drm/amdgpu: correctly sign extend 48bit addresses v3

Correct sign extend the GMC addresses to 48bit.

v2: sign extending turned out easier than thought.
v3: clean up the defines and move them into amdgpu_gmc.h as well

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c

@@ -135,7 +135,7 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 			.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),
+					  AMDGPU_GMC_HOLE_START),
 			.drm_render_minor = adev->ddev->render->index
 		};
 

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

@@ -835,7 +835,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 			if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
 				continue;
 
-			va_start = chunk_ib->va_start & AMDGPU_VA_HOLE_MASK;
+			va_start = chunk_ib->va_start & AMDGPU_GMC_HOLE_MASK;
 			r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
 			if (r) {
 				DRM_ERROR("IB va_start is invalid\n");

drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c

@@ -572,16 +572,16 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 		return -EINVAL;
 	}
 
-	if (args->va_address >= AMDGPU_VA_HOLE_START &&
-	    args->va_address < AMDGPU_VA_HOLE_END) {
+	if (args->va_address >= AMDGPU_GMC_HOLE_START &&
+	    args->va_address < AMDGPU_GMC_HOLE_END) {
 		dev_dbg(&dev->pdev->dev,
 			"va_address 0x%LX is in VA hole 0x%LX-0x%LX\n",
-			args->va_address, AMDGPU_VA_HOLE_START,
-			AMDGPU_VA_HOLE_END);
+			args->va_address, AMDGPU_GMC_HOLE_START,
+			AMDGPU_GMC_HOLE_END);
 		return -EINVAL;
 	}
 
-	args->va_address &= AMDGPU_VA_HOLE_MASK;
+	args->va_address &= AMDGPU_GMC_HOLE_MASK;
 
 	if ((args->flags & ~valid_flags) && (args->flags & ~prt_flags)) {
 		dev_dbg(&dev->pdev->dev, "invalid flags combination 0x%08X\n",

drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h

@@ -30,6 +30,19 @@
 
 #include "amdgpu_irq.h"
 
+/* VA hole for 48bit addresses on Vega10 */
+#define AMDGPU_GMC_HOLE_START	0x0000800000000000ULL
+#define AMDGPU_GMC_HOLE_END	0xffff800000000000ULL
+
+/*
+ * Hardware is programmed as if the hole doesn't exists with start and end
+ * address values.
+ *
+ * This mask is used to remove the upper 16bits of the VA and so come up with
+ * the linear addr value.
+ */
+#define AMDGPU_GMC_HOLE_MASK	0x0000ffffffffffffULL
+
 struct firmware;
 
 /*
@@ -133,6 +146,19 @@ static inline bool amdgpu_gmc_vram_full_visible(struct amdgpu_gmc *gmc)
 	return (gmc->real_vram_size == gmc->visible_vram_size);
 }
 
+/**
+ * amdgpu_gmc_sign_extend - sign extend the given gmc address
+ *
+ * @addr: address to extend
+ */
+static inline uint64_t amdgpu_gmc_sign_extend(uint64_t addr)
+{
+	if (addr >= AMDGPU_GMC_HOLE_START)
+		addr |= AMDGPU_GMC_HOLE_END;
+
+	return addr;
+}
+
 void amdgpu_gmc_get_pde_for_bo(struct amdgpu_bo *bo, int level,
 			       uint64_t *addr, uint64_t *flags);
 uint64_t amdgpu_gmc_pd_addr(struct amdgpu_bo *bo);

drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c

@@ -655,11 +655,11 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 
 		dev_info.virtual_address_offset = AMDGPU_VA_RESERVED_SIZE;
 		dev_info.virtual_address_max =
-			min(vm_size, AMDGPU_VA_HOLE_START);
+			min(vm_size, AMDGPU_GMC_HOLE_START);
 
-		if (vm_size > AMDGPU_VA_HOLE_START) {
-			dev_info.high_va_offset = AMDGPU_VA_HOLE_END;
-			dev_info.high_va_max = AMDGPU_VA_HOLE_END | vm_size;
+		if (vm_size > AMDGPU_GMC_HOLE_START) {
+			dev_info.high_va_offset = AMDGPU_GMC_HOLE_END;
+			dev_info.high_va_max = AMDGPU_GMC_HOLE_END | vm_size;
 		}
 		dev_info.virtual_address_alignment = max((int)PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
 		dev_info.pte_fragment_size = (1 << adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;

drivers/gpu/drm/amd/amdgpu/amdgpu_object.c

@@ -1368,7 +1368,7 @@ u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo)
 	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_VRAM &&
 		     !(bo->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS));
 
-	return bo->tbo.offset;
+	return amdgpu_gmc_sign_extend(bo->tbo.offset);
 }
 
 /**

drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c

@@ -28,9 +28,7 @@ uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev)
 	uint64_t addr = adev->vm_manager.max_pfn << AMDGPU_GPU_PAGE_SHIFT;
 
 	addr -= AMDGPU_VA_RESERVED_SIZE;
-
-	if (addr >= AMDGPU_VA_HOLE_START)
-		addr |= AMDGPU_VA_HOLE_END;
+	addr = amdgpu_gmc_sign_extend(addr);
 
 	return addr;
 }
@@ -73,7 +71,7 @@ void amdgpu_free_static_csa(struct amdgpu_device *adev) {
 int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			  struct amdgpu_bo_va **bo_va)
 {
-	uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_VA_HOLE_MASK;
+	uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK;
 	struct ww_acquire_ctx ticket;
 	struct list_head list;
 	struct amdgpu_bo_list_entry pd;

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c

@@ -492,7 +492,7 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
 		if (level == adev->vm_manager.root_level) {
 			ats_entries = amdgpu_vm_level_shift(adev, level);
 			ats_entries += AMDGPU_GPU_PAGE_SHIFT;
-			ats_entries = AMDGPU_VA_HOLE_START >> ats_entries;
+			ats_entries = AMDGPU_GMC_HOLE_START >> ats_entries;
 			ats_entries = min(ats_entries, entries);
 			entries -= ats_entries;
 		} else {
@@ -722,7 +722,7 @@ int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
 	eaddr = saddr + size - 1;
 
 	if (vm->pte_support_ats)
-		ats = saddr < AMDGPU_VA_HOLE_START;
+		ats = saddr < AMDGPU_GMC_HOLE_START;
 
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
 	eaddr /= AMDGPU_GPU_PAGE_SIZE;
@@ -2016,7 +2016,8 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 			struct amdgpu_bo_va_mapping, list);
 		list_del(&mapping->list);
 
-		if (vm->pte_support_ats && mapping->start < AMDGPU_VA_HOLE_START)
+		if (vm->pte_support_ats &&
+		    mapping->start < AMDGPU_GMC_HOLE_START)
 			init_pte_value = AMDGPU_PTE_DEFAULT_ATC;
 
 		r = amdgpu_vm_bo_update_mapping(adev, NULL, NULL, vm,

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h

@@ -101,19 +101,6 @@ struct amdgpu_bo_list_entry;
 /* hardcode that limit for now */
 #define AMDGPU_VA_RESERVED_SIZE			(1ULL << 20)
 
-/* VA hole for 48bit addresses on Vega10 */
-#define AMDGPU_VA_HOLE_START			0x0000800000000000ULL
-#define AMDGPU_VA_HOLE_END			0xffff800000000000ULL
-
-/*
- * Hardware is programmed as if the hole doesn't exists with start and end
- * address values.
- *
- * This mask is used to remove the upper 16bits of the VA and so come up with
- * the linear addr value.
- */
-#define AMDGPU_VA_HOLE_MASK			0x0000ffffffffffffULL
-
 /* max vmids dedicated for process */
 #define AMDGPU_VM_MAX_RESERVED_VMID	1