1 /**************************************************************************
3 * Copyright 2011 Christian König
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sub license, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21 * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 **************************************************************************/
30 #include "pipe/p_screen.h"
31 #include "pipe/p_context.h"
33 #include "util/u_draw.h"
34 #include "util/u_sampler.h"
35 #include "util/u_inlines.h"
36 #include "util/u_memory.h"
38 #include "tgsi/tgsi_ureg.h"
40 #include "vl_defines.h"
44 #include "vl_vertex_buffers.h"
52 const int vl_zscan_normal_16
[] =
54 /* Zig-Zag scan pattern */
55 0, 1, 4, 8, 5, 2, 3, 6,
56 9,12,13,10, 7,11,14,15
59 const int vl_zscan_linear
[] =
61 /* Linear scan pattern */
62 0, 1, 2, 3, 4, 5, 6, 7,
63 8, 9,10,11,12,13,14,15,
64 16,17,18,19,20,21,22,23,
65 24,25,26,27,28,29,30,31,
66 32,33,34,35,36,37,38,39,
67 40,41,42,43,44,45,46,47,
68 48,49,50,51,52,53,54,55,
69 56,57,58,59,60,61,62,63
72 const int vl_zscan_normal
[] =
74 /* Zig-Zag scan pattern */
75 0, 1, 8,16, 9, 2, 3,10,
76 17,24,32,25,18,11, 4, 5,
77 12,19,26,33,40,48,41,34,
78 27,20,13, 6, 7,14,21,28,
79 35,42,49,56,57,50,43,36,
80 29,22,15,23,30,37,44,51,
81 58,59,52,45,38,31,39,46,
82 53,60,61,54,47,55,62,63
85 const int vl_zscan_alternate
[] =
87 /* Alternate scan pattern */
88 0, 8,16,24, 1, 9, 2,10,
89 17,25,32,40,48,56,57,49,
90 41,33,26,18, 3,11, 4,12,
91 19,27,34,42,50,58,35,43,
92 51,59,20,28, 5,13, 6,14,
93 21,29,36,44,52,60,37,45,
94 53,61,22,30, 7,15,23,31,
95 38,46,54,62,39,47,55,63
99 create_vert_shader(struct vl_zscan
*zscan
)
101 struct ureg_program
*shader
;
103 struct ureg_src scale
;
104 struct ureg_src vrect
, vpos
, block_num
;
107 struct ureg_dst o_vpos
;
108 struct ureg_dst
*o_vtex
;
112 shader
= ureg_create(PIPE_SHADER_VERTEX
);
116 o_vtex
= MALLOC(zscan
->num_channels
* sizeof(struct ureg_dst
));
118 scale
= ureg_imm2f(shader
,
119 (float)VL_BLOCK_WIDTH
/ zscan
->buffer_width
,
120 (float)VL_BLOCK_HEIGHT
/ zscan
->buffer_height
);
122 vrect
= ureg_DECL_vs_input(shader
, VS_I_RECT
);
123 vpos
= ureg_DECL_vs_input(shader
, VS_I_VPOS
);
124 block_num
= ureg_DECL_vs_input(shader
, VS_I_BLOCK_NUM
);
126 tmp
= ureg_DECL_temporary(shader
);
128 o_vpos
= ureg_DECL_output(shader
, TGSI_SEMANTIC_POSITION
, VS_O_VPOS
);
130 for (i
= 0; i
< zscan
->num_channels
; ++i
)
131 o_vtex
[i
] = ureg_DECL_output(shader
, TGSI_SEMANTIC_GENERIC
, VS_O_VTEX
+ i
);
134 * o_vpos.xy = (vpos + vrect) * scale
137 * tmp.xy = InstanceID / blocks_per_line
138 * tmp.x = frac(tmp.x)
139 * tmp.y = floor(tmp.y)
141 * o_vtex.x = vrect.x / blocks_per_line + tmp.x
143 * o_vtex.z = tmp.z * blocks_per_line / blocks_total
145 ureg_ADD(shader
, ureg_writemask(tmp
, TGSI_WRITEMASK_XY
), vpos
, vrect
);
146 ureg_MUL(shader
, ureg_writemask(o_vpos
, TGSI_WRITEMASK_XY
), ureg_src(tmp
), scale
);
147 ureg_MOV(shader
, ureg_writemask(o_vpos
, TGSI_WRITEMASK_ZW
), ureg_imm1f(shader
, 1.0f
));
149 ureg_MUL(shader
, ureg_writemask(tmp
, TGSI_WRITEMASK_XW
), ureg_scalar(block_num
, TGSI_SWIZZLE_X
),
150 ureg_imm1f(shader
, 1.0f
/ zscan
->blocks_per_line
));
152 ureg_FRC(shader
, ureg_writemask(tmp
, TGSI_WRITEMASK_Y
), ureg_scalar(ureg_src(tmp
), TGSI_SWIZZLE_X
));
153 ureg_FLR(shader
, ureg_writemask(tmp
, TGSI_WRITEMASK_W
), ureg_src(tmp
));
155 for (i
= 0; i
< zscan
->num_channels
; ++i
) {
156 ureg_ADD(shader
, ureg_writemask(tmp
, TGSI_WRITEMASK_X
), ureg_scalar(ureg_src(tmp
), TGSI_SWIZZLE_Y
),
157 ureg_imm1f(shader
, 1.0f
/ (zscan
->blocks_per_line
* VL_BLOCK_WIDTH
)
158 * (i
- (signed)zscan
->num_channels
/ 2)));
160 ureg_MAD(shader
, ureg_writemask(o_vtex
[i
], TGSI_WRITEMASK_X
), vrect
,
161 ureg_imm1f(shader
, 1.0f
/ zscan
->blocks_per_line
), ureg_src(tmp
));
162 ureg_MOV(shader
, ureg_writemask(o_vtex
[i
], TGSI_WRITEMASK_Y
), vrect
);
163 ureg_MOV(shader
, ureg_writemask(o_vtex
[i
], TGSI_WRITEMASK_Z
), vpos
);
164 ureg_MUL(shader
, ureg_writemask(o_vtex
[i
], TGSI_WRITEMASK_W
), ureg_src(tmp
),
165 ureg_imm1f(shader
, (float)zscan
->blocks_per_line
/ zscan
->blocks_total
));
168 ureg_release_temporary(shader
, tmp
);
173 return ureg_create_shader_and_destroy(shader
, zscan
->pipe
);
177 create_frag_shader(struct vl_zscan
*zscan
)
179 struct ureg_program
*shader
;
180 struct ureg_src
*vtex
;
182 struct ureg_src samp_src
, samp_scan
, samp_quant
;
184 struct ureg_dst
*tmp
;
185 struct ureg_dst quant
, fragment
;
189 shader
= ureg_create(PIPE_SHADER_FRAGMENT
);
193 vtex
= MALLOC(zscan
->num_channels
* sizeof(struct ureg_src
));
194 tmp
= MALLOC(zscan
->num_channels
* sizeof(struct ureg_dst
));
196 for (i
= 0; i
< zscan
->num_channels
; ++i
)
197 vtex
[i
] = ureg_DECL_fs_input(shader
, TGSI_SEMANTIC_GENERIC
, VS_O_VTEX
+ i
, TGSI_INTERPOLATE_LINEAR
);
199 samp_src
= ureg_DECL_sampler(shader
, 0);
200 samp_scan
= ureg_DECL_sampler(shader
, 1);
201 samp_quant
= ureg_DECL_sampler(shader
, 2);
203 for (i
= 0; i
< zscan
->num_channels
; ++i
)
204 tmp
[i
] = ureg_DECL_temporary(shader
);
205 quant
= ureg_DECL_temporary(shader
);
207 fragment
= ureg_DECL_output(shader
, TGSI_SEMANTIC_COLOR
, 0);
210 * tmp.x = tex(vtex, 1)
212 * fragment = tex(tmp, 0) * quant
214 for (i
= 0; i
< zscan
->num_channels
; ++i
)
215 ureg_TEX(shader
, ureg_writemask(tmp
[i
], TGSI_WRITEMASK_X
), TGSI_TEXTURE_2D
, vtex
[i
], samp_scan
);
217 for (i
= 0; i
< zscan
->num_channels
; ++i
)
218 ureg_MOV(shader
, ureg_writemask(tmp
[i
], TGSI_WRITEMASK_Y
), ureg_scalar(vtex
[i
], TGSI_SWIZZLE_W
));
220 for (i
= 0; i
< zscan
->num_channels
; ++i
) {
221 ureg_TEX(shader
, ureg_writemask(tmp
[0], TGSI_WRITEMASK_X
<< i
), TGSI_TEXTURE_2D
, ureg_src(tmp
[i
]), samp_src
);
222 ureg_TEX(shader
, ureg_writemask(quant
, TGSI_WRITEMASK_X
<< i
), TGSI_TEXTURE_3D
, vtex
[i
], samp_quant
);
225 ureg_MUL(shader
, quant
, ureg_src(quant
), ureg_imm1f(shader
, 16.0f
));
226 ureg_MUL(shader
, fragment
, ureg_src(tmp
[0]), ureg_src(quant
));
228 for (i
= 0; i
< zscan
->num_channels
; ++i
)
229 ureg_release_temporary(shader
, tmp
[i
]);
235 return ureg_create_shader_and_destroy(shader
, zscan
->pipe
);
239 init_shaders(struct vl_zscan
*zscan
)
243 zscan
->vs
= create_vert_shader(zscan
);
247 zscan
->fs
= create_frag_shader(zscan
);
254 zscan
->pipe
->delete_vs_state(zscan
->pipe
, zscan
->vs
);
261 cleanup_shaders(struct vl_zscan
*zscan
)
265 zscan
->pipe
->delete_vs_state(zscan
->pipe
, zscan
->vs
);
266 zscan
->pipe
->delete_fs_state(zscan
->pipe
, zscan
->fs
);
270 init_state(struct vl_zscan
*zscan
)
272 struct pipe_blend_state blend
;
273 struct pipe_rasterizer_state rs_state
;
274 struct pipe_sampler_state sampler
;
279 memset(&rs_state
, 0, sizeof(rs_state
));
280 rs_state
.half_pixel_center
= true;
281 rs_state
.bottom_edge_rule
= true;
282 rs_state
.depth_clip
= 1;
283 zscan
->rs_state
= zscan
->pipe
->create_rasterizer_state(zscan
->pipe
, &rs_state
);
284 if (!zscan
->rs_state
)
287 memset(&blend
, 0, sizeof blend
);
289 blend
.independent_blend_enable
= 0;
290 blend
.rt
[0].blend_enable
= 0;
291 blend
.rt
[0].rgb_func
= PIPE_BLEND_ADD
;
292 blend
.rt
[0].rgb_src_factor
= PIPE_BLENDFACTOR_ONE
;
293 blend
.rt
[0].rgb_dst_factor
= PIPE_BLENDFACTOR_ONE
;
294 blend
.rt
[0].alpha_func
= PIPE_BLEND_ADD
;
295 blend
.rt
[0].alpha_src_factor
= PIPE_BLENDFACTOR_ONE
;
296 blend
.rt
[0].alpha_dst_factor
= PIPE_BLENDFACTOR_ONE
;
297 blend
.logicop_enable
= 0;
298 blend
.logicop_func
= PIPE_LOGICOP_CLEAR
;
299 /* Needed to allow color writes to FB, even if blending disabled */
300 blend
.rt
[0].colormask
= PIPE_MASK_RGBA
;
302 zscan
->blend
= zscan
->pipe
->create_blend_state(zscan
->pipe
, &blend
);
306 for (i
= 0; i
< 3; ++i
) {
307 memset(&sampler
, 0, sizeof(sampler
));
308 sampler
.wrap_s
= PIPE_TEX_WRAP_REPEAT
;
309 sampler
.wrap_t
= PIPE_TEX_WRAP_REPEAT
;
310 sampler
.wrap_r
= PIPE_TEX_WRAP_CLAMP_TO_EDGE
;
311 sampler
.min_img_filter
= PIPE_TEX_FILTER_NEAREST
;
312 sampler
.min_mip_filter
= PIPE_TEX_MIPFILTER_NONE
;
313 sampler
.mag_img_filter
= PIPE_TEX_FILTER_NEAREST
;
314 sampler
.compare_mode
= PIPE_TEX_COMPARE_NONE
;
315 sampler
.compare_func
= PIPE_FUNC_ALWAYS
;
316 sampler
.normalized_coords
= 1;
317 zscan
->samplers
[i
] = zscan
->pipe
->create_sampler_state(zscan
->pipe
, &sampler
);
318 if (!zscan
->samplers
[i
])
325 for (i
= 0; i
< 2; ++i
)
326 if (zscan
->samplers
[i
])
327 zscan
->pipe
->delete_sampler_state(zscan
->pipe
, zscan
->samplers
[i
]);
329 zscan
->pipe
->delete_rasterizer_state(zscan
->pipe
, zscan
->rs_state
);
332 zscan
->pipe
->delete_blend_state(zscan
->pipe
, zscan
->blend
);
339 cleanup_state(struct vl_zscan
*zscan
)
345 for (i
= 0; i
< 3; ++i
)
346 zscan
->pipe
->delete_sampler_state(zscan
->pipe
, zscan
->samplers
[i
]);
348 zscan
->pipe
->delete_rasterizer_state(zscan
->pipe
, zscan
->rs_state
);
349 zscan
->pipe
->delete_blend_state(zscan
->pipe
, zscan
->blend
);
352 struct pipe_sampler_view
*
353 vl_zscan_layout(struct pipe_context
*pipe
, const int layout
[64], unsigned blocks_per_line
)
355 const unsigned total_size
= blocks_per_line
* VL_BLOCK_WIDTH
* VL_BLOCK_HEIGHT
;
357 int patched_layout
[64];
359 struct pipe_resource res_tmpl
, *res
;
360 struct pipe_sampler_view sv_tmpl
, *sv
;
361 struct pipe_transfer
*buf_transfer
;
362 unsigned x
, y
, i
, pitch
;
365 struct pipe_box rect
=
368 VL_BLOCK_WIDTH
* blocks_per_line
,
373 assert(pipe
&& layout
&& blocks_per_line
);
375 for (i
= 0; i
< 64; ++i
)
376 patched_layout
[layout
[i
]] = i
;
378 memset(&res_tmpl
, 0, sizeof(res_tmpl
));
379 res_tmpl
.target
= PIPE_TEXTURE_2D
;
380 res_tmpl
.format
= PIPE_FORMAT_R32_FLOAT
;
381 res_tmpl
.width0
= VL_BLOCK_WIDTH
* blocks_per_line
;
382 res_tmpl
.height0
= VL_BLOCK_HEIGHT
;
384 res_tmpl
.array_size
= 1;
385 res_tmpl
.usage
= PIPE_USAGE_IMMUTABLE
;
386 res_tmpl
.bind
= PIPE_BIND_SAMPLER_VIEW
;
388 res
= pipe
->screen
->resource_create(pipe
->screen
, &res_tmpl
);
392 f
= pipe
->transfer_map(pipe
, res
,
393 0, PIPE_TRANSFER_WRITE
| PIPE_TRANSFER_DISCARD_RANGE
,
394 &rect
, &buf_transfer
);
398 pitch
= buf_transfer
->stride
/ sizeof(float);
400 for (i
= 0; i
< blocks_per_line
; ++i
)
401 for (y
= 0; y
< VL_BLOCK_HEIGHT
; ++y
)
402 for (x
= 0; x
< VL_BLOCK_WIDTH
; ++x
) {
403 float addr
= patched_layout
[x
+ y
* VL_BLOCK_WIDTH
] +
404 i
* VL_BLOCK_WIDTH
* VL_BLOCK_HEIGHT
;
408 f
[i
* VL_BLOCK_WIDTH
+ y
* pitch
+ x
] = addr
;
411 pipe
->transfer_unmap(pipe
, buf_transfer
);
413 memset(&sv_tmpl
, 0, sizeof(sv_tmpl
));
414 u_sampler_view_default_template(&sv_tmpl
, res
, res
->format
);
415 sv
= pipe
->create_sampler_view(pipe
, res
, &sv_tmpl
);
416 pipe_resource_reference(&res
, NULL
);
423 pipe_resource_reference(&res
, NULL
);
430 vl_zscan_init(struct vl_zscan
*zscan
, struct pipe_context
*pipe
,
431 unsigned buffer_width
, unsigned buffer_height
,
432 unsigned blocks_per_line
, unsigned blocks_total
,
433 unsigned num_channels
)
435 assert(zscan
&& pipe
);
438 zscan
->buffer_width
= buffer_width
;
439 zscan
->buffer_height
= buffer_height
;
440 zscan
->num_channels
= num_channels
;
441 zscan
->blocks_per_line
= blocks_per_line
;
442 zscan
->blocks_total
= blocks_total
;
444 if(!init_shaders(zscan
))
447 if(!init_state(zscan
)) {
448 cleanup_shaders(zscan
);
456 vl_zscan_cleanup(struct vl_zscan
*zscan
)
460 cleanup_shaders(zscan
);
461 cleanup_state(zscan
);
465 vl_zscan_init_buffer(struct vl_zscan
*zscan
, struct vl_zscan_buffer
*buffer
,
466 struct pipe_sampler_view
*src
, struct pipe_surface
*dst
)
468 struct pipe_resource res_tmpl
, *res
;
469 struct pipe_sampler_view sv_tmpl
;
471 assert(zscan
&& buffer
);
473 memset(buffer
, 0, sizeof(struct vl_zscan_buffer
));
475 pipe_sampler_view_reference(&buffer
->src
, src
);
477 buffer
->viewport
.scale
[0] = dst
->width
;
478 buffer
->viewport
.scale
[1] = dst
->height
;
479 buffer
->viewport
.scale
[2] = 1;
480 buffer
->viewport
.translate
[0] = 0;
481 buffer
->viewport
.translate
[1] = 0;
482 buffer
->viewport
.translate
[2] = 0;
484 buffer
->fb_state
.width
= dst
->width
;
485 buffer
->fb_state
.height
= dst
->height
;
486 buffer
->fb_state
.nr_cbufs
= 1;
487 pipe_surface_reference(&buffer
->fb_state
.cbufs
[0], dst
);
489 memset(&res_tmpl
, 0, sizeof(res_tmpl
));
490 res_tmpl
.target
= PIPE_TEXTURE_3D
;
491 res_tmpl
.format
= PIPE_FORMAT_R8_UNORM
;
492 res_tmpl
.width0
= VL_BLOCK_WIDTH
* zscan
->blocks_per_line
;
493 res_tmpl
.height0
= VL_BLOCK_HEIGHT
;
495 res_tmpl
.array_size
= 1;
496 res_tmpl
.usage
= PIPE_USAGE_IMMUTABLE
;
497 res_tmpl
.bind
= PIPE_BIND_SAMPLER_VIEW
;
499 res
= zscan
->pipe
->screen
->resource_create(zscan
->pipe
->screen
, &res_tmpl
);
503 memset(&sv_tmpl
, 0, sizeof(sv_tmpl
));
504 u_sampler_view_default_template(&sv_tmpl
, res
, res
->format
);
505 sv_tmpl
.swizzle_r
= sv_tmpl
.swizzle_g
= sv_tmpl
.swizzle_b
= sv_tmpl
.swizzle_a
= TGSI_SWIZZLE_X
;
506 buffer
->quant
= zscan
->pipe
->create_sampler_view(zscan
->pipe
, res
, &sv_tmpl
);
507 pipe_resource_reference(&res
, NULL
);
515 vl_zscan_cleanup_buffer(struct vl_zscan_buffer
*buffer
)
519 pipe_sampler_view_reference(&buffer
->src
, NULL
);
520 pipe_sampler_view_reference(&buffer
->layout
, NULL
);
521 pipe_sampler_view_reference(&buffer
->quant
, NULL
);
522 pipe_surface_reference(&buffer
->fb_state
.cbufs
[0], NULL
);
526 vl_zscan_set_layout(struct vl_zscan_buffer
*buffer
, struct pipe_sampler_view
*layout
)
531 pipe_sampler_view_reference(&buffer
->layout
, layout
);
535 vl_zscan_upload_quant(struct vl_zscan
*zscan
, struct vl_zscan_buffer
*buffer
,
536 const uint8_t matrix
[64], bool intra
)
538 struct pipe_context
*pipe
;
539 struct pipe_transfer
*buf_transfer
;
540 unsigned x
, y
, i
, pitch
;
543 struct pipe_box rect
=
556 rect
.width
*= zscan
->blocks_per_line
;
558 data
= pipe
->transfer_map(pipe
, buffer
->quant
->texture
,
559 0, PIPE_TRANSFER_WRITE
|
560 PIPE_TRANSFER_DISCARD_RANGE
,
561 &rect
, &buf_transfer
);
565 pitch
= buf_transfer
->stride
;
567 for (i
= 0; i
< zscan
->blocks_per_line
; ++i
)
568 for (y
= 0; y
< VL_BLOCK_HEIGHT
; ++y
)
569 for (x
= 0; x
< VL_BLOCK_WIDTH
; ++x
)
570 data
[i
* VL_BLOCK_WIDTH
+ y
* pitch
+ x
] = matrix
[x
+ y
* VL_BLOCK_WIDTH
];
572 pipe
->transfer_unmap(pipe
, buf_transfer
);
576 vl_zscan_render(struct vl_zscan
*zscan
, struct vl_zscan_buffer
*buffer
, unsigned num_instances
)
580 zscan
->pipe
->bind_rasterizer_state(zscan
->pipe
, zscan
->rs_state
);
581 zscan
->pipe
->bind_blend_state(zscan
->pipe
, zscan
->blend
);
582 zscan
->pipe
->bind_sampler_states(zscan
->pipe
, PIPE_SHADER_FRAGMENT
,
583 0, 3, zscan
->samplers
);
584 zscan
->pipe
->set_framebuffer_state(zscan
->pipe
, &buffer
->fb_state
);
585 zscan
->pipe
->set_viewport_states(zscan
->pipe
, 0, 1, &buffer
->viewport
);
586 zscan
->pipe
->set_sampler_views(zscan
->pipe
, PIPE_SHADER_FRAGMENT
,
588 zscan
->pipe
->bind_vs_state(zscan
->pipe
, zscan
->vs
);
589 zscan
->pipe
->bind_fs_state(zscan
->pipe
, zscan
->fs
);
590 util_draw_arrays_instanced(zscan
->pipe
, PIPE_PRIM_QUADS
, 0, 4, 0, num_instances
);