drm/amdgpu: Add EXT_COHERENT memory allocation flags

These flags (for GEM and SVM allocations) allocate
memory that allows for system-scope atomic semantics.

On GFX943 these flags cause caches to be avoided on
non-local memory.

On all other ASICs they are identical in functionality to the
equivalent COHERENT flags.

Corresponding Thunk patch is at
https://github.com/RadeonOpenCompute/ROCT-Thunk-Interface/pull/88

Reviewed-by: David Yat Sin <David.YatSin@amd.com>
Signed-off-by: David Francis <David.Francis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

6 files changed, 18 insertions, 2 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 57189b8bc214..c48ca5adad8a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -1691,6 +1691,8 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
 
 	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_COHERENT)
 		alloc_flags |= AMDGPU_GEM_CREATE_COHERENT;
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_EXT_COHERENT)
+		alloc_flags |= AMDGPU_GEM_CREATE_EXT_COHERENT;
 	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED)
 		alloc_flags |= AMDGPU_GEM_CREATE_UNCACHED;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
index 12210598e5b8..76b618735dc0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
@@ -331,6 +331,7 @@ amdgpu_dma_buf_create_obj(struct drm_device *dev, struct dma_buf *dma_buf)
 
 		flags |= other->flags & (AMDGPU_GEM_CREATE_CPU_GTT_USWC |
 					 AMDGPU_GEM_CREATE_COHERENT |
+					 AMDGPU_GEM_CREATE_EXT_COHERENT |
 					 AMDGPU_GEM_CREATE_UNCACHED);
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
index d3da13f4c80e..e582073b57c8 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@@ -635,6 +635,7 @@ static void gmc_v10_0_get_vm_pte(struct amdgpu_device *adev,
 	}
 
 	if (bo && bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+			       AMDGPU_GEM_CREATE_EXT_COHERENT |
 			       AMDGPU_GEM_CREATE_UNCACHED))
 		*flags = (*flags & ~AMDGPU_PTE_MTYPE_NV10_MASK) |
 			 AMDGPU_PTE_MTYPE_NV10(MTYPE_UC);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
index e1f47f9c1881..69f65e9c4f93 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
@@ -543,6 +543,7 @@ static void gmc_v11_0_get_vm_pte(struct amdgpu_device *adev,
 	}
 
 	if (bo && bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+			       AMDGPU_GEM_CREATE_EXT_COHERENT |
 			       AMDGPU_GEM_CREATE_UNCACHED))
 		*flags = (*flags & ~AMDGPU_PTE_MTYPE_NV10_MASK) |
 			 AMDGPU_PTE_MTYPE_NV10(MTYPE_UC);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 268ee533e7c1..2936a0fb7527 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1187,7 +1187,8 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 {
 	struct amdgpu_device *bo_adev = amdgpu_ttm_adev(bo->tbo.bdev);
 	bool is_vram = bo->tbo.resource->mem_type == TTM_PL_VRAM;
-	bool coherent = bo->flags & AMDGPU_GEM_CREATE_COHERENT;
+	bool coherent = bo->flags & (AMDGPU_GEM_CREATE_COHERENT | AMDGPU_GEM_CREATE_EXT_COHERENT);
+	bool ext_coherent = bo->flags & AMDGPU_GEM_CREATE_EXT_COHERENT;
 	bool uncached = bo->flags & AMDGPU_GEM_CREATE_UNCACHED;
 	struct amdgpu_vm *vm = mapping->bo_va->base.vm;
 	unsigned int mtype_local, mtype;
@@ -1257,6 +1258,8 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 		snoop = true;
 		if (uncached) {
 			mtype = MTYPE_UC;
+		} else if (ext_coherent) {
+			mtype = is_local ? MTYPE_CC : MTYPE_UC;
 		} else if (adev->flags & AMD_IS_APU) {
 			mtype = is_local ? mtype_local : MTYPE_NC;
 		} else {
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index ae5660861133..9a1952114e5b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1189,7 +1189,8 @@ svm_range_get_pte_flags(struct kfd_node *node,
 	uint32_t mapping_flags = 0;
 	uint64_t pte_flags;
 	bool snoop = (domain != SVM_RANGE_VRAM_DOMAIN);
-	bool coherent = flags & KFD_IOCTL_SVM_FLAG_COHERENT;
+	bool coherent = flags & (KFD_IOCTL_SVM_FLAG_COHERENT | KFD_IOCTL_SVM_FLAG_EXT_COHERENT);
+	bool ext_coherent = flags & KFD_IOCTL_SVM_FLAG_EXT_COHERENT;
 	bool uncached = false; /*flags & KFD_IOCTL_SVM_FLAG_UNCACHED;*/
 	unsigned int mtype_local;
 
@@ -1237,6 +1238,13 @@ svm_range_get_pte_flags(struct kfd_node *node,
 		snoop = true;
 		if (uncached) {
 			mapping_flags |= AMDGPU_VM_MTYPE_UC;
+		} else if (ext_coherent) {
+			/* local HBM region close to partition */
+			if (bo_node->adev == node->adev &&
+			    (!bo_node->xcp || !node->xcp || bo_node->xcp->mem_id == node->xcp->mem_id))
+				mapping_flags |= AMDGPU_VM_MTYPE_CC;
+			else
+				mapping_flags |= AMDGPU_VM_MTYPE_UC;
 		} else if (domain == SVM_RANGE_VRAM_DOMAIN) {
 			/* local HBM region close to partition */
 			if (bo_node->adev == node->adev &&