From b85891bd6d1bf887b3398f4c44b7a30b37f4485e Mon Sep 17 00:00:00 2001
From: Junwei Zhang <Jerry.Zhang@amd.com>
Date: Mon, 16 Jan 2017 13:59:01 +0800
Subject: drm/amdgpu: IOCTL interface for PRT support v4
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Till GFX8 we can only enable PRT support globally, but with the next hardware
generation we can do this on a per page basis.

Keep the interface consistent by adding PRT mappings and enable
support globally on current hardware when the first mapping is made.

v2: disable PRT support delayed and on all error paths
v3: PRT and other permissions are mutal exclusive,
    PRT mappings don't need a BO.
v4: update PRT mappings durign CS as well, make va_flags 64bit

Signed-off-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 62 ++++++++++++++++++++-------------
 1 file changed, 38 insertions(+), 24 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 106cf83c2e6b..3c22656aa1bf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -553,6 +553,12 @@ error:
 int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 			  struct drm_file *filp)
 {
+	const uint32_t valid_flags = AMDGPU_VM_DELAY_UPDATE |
+		AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_PAGE_WRITEABLE |
+		AMDGPU_VM_PAGE_EXECUTABLE;
+	const uint32_t prt_flags = AMDGPU_VM_DELAY_UPDATE |
+		AMDGPU_VM_PAGE_PRT;
+
 	struct drm_amdgpu_gem_va *args = data;
 	struct drm_gem_object *gobj;
 	struct amdgpu_device *adev = dev->dev_private;
@@ -563,7 +569,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	struct ttm_validate_buffer tv;
 	struct ww_acquire_ctx ticket;
 	struct list_head list;
-	uint32_t invalid_flags, va_flags = 0;
+	uint64_t va_flags = 0;
 	int r = 0;
 
 	if (!adev->vm_manager.enabled)
@@ -577,11 +583,9 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 		return -EINVAL;
 	}
 
-	invalid_flags = ~(AMDGPU_VM_DELAY_UPDATE | AMDGPU_VM_PAGE_READABLE |
-			AMDGPU_VM_PAGE_WRITEABLE | AMDGPU_VM_PAGE_EXECUTABLE);
-	if ((args->flags & invalid_flags)) {
-		dev_err(&dev->pdev->dev, "invalid flags 0x%08X vs 0x%08X\n",
-			args->flags, invalid_flags);
+	if ((args->flags & ~valid_flags) && (args->flags & ~prt_flags)) {
+		dev_err(&dev->pdev->dev, "invalid flags combination 0x%08X\n",
+			args->flags);
 		return -EINVAL;
 	}
 
@@ -595,28 +599,34 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 		return -EINVAL;
 	}
 
-	gobj = drm_gem_object_lookup(filp, args->handle);
-	if (gobj == NULL)
-		return -ENOENT;
-	abo = gem_to_amdgpu_bo(gobj);
 	INIT_LIST_HEAD(&list);
-	tv.bo = &abo->tbo;
-	tv.shared = false;
-	list_add(&tv.head, &list);
+	if (!(args->flags & AMDGPU_VM_PAGE_PRT)) {
+		gobj = drm_gem_object_lookup(filp, args->handle);
+		if (gobj == NULL)
+			return -ENOENT;
+		abo = gem_to_amdgpu_bo(gobj);
+		tv.bo = &abo->tbo;
+		tv.shared = false;
+		list_add(&tv.head, &list);
+	} else {
+		gobj = NULL;
+		abo = NULL;
+	}
 
 	amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
 
 	r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL);
-	if (r) {
-		drm_gem_object_unreference_unlocked(gobj);
-		return r;
-	}
+	if (r)
+		goto error_unref;
 
-	bo_va = amdgpu_vm_bo_find(&fpriv->vm, abo);
-	if (!bo_va) {
-		ttm_eu_backoff_reservation(&ticket, &list);
-		drm_gem_object_unreference_unlocked(gobj);
-		return -ENOENT;
+	if (abo) {
+		bo_va = amdgpu_vm_bo_find(&fpriv->vm, abo);
+		if (!bo_va) {
+			r = -ENOENT;
+			goto error_backoff;
+		}
+	} else {
+		bo_va = fpriv->prt_va;
 	}
 
 	switch (args->operation) {
@@ -627,6 +637,8 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 			va_flags |= AMDGPU_PTE_WRITEABLE;
 		if (args->flags & AMDGPU_VM_PAGE_EXECUTABLE)
 			va_flags |= AMDGPU_PTE_EXECUTABLE;
+		if (args->flags & AMDGPU_VM_PAGE_PRT)
+			va_flags |= AMDGPU_PTE_PRT;
 		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,
 				     args->offset_in_bo, args->map_size,
 				     va_flags);
@@ -637,11 +649,13 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	default:
 		break;
 	}
-	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) &&
-	    !amdgpu_vm_debug)
+	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) && !amdgpu_vm_debug)
 		amdgpu_gem_va_update_vm(adev, bo_va, &list, args->operation);
+
+error_backoff:
 	ttm_eu_backoff_reservation(&ticket, &list);
 
+error_unref:
 	drm_gem_object_unreference_unlocked(gobj);
 	return r;
 }
-- 
cgit 


From 5463545b9280093f4614049df2dfee3f9e662172 Mon Sep 17 00:00:00 2001
From: Alex Xie <AlexBin.Xie@amd.com>
Date: Tue, 14 Feb 2017 12:22:57 -0500
Subject: drm/amdgpu: add a callback to set vm mapping flags
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This lets each asic set whichever flags it supports.

Signed-off-by: Alex Xie <AlexBin.Xie@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@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.h     |  4 ++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 12 +++---------
 drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c   | 16 ++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c   | 16 ++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c   | 18 ++++++++++++++++++
 5 files changed, 57 insertions(+), 9 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index abf2b39fd968..00553bf26e66 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -296,6 +296,9 @@ struct amdgpu_gart_funcs {
 			   uint64_t flags); /* access flags */
 	/* enable/disable PRT support */
 	void (*set_prt)(struct amdgpu_device *adev, bool enable);
+	/* set pte flags based per asic */
+	uint64_t (*get_vm_pte_flags)(struct amdgpu_device *adev,
+				     uint32_t flags);
 };
 
 /* provided by the ih block */
@@ -1682,6 +1685,7 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_vm_copy_pte(adev, ib, pe, src, count) ((adev)->vm_manager.vm_pte_funcs->copy_pte((ib), (pe), (src), (count)))
 #define amdgpu_vm_write_pte(adev, ib, pe, value, count, incr) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pe), (value), (count), (incr)))
 #define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
+#define amdgpu_vm_get_pte_flags(adev, flags) (adev)->gart.gart_funcs->get_vm_pte_flags((adev),(flags))
 #define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
 #define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
 #define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t))
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 3c22656aa1bf..6daf004955b0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -569,7 +569,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	struct ttm_validate_buffer tv;
 	struct ww_acquire_ctx ticket;
 	struct list_head list;
-	uint64_t va_flags = 0;
+	uint64_t va_flags;
 	int r = 0;
 
 	if (!adev->vm_manager.enabled)
@@ -631,14 +631,8 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 
 	switch (args->operation) {
 	case AMDGPU_VA_OP_MAP:
-		if (args->flags & AMDGPU_VM_PAGE_READABLE)
-			va_flags |= AMDGPU_PTE_READABLE;
-		if (args->flags & AMDGPU_VM_PAGE_WRITEABLE)
-			va_flags |= AMDGPU_PTE_WRITEABLE;
-		if (args->flags & AMDGPU_VM_PAGE_EXECUTABLE)
-			va_flags |= AMDGPU_PTE_EXECUTABLE;
-		if (args->flags & AMDGPU_VM_PAGE_PRT)
-			va_flags |= AMDGPU_PTE_PRT;
+		va_flags = amdgpu_vm_get_pte_flags(adev, args->flags);
+
 		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,
 				     args->offset_in_bo, args->map_size,
 				     va_flags);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
index 6a1f566632be..35f74c6d1f29 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
@@ -379,6 +379,21 @@ static int gmc_v6_0_gart_set_pte_pde(struct amdgpu_device *adev,
 	return 0;
 }
 
+static uint64_t gmc_v6_0_get_vm_pte_flags(struct amdgpu_device *adev,
+					  uint32_t flags)
+{
+	uint64_t pte_flag = 0;
+
+	if (flags & AMDGPU_VM_PAGE_READABLE)
+		pte_flag |= AMDGPU_PTE_READABLE;
+	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
+		pte_flag |= AMDGPU_PTE_WRITEABLE;
+	if (flags & AMDGPU_VM_PAGE_PRT)
+		pte_flag |= AMDGPU_PTE_PRT;
+
+	return pte_flag;
+}
+
 static void gmc_v6_0_set_fault_enable_default(struct amdgpu_device *adev,
 					      bool value)
 {
@@ -1138,6 +1153,7 @@ static const struct amdgpu_gart_funcs gmc_v6_0_gart_funcs = {
 	.flush_gpu_tlb = gmc_v6_0_gart_flush_gpu_tlb,
 	.set_pte_pde = gmc_v6_0_gart_set_pte_pde,
 	.set_prt = gmc_v6_0_set_prt,
+	.get_vm_pte_flags = gmc_v6_0_get_vm_pte_flags
 };
 
 static const struct amdgpu_irq_src_funcs gmc_v6_0_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
index 9f41d9fa5ed5..dc9b6d62d1ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -451,6 +451,21 @@ static int gmc_v7_0_gart_set_pte_pde(struct amdgpu_device *adev,
 	return 0;
 }
 
+static uint64_t gmc_v7_0_get_vm_pte_flags(struct amdgpu_device *adev,
+					  uint32_t flags)
+{
+	uint64_t pte_flag = 0;
+
+	if (flags & AMDGPU_VM_PAGE_READABLE)
+		pte_flag |= AMDGPU_PTE_READABLE;
+	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
+		pte_flag |= AMDGPU_PTE_WRITEABLE;
+	if (flags & AMDGPU_VM_PAGE_PRT)
+		pte_flag |= AMDGPU_PTE_PRT;
+
+	return pte_flag;
+}
+
 /**
  * gmc_v8_0_set_fault_enable_default - update VM fault handling
  *
@@ -1323,6 +1338,7 @@ static const struct amdgpu_gart_funcs gmc_v7_0_gart_funcs = {
 	.flush_gpu_tlb = gmc_v7_0_gart_flush_gpu_tlb,
 	.set_pte_pde = gmc_v7_0_gart_set_pte_pde,
 	.set_prt = gmc_v7_0_set_prt,
+	.get_vm_pte_flags = gmc_v7_0_get_vm_pte_flags
 };
 
 static const struct amdgpu_irq_src_funcs gmc_v7_0_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index 233ac3504fe9..c087b00598cd 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -563,6 +563,23 @@ static int gmc_v8_0_gart_set_pte_pde(struct amdgpu_device *adev,
 	return 0;
 }
 
+static uint64_t gmc_v8_0_get_vm_pte_flags(struct amdgpu_device *adev,
+					  uint32_t flags)
+{
+	uint64_t pte_flag = 0;
+
+	if (flags & AMDGPU_VM_PAGE_EXECUTABLE)
+		pte_flag |= AMDGPU_PTE_EXECUTABLE;
+	if (flags & AMDGPU_VM_PAGE_READABLE)
+		pte_flag |= AMDGPU_PTE_READABLE;
+	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
+		pte_flag |= AMDGPU_PTE_WRITEABLE;
+	if (flags & AMDGPU_VM_PAGE_PRT)
+		pte_flag |= AMDGPU_PTE_PRT;
+
+	return pte_flag;
+}
+
 /**
  * gmc_v8_0_set_fault_enable_default - update VM fault handling
  *
@@ -1562,6 +1579,7 @@ static const struct amdgpu_gart_funcs gmc_v8_0_gart_funcs = {
 	.flush_gpu_tlb = gmc_v8_0_gart_flush_gpu_tlb,
 	.set_pte_pde = gmc_v8_0_gart_set_pte_pde,
 	.set_prt = gmc_v8_0_set_prt,
+	.get_vm_pte_flags = gmc_v8_0_get_vm_pte_flags
 };
 
 static const struct amdgpu_irq_src_funcs gmc_v8_0_irq_funcs = {
-- 
cgit 


From 663e4577a5733fab18d601128f54486d78595bc0 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Mon, 13 Mar 2017 10:13:37 +0100
Subject: drm/amdgpu: separate page table allocation from mapping
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This makes it easier to implement a replace operation.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c  |   6 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c |   9 ++
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c   | 136 +++++++++++++++++--------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h   |   3 +
 4 files changed, 92 insertions(+), 62 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 6daf004955b0..0240f108f90e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -631,8 +631,12 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 
 	switch (args->operation) {
 	case AMDGPU_VA_OP_MAP:
-		va_flags = amdgpu_vm_get_pte_flags(adev, args->flags);
+		r = amdgpu_vm_alloc_pts(adev, bo_va->vm, args->va_address,
+					args->map_size);
+		if (r)
+			goto error_backoff;
 
+		va_flags = amdgpu_vm_get_pte_flags(adev, args->flags);
 		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,
 				     args->offset_in_bo, args->map_size,
 				     va_flags);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
index 2feb9aa3b1f4..ecef35a1fe33 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
@@ -75,6 +75,15 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 		return -ENOMEM;
 	}
 
+	r = amdgpu_vm_alloc_pts(adev, bo_va->vm, AMDGPU_CSA_VADDR,
+				   AMDGPU_CSA_SIZE);
+	if (r) {
+		DRM_ERROR("failed to allocate pts for static CSA, err=%d\n", r);
+		amdgpu_vm_bo_rmv(adev, bo_va);
+		ttm_eu_backoff_reservation(&ticket, &list);
+		return r;
+	}
+
 	r = amdgpu_vm_bo_map(adev, bo_va, AMDGPU_CSA_VADDR, 0,AMDGPU_CSA_SIZE,
 						AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
 						AMDGPU_PTE_EXECUTABLE);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index f9f4e2055084..296e985d0b65 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -187,6 +187,78 @@ void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev,
 	spin_unlock(&glob->lru_lock);
 }
 
+/**
+ * amdgpu_vm_alloc_pts - Allocate page tables.
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: VM to allocate page tables for
+ * @saddr: Start address which needs to be allocated
+ * @size: Size from start address we need.
+ *
+ * Make sure the page tables are allocated.
+ */
+int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
+			struct amdgpu_vm *vm,
+			uint64_t saddr, uint64_t size)
+{
+	unsigned last_pfn, pt_idx;
+	uint64_t eaddr;
+	int r;
+
+	/* validate the parameters */
+	if (saddr & AMDGPU_GPU_PAGE_MASK || size & AMDGPU_GPU_PAGE_MASK)
+		return -EINVAL;
+
+	eaddr = saddr + size - 1;
+	last_pfn = eaddr / AMDGPU_GPU_PAGE_SIZE;
+	if (last_pfn >= adev->vm_manager.max_pfn) {
+		dev_err(adev->dev, "va above limit (0x%08X >= 0x%08X)\n",
+			last_pfn, adev->vm_manager.max_pfn);
+		return -EINVAL;
+	}
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	saddr >>= amdgpu_vm_block_size;
+	eaddr >>= amdgpu_vm_block_size;
+
+	BUG_ON(eaddr >= amdgpu_vm_num_pdes(adev));
+
+	if (eaddr > vm->max_pde_used)
+		vm->max_pde_used = eaddr;
+
+	/* walk over the address space and allocate the page tables */
+	for (pt_idx = saddr; pt_idx <= eaddr; ++pt_idx) {
+		struct reservation_object *resv = vm->page_directory->tbo.resv;
+		struct amdgpu_bo *pt;
+
+		if (vm->page_tables[pt_idx].bo)
+			continue;
+
+		r = amdgpu_bo_create(adev, AMDGPU_VM_PTE_COUNT * 8,
+				     AMDGPU_GPU_PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_VRAM,
+				     AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
+				     AMDGPU_GEM_CREATE_SHADOW |
+				     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
+				     AMDGPU_GEM_CREATE_VRAM_CLEARED,
+				     NULL, resv, &pt);
+		if (r)
+			return r;
+
+		/* Keep a reference to the page table to avoid freeing
+		 * them up in the wrong order.
+		 */
+		pt->parent = amdgpu_bo_ref(vm->page_directory);
+
+		vm->page_tables[pt_idx].bo = pt;
+		vm->page_tables[pt_idx].addr = 0;
+	}
+
+	return 0;
+}
+
 static bool amdgpu_vm_is_gpu_reset(struct amdgpu_device *adev,
 			      struct amdgpu_vm_id *id)
 {
@@ -1442,9 +1514,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 	struct amdgpu_bo_va_mapping *mapping;
 	struct amdgpu_vm *vm = bo_va->vm;
 	struct interval_tree_node *it;
-	unsigned last_pfn, pt_idx;
 	uint64_t eaddr;
-	int r;
 
 	/* validate the parameters */
 	if (saddr & AMDGPU_GPU_PAGE_MASK || offset & AMDGPU_GPU_PAGE_MASK ||
@@ -1457,13 +1527,6 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 	    (bo_va->bo && offset + size > amdgpu_bo_size(bo_va->bo)))
 		return -EINVAL;
 
-	last_pfn = eaddr / AMDGPU_GPU_PAGE_SIZE;
-	if (last_pfn >= adev->vm_manager.max_pfn) {
-		dev_err(adev->dev, "va above limit (0x%08X >= 0x%08X)\n",
-			last_pfn, adev->vm_manager.max_pfn);
-		return -EINVAL;
-	}
-
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
 	eaddr /= AMDGPU_GPU_PAGE_SIZE;
 
@@ -1475,15 +1538,12 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		dev_err(adev->dev, "bo %p va 0x%010Lx-0x%010Lx conflict with "
 			"0x%010lx-0x%010lx\n", bo_va->bo, saddr, eaddr,
 			tmp->it.start, tmp->it.last + 1);
-		r = -EINVAL;
-		goto error;
+		return -EINVAL;
 	}
 
 	mapping = kmalloc(sizeof(*mapping), GFP_KERNEL);
-	if (!mapping) {
-		r = -ENOMEM;
-		goto error;
-	}
+	if (!mapping)
+		return -ENOMEM;
 
 	INIT_LIST_HEAD(&mapping->list);
 	mapping->it.start = saddr;
@@ -1494,56 +1554,10 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 	list_add(&mapping->list, &bo_va->invalids);
 	interval_tree_insert(&mapping->it, &vm->va);
 
-	/* Make sure the page tables are allocated */
-	saddr >>= amdgpu_vm_block_size;
-	eaddr >>= amdgpu_vm_block_size;
-
-	BUG_ON(eaddr >= amdgpu_vm_num_pdes(adev));
-
-	if (eaddr > vm->max_pde_used)
-		vm->max_pde_used = eaddr;
-
-	/* walk over the address space and allocate the page tables */
-	for (pt_idx = saddr; pt_idx <= eaddr; ++pt_idx) {
-		struct reservation_object *resv = vm->page_directory->tbo.resv;
-		struct amdgpu_bo *pt;
-
-		if (vm->page_tables[pt_idx].bo)
-			continue;
-
-		r = amdgpu_bo_create(adev, AMDGPU_VM_PTE_COUNT * 8,
-				     AMDGPU_GPU_PAGE_SIZE, true,
-				     AMDGPU_GEM_DOMAIN_VRAM,
-				     AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
-				     AMDGPU_GEM_CREATE_SHADOW |
-				     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
-				     AMDGPU_GEM_CREATE_VRAM_CLEARED,
-				     NULL, resv, &pt);
-		if (r)
-			goto error_free;
-
-		/* Keep a reference to the page table to avoid freeing
-		 * them up in the wrong order.
-		 */
-		pt->parent = amdgpu_bo_ref(vm->page_directory);
-
-		vm->page_tables[pt_idx].bo = pt;
-		vm->page_tables[pt_idx].addr = 0;
-	}
-
 	if (flags & AMDGPU_PTE_PRT)
 		amdgpu_vm_prt_get(adev);
 
 	return 0;
-
-error_free:
-	list_del(&mapping->list);
-	interval_tree_remove(&mapping->it, &vm->va);
-	trace_amdgpu_vm_bo_unmap(bo_va, mapping);
-	amdgpu_vm_free_mapping(adev, vm, mapping, NULL);
-
-error:
-	return r;
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 57eaf71a5b1c..1e5a3b2c7927 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -179,6 +179,9 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			      void *param);
 void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev,
 				  struct amdgpu_vm *vm);
+int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
+			struct amdgpu_vm *vm,
+			uint64_t saddr, uint64_t size);
 int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 		      struct amdgpu_sync *sync, struct dma_fence *fence,
 		      struct amdgpu_job *job);
-- 
cgit 


From dc54d3d1744d23ed0b345fd8bc1c493b74e8df44 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Mon, 13 Mar 2017 10:13:38 +0100
Subject: drm/amdgpu: implement AMDGPU_VA_OP_CLEAR v2
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

A new VM operation to remove all mappings in a range.

v2: limit unmapped area as noted by Jerry

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c   | 27 ++++++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c    | 99 +++++++++++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h    |  3 +
 include/uapi/drm/amdgpu_drm.h             |  1 +
 5 files changed, 124 insertions(+), 8 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 0240f108f90e..b311b389bd5a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -507,14 +507,16 @@ static int amdgpu_gem_va_check(void *param, struct amdgpu_bo *bo)
  * amdgpu_gem_va_update_vm -update the bo_va in its VM
  *
  * @adev: amdgpu_device pointer
+ * @vm: vm to update
  * @bo_va: bo_va to update
  * @list: validation list
- * @operation: map or unmap
+ * @operation: map, unmap or clear
  *
  * Update the bo_va directly after setting its address. Errors are not
  * vital here, so they are not reported back to userspace.
  */
 static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm,
 				    struct amdgpu_bo_va *bo_va,
 				    struct list_head *list,
 				    uint32_t operation)
@@ -529,16 +531,16 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 			goto error;
 	}
 
-	r = amdgpu_vm_validate_pt_bos(adev, bo_va->vm, amdgpu_gem_va_check,
+	r = amdgpu_vm_validate_pt_bos(adev, vm, amdgpu_gem_va_check,
 				      NULL);
 	if (r)
 		goto error;
 
-	r = amdgpu_vm_update_page_directory(adev, bo_va->vm);
+	r = amdgpu_vm_update_page_directory(adev, vm);
 	if (r)
 		goto error;
 
-	r = amdgpu_vm_clear_freed(adev, bo_va->vm);
+	r = amdgpu_vm_clear_freed(adev, vm);
 	if (r)
 		goto error;
 
@@ -592,6 +594,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	switch (args->operation) {
 	case AMDGPU_VA_OP_MAP:
 	case AMDGPU_VA_OP_UNMAP:
+	case AMDGPU_VA_OP_CLEAR:
 		break;
 	default:
 		dev_err(&dev->pdev->dev, "unsupported operation %d\n",
@@ -600,7 +603,8 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	}
 
 	INIT_LIST_HEAD(&list);
-	if (!(args->flags & AMDGPU_VM_PAGE_PRT)) {
+	if ((args->operation != AMDGPU_VA_OP_CLEAR) &&
+	    !(args->flags & AMDGPU_VM_PAGE_PRT)) {
 		gobj = drm_gem_object_lookup(filp, args->handle);
 		if (gobj == NULL)
 			return -ENOENT;
@@ -625,8 +629,10 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 			r = -ENOENT;
 			goto error_backoff;
 		}
-	} else {
+	} else if (args->operation != AMDGPU_VA_OP_CLEAR) {
 		bo_va = fpriv->prt_va;
+	} else {
+		bo_va = NULL;
 	}
 
 	switch (args->operation) {
@@ -644,11 +650,18 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	case AMDGPU_VA_OP_UNMAP:
 		r = amdgpu_vm_bo_unmap(adev, bo_va, args->va_address);
 		break;
+
+	case AMDGPU_VA_OP_CLEAR:
+		r = amdgpu_vm_bo_clear_mappings(adev, &fpriv->vm,
+						args->va_address,
+						args->map_size);
+		break;
 	default:
 		break;
 	}
 	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE) && !amdgpu_vm_debug)
-		amdgpu_gem_va_update_vm(adev, bo_va, &list, args->operation);
+		amdgpu_gem_va_update_vm(adev, &fpriv->vm, bo_va, &list,
+					args->operation);
 
 error_backoff:
 	ttm_eu_backoff_reservation(&ticket, &list);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 08ccb3d34b21..3e955190f013 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -188,7 +188,7 @@ TRACE_EVENT(amdgpu_vm_bo_map,
 			     ),
 
 	    TP_fast_assign(
-			   __entry->bo = bo_va->bo;
+			   __entry->bo = bo_va ? bo_va->bo : NULL;
 			   __entry->start = mapping->it.start;
 			   __entry->last = mapping->it.last;
 			   __entry->offset = mapping->offset;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 296e985d0b65..b67e94e25cfc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1612,6 +1612,105 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 	return 0;
 }
 
+/**
+ * amdgpu_vm_bo_clear_mappings - remove all mappings in a specific range
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: VM structure to use
+ * @saddr: start of the range
+ * @size: size of the range
+ *
+ * Remove all mappings in a range, split them as appropriate.
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				uint64_t saddr, uint64_t size)
+{
+	struct amdgpu_bo_va_mapping *before, *after, *tmp, *next;
+	struct interval_tree_node *it;
+	LIST_HEAD(removed);
+	uint64_t eaddr;
+
+	eaddr = saddr + size - 1;
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	/* Allocate all the needed memory */
+	before = kzalloc(sizeof(*before), GFP_KERNEL);
+	if (!before)
+		return -ENOMEM;
+
+	after = kzalloc(sizeof(*after), GFP_KERNEL);
+	if (!after) {
+		kfree(before);
+		return -ENOMEM;
+	}
+
+	/* Now gather all removed mappings */
+	it = interval_tree_iter_first(&vm->va, saddr, eaddr);
+	while (it) {
+		tmp = container_of(it, struct amdgpu_bo_va_mapping, it);
+		it = interval_tree_iter_next(it, saddr, eaddr);
+
+		/* Remember mapping split at the start */
+		if (tmp->it.start < saddr) {
+			before->it.start = tmp->it.start;;
+			before->it.last = saddr - 1;
+			before->offset = tmp->offset;
+			before->flags = tmp->flags;
+			list_add(&before->list, &tmp->list);
+		}
+
+		/* Remember mapping split at the end */
+		if (tmp->it.last > eaddr) {
+			after->it.start = eaddr + 1;
+			after->it.last = tmp->it.last;
+			after->offset = tmp->offset;
+			after->offset += after->it.start - tmp->it.start;
+			after->flags = tmp->flags;
+			list_add(&after->list, &tmp->list);
+		}
+
+		list_del(&tmp->list);
+		list_add(&tmp->list, &removed);
+	}
+
+	/* And free them up */
+	list_for_each_entry_safe(tmp, next, &removed, list) {
+		interval_tree_remove(&tmp->it, &vm->va);
+		list_del(&tmp->list);
+
+		if (tmp->it.start < saddr)
+		    tmp->it.start = saddr;
+		if (tmp->it.last > eaddr)
+		    tmp->it.last = eaddr;
+
+		list_add(&tmp->list, &vm->freed);
+		trace_amdgpu_vm_bo_unmap(NULL, tmp);
+	}
+
+	/* Insert partial mapping before the range*/
+	if (before->it.start != before->it.last) {
+		interval_tree_insert(&before->it, &vm->va);
+		if (before->flags & AMDGPU_PTE_PRT)
+			amdgpu_vm_prt_get(adev);
+	} else {
+		kfree(before);
+	}
+
+	/* Insert partial mapping after the range */
+	if (after->it.start != after->it.last) {
+		interval_tree_insert(&after->it, &vm->va);
+		if (after->flags & AMDGPU_PTE_PRT)
+			amdgpu_vm_prt_get(adev);
+	} else {
+		kfree(after);
+	}
+
+	return 0;
+}
+
 /**
  * amdgpu_vm_bo_rmv - remove a bo to a specific vm
  *
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 1e5a3b2c7927..95fe47733b7f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -210,6 +210,9 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 		       struct amdgpu_bo_va *bo_va,
 		       uint64_t addr);
+int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				uint64_t saddr, uint64_t size);
 void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 		      struct amdgpu_bo_va *bo_va);
 
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index 2c30e324cb12..199f1b46fd2c 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -350,6 +350,7 @@ struct drm_amdgpu_gem_op {
 
 #define AMDGPU_VA_OP_MAP			1
 #define AMDGPU_VA_OP_UNMAP			2
+#define AMDGPU_VA_OP_CLEAR			3
 
 /* Delay the page table update till the next CS */
 #define AMDGPU_VM_DELAY_UPDATE		(1 << 0)
-- 
cgit 


From 80f95c579d800fa22e9e57ecb3d50b9e93bc1f82 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Mon, 13 Mar 2017 10:13:39 +0100
Subject: drm/amdgpu: add a VM mapping replace operation v2
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Add a new operation to replace mappings in a VM with a new one.

v2: Fix Jerry's comment, separate out clear operation.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 15 +++++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c  | 64 +++++++++++++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h  |  4 +++
 include/uapi/drm/amdgpu_drm.h           |  1 +
 4 files changed, 83 insertions(+), 1 deletion(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index b311b389bd5a..c71c087727b8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -544,7 +544,8 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 	if (r)
 		goto error;
 
-	if (operation == AMDGPU_VA_OP_MAP)
+	if (operation == AMDGPU_VA_OP_MAP ||
+	    operation == AMDGPU_VA_OP_REPLACE)
 		r = amdgpu_vm_bo_update(adev, bo_va, false);
 
 error:
@@ -595,6 +596,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 	case AMDGPU_VA_OP_MAP:
 	case AMDGPU_VA_OP_UNMAP:
 	case AMDGPU_VA_OP_CLEAR:
+	case AMDGPU_VA_OP_REPLACE:
 		break;
 	default:
 		dev_err(&dev->pdev->dev, "unsupported operation %d\n",
@@ -656,6 +658,17 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 						args->va_address,
 						args->map_size);
 		break;
+	case AMDGPU_VA_OP_REPLACE:
+		r = amdgpu_vm_alloc_pts(adev, bo_va->vm, args->va_address,
+					args->map_size);
+		if (r)
+			goto error_backoff;
+
+		va_flags = amdgpu_vm_get_pte_flags(adev, args->flags);
+		r = amdgpu_vm_bo_replace_map(adev, bo_va, args->va_address,
+					     args->offset_in_bo, args->map_size,
+					     va_flags);
+		break;
 	default:
 		break;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index b67e94e25cfc..2da08027ff29 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1560,6 +1560,70 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 	return 0;
 }
 
+/**
+ * amdgpu_vm_bo_replace_map - map bo inside a vm, replacing existing mappings
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to store the address
+ * @saddr: where to map the BO
+ * @offset: requested offset in the BO
+ * @flags: attributes of pages (read/write/valid/etc.)
+ *
+ * Add a mapping of the BO at the specefied addr into the VM. Replace existing
+ * mappings as we do so.
+ * Returns 0 for success, error for failure.
+ *
+ * Object has to be reserved and unreserved outside!
+ */
+int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
+			     struct amdgpu_bo_va *bo_va,
+			     uint64_t saddr, uint64_t offset,
+			     uint64_t size, uint64_t flags)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_vm *vm = bo_va->vm;
+	uint64_t eaddr;
+	int r;
+
+	/* validate the parameters */
+	if (saddr & AMDGPU_GPU_PAGE_MASK || offset & AMDGPU_GPU_PAGE_MASK ||
+	    size == 0 || size & AMDGPU_GPU_PAGE_MASK)
+		return -EINVAL;
+
+	/* make sure object fit at this offset */
+	eaddr = saddr + size - 1;
+	if (saddr >= eaddr ||
+	    (bo_va->bo && offset + size > amdgpu_bo_size(bo_va->bo)))
+		return -EINVAL;
+
+	/* Allocate all the needed memory */
+	mapping = kmalloc(sizeof(*mapping), GFP_KERNEL);
+	if (!mapping)
+		return -ENOMEM;
+
+	r = amdgpu_vm_bo_clear_mappings(adev, bo_va->vm, saddr, size);
+	if (r) {
+		kfree(mapping);
+		return r;
+	}
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	mapping->it.start = saddr;
+	mapping->it.last = eaddr;
+	mapping->offset = offset;
+	mapping->flags = flags;
+
+	list_add(&mapping->list, &bo_va->invalids);
+	interval_tree_insert(&mapping->it, &vm->va);
+
+	if (flags & AMDGPU_PTE_PRT)
+		amdgpu_vm_prt_get(adev);
+
+	return 0;
+}
+
 /**
  * amdgpu_vm_bo_unmap - remove bo mapping from vm
  *
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 95fe47733b7f..ab0429d12992 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -207,6 +207,10 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		     struct amdgpu_bo_va *bo_va,
 		     uint64_t addr, uint64_t offset,
 		     uint64_t size, uint64_t flags);
+int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
+			     struct amdgpu_bo_va *bo_va,
+			     uint64_t addr, uint64_t offset,
+			     uint64_t size, uint64_t flags);
 int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 		       struct amdgpu_bo_va *bo_va,
 		       uint64_t addr);
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index 199f1b46fd2c..37e2c0da01fb 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -351,6 +351,7 @@ struct drm_amdgpu_gem_op {
 #define AMDGPU_VA_OP_MAP			1
 #define AMDGPU_VA_OP_UNMAP			2
 #define AMDGPU_VA_OP_CLEAR			3
+#define AMDGPU_VA_OP_REPLACE			4
 
 /* Delay the page table update till the next CS */
 #define AMDGPU_VM_DELAY_UPDATE		(1 << 0)
-- 
cgit 


From f34678187a339702256c70c88c3db2d1975484ce Mon Sep 17 00:00:00 2001
From: Nicolai Hähnle <nicolai.haehnle@amd.com>
Date: Thu, 23 Mar 2017 19:36:31 +0100
Subject: drm/amdgpu: add optional fence out-parameter to amdgpu_vm_clear_freed
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

We will add the fence to freed buffer objects in a later commit, to ensure
that the underlying memory can only be re-used after all references in
page tables have been cleared.

Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Junwei Zhang <Jerry.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_cs.c  |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c  | 21 +++++++++++++++------
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h  |  3 ++-
 4 files changed, 19 insertions(+), 9 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 419fab50dc54..e83a6dcf2f68 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -785,7 +785,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_vm_clear_freed(adev, vm);
+	r = amdgpu_vm_clear_freed(adev, vm, 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 c71c087727b8..5e068eefcaec 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -540,7 +540,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 	if (r)
 		goto error;
 
-	r = amdgpu_vm_clear_freed(adev, vm);
+	r = amdgpu_vm_clear_freed(adev, vm, 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 4fe892e3600d..696860132f04 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -1382,6 +1382,8 @@ static void amdgpu_vm_prt_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
+ * @fence: optional resulting fence (unchanged if no work needed to be done
+ * or if an error occurred)
  *
  * Make sure all freed BOs are cleared in the PT.
  * Returns 0 for success.
@@ -1389,10 +1391,11 @@ static void amdgpu_vm_prt_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
  * PTs have to be reserved and mutex must be locked!
  */
 int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
-			  struct amdgpu_vm *vm)
+			  struct amdgpu_vm *vm,
+			  struct dma_fence **fence)
 {
 	struct amdgpu_bo_va_mapping *mapping;
-	struct dma_fence *fence = NULL;
+	struct dma_fence *f = NULL;
 	int r;
 
 	while (!list_empty(&vm->freed)) {
@@ -1401,15 +1404,21 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 		list_del(&mapping->list);
 
 		r = amdgpu_vm_bo_split_mapping(adev, NULL, 0, NULL, vm, mapping,
-					       0, 0, &fence);
-		amdgpu_vm_free_mapping(adev, vm, mapping, fence);
+					       0, 0, &f);
+		amdgpu_vm_free_mapping(adev, vm, mapping, f);
 		if (r) {
-			dma_fence_put(fence);
+			dma_fence_put(f);
 			return r;
 		}
+	}
 
+	if (fence && f) {
+		dma_fence_put(*fence);
+		*fence = f;
+	} else {
+		dma_fence_put(f);
 	}
-	dma_fence_put(fence);
+
 	return 0;
 
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index ab0429d12992..cbef85bef296 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -190,7 +190,8 @@ void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vm_id);
 int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
 				    struct amdgpu_vm *vm);
 int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
-			  struct amdgpu_vm *vm);
+			  struct amdgpu_vm *vm,
+			  struct dma_fence **fence);
 int amdgpu_vm_clear_invalids(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			     struct amdgpu_sync *sync);
 int amdgpu_vm_bo_update(struct amdgpu_device *adev,
-- 
cgit 


From 66e02bc343bafac589875a72607a1ae6c4d25bcb Mon Sep 17 00:00:00 2001
From: Alex Xie <AlexBin.Xie@amd.com>
Date: Tue, 14 Feb 2017 12:04:52 -0500
Subject: drm/amdgpu: Add MTYPE flags to GPU VM IOCTL interface
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Signed-off-by: Alex Xie <AlexBin.Xie@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@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 +-
 include/uapi/drm/amdgpu_drm.h           | 12 ++++++++++++
 2 files changed, 13 insertions(+), 1 deletion(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 5e068eefcaec..b4f52fd7e237 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -558,7 +558,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 {
 	const uint32_t valid_flags = AMDGPU_VM_DELAY_UPDATE |
 		AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_PAGE_WRITEABLE |
-		AMDGPU_VM_PAGE_EXECUTABLE;
+		AMDGPU_VM_PAGE_EXECUTABLE | AMDGPU_VM_MTYPE_MASK;
 	const uint32_t prt_flags = AMDGPU_VM_DELAY_UPDATE |
 		AMDGPU_VM_PAGE_PRT;
 
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index 37e2c0da01fb..5d8e7090fe9e 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -365,6 +365,18 @@ struct drm_amdgpu_gem_op {
 #define AMDGPU_VM_PAGE_EXECUTABLE	(1 << 3)
 /* partially resident texture */
 #define AMDGPU_VM_PAGE_PRT		(1 << 4)
+/* MTYPE flags use bit 5 to 8 */
+#define AMDGPU_VM_MTYPE_MASK		(0xf << 5)
+/* Default MTYPE. Pre-AI must use this.  Recommended for newer ASICs. */
+#define AMDGPU_VM_MTYPE_DEFAULT		(0 << 5)
+/* Use NC MTYPE instead of default MTYPE */
+#define AMDGPU_VM_MTYPE_NC		(1 << 5)
+/* Use WC MTYPE instead of default MTYPE */
+#define AMDGPU_VM_MTYPE_WC		(2 << 5)
+/* Use CC MTYPE instead of default MTYPE */
+#define AMDGPU_VM_MTYPE_CC		(3 << 5)
+/* Use UC MTYPE instead of default MTYPE */
+#define AMDGPU_VM_MTYPE_UC		(4 << 5)
 
 struct drm_amdgpu_gem_va {
 	/** GEM object handle */
-- 
cgit 


From 194d2161131b35f9cd4cf3e355f2d7cd3b296898 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Wed, 12 Oct 2016 15:13:52 +0200
Subject: drm/amdgpu: handle multi level PD updates V2
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Update all levels of the page directory.

V2:
a. sub level pdes always are written to incorrect place.
b. sub levels need to update regardless of parent updates.

Signed-off-by: Christian König <christian.koenig@amd.com> (V1)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (V1)
Signed-off-by: Chunming Zhou <David1.Zhou@amd.com> (V2)
Acked-by: Alex Deucher <alexander.deucher@amd.com> (V2)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c  |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c  | 97 ++++++++++++++++++++++-----------
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h  |  4 +-
 4 files changed, 68 insertions(+), 37 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 2e11b48d92e7..bf64a0d1c1f4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -777,7 +777,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 	struct amdgpu_bo *bo;
 	int i, r;
 
-	r = amdgpu_vm_update_page_directory(adev, vm);
+	r = amdgpu_vm_update_directories(adev, vm);
 	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 b4f52fd7e237..7a37b93a0dfd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -536,7 +536,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 	if (r)
 		goto error;
 
-	r = amdgpu_vm_update_page_directory(adev, vm);
+	r = amdgpu_vm_update_directories(adev, vm);
 	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 d48ea0f0d825..e3c6ffac7f5c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -700,24 +700,24 @@ static uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr)
 }
 
 /*
- * amdgpu_vm_update_pdes - make sure that page directory is valid
+ * amdgpu_vm_update_level - update a single level in the hierarchy
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
- * @start: start of GPU address range
- * @end: end of GPU address range
+ * @parent: parent directory
  *
- * Allocates new page tables if necessary
- * and updates the page directory.
+ * Makes sure all entries in @parent are up to date.
  * Returns 0 for success, error for failure.
  */
-int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
-				    struct amdgpu_vm *vm)
+static int amdgpu_vm_update_level(struct amdgpu_device *adev,
+				  struct amdgpu_vm *vm,
+				  struct amdgpu_vm_pt *parent,
+				  unsigned level)
 {
 	struct amdgpu_bo *shadow;
 	struct amdgpu_ring *ring;
 	uint64_t pd_addr, shadow_addr;
-	uint32_t incr = AMDGPU_VM_PTE_COUNT * 8;
+	uint32_t incr = amdgpu_vm_bo_size(adev, level + 1);
 	uint64_t last_pde = ~0, last_pt = ~0, last_shadow = ~0;
 	unsigned count = 0, pt_idx, ndw;
 	struct amdgpu_job *job;
@@ -726,16 +726,19 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
 
 	int r;
 
+	if (!parent->entries)
+		return 0;
 	ring = container_of(vm->entity.sched, struct amdgpu_ring, sched);
-	shadow = vm->root.bo->shadow;
 
 	/* padding, etc. */
 	ndw = 64;
 
 	/* assume the worst case */
-	ndw += vm->root.last_entry_used * 6;
+	ndw += parent->last_entry_used * 6;
+
+	pd_addr = amdgpu_bo_gpu_offset(parent->bo);
 
-	pd_addr = amdgpu_bo_gpu_offset(vm->root.bo);
+	shadow = parent->bo->shadow;
 	if (shadow) {
 		r = amdgpu_ttm_bind(&shadow->tbo, &shadow->tbo.mem);
 		if (r)
@@ -754,9 +757,9 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
 	params.adev = adev;
 	params.ib = &job->ibs[0];
 
-	/* walk over the address space and update the page directory */
-	for (pt_idx = 0; pt_idx <= vm->root.last_entry_used; ++pt_idx) {
-		struct amdgpu_bo *bo = vm->root.entries[pt_idx].bo;
+	/* walk over the address space and update the directory */
+	for (pt_idx = 0; pt_idx <= parent->last_entry_used; ++pt_idx) {
+		struct amdgpu_bo *bo = parent->entries[pt_idx].bo;
 		uint64_t pde, pt;
 
 		if (bo == NULL)
@@ -772,10 +775,10 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
 		}
 
 		pt = amdgpu_bo_gpu_offset(bo);
-		if (vm->root.entries[pt_idx].addr == pt)
+		if (parent->entries[pt_idx].addr == pt)
 			continue;
 
-		vm->root.entries[pt_idx].addr = pt;
+		parent->entries[pt_idx].addr = pt;
 
 		pde = pd_addr + pt_idx * 8;
 		if (((last_pde + 8 * count) != pde) ||
@@ -820,26 +823,39 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
 
 	if (params.ib->length_dw == 0) {
 		amdgpu_job_free(job);
-		return 0;
-	}
-
-	amdgpu_ring_pad_ib(ring, params.ib);
-	amdgpu_sync_resv(adev, &job->sync, vm->root.bo->tbo.resv,
-			 AMDGPU_FENCE_OWNER_VM);
-	if (shadow)
-		amdgpu_sync_resv(adev, &job->sync, shadow->tbo.resv,
+	} else {
+		amdgpu_ring_pad_ib(ring, params.ib);
+		amdgpu_sync_resv(adev, &job->sync, parent->bo->tbo.resv,
 				 AMDGPU_FENCE_OWNER_VM);
+		if (shadow)
+			amdgpu_sync_resv(adev, &job->sync, shadow->tbo.resv,
+					 AMDGPU_FENCE_OWNER_VM);
 
-	WARN_ON(params.ib->length_dw > ndw);
-	r = amdgpu_job_submit(job, ring, &vm->entity,
-			      AMDGPU_FENCE_OWNER_VM, &fence);
-	if (r)
-		goto error_free;
+		WARN_ON(params.ib->length_dw > ndw);
+		r = amdgpu_job_submit(job, ring, &vm->entity,
+				AMDGPU_FENCE_OWNER_VM, &fence);
+		if (r)
+			goto error_free;
 
-	amdgpu_bo_fence(vm->root.bo, fence, true);
-	dma_fence_put(vm->last_dir_update);
-	vm->last_dir_update = dma_fence_get(fence);
-	dma_fence_put(fence);
+		amdgpu_bo_fence(parent->bo, fence, true);
+		dma_fence_put(vm->last_dir_update);
+		vm->last_dir_update = dma_fence_get(fence);
+		dma_fence_put(fence);
+	}
+	/*
+	 * Recurse into the subdirectories. This recursion is harmless because
+	 * we only have a maximum of 5 layers.
+	 */
+	for (pt_idx = 0; pt_idx <= parent->last_entry_used; ++pt_idx) {
+		struct amdgpu_vm_pt *entry = &parent->entries[pt_idx];
+
+		if (!entry->bo)
+			continue;
+
+		r = amdgpu_vm_update_level(adev, vm, entry, level + 1);
+		if (r)
+			return r;
+	}
 
 	return 0;
 
@@ -848,6 +864,21 @@ error_free:
 	return r;
 }
 
+/*
+ * amdgpu_vm_update_directories - make sure that all directories are valid
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Makes sure all directories are up to date.
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_vm_update_directories(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm)
+{
+	return amdgpu_vm_update_level(adev, vm, &vm->root, 0);
+}
+
 /**
  * amdgpu_vm_update_ptes - make sure that page tables are valid
  *
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 6c8380d89148..abb4d27b2326 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -192,8 +192,8 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 		      struct amdgpu_job *job);
 int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job);
 void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vm_id);
-int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
-				    struct amdgpu_vm *vm);
+int amdgpu_vm_update_directories(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm);
 int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 			  struct amdgpu_vm *vm,
 			  struct dma_fence **fence);
-- 
cgit 


From 23e0563e48f7e9e98003df5b43d6a48e162782c6 Mon Sep 17 00:00:00 2001
From: Nicolai Hähnle <nicolai.haehnle@amd.com>
Date: Thu, 23 Mar 2017 19:34:11 +0100
Subject: drm/amdgpu: clear freed mappings immediately when BO may be freed
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Also, add the fence of the clear operations to the BO to ensure that
the underlying memory can only be re-used after all PTEs pointing to
it have been cleared.

This avoids the following sequence of events that could be triggered
by user space:

1. Submit a CS that accesses some BO _without_ adding that BO to the
   buffer list.
2. Free that BO.
3. Some other task re-uses the memory underlying the BO.
4. The CS is submitted to the hardware and accesses memory that is
   now already in use by somebody else.

By clearing the page tables immediately in step 2, a GPU VM fault will
be triggered in step 4 instead of wild memory accesses.

v2: use amdgpu_bo_fence directly

Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Junwei Zhang <Jerry.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 | 12 ++++++++++++
 1 file changed, 12 insertions(+)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 7a37b93a0dfd..f85520d4e711 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -152,6 +152,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
 	struct ttm_validate_buffer tv;
 	struct ww_acquire_ctx ticket;
 	struct amdgpu_bo_va *bo_va;
+	struct dma_fence *fence = NULL;
 	int r;
 
 	INIT_LIST_HEAD(&list);
@@ -173,6 +174,17 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
 	if (bo_va) {
 		if (--bo_va->ref_count == 0) {
 			amdgpu_vm_bo_rmv(adev, bo_va);
+
+			r = amdgpu_vm_clear_freed(adev, vm, &fence);
+			if (unlikely(r)) {
+				dev_err(adev->dev, "failed to clear page "
+					"tables on GEM object close (%d)\n", r);
+			}
+
+			if (fence) {
+				amdgpu_bo_fence(bo, fence, true);
+				dma_fence_put(fence);
+			}
 		}
 	}
 	ttm_eu_backoff_reservation(&ticket, &list);
-- 
cgit 


From ec2c467efe9b9d781e08a4ac827c5106e3c775c4 Mon Sep 17 00:00:00 2001
From: Alex Xie <AlexBin.Xie@amd.com>
Date: Wed, 5 Apr 2017 16:33:00 -0400
Subject: drm/amdgpu: Avoid using signed integer to store pointer value

Signed-off-by: Alex Xie <AlexBin.Xie@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c | 2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c     | 2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c     | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index 0218cea6be4d..a6649874e6ce 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -237,7 +237,7 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	union drm_amdgpu_bo_list *args = data;
 	uint32_t handle = args->in.list_handle;
-	const void __user *uptr = (const void*)(long)args->in.bo_info_ptr;
+	const void __user *uptr = (const void*)(uintptr_t)args->in.bo_info_ptr;
 
 	struct drm_amdgpu_bo_list_entry *info;
 	struct amdgpu_bo_list *list;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index f85520d4e711..bb710e0ace01 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -717,7 +717,7 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 	switch (args->op) {
 	case AMDGPU_GEM_OP_GET_GEM_CREATE_INFO: {
 		struct drm_amdgpu_gem_create_in info;
-		void __user *out = (void __user *)(long)args->value;
+		void __user *out = (void __user *)(uintptr_t)args->value;
 
 		info.bo_size = robj->gem_base.size;
 		info.alignment = robj->tbo.mem.page_alignment << PAGE_SHIFT;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 055d10a79ea5..832be632478f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -237,7 +237,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 	struct amdgpu_device *adev = dev->dev_private;
 	struct drm_amdgpu_info *info = data;
 	struct amdgpu_mode_info *minfo = &adev->mode_info;
-	void __user *out = (void __user *)(long)info->return_pointer;
+	void __user *out = (void __user *)(uintptr_t)info->return_pointer;
 	uint32_t size = info->return_size;
 	struct drm_crtc *crtc;
 	uint32_t ui32 = 0;
-- 
cgit 


From 803d89ade4ede131405c57ec1a1db32673995b2e Mon Sep 17 00:00:00 2001
From: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Date: Mon, 3 Apr 2017 13:31:22 +1000
Subject: drm/amdgpu: Refuse to pin or change acceptable domains of prime BOs
 to VRAM. (v2)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Migration to VRAM will break the sharing, resulting in rendering on the exporting GPU never becoming
visible on the importing GPU.

v2: Don't pin BOs to GTT. Instead, refuse to migrate them out of GTT.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c    | 5 +++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 4 ++++
 2 files changed, 9 insertions(+)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index bb710e0ace01..03a9c5cad222 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -729,6 +729,11 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 		break;
 	}
 	case AMDGPU_GEM_OP_SET_PLACEMENT:
+		if (robj->prime_shared_count && (args->value & AMDGPU_GEM_DOMAIN_VRAM)) {
+			r = -EINVAL;
+			amdgpu_bo_unreserve(robj);
+			break;
+		}
 		if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm)) {
 			r = -EPERM;
 			amdgpu_bo_unreserve(robj);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 758d8cf8fd17..cb89fff863c0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -650,6 +650,10 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
 	if (WARN_ON_ONCE(min_offset > max_offset))
 		return -EINVAL;
 
+	/* A shared bo cannot be migrated to VRAM */
+	if (bo->prime_shared_count && (domain == AMDGPU_GEM_DOMAIN_VRAM))
+		return -EINVAL;
+
 	if (bo->pin_count) {
 		uint32_t mem_type = bo->tbo.mem.mem_type;
 
-- 
cgit 


From 5a0f3b5f6d798637b0af5d6d3ab3eb02063e0317 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Fri, 21 Apr 2017 10:05:56 +0200
Subject: drm/amdgpu: fix VM clearing in amdgpu_gem_object_close
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

We need to check if the VM is swapped out before trying to update it.

Fixes: 23e0563e48f7 ("drm/amdgpu: clear freed mappings immediately when BO may be freed")
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 68 ++++++++++++++++++---------------
 1 file changed, 37 insertions(+), 31 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 03a9c5cad222..94cb91cf93eb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -139,6 +139,35 @@ int amdgpu_gem_object_open(struct drm_gem_object *obj,
 	return 0;
 }
 
+static int amdgpu_gem_vm_check(void *param, struct amdgpu_bo *bo)
+{
+	/* if anything is swapped out don't swap it in here,
+	   just abort and wait for the next CS */
+	if (!amdgpu_bo_gpu_accessible(bo))
+		return -ERESTARTSYS;
+
+	if (bo->shadow && !amdgpu_bo_gpu_accessible(bo->shadow))
+		return -ERESTARTSYS;
+
+	return 0;
+}
+
+static bool amdgpu_gem_vm_ready(struct amdgpu_device *adev,
+				struct amdgpu_vm *vm,
+				struct list_head *list)
+{
+	struct ttm_validate_buffer *entry;
+
+	list_for_each_entry(entry, list, head) {
+		struct amdgpu_bo *bo =
+			container_of(entry->bo, struct amdgpu_bo, tbo);
+		if (amdgpu_gem_vm_check(NULL, bo))
+			return false;
+	}
+
+	return !amdgpu_vm_validate_pt_bos(adev, vm, amdgpu_gem_vm_check, NULL);
+}
+
 void amdgpu_gem_object_close(struct drm_gem_object *obj,
 			     struct drm_file *file_priv)
 {
@@ -148,15 +177,13 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
 	struct amdgpu_vm *vm = &fpriv->vm;
 
 	struct amdgpu_bo_list_entry vm_pd;
-	struct list_head list, duplicates;
+	struct list_head list;
 	struct ttm_validate_buffer tv;
 	struct ww_acquire_ctx ticket;
 	struct amdgpu_bo_va *bo_va;
-	struct dma_fence *fence = NULL;
 	int r;
 
 	INIT_LIST_HEAD(&list);
-	INIT_LIST_HEAD(&duplicates);
 
 	tv.bo = &bo->tbo;
 	tv.shared = true;
@@ -164,16 +191,18 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
 
 	amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
 
-	r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates);
+	r = ttm_eu_reserve_buffers(&ticket, &list, false, NULL);
 	if (r) {
 		dev_err(adev->dev, "leaking bo va because "
 			"we fail to reserve bo (%d)\n", r);
 		return;
 	}
 	bo_va = amdgpu_vm_bo_find(vm, bo);
-	if (bo_va) {
-		if (--bo_va->ref_count == 0) {
-			amdgpu_vm_bo_rmv(adev, bo_va);
+	if (bo_va && --bo_va->ref_count == 0) {
+		amdgpu_vm_bo_rmv(adev, bo_va);
+
+		if (amdgpu_gem_vm_ready(adev, vm, &list)) {
+			struct dma_fence *fence = NULL;
 
 			r = amdgpu_vm_clear_freed(adev, vm, &fence);
 			if (unlikely(r)) {
@@ -502,19 +531,6 @@ out:
 	return r;
 }
 
-static int amdgpu_gem_va_check(void *param, struct amdgpu_bo *bo)
-{
-	/* if anything is swapped out don't swap it in here,
-	   just abort and wait for the next CS */
-	if (!amdgpu_bo_gpu_accessible(bo))
-		return -ERESTARTSYS;
-
-	if (bo->shadow && !amdgpu_bo_gpu_accessible(bo->shadow))
-		return -ERESTARTSYS;
-
-	return 0;
-}
-
 /**
  * amdgpu_gem_va_update_vm -update the bo_va in its VM
  *
@@ -533,19 +549,9 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 				    struct list_head *list,
 				    uint32_t operation)
 {
-	struct ttm_validate_buffer *entry;
 	int r = -ERESTARTSYS;
 
-	list_for_each_entry(entry, list, head) {
-		struct amdgpu_bo *bo =
-			container_of(entry->bo, struct amdgpu_bo, tbo);
-		if (amdgpu_gem_va_check(NULL, bo))
-			goto error;
-	}
-
-	r = amdgpu_vm_validate_pt_bos(adev, vm, amdgpu_gem_va_check,
-				      NULL);
-	if (r)
+	if (!amdgpu_gem_vm_ready(adev, vm, list))
 		goto error;
 
 	r = amdgpu_vm_update_directories(adev, vm);
-- 
cgit