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 ++-
1 file changed, 2 insertions(+), 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..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;
}
--
cgit v1.2.3
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 v1.2.3
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 v1.2.3
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 v1.2.3