drm/radeon: implement blit copy callback for CIK

Uses the CP ring rather than the DMA ring.  Useful
for debugging and benchmarking.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/radeon/cik.c

@@ -3094,6 +3094,85 @@ void cik_semaphore_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, (upper_32_bits(addr) & 0xffff) | sel);
 }
 
+/**
+ * cik_copy_cpdma - copy pages using the CP DMA engine
+ *
+ * @rdev: radeon_device pointer
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @num_gpu_pages: number of GPU pages to xfer
+ * @fence: radeon fence object
+ *
+ * Copy GPU paging using the CP DMA engine (CIK+).
+ * Used by the radeon ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+int cik_copy_cpdma(struct radeon_device *rdev,
+		   uint64_t src_offset, uint64_t dst_offset,
+		   unsigned num_gpu_pages,
+		   struct radeon_fence **fence)
+{
+	struct radeon_semaphore *sem = NULL;
+	int ring_index = rdev->asic->copy.blit_ring_index;
+	struct radeon_ring *ring = &rdev->ring[ring_index];
+	u32 size_in_bytes, cur_size_in_bytes, control;
+	int i, num_loops;
+	int r = 0;
+
+	r = radeon_semaphore_create(rdev, &sem);
+	if (r) {
+		DRM_ERROR("radeon: moving bo (%d).\n", r);
+		return r;
+	}
+
+	size_in_bytes = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT);
+	num_loops = DIV_ROUND_UP(size_in_bytes, 0x1fffff);
+	r = radeon_ring_lock(rdev, ring, num_loops * 7 + 18);
+	if (r) {
+		DRM_ERROR("radeon: moving bo (%d).\n", r);
+		radeon_semaphore_free(rdev, &sem, NULL);
+		return r;
+	}
+
+	if (radeon_fence_need_sync(*fence, ring->idx)) {
+		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
+					    ring->idx);
+		radeon_fence_note_sync(*fence, ring->idx);
+	} else {
+		radeon_semaphore_free(rdev, &sem, NULL);
+	}
+
+	for (i = 0; i < num_loops; i++) {
+		cur_size_in_bytes = size_in_bytes;
+		if (cur_size_in_bytes > 0x1fffff)
+			cur_size_in_bytes = 0x1fffff;
+		size_in_bytes -= cur_size_in_bytes;
+		control = 0;
+		if (size_in_bytes == 0)
+			control |= PACKET3_DMA_DATA_CP_SYNC;
+		radeon_ring_write(ring, PACKET3(PACKET3_DMA_DATA, 5));
+		radeon_ring_write(ring, control);
+		radeon_ring_write(ring, lower_32_bits(src_offset));
+		radeon_ring_write(ring, upper_32_bits(src_offset));
+		radeon_ring_write(ring, lower_32_bits(dst_offset));
+		radeon_ring_write(ring, upper_32_bits(dst_offset));
+		radeon_ring_write(ring, cur_size_in_bytes);
+		src_offset += cur_size_in_bytes;
+		dst_offset += cur_size_in_bytes;
+	}
+
+	r = radeon_fence_emit(rdev, fence, ring->idx);
+	if (r) {
+		radeon_ring_unlock_undo(rdev, ring);
+		return r;
+	}
+
+	radeon_ring_unlock_commit(rdev, ring);
+	radeon_semaphore_free(rdev, &sem, *fence);
+
+	return r;
+}
+
 /*
  * IB stuff
  */

drivers/gpu/drm/radeon/cikd.h

@@ -1747,6 +1747,68 @@
 #              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
 #              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
 #define	PACKET3_DMA_DATA				0x50
+/* 1. header
+ * 2. CONTROL
+ * 3. SRC_ADDR_LO or DATA [31:0]
+ * 4. SRC_ADDR_HI [31:0]
+ * 5. DST_ADDR_LO [31:0]
+ * 6. DST_ADDR_HI [7:0]
+ * 7. COMMAND [30:21] | BYTE_COUNT [20:0]
+ */
+/* CONTROL */
+#              define PACKET3_DMA_DATA_ENGINE(x)     ((x) << 0)
+                /* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
+                /* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_SRC_VOLATILE (1 << 15)
+#              define PACKET3_DMA_DATA_DST_SEL(x)  ((x) << 20)
+                /* 0 - DST_ADDR using DAS
+		 * 1 - GDS
+		 * 3 - DST_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
+                /* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_DST_VOLATILE (1 << 27)
+#              define PACKET3_DMA_DATA_SRC_SEL(x)  ((x) << 29)
+                /* 0 - SRC_ADDR using SAS
+		 * 1 - GDS
+		 * 2 - DATA
+		 * 3 - SRC_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_CP_SYNC     (1 << 31)
+/* COMMAND */
+#              define PACKET3_DMA_DATA_DIS_WC      (1 << 21)
+#              define PACKET3_DMA_DATA_CMD_SRC_SWAP(x) ((x) << 22)
+                /* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_DST_SWAP(x) ((x) << 24)
+                /* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAS     (1 << 26)
+                /* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_DAS     (1 << 27)
+                /* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
+#              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
+#              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
 #define	PACKET3_AQUIRE_MEM				0x58
 #define	PACKET3_REWIND					0x59
 #define	PACKET3_LOAD_UCONFIG_REG			0x5E

drivers/gpu/drm/radeon/radeon_asic.h

@@ -705,6 +705,10 @@ int cik_copy_dma(struct radeon_device *rdev,
 		 uint64_t src_offset, uint64_t dst_offset,
 		 unsigned num_gpu_pages,
 		 struct radeon_fence **fence);
+int cik_copy_cpdma(struct radeon_device *rdev,
+		   uint64_t src_offset, uint64_t dst_offset,
+		   unsigned num_gpu_pages,
+		   struct radeon_fence **fence);
 int cik_sdma_ring_test(struct radeon_device *rdev, struct radeon_ring *ring);
 int cik_sdma_ib_test(struct radeon_device *rdev, struct radeon_ring *ring);
 bool cik_sdma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring);