From e52482dec8366a98ac380b3bdc1a4abb8a390914 Mon Sep 17 00:00:00 2001
From: Felix Kuehling <Felix.Kuehling@amd.com>
Date: Fri, 23 Mar 2018 15:32:28 -0400
Subject: drm/amdgpu: Add MMU notifier type for KFD userptr

This commit adds the notion of MMU notifier types GFX and HSA. GFX
continues to work like MMU notifiers did before. HSA adds support for
KFD userptr BOs. The implementation of KFD userptr eviction is a stub
for now.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index dc34b50e6b29..8e66f3702b7c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -536,7 +536,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	if (p->bo_list) {
 		amdgpu_bo_list_get_list(p->bo_list, &p->validated);
 		if (p->bo_list->first_userptr != p->bo_list->num_entries)
-			p->mn = amdgpu_mn_get(p->adev);
+			p->mn = amdgpu_mn_get(p->adev, AMDGPU_MN_TYPE_GFX);
 	}
 
 	INIT_LIST_HEAD(&duplicates);
-- 
cgit 


From 552825b28ddac200b6080d9e79f4121b68e1517d Mon Sep 17 00:00:00 2001
From: Chunming Zhou <david1.zhou@amd.com>
Date: Mon, 2 Apr 2018 11:20:44 +0800
Subject: drm/amdgpu: add new bo flag that indicates BOs don't need fallback
 (v2)

user cases:
1. KFD wraps amdgpu_bo_create, they have no fallback case which is different
with amdgpu_gem_object_create.
since upstream branch has no amdgpu_amdkfd_gpuvm.c, which need KFD
guys add this flag to __alloc_memory_of_gpu:
+       flags |= AMDGPU_GEM_CREATE_NO_FALLBACK;
2. UMD can specify this flag for their allocation as well if they like.

v2: squash in merge conflict fix (Chunming)

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: felix.kuehling@amd.com
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c     | 3 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 5 ++++-
 include/uapi/drm/amdgpu_drm.h              | 2 ++
 3 files changed, 8 insertions(+), 2 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index dc34b50e6b29..d7d7ce1507ec 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -386,7 +386,8 @@ retry:
 	    bo->tbo.mem.start < adev->gmc.visible_vram_size >> PAGE_SHIFT)
 		p->bytes_moved_vis += ctx.bytes_moved;
 
-	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
+	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains &&
+	    !(bo->flags & AMDGPU_GEM_CREATE_NO_FALLBACK)) {
 		domain = bo->allowed_domains;
 		goto retry;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 04d6830347ec..9e23d6f6f3f3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -388,6 +388,8 @@ retry:
 	drm_gem_private_object_init(adev->ddev, &bo->gem_base, size);
 	INIT_LIST_HEAD(&bo->shadow_list);
 	INIT_LIST_HEAD(&bo->va);
+	bo->preferred_domains = preferred_domains;
+	bo->allowed_domains = allowed_domains;
 
 	bo->flags = flags;
 
@@ -424,7 +426,8 @@ retry:
 	r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, size, type,
 				 &bo->placement, page_align, &ctx, acc_size,
 				 NULL, resv, &amdgpu_ttm_bo_destroy);
-	if (unlikely(r && r != -ERESTARTSYS) && type == ttm_bo_type_device) {
+	if (unlikely(r && r != -ERESTARTSYS) && type == ttm_bo_type_device &&
+	    !(flags & AMDGPU_GEM_CREATE_NO_FALLBACK)) {
 		if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) {
 			flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
 			goto retry;
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index c363b67f2d0a..4f5a27d64c54 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -95,6 +95,8 @@ extern "C" {
 #define AMDGPU_GEM_CREATE_VM_ALWAYS_VALID	(1 << 6)
 /* Flag that BO sharing will be explicitly synchronized */
 #define AMDGPU_GEM_CREATE_EXPLICIT_SYNC		(1 << 7)
+/* Flag that BO doesn't need fallback */
+#define AMDGPU_GEM_CREATE_NO_FALLBACK		(1 << 8)
 
 struct drm_amdgpu_gem_create_in  {
 	/** the requested memory size */
-- 
cgit 


From f1018f50d48395b4a189bf8ea9af1e4441209cfd Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Thu, 5 Apr 2018 14:46:41 +0200
Subject: drm/amdgpu: use ctx bytes_moved
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Instead of the global (inaccurate) counter.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index d7d7ce1507ec..de69ab12bb55 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -412,7 +412,6 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 		struct amdgpu_bo_list_entry *candidate = p->evictable;
 		struct amdgpu_bo *bo = candidate->robj;
 		struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
-		u64 initial_bytes_moved, bytes_moved;
 		bool update_bytes_moved_vis;
 		uint32_t other;
 
@@ -436,18 +435,15 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 			continue;
 
 		/* Good we can try to move this BO somewhere else */
-		amdgpu_ttm_placement_from_domain(bo, other);
 		update_bytes_moved_vis =
 			adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
 			bo->tbo.mem.mem_type == TTM_PL_VRAM &&
 			bo->tbo.mem.start < adev->gmc.visible_vram_size >> PAGE_SHIFT;
-		initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
+		amdgpu_ttm_placement_from_domain(bo, other);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
-		bytes_moved = atomic64_read(&adev->num_bytes_moved) -
-			initial_bytes_moved;
-		p->bytes_moved += bytes_moved;
+		p->bytes_moved += ctx.bytes_moved;
 		if (update_bytes_moved_vis)
-			p->bytes_moved_vis += bytes_moved;
+			p->bytes_moved_vis += ctx.bytes_moved;
 
 		if (unlikely(r))
 			break;
-- 
cgit 


From 5422a28fe86f9f77480471385e0a416c27a9ca72 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Thu, 5 Apr 2018 16:42:03 +0200
Subject: drm/amdgpu: fix and cleanup cpu visible VRAM handling
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The detection if a BO was placed in CPU visible VRAM was incorrect.

Fix it and merge it with the correct detection in amdgpu_ttm.c

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@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_object.h | 21 +++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c    | 19 +++----------------
 3 files changed, 26 insertions(+), 20 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index de69ab12bb55..68af2f878bc9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -382,8 +382,7 @@ retry:
 
 	p->bytes_moved += ctx.bytes_moved;
 	if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
-	    bo->tbo.mem.mem_type == TTM_PL_VRAM &&
-	    bo->tbo.mem.start < adev->gmc.visible_vram_size >> PAGE_SHIFT)
+	    amdgpu_bo_in_cpu_visible_vram(bo))
 		p->bytes_moved_vis += ctx.bytes_moved;
 
 	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains &&
@@ -437,8 +436,7 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 		/* Good we can try to move this BO somewhere else */
 		update_bytes_moved_vis =
 			adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
-			bo->tbo.mem.mem_type == TTM_PL_VRAM &&
-			bo->tbo.mem.start < adev->gmc.visible_vram_size >> PAGE_SHIFT;
+			amdgpu_bo_in_cpu_visible_vram(bo);
 		amdgpu_ttm_placement_from_domain(bo, other);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 		p->bytes_moved += ctx.bytes_moved;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index 546f77cb7882..3bee13344065 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -195,6 +195,27 @@ static inline bool amdgpu_bo_gpu_accessible(struct amdgpu_bo *bo)
 	}
 }
 
+/**
+ * amdgpu_bo_in_cpu_visible_vram - check if BO is (partly) in visible VRAM
+ */
+static inline bool amdgpu_bo_in_cpu_visible_vram(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	unsigned fpfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
+	struct drm_mm_node *node = bo->tbo.mem.mm_node;
+	unsigned long pages_left;
+
+	if (bo->tbo.mem.mem_type != TTM_PL_VRAM)
+		return false;
+
+	for (pages_left = bo->tbo.mem.num_pages; pages_left;
+	     pages_left -= node->size, node++)
+		if (node->start < fpfn)
+			return true;
+
+	return false;
+}
+
 /**
  * amdgpu_bo_explicit_sync - return whether the bo is explicitly synced
  */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 205da3ff9cd0..ab73300e6c7f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -223,20 +223,8 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 		if (!adev->mman.buffer_funcs_enabled) {
 			amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
 		} else if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
-			   !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
-			unsigned fpfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
-			struct drm_mm_node *node = bo->mem.mm_node;
-			unsigned long pages_left;
-
-			for (pages_left = bo->mem.num_pages;
-			     pages_left;
-			     pages_left -= node->size, node++) {
-				if (node->start < fpfn)
-					break;
-			}
-
-			if (!pages_left)
-				goto gtt;
+			   !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
+			   amdgpu_bo_in_cpu_visible_vram(abo)) {
 
 			/* Try evicting to the CPU inaccessible part of VRAM
 			 * first, but only set GTT as busy placement, so this
@@ -245,12 +233,11 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 			 */
 			amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
 							 AMDGPU_GEM_DOMAIN_GTT);
-			abo->placements[0].fpfn = fpfn;
+			abo->placements[0].fpfn = adev->gmc.visible_vram_size >> PAGE_SHIFT;
 			abo->placements[0].lpfn = 0;
 			abo->placement.busy_placement = &abo->placements[1];
 			abo->placement.num_busy_placement = 1;
 		} else {
-gtt:
 			amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
 		}
 		break;
-- 
cgit 


From 1afd30efeddbb1b32cf35d3bf6477b35690eeca6 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Tue, 10 Apr 2018 13:42:29 +0200
Subject: drm/amdgpu: revert "add new bo flag that indicates BOs don't need
 fallback (v2)"
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This reverts commit 6f51d28bfe8e1a676de5cd877639245bed3cc818.

Makes fallback handling to complicated. This is just a feature for the
GEM interface and shouldn't leak into the core BO create function.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c     | 3 +--
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 5 +----
 include/uapi/drm/amdgpu_drm.h              | 2 --
 3 files changed, 2 insertions(+), 8 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 68af2f878bc9..e1756b68a17b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -385,8 +385,7 @@ retry:
 	    amdgpu_bo_in_cpu_visible_vram(bo))
 		p->bytes_moved_vis += ctx.bytes_moved;
 
-	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains &&
-	    !(bo->flags & AMDGPU_GEM_CREATE_NO_FALLBACK)) {
+	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
 		domain = bo->allowed_domains;
 		goto retry;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index a160ef0332d6..1de6864da717 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -388,8 +388,6 @@ retry:
 	drm_gem_private_object_init(adev->ddev, &bo->gem_base, size);
 	INIT_LIST_HEAD(&bo->shadow_list);
 	INIT_LIST_HEAD(&bo->va);
-	bo->preferred_domains = preferred_domains;
-	bo->allowed_domains = allowed_domains;
 
 	bo->flags = flags;
 
@@ -426,8 +424,7 @@ retry:
 	r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, size, type,
 				 &bo->placement, page_align, &ctx, acc_size,
 				 NULL, resv, &amdgpu_ttm_bo_destroy);
-	if (unlikely(r && r != -ERESTARTSYS) && type == ttm_bo_type_device &&
-	    !(flags & AMDGPU_GEM_CREATE_NO_FALLBACK)) {
+	if (unlikely(r && r != -ERESTARTSYS) && type == ttm_bo_type_device) {
 		if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) {
 			flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
 			goto retry;
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index 4f5a27d64c54..c363b67f2d0a 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -95,8 +95,6 @@ extern "C" {
 #define AMDGPU_GEM_CREATE_VM_ALWAYS_VALID	(1 << 6)
 /* Flag that BO sharing will be explicitly synchronized */
 #define AMDGPU_GEM_CREATE_EXPLICIT_SYNC		(1 << 7)
-/* Flag that BO doesn't need fallback */
-#define AMDGPU_GEM_CREATE_NO_FALLBACK		(1 << 8)
 
 struct drm_amdgpu_gem_create_in  {
 	/** the requested memory size */
-- 
cgit 


From 01d98506eca24f1a6b67374ffb65c2e62a56e8e7 Mon Sep 17 00:00:00 2001
From: Emily Deng <Emily.Deng@amd.com>
Date: Wed, 30 May 2018 10:04:25 +0800
Subject: drm/amdgpu: To get gds, gws and oa from adev->gds (v2)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

As now enabled per vm bo feature, the user mode driver won't supply the
bo_list generally, for this case, the gdb_base, gds_size, gws_base, gws_size and
oa_base, oa_size won't be set.

v2: fix warning (Chunming)

Signed-off-by: Emily Deng <Emily.Deng@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 | 38 ++++++++++++++++++++--------------
 1 file changed, 23 insertions(+), 15 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 9c1d491d742e..82312a7bc6ad 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -522,6 +522,9 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	struct amdgpu_bo_list_entry *e;
 	struct list_head duplicates;
 	unsigned i, tries = 10;
+	struct amdgpu_bo *gds;
+	struct amdgpu_bo *gws;
+	struct amdgpu_bo *oa;
 	int r;
 
 	INIT_LIST_HEAD(&p->validated);
@@ -652,31 +655,36 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
 				     p->bytes_moved_vis);
+
 	if (p->bo_list) {
-		struct amdgpu_bo *gds = p->bo_list->gds_obj;
-		struct amdgpu_bo *gws = p->bo_list->gws_obj;
-		struct amdgpu_bo *oa = p->bo_list->oa_obj;
 		struct amdgpu_vm *vm = &fpriv->vm;
 		unsigned i;
 
+		gds = p->bo_list->gds_obj;
+		gws = p->bo_list->gws_obj;
+		oa = p->bo_list->oa_obj;
 		for (i = 0; i < p->bo_list->num_entries; i++) {
 			struct amdgpu_bo *bo = p->bo_list->array[i].robj;
 
 			p->bo_list->array[i].bo_va = amdgpu_vm_bo_find(vm, bo);
 		}
+	} else {
+		gds = p->adev->gds.gds_gfx_bo;
+		gws = p->adev->gds.gws_gfx_bo;
+		oa = p->adev->gds.oa_gfx_bo;
+	}
 
-		if (gds) {
-			p->job->gds_base = amdgpu_bo_gpu_offset(gds);
-			p->job->gds_size = amdgpu_bo_size(gds);
-		}
-		if (gws) {
-			p->job->gws_base = amdgpu_bo_gpu_offset(gws);
-			p->job->gws_size = amdgpu_bo_size(gws);
-		}
-		if (oa) {
-			p->job->oa_base = amdgpu_bo_gpu_offset(oa);
-			p->job->oa_size = amdgpu_bo_size(oa);
-		}
+	if (gds) {
+		p->job->gds_base = amdgpu_bo_gpu_offset(gds);
+		p->job->gds_size = amdgpu_bo_size(gds);
+	}
+	if (gws) {
+		p->job->gws_base = amdgpu_bo_gpu_offset(gws);
+		p->job->gws_size = amdgpu_bo_size(gws);
+	}
+	if (oa) {
+		p->job->oa_base = amdgpu_bo_gpu_offset(oa);
+		p->job->oa_size = amdgpu_bo_size(oa);
 	}
 
 	if (!r && p->uf_entry.robj) {
-- 
cgit 


From c8c5e569c5b0c9ad523c35497af2ae5788a29581 Mon Sep 17 00:00:00 2001
From: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Date: Tue, 12 Jun 2018 14:28:20 -0400
Subject: drm/amdgpu: Consolidate visible vs. real vram check v2.
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Move all instnaces of this check into a function in amdgpu_gmc.h
Rename the original function to a more proper name.

v2:
Add more places to cleanup.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c     | 11 ++++++-----
 drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h    | 15 +++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c    |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c     | 20 ++++----------------
 5 files changed, 27 insertions(+), 23 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 82312a7bc6ad..7a625f3989a0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -31,6 +31,7 @@
 #include <drm/drm_syncobj.h>
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
+#include "amdgpu_gmc.h"
 
 static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 				      struct drm_amdgpu_cs_chunk_fence *data,
@@ -302,7 +303,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	*max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
 
 	/* Do the same for visible VRAM if half of it is free */
-	if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size) {
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc)) {
 		u64 total_vis_vram = adev->gmc.visible_vram_size;
 		u64 used_vis_vram =
 			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
@@ -359,7 +360,7 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	 * to move it. Don't move anything if the threshold is zero.
 	 */
 	if (p->bytes_moved < p->bytes_moved_threshold) {
-		if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+		if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
 		    (bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
 			/* And don't move a CPU_ACCESS_REQUIRED BO to limited
 			 * visible VRAM if we've depleted our allowance to do
@@ -381,7 +382,7 @@ retry:
 	r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 
 	p->bytes_moved += ctx.bytes_moved;
-	if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
 	    amdgpu_bo_in_cpu_visible_vram(bo))
 		p->bytes_moved_vis += ctx.bytes_moved;
 
@@ -434,8 +435,8 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
 
 		/* Good we can try to move this BO somewhere else */
 		update_bytes_moved_vis =
-			adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
-			amdgpu_bo_in_cpu_visible_vram(bo);
+				!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
+				amdgpu_bo_in_cpu_visible_vram(bo);
 		amdgpu_ttm_placement_from_domain(bo, other);
 		r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
 		p->bytes_moved += ctx.bytes_moved;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
index 893c2490b783..6cb4948233cb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@@ -109,4 +109,19 @@ struct amdgpu_gmc {
 	const struct amdgpu_gmc_funcs	*gmc_funcs;
 };
 
+/**
+ * amdgpu_gmc_vram_full_visible - Check if full VRAM is visible through the BAR
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns:
+ * True if full VRAM is visible through the BAR
+ */
+static inline bool amdgpu_gmc_vram_full_visible(struct amdgpu_gmc *gmc)
+{
+	WARN_ON(gmc->real_vram_size < gmc->visible_vram_size);
+
+	return (gmc->real_vram_size == gmc->visible_vram_size);
+}
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 987a9fa33d35..f5b0b180a6cc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -463,7 +463,7 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev,
 	if (unlikely(r != 0))
 		return r;
 
-	if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+	if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
 	    bo->tbo.mem.mem_type == TTM_PL_VRAM &&
 	    bo->tbo.mem.start < adev->gmc.visible_vram_size >> PAGE_SHIFT)
 		amdgpu_cs_report_moved_bytes(adev, ctx.bytes_moved,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index e93a0a237dc3..0c084d3d0865 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -277,7 +277,7 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 		if (!adev->mman.buffer_funcs_enabled) {
 			/* Move to system memory */
 			amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
-		} else if (adev->gmc.visible_vram_size < adev->gmc.real_vram_size &&
+		} else if (!amdgpu_gmc_vram_full_visible(&adev->gmc) &&
 			   !(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
 			   amdgpu_bo_in_cpu_visible_vram(abo)) {
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 7f03f8c38708..6d2294203bd7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -33,6 +33,7 @@
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
 #include "amdgpu_amdkfd.h"
+#include "amdgpu_gmc.h"
 
 /**
  * DOC: GPUVM
@@ -668,19 +669,6 @@ bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
 	return vm_flush_needed || gds_switch_needed;
 }
 
-/**
- * amdgpu_vm_is_large_bar - Check if BAR is large enough
- *
- * @adev: amdgpu_device pointer
- *
- * Returns:
- * True if BAR is large enough.
- */
-static bool amdgpu_vm_is_large_bar(struct amdgpu_device *adev)
-{
-	return (adev->gmc.real_vram_size == adev->gmc.visible_vram_size);
-}
-
 /**
  * amdgpu_vm_flush - hardware flush the vm
  *
@@ -2579,7 +2567,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	}
 	DRM_DEBUG_DRIVER("VM update mode is %s\n",
 			 vm->use_cpu_for_update ? "CPU" : "SDMA");
-	WARN_ONCE((vm->use_cpu_for_update & !amdgpu_vm_is_large_bar(adev)),
+	WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
 		  "CPU update of VM recommended only for large BAR system\n");
 	vm->last_update = NULL;
 
@@ -2699,7 +2687,7 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	vm->pte_support_ats = pte_support_ats;
 	DRM_DEBUG_DRIVER("VM update mode is %s\n",
 			 vm->use_cpu_for_update ? "CPU" : "SDMA");
-	WARN_ONCE((vm->use_cpu_for_update & !amdgpu_vm_is_large_bar(adev)),
+	WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
 		  "CPU update of VM recommended only for large BAR system\n");
 
 	if (vm->pasid) {
@@ -2877,7 +2865,7 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 	 */
 #ifdef CONFIG_X86_64
 	if (amdgpu_vm_update_mode == -1) {
-		if (amdgpu_vm_is_large_bar(adev))
+		if (amdgpu_gmc_vram_full_visible(&adev->gmc))
 			adev->vm_manager.vm_update_mode =
 				AMDGPU_VM_USE_CPU_FOR_COMPUTE;
 		else
-- 
cgit 


From 02374bbd3bfa38cc6922fe56736716308c48f538 Mon Sep 17 00:00:00 2001
From: Michel Dänzer <michel.daenzer@amd.com>
Date: Mon, 25 Jun 2018 11:07:17 +0200
Subject: drm/amdgpu: Reserve VM root shared fence slot for command submission
 (v3)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Without this, there could not be enough slots, which could trigger the
BUG_ON in reservation_object_add_shared_fence.

v2:
* Jump to the error label instead of returning directly (Jerry Zhang)
v3:
* Reserve slots for command submission after VM updates (Christian König)

Cc: stable@vger.kernel.org
Bugzilla: https://bugs.freedesktop.org/106418
Reported-by: mikhail.v.gavrilov@gmail.com
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-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 | 4 ++++
 1 file changed, 4 insertions(+)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 7a625f3989a0..1bc02816989d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -928,6 +928,10 @@ static int amdgpu_cs_ib_vm_chunk(struct amdgpu_device *adev,
 		r = amdgpu_bo_vm_update_pte(p);
 		if (r)
 			return r;
+
+		r = reservation_object_reserve_shared(vm->root.base.bo->tbo.resv);
+		if (r)
+			return r;
 	}
 
 	return amdgpu_cs_sync_rings(p);
-- 
cgit 


From efaa96463929684ceca4be96d0959e9ced1fa093 Mon Sep 17 00:00:00 2001
From: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Date: Thu, 28 Jun 2018 22:55:27 -0400
Subject: drm/amdgpu: Present amdgpu_task_info in VM_FAULTS.

Extract and present the reposnsible process and thread when
VM_FAULT happens.

v2: Use getter and setter functions.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Jim Qu <Jim.Qu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c |  4 ++++
 drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c  | 10 +++++++---
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c  |  9 +++++++--
 3 files changed, 18 insertions(+), 5 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 1bc02816989d..9881a1e55df3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -187,6 +187,10 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 	if (p->uf_entry.robj)
 		p->job->uf_addr = uf_offset;
 	kfree(chunk_array);
+
+	/* Use this opportunity to fill in task info for the vm */
+	amdgpu_vm_set_task_info(vm);
+
 	return 0;
 
 free_all_kdata:
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index 1edbe6b477b5..a86332f36c3e 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -44,7 +44,6 @@
 
 #include "amdgpu_atombios.h"
 
-
 static void gmc_v8_0_set_gmc_funcs(struct amdgpu_device *adev);
 static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev);
 static int gmc_v8_0_wait_for_idle(void *handle);
@@ -1447,8 +1446,13 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 		gmc_v8_0_set_fault_enable_default(adev, false);
 
 	if (printk_ratelimit()) {
-		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
-			entry->src_id, entry->src_data[0]);
+		struct amdgpu_task_info task_info = { 0 };
+
+		amdgpu_vm_get_task_info(adev, entry->pasid, &task_info);
+
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x for process %s pid %d thread %s pid %d\n",
+			entry->src_id, entry->src_data[0], task_info.process_name,
+			task_info.tgid, task_info.task_name, task_info.pid);
 		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
 			addr);
 		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 3c0a85d4e4ab..7f238149ba54 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -257,11 +257,16 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
 	}
 
 	if (printk_ratelimit()) {
+		struct amdgpu_task_info task_info = { 0 };
+
+		amdgpu_vm_get_task_info(adev, entry->pasid, &task_info);
+
 		dev_err(adev->dev,
-			"[%s] VMC page fault (src_id:%u ring:%u vmid:%u pasid:%u)\n",
+			"[%s] VMC page fault (src_id:%u ring:%u vmid:%u pasid:%u, for process %s pid %d thread %s pid %d\n)\n",
 			entry->vmid_src ? "mmhub" : "gfxhub",
 			entry->src_id, entry->ring_id, entry->vmid,
-			entry->pasid);
+			entry->pasid, task_info.process_name, task_info.tgid,
+			task_info.task_name, task_info.pid);
 		dev_err(adev->dev, "  at page 0x%016llx from %d\n",
 			addr, entry->client_id);
 		if (!amdgpu_sriov_vf(adev))
-- 
cgit 


From a5ccfe5c20740f2fbf00291490cdf8d2373ec255 Mon Sep 17 00:00:00 2001
From: Michel Dänzer <michel.daenzer@amd.com>
Date: Wed, 11 Jul 2018 12:00:40 +0200
Subject: drm/amdgpu: Make pin_size values atomic
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Concurrent execution of the non-atomic arithmetic could result in
completely bogus values.

v2:
* Rebased on v2 of the previous patch

Cc: stable@vger.kernel.org
Bugzilla: https://bugs.freedesktop.org/106872
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  6 +++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c     |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c    | 22 +++++++++++-----------
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 14 ++++++++------
 4 files changed, 23 insertions(+), 21 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 51b16de008bb..53435da158c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -1589,9 +1589,9 @@ struct amdgpu_device {
 	DECLARE_HASHTABLE(mn_hash, 7);
 
 	/* tracking pinned memory */
-	u64 vram_pin_size;
-	u64 visible_pin_size;
-	u64 gart_pin_size;
+	atomic64_t vram_pin_size;
+	atomic64_t visible_pin_size;
+	atomic64_t gart_pin_size;
 
 	/* amdkfd interface */
 	struct kfd_dev          *kfd;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 9881a1e55df3..5a2a5ba29f9a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -262,7 +262,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 		return;
 	}
 
-	total_vram = adev->gmc.real_vram_size - adev->vram_pin_size;
+	total_vram = adev->gmc.real_vram_size - atomic64_read(&adev->vram_pin_size);
 	used_vram = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 	free_vram = used_vram >= total_vram ? 0 : total_vram - used_vram;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index c00bd9591dad..207f238649b4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -501,13 +501,13 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 	case AMDGPU_INFO_VRAM_GTT: {
 		struct drm_amdgpu_info_vram_gtt vram_gtt;
 
-		vram_gtt.vram_size = adev->gmc.real_vram_size;
-		vram_gtt.vram_size -= adev->vram_pin_size;
-		vram_gtt.vram_cpu_accessible_size = adev->gmc.visible_vram_size;
-		vram_gtt.vram_cpu_accessible_size -= adev->visible_pin_size;
+		vram_gtt.vram_size = adev->gmc.real_vram_size -
+			atomic64_read(&adev->vram_pin_size);
+		vram_gtt.vram_cpu_accessible_size = adev->gmc.visible_vram_size -
+			atomic64_read(&adev->visible_pin_size);
 		vram_gtt.gtt_size = adev->mman.bdev.man[TTM_PL_TT].size;
 		vram_gtt.gtt_size *= PAGE_SIZE;
-		vram_gtt.gtt_size -= adev->gart_pin_size;
+		vram_gtt.gtt_size -= atomic64_read(&adev->gart_pin_size);
 		return copy_to_user(out, &vram_gtt,
 				    min((size_t)size, sizeof(vram_gtt))) ? -EFAULT : 0;
 	}
@@ -516,16 +516,16 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 
 		memset(&mem, 0, sizeof(mem));
 		mem.vram.total_heap_size = adev->gmc.real_vram_size;
-		mem.vram.usable_heap_size =
-			adev->gmc.real_vram_size - adev->vram_pin_size;
+		mem.vram.usable_heap_size = adev->gmc.real_vram_size -
+			atomic64_read(&adev->vram_pin_size);
 		mem.vram.heap_usage =
 			amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 		mem.vram.max_allocation = mem.vram.usable_heap_size * 3 / 4;
 
 		mem.cpu_accessible_vram.total_heap_size =
 			adev->gmc.visible_vram_size;
-		mem.cpu_accessible_vram.usable_heap_size =
-			adev->gmc.visible_vram_size - adev->visible_pin_size;
+		mem.cpu_accessible_vram.usable_heap_size = adev->gmc.visible_vram_size -
+			atomic64_read(&adev->visible_pin_size);
 		mem.cpu_accessible_vram.heap_usage =
 			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 		mem.cpu_accessible_vram.max_allocation =
@@ -533,8 +533,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 
 		mem.gtt.total_heap_size = adev->mman.bdev.man[TTM_PL_TT].size;
 		mem.gtt.total_heap_size *= PAGE_SIZE;
-		mem.gtt.usable_heap_size = mem.gtt.total_heap_size
-			- adev->gart_pin_size;
+		mem.gtt.usable_heap_size = mem.gtt.total_heap_size -
+			atomic64_read(&adev->gart_pin_size);
 		mem.gtt.heap_usage =
 			amdgpu_gtt_mgr_usage(&adev->mman.bdev.man[TTM_PL_TT]);
 		mem.gtt.max_allocation = mem.gtt.usable_heap_size * 3 / 4;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index f0239feceab4..1fd83ed175fa 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -916,10 +916,11 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
 
 	domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
 	if (domain == AMDGPU_GEM_DOMAIN_VRAM) {
-		adev->vram_pin_size += amdgpu_bo_size(bo);
-		adev->visible_pin_size += amdgpu_vram_mgr_bo_visible_size(bo);
+		atomic64_add(amdgpu_bo_size(bo), &adev->vram_pin_size);
+		atomic64_add(amdgpu_vram_mgr_bo_visible_size(bo),
+			     &adev->visible_pin_size);
 	} else if (domain == AMDGPU_GEM_DOMAIN_GTT) {
-		adev->gart_pin_size += amdgpu_bo_size(bo);
+		atomic64_add(amdgpu_bo_size(bo), &adev->gart_pin_size);
 	}
 
 error:
@@ -968,10 +969,11 @@ int amdgpu_bo_unpin(struct amdgpu_bo *bo)
 		return 0;
 
 	if (bo->tbo.mem.mem_type == TTM_PL_VRAM) {
-		adev->vram_pin_size -= amdgpu_bo_size(bo);
-		adev->visible_pin_size -= amdgpu_vram_mgr_bo_visible_size(bo);
+		atomic64_sub(amdgpu_bo_size(bo), &adev->vram_pin_size);
+		atomic64_sub(amdgpu_vram_mgr_bo_visible_size(bo),
+			     &adev->visible_pin_size);
 	} else if (bo->tbo.mem.mem_type == TTM_PL_TT) {
-		adev->gart_pin_size -= amdgpu_bo_size(bo);
+		atomic64_sub(amdgpu_bo_size(bo), &adev->gart_pin_size);
 	}
 
 	for (i = 0; i < bo->placement.num_placement; i++) {
-- 
cgit 


From 964d0fbf6301d3dc8dfad19ffab5a06d002d27f1 Mon Sep 17 00:00:00 2001
From: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Date: Fri, 6 Jul 2018 14:16:54 -0400
Subject: drm/amdgpu: Allow to create BO lists in CS ioctl v3
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This change is to support MESA performace optimization.
Modify CS IOCTL to allow its input as command buffer and an array of
buffer handles to create a temporay bo list and then destroy it
when IOCTL completes.
This saves on calling for BO_LIST create and destry IOCTLs in MESA
and by this improves performance.

v2: Avoid inserting the temp list into idr struct.

v3:
Remove idr alloation from amdgpu_bo_list_create.
Remove useless argument from amdgpu_cs_parser_fini
Minor cosmetic stuff.

v4: Revert amdgpu_bo_list_destroy back to static

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h         |  8 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c | 88 +++++++++++++++++------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c      | 48 +++++++++++++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c     |  3 +-
 include/uapi/drm/amdgpu_drm.h               |  1 +
 5 files changed, 108 insertions(+), 40 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 53435da158c2..c6c1e8dc919f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -732,6 +732,14 @@ void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
 			     struct list_head *validated);
 void amdgpu_bo_list_put(struct amdgpu_bo_list *list);
 void amdgpu_bo_list_free(struct amdgpu_bo_list *list);
+int amdgpu_bo_create_list_entry_array(struct drm_amdgpu_bo_list_in *in,
+				      struct drm_amdgpu_bo_list_entry **info_param);
+
+int amdgpu_bo_list_create(struct amdgpu_device *adev,
+				 struct drm_file *filp,
+				 struct drm_amdgpu_bo_list_entry *info,
+				 unsigned num_entries,
+				 struct amdgpu_bo_list **list);
 
 /*
  * GFX stuff
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index 92be7f6de197..7679c068c89a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -55,15 +55,15 @@ static void amdgpu_bo_list_release_rcu(struct kref *ref)
 	kfree_rcu(list, rhead);
 }
 
-static int amdgpu_bo_list_create(struct amdgpu_device *adev,
+int amdgpu_bo_list_create(struct amdgpu_device *adev,
 				 struct drm_file *filp,
 				 struct drm_amdgpu_bo_list_entry *info,
 				 unsigned num_entries,
-				 int *id)
+				 struct amdgpu_bo_list **list_out)
 {
-	int r;
-	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	struct amdgpu_bo_list *list;
+	int r;
+
 
 	list = kzalloc(sizeof(struct amdgpu_bo_list), GFP_KERNEL);
 	if (!list)
@@ -78,16 +78,7 @@ static int amdgpu_bo_list_create(struct amdgpu_device *adev,
 		return r;
 	}
 
-	/* idr alloc should be called only after initialization of bo list. */
-	mutex_lock(&fpriv->bo_list_lock);
-	r = idr_alloc(&fpriv->bo_list_handles, list, 1, 0, GFP_KERNEL);
-	mutex_unlock(&fpriv->bo_list_lock);
-	if (r < 0) {
-		amdgpu_bo_list_free(list);
-		return r;
-	}
-	*id = r;
-
+	*list_out = list;
 	return 0;
 }
 
@@ -263,55 +254,79 @@ void amdgpu_bo_list_free(struct amdgpu_bo_list *list)
 	kfree(list);
 }
 
-int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
-				struct drm_file *filp)
+int amdgpu_bo_create_list_entry_array(struct drm_amdgpu_bo_list_in *in,
+				      struct drm_amdgpu_bo_list_entry **info_param)
 {
+	const void __user *uptr = u64_to_user_ptr(in->bo_info_ptr);
 	const uint32_t info_size = sizeof(struct drm_amdgpu_bo_list_entry);
-
-	struct amdgpu_device *adev = dev->dev_private;
-	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 = u64_to_user_ptr(args->in.bo_info_ptr);
-
 	struct drm_amdgpu_bo_list_entry *info;
-	struct amdgpu_bo_list *list;
-
 	int r;
 
-	info = kvmalloc_array(args->in.bo_number,
-			     sizeof(struct drm_amdgpu_bo_list_entry), GFP_KERNEL);
+	info = kvmalloc_array(in->bo_number, info_size, GFP_KERNEL);
 	if (!info)
 		return -ENOMEM;
 
 	/* copy the handle array from userspace to a kernel buffer */
 	r = -EFAULT;
-	if (likely(info_size == args->in.bo_info_size)) {
-		unsigned long bytes = args->in.bo_number *
-			args->in.bo_info_size;
+	if (likely(info_size == in->bo_info_size)) {
+		unsigned long bytes = in->bo_number *
+			in->bo_info_size;
 
 		if (copy_from_user(info, uptr, bytes))
 			goto error_free;
 
 	} else {
-		unsigned long bytes = min(args->in.bo_info_size, info_size);
+		unsigned long bytes = min(in->bo_info_size, info_size);
 		unsigned i;
 
-		memset(info, 0, args->in.bo_number * info_size);
-		for (i = 0; i < args->in.bo_number; ++i) {
+		memset(info, 0, in->bo_number * info_size);
+		for (i = 0; i < in->bo_number; ++i) {
 			if (copy_from_user(&info[i], uptr, bytes))
 				goto error_free;
 
-			uptr += args->in.bo_info_size;
+			uptr += in->bo_info_size;
 		}
 	}
 
+	*info_param = info;
+	return 0;
+
+error_free:
+	kvfree(info);
+	return r;
+}
+
+int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	union drm_amdgpu_bo_list *args = data;
+	uint32_t handle = args->in.list_handle;
+	struct drm_amdgpu_bo_list_entry *info = NULL;
+	struct amdgpu_bo_list *list;
+	int r;
+
+	r = amdgpu_bo_create_list_entry_array(&args->in, &info);
+	if (r)
+		goto error_free;
+
 	switch (args->in.operation) {
 	case AMDGPU_BO_LIST_OP_CREATE:
 		r = amdgpu_bo_list_create(adev, filp, info, args->in.bo_number,
-					  &handle);
+					  &list);
 		if (r)
 			goto error_free;
+
+		mutex_lock(&fpriv->bo_list_lock);
+		r = idr_alloc(&fpriv->bo_list_handles, list, 1, 0, GFP_KERNEL);
+		mutex_unlock(&fpriv->bo_list_lock);
+		if (r < 0) {
+			amdgpu_bo_list_free(list);
+			return r;
+		}
+
+		handle = r;
 		break;
 
 	case AMDGPU_BO_LIST_OP_DESTROY:
@@ -345,6 +360,7 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
 	return 0;
 
 error_free:
-	kvfree(info);
+	if (info)
+		kvfree(info);
 	return r;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 5a2a5ba29f9a..6d8df76b5a5d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -66,11 +66,35 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 	return 0;
 }
 
-static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
+static int amdgpu_cs_bo_handles_chunk(struct amdgpu_cs_parser *p,
+				      struct drm_amdgpu_bo_list_in *data)
+{
+	int r;
+	struct drm_amdgpu_bo_list_entry *info = NULL;
+
+	r = amdgpu_bo_create_list_entry_array(data, &info);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_list_create(p->adev, p->filp, info, data->bo_number,
+				  &p->bo_list);
+	if (r)
+		goto error_free;
+
+	kvfree(info);
+	return 0;
+
+error_free:
+	if (info)
+		kvfree(info);
+
+	return r;
+}
+
+static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
 	struct amdgpu_vm *vm = &fpriv->vm;
-	union drm_amdgpu_cs *cs = data;
 	uint64_t *chunk_array_user;
 	uint64_t *chunk_array;
 	unsigned size, num_ibs = 0;
@@ -164,6 +188,19 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 
 			break;
 
+		case AMDGPU_CHUNK_ID_BO_HANDLES:
+			size = sizeof(struct drm_amdgpu_bo_list_in);
+			if (p->chunks[i].length_dw * sizeof(uint32_t) < size) {
+				ret = -EINVAL;
+				goto free_partial_kdata;
+			}
+
+			ret = amdgpu_cs_bo_handles_chunk(p, p->chunks[i].kdata);
+			if (ret)
+				goto free_partial_kdata;
+
+			break;
+
 		case AMDGPU_CHUNK_ID_DEPENDENCIES:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
 		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
@@ -534,7 +571,12 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 	INIT_LIST_HEAD(&p->validated);
 
-	p->bo_list = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle);
+	/* p->bo_list could already be assigned if AMDGPU_CHUNK_ID_BO_HANDLES is present */
+	if (!p->bo_list)
+		p->bo_list = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle);
+	else
+		mutex_lock(&p->bo_list->lock);
+
 	if (p->bo_list) {
 		amdgpu_bo_list_get_list(p->bo_list, &p->validated);
 		if (p->bo_list->first_userptr != p->bo_list->num_entries)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 06aede194bf8..529500c94675 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -69,9 +69,10 @@
  * - 3.24.0 - Add high priority compute support for gfx9
  * - 3.25.0 - Add support for sensor query info (stable pstate sclk/mclk).
  * - 3.26.0 - GFX9: Process AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE.
+ * - 3.27.0 - Add new chunk to to AMDGPU_CS to enable BO_LIST creation.
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	26
+#define KMS_DRIVER_MINOR	27
 #define KMS_DRIVER_PATCHLEVEL	0
 
 int amdgpu_vram_limit = 0;
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h
index 784b0fe470ee..1ceec56de015 100644
--- a/include/uapi/drm/amdgpu_drm.h
+++ b/include/uapi/drm/amdgpu_drm.h
@@ -516,6 +516,7 @@ struct drm_amdgpu_gem_va {
 #define AMDGPU_CHUNK_ID_DEPENDENCIES	0x03
 #define AMDGPU_CHUNK_ID_SYNCOBJ_IN      0x04
 #define AMDGPU_CHUNK_ID_SYNCOBJ_OUT     0x05
+#define AMDGPU_CHUNK_ID_BO_HANDLES      0x06
 
 struct drm_amdgpu_cs_chunk {
 	__u32		chunk_id;
-- 
cgit 


From eb3961a57424a5c3dae44576d1c88e64a818d871 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Fri, 13 Jul 2018 09:58:49 +0200
Subject: drm/amdgpu: remove fence context from the job
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Can be obtained directly from the fence as well.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Acked-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c  | 1 -
 drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c  | 2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 1 -
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.h | 1 -
 4 files changed, 1 insertion(+), 4 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 6d8df76b5a5d..6eb7ee859ffd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -1239,7 +1239,6 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	}
 
 	job->owner = p->filp;
-	job->fence_ctx = entity->fence_context;
 	p->fence = dma_fence_get(&job->base.s_fence->finished);
 
 	r = amdgpu_ctx_add_fence(p->ctx, ring, p->fence, &seq);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index ce7739832d29..5518e623fed2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -139,7 +139,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	/* ring tests don't use a job */
 	if (job) {
 		vm = job->vm;
-		fence_ctx = job->fence_ctx;
+		fence_ctx = job->base.s_fence->scheduled.context;
 	} else {
 		vm = NULL;
 		fence_ctx = 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 2bd56760c744..10e0a97c7c03 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -132,7 +132,6 @@ int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
 		return r;
 
 	job->owner = owner;
-	job->fence_ctx = entity->fence_context;
 	*f = dma_fence_get(&job->base.s_fence->finished);
 	amdgpu_job_free_resources(job);
 	amdgpu_ring_priority_get(job->ring, job->base.s_priority);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
index 35bb93254eb2..3151692312bd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
@@ -45,7 +45,6 @@ struct amdgpu_job {
 	uint32_t		preamble_status;
 	uint32_t		num_ibs;
 	void			*owner;
-	uint64_t		fence_ctx; /* the fence_context this job uses */
 	bool                    vm_needs_flush;
 	uint64_t		vm_pd_addr;
 	unsigned		vmid;
-- 
cgit 


From 3320b8d2acd3d480d0dd4835d970067354eac915 Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Fri, 13 Jul 2018 15:08:44 +0200
Subject: drm/amdgpu: remove job->ring
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

We can easily get that from the scheduler.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Acked-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c     | 18 +++++++++---------
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c |  4 ++--
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.c    | 23 ++++++++++++-----------
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.h    |  1 -
 drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h  |  8 ++++----
 drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c    |  2 +-
 7 files changed, 29 insertions(+), 28 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index e0cc9f878e80..90780b0f0ce5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -1027,6 +1027,7 @@ struct amdgpu_cs_parser {
 
 	/* scheduler job object */
 	struct amdgpu_job	*job;
+	struct amdgpu_ring	*ring;
 
 	/* buffer objects */
 	struct ww_acquire_ctx		ticket;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 6eb7ee859ffd..72dc9b36b937 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -912,11 +912,11 @@ static int amdgpu_cs_ib_vm_chunk(struct amdgpu_device *adev,
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
 	struct amdgpu_vm *vm = &fpriv->vm;
-	struct amdgpu_ring *ring = p->job->ring;
+	struct amdgpu_ring *ring = p->ring;
 	int r;
 
 	/* Only for UVD/VCE VM emulation */
-	if (p->job->ring->funcs->parse_cs) {
+	if (p->ring->funcs->parse_cs) {
 		unsigned i, j;
 
 		for (i = 0, j = 0; i < p->nchunks && j < p->job->num_ibs; i++) {
@@ -1030,10 +1030,10 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 			}
 		}
 
-		if (parser->job->ring && parser->job->ring != ring)
+		if (parser->ring && parser->ring != ring)
 			return -EINVAL;
 
-		parser->job->ring = ring;
+		parser->ring = ring;
 
 		r =  amdgpu_ib_get(adev, vm,
 					ring->funcs->parse_cs ? chunk_ib->ib_bytes : 0,
@@ -1052,11 +1052,11 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 
 	/* UVD & VCE fw doesn't support user fences */
 	if (parser->job->uf_addr && (
-	    parser->job->ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
-	    parser->job->ring->funcs->type == AMDGPU_RING_TYPE_VCE))
+	    parser->ring->funcs->type == AMDGPU_RING_TYPE_UVD ||
+	    parser->ring->funcs->type == AMDGPU_RING_TYPE_VCE))
 		return -EINVAL;
 
-	return amdgpu_ctx_wait_prev_fence(parser->ctx, parser->job->ring->idx);
+	return amdgpu_ctx_wait_prev_fence(parser->ctx, parser->ring->idx);
 }
 
 static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
@@ -1207,7 +1207,7 @@ static void amdgpu_cs_post_dependencies(struct amdgpu_cs_parser *p)
 static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 			    union drm_amdgpu_cs *cs)
 {
-	struct amdgpu_ring *ring = p->job->ring;
+	struct amdgpu_ring *ring = p->ring;
 	struct drm_sched_entity *entity = &p->ctx->rings[ring->idx].entity;
 	struct amdgpu_job *job;
 	unsigned i;
@@ -1256,7 +1256,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	job->uf_sequence = seq;
 
 	amdgpu_job_free_resources(job);
-	amdgpu_ring_priority_get(job->ring, job->base.s_priority);
+	amdgpu_ring_priority_get(p->ring, job->base.s_priority);
 
 	trace_amdgpu_cs_ioctl(job);
 	drm_sched_entity_push_job(&job->base, entity);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 2720444ff23a..2b2de5f3e6e3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -3253,7 +3253,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 
 		kthread_park(ring->sched.thread);
 
-		if (job && job->ring->idx != i)
+		if (job && job->base.sched == &ring->sched)
 			continue;
 
 		drm_sched_hw_job_reset(&ring->sched, &job->base);
@@ -3277,7 +3277,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 		 * or all rings (in the case @job is NULL)
 		 * after above amdgpu_reset accomplished
 		 */
-		if ((!job || job->ring->idx == i) && !r)
+		if ((!job || job->base.sched == &ring->sched) && !r)
 			drm_sched_job_recovery(&ring->sched);
 
 		kthread_unpark(ring->sched.thread);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 51ff751e093b..2496f2269bcb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -30,12 +30,12 @@
 
 static void amdgpu_job_timedout(struct drm_sched_job *s_job)
 {
-	struct amdgpu_job *job = container_of(s_job, struct amdgpu_job, base);
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched);
+	struct amdgpu_job *job = to_amdgpu_job(s_job);
 
 	DRM_ERROR("ring %s timeout, last signaled seq=%u, last emitted seq=%u\n",
-		  job->base.sched->name,
-		  atomic_read(&job->ring->fence_drv.last_seq),
-		  job->ring->fence_drv.sync_seq);
+		  job->base.sched->name, atomic_read(&ring->fence_drv.last_seq),
+		  ring->fence_drv.sync_seq);
 
 	amdgpu_device_gpu_recover(job->adev, job, false);
 }
@@ -98,9 +98,10 @@ void amdgpu_job_free_resources(struct amdgpu_job *job)
 
 static void amdgpu_job_free_cb(struct drm_sched_job *s_job)
 {
-	struct amdgpu_job *job = container_of(s_job, struct amdgpu_job, base);
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_job->sched);
+	struct amdgpu_job *job = to_amdgpu_job(s_job);
 
-	amdgpu_ring_priority_put(job->ring, s_job->s_priority);
+	amdgpu_ring_priority_put(ring, s_job->s_priority);
 	dma_fence_put(job->fence);
 	amdgpu_sync_free(&job->sync);
 	amdgpu_sync_free(&job->sched_sync);
@@ -120,6 +121,7 @@ void amdgpu_job_free(struct amdgpu_job *job)
 int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
 		      void *owner, struct dma_fence **f)
 {
+	struct amdgpu_ring *ring = to_amdgpu_ring(entity->sched);
 	int r;
 
 	if (!f)
@@ -130,10 +132,9 @@ int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
 		return r;
 
 	job->owner = owner;
-	job->ring = to_amdgpu_ring(entity->sched);
 	*f = dma_fence_get(&job->base.s_fence->finished);
 	amdgpu_job_free_resources(job);
-	amdgpu_ring_priority_get(job->ring, job->base.s_priority);
+	amdgpu_ring_priority_get(ring, job->base.s_priority);
 	drm_sched_entity_push_job(&job->base, entity);
 
 	return 0;
@@ -142,6 +143,7 @@ int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
 static struct dma_fence *amdgpu_job_dependency(struct drm_sched_job *sched_job,
 					       struct drm_sched_entity *s_entity)
 {
+	struct amdgpu_ring *ring = to_amdgpu_ring(s_entity->sched);
 	struct amdgpu_job *job = to_amdgpu_job(sched_job);
 	struct amdgpu_vm *vm = job->vm;
 	bool explicit = false;
@@ -157,8 +159,6 @@ static struct dma_fence *amdgpu_job_dependency(struct drm_sched_job *sched_job,
 	}
 
 	while (fence == NULL && vm && !job->vmid) {
-		struct amdgpu_ring *ring = job->ring;
-
 		r = amdgpu_vmid_grab(vm, ring, &job->sync,
 				     &job->base.s_fence->finished,
 				     job);
@@ -173,6 +173,7 @@ static struct dma_fence *amdgpu_job_dependency(struct drm_sched_job *sched_job,
 
 static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 {
+	struct amdgpu_ring *ring = to_amdgpu_ring(sched_job->sched);
 	struct dma_fence *fence = NULL, *finished;
 	struct amdgpu_device *adev;
 	struct amdgpu_job *job;
@@ -196,7 +197,7 @@ static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 	if (finished->error < 0) {
 		DRM_INFO("Skip scheduling IBs!\n");
 	} else {
-		r = amdgpu_ib_schedule(job->ring, job->num_ibs, job->ibs, job,
+		r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, job,
 				       &fence);
 		if (r)
 			DRM_ERROR("Error scheduling IBs (%d)\n", r);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
index 39f4230e1d37..c663c1925f91 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
@@ -37,7 +37,6 @@ struct amdgpu_job {
 	struct drm_sched_job    base;
 	struct amdgpu_device	*adev;
 	struct amdgpu_vm	*vm;
-	struct amdgpu_ring	*ring;
 	struct amdgpu_sync	sync;
 	struct amdgpu_sync	sched_sync;
 	struct amdgpu_ib	*ibs;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index e96e26d3f3b0..76920035eb22 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -150,10 +150,10 @@ TRACE_EVENT(amdgpu_cs,
 
 	    TP_fast_assign(
 			   __entry->bo_list = p->bo_list;
-			   __entry->ring = p->job->ring->idx;
+			   __entry->ring = p->ring->idx;
 			   __entry->dw = p->job->ibs[i].length_dw;
 			   __entry->fences = amdgpu_fence_count_emitted(
-				p->job->ring);
+				p->ring);
 			   ),
 	    TP_printk("bo_list=%p, ring=%u, dw=%u, fences=%u",
 		      __entry->bo_list, __entry->ring, __entry->dw,
@@ -178,7 +178,7 @@ TRACE_EVENT(amdgpu_cs_ioctl,
 			   __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
 			   __entry->context = job->base.s_fence->finished.context;
 			   __entry->seqno = job->base.s_fence->finished.seqno;
-			   __entry->ring_name = job->ring->name;
+			   __entry->ring_name = to_amdgpu_ring(job->base.sched)->name;
 			   __entry->num_ibs = job->num_ibs;
 			   ),
 	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
@@ -203,7 +203,7 @@ TRACE_EVENT(amdgpu_sched_run_job,
 			   __assign_str(timeline, AMDGPU_JOB_GET_TIMELINE_NAME(job))
 			   __entry->context = job->base.s_fence->finished.context;
 			   __entry->seqno = job->base.s_fence->finished.seqno;
-			   __entry->ring_name = job->ring->name;
+			   __entry->ring_name = to_amdgpu_ring(job->base.sched)->name;
 			   __entry->num_ibs = job->num_ibs;
 			   ),
 	    TP_printk("sched_job=%llu, timeline=%s, context=%u, seqno=%u, ring_name=%s, num_ibs=%u",
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index 848b2e898818..7738d2b90dae 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -692,11 +692,11 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
 			     struct amdgpu_bo *bo, unsigned offset)
 {
 	struct amdgpu_device *adev = ctx->parser->adev;
+	uint32_t ip_instance = ctx->parser->ring->me;
 	int32_t *msg, msg_type, handle;
 	void *ptr;
 	long r;
 	int i;
-	uint32_t ip_instance = ctx->parser->job->ring->me;
 
 	if (offset & 0x3F) {
 		DRM_ERROR("UVD(%d) messages must be 64 byte aligned!\n", ip_instance);
-- 
cgit 


From b5286801705b39e44b824520714d503f882fef1c Mon Sep 17 00:00:00 2001
From: Christian König <christian.koenig@amd.com>
Date: Mon, 16 Jul 2018 14:58:48 +0200
Subject: drm/amdgpu: change ring priority after pushing the job (v2)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Pushing a job can change the ring assignment of an entity.

v2: squash in:
"drm/amdgpu: fix job priority handling" (Christian)

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@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_job.c | 8 ++++++--
 2 files changed, 11 insertions(+), 3 deletions(-)

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 72dc9b36b937..7c5cc33d0cda 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -1209,6 +1209,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 {
 	struct amdgpu_ring *ring = p->ring;
 	struct drm_sched_entity *entity = &p->ctx->rings[ring->idx].entity;
+	enum drm_sched_priority priority;
 	struct amdgpu_job *job;
 	unsigned i;
 	uint64_t seq;
@@ -1256,11 +1257,14 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	job->uf_sequence = seq;
 
 	amdgpu_job_free_resources(job);
-	amdgpu_ring_priority_get(p->ring, job->base.s_priority);
 
 	trace_amdgpu_cs_ioctl(job);
+	priority = job->base.s_priority;
 	drm_sched_entity_push_job(&job->base, entity);
 
+	ring = to_amdgpu_ring(entity->sched);
+	amdgpu_ring_priority_get(ring, priority);
+
 	ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);
 	amdgpu_mn_unlock(p->mn);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 42a4764d728e..5a2c26a85984 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -126,7 +126,8 @@ void amdgpu_job_free(struct amdgpu_job *job)
 int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
 		      void *owner, struct dma_fence **f)
 {
-	struct amdgpu_ring *ring = to_amdgpu_ring(entity->sched);
+	enum drm_sched_priority priority;
+	struct amdgpu_ring *ring;
 	int r;
 
 	if (!f)
@@ -139,9 +140,12 @@ int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
 	job->owner = owner;
 	*f = dma_fence_get(&job->base.s_fence->finished);
 	amdgpu_job_free_resources(job);
-	amdgpu_ring_priority_get(ring, job->base.s_priority);
+	priority = job->base.s_priority;
 	drm_sched_entity_push_job(&job->base, entity);
 
+	ring = to_amdgpu_ring(entity->sched);
+	amdgpu_ring_priority_get(ring, priority);
+
 	return 0;
 }
 
-- 
cgit