drm/amdgpu: use amdgpu_device_vram_access in amdgpu_ttm_access_memory v2

Make use of the better performance here as well.

This patch is only compile tested!

v2: fix calculation bug pointed out by Felix
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Jonathan Kim <Jonathan.Kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -1598,7 +1598,7 @@ static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo,
 
 	while (len && pos < adev->gmc.mc_vram_size) {
 		uint64_t aligned_pos = pos & ~(uint64_t)3;
-		uint32_t bytes = 4 - (pos & 3);
+		uint64_t bytes = 4 - (pos & 3);
 		uint32_t shift = (pos & 3) * 8;
 		uint32_t mask = 0xffffffff << shift;
 
@@ -1607,6 +1607,7 @@ static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo,
 			bytes = len;
 		}
 
+		if (mask != 0xffffffff) {
 			spin_lock_irqsave(&adev->mmio_idx_lock, flags);
 			WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t)aligned_pos) | 0x80000000);
 			WREG32_NO_KIQ(mmMM_INDEX_HI, aligned_pos >> 31);
@@ -1622,6 +1623,13 @@ static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo,
 				value = (value & mask) >> shift;
 				memcpy(buf, &value, bytes);
 			}
+		} else {
+			bytes = (nodes->start + nodes->size) << PAGE_SHIFT;
+			bytes = min(bytes - pos, (uint64_t)len & ~0x3ull);
+
+			amdgpu_device_vram_access(adev, pos, (uint32_t *)buf,
+						  bytes, write);
+		}
 
 		ret += bytes;
 		buf = (uint8_t *)buf + bytes;