drm/nouveau/bo: convert move move() to new push macros

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>

8 files changed, 114 insertions, 131 deletions

diff --git a/drivers/gpu/drm/nouveau/nouveau_bo0039.c b/drivers/gpu/drm/nouveau/nouveau_bo0039.c
index ddf2f5ee1140..e00ec7cfce5f 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo0039.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo0039.c
@@ -45,39 +45,36 @@ int
 nv04_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		  struct ttm_mem_reg *old_reg, struct ttm_mem_reg *new_reg)
 {
+	struct nvif_push *push = chan->chan.push;
 	u32 src_offset = old_reg->start << PAGE_SHIFT;
 	u32 dst_offset = new_reg->start << PAGE_SHIFT;
 	u32 page_count = new_reg->num_pages;
 	int ret;
 
-	ret = RING_SPACE(chan, 3);
+	ret = PUSH_WAIT(push, 3);
 	if (ret)
 		return ret;
 
-	BEGIN_NV04(chan, NvSubCopy, NV_MEMORY_TO_MEMORY_FORMAT_DMA_SOURCE, 2);
-	OUT_RING  (chan, nouveau_bo_mem_ctxdma(bo, chan, old_reg));
-	OUT_RING  (chan, nouveau_bo_mem_ctxdma(bo, chan, new_reg));
+	PUSH_NVSQ(push, NV039, 0x0184, nouveau_bo_mem_ctxdma(bo, chan, old_reg),
+			       0x0188, nouveau_bo_mem_ctxdma(bo, chan, new_reg));
 
 	page_count = new_reg->num_pages;
 	while (page_count) {
 		int line_count = (page_count > 2047) ? 2047 : page_count;
 
-		ret = RING_SPACE(chan, 11);
+		ret = PUSH_WAIT(push, 11);
 		if (ret)
 			return ret;
 
-		BEGIN_NV04(chan, NvSubCopy,
-				 NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
-		OUT_RING  (chan, src_offset);
-		OUT_RING  (chan, dst_offset);
-		OUT_RING  (chan, PAGE_SIZE); /* src_pitch */
-		OUT_RING  (chan, PAGE_SIZE); /* dst_pitch */
-		OUT_RING  (chan, PAGE_SIZE); /* line_length */
-		OUT_RING  (chan, line_count);
-		OUT_RING  (chan, 0x00000101);
-		OUT_RING  (chan, 0x00000000);
-		BEGIN_NV04(chan, NvSubCopy, NV_MEMORY_TO_MEMORY_FORMAT_NOP, 1);
-		OUT_RING  (chan, 0);
+		PUSH_NVSQ(push, NV039, 0x030c, src_offset,
+				       0x0310, dst_offset,
+				       0x0314, PAGE_SIZE, /* src_pitch */
+				       0x0318, PAGE_SIZE, /* dst_pitch */
+				       0x031c, PAGE_SIZE, /* line_length */
+				       0x0320, line_count,
+				       0x0324, 0x00000101,
+				       0x0328, 0x00000000);
+		PUSH_NVSQ(push, NV039, 0x0100, 0x00000000);
 
 		page_count -= line_count;
 		src_offset += (PAGE_SIZE * line_count);
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo5039.c b/drivers/gpu/drm/nouveau/nouveau_bo5039.c
index 9a667dc93dd3..19fb36b35ff9 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo5039.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo5039.c
@@ -38,6 +38,7 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		  struct ttm_mem_reg *old_reg, struct ttm_mem_reg *new_reg)
 {
 	struct nouveau_mem *mem = nouveau_mem(old_reg);
+	struct nvif_push *push = chan->chan.push;
 	u64 length = (new_reg->num_pages << PAGE_SHIFT);
 	u64 src_offset = mem->vma[0].addr;
 	u64 dst_offset = mem->vma[1].addr;
@@ -48,7 +49,7 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 	while (length) {
 		u32 amount, stride, height;
 
-		ret = RING_SPACE(chan, 18 + 6 * (src_tiled + dst_tiled));
+		ret = PUSH_WAIT(push, 18 + 6 * (src_tiled + dst_tiled));
 		if (ret)
 			return ret;
 
@@ -57,46 +58,40 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		height  = amount / stride;
 
 		if (src_tiled) {
-			BEGIN_NV04(chan, NvSubCopy, 0x0200, 7);
-			OUT_RING  (chan, 0);
-			OUT_RING  (chan, 0);
-			OUT_RING  (chan, stride);
-			OUT_RING  (chan, height);
-			OUT_RING  (chan, 1);
-			OUT_RING  (chan, 0);
-			OUT_RING  (chan, 0);
+			PUSH_NVSQ(push, NV5039, 0x0200, 0,
+						0x0204, 0,
+						0x0208, stride,
+						0x020c, height,
+						0x0210, 1,
+						0x0214, 0,
+						0x0218, 0);
 		} else {
-			BEGIN_NV04(chan, NvSubCopy, 0x0200, 1);
-			OUT_RING  (chan, 1);
+			PUSH_NVSQ(push, NV5039, 0x0200, 1);
 		}
+
 		if (dst_tiled) {
-			BEGIN_NV04(chan, NvSubCopy, 0x021c, 7);
-			OUT_RING  (chan, 0);
-			OUT_RING  (chan, 0);
-			OUT_RING  (chan, stride);
-			OUT_RING  (chan, height);
-			OUT_RING  (chan, 1);
-			OUT_RING  (chan, 0);
-			OUT_RING  (chan, 0);
+			PUSH_NVSQ(push, NV5039, 0x021c, 0,
+						0x0220, 0,
+						0x0224, stride,
+						0x0228, height,
+						0x022c, 1,
+						0x0230, 0,
+						0x0234, 0);
 		} else {
-			BEGIN_NV04(chan, NvSubCopy, 0x021c, 1);
-			OUT_RING  (chan, 1);
+			PUSH_NVSQ(push, NV5039, 0x021c, 1);
 		}
 
-		BEGIN_NV04(chan, NvSubCopy, 0x0238, 2);
-		OUT_RING  (chan, upper_32_bits(src_offset));
-		OUT_RING  (chan, upper_32_bits(dst_offset));
-		BEGIN_NV04(chan, NvSubCopy, 0x030c, 8);
-		OUT_RING  (chan, lower_32_bits(src_offset));
-		OUT_RING  (chan, lower_32_bits(dst_offset));
-		OUT_RING  (chan, stride);
-		OUT_RING  (chan, stride);
-		OUT_RING  (chan, stride);
-		OUT_RING  (chan, height);
-		OUT_RING  (chan, 0x00000101);
-		OUT_RING  (chan, 0x00000000);
-		BEGIN_NV04(chan, NvSubCopy, NV_MEMORY_TO_MEMORY_FORMAT_NOP, 1);
-		OUT_RING  (chan, 0);
+		PUSH_NVSQ(push, NV5039, 0x0238, upper_32_bits(src_offset),
+					0x023c, upper_32_bits(dst_offset));
+		PUSH_NVSQ(push, NV5039, 0x030c, lower_32_bits(src_offset),
+					0x0310, lower_32_bits(dst_offset),
+					0x0314, stride,
+					0x0318, stride,
+					0x031c, stride,
+					0x0320, height,
+					0x0324, 0x00000101,
+					0x0328, 0x00000000);
+		PUSH_NVSQ(push, NV5039, 0x0100, 0x00000000);
 
 		length -= amount;
 		src_offset += amount;
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo74c1.c b/drivers/gpu/drm/nouveau/nouveau_bo74c1.c
index 7528a03229b2..1b5fd78ddcba 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo74c1.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo74c1.c
@@ -30,20 +30,25 @@
 #include "nouveau_dma.h"
 #include "nouveau_mem.h"
 
+#include <nvif/push206e.h>
+
 int
 nv84_bo_move_exec(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		  struct ttm_mem_reg *old_reg, struct ttm_mem_reg *new_reg)
 {
 	struct nouveau_mem *mem = nouveau_mem(old_reg);
-	int ret = RING_SPACE(chan, 7);
-	if (ret == 0) {
-		BEGIN_NV04(chan, NvSubCopy, 0x0304, 6);
-		OUT_RING  (chan, new_reg->num_pages << PAGE_SHIFT);
-		OUT_RING  (chan, upper_32_bits(mem->vma[0].addr));
-		OUT_RING  (chan, lower_32_bits(mem->vma[0].addr));
-		OUT_RING  (chan, upper_32_bits(mem->vma[1].addr));
-		OUT_RING  (chan, lower_32_bits(mem->vma[1].addr));
-		OUT_RING  (chan, 0x00000000 /* MODE_COPY, QUERY_NONE */);
-	}
-	return ret;
+	struct nvif_push *push = chan->chan.push;
+	int ret;
+
+	ret = PUSH_WAIT(push, 7);
+	if (ret)
+		return ret;
+
+	PUSH_NVSQ(push, NV74C1, 0x0304, new_reg->num_pages << PAGE_SHIFT,
+				0x0308, upper_32_bits(mem->vma[0].addr),
+				0x030c, lower_32_bits(mem->vma[0].addr),
+				0x0310, upper_32_bits(mem->vma[1].addr),
+				0x0314, lower_32_bits(mem->vma[1].addr),
+				0x0318, 0x00000000 /* MODE_COPY, QUERY_NONE */);
+	return 0;
 }
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo85b5.c b/drivers/gpu/drm/nouveau/nouveau_bo85b5.c
index c658c5e5fe04..f0df172b029e 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo85b5.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo85b5.c
@@ -30,6 +30,8 @@
 #include "nouveau_dma.h"
 #include "nouveau_mem.h"
 
+#include <nvif/push206e.h>
+
 /*XXX: Fixup class to be compatible with NVIDIA's, which will allow sharing
  *     code with KeplerDmaCopyA.
  */
@@ -39,6 +41,7 @@ nva3_bo_move_copy(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		  struct ttm_mem_reg *old_reg, struct ttm_mem_reg *new_reg)
 {
 	struct nouveau_mem *mem = nouveau_mem(old_reg);
+	struct nvif_push *push = chan->chan.push;
 	u64 src_offset = mem->vma[0].addr;
 	u64 dst_offset = mem->vma[1].addr;
 	u32 page_count = new_reg->num_pages;
@@ -48,21 +51,19 @@ nva3_bo_move_copy(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 	while (page_count) {
 		int line_count = (page_count > 8191) ? 8191 : page_count;
 
-		ret = RING_SPACE(chan, 11);
+		ret = PUSH_WAIT(push, 11);
 		if (ret)
 			return ret;
 
-		BEGIN_NV04(chan, NvSubCopy, 0x030c, 8);
-		OUT_RING  (chan, upper_32_bits(src_offset));
-		OUT_RING  (chan, lower_32_bits(src_offset));
-		OUT_RING  (chan, upper_32_bits(dst_offset));
-		OUT_RING  (chan, lower_32_bits(dst_offset));
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, line_count);
-		BEGIN_NV04(chan, NvSubCopy, 0x0300, 1);
-		OUT_RING  (chan, 0x00000110);
+		PUSH_NVSQ(push, NV85B5, 0x030c, upper_32_bits(src_offset),
+					0x0310, lower_32_bits(src_offset),
+					0x0314, upper_32_bits(dst_offset),
+					0x0318, lower_32_bits(dst_offset),
+					0x031c, PAGE_SIZE,
+					0x0320, PAGE_SIZE,
+					0x0324, PAGE_SIZE,
+					0x0328, line_count);
+		PUSH_NVSQ(push, NV85B5, 0x0300, 0x00000110);
 
 		page_count -= line_count;
 		src_offset += (PAGE_SIZE * line_count);
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo9039.c b/drivers/gpu/drm/nouveau/nouveau_bo9039.c
index f9ba04faf1a2..995ebe7ffe00 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo9039.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo9039.c
@@ -36,6 +36,7 @@ int
 nvc0_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		  struct ttm_mem_reg *old_reg, struct ttm_mem_reg *new_reg)
 {
+	struct nvif_push *push = chan->chan.push;
 	struct nouveau_mem *mem = nouveau_mem(old_reg);
 	u64 src_offset = mem->vma[0].addr;
 	u64 dst_offset = mem->vma[1].addr;
@@ -46,22 +47,19 @@ nvc0_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 	while (page_count) {
 		int line_count = (page_count > 2047) ? 2047 : page_count;
 
-		ret = RING_SPACE(chan, 12);
+		ret = PUSH_WAIT(push, 12);
 		if (ret)
 			return ret;
 
-		BEGIN_NVC0(chan, NvSubCopy, 0x0238, 2);
-		OUT_RING  (chan, upper_32_bits(dst_offset));
-		OUT_RING  (chan, lower_32_bits(dst_offset));
-		BEGIN_NVC0(chan, NvSubCopy, 0x030c, 6);
-		OUT_RING  (chan, upper_32_bits(src_offset));
-		OUT_RING  (chan, lower_32_bits(src_offset));
-		OUT_RING  (chan, PAGE_SIZE); /* src_pitch */
-		OUT_RING  (chan, PAGE_SIZE); /* dst_pitch */
-		OUT_RING  (chan, PAGE_SIZE); /* line_length */
-		OUT_RING  (chan, line_count);
-		BEGIN_NVC0(chan, NvSubCopy, 0x0300, 1);
-		OUT_RING  (chan, 0x00100110);
+		PUSH_NVSQ(push, NV9039, 0x0238, upper_32_bits(dst_offset),
+					0x023c, lower_32_bits(dst_offset));
+		PUSH_NVSQ(push, NV9039, 0x030c, upper_32_bits(src_offset),
+					0x0310, lower_32_bits(src_offset),
+					0x0314, PAGE_SIZE, /* src_pitch */
+					0x0318, PAGE_SIZE, /* dst_pitch */
+					0x031c, PAGE_SIZE, /* line_length */
+					0x0320, line_count);
+		PUSH_NVSQ(push, NV9039, 0x0300, 0x00100110);
 
 		page_count -= line_count;
 		src_offset += (PAGE_SIZE * line_count);
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo90b5.c b/drivers/gpu/drm/nouveau/nouveau_bo90b5.c
index 4c014f9a641f..34b79d561c7f 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo90b5.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo90b5.c
@@ -23,6 +23,8 @@
 #include "nouveau_dma.h"
 #include "nouveau_mem.h"
 
+#include <nvif/push906f.h>
+
 /*XXX: Fixup class to be compatible with NVIDIA's, which will allow sharing
  *     code with KeplerDmaCopyA.
  */
@@ -32,6 +34,7 @@ nvc0_bo_move_copy(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		  struct ttm_mem_reg *old_reg, struct ttm_mem_reg *new_reg)
 {
 	struct nouveau_mem *mem = nouveau_mem(old_reg);
+	struct nvif_push *push = chan->chan.push;
 	u64 src_offset = mem->vma[0].addr;
 	u64 dst_offset = mem->vma[1].addr;
 	u32 page_count = new_reg->num_pages;
@@ -41,21 +44,19 @@ nvc0_bo_move_copy(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 	while (page_count) {
 		int line_count = (page_count > 8191) ? 8191 : page_count;
 
-		ret = RING_SPACE(chan, 11);
+		ret = PUSH_WAIT(push, 10);
 		if (ret)
 			return ret;
 
-		BEGIN_NVC0(chan, NvSubCopy, 0x030c, 8);
-		OUT_RING  (chan, upper_32_bits(src_offset));
-		OUT_RING  (chan, lower_32_bits(src_offset));
-		OUT_RING  (chan, upper_32_bits(dst_offset));
-		OUT_RING  (chan, lower_32_bits(dst_offset));
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, line_count);
-		BEGIN_NVC0(chan, NvSubCopy, 0x0300, 1);
-		OUT_RING  (chan, 0x00000110);
+		PUSH_NVSQ(push, NV90B5, 0x030c, upper_32_bits(src_offset),
+					0x0310, lower_32_bits(src_offset),
+					0x0314, upper_32_bits(dst_offset),
+					0x0318, lower_32_bits(dst_offset),
+					0x031c, PAGE_SIZE,
+					0x0320, PAGE_SIZE,
+					0x0324, PAGE_SIZE,
+					0x0328, line_count);
+		PUSH_NVIM(push, NV90B5, 0x0300, 0x0110);
 
 		page_count -= line_count;
 		src_offset += (PAGE_SIZE * line_count);
diff --git a/drivers/gpu/drm/nouveau/nouveau_boa0b5.c b/drivers/gpu/drm/nouveau/nouveau_boa0b5.c
index 9c09691623d0..b1afb2724fb7 100644
--- a/drivers/gpu/drm/nouveau/nouveau_boa0b5.c
+++ b/drivers/gpu/drm/nouveau/nouveau_boa0b5.c
@@ -37,20 +37,23 @@ nve0_bo_move_copy(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		  struct ttm_mem_reg *old_reg, struct ttm_mem_reg *new_reg)
 {
 	struct nouveau_mem *mem = nouveau_mem(old_reg);
-	int ret = RING_SPACE(chan, 10);
-	if (ret == 0) {
-		BEGIN_NVC0(chan, NvSubCopy, 0x0400, 8);
-		OUT_RING  (chan, upper_32_bits(mem->vma[0].addr));
-		OUT_RING  (chan, lower_32_bits(mem->vma[0].addr));
-		OUT_RING  (chan, upper_32_bits(mem->vma[1].addr));
-		OUT_RING  (chan, lower_32_bits(mem->vma[1].addr));
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, PAGE_SIZE);
-		OUT_RING  (chan, new_reg->num_pages);
-		BEGIN_IMC0(chan, NvSubCopy, 0x0300, 0x0386);
-	}
-	return ret;
+	struct nvif_push *push = chan->chan.push;
+	int ret;
+
+	ret = PUSH_WAIT(push, 10);
+	if (ret)
+		return ret;
+
+	PUSH_NVSQ(push, NVA0B5, 0x0400, upper_32_bits(mem->vma[0].addr),
+				0x0404, lower_32_bits(mem->vma[0].addr),
+				0x0408, upper_32_bits(mem->vma[1].addr),
+				0x040c, lower_32_bits(mem->vma[1].addr),
+				0x0410, PAGE_SIZE,
+				0x0414, PAGE_SIZE,
+				0x0418, PAGE_SIZE,
+				0x041c, new_reg->num_pages);
+	PUSH_NVIM(push, NVA0B5, 0x0300, 0x0386);
+	return 0;
 }
 
 int
diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.h b/drivers/gpu/drm/nouveau/nouveau_dma.h
index 72de20437542..8778fd6002c0 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dma.h
+++ b/drivers/gpu/drm/nouveau/nouveau_dma.h
@@ -63,23 +63,6 @@ enum {
 	NvEvoSema1	= 0x80000011,
 };
 
-#define NV_MEMORY_TO_MEMORY_FORMAT                                    0x00000039
-#define NV_MEMORY_TO_MEMORY_FORMAT_NAME                               0x00000000
-#define NV_MEMORY_TO_MEMORY_FORMAT_SET_REF                            0x00000050
-#define NV_MEMORY_TO_MEMORY_FORMAT_NOP                                0x00000100
-#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY                             0x00000104
-#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY_STYLE_WRITE                 0x00000000
-#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY_STYLE_WRITE_LE_AWAKEN       0x00000001
-#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY                         0x00000180
-#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_SOURCE                         0x00000184
-#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN                          0x0000030c
-
-#define NV50_MEMORY_TO_MEMORY_FORMAT                                  0x00005039
-#define NV50_MEMORY_TO_MEMORY_FORMAT_UNK200                           0x00000200
-#define NV50_MEMORY_TO_MEMORY_FORMAT_UNK21C                           0x0000021c
-#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN_HIGH                   0x00000238
-#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT_HIGH                  0x0000023c
-
 static __must_check inline int
 RING_SPACE(struct nouveau_channel *chan, int size)
 {