1 files changed, 181 insertions, 112 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index df110afa97bf..0768c8686983 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -300,7 +300,7 @@ static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
 static void amdgpu_vm_bo_done(struct amdgpu_vm_bo_base *vm_bo)
 {
 	spin_lock(&vm_bo->vm->invalidated_lock);
-	list_del_init(&vm_bo->vm_status);
+	list_move(&vm_bo->vm_status, &vm_bo->vm->done);
 	spin_unlock(&vm_bo->vm->invalidated_lock);
 }
 
@@ -609,7 +609,7 @@ void amdgpu_vm_del_from_lru_notify(struct ttm_buffer_object *bo)
 	if (!amdgpu_bo_is_amdgpu_bo(bo))
 		return;
 
-	if (bo->mem.placement & TTM_PL_FLAG_NO_EVICT)
+	if (bo->pin_count)
 		return;
 
 	abo = ttm_to_amdgpu_bo(bo);
@@ -1570,7 +1570,8 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params,
 /**
  * amdgpu_vm_bo_update_mapping - update a mapping in the vm page table
  *
- * @adev: amdgpu_device pointer
+ * @adev: amdgpu_device pointer of the VM
+ * @bo_adev: amdgpu_device pointer of the mapped BO
  * @vm: requested vm
  * @immediate: immediate submission in a page fault
  * @unlocked: unlocked invalidation during MM callback
@@ -1578,7 +1579,8 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params,
  * @start: start of mapped range
  * @last: last mapped entry
  * @flags: flags for the entries
- * @addr: addr to set the area to
+ * @offset: offset into nodes and pages_addr
+ * @nodes: array of drm_mm_nodes with the MC addresses
  * @pages_addr: DMA addresses to use for mapping
  * @fence: optional resulting fence
  *
@@ -1588,15 +1590,18 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params,
  * 0 for success, -EINVAL for failure.
  */
 static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
+				       struct amdgpu_device *bo_adev,
 				       struct amdgpu_vm *vm, bool immediate,
 				       bool unlocked, struct dma_resv *resv,
 				       uint64_t start, uint64_t last,
-				       uint64_t flags, uint64_t addr,
+				       uint64_t flags, uint64_t offset,
+				       struct drm_mm_node *nodes,
 				       dma_addr_t *pages_addr,
 				       struct dma_fence **fence)
 {
 	struct amdgpu_vm_update_params params;
 	enum amdgpu_sync_mode sync_mode;
+	uint64_t pfn;
 	int r;
 
 	memset(&params, 0, sizeof(params));
@@ -1614,6 +1619,14 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 	else
 		sync_mode = AMDGPU_SYNC_EXPLICIT;
 
+	pfn = offset >> PAGE_SHIFT;
+	if (nodes) {
+		while (pfn >= nodes->size) {
+			pfn -= nodes->size;
+			++nodes;
+		}
+	}
+
 	amdgpu_vm_eviction_lock(vm);
 	if (vm->evicting) {
 		r = -EBUSY;
@@ -1632,105 +1645,47 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 	if (r)
 		goto error_unlock;
 
-	r = amdgpu_vm_update_ptes(&params, start, last + 1, addr, flags);
-	if (r)
-		goto error_unlock;
-
-	r = vm->update_funcs->commit(&params, fence);
-
-error_unlock:
-	amdgpu_vm_eviction_unlock(vm);
-	return r;
-}
-
-/**
- * amdgpu_vm_bo_split_mapping - split a mapping into smaller chunks
- *
- * @adev: amdgpu_device pointer
- * @resv: fences we need to sync to
- * @pages_addr: DMA addresses to use for mapping
- * @vm: requested vm
- * @mapping: mapped range and flags to use for the update
- * @flags: HW flags for the mapping
- * @bo_adev: amdgpu_device pointer that bo actually been allocated
- * @nodes: array of drm_mm_nodes with the MC addresses
- * @fence: optional resulting fence
- *
- * Split the mapping into smaller chunks so that each update fits
- * into a SDMA IB.
- *
- * Returns:
- * 0 for success, -EINVAL for failure.
- */
-static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
-				      struct dma_resv *resv,
-				      dma_addr_t *pages_addr,
-				      struct amdgpu_vm *vm,
-				      struct amdgpu_bo_va_mapping *mapping,
-				      uint64_t flags,
-				      struct amdgpu_device *bo_adev,
-				      struct drm_mm_node *nodes,
-				      struct dma_fence **fence)
-{
-	unsigned min_linear_pages = 1 << adev->vm_manager.fragment_size;
-	uint64_t pfn, start = mapping->start;
-	int r;
-
-	/* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
-	 * but in case of something, we filter the flags in first place
-	 */
-	if (!(mapping->flags & AMDGPU_PTE_READABLE))
-		flags &= ~AMDGPU_PTE_READABLE;
-	if (!(mapping->flags & AMDGPU_PTE_WRITEABLE))
-		flags &= ~AMDGPU_PTE_WRITEABLE;
-
-	/* Apply ASIC specific mapping flags */
-	amdgpu_gmc_get_vm_pte(adev, mapping, &flags);
-
-	trace_amdgpu_vm_bo_update(mapping);
-
-	pfn = mapping->offset >> PAGE_SHIFT;
-	if (nodes) {
-		while (pfn >= nodes->size) {
-			pfn -= nodes->size;
-			++nodes;
-		}
-	}
-
 	do {
-		dma_addr_t *dma_addr = NULL;
-		uint64_t max_entries;
-		uint64_t addr, last;
+		uint64_t tmp, num_entries, addr;
+
 
-		max_entries = mapping->last - start + 1;
+		num_entries = last - start + 1;
 		if (nodes) {
 			addr = nodes->start << PAGE_SHIFT;
-			max_entries = min((nodes->size - pfn) *
-				AMDGPU_GPU_PAGES_IN_CPU_PAGE, max_entries);
+			num_entries = min((nodes->size - pfn) *
+				AMDGPU_GPU_PAGES_IN_CPU_PAGE, num_entries);
 		} else {
 			addr = 0;
 		}
 
 		if (pages_addr) {
-			uint64_t count;
+			bool contiguous = true;
 
-			for (count = 1;
-			     count < max_entries / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
-			     ++count) {
-				uint64_t idx = pfn + count;
+			if (num_entries > AMDGPU_GPU_PAGES_IN_CPU_PAGE) {
+				uint64_t count;
 
-				if (pages_addr[idx] !=
-				    (pages_addr[idx - 1] + PAGE_SIZE))
-					break;
+				contiguous = pages_addr[pfn + 1] ==
+					pages_addr[pfn] + PAGE_SIZE;
+
+				tmp = num_entries /
+					AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+				for (count = 2; count < tmp; ++count) {
+					uint64_t idx = pfn + count;
+
+					if (contiguous != (pages_addr[idx] ==
+					    pages_addr[idx - 1] + PAGE_SIZE))
+						break;
+				}
+				num_entries = count *
+					AMDGPU_GPU_PAGES_IN_CPU_PAGE;
 			}
 
-			if (count < min_linear_pages) {
+			if (!contiguous) {
 				addr = pfn << PAGE_SHIFT;
-				dma_addr = pages_addr;
+				params.pages_addr = pages_addr;
 			} else {
 				addr = pages_addr[pfn];
-				max_entries = count *
-					AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+				params.pages_addr = NULL;
 			}
 
 		} else if (flags & (AMDGPU_PTE_VALID | AMDGPU_PTE_PRT)) {
@@ -1738,23 +1693,25 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
 			addr += pfn << PAGE_SHIFT;
 		}
 
-		last = start + max_entries - 1;
-		r = amdgpu_vm_bo_update_mapping(adev, vm, false, false, resv,
-						start, last, flags, addr,
-						dma_addr, fence);
+		tmp = start + num_entries;
+		r = amdgpu_vm_update_ptes(&params, start, tmp, addr, flags);
 		if (r)
-			return r;
+			goto error_unlock;
 
-		pfn += (last - start + 1) / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
+		pfn += num_entries / AMDGPU_GPU_PAGES_IN_CPU_PAGE;
 		if (nodes && nodes->size == pfn) {
 			pfn = 0;
 			++nodes;
 		}
-		start = last + 1;
+		start = tmp;
 
-	} while (unlikely(start != mapping->last + 1));
+	} while (unlikely(start != last + 1));
 
-	return 0;
+	r = vm->update_funcs->commit(&params, fence);
+
+error_unlock:
+	amdgpu_vm_eviction_unlock(vm);
+	return r;
 }
 
 /**
@@ -1790,7 +1747,6 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 		resv = vm->root.base.bo->tbo.base.resv;
 	} else {
 		struct drm_gem_object *obj = &bo->tbo.base;
-		struct ttm_dma_tt *ttm;
 
 		resv = bo->tbo.base.resv;
 		if (obj->import_attach && bo_va->is_xgmi) {
@@ -1803,10 +1759,8 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 		}
 		mem = &bo->tbo.mem;
 		nodes = mem->mm_node;
-		if (mem->mem_type == TTM_PL_TT) {
-			ttm = container_of(bo->tbo.ttm, struct ttm_dma_tt, ttm);
-			pages_addr = ttm->dma_address;
-		}
+		if (mem->mem_type == TTM_PL_TT)
+			pages_addr = bo->tbo.ttm->dma_address;
 	}
 
 	if (bo) {
@@ -1835,9 +1789,26 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 	}
 
 	list_for_each_entry(mapping, &bo_va->invalids, list) {
-		r = amdgpu_vm_bo_split_mapping(adev, resv, pages_addr, vm,
-					       mapping, flags, bo_adev, nodes,
-					       last_update);
+		uint64_t update_flags = flags;
+
+		/* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
+		 * but in case of something, we filter the flags in first place
+		 */
+		if (!(mapping->flags & AMDGPU_PTE_READABLE))
+			update_flags &= ~AMDGPU_PTE_READABLE;
+		if (!(mapping->flags & AMDGPU_PTE_WRITEABLE))
+			update_flags &= ~AMDGPU_PTE_WRITEABLE;
+
+		/* Apply ASIC specific mapping flags */
+		amdgpu_gmc_get_vm_pte(adev, mapping, &update_flags);
+
+		trace_amdgpu_vm_bo_update(mapping);
+
+		r = amdgpu_vm_bo_update_mapping(adev, bo_adev, vm, false, false,
+						resv, mapping->start,
+						mapping->last, update_flags,
+						mapping->offset, nodes,
+						pages_addr, last_update);
 		if (r)
 			return r;
 	}
@@ -2045,9 +2016,10 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 		    mapping->start < AMDGPU_GMC_HOLE_START)
 			init_pte_value = AMDGPU_PTE_DEFAULT_ATC;
 
-		r = amdgpu_vm_bo_update_mapping(adev, vm, false, false, resv,
-						mapping->start, mapping->last,
-						init_pte_value, 0, NULL, &f);
+		r = amdgpu_vm_bo_update_mapping(adev, adev, vm, false, false,
+						resv, mapping->start,
+						mapping->last, init_pte_value,
+						0, NULL, NULL, &f);
 		amdgpu_vm_free_mapping(adev, vm, mapping, f);
 		if (r) {
 			dma_fence_put(f);
@@ -2166,7 +2138,7 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
 
 
 /**
- * amdgpu_vm_bo_insert_mapping - insert a new mapping
+ * amdgpu_vm_bo_insert_map - insert a new mapping
  *
  * @adev: amdgpu_device pointer
  * @bo_va: bo_va to store the address
@@ -2823,7 +2795,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	INIT_LIST_HEAD(&vm->invalidated);
 	spin_lock_init(&vm->invalidated_lock);
 	INIT_LIST_HEAD(&vm->freed);
-
+	INIT_LIST_HEAD(&vm->done);
 
 	/* create scheduler entities for page table updates */
 	r = drm_sched_entity_init(&vm->immediate, DRM_SCHED_PRIORITY_NORMAL,
@@ -3375,8 +3347,9 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 		value = 0;
 	}
 
-	r = amdgpu_vm_bo_update_mapping(adev, vm, true, false, NULL, addr,
-					addr + 1, flags, value, NULL, NULL);
+	r = amdgpu_vm_bo_update_mapping(adev, adev, vm, true, false, NULL, addr,
+					addr, flags, value, NULL, NULL,
+					NULL);
 	if (r)
 		goto error_unlock;
 
@@ -3392,3 +3365,99 @@ error_unref:
 
 	return false;
 }
+
+#if defined(CONFIG_DEBUG_FS)
+/**
+ * amdgpu_debugfs_vm_bo_info  - print BO info for the VM
+ *
+ * @vm: Requested VM for printing BO info
+ * @m: debugfs file
+ *
+ * Print BO information in debugfs file for the VM
+ */
+void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
+{
+	struct amdgpu_bo_va *bo_va, *tmp;
+	u64 total_idle = 0;
+	u64 total_evicted = 0;
+	u64 total_relocated = 0;
+	u64 total_moved = 0;
+	u64 total_invalidated = 0;
+	u64 total_done = 0;
+	unsigned int total_idle_objs = 0;
+	unsigned int total_evicted_objs = 0;
+	unsigned int total_relocated_objs = 0;
+	unsigned int total_moved_objs = 0;
+	unsigned int total_invalidated_objs = 0;
+	unsigned int total_done_objs = 0;
+	unsigned int id = 0;
+
+	seq_puts(m, "\tIdle BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_idle += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_idle_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tEvicted BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_evicted += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_evicted_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tRelocated BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_relocated += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_relocated_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tMoved BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_moved += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	total_moved_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tInvalidated BOs:\n");
+	spin_lock(&vm->invalidated_lock);
+	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_invalidated += amdgpu_bo_print_info(id++,	bo_va->base.bo, m);
+	}
+	total_invalidated_objs = id;
+	id = 0;
+
+	seq_puts(m, "\tDone BOs:\n");
+	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status) {
+		if (!bo_va->base.bo)
+			continue;
+		total_done += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
+	}
+	spin_unlock(&vm->invalidated_lock);
+	total_done_objs = id;
+
+	seq_printf(m, "\tTotal idle size:        %12lld\tobjs:\t%d\n", total_idle,
+		   total_idle_objs);
+	seq_printf(m, "\tTotal evicted size:     %12lld\tobjs:\t%d\n", total_evicted,
+		   total_evicted_objs);
+	seq_printf(m, "\tTotal relocated size:   %12lld\tobjs:\t%d\n", total_relocated,
+		   total_relocated_objs);
+	seq_printf(m, "\tTotal moved size:       %12lld\tobjs:\t%d\n", total_moved,
+		   total_moved_objs);
+	seq_printf(m, "\tTotal invalidated size: %12lld\tobjs:\t%d\n", total_invalidated,
+		   total_invalidated_objs);
+	seq_printf(m, "\tTotal done size:        %12lld\tobjs:\t%d\n", total_done,
+		   total_done_objs);
+}
+#endif