deps[i].syncobj =
-			drm_syncobj_find(file, syncobj_desc.handle);
-		if (!post_deps[i].syncobj) {
-			ret = -EINVAL;
-			break;
-		}
-	}
-
-	if (ret) {
-		for (j = 0; j <= i; ++j) {
-			kfree(post_deps[j].chain);
-			if (post_deps[j].syncobj)
-				drm_syncobj_put(post_deps[j].syncobj);
-		}
-
-		kfree(post_deps);
-		return ERR_PTR(ret);
-	}
-
-	return post_deps;
-}
+	for (i = 0; i < submit->nr_bos; i++) {
+		struct drm_gem_object *obj = submit->bos[i].obj;
+		struct drm_gpuvm_bo *vm_bo = submit->bos[i].vm_bo;
 
-static void msm_process_post_deps(struct msm_submit_post_dep *post_deps,
-                                  uint32_t count, struct dma_fence *fence)
-{
-	uint32_t i;
-
-	for (i = 0; post_deps && i < count; ++i) {
-		if (post_deps[i].chain) {
-			drm_syncobj_add_point(post_deps[i].syncobj,
-			                      post_deps[i].chain,
-			                      fence, post_deps[i].point);
-			post_deps[i].chain = NULL;
-		} else {
-			drm_syncobj_replace_fence(post_deps[i].syncobj,
-			                          fence);
-		}
+		msm_gem_lock(obj);
+		drm_gpuvm_bo_put(vm_bo);
+		msm_gem_unlock(obj);
+		drm_gem_object_put(obj);
 	}
 }
 
 int msm_ioctl_gem_submit(struct drm_device *dev, void *data,
 		struct drm_file *file)
 {
-	static atomic_t ident = ATOMIC_INIT(0);
 	struct msm_drm_private *priv = dev->dev_private;
 	struct drm_msm_gem_submit *args = data;
-	struct msm_file_private *ctx = file->driver_priv;
-	struct msm_gem_submit *submit;
+	struct msm_context *ctx = file->driver_priv;
+	struct msm_gem_submit *submit = NULL;
 	struct msm_gpu *gpu = priv->gpu;
-	struct sync_file *sync_file = NULL;
 	struct msm_gpu_submitqueue *queue;
 	struct msm_ringbuffer *ring;
-	struct msm_submit_post_dep *post_deps = NULL;
+	struct msm_syncobj_post_dep *post_deps = NULL;
 	struct drm_syncobj **syncobjs_to_reset = NULL;
+	struct sync_file *sync_file = NULL;
+	unsigned cmds_to_parse;
 	int out_fence_fd = -1;
-	struct pid *pid = get_pid(task_pid(current));
-	bool has_ww_ticket = false;
 	unsigned i;
-	int ret, submitid;
+	int ret;
+
 	if (!gpu)
 		return -ENXIO;
 
 	if (args->pad)
 		return -EINVAL;
 
+	if (to_msm_vm(ctx->vm)->unusable)
+		return UERR(EPIPE, dev, "context is unusable");
+
 	/* for now, we just have 3d pipe.. eventually this would need to
 	 * be more clever to dispatch to appropriate gpu module:
 	 */
 	if (MSM_PIPE_ID(args->flags) != MSM_PIPE_3D0)
-		return -EINVAL;
+		return UERR(EINVAL, dev, "invalid pipe");
 
 	if (MSM_PIPE_FLAGS(args->flags) & ~MSM_SUBMIT_FLAGS)
-		return -EINVAL;
+		return UERR(EINVAL, dev, "invalid flags");
 
 	if (args->flags & MSM_SUBMIT_SUDO) {
 		if (!IS_ENABLED(CONFIG_DRM_MSM_GPU_SUDO) ||
@@ -680,183 +592,248 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data,
 	if (!queue)
 		return -ENOENT;
 
-	/* Get a unique identifier for the submission for logging purposes */
-	submitid = atomic_inc_return(&ident) - 1;
+	if (queue->flags & MSM_SUBMITQUEUE_VM_BIND) {
+		ret = UERR(EINVAL, dev, "Invalid queue type");
+		goto out_post_unlock;
+	}
+
+	ring = gpu->rb[queue->ring_nr];
+
+	if (args->flags & MSM_SUBMIT_FENCE_FD_OUT) {
+		out_fence_fd = get_unused_fd_flags(O_CLOEXEC);
+		if (out_fence_fd < 0) {
+			ret = out_fence_fd;
+			goto out_post_unlock;
+		}
+	}
+
+	submit = submit_create(dev, gpu, queue, args->nr_bos, args->nr_cmds,
+			       file->client_id);
+	if (IS_ERR(submit)) {
+		ret = PTR_ERR(submit);
+		goto out_post_unlock;
+	}
 
-	ring = gpu->rb[queue->prio];
-	trace_msm_gpu_submit(pid_nr(pid), ring->id, submitid,
+	trace_msm_gpu_submit(pid_nr(submit->pid), ring->id, submit->ident,
 		args->nr_bos, args->nr_cmds);
 
+	ret = mutex_lock_interruptible(&queue->lock);
+	if (ret)
+		goto out_post_unlock;
+
+	if (args->flags & MSM_SUBMIT_SUDO)
+		submit->in_rb = true;
+
 	if (args->flags & MSM_SUBMIT_FENCE_FD_IN) {
 		struct dma_fence *in_fence;
 
 		in_fence = sync_file_get_fence(args->fence_fd);
 
-		if (!in_fence)
-			return -EINVAL;
-
-		/*
-		 * Wait if the fence is from a foreign context, or if the fence
-		 * array contains any fence from a foreign context.
-		 */
-		ret = 0;
-		if (!dma_fence_match_context(in_fence, ring->fctx->context))
-			ret = dma_fence_wait(in_fence, true);
+		if (!in_fence) {
+			ret = UERR(EINVAL, dev, "invalid in-fence");
+			goto out_unlock;
+		}
 
-		dma_fence_put(in_fence);
+		ret = drm_sched_job_add_dependency(&submit->base, in_fence);
 		if (ret)
-			return ret;
+			goto out_unlock;
 	}
 
 	if (args->flags & MSM_SUBMIT_SYNCOBJ_IN) {
-		syncobjs_to_reset = msm_wait_deps(dev, file,
-		                                  args->in_syncobjs,
-		                                  args->nr_in_syncobjs,
-		                                  args->syncobj_stride, ring);
-		if (IS_ERR(syncobjs_to_reset))
-			return PTR_ERR(syncobjs_to_reset);
+		syncobjs_to_reset = msm_syncobj_parse_deps(dev, &submit->base,
+							   file, args->in_syncobjs,
+							   args->nr_in_syncobjs,
+							   args->syncobj_stride);
+		if (IS_ERR(syncobjs_to_reset)) {
+			ret = PTR_ERR(syncobjs_to_reset);
+			goto out_unlock;
+		}
 	}
 
 	if (args->flags & MSM_SUBMIT_SYNCOBJ_OUT) {
-		post_deps = msm_parse_post_deps(dev, file,
-		                                args->out_syncobjs,
-		                                args->nr_out_syncobjs,
-		                                args->syncobj_stride);
+		post_deps = msm_syncobj_parse_post_deps(dev, file,
+							args->out_syncobjs,
+							args->nr_out_syncobjs,
+							args->syncobj_stride);
 		if (IS_ERR(post_deps)) {
 			ret = PTR_ERR(post_deps);
-			goto out_post_unlock;
-		}
-	}
-
-	ret = mutex_lock_interruptible(&dev->struct_mutex);
-	if (ret)
-		goto out_post_unlock;
-
-	if (args->flags & MSM_SUBMIT_FENCE_FD_OUT) {
-		out_fence_fd = get_unused_fd_flags(O_CLOEXEC);
-		if (out_fence_fd < 0) {
-			ret = out_fence_fd;
 			goto out_unlock;
 		}
 	}
 
-	submit = submit_create(dev, gpu, queue, args->nr_bos,
-		args->nr_cmds);
-	if (!submit) {
-		ret = -ENOMEM;
-		goto out_unlock;
-	}
-
-	submit->pid = pid;
-	submit->ident = submitid;
-
-	if (args->flags & MSM_SUBMIT_SUDO)
-		submit->in_rb = true;
-
 	ret = submit_lookup_objects(submit, args, file);
 	if (ret)
-		goto out_pre_pm;
+		goto out;
 
 	ret = submit_lookup_cmds(submit, args, file);
 	if (ret)
-		goto out_pre_pm;
-
-	/*
-	 * Thanks to dev_pm_opp opp_table_lock interactions with mm->mmap_sem
-	 * in the resume path, we need to to rpm get before we lock objs.
-	 * Which unfortunately might involve powering up the GPU sooner than
-	 * is necessary.  But at least in the explicit fencing case, we will
-	 * have already done all the fence waiting.
-	 */
-	pm_runtime_get_sync(&gpu->pdev->dev);
+		goto out;
 
 	/* copy_*_user while holding a ww ticket upsets lockdep */
-	ww_acquire_init(&submit->ticket, &reservation_ww_class);
-	has_ww_ticket = true;
 	ret = submit_lock_objects(submit);
 	if (ret)
 		goto out;
 
-	ret = submit_fence_sync(submit, !!(args->flags & MSM_SUBMIT_NO_IMPLICIT));
-	if (ret)
-		goto out;
+	if (!(args->flags & MSM_SUBMIT_NO_IMPLICIT)) {
+		ret = submit_fence_sync(submit);
+		if (ret)
+			goto out;
+	}
 
 	ret = submit_pin_objects(submit);
 	if (ret)
 		goto out;
 
-	for (i = 0; i < args->nr_cmds; i++) {
-		struct msm_gem_object *msm_obj;
+	cmds_to_parse = msm_context_is_vmbind(ctx) ? 0 : args->nr_cmds;
+
+	for (i = 0; i < cmds_to_parse; i++) {
+		struct drm_gem_object *obj;
 		uint64_t iova;
 
-		ret = submit_bo(submit, submit->cmd[i].idx,
-				&msm_obj, &iova, NULL);
+		ret = submit_bo(submit, submit->cmd[i].idx, &obj, &iova);
 		if (ret)
 			goto out;
 
 		if (!submit->cmd[i].size ||
-			((submit->cmd[i].size + submit->cmd[i].offset) >
-				msm_obj->base.size / 4)) {
-			DRM_ERROR("invalid cmdstream size: %u\n", submit->cmd[i].size * 4);
-			ret = -EINVAL;
+		    (size_add(submit->cmd[i].size, submit->cmd[i].offset) > obj->size / 4)) {
+			ret = UERR(EINVAL, dev, "invalid cmdstream size: %u\n",
+				   submit->cmd[i].size * 4);
 			goto out;
 		}
 
 		submit->cmd[i].iova = iova + (submit->cmd[i].offset * 4);
 
-		if (submit->valid)
+		if (likely(!submit->cmd[i].nr_relocs))
 			continue;
 
-		ret = submit_reloc(submit, msm_obj, submit->cmd[i].offset * 4,
+		if (!gpu->allow_relocs) {
+			ret = UERR(EINVAL, dev, "relocs not allowed\n");
+			goto out;
+		}
+
+		ret = submit_reloc(submit, obj, submit->cmd[i].offset * 4,
 				submit->cmd[i].nr_relocs, submit->cmd[i].relocs);
 		if (ret)
 			goto out;
 	}
 
-	submit->nr_cmds = i;
+	submit->nr_cmds = args->nr_cmds;
+
+	idr_preload(GFP_KERNEL);
+
+	spin_lock(&queue->idr_lock);
 
-	submit->fence = msm_fence_alloc(ring->fctx);
-	if (IS_ERR(submit->fence)) {
-		ret = PTR_ERR(submit->fence);
-		submit->fence = NULL;
+	/*
+	 * If using userspace provided seqno fence, validate that the id
+	 * is available before arming sched job.  Since access to fence_idr
+	 * is serialized on the queue lock, the slot should be still avail
+	 * after the job is armed
+	 */
+	if ((args->flags & MSM_SUBMIT_FENCE_SN_IN) &&
+			(!args->fence || idr_find(&queue->fence_idr, args->fence))) {
+		spin_unlock(&queue->idr_lock);
+		idr_preload_end();
+		ret = UERR(EINVAL, dev, "invalid in-fence-sn");
 		goto out;
 	}
 
-	if (args->flags & MSM_SUBMIT_FENCE_FD_OUT) {
-		sync_file = sync_file_create(submit->fence);
-		if (!sync_file) {
+	drm_sched_job_arm(&submit->base);
+
+	submit->user_fence = dma_fence_get(&submit->base.s_fence->finished);
+
+	if (args->flags & MSM_SUBMIT_FENCE_SN_IN) {
+		/*
+		 * Userspace has assigned the seqno fence that it wants
+		 * us to use.  It is an error to pick a fence sequence
+		 * number that is not available.
+		 */
+		submit->fence_id = args->fence;
+		ret = idr_alloc_u32(&queue->fence_idr, submit->user_fence,
+				    &submit->fence_id, submit->fence_id,
+				    GFP_NOWAIT);
+		/*
+		 * We've already validated that the fence_id slot is valid,
+		 * so if idr_alloc_u32 failed, it is a kernel bug
+		 */
+		WARN_ON(ret);
+	} else {
+		/*
+		 * Allocate an id which can be used by WAIT_FENCE ioctl to map
+		 * back to the underlying fence.
+		 */
+		submit->fence_id = idr_alloc_cyclic(&queue->fence_idr,
+						    submit->user_fence, 1,
+						    INT_MAX, GFP_NOWAIT);
+	}
+
+	spin_unlock(&queue->idr_lock);
+	idr_preload_end();
+
+	if (submit->fence_id < 0) {
+		ret = submit->fence_id;
+		submit->fence_id = 0;
+	}
+
+	if (ret == 0 && args->flags & MSM_SUBMIT_FENCE_FD_OUT) {
+		sync_file = sync_file_create(submit->user_fence);
+		if (!sync_file)
 			ret = -ENOMEM;
-			goto out;
-		}
 	}
 
-	msm_gpu_submit(gpu, submit);
+	if (ret)
+		goto out;
 
-	args->fence = submit->fence->seqno;
+	submit_attach_object_fences(submit);
 
-	if (args->flags & MSM_SUBMIT_FENCE_FD_OUT) {
-		fd_install(out_fence_fd, sync_file->file);
-		args->fence_fd = out_fence_fd;
+	if (msm_context_is_vmbind(ctx)) {
+		/*
+		 * If we are not using VM_BIND, submit_pin_vmas() will validate
+		 * just the BOs attached to the submit.  In that case we don't
+		 * need to validate the _entire_ vm, because userspace tracked
+		 * what BOs are associated with the submit.
+		 */
+		ret = drm_gpuvm_validate(submit->vm, &submit->exec);
+		if (ret)
+			goto out;
 	}
 
-	msm_reset_syncobjs(syncobjs_to_reset, args->nr_in_syncobjs);
-	msm_process_post_deps(post_deps, args->nr_out_syncobjs,
-	                      submit->fence);
+	/* The scheduler owns a ref now: */
+	msm_gem_submit_get(submit);
 
+	msm_rd_dump_submit(priv->rd, submit, NULL);
+
+	drm_sched_entity_push_job(&submit->base);
+
+	args->fence = submit->fence_id;
+	queue->last_fence = submit->fence_id;
+
+	msm_syncobj_reset(syncobjs_to_reset, args->nr_in_syncobjs);
+	msm_syncobj_process_post_deps(post_deps, args->nr_out_syncobjs, submit->user_fence);
 
 out:
-	pm_runtime_put(&gpu->pdev->dev);
-out_pre_pm:
-	submit_cleanup(submit);
-	if (has_ww_ticket)
-		ww_acquire_fini(&submit->ticket);
-	msm_gem_submit_put(submit);
+	submit_cleanup(submit, !!ret);
 out_unlock:
-	if (ret && (out_fence_fd >= 0))
-		put_unused_fd(out_fence_fd);
-	mutex_unlock(&dev->struct_mutex);
-
+	mutex_unlock(&queue->lock);
 out_post_unlock:
+	if (ret) {
+		if (out_fence_fd >= 0)
+			put_unused_fd(out_fence_fd);
+		if (sync_file)
+			fput(sync_file->file);
+	} else if (sync_file) {
+		fd_install(out_fence_fd, sync_file->file);
+		args->fence_fd = out_fence_fd;
+	}
+
+	if (!IS_ERR_OR_NULL(submit)) {
+		msm_gem_submit_put(submit);
+	} else {
+		/*
+		 * If the submit hasn't yet taken ownership of the queue
+		 * then we need to drop the reference ourself:
+		 */
+		msm_submitqueue_put(queue);
+	}
 	if (!IS_ERR_OR_NULL(post_deps)) {
 		for (i = 0; i < args->nr_out_syncobjs; ++i) {
 			kfree(post_deps[i].chain);