Merge tag 'amd-drm-next-5.14-2021-05-19' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-5.14-2021-05-19:

amdgpu:
- Aldebaran updates
- More LTTPR display work
- Vangogh updates
- SDMA 5.x GCR fixes
- RAS fixes
- PCIe ASPM support
- Modifier fixes
- Enable TMZ on Renoir
- Buffer object code cleanup
- Display overlay fixes
- Initial support for multiple eDP panels
- Initial SR-IOV support for Aldebaran
- DP link training refactor
- Misc code cleanups and bug fixes
- SMU regression fixes for variable sized arrays
- MAINTAINERS fixes for amdgpu

amdkfd:
- Initial SR-IOV support for Aldebaran
- Topology fixes
- Initial HMM SVM support
- Misc code cleanups and bug fixes

radeon:
- Misc code cleanups and bug fixes
- SMU regression fixes for variable sized arrays
- Flickering fix for Oland with multiple 4K displays

UAPI:
- amdgpu: Drop AMDGPU_GEM_CREATE_SHADOW flag.
  This was always a kernel internal flag and userspace use of it has always been blocked.
  It's no longer needed so remove it.
- amdkgd: HMM SVM support
  Overview: https://patchwork.freedesktop.org/series/85562/
  Porposed userspace: https://github.com/RadeonOpenCompute/ROCT-Thunk-Interface/tree/fxkamd/hmm-wip

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210520031258.231896-1-alexander.deucher@amd.com

1 files changed, 72 insertions, 52 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 4a3e3f72e127..edc63d3e087e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -38,6 +38,7 @@
 #include "amdgpu_gmc.h"
 #include "amdgpu_xgmi.h"
 #include "amdgpu_dma_buf.h"
+#include "kfd_svm.h"
 
 /**
  * DOC: GPUVM
@@ -850,35 +851,60 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_vm_bo_param - fill in parameters for PD/PT allocation
+ * amdgpu_vm_pt_create - create bo for PD/PT
  *
  * @adev: amdgpu_device pointer
  * @vm: requesting vm
  * @level: the page table level
  * @immediate: use a immediate update
- * @bp: resulting BO allocation parameters
+ * @bo: pointer to the buffer object pointer
  */
-static void amdgpu_vm_bo_param(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+static int amdgpu_vm_pt_create(struct amdgpu_device *adev,
+			       struct amdgpu_vm *vm,
 			       int level, bool immediate,
-			       struct amdgpu_bo_param *bp)
+			       struct amdgpu_bo **bo)
 {
-	memset(bp, 0, sizeof(*bp));
+	struct amdgpu_bo_param bp;
+	int r;
 
-	bp->size = amdgpu_vm_bo_size(adev, level);
-	bp->byte_align = AMDGPU_GPU_PAGE_SIZE;
-	bp->domain = AMDGPU_GEM_DOMAIN_VRAM;
-	bp->domain = amdgpu_bo_get_preferred_pin_domain(adev, bp->domain);
-	bp->flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
+	memset(&bp, 0, sizeof(bp));
+
+	bp.size = amdgpu_vm_bo_size(adev, level);
+	bp.byte_align = AMDGPU_GPU_PAGE_SIZE;
+	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
+	bp.domain = amdgpu_bo_get_preferred_pin_domain(adev, bp.domain);
+	bp.flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
 		AMDGPU_GEM_CREATE_CPU_GTT_USWC;
-	bp->bo_ptr_size = sizeof(struct amdgpu_bo);
+	bp.bo_ptr_size = sizeof(struct amdgpu_bo);
 	if (vm->use_cpu_for_update)
-		bp->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
-	else if (!vm->root.base.bo || vm->root.base.bo->shadow)
-		bp->flags |= AMDGPU_GEM_CREATE_SHADOW;
-	bp->type = ttm_bo_type_kernel;
-	bp->no_wait_gpu = immediate;
+		bp.flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+
+	bp.type = ttm_bo_type_kernel;
+	bp.no_wait_gpu = immediate;
 	if (vm->root.base.bo)
-		bp->resv = vm->root.base.bo->tbo.base.resv;
+		bp.resv = vm->root.base.bo->tbo.base.resv;
+
+	r = amdgpu_bo_create(adev, &bp, bo);
+	if (r)
+		return r;
+
+	if (vm->is_compute_context && (adev->flags & AMD_IS_APU))
+		return 0;
+
+	if (!bp.resv)
+		WARN_ON(dma_resv_lock((*bo)->tbo.base.resv,
+				      NULL));
+	r = amdgpu_bo_create_shadow(adev, bp.size, *bo);
+
+	if (!bp.resv)
+		dma_resv_unlock((*bo)->tbo.base.resv);
+
+	if (r) {
+		amdgpu_bo_unref(bo);
+		return r;
+	}
+
+	return 0;
 }
 
 /**
@@ -901,7 +927,6 @@ static int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
 			       bool immediate)
 {
 	struct amdgpu_vm_pt *entry = cursor->entry;
-	struct amdgpu_bo_param bp;
 	struct amdgpu_bo *pt;
 	int r;
 
@@ -919,9 +944,7 @@ static int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
 	if (entry->base.bo)
 		return 0;
 
-	amdgpu_vm_bo_param(adev, vm, cursor->level, immediate, &bp);
-
-	r = amdgpu_bo_create(adev, &bp, &pt);
+	r = amdgpu_vm_pt_create(adev, vm, cursor->level, immediate, &pt);
 	if (r)
 		return r;
 
@@ -1593,15 +1616,15 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params,
  * Returns:
  * 0 for success, -EINVAL for failure.
  */
-static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
-				       struct amdgpu_device *bo_adev,
-				       struct amdgpu_vm *vm, bool immediate,
-				       bool unlocked, struct dma_resv *resv,
-				       uint64_t start, uint64_t last,
-				       uint64_t flags, uint64_t offset,
-				       struct drm_mm_node *nodes,
-				       dma_addr_t *pages_addr,
-				       struct dma_fence **fence)
+int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
+				struct amdgpu_device *bo_adev,
+				struct amdgpu_vm *vm, bool immediate,
+				bool unlocked, struct dma_resv *resv,
+				uint64_t start, uint64_t last,
+				uint64_t flags, uint64_t offset,
+				struct drm_mm_node *nodes,
+				dma_addr_t *pages_addr,
+				struct dma_fence **fence)
 {
 	struct amdgpu_vm_update_params params;
 	enum amdgpu_sync_mode sync_mode;
@@ -2818,7 +2841,6 @@ long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
- * @vm_context: Indicates if it GFX or Compute context
  * @pasid: Process address space identifier
  *
  * Init @vm fields.
@@ -2826,10 +2848,8 @@ long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
  * Returns:
  * 0 for success, error for failure.
  */
-int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-		   int vm_context, u32 pasid)
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, u32 pasid)
 {
-	struct amdgpu_bo_param bp;
 	struct amdgpu_bo *root;
 	int r, i;
 
@@ -2861,16 +2881,9 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	vm->pte_support_ats = false;
 	vm->is_compute_context = false;
 
-	if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE) {
-		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
-						AMDGPU_VM_USE_CPU_FOR_COMPUTE);
+	vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
+				    AMDGPU_VM_USE_CPU_FOR_GFX);
 
-		if (adev->asic_type == CHIP_RAVEN)
-			vm->pte_support_ats = true;
-	} else {
-		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
-						AMDGPU_VM_USE_CPU_FOR_GFX);
-	}
 	DRM_DEBUG_DRIVER("VM update mode is %s\n",
 			 vm->use_cpu_for_update ? "CPU" : "SDMA");
 	WARN_ONCE((vm->use_cpu_for_update &&
@@ -2887,10 +2900,8 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	mutex_init(&vm->eviction_lock);
 	vm->evicting = false;
 
-	amdgpu_vm_bo_param(adev, vm, adev->vm_manager.root_level, false, &bp);
-	if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE)
-		bp.flags &= ~AMDGPU_GEM_CREATE_SHADOW;
-	r = amdgpu_bo_create(adev, &bp, &root);
+	r = amdgpu_vm_pt_create(adev, vm, adev->vm_manager.root_level,
+				false, &root);
 	if (r)
 		goto error_free_delayed;
 
@@ -3349,6 +3360,7 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
 bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 			    uint64_t addr)
 {
+	bool is_compute_context = false;
 	struct amdgpu_bo *root;
 	uint64_t value, flags;
 	struct amdgpu_vm *vm;
@@ -3356,15 +3368,25 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 
 	spin_lock(&adev->vm_manager.pasid_lock);
 	vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
-	if (vm)
+	if (vm) {
 		root = amdgpu_bo_ref(vm->root.base.bo);
-	else
+		is_compute_context = vm->is_compute_context;
+	} else {
 		root = NULL;
+	}
 	spin_unlock(&adev->vm_manager.pasid_lock);
 
 	if (!root)
 		return false;
 
+	addr /= AMDGPU_GPU_PAGE_SIZE;
+
+	if (is_compute_context &&
+	    !svm_range_restore_pages(adev, pasid, addr)) {
+		amdgpu_bo_unref(&root);
+		return true;
+	}
+
 	r = amdgpu_bo_reserve(root, true);
 	if (r)
 		goto error_unref;
@@ -3378,18 +3400,16 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 	if (!vm)
 		goto error_unlock;
 
-	addr /= AMDGPU_GPU_PAGE_SIZE;
 	flags = AMDGPU_PTE_VALID | AMDGPU_PTE_SNOOPED |
 		AMDGPU_PTE_SYSTEM;
 
-	if (vm->is_compute_context) {
+	if (is_compute_context) {
 		/* Intentionally setting invalid PTE flag
 		 * combination to force a no-retry-fault
 		 */
 		flags = AMDGPU_PTE_EXECUTABLE | AMDGPU_PDE_PTE |
 			AMDGPU_PTE_TF;
 		value = 0;
-
 	} else if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_NEVER) {
 		/* Redirect the access to the dummy page */
 		value = adev->dummy_page_addr;