This code is a slightly modified version of evergreen_dma_blit (and
evergreen_dma_copy as well as evergreen_dma_copy_tile).
It would be nice to share some of the code in the long term.
I have reused some "cik"-prefixed functions that also return the right
value for SI. I am not sure if they should be renamed.
Signed-off-by: Niels Ole Salscheider <niels_...@salscheider-online.de>
---
src/gallium/drivers/radeonsi/Makefile.sources | 1 +
src/gallium/drivers/radeonsi/si_dma.c | 352 ++++++++++++++++++++++++++
src/gallium/drivers/radeonsi/si_pipe.h | 9 +
src/gallium/drivers/radeonsi/si_state.c | 25 +-
src/gallium/drivers/radeonsi/si_state.h | 7 +
src/gallium/drivers/radeonsi/sid.h | 20 ++
6 files changed, 394 insertions(+), 20 deletions(-)
create mode 100644 src/gallium/drivers/radeonsi/si_dma.c
diff --git a/src/gallium/drivers/radeonsi/Makefile.sources
b/src/gallium/drivers/radeonsi/Makefile.sources
index 11b3319..6a24cde 100644
--- a/src/gallium/drivers/radeonsi/Makefile.sources
+++ b/src/gallium/drivers/radeonsi/Makefile.sources
@@ -3,6 +3,7 @@ C_SOURCES := \
si_commands.c \
si_compute.c \
si_descriptors.c \
+ si_dma.c \
si_hw_context.c \
si_pipe.c \
si_pm4.c \
diff --git a/src/gallium/drivers/radeonsi/si_dma.c
b/src/gallium/drivers/radeonsi/si_dma.c
new file mode 100644
index 0000000..61078eb
--- /dev/null
+++ b/src/gallium/drivers/radeonsi/si_dma.c
@@ -0,0 +1,352 @@
+/*
+ * Copyright 2010 Jerome Glisse <gli...@freedesktop.org>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ * Jerome Glisse
+ */
+
+#include "sid.h"
+#include "si_pipe.h"
+#include "../radeon/r600_cs.h"
+
+#include "util/u_format.h"
+
+static unsigned si_array_mode(unsigned mode)
+{
+ switch (mode) {
+ case RADEON_SURF_MODE_LINEAR_ALIGNED:
+ return V_009910_ARRAY_LINEAR_ALIGNED;
+ case RADEON_SURF_MODE_1D:
+ return V_009910_ARRAY_1D_TILED_THIN1;
+ case RADEON_SURF_MODE_2D:
+ return V_009910_ARRAY_2D_TILED_THIN1;
+ default:
+ case RADEON_SURF_MODE_LINEAR:
+ return V_009910_ARRAY_LINEAR_GENERAL;
+ }
+}
+
+static uint32_t si_num_banks(uint32_t nbanks)
+{
+ switch (nbanks) {
+ case 2:
+ return V_009910_ADDR_SURF_2_BANK;
+ case 4:
+ return V_009910_ADDR_SURF_4_BANK;
+ case 8:
+ default:
+ return V_009910_ADDR_SURF_8_BANK;
+ case 16:
+ return V_009910_ADDR_SURF_16_BANK;
+ }
+}
+
+static uint32_t si_micro_tile_mode(struct si_screen *sscreen, unsigned
tile_mode)
+{
+ if (sscreen->b.info.si_tile_mode_array_valid) {
+ uint32_t gb_tile_mode =
sscreen->b.info.si_tile_mode_array[tile_mode];
+
+ return G_009910_MICRO_TILE_MODE(gb_tile_mode);
+ }
+
+ /* The kernel cannod return the tile mode array. Guess? */
+ return V_009910_ADDR_SURF_THIN_MICRO_TILING;
+}
+
+static void si_dma_copy_buffer(struct si_context *ctx,
+ struct pipe_resource *dst,
+ struct pipe_resource *src,
+ uint64_t dst_offset,
+ uint64_t src_offset,
+ uint64_t size)
+{
+ struct radeon_winsys_cs *cs = ctx->b.rings.dma.cs;
+ unsigned i, ncopy, csize, max_csize, sub_cmd, shift;
+ struct r600_resource *rdst = (struct r600_resource*)dst;
+ struct r600_resource *rsrc = (struct r600_resource*)src;
+
+ /* Mark the buffer range of destination as valid (initialized),
+ * so that transfer_map knows it should wait for the GPU when mapping
+ * that range. */
+ util_range_add(&rdst->valid_buffer_range, dst_offset,
+ dst_offset + size);
+
+ dst_offset += r600_resource_va(&ctx->screen->b.b, dst);
+ src_offset += r600_resource_va(&ctx->screen->b.b, src);
+
+ /* see if we use dword or byte copy */
+ if (!(dst_offset % 4) && !(src_offset % 4) && !(size % 4)) {
+ size >>= 2;
+ sub_cmd = SI_DMA_COPY_DWORD_ALIGNED;
+ shift = 2;
+ max_csize = SI_DMA_COPY_MAX_SIZE_DW;
+ } else {
+ sub_cmd = SI_DMA_COPY_BYTE_ALIGNED;
+ shift = 0;
+ max_csize = SI_DMA_COPY_MAX_SIZE;
+ }
+ ncopy = (size / max_csize) + !!(size % max_csize);
+
+ r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, rsrc,
RADEON_USAGE_READ,
+ RADEON_PRIO_MIN);
+ r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, rdst,
RADEON_USAGE_WRITE,
+ RADEON_PRIO_MIN);
+
+ r600_need_dma_space(&ctx->b, ncopy * 5);
+ for (i = 0; i < ncopy; i++) {
+ csize = size < max_csize ? size : max_csize;
+ cs->buf[cs->cdw++] = SI_DMA_PACKET(SI_DMA_PACKET_COPY, sub_cmd,
csize);
+ cs->buf[cs->cdw++] = dst_offset & 0xffffffff;
+ cs->buf[cs->cdw++] = src_offset & 0xffffffff;
+ cs->buf[cs->cdw++] = (dst_offset >> 32UL) & 0xff;
+ cs->buf[cs->cdw++] = (src_offset >> 32UL) & 0xff;
+ dst_offset += csize << shift;
+ src_offset += csize << shift;
+ size -= csize;
+ }
+}
+
+static void si_dma_copy_tile(struct si_context *ctx,
+ struct pipe_resource *dst,
+ unsigned dst_level,
+ unsigned dst_x,
+ unsigned dst_y,
+ unsigned dst_z,
+ struct pipe_resource *src,
+ unsigned src_level,
+ unsigned src_x,
+ unsigned src_y,
+ unsigned src_z,
+ unsigned copy_height,
+ unsigned pitch,
+ unsigned bpp)
+{
+ struct radeon_winsys_cs *cs = ctx->b.rings.dma.cs;
+ struct si_screen *sscreen = ctx->screen;
+ struct r600_texture *rsrc = (struct r600_texture*)src;
+ struct r600_texture *rdst = (struct r600_texture*)dst;
+ unsigned array_mode, lbpp, pitch_tile_max, slice_tile_max, size;
+ unsigned ncopy, height, cheight, detile, i, x, y, z, src_mode, dst_mode;
+ unsigned sub_cmd, bank_h, bank_w, mt_aspect, nbanks, tile_split, mt;
+ uint64_t base, addr;
+ unsigned pipe_config, tile_mode_index;
+
+ /* make sure that the dma ring is only one active */
+ ctx->b.rings.gfx.flush(ctx, RADEON_FLUSH_ASYNC);
+
+ dst_mode = rdst->surface.level[dst_level].mode;
+ src_mode = rsrc->surface.level[src_level].mode;
+ /* downcast linear aligned to linear to simplify test */
+ src_mode = src_mode == RADEON_SURF_MODE_LINEAR_ALIGNED ?
RADEON_SURF_MODE_LINEAR : src_mode;
+ dst_mode = dst_mode == RADEON_SURF_MODE_LINEAR_ALIGNED ?
RADEON_SURF_MODE_LINEAR : dst_mode;
+ assert(dst_mode != src_mode);
+
+ y = 0;
+ sub_cmd = SI_DMA_COPY_TILED;
+ lbpp = util_logbase2(bpp);
+ pitch_tile_max = ((pitch / bpp) / 8) - 1;
+ nbanks = si_num_banks(ctx->screen->b.tiling_info.num_banks);
+
+ if (dst_mode == RADEON_SURF_MODE_LINEAR) {
+ /* T2L */
+ array_mode = si_array_mode(src_mode);
+ slice_tile_max = (rsrc->surface.level[src_level].nblk_x *
rsrc->surface.level[src_level].nblk_y) / (8*8);
+ slice_tile_max = slice_tile_max ? slice_tile_max - 1 : 0;
+ /* linear height must be the same as the slice tile max height,
it's ok even
+ * if the linear destination/source have smaller heigh as the
size of the
+ * dma packet will be using the copy_height which is always
smaller or equal
+ * to the linear height
+ */
+ height = rsrc->surface.level[src_level].npix_y;
+ detile = 1;
+ x = src_x;
+ y = src_y;
+ z = src_z;
+ base = rsrc->surface.level[src_level].offset;
+ addr = rdst->surface.level[dst_level].offset;
+ addr += rdst->surface.level[dst_level].slice_size * dst_z;
+ addr += dst_y * pitch + dst_x * bpp;
+ bank_h = cik_bank_wh(rsrc->surface.bankh);
+ bank_w = cik_bank_wh(rsrc->surface.bankw);
+ mt_aspect = cik_macro_tile_aspect(rsrc->surface.mtilea);
+ tile_split = cik_tile_split(rsrc->surface.tile_split);
+ tile_mode_index = si_tile_mode_index(rsrc, src_level,
+
util_format_has_stencil(util_format_description(src->format)));
+ base += r600_resource_va(&ctx->screen->b.b, src);
+ addr += r600_resource_va(&ctx->screen->b.b, dst);
+ } else {
+ /* L2T */
+ array_mode = si_array_mode(dst_mode);
+ slice_tile_max = (rdst->surface.level[dst_level].nblk_x *
rdst->surface.level[dst_level].nblk_y) / (8*8);
+ slice_tile_max = slice_tile_max ? slice_tile_max - 1 : 0;
+ /* linear height must be the same as the slice tile max height,
it's ok even
+ * if the linear destination/source have smaller heigh as the
size of the
+ * dma packet will be using the copy_height which is always
smaller or equal
+ * to the linear height
+ */
+ height = rdst->surface.level[dst_level].npix_y;
+ detile = 0;
+ x = dst_x;
+ y = dst_y;
+ z = dst_z;
+ base = rdst->surface.level[dst_level].offset;
+ addr = rsrc->surface.level[src_level].offset;
+ addr += rsrc->surface.level[src_level].slice_size * src_z;
+ addr += src_y * pitch + src_x * bpp;
+ bank_h = cik_bank_wh(rdst->surface.bankh);
+ bank_w = cik_bank_wh(rdst->surface.bankw);
+ mt_aspect = cik_macro_tile_aspect(rdst->surface.mtilea);
+ tile_split = cik_tile_split(rdst->surface.tile_split);
+ tile_mode_index = si_tile_mode_index(rdst, dst_level,
+
util_format_has_stencil(util_format_description(dst->format)));
+ base += r600_resource_va(&ctx->screen->b.b, dst);
+ addr += r600_resource_va(&ctx->screen->b.b, src);
+ }
+
+ pipe_config = cik_db_pipe_config(sscreen, tile_mode_index);
+ mt = si_micro_tile_mode(sscreen, tile_mode_index);
+ size = (copy_height * pitch) / 4;
+ ncopy = (size / SI_DMA_COPY_MAX_SIZE_DW) + !!(size %
SI_DMA_COPY_MAX_SIZE_DW);
+ r600_need_dma_space(&ctx->b, ncopy * 9);
+
+ r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, &rsrc->resource,
+ RADEON_USAGE_READ, RADEON_PRIO_MIN);
+ r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, &rdst->resource,
+ RADEON_USAGE_WRITE, RADEON_PRIO_MIN);
+
+ for (i = 0; i < ncopy; i++) {
+ cheight = copy_height;
+ if (((cheight * pitch) / 4) > SI_DMA_COPY_MAX_SIZE_DW) {
+ cheight = (SI_DMA_COPY_MAX_SIZE_DW * 4) / pitch;
+ }
+ size = (cheight * pitch) / 4;
+ cs->buf[cs->cdw++] = SI_DMA_PACKET(SI_DMA_PACKET_COPY, sub_cmd,
size);
+ cs->buf[cs->cdw++] = base >> 8;
+ cs->buf[cs->cdw++] = (detile << 31) | (array_mode << 27) |
+ (lbpp << 24) | (bank_h << 21) |
+ (bank_w << 18) | (mt_aspect << 16);
+ cs->buf[cs->cdw++] = (pitch_tile_max << 0) | ((height - 1) <<
16);
+ cs->buf[cs->cdw++] = (slice_tile_max << 0) | (pipe_config <<
26);
+ cs->buf[cs->cdw++] = (x << 0) | (z << 18);
+ cs->buf[cs->cdw++] = (y << 0) | (tile_split << 21) | (nbanks <<
25) | (mt << 27);
+ cs->buf[cs->cdw++] = addr & 0xfffffffc;
+ cs->buf[cs->cdw++] = (addr >> 32UL) & 0xff;
+ copy_height -= cheight;
+ addr += cheight * pitch;
+ y += cheight;
+ }
+}
+
+void si_dma_copy(struct pipe_context *ctx,
+ struct pipe_resource *dst,
+ unsigned dst_level,
+ unsigned dstx, unsigned dsty, unsigned dstz,
+ struct pipe_resource *src,
+ unsigned src_level,
+ const struct pipe_box *src_box)
+{
+ struct si_context *sctx = (struct si_context *)ctx;
+ struct r600_texture *rsrc = (struct r600_texture*)src;
+ struct r600_texture *rdst = (struct r600_texture*)dst;
+ unsigned dst_pitch, src_pitch, bpp, dst_mode, src_mode, copy_height;
+ unsigned src_w, dst_w;
+ unsigned src_x, src_y;
+ unsigned dst_x = dstx, dst_y = dsty, dst_z = dstz;
+
+ if (sctx->b.rings.dma.cs == NULL) {
+ goto fallback;
+ }
+
+ /* TODO: Implement DMA copy for CIK */
+ if (sctx->b.chip_class >= CIK) {
+ goto fallback;
+ }
+
+ if (dst->target == PIPE_BUFFER && src->target == PIPE_BUFFER) {
+ si_dma_copy_buffer(sctx, dst, src, dst_x, src_box->x,
src_box->width);
+ return;
+ }
+
+ if (src->format != dst->format || src_box->depth > 1 ||
+ rdst->dirty_level_mask != 0) {
+ goto fallback;
+ }
+
+ if (rsrc->dirty_level_mask) {
+ ctx->flush_resource(ctx, src);
+ }
+
+ src_x = util_format_get_nblocksx(src->format, src_box->x);
+ dst_x = util_format_get_nblocksx(src->format, dst_x);
+ src_y = util_format_get_nblocksy(src->format, src_box->y);
+ dst_y = util_format_get_nblocksy(src->format, dst_y);
+
+ bpp = rdst->surface.bpe;
+ dst_pitch = rdst->surface.level[dst_level].pitch_bytes;
+ src_pitch = rsrc->surface.level[src_level].pitch_bytes;
+ src_w = rsrc->surface.level[src_level].npix_x;
+ dst_w = rdst->surface.level[dst_level].npix_x;
+ copy_height = src_box->height / rsrc->surface.blk_h;
+
+ dst_mode = rdst->surface.level[dst_level].mode;
+ src_mode = rsrc->surface.level[src_level].mode;
+ /* downcast linear aligned to linear to simplify test */
+ src_mode = src_mode == RADEON_SURF_MODE_LINEAR_ALIGNED ?
RADEON_SURF_MODE_LINEAR : src_mode;
+ dst_mode = dst_mode == RADEON_SURF_MODE_LINEAR_ALIGNED ?
RADEON_SURF_MODE_LINEAR : dst_mode;
+
+ if (src_pitch != dst_pitch || src_box->x || dst_x || src_w != dst_w) {
+ /* FIXME si can do partial blit */
+ goto fallback;
+ }
+ /* the x test here are currently useless (because we don't support
partial blit)
+ * but keep them around so we don't forget about those
+ */
+ if ((src_pitch % 8) || (src_box->x % 8) || (dst_x % 8) || (src_box->y %
8) || (dst_y % 8)) {
+ goto fallback;
+ }
+
+ if (src_mode == dst_mode) {
+ uint64_t dst_offset, src_offset;
+ /* simple dma blit would do NOTE code here assume :
+ * src_box.x/y == 0
+ * dst_x/y == 0
+ * dst_pitch == src_pitch
+ */
+ src_offset= rsrc->surface.level[src_level].offset;
+ src_offset += rsrc->surface.level[src_level].slice_size *
src_box->z;
+ src_offset += src_y * src_pitch + src_x * bpp;
+ dst_offset = rdst->surface.level[dst_level].offset;
+ dst_offset += rdst->surface.level[dst_level].slice_size * dst_z;
+ dst_offset += dst_y * dst_pitch + dst_x * bpp;
+ si_dma_copy_buffer(sctx, dst, src, dst_offset, src_offset,
+ src_box->height * src_pitch);
+ } else {
+ si_dma_copy_tile(sctx, dst, dst_level, dst_x, dst_y, dst_z,
+ src, src_level, src_x, src_y, src_box->z,
+ copy_height, dst_pitch, bpp);
+ }
+ return;
+
+fallback:
+ ctx->resource_copy_region(ctx, dst, dst_level, dstx, dsty, dstz,
+ src, src_level, src_box);
+}
diff --git a/src/gallium/drivers/radeonsi/si_pipe.h
b/src/gallium/drivers/radeonsi/si_pipe.h
index 47dc8e7..f1ef2ac 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.h
+++ b/src/gallium/drivers/radeonsi/si_pipe.h
@@ -167,6 +167,15 @@ void si_flush_depth_textures(struct si_context *sctx,
void si_decompress_color_textures(struct si_context *sctx,
struct si_textures_info *textures);
+/* si_dma.c */
+void si_dma_copy(struct pipe_context *ctx,
+ struct pipe_resource *dst,
+ unsigned dst_level,
+ unsigned dstx, unsigned dsty, unsigned dstz,
+ struct pipe_resource *src,
+ unsigned src_level,
+ const struct pipe_box *src_box);
+
/* si_hw_context.c */
void si_context_flush(struct si_context *ctx, unsigned flags);
void si_begin_new_cs(struct si_context *ctx);
diff --git a/src/gallium/drivers/radeonsi/si_state.c
b/src/gallium/drivers/radeonsi/si_state.c
index 3843330..6c345ff 100644
--- a/src/gallium/drivers/radeonsi/si_state.c
+++ b/src/gallium/drivers/radeonsi/si_state.c
@@ -78,7 +78,7 @@ static uint32_t cik_num_banks(struct si_screen *sscreen,
unsigned bpe, unsigned
}
}
-static unsigned cik_tile_split(unsigned tile_split)
+unsigned cik_tile_split(unsigned tile_split)
{
switch (tile_split) {
case 64:
@@ -107,7 +107,7 @@ static unsigned cik_tile_split(unsigned tile_split)
return tile_split;
}
-static unsigned cik_macro_tile_aspect(unsigned macro_tile_aspect)
+unsigned cik_macro_tile_aspect(unsigned macro_tile_aspect)
{
switch (macro_tile_aspect) {
default:
@@ -127,7 +127,7 @@ static unsigned cik_macro_tile_aspect(unsigned
macro_tile_aspect)
return macro_tile_aspect;
}
-static unsigned cik_bank_wh(unsigned bankwh)
+unsigned cik_bank_wh(unsigned bankwh)
{
switch (bankwh) {
default:
@@ -147,7 +147,7 @@ static unsigned cik_bank_wh(unsigned bankwh)
return bankwh;
}
-static unsigned cik_db_pipe_config(struct si_screen *sscreen, unsigned
tile_mode)
+unsigned cik_db_pipe_config(struct si_screen *sscreen, unsigned tile_mode)
{
if (sscreen->b.info.si_tile_mode_array_valid) {
uint32_t gb_tile_mode =
sscreen->b.info.si_tile_mode_array[tile_mode];
@@ -1514,7 +1514,7 @@ boolean si_is_format_supported(struct pipe_screen *screen,
return retval == usage;
}
-static unsigned si_tile_mode_index(struct r600_texture *rtex, unsigned level,
bool stencil)
+unsigned si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool
stencil)
{
unsigned tile_mode_index = 0;
@@ -2926,21 +2926,6 @@ static void *si_create_blend_custom(struct si_context
*sctx, unsigned mode)
return si_create_blend_state_mode(&sctx->b.b, &blend, mode);
}
-static void si_dma_copy(struct pipe_context *ctx,
- struct pipe_resource *dst,
- unsigned dst_level,
- unsigned dst_x, unsigned dst_y, unsigned dst_z,
- struct pipe_resource *src,
- unsigned src_level,
- const struct pipe_box *src_box)
-{
- /* XXX implement this or share evergreen_dma_blit with r600g */
-
- /* Fallback: */
- ctx->resource_copy_region(ctx, dst, dst_level, dst_x, dst_y, dst_z,
- src, src_level, src_box);
-}
-
static void si_set_occlusion_query_state(struct pipe_context *ctx, bool enable)
{
/* XXX Turn this into a proper state. Right now the queries are
diff --git a/src/gallium/drivers/radeonsi/si_state.h
b/src/gallium/drivers/radeonsi/si_state.h
index f91d407..c080600 100644
--- a/src/gallium/drivers/radeonsi/si_state.h
+++ b/src/gallium/drivers/radeonsi/si_state.h
@@ -30,6 +30,8 @@
#include "si_pm4.h"
#include "../radeon/r600_pipe_common.h"
+struct si_screen;
+
struct si_state_blend {
struct si_pm4_state pm4;
uint32_t cb_target_mask;
@@ -227,6 +229,11 @@ int si_shader_select(struct pipe_context *ctx,
struct si_pipe_shader_selector *sel);
void si_init_state_functions(struct si_context *sctx);
void si_init_config(struct si_context *sctx);
+unsigned cik_bank_wh(unsigned bankwh);
+unsigned cik_db_pipe_config(struct si_screen *sscreen, unsigned tile_mode);
+unsigned cik_macro_tile_aspect(unsigned macro_tile_aspect);
+unsigned cik_tile_split(unsigned tile_split);
+unsigned si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool
stencil);
/* si_state_draw.c */
extern const struct r600_atom si_atom_cache_flush;
diff --git a/src/gallium/drivers/radeonsi/sid.h
b/src/gallium/drivers/radeonsi/sid.h
index 876d0a3..2bd2cb4 100644
--- a/src/gallium/drivers/radeonsi/sid.h
+++ b/src/gallium/drivers/radeonsi/sid.h
@@ -8630,5 +8630,25 @@
#define R_028E30_CB_COLOR7_CLEAR_WORD0
0x028E30
#define R_028E34_CB_COLOR7_CLEAR_WORD1
0x028E34
+/* SI async DMA packets */
+#define SI_DMA_PACKET(cmd, sub_cmd, n) ((((cmd) & 0xF) << 28) | \
+ (((sub_cmd) & 0xFF) << 20) |\
+ (((n) & 0xFFFFF) << 0))
+/* SI async DMA Packet types */
+#define SI_DMA_PACKET_WRITE 0x2
+#define SI_DMA_PACKET_COPY 0x3
+#define SI_DMA_COPY_MAX_SIZE 0xfffe0
+#define SI_DMA_COPY_MAX_SIZE_DW 0xffff8
+#define SI_DMA_COPY_DWORD_ALIGNED 0x00
+#define SI_DMA_COPY_BYTE_ALIGNED 0x40
+#define SI_DMA_COPY_TILED 0x8
+#define SI_DMA_PACKET_INDIRECT_BUFFER 0x4
+#define SI_DMA_PACKET_SEMAPHORE 0x5
+#define SI_DMA_PACKET_FENCE 0x6
+#define SI_DMA_PACKET_TRAP 0x7
+#define SI_DMA_PACKET_SRBM_WRITE 0x9
+#define SI_DMA_PACKET_CONSTANT_FILL 0xd
+#define SI_DMA_PACKET_NOP 0xf
+
#endif /* _SID_H */
--
1.9.0
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
