drm_amdgpu: Add job fence to resv conditionally

Job fence on page table should be a shared one, so add it to the root
page talbe bo resv.
last_delayed field is not needed anymore. so remove it.

Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c

@@ -1608,9 +1608,6 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 
 		if (!dma_fence_is_signaled(vm->last_direct))
 			amdgpu_bo_fence(root, vm->last_direct, true);
-
-		if (!dma_fence_is_signaled(vm->last_delayed))
-			amdgpu_bo_fence(root, vm->last_delayed, true);
 	}
 
 	r = vm->update_funcs->prepare(&params, resv, sync_mode);
@@ -2588,8 +2585,7 @@ bool amdgpu_vm_evictable(struct amdgpu_bo *bo)
 		return false;
 
 	/* Don't evict VM page tables while they are updated */
-	if (!dma_fence_is_signaled(bo_base->vm->last_direct) ||
-	    !dma_fence_is_signaled(bo_base->vm->last_delayed)) {
+	if (!dma_fence_is_signaled(bo_base->vm->last_direct)) {
 		amdgpu_vm_eviction_unlock(bo_base->vm);
 		return false;
 	}
@@ -2766,11 +2762,7 @@ long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
 	if (timeout <= 0)
 		return timeout;
 
-	timeout = dma_fence_wait_timeout(vm->last_direct, true, timeout);
-	if (timeout <= 0)
-		return timeout;
-
-	return dma_fence_wait_timeout(vm->last_delayed, true, timeout);
+	return dma_fence_wait_timeout(vm->last_direct, true, timeout);
 }
 
 /**
@@ -2843,7 +2835,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		vm->update_funcs = &amdgpu_vm_sdma_funcs;
 	vm->last_update = NULL;
 	vm->last_direct = dma_fence_get_stub();
-	vm->last_delayed = dma_fence_get_stub();
 
 	mutex_init(&vm->eviction_lock);
 	vm->evicting = false;
@@ -2898,7 +2889,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 
 error_free_delayed:
 	dma_fence_put(vm->last_direct);
-	dma_fence_put(vm->last_delayed);
 	drm_sched_entity_destroy(&vm->delayed);
 
 error_free_direct:
@@ -3101,8 +3091,6 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 
 	dma_fence_wait(vm->last_direct, false);
 	dma_fence_put(vm->last_direct);
-	dma_fence_wait(vm->last_delayed, false);
-	dma_fence_put(vm->last_delayed);
 
 	list_for_each_entry_safe(mapping, tmp, &vm->freed, list) {
 		if (mapping->flags & AMDGPU_PTE_PRT && prt_fini_needed) {

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h

@@ -276,7 +276,6 @@ struct amdgpu_vm {
 
 	/* Last submission to the scheduler entities */
 	struct dma_fence	*last_direct;
-	struct dma_fence	*last_delayed;
 
 	unsigned int		pasid;
 	/* dedicated to vm */

drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c

@@ -104,12 +104,13 @@ static int amdgpu_vm_sdma_commit(struct amdgpu_vm_update_params *p,
 	if (r)
 		goto error;
 
+	if (p->direct) {
 		tmp = dma_fence_get(f);
-	if (p->direct)
 		swap(p->vm->last_direct, tmp);
-	else
-		swap(p->vm->last_delayed, tmp);
 		dma_fence_put(tmp);
+	} else {
+		dma_resv_add_shared_fence(p->vm->root.base.bo->tbo.base.resv, f);
+	}
 
 	if (fence && !p->direct)
 		swap(*fence, f);