Merge tag 'drm-amdkfd-next-2018-03-11' of git://people.freedesktop.org/~gabbayo/linux into drm-next

Major points for this pull request:
- Add dGPU support for amdkfd initialization code and queue handling. It's
  not complete support since the GPUVM part is missing (the under debate stuff).
- Enable PCIe atomics for dGPU if present
- Various adjustments to the amdgpu<-->amdkfd interface for dGPUs
- Refactor IOMMUv2 code to allow loading amdkfd without IOMMUv2 in the system
- Add HSA process eviction code in case of system memory pressure
- Various fixes and small changes

* tag 'drm-amdkfd-next-2018-03-11' of git://people.freedesktop.org/~gabbayo/linux: (24 commits)
  uapi: Fix type used in ioctl parameter structures
  drm/amdkfd: Implement KFD process eviction/restore
  drm/amdkfd: Add GPUVM virtual address space to PDD
  drm/amdkfd: Remove unaligned memory access
  drm/amdkfd: Centralize IOMMUv2 code and make it conditional
  drm/amdgpu: Add submit IB function for KFD
  drm/amdgpu: Add GPUVM memory management functions for KFD
  drm/amdgpu: add amdgpu_sync_clone
  drm/amdgpu: Update kgd2kfd_shared_resources for dGPU support
  drm/amdgpu: Add KFD eviction fence
  drm/amdgpu: Remove unused kfd2kgd interface
  drm/amdgpu: Fix wrong mask in get_atc_vmid_pasid_mapping_pasid
  drm/amdgpu: Fix header file dependencies
  drm/amdgpu: Replace kgd_mem with amdgpu_bo for kernel pinned gtt mem
  drm/amdgpu: remove useless BUG_ONs
  drm/amdgpu: Enable KFD initialization on dGPUs
  drm/amdkfd: Add dGPU device IDs and device info
  drm/amdkfd: Add dGPU support to kernel_queue_init
  drm/amdkfd: Add dGPU support to the MQD manager
  drm/amdkfd: Add dGPU support to the device queue manager
  ...

1 files changed, 101 insertions, 31 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 450426dbed92..8a23aa8f9c73 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -30,6 +30,8 @@
 const struct kgd2kfd_calls *kgd2kfd;
 bool (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
 
+static const unsigned int compute_vmid_bitmap = 0xFF00;
+
 int amdgpu_amdkfd_init(void)
 {
 	int ret;
@@ -56,6 +58,7 @@ int amdgpu_amdkfd_init(void)
 #else
 	ret = -ENOENT;
 #endif
+	amdgpu_amdkfd_gpuvm_init_mem_limits();
 
 	return ret;
 }
@@ -78,10 +81,15 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
 	switch (adev->asic_type) {
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	case CHIP_KAVERI:
+	case CHIP_HAWAII:
 		kfd2kgd = amdgpu_amdkfd_gfx_7_get_functions();
 		break;
 #endif
 	case CHIP_CARRIZO:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_POLARIS10:
+	case CHIP_POLARIS11:
 		kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
 		break;
 	default:
@@ -132,9 +140,13 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 	int last_valid_bit;
 	if (adev->kfd) {
 		struct kgd2kfd_shared_resources gpu_resources = {
-			.compute_vmid_bitmap = 0xFF00,
+			.compute_vmid_bitmap = compute_vmid_bitmap,
 			.num_pipe_per_mec = adev->gfx.mec.num_pipe_per_mec,
-			.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe
+			.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe,
+			.gpuvm_size = min(adev->vm_manager.max_pfn
+					  << AMDGPU_GPU_PAGE_SHIFT,
+					  AMDGPU_VA_HOLE_START),
+			.drm_render_minor = adev->ddev->render->index
 		};
 
 		/* this is going to have a few of the MSBs set that we need to
@@ -204,19 +216,13 @@ int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
 			void **cpu_ptr)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
-	struct kgd_mem **mem = (struct kgd_mem **) mem_obj;
+	struct amdgpu_bo *bo = NULL;
 	int r;
-
-	BUG_ON(kgd == NULL);
-	BUG_ON(gpu_addr == NULL);
-	BUG_ON(cpu_ptr == NULL);
-
-	*mem = kmalloc(sizeof(struct kgd_mem), GFP_KERNEL);
-	if ((*mem) == NULL)
-		return -ENOMEM;
+	uint64_t gpu_addr_tmp = 0;
+	void *cpu_ptr_tmp = NULL;
 
 	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_GTT,
-			     AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, &(*mem)->bo);
+			AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, &bo);
 	if (r) {
 		dev_err(adev->dev,
 			"failed to allocate BO for amdkfd (%d)\n", r);
@@ -224,54 +230,53 @@ int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
 	}
 
 	/* map the buffer */
-	r = amdgpu_bo_reserve((*mem)->bo, true);
+	r = amdgpu_bo_reserve(bo, true);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to reserve bo for amdkfd\n", r);
 		goto allocate_mem_reserve_bo_failed;
 	}
 
-	r = amdgpu_bo_pin((*mem)->bo, AMDGPU_GEM_DOMAIN_GTT,
-				&(*mem)->gpu_addr);
+	r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT,
+				&gpu_addr_tmp);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to pin bo for amdkfd\n", r);
 		goto allocate_mem_pin_bo_failed;
 	}
-	*gpu_addr = (*mem)->gpu_addr;
 
-	r = amdgpu_bo_kmap((*mem)->bo, &(*mem)->cpu_ptr);
+	r = amdgpu_bo_kmap(bo, &cpu_ptr_tmp);
 	if (r) {
 		dev_err(adev->dev,
 			"(%d) failed to map bo to kernel for amdkfd\n", r);
 		goto allocate_mem_kmap_bo_failed;
 	}
-	*cpu_ptr = (*mem)->cpu_ptr;
 
-	amdgpu_bo_unreserve((*mem)->bo);
+	*mem_obj = bo;
+	*gpu_addr = gpu_addr_tmp;
+	*cpu_ptr = cpu_ptr_tmp;
+
+	amdgpu_bo_unreserve(bo);
 
 	return 0;
 
 allocate_mem_kmap_bo_failed:
-	amdgpu_bo_unpin((*mem)->bo);
+	amdgpu_bo_unpin(bo);
 allocate_mem_pin_bo_failed:
-	amdgpu_bo_unreserve((*mem)->bo);
+	amdgpu_bo_unreserve(bo);
 allocate_mem_reserve_bo_failed:
-	amdgpu_bo_unref(&(*mem)->bo);
+	amdgpu_bo_unref(&bo);
 
 	return r;
 }
 
 void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj)
 {
-	struct kgd_mem *mem = (struct kgd_mem *) mem_obj;
+	struct amdgpu_bo *bo = (struct amdgpu_bo *) mem_obj;
 
-	BUG_ON(mem == NULL);
-
-	amdgpu_bo_reserve(mem->bo, true);
-	amdgpu_bo_kunmap(mem->bo);
-	amdgpu_bo_unpin(mem->bo);
-	amdgpu_bo_unreserve(mem->bo);
-	amdgpu_bo_unref(&(mem->bo));
-	kfree(mem);
+	amdgpu_bo_reserve(bo, true);
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unpin(bo);
+	amdgpu_bo_unreserve(bo);
+	amdgpu_bo_unref(&(bo));
 }
 
 void get_local_mem_info(struct kgd_dev *kgd,
@@ -361,3 +366,68 @@ uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd)
 
 	return amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 }
+
+int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
+				uint32_t vmid, uint64_t gpu_addr,
+				uint32_t *ib_cmd, uint32_t ib_len)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	struct amdgpu_ring *ring;
+	struct dma_fence *f = NULL;
+	int ret;
+
+	switch (engine) {
+	case KGD_ENGINE_MEC1:
+		ring = &adev->gfx.compute_ring[0];
+		break;
+	case KGD_ENGINE_SDMA1:
+		ring = &adev->sdma.instance[0].ring;
+		break;
+	case KGD_ENGINE_SDMA2:
+		ring = &adev->sdma.instance[1].ring;
+		break;
+	default:
+		pr_err("Invalid engine in IB submission: %d\n", engine);
+		ret = -EINVAL;
+		goto err;
+	}
+
+	ret = amdgpu_job_alloc(adev, 1, &job, NULL);
+	if (ret)
+		goto err;
+
+	ib = &job->ibs[0];
+	memset(ib, 0, sizeof(struct amdgpu_ib));
+
+	ib->gpu_addr = gpu_addr;
+	ib->ptr = ib_cmd;
+	ib->length_dw = ib_len;
+	/* This works for NO_HWS. TODO: need to handle without knowing VMID */
+	job->vmid = vmid;
+
+	ret = amdgpu_ib_schedule(ring, 1, ib, job, &f);
+	if (ret) {
+		DRM_ERROR("amdgpu: failed to schedule IB.\n");
+		goto err_ib_sched;
+	}
+
+	ret = dma_fence_wait(f, false);
+
+err_ib_sched:
+	dma_fence_put(f);
+	amdgpu_job_free(job);
+err:
+	return ret;
+}
+
+bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid)
+{
+	if (adev->kfd) {
+		if ((1 << vmid) & compute_vmid_bitmap)
+			return true;
+	}
+
+	return false;
+}