From dcb388eddb5f1bbc817d0e9ba341908b0a201ff2 Mon Sep 17 00:00:00 2001
From: Nirmoy Das <nirmoy.das@amd.com>
Date: Mon, 28 Jun 2021 23:29:39 +0200
Subject: drm/amdgpu: use xarray for storing pasid in vm
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Replace idr with xarray as we actually need hash functionality.
Cleanup code related to vm pasid by adding helper function.

Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@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 | 149 +++++++++++++++------------------
 1 file changed, 69 insertions(+), 80 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 79cfa2d68487..9ea527ea3238 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -88,6 +88,46 @@ struct amdgpu_prt_cb {
 	struct dma_fence_cb cb;
 };
 
+/**
+ * amdgpu_vm_set_pasid - manage pasid and vm ptr mapping
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: amdgpu_vm pointer
+ * @pasid: the pasid the VM is using on this GPU
+ *
+ * Set the pasid this VM is using on this GPU, can also be used to remove the
+ * pasid by passing in zero.
+ *
+ */
+int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			u32 pasid)
+{
+	int r;
+
+	if (vm->pasid == pasid)
+		return 0;
+
+	if (vm->pasid) {
+		r = xa_err(xa_erase_irq(&adev->vm_manager.pasids, vm->pasid));
+		if (r < 0)
+			return r;
+
+		vm->pasid = 0;
+	}
+
+	if (pasid) {
+		r = xa_err(xa_store_irq(&adev->vm_manager.pasids, pasid, vm,
+					GFP_KERNEL));
+		if (r < 0)
+			return r;
+
+		vm->pasid = pasid;
+	}
+
+
+	return 0;
+}
+
 /*
  * vm eviction_lock can be taken in MMU notifiers. Make sure no reclaim-FS
  * happens while holding this lock anywhere to prevent deadlocks when
@@ -2945,18 +2985,9 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, u32 pasid)
 
 	amdgpu_bo_unreserve(vm->root.bo);
 
-	if (pasid) {
-		unsigned long flags;
-
-		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
-		r = idr_alloc(&adev->vm_manager.pasid_idr, vm, pasid, pasid + 1,
-			      GFP_ATOMIC);
-		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
-		if (r < 0)
-			goto error_free_root;
-
-		vm->pasid = pasid;
-	}
+	r = amdgpu_vm_set_pasid(adev, vm, pasid);
+	if (r)
+		goto error_free_root;
 
 	INIT_KFIFO(vm->faults);
 
@@ -3044,18 +3075,15 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (r)
 		goto unreserve_bo;
 
-	if (pasid) {
-		unsigned long flags;
-
-		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
-		r = idr_alloc(&adev->vm_manager.pasid_idr, vm, pasid, pasid + 1,
-			      GFP_ATOMIC);
-		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
+	/* Free the original amdgpu allocated pasid,
+	 * will be replaced with kfd allocated pasid.
+	 */
+	if (vm->pasid)
+		amdgpu_pasid_free(vm->pasid);
 
-		if (r == -ENOSPC)
-			goto unreserve_bo;
-		r = 0;
-	}
+	r = amdgpu_vm_set_pasid(adev, vm, pasid);
+	if (r)
+		goto unreserve_bo;
 
 	/* Check if PD needs to be reinitialized and do it before
 	 * changing any other state, in case it fails.
@@ -3066,7 +3094,7 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 				       to_amdgpu_bo_vm(vm->root.bo),
 				       false);
 		if (r)
-			goto free_idr;
+			goto free_pasid_entry;
 	}
 
 	/* Update VM state */
@@ -3083,7 +3111,7 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		r = amdgpu_bo_sync_wait(vm->root.bo,
 					AMDGPU_FENCE_OWNER_UNDEFINED, true);
 		if (r)
-			goto free_idr;
+			goto free_pasid_entry;
 
 		vm->update_funcs = &amdgpu_vm_cpu_funcs;
 	} else {
@@ -3093,36 +3121,13 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	vm->last_update = NULL;
 	vm->is_compute_context = true;
 
-	if (vm->pasid) {
-		unsigned long flags;
-
-		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
-		idr_remove(&adev->vm_manager.pasid_idr, vm->pasid);
-		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
-
-		/* Free the original amdgpu allocated pasid
-		 * Will be replaced with kfd allocated pasid
-		 */
-		amdgpu_pasid_free(vm->pasid);
-		vm->pasid = 0;
-	}
-
 	/* Free the shadow bo for compute VM */
 	amdgpu_bo_unref(&to_amdgpu_bo_vm(vm->root.bo)->shadow);
 
-	if (pasid)
-		vm->pasid = pasid;
-
 	goto unreserve_bo;
 
-free_idr:
-	if (pasid) {
-		unsigned long flags;
-
-		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
-		idr_remove(&adev->vm_manager.pasid_idr, pasid);
-		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
-	}
+free_pasid_entry:
+	amdgpu_vm_set_pasid(adev, vm, 0);
 unreserve_bo:
 	amdgpu_bo_unreserve(vm->root.bo);
 	return r;
@@ -3138,14 +3143,7 @@ unreserve_bo:
  */
 void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
-	if (vm->pasid) {
-		unsigned long flags;
-
-		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
-		idr_remove(&adev->vm_manager.pasid_idr, vm->pasid);
-		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
-	}
-	vm->pasid = 0;
+	amdgpu_vm_set_pasid(adev, vm, 0);
 	vm->is_compute_context = false;
 }
 
@@ -3169,15 +3167,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 
 	root = amdgpu_bo_ref(vm->root.bo);
 	amdgpu_bo_reserve(root, true);
-	if (vm->pasid) {
-		unsigned long flags;
-
-		spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
-		idr_remove(&adev->vm_manager.pasid_idr, vm->pasid);
-		spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
-		vm->pasid = 0;
-	}
-
+	amdgpu_vm_set_pasid(adev, vm, 0);
 	dma_fence_wait(vm->last_unlocked, false);
 	dma_fence_put(vm->last_unlocked);
 
@@ -3259,8 +3249,7 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 	adev->vm_manager.vm_update_mode = 0;
 #endif
 
-	idr_init(&adev->vm_manager.pasid_idr);
-	spin_lock_init(&adev->vm_manager.pasid_lock);
+	xa_init_flags(&adev->vm_manager.pasids, XA_FLAGS_LOCK_IRQ);
 }
 
 /**
@@ -3272,8 +3261,8 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
  */
 void amdgpu_vm_manager_fini(struct amdgpu_device *adev)
 {
-	WARN_ON(!idr_is_empty(&adev->vm_manager.pasid_idr));
-	idr_destroy(&adev->vm_manager.pasid_idr);
+	WARN_ON(!xa_empty(&adev->vm_manager.pasids));
+	xa_destroy(&adev->vm_manager.pasids);
 
 	amdgpu_vmid_mgr_fini(adev);
 }
@@ -3342,13 +3331,13 @@ void amdgpu_vm_get_task_info(struct amdgpu_device *adev, u32 pasid,
 	struct amdgpu_vm *vm;
 	unsigned long flags;
 
-	spin_lock_irqsave(&adev->vm_manager.pasid_lock, flags);
+	xa_lock_irqsave(&adev->vm_manager.pasids, flags);
 
-	vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
 	if (vm)
 		*task_info = vm->task_info;
 
-	spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, flags);
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, flags);
 }
 
 /**
@@ -3390,15 +3379,15 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 	struct amdgpu_vm *vm;
 	int r;
 
-	spin_lock_irqsave(&adev->vm_manager.pasid_lock, irqflags);
-	vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
+	xa_lock_irqsave(&adev->vm_manager.pasids, irqflags);
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
 	if (vm) {
 		root = amdgpu_bo_ref(vm->root.bo);
 		is_compute_context = vm->is_compute_context;
 	} else {
 		root = NULL;
 	}
-	spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, irqflags);
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, irqflags);
 
 	if (!root)
 		return false;
@@ -3416,11 +3405,11 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 		goto error_unref;
 
 	/* Double check that the VM still exists */
-	spin_lock_irqsave(&adev->vm_manager.pasid_lock, irqflags);
-	vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
+	xa_lock_irqsave(&adev->vm_manager.pasids, irqflags);
+	vm = xa_load(&adev->vm_manager.pasids, pasid);
 	if (vm && vm->root.bo != root)
 		vm = NULL;
-	spin_unlock_irqrestore(&adev->vm_manager.pasid_lock, irqflags);
+	xa_unlock_irqrestore(&adev->vm_manager.pasids, irqflags);
 	if (!vm)
 		goto error_unlock;
 
-- 
cgit v1.2.3


From 88f7f88159bcdff96b2a5d244b26c8ba99b5e773 Mon Sep 17 00:00:00 2001
From: Nirmoy Das <nirmoy.das@amd.com>
Date: Fri, 2 Jul 2021 11:10:52 +0200
Subject: drm/amdgpu: separate out vm pasid assignment
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Use new helper function amdgpu_vm_set_pasid() to
assign vm pasid value. This also ensures that we don't free
a pasid from vm code as pasids are allocated somewhere else.

Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@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_amdkfd_gpuvm.c | 13 ++++++++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c          | 10 +++++++--
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c           | 28 ++++--------------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h           |  4 ++--
 4 files changed, 26 insertions(+), 29 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 3b8e1ee8c475..da02bd1bbb31 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -1287,11 +1287,22 @@ int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct kgd_dev *kgd,
 	if (avm->process_info)
 		return -EINVAL;
 
+	/* Free the original amdgpu allocated pasid,
+	 * will be replaced with kfd allocated pasid.
+	 */
+	if (avm->pasid) {
+		amdgpu_pasid_free(avm->pasid);
+		amdgpu_vm_set_pasid(adev, avm, 0);
+	}
+
 	/* Convert VM into a compute VM */
-	ret = amdgpu_vm_make_compute(adev, avm, pasid);
+	ret = amdgpu_vm_make_compute(adev, avm);
 	if (ret)
 		return ret;
 
+	ret = amdgpu_vm_set_pasid(adev, avm, pasid);
+	if (ret)
+		return ret;
 	/* Initialize KFD part of the VM and process info */
 	ret = init_kfd_vm(avm, process_info, ef);
 	if (ret)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 96ef3f1051d8..0de9ccd47eb9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -1179,10 +1179,14 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 		pasid = 0;
 	}
 
-	r = amdgpu_vm_init(adev, &fpriv->vm, pasid);
+	r = amdgpu_vm_init(adev, &fpriv->vm);
 	if (r)
 		goto error_pasid;
 
+	r = amdgpu_vm_set_pasid(adev, &fpriv->vm, pasid);
+	if (r)
+		goto error_vm;
+
 	fpriv->prt_va = amdgpu_vm_bo_add(adev, &fpriv->vm, NULL);
 	if (!fpriv->prt_va) {
 		r = -ENOMEM;
@@ -1210,8 +1214,10 @@ error_vm:
 	amdgpu_vm_fini(adev, &fpriv->vm);
 
 error_pasid:
-	if (pasid)
+	if (pasid) {
 		amdgpu_pasid_free(pasid);
+		amdgpu_vm_set_pasid(adev, &fpriv->vm, 0);
+	}
 
 	kfree(fpriv);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 9ea527ea3238..2a88ed5d983b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -2904,14 +2904,13 @@ long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
- * @pasid: Process address space identifier
  *
  * Init @vm fields.
  *
  * Returns:
  * 0 for success, error for failure.
  */
-int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, u32 pasid)
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
 	struct amdgpu_bo *root_bo;
 	struct amdgpu_bo_vm *root;
@@ -2985,10 +2984,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, u32 pasid)
 
 	amdgpu_bo_unreserve(vm->root.bo);
 
-	r = amdgpu_vm_set_pasid(adev, vm, pasid);
-	if (r)
-		goto error_free_root;
-
 	INIT_KFIFO(vm->faults);
 
 	return 0;
@@ -3044,7 +3039,6 @@ static int amdgpu_vm_check_clean_reserved(struct amdgpu_device *adev,
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
- * @pasid: pasid to use
  *
  * This only works on GFX VMs that don't have any BOs added and no
  * page tables allocated yet.
@@ -3052,7 +3046,6 @@ static int amdgpu_vm_check_clean_reserved(struct amdgpu_device *adev,
  * Changes the following VM parameters:
  * - use_cpu_for_update
  * - pte_supports_ats
- * - pasid (old PASID is released, because compute manages its own PASIDs)
  *
  * Reinitializes the page directory to reflect the changed ATS
  * setting.
@@ -3060,8 +3053,7 @@ static int amdgpu_vm_check_clean_reserved(struct amdgpu_device *adev,
  * Returns:
  * 0 for success, -errno for errors.
  */
-int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			   u32 pasid)
+int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
 	bool pte_support_ats = (adev->asic_type == CHIP_RAVEN);
 	int r;
@@ -3075,16 +3067,6 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (r)
 		goto unreserve_bo;
 
-	/* Free the original amdgpu allocated pasid,
-	 * will be replaced with kfd allocated pasid.
-	 */
-	if (vm->pasid)
-		amdgpu_pasid_free(vm->pasid);
-
-	r = amdgpu_vm_set_pasid(adev, vm, pasid);
-	if (r)
-		goto unreserve_bo;
-
 	/* Check if PD needs to be reinitialized and do it before
 	 * changing any other state, in case it fails.
 	 */
@@ -3094,7 +3076,7 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 				       to_amdgpu_bo_vm(vm->root.bo),
 				       false);
 		if (r)
-			goto free_pasid_entry;
+			goto unreserve_bo;
 	}
 
 	/* Update VM state */
@@ -3111,7 +3093,7 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		r = amdgpu_bo_sync_wait(vm->root.bo,
 					AMDGPU_FENCE_OWNER_UNDEFINED, true);
 		if (r)
-			goto free_pasid_entry;
+			goto unreserve_bo;
 
 		vm->update_funcs = &amdgpu_vm_cpu_funcs;
 	} else {
@@ -3126,8 +3108,6 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 
 	goto unreserve_bo;
 
-free_pasid_entry:
-	amdgpu_vm_set_pasid(adev, vm, 0);
 unreserve_bo:
 	amdgpu_bo_unreserve(vm->root.bo);
 	return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index bcbe362e7d5b..80cc9ab2c1d0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -378,8 +378,8 @@ int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			u32 pasid);
 
 long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout);
-int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, u32 pasid);
-int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, u32 pasid);
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
-- 
cgit v1.2.3


From 4e7b93ca52fb228b177168d436449c5671415a72 Mon Sep 17 00:00:00 2001
From: Eric Huang <JinhuiEric.Huang@amd.com>
Date: Fri, 9 Jul 2021 14:55:03 -0400
Subject: Revert "drm/amdgpu: Fix warning of Function parameter or member not
 described"

This reverts commit 7a68d188d1c4a9d947369acaa19040a58baaaeda.

Reason for revert: the related commit is reverted.

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 2a88ed5d983b..f735282ba878 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1856,7 +1856,6 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
  * @adev: amdgpu_device pointer
  * @bo_va: requested BO and VM object
  * @clear: if true clear the entries
- * @table_freed: return true if page table is freed
  *
  * Fill in the page table entries for @bo_va.
  *
-- 
cgit v1.2.3


From 024d8811c90ed56d8b90cdcf71e51c9fedeff460 Mon Sep 17 00:00:00 2001
From: Eric Huang <JinhuiEric.Huang@amd.com>
Date: Fri, 9 Jul 2021 14:58:16 -0400
Subject: Revert "drm/amdgpu: Add table_freed parameter to amdgpu_vm_bo_update"

This reverts commit 075e8080c1a7571563171a07fa9ce47c4bc80044.

Reason for revert: the related commit is reverted.

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c  |  6 +++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c  | 10 +++++-----
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h  |  2 +-
 4 files changed, 10 insertions(+), 10 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 76fe5b71e35d..30fa1f61e0e5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -781,7 +781,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_vm_bo_update(adev, fpriv->prt_va, false, NULL);
+	r = amdgpu_vm_bo_update(adev, fpriv->prt_va, false);
 	if (r)
 		return r;
 
@@ -792,7 +792,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) {
 		bo_va = fpriv->csa_va;
 		BUG_ON(!bo_va);
-		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			return r;
 
@@ -811,7 +811,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 		if (bo_va == NULL)
 			continue;
 
-		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index b3404c43a911..d0d9bc445d7b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -612,7 +612,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 
 	if (operation == AMDGPU_VA_OP_MAP ||
 	    operation == AMDGPU_VA_OP_REPLACE) {
-		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			goto error;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index f735282ba878..6362e861a3f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1798,7 +1798,7 @@ int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 	r = vm->update_funcs->commit(&params, fence);
 
 	if (table_freed)
-		*table_freed = *table_freed || params.table_freed;
+		*table_freed = params.table_freed;
 
 error_unlock:
 	amdgpu_vm_eviction_unlock(vm);
@@ -1863,7 +1863,7 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
  * 0 for success, -EINVAL for failure.
  */
 int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
-			bool clear, bool *table_freed)
+			bool clear)
 {
 	struct amdgpu_bo *bo = bo_va->base.bo;
 	struct amdgpu_vm *vm = bo_va->base.vm;
@@ -1942,7 +1942,7 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 						resv, mapping->start,
 						mapping->last, update_flags,
 						mapping->offset, mem,
-						pages_addr, last_update, table_freed);
+						pages_addr, last_update, NULL);
 		if (r)
 			return r;
 	}
@@ -2194,7 +2194,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 
 	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
 		/* Per VM BOs never need to bo cleared in the page tables */
-		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
+		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			return r;
 	}
@@ -2213,7 +2213,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 		else
 			clear = true;
 
-		r = amdgpu_vm_bo_update(adev, bo_va, clear, NULL);
+		r = amdgpu_vm_bo_update(adev, bo_va, clear);
 		if (r)
 			return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 80cc9ab2c1d0..ee0028ec2c4f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -408,7 +408,7 @@ int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 				struct dma_fence **fence, bool *free_table);
 int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 			struct amdgpu_bo_va *bo_va,
-			bool clear, bool *table_freed);
+			bool clear);
 bool amdgpu_vm_evictable(struct amdgpu_bo *bo);
 void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
 			     struct amdgpu_bo *bo, bool evicted);
-- 
cgit v1.2.3


From e9949dd79182a29410379dfebea8870f6abfa6e7 Mon Sep 17 00:00:00 2001
From: Eric Huang <JinhuiEric.Huang@amd.com>
Date: Mon, 26 Jul 2021 11:23:17 -0400
Subject: Revert "Revert "drm/amdgpu: Add table_freed parameter to
 amdgpu_vm_bo_update""

This reverts commit 024d8811c90ed56d8b90cdcf71e51c9fedeff460.

Revert reason: The issue has been resolved.

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c  |  6 +++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c  | 10 +++++-----
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h  |  2 +-
 4 files changed, 10 insertions(+), 10 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 30fa1f61e0e5..76fe5b71e35d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -781,7 +781,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_vm_bo_update(adev, fpriv->prt_va, false);
+	r = amdgpu_vm_bo_update(adev, fpriv->prt_va, false, NULL);
 	if (r)
 		return r;
 
@@ -792,7 +792,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) {
 		bo_va = fpriv->csa_va;
 		BUG_ON(!bo_va);
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			return r;
 
@@ -811,7 +811,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 		if (bo_va == NULL)
 			continue;
 
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index d0d9bc445d7b..b3404c43a911 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -612,7 +612,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 
 	if (operation == AMDGPU_VA_OP_MAP ||
 	    operation == AMDGPU_VA_OP_REPLACE) {
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			goto error;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 6362e861a3f5..f735282ba878 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1798,7 +1798,7 @@ int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 	r = vm->update_funcs->commit(&params, fence);
 
 	if (table_freed)
-		*table_freed = params.table_freed;
+		*table_freed = *table_freed || params.table_freed;
 
 error_unlock:
 	amdgpu_vm_eviction_unlock(vm);
@@ -1863,7 +1863,7 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
  * 0 for success, -EINVAL for failure.
  */
 int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
-			bool clear)
+			bool clear, bool *table_freed)
 {
 	struct amdgpu_bo *bo = bo_va->base.bo;
 	struct amdgpu_vm *vm = bo_va->base.vm;
@@ -1942,7 +1942,7 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 						resv, mapping->start,
 						mapping->last, update_flags,
 						mapping->offset, mem,
-						pages_addr, last_update, NULL);
+						pages_addr, last_update, table_freed);
 		if (r)
 			return r;
 	}
@@ -2194,7 +2194,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 
 	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
 		/* Per VM BOs never need to bo cleared in the page tables */
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			return r;
 	}
@@ -2213,7 +2213,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 		else
 			clear = true;
 
-		r = amdgpu_vm_bo_update(adev, bo_va, clear);
+		r = amdgpu_vm_bo_update(adev, bo_va, clear, NULL);
 		if (r)
 			return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index ee0028ec2c4f..80cc9ab2c1d0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -408,7 +408,7 @@ int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 				struct dma_fence **fence, bool *free_table);
 int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 			struct amdgpu_bo_va *bo_va,
-			bool clear);
+			bool clear, bool *table_freed);
 bool amdgpu_vm_evictable(struct amdgpu_bo *bo);
 void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
 			     struct amdgpu_bo *bo, bool evicted);
-- 
cgit v1.2.3


From 3b2b254425cc92c2c603efac488188d3084595ac Mon Sep 17 00:00:00 2001
From: Eric Huang <JinhuiEric.Huang@amd.com>
Date: Mon, 26 Jul 2021 11:27:53 -0400
Subject: Revert "Revert "drm/amdgpu: Fix warning of Function parameter or
 member not described""

This reverts commit 4e7b93ca52fb228b177168d436449c5671415a72.

Revert reason: The issue has been resolved.

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index f735282ba878..2a88ed5d983b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1856,6 +1856,7 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
  * @adev: amdgpu_device pointer
  * @bo_va: requested BO and VM object
  * @clear: if true clear the entries
+ * @table_freed: return true if page table is freed
  *
  * Fill in the page table entries for @bo_va.
  *
-- 
cgit v1.2.3


From cc6152ff4ff339f4d3a3f35d43d416010d62842d Mon Sep 17 00:00:00 2001
From: Eric Huang <JinhuiEric.Huang@amd.com>
Date: Mon, 26 Jul 2021 11:23:17 -0400
Subject: Revert "Revert "drm/amdgpu: Add table_freed parameter to
 amdgpu_vm_bo_update""

This reverts commit 024d8811c90ed56d8b90cdcf71e51c9fedeff460.

Revert reason: The issue has been resolved.

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c  |  6 +++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c  | 10 +++++-----
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h  |  2 +-
 4 files changed, 10 insertions(+), 10 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index a152363b0254..913f9eaa9cd6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -799,7 +799,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_vm_bo_update(adev, fpriv->prt_va, false);
+	r = amdgpu_vm_bo_update(adev, fpriv->prt_va, false, NULL);
 	if (r)
 		return r;
 
@@ -810,7 +810,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) {
 		bo_va = fpriv->csa_va;
 		BUG_ON(!bo_va);
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			return r;
 
@@ -829,7 +829,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 		if (bo_va == NULL)
 			continue;
 
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 611fd10c3410..c6f2fb9557ff 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -621,7 +621,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 
 	if (operation == AMDGPU_VA_OP_MAP ||
 	    operation == AMDGPU_VA_OP_REPLACE) {
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			goto error;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 6362e861a3f5..f735282ba878 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1798,7 +1798,7 @@ int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 	r = vm->update_funcs->commit(&params, fence);
 
 	if (table_freed)
-		*table_freed = params.table_freed;
+		*table_freed = *table_freed || params.table_freed;
 
 error_unlock:
 	amdgpu_vm_eviction_unlock(vm);
@@ -1863,7 +1863,7 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
  * 0 for success, -EINVAL for failure.
  */
 int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
-			bool clear)
+			bool clear, bool *table_freed)
 {
 	struct amdgpu_bo *bo = bo_va->base.bo;
 	struct amdgpu_vm *vm = bo_va->base.vm;
@@ -1942,7 +1942,7 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va,
 						resv, mapping->start,
 						mapping->last, update_flags,
 						mapping->offset, mem,
-						pages_addr, last_update, NULL);
+						pages_addr, last_update, table_freed);
 		if (r)
 			return r;
 	}
@@ -2194,7 +2194,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 
 	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
 		/* Per VM BOs never need to bo cleared in the page tables */
-		r = amdgpu_vm_bo_update(adev, bo_va, false);
+		r = amdgpu_vm_bo_update(adev, bo_va, false, NULL);
 		if (r)
 			return r;
 	}
@@ -2213,7 +2213,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 		else
 			clear = true;
 
-		r = amdgpu_vm_bo_update(adev, bo_va, clear);
+		r = amdgpu_vm_bo_update(adev, bo_va, clear, NULL);
 		if (r)
 			return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index ee0028ec2c4f..80cc9ab2c1d0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -408,7 +408,7 @@ int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 				struct dma_fence **fence, bool *free_table);
 int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 			struct amdgpu_bo_va *bo_va,
-			bool clear);
+			bool clear, bool *table_freed);
 bool amdgpu_vm_evictable(struct amdgpu_bo *bo);
 void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
 			     struct amdgpu_bo *bo, bool evicted);
-- 
cgit v1.2.3


From b1f21482affa6b4509077b0d58b6edc11f2ff75d Mon Sep 17 00:00:00 2001
From: Eric Huang <JinhuiEric.Huang@amd.com>
Date: Mon, 26 Jul 2021 11:27:53 -0400
Subject: Revert "Revert "drm/amdgpu: Fix warning of Function parameter or
 member not described""

This reverts commit 4e7b93ca52fb228b177168d436449c5671415a72.

Revert reason: The issue has been resolved.

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index f735282ba878..2a88ed5d983b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1856,6 +1856,7 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
  * @adev: amdgpu_device pointer
  * @bo_va: requested BO and VM object
  * @clear: if true clear the entries
+ * @table_freed: return true if page table is freed
  *
  * Fill in the page table entries for @bo_va.
  *
-- 
cgit v1.2.3


From c530b02f39850a639b72d01ebbf7e5d745c60831 Mon Sep 17 00:00:00 2001
From: Jack Zhang <Jack.Zhang1@amd.com>
Date: Wed, 12 May 2021 15:06:35 +0800
Subject: drm/amd/amdgpu embed hw_fence into amdgpu_job

Why: Previously hw fence is alloced separately with job.
It caused historical lifetime issues and corner cases.
The ideal situation is to take fence to manage both job
and fence's lifetime, and simplify the design of gpu-scheduler.

How:
We propose to embed hw_fence into amdgpu_job.
1. We cover the normal job submission by this method.
2. For ib_test, and submit without a parent job keep the
legacy way to create a hw fence separately.
v2:
use AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT to show that the fence is
embedded in a job.
v3:
remove redundant variable ring in amdgpu_job
v4:
add tdr sequence support for this feature. Add a job_run_counter to
indicate whether this job is a resubmit job.
v5
add missing handling in amdgpu_fence_enable_signaling

Signed-off-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Signed-off-by: Jack Zhang <Jack.Zhang7@hotmail.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c  |  1 -
 drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c  | 13 ++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c   | 86 ++++++++++++++++++++++-------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c      |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.c     | 39 +++++++++----
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.h     |  6 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h    |  5 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c      |  2 +-
 9 files changed, 119 insertions(+), 37 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 7b46ba551cb2..3003ee1c9487 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -714,7 +714,6 @@ int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
 	ret = dma_fence_wait(f, false);
 
 err_ib_sched:
-	dma_fence_put(f);
 	amdgpu_job_free(job);
 err:
 	return ret;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
index 536005bff24a..277128846dd1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
@@ -1414,7 +1414,7 @@ no_preempt:
 			continue;
 		}
 		job = to_amdgpu_job(s_job);
-		if (preempted && job->fence == fence)
+		if (preempted && (&job->hw_fence) == fence)
 			/* mark the job as preempted */
 			job->preemption_status |= AMDGPU_IB_PREEMPTED;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 65ccd1144863..395779ee44d0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -4449,7 +4449,7 @@ int amdgpu_device_mode1_reset(struct amdgpu_device *adev)
 int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 				 struct amdgpu_reset_context *reset_context)
 {
-	int i, r = 0;
+	int i, j, r = 0;
 	struct amdgpu_job *job = NULL;
 	bool need_full_reset =
 		test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
@@ -4473,6 +4473,17 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 		if (!ring || !ring->sched.thread)
 			continue;
 
+		/*clear job fence from fence drv to avoid force_completion
+		 *leave NULL and vm flush fence in fence drv */
+		for (j = 0; j <= ring->fence_drv.num_fences_mask; j++) {
+			struct dma_fence *old, **ptr;
+
+			ptr = &ring->fence_drv.fences[j];
+			old = rcu_dereference_protected(*ptr, 1);
+			if (old && test_bit(AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT, &old->flags)) {
+				RCU_INIT_POINTER(*ptr, NULL);
+			}
+		}
 		/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
 		amdgpu_fence_driver_force_completion(ring);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index 6ed53669f3e0..14499f0de32d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -129,30 +129,50 @@ static u32 amdgpu_fence_read(struct amdgpu_ring *ring)
  *
  * @ring: ring the fence is associated with
  * @f: resulting fence object
+ * @job: job the fence is embedded in
  * @flags: flags to pass into the subordinate .emit_fence() call
  *
  * Emits a fence command on the requested ring (all asics).
  * Returns 0 on success, -ENOMEM on failure.
  */
-int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f,
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, struct amdgpu_job *job,
 		      unsigned flags)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_fence *fence;
+	struct dma_fence *fence;
+	struct amdgpu_fence *am_fence;
 	struct dma_fence __rcu **ptr;
 	uint32_t seq;
 	int r;
 
-	fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_KERNEL);
-	if (fence == NULL)
-		return -ENOMEM;
+	if (job == NULL) {
+		/* create a sperate hw fence */
+		am_fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_ATOMIC);
+		if (am_fence == NULL)
+			return -ENOMEM;
+		fence = &am_fence->base;
+		am_fence->ring = ring;
+	} else {
+		/* take use of job-embedded fence */
+		fence = &job->hw_fence;
+	}
 
 	seq = ++ring->fence_drv.sync_seq;
-	fence->ring = ring;
-	dma_fence_init(&fence->base, &amdgpu_fence_ops,
-		       &ring->fence_drv.lock,
-		       adev->fence_context + ring->idx,
-		       seq);
+	if (job != NULL && job->job_run_counter) {
+		/* reinit seq for resubmitted jobs */
+		fence->seqno = seq;
+	} else {
+		dma_fence_init(fence, &amdgpu_fence_ops,
+				&ring->fence_drv.lock,
+				adev->fence_context + ring->idx,
+				seq);
+	}
+
+	if (job != NULL) {
+		/* mark this fence has a parent job */
+		set_bit(AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT, &fence->flags);
+	}
+
 	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
 			       seq, flags | AMDGPU_FENCE_FLAG_INT);
 	pm_runtime_get_noresume(adev_to_drm(adev)->dev);
@@ -175,9 +195,9 @@ int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f,
 	/* This function can't be called concurrently anyway, otherwise
 	 * emitting the fence would mess up the hardware ring buffer.
 	 */
-	rcu_assign_pointer(*ptr, dma_fence_get(&fence->base));
+	rcu_assign_pointer(*ptr, dma_fence_get(fence));
 
-	*f = &fence->base;
+	*f = fence;
 
 	return 0;
 }
@@ -621,8 +641,16 @@ static const char *amdgpu_fence_get_driver_name(struct dma_fence *fence)
 
 static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f)
 {
-	struct amdgpu_fence *fence = to_amdgpu_fence(f);
-	return (const char *)fence->ring->name;
+	struct amdgpu_ring *ring;
+
+	if (test_bit(AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT, &f->flags)) {
+		struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence);
+
+		ring = to_amdgpu_ring(job->base.sched);
+	} else {
+		ring = to_amdgpu_fence(f)->ring;
+	}
+	return (const char *)ring->name;
 }
 
 /**
@@ -635,13 +663,20 @@ static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f)
  */
 static bool amdgpu_fence_enable_signaling(struct dma_fence *f)
 {
-	struct amdgpu_fence *fence = to_amdgpu_fence(f);
-	struct amdgpu_ring *ring = fence->ring;
+	struct amdgpu_ring *ring;
+
+	if (test_bit(AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT, &f->flags)) {
+		struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence);
+
+		ring = to_amdgpu_ring(job->base.sched);
+	} else {
+		ring = to_amdgpu_fence(f)->ring;
+	}
 
 	if (!timer_pending(&ring->fence_drv.fallback_timer))
 		amdgpu_fence_schedule_fallback(ring);
 
-	DMA_FENCE_TRACE(&fence->base, "armed on ring %i!\n", ring->idx);
+	DMA_FENCE_TRACE(f, "armed on ring %i!\n", ring->idx);
 
 	return true;
 }
@@ -656,8 +691,20 @@ static bool amdgpu_fence_enable_signaling(struct dma_fence *f)
 static void amdgpu_fence_free(struct rcu_head *rcu)
 {
 	struct dma_fence *f = container_of(rcu, struct dma_fence, rcu);
-	struct amdgpu_fence *fence = to_amdgpu_fence(f);
-	kmem_cache_free(amdgpu_fence_slab, fence);
+
+	if (test_bit(AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT, &f->flags)) {
+	/* free job if fence has a parent job */
+		struct amdgpu_job *job;
+
+		job = container_of(f, struct amdgpu_job, hw_fence);
+		kfree(job);
+	} else {
+	/* free fence_slab if it's separated fence*/
+		struct amdgpu_fence *fence;
+
+		fence = to_amdgpu_fence(f);
+		kmem_cache_free(amdgpu_fence_slab, fence);
+	}
 }
 
 /**
@@ -680,6 +727,7 @@ static const struct dma_fence_ops amdgpu_fence_ops = {
 	.release = amdgpu_fence_release,
 };
 
+
 /*
  * Fence debugfs
  */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index ec65ab0ddf89..c076a6b9a5a2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -262,7 +262,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 				       fence_flags | AMDGPU_FENCE_FLAG_64BIT);
 	}
 
-	r = amdgpu_fence_emit(ring, f, fence_flags);
+	r = amdgpu_fence_emit(ring, f, job, fence_flags);
 	if (r) {
 		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
 		if (job && job->vmid)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index d33e6d97cc89..de29518673dd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -127,11 +127,16 @@ void amdgpu_job_free_resources(struct amdgpu_job *job)
 {
 	struct amdgpu_ring *ring = to_amdgpu_ring(job->base.sched);
 	struct dma_fence *f;
+	struct dma_fence *hw_fence;
 	unsigned i;
 
-	/* use sched fence if available */
-	f = job->base.s_fence ? &job->base.s_fence->finished : job->fence;
+	if (job->hw_fence.ops == NULL)
+		hw_fence = job->external_hw_fence;
+	else
+		hw_fence = &job->hw_fence;
 
+	/* use sched fence if available */
+	f = job->base.s_fence ? &job->base.s_fence->finished : hw_fence;
 	for (i = 0; i < job->num_ibs; ++i)
 		amdgpu_ib_free(ring->adev, &job->ibs[i], f);
 }
@@ -142,20 +147,27 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job)
 
 	drm_sched_job_cleanup(s_job);
 
-	dma_fence_put(job->fence);
 	amdgpu_sync_free(&job->sync);
 	amdgpu_sync_free(&job->sched_sync);
-	kfree(job);
+
+    /* only put the hw fence if has embedded fence */
+	if (job->hw_fence.ops != NULL)
+		dma_fence_put(&job->hw_fence);
+	else
+		kfree(job);
 }
 
 void amdgpu_job_free(struct amdgpu_job *job)
 {
 	amdgpu_job_free_resources(job);
-
-	dma_fence_put(job->fence);
 	amdgpu_sync_free(&job->sync);
 	amdgpu_sync_free(&job->sched_sync);
-	kfree(job);
+
+	/* only put the hw fence if has embedded fence */
+	if (job->hw_fence.ops != NULL)
+		dma_fence_put(&job->hw_fence);
+	else
+		kfree(job);
 }
 
 int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
@@ -184,11 +196,14 @@ int amdgpu_job_submit_direct(struct amdgpu_job *job, struct amdgpu_ring *ring,
 
 	job->base.sched = &ring->sched;
 	r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, NULL, fence);
-	job->fence = dma_fence_get(*fence);
+	/* record external_hw_fence for direct submit */
+	job->external_hw_fence = dma_fence_get(*fence);
 	if (r)
 		return r;
 
 	amdgpu_job_free(job);
+	dma_fence_put(*fence);
+
 	return 0;
 }
 
@@ -246,10 +261,12 @@ static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 		if (r)
 			DRM_ERROR("Error scheduling IBs (%d)\n", r);
 	}
-	/* if gpu reset, hw fence will be replaced here */
-	dma_fence_put(job->fence);
-	job->fence = dma_fence_get(fence);
 
+	if (!job->job_run_counter)
+		dma_fence_get(fence);
+	else if (finished->error < 0)
+		dma_fence_put(&job->hw_fence);
+	job->job_run_counter++;
 	amdgpu_job_free_resources(job);
 
 	fence = r ? ERR_PTR(r) : fence;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
index 81caac9b958a..9e65730193b8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
@@ -46,7 +46,8 @@ struct amdgpu_job {
 	struct amdgpu_sync	sync;
 	struct amdgpu_sync	sched_sync;
 	struct amdgpu_ib	*ibs;
-	struct dma_fence	*fence; /* the hw fence */
+	struct dma_fence	hw_fence;
+	struct dma_fence	*external_hw_fence;
 	uint32_t		preamble_status;
 	uint32_t                preemption_status;
 	uint32_t		num_ibs;
@@ -62,6 +63,9 @@ struct amdgpu_job {
 	/* user fence handling */
 	uint64_t		uf_addr;
 	uint64_t		uf_sequence;
+
+	/* job_run_counter >= 1 means a resubmit job */
+	uint32_t		job_run_counter;
 };
 
 int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
index 9c11ced4312c..e713d31619fe 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
@@ -48,6 +48,9 @@
 #define AMDGPU_FENCE_FLAG_INT           (1 << 1)
 #define AMDGPU_FENCE_FLAG_TC_WB_ONLY    (1 << 2)
 
+/* fence flag bit to indicate the face is embedded in job*/
+#define AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT		(DMA_FENCE_FLAG_USER_BITS + 1)
+
 #define to_amdgpu_ring(s) container_of((s), struct amdgpu_ring, sched)
 
 #define AMDGPU_IB_POOL_SIZE	(1024 * 1024)
@@ -118,7 +121,7 @@ void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev);
 void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev);
 int amdgpu_fence_driver_sw_init(struct amdgpu_device *adev);
 void amdgpu_fence_driver_sw_fini(struct amdgpu_device *adev);
-int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence,
+int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence, struct amdgpu_job *job,
 		      unsigned flags);
 int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s,
 			      uint32_t timeout);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 2a88ed5d983b..2af8860d74cc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1218,7 +1218,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
 		amdgpu_gmc_emit_pasid_mapping(ring, job->vmid, job->pasid);
 
 	if (vm_flush_needed || pasid_mapping_needed) {
-		r = amdgpu_fence_emit(ring, &fence, 0);
+		r = amdgpu_fence_emit(ring, &fence, NULL, 0);
 		if (r)
 			return r;
 	}
-- 
cgit v1.2.3


From ff891a2e6431f79c239ab0c5bc05834c0235821c Mon Sep 17 00:00:00 2001
From: Philip Yang <Philip.Yang@amd.com>
Date: Sun, 15 Aug 2021 14:42:33 -0400
Subject: drm/amdkfd: check access permisson to restore retry fault

Check range access permission to restore GPU retry fault, if GPU retry
fault on address which belongs to VMA, and VMA has no read or write
permission requested by GPU, failed to restore the address. The vm fault
event will pass back to user space.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c |  5 +++--
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h |  2 +-
 drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c |  3 ++-
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c  |  3 ++-
 drivers/gpu/drm/amd/amdkfd/kfd_svm.c   | 29 ++++++++++++++++++++++++++++-
 drivers/gpu/drm/amd/amdkfd/kfd_svm.h   |  5 +++--
 6 files changed, 39 insertions(+), 8 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 2af8860d74cc..0b893aa21ffe 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -3345,12 +3345,13 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
  * @adev: amdgpu device pointer
  * @pasid: PASID of the VM
  * @addr: Address of the fault
+ * @write_fault: true is write fault, false is read fault
  *
  * Try to gracefully handle a VM fault. Return true if the fault was handled and
  * shouldn't be reported any more.
  */
 bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
-			    uint64_t addr)
+			    uint64_t addr, bool write_fault)
 {
 	bool is_compute_context = false;
 	struct amdgpu_bo *root;
@@ -3375,7 +3376,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 	addr /= AMDGPU_GPU_PAGE_SIZE;
 
 	if (is_compute_context &&
-	    !svm_range_restore_pages(adev, pasid, addr)) {
+	    !svm_range_restore_pages(adev, pasid, addr, write_fault)) {
 		amdgpu_bo_unref(&root);
 		return true;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 80cc9ab2c1d0..85fcfb8c5efd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -448,7 +448,7 @@ void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev);
 void amdgpu_vm_get_task_info(struct amdgpu_device *adev, u32 pasid,
 			     struct amdgpu_task_info *task_info);
 bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
-			    uint64_t addr);
+			    uint64_t addr, bool write_fault);
 
 void amdgpu_vm_set_task_info(struct amdgpu_vm *vm);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
index 24b781e90bef..41c3a0d70b7c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@@ -93,6 +93,7 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev,
 				       struct amdgpu_iv_entry *entry)
 {
 	bool retry_fault = !!(entry->src_data[1] & 0x80);
+	bool write_fault = !!(entry->src_data[1] & 0x20);
 	struct amdgpu_vmhub *hub = &adev->vmhub[entry->vmid_src];
 	struct amdgpu_task_info task_info;
 	uint32_t status = 0;
@@ -121,7 +122,7 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev,
 		/* Try to handle the recoverable page faults by filling page
 		 * tables
 		 */
-		if (amdgpu_vm_handle_fault(adev, entry->pasid, addr))
+		if (amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault))
 			return 1;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 085fab45245d..d90c16a6b2b8 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -507,6 +507,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 				      struct amdgpu_iv_entry *entry)
 {
 	bool retry_fault = !!(entry->src_data[1] & 0x80);
+	bool write_fault = !!(entry->src_data[1] & 0x20);
 	uint32_t status = 0, cid = 0, rw = 0;
 	struct amdgpu_task_info task_info;
 	struct amdgpu_vmhub *hub;
@@ -537,7 +538,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 		/* Try to handle the recoverable page faults by filling page
 		 * tables
 		 */
-		if (amdgpu_vm_handle_fault(adev, entry->pasid, addr))
+		if (amdgpu_vm_handle_fault(adev, entry->pasid, addr, write_fault))
 			return 1;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index ecc390c2f28c..e4bdfd77669e 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -2400,9 +2400,29 @@ svm_range_count_fault(struct amdgpu_device *adev, struct kfd_process *p,
 		WRITE_ONCE(pdd->faults, pdd->faults + 1);
 }
 
+static bool
+svm_fault_allowed(struct mm_struct *mm, uint64_t addr, bool write_fault)
+{
+	unsigned long requested = VM_READ;
+	struct vm_area_struct *vma;
+
+	if (write_fault)
+		requested |= VM_WRITE;
+
+	vma = find_vma(mm, addr << PAGE_SHIFT);
+	if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
+		pr_debug("address 0x%llx VMA is removed\n", addr);
+		return true;
+	}
+
+	pr_debug("requested 0x%lx, vma permission flags 0x%lx\n", requested,
+		vma->vm_flags);
+	return (vma->vm_flags & requested) == requested;
+}
+
 int
 svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
-			uint64_t addr)
+			uint64_t addr, bool write_fault)
 {
 	struct mm_struct *mm = NULL;
 	struct svm_range_list *svms;
@@ -2484,6 +2504,13 @@ retry_write_locked:
 		goto out_unlock_range;
 	}
 
+	if (!svm_fault_allowed(mm, addr, write_fault)) {
+		pr_debug("fault addr 0x%llx no %s permission\n", addr,
+			write_fault ? "write" : "read");
+		r = -EPERM;
+		goto out_unlock_range;
+	}
+
 	best_loc = svm_range_best_restore_location(prange, adev, &gpuidx);
 	if (best_loc == -1) {
 		pr_debug("svms %p failed get best restore loc [0x%lx 0x%lx]\n",
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
index 3fc1fd8b4fbc..c6ec55354c7b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
@@ -175,7 +175,7 @@ int svm_range_split_by_granularity(struct kfd_process *p, struct mm_struct *mm,
 			       unsigned long addr, struct svm_range *parent,
 			       struct svm_range *prange);
 int svm_range_restore_pages(struct amdgpu_device *adev,
-			    unsigned int pasid, uint64_t addr);
+			    unsigned int pasid, uint64_t addr, bool write_fault);
 int svm_range_schedule_evict_svm_bo(struct amdgpu_amdkfd_fence *fence);
 void svm_range_add_list_work(struct svm_range_list *svms,
 			     struct svm_range *prange, struct mm_struct *mm,
@@ -209,7 +209,8 @@ static inline void svm_range_list_fini(struct kfd_process *p)
 }
 
 static inline int svm_range_restore_pages(struct amdgpu_device *adev,
-					  unsigned int pasid, uint64_t addr)
+					  unsigned int pasid, uint64_t addr,
+					  bool write_fault)
 {
 	return -EFAULT;
 }
-- 
cgit v1.2.3


From d035f84d834cd09d0c083f4d4cb3e46c2ae56cb2 Mon Sep 17 00:00:00 2001
From: Yifan Zhang <yifan1.zhang@amd.com>
Date: Wed, 25 Aug 2021 17:05:09 +0800
Subject: drm/amdgpu: rename amdgpu_bo_get_preferred_pin_domain
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

amdgpu_bo_get_preferred_pin_domain is used for page tables
creation, which is not involved with page pinning. And it is used in
more cases than display scanout, modify its documentation as well.

Signed-off-by: Yifan Zhang <yifan1.zhang@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_gem.c    | 2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 8 ++++----
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.h | 2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c     | 2 +-
 4 files changed, 7 insertions(+), 7 deletions(-)

(limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c')

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index c6f2fb9557ff..cb07cc3b06ed 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -904,7 +904,7 @@ int amdgpu_mode_dumb_create(struct drm_file *file_priv,
 					 DIV_ROUND_UP(args->bpp, 8), 0);
 	args->size = (u64)args->pitch * args->height;
 	args->size = ALIGN(args->size, PAGE_SIZE);
-	domain = amdgpu_bo_get_preferred_pin_domain(adev,
+	domain = amdgpu_bo_get_preferred_domain(adev,
 				amdgpu_display_supported_domains(adev, flags));
 	r = amdgpu_gem_object_create(adev, args->size, 0, domain, flags,
 				     ttm_bo_type_device, NULL, &gobj);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 7734c10ae74e..01a78c786536 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -947,7 +947,7 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
 	/* This assumes only APU display buffers are pinned with (VRAM|GTT).
 	 * See function amdgpu_display_supported_domains()
 	 */
-	domain = amdgpu_bo_get_preferred_pin_domain(adev, domain);
+	domain = amdgpu_bo_get_preferred_domain(adev, domain);
 
 	if (bo->tbo.base.import_attach)
 		dma_buf_pin(bo->tbo.base.import_attach);
@@ -1518,14 +1518,14 @@ u64 amdgpu_bo_gpu_offset_no_check(struct amdgpu_bo *bo)
 }
 
 /**
- * amdgpu_bo_get_preferred_pin_domain - get preferred domain for scanout
+ * amdgpu_bo_get_preferred_domain - get preferred domain
  * @adev: amdgpu device object
  * @domain: allowed :ref:`memory domains <amdgpu_memory_domains>`
  *
  * Returns:
- * Which of the allowed domains is preferred for pinning the BO for scanout.
+ * Which of the allowed domains is preferred for allocating the BO.
  */
-uint32_t amdgpu_bo_get_preferred_pin_domain(struct amdgpu_device *adev,
+uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev,
 					    uint32_t domain)
 {
 	if (domain == (AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT)) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index e72f329e7f18..9d6c001c15f8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -333,7 +333,7 @@ void amdgpu_bo_get_memory(struct amdgpu_bo *bo, uint64_t *vram_mem,
 void amdgpu_bo_add_to_shadow_list(struct amdgpu_bo_vm *vmbo);
 int amdgpu_bo_restore_shadow(struct amdgpu_bo *shadow,
 			     struct dma_fence **fence);
-uint32_t amdgpu_bo_get_preferred_pin_domain(struct amdgpu_device *adev,
+uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev,
 					    uint32_t domain);
 
 /*
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 0b893aa21ffe..6b15cad78de9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -926,7 +926,7 @@ static int amdgpu_vm_pt_create(struct amdgpu_device *adev,
 	bp.size = amdgpu_vm_bo_size(adev, level);
 	bp.byte_align = AMDGPU_GPU_PAGE_SIZE;
 	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
-	bp.domain = amdgpu_bo_get_preferred_pin_domain(adev, bp.domain);
+	bp.domain = amdgpu_bo_get_preferred_domain(adev, bp.domain);
 	bp.flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
 		AMDGPU_GEM_CREATE_CPU_GTT_USWC;
 
-- 
cgit v1.2.3