1/**********************************************************
2 * Copyright 2009-2011 VMware, Inc. All rights reserved.
3 *
4 * Permission is hereby granted, free of charge, to any person
5 * obtaining a copy of this software and associated documentation
6 * files (the "Software"), to deal in the Software without
7 * restriction, including without limitation the rights to use, copy,
8 * modify, merge, publish, distribute, sublicense, and/or sell copies
9 * of the Software, and to permit persons to whom the Software is
10 * furnished to do so, subject to the following conditions:
11 *
12 * The above copyright notice and this permission notice shall be
13 * included in all copies or substantial portions of the Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
16 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
17 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
18 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
19 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
20 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
21 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22 * SOFTWARE.
23 *
24 *********************************************************
25 * Authors:
26 * Zack Rusin <zackr-at-vmware-dot-com>
27 */
28
29#include "xa_context.h"
30#include "xa_priv.h"
31#include <math.h>
32#include "cso_cache/cso_context.h"
33#include "util/u_inlines.h"
34#include "util/u_sampler.h"
35#include "util/u_draw_quad.h"
36
37#define floatsEqual(x, y) (fabsf(x - y) <= 0.00001f * MIN2(fabsf(x), fabsf(y)))
38#define floatIsZero(x) (floatsEqual((x) + 1.0f, 1.0f))
39
40#define NUM_COMPONENTS 4
41
42void
43
44
45renderer_set_constants(struct xa_context *r,
46		       int shader_type, const float *params, int param_bytes);
47
48static inline boolean
49is_affine(const float *matrix)
50{
51    return floatIsZero(matrix[2]) && floatIsZero(matrix[5])
52	&& floatsEqual(matrix[8], 1.0f);
53}
54
55static inline void
56map_point(const float *mat, float x, float y, float *out_x, float *out_y)
57{
58    if (!mat) {
59	*out_x = x;
60	*out_y = y;
61	return;
62    }
63
64    *out_x = mat[0] * x + mat[3] * y + mat[6];
65    *out_y = mat[1] * x + mat[4] * y + mat[7];
66    if (!is_affine(mat)) {
67	float w = 1 / (mat[2] * x + mat[5] * y + mat[8]);
68
69	*out_x *= w;
70	*out_y *= w;
71    }
72}
73
74static inline void
75renderer_draw(struct xa_context *r)
76{
77    int num_verts = r->buffer_size / (r->attrs_per_vertex * NUM_COMPONENTS);
78
79    if (!r->buffer_size)
80	return;
81
82    if (!r->scissor_valid) {
83	r->scissor.minx = 0;
84	r->scissor.miny = 0;
85	r->scissor.maxx = r->dst->tex->width0;
86	r->scissor.maxy = r->dst->tex->height0;
87    }
88
89    r->pipe->set_scissor_states(r->pipe, 0, 1, &r->scissor);
90
91    struct cso_velems_state velems;
92    velems.count = r->attrs_per_vertex;
93    memcpy(velems.velems, r->velems, sizeof(r->velems[0]) * velems.count);
94
95    cso_set_vertex_elements(r->cso, &velems);
96    util_draw_user_vertex_buffer(r->cso, r->buffer, PIPE_PRIM_QUADS,
97                                 num_verts,	/* verts */
98                                 r->attrs_per_vertex);	/* attribs/vert */
99    r->buffer_size = 0;
100
101    xa_scissor_reset(r);
102}
103
104static inline void
105renderer_draw_conditional(struct xa_context *r, int next_batch)
106{
107    if (r->buffer_size + next_batch >= XA_VB_SIZE ||
108	(next_batch == 0 && r->buffer_size)) {
109	renderer_draw(r);
110    }
111}
112
113void
114renderer_init_state(struct xa_context *r)
115{
116    struct pipe_depth_stencil_alpha_state dsa;
117    struct pipe_rasterizer_state raster;
118    unsigned i;
119
120    /* set common initial clip state */
121    memset(&dsa, 0, sizeof(struct pipe_depth_stencil_alpha_state));
122    cso_set_depth_stencil_alpha(r->cso, &dsa);
123
124    /* XXX: move to renderer_init_state? */
125    memset(&raster, 0, sizeof(struct pipe_rasterizer_state));
126    raster.half_pixel_center = 1;
127    raster.bottom_edge_rule = 1;
128    raster.depth_clip_near = 1;
129    raster.depth_clip_far = 1;
130    raster.scissor = 1;
131    cso_set_rasterizer(r->cso, &raster);
132
133    /* vertex elements state */
134    memset(&r->velems[0], 0, sizeof(r->velems[0]) * 3);
135    for (i = 0; i < 3; i++) {
136	r->velems[i].src_offset = i * 4 * sizeof(float);
137	r->velems[i].instance_divisor = 0;
138	r->velems[i].vertex_buffer_index = 0;
139	r->velems[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
140    }
141}
142
143static inline void
144add_vertex_none(struct xa_context *r, float x, float y)
145{
146    float *vertex = r->buffer + r->buffer_size;
147
148    vertex[0] = x;
149    vertex[1] = y;
150    vertex[2] = 0.f;		/*z */
151    vertex[3] = 1.f;		/*w */
152
153    r->buffer_size += 4;
154}
155
156static inline void
157add_vertex_1tex(struct xa_context *r, float x, float y, float s, float t)
158{
159    float *vertex = r->buffer + r->buffer_size;
160
161    vertex[0] = x;
162    vertex[1] = y;
163    vertex[2] = 0.f;		/*z */
164    vertex[3] = 1.f;		/*w */
165
166    vertex[4] = s;		/*s */
167    vertex[5] = t;		/*t */
168    vertex[6] = 0.f;		/*r */
169    vertex[7] = 1.f;		/*q */
170
171    r->buffer_size += 8;
172}
173
174static inline void
175add_vertex_2tex(struct xa_context *r,
176		float x, float y, float s0, float t0, float s1, float t1)
177{
178    float *vertex = r->buffer + r->buffer_size;
179
180    vertex[0] = x;
181    vertex[1] = y;
182    vertex[2] = 0.f;		/*z */
183    vertex[3] = 1.f;		/*w */
184
185    vertex[4] = s0;		/*s */
186    vertex[5] = t0;		/*t */
187    vertex[6] = 0.f;		/*r */
188    vertex[7] = 1.f;		/*q */
189
190    vertex[8] = s1;		/*s */
191    vertex[9] = t1;		/*t */
192    vertex[10] = 0.f;		/*r */
193    vertex[11] = 1.f;		/*q */
194
195    r->buffer_size += 12;
196}
197
198static void
199compute_src_coords(float sx, float sy, const struct pipe_resource *src,
200                   const float *src_matrix,
201                   float width, float height,
202                   float tc0[2], float tc1[2], float tc2[2], float tc3[2])
203{
204    tc0[0] = sx;
205    tc0[1] = sy;
206    tc1[0] = sx + width;
207    tc1[1] = sy;
208    tc2[0] = sx + width;
209    tc2[1] = sy + height;
210    tc3[0] = sx;
211    tc3[1] = sy + height;
212
213    if (src_matrix) {
214	map_point(src_matrix, tc0[0], tc0[1], &tc0[0], &tc0[1]);
215	map_point(src_matrix, tc1[0], tc1[1], &tc1[0], &tc1[1]);
216	map_point(src_matrix, tc2[0], tc2[1], &tc2[0], &tc2[1]);
217	map_point(src_matrix, tc3[0], tc3[1], &tc3[0], &tc3[1]);
218    }
219
220    tc0[0] /= src->width0;
221    tc1[0] /= src->width0;
222    tc2[0] /= src->width0;
223    tc3[0] /= src->width0;
224    tc0[1] /= src->height0;
225    tc1[1] /= src->height0;
226    tc2[1] /= src->height0;
227    tc3[1] /= src->height0;
228}
229
230static void
231add_vertex_data1(struct xa_context *r,
232                 float srcX, float srcY,  float dstX, float dstY,
233                 float width, float height,
234                 const struct pipe_resource *src, const float *src_matrix)
235{
236    float tc0[2], tc1[2], tc2[2], tc3[2];
237
238    compute_src_coords(srcX, srcY, src, src_matrix, width, height,
239                       tc0, tc1, tc2, tc3);
240    /* 1st vertex */
241    add_vertex_1tex(r, dstX, dstY, tc0[0], tc0[1]);
242    /* 2nd vertex */
243    add_vertex_1tex(r, dstX + width, dstY, tc1[0], tc1[1]);
244    /* 3rd vertex */
245    add_vertex_1tex(r, dstX + width, dstY + height, tc2[0], tc2[1]);
246    /* 4th vertex */
247    add_vertex_1tex(r, dstX, dstY + height, tc3[0], tc3[1]);
248}
249
250static void
251add_vertex_data2(struct xa_context *r,
252                 float srcX, float srcY, float maskX, float maskY,
253                 float dstX, float dstY, float width, float height,
254                 struct pipe_resource *src,
255                 struct pipe_resource *mask,
256                 const float *src_matrix, const float *mask_matrix)
257{
258    float spt0[2], spt1[2], spt2[2], spt3[2];
259    float mpt0[2], mpt1[2], mpt2[2], mpt3[2];
260
261    compute_src_coords(srcX, srcY, src, src_matrix, width, height,
262                       spt0, spt1, spt2, spt3);
263    compute_src_coords(maskX, maskY, mask, mask_matrix, width, height,
264                       mpt0, mpt1, mpt2, mpt3);
265
266    /* 1st vertex */
267    add_vertex_2tex(r, dstX, dstY,
268		    spt0[0], spt0[1], mpt0[0], mpt0[1]);
269    /* 2nd vertex */
270    add_vertex_2tex(r, dstX + width, dstY,
271		    spt1[0], spt1[1], mpt1[0], mpt1[1]);
272    /* 3rd vertex */
273    add_vertex_2tex(r, dstX + width, dstY + height,
274		    spt2[0], spt2[1], mpt2[0], mpt2[1]);
275    /* 4th vertex */
276    add_vertex_2tex(r, dstX, dstY + height,
277		    spt3[0], spt3[1], mpt3[0], mpt3[1]);
278}
279
280static void
281setup_vertex_data_yuv(struct xa_context *r,
282		      float srcX,
283		      float srcY,
284		      float srcW,
285		      float srcH,
286		      float dstX,
287		      float dstY,
288		      float dstW, float dstH, struct xa_surface *srf[])
289{
290    float s0, t0, s1, t1;
291    float spt0[2], spt1[2];
292    struct pipe_resource *tex;
293
294    spt0[0] = srcX;
295    spt0[1] = srcY;
296    spt1[0] = srcX + srcW;
297    spt1[1] = srcY + srcH;
298
299    tex = srf[0]->tex;
300    s0 = spt0[0] / tex->width0;
301    t0 = spt0[1] / tex->height0;
302    s1 = spt1[0] / tex->width0;
303    t1 = spt1[1] / tex->height0;
304
305    /* 1st vertex */
306    add_vertex_1tex(r, dstX, dstY, s0, t0);
307    /* 2nd vertex */
308    add_vertex_1tex(r, dstX + dstW, dstY, s1, t0);
309    /* 3rd vertex */
310    add_vertex_1tex(r, dstX + dstW, dstY + dstH, s1, t1);
311    /* 4th vertex */
312    add_vertex_1tex(r, dstX, dstY + dstH, s0, t1);
313}
314
315/* Set up framebuffer, viewport and vertex shader constant buffer
316 * state for a particular destinaton surface.  In all our rendering,
317 * these concepts are linked.
318 */
319void
320renderer_bind_destination(struct xa_context *r,
321			  struct pipe_surface *surface)
322{
323    int width = surface->width;
324    int height = surface->height;
325
326    struct pipe_framebuffer_state fb;
327    struct pipe_viewport_state viewport;
328
329    xa_scissor_reset(r);
330
331    /* Framebuffer uses actual surface width/height
332     */
333    memset(&fb, 0, sizeof fb);
334    fb.width = surface->width;
335    fb.height = surface->height;
336    fb.nr_cbufs = 1;
337    fb.cbufs[0] = surface;
338    fb.zsbuf = NULL;
339
340    /* Viewport just touches the bit we're interested in:
341     */
342    viewport.scale[0] = width / 2.f;
343    viewport.scale[1] = height / 2.f;
344    viewport.scale[2] = 1.0;
345    viewport.translate[0] = width / 2.f;
346    viewport.translate[1] = height / 2.f;
347    viewport.translate[2] = 0.0;
348    viewport.swizzle_x = PIPE_VIEWPORT_SWIZZLE_POSITIVE_X;
349    viewport.swizzle_y = PIPE_VIEWPORT_SWIZZLE_POSITIVE_Y;
350    viewport.swizzle_z = PIPE_VIEWPORT_SWIZZLE_POSITIVE_Z;
351    viewport.swizzle_w = PIPE_VIEWPORT_SWIZZLE_POSITIVE_W;
352
353    /* Constant buffer set up to match viewport dimensions:
354     */
355    if (r->fb_width != width || r->fb_height != height) {
356	float vs_consts[8] = {
357	    2.f / width, 2.f / height, 1, 1,
358	    -1, -1, 0, 0
359	};
360
361	r->fb_width = width;
362	r->fb_height = height;
363
364	renderer_set_constants(r, PIPE_SHADER_VERTEX,
365			       vs_consts, sizeof vs_consts);
366    }
367
368    cso_set_framebuffer(r->cso, &fb);
369    cso_set_viewport(r->cso, &viewport);
370}
371
372void
373renderer_set_constants(struct xa_context *r,
374		       int shader_type, const float *params, int param_bytes)
375{
376    struct pipe_resource **cbuf =
377	(shader_type == PIPE_SHADER_VERTEX) ? &r->vs_const_buffer :
378	&r->fs_const_buffer;
379
380    pipe_resource_reference(cbuf, NULL);
381    *cbuf = pipe_buffer_create_const0(r->pipe->screen,
382                                      PIPE_BIND_CONSTANT_BUFFER,
383                                      PIPE_USAGE_DEFAULT,
384                                      param_bytes);
385
386    if (*cbuf) {
387	pipe_buffer_write(r->pipe, *cbuf, 0, param_bytes, params);
388    }
389    pipe_set_constant_buffer(r->pipe, shader_type, 0, *cbuf);
390}
391
392void
393renderer_copy_prepare(struct xa_context *r,
394		      struct pipe_surface *dst_surface,
395		      struct pipe_resource *src_texture,
396		      const enum xa_formats src_xa_format,
397		      const enum xa_formats dst_xa_format)
398{
399    struct pipe_context *pipe = r->pipe;
400    struct pipe_screen *screen = pipe->screen;
401    struct xa_shader shader;
402    uint32_t fs_traits = FS_COMPOSITE;
403
404    assert(screen->is_format_supported(screen, dst_surface->format,
405				       PIPE_TEXTURE_2D, 0, 0,
406				       PIPE_BIND_RENDER_TARGET));
407    (void)screen;
408
409    renderer_bind_destination(r, dst_surface);
410
411    /* set misc state we care about */
412    {
413	struct pipe_blend_state blend;
414
415	memset(&blend, 0, sizeof(blend));
416	blend.rt[0].rgb_src_factor = PIPE_BLENDFACTOR_ONE;
417	blend.rt[0].alpha_src_factor = PIPE_BLENDFACTOR_ONE;
418	blend.rt[0].rgb_dst_factor = PIPE_BLENDFACTOR_ZERO;
419	blend.rt[0].alpha_dst_factor = PIPE_BLENDFACTOR_ZERO;
420	blend.rt[0].colormask = PIPE_MASK_RGBA;
421	cso_set_blend(r->cso, &blend);
422    }
423
424    /* sampler */
425    {
426	struct pipe_sampler_state sampler;
427        const struct pipe_sampler_state *p_sampler = &sampler;
428
429	memset(&sampler, 0, sizeof(sampler));
430	sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
431	sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
432	sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
433	sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
434	sampler.min_img_filter = PIPE_TEX_FILTER_NEAREST;
435	sampler.mag_img_filter = PIPE_TEX_FILTER_NEAREST;
436	sampler.normalized_coords = 1;
437        cso_set_samplers(r->cso, PIPE_SHADER_FRAGMENT, 1, &p_sampler);
438        r->num_bound_samplers = 1;
439    }
440
441    /* texture/sampler view */
442    {
443	struct pipe_sampler_view templ;
444	struct pipe_sampler_view *src_view;
445
446	u_sampler_view_default_template(&templ,
447					src_texture, src_texture->format);
448	src_view = pipe->create_sampler_view(pipe, src_texture, &templ);
449	pipe->set_sampler_views(pipe, PIPE_SHADER_FRAGMENT, 0, 1, 0, false, &src_view);
450	pipe_sampler_view_reference(&src_view, NULL);
451    }
452
453    /* shaders */
454    if (src_texture->format == PIPE_FORMAT_L8_UNORM ||
455        src_texture->format == PIPE_FORMAT_R8_UNORM)
456	fs_traits |= FS_SRC_LUMINANCE;
457    if (dst_surface->format == PIPE_FORMAT_L8_UNORM ||
458        dst_surface->format == PIPE_FORMAT_R8_UNORM)
459	fs_traits |= FS_DST_LUMINANCE;
460    if (xa_format_a(dst_xa_format) != 0 &&
461	xa_format_a(src_xa_format) == 0)
462	fs_traits |= FS_SRC_SET_ALPHA;
463
464    shader = xa_shaders_get(r->shaders, VS_COMPOSITE, fs_traits);
465    cso_set_vertex_shader_handle(r->cso, shader.vs);
466    cso_set_fragment_shader_handle(r->cso, shader.fs);
467
468    r->buffer_size = 0;
469    r->attrs_per_vertex = 2;
470}
471
472void
473renderer_copy(struct xa_context *r,
474	      int dx,
475	      int dy,
476	      int sx,
477	      int sy,
478	      int width, int height, float src_width, float src_height)
479{
480    float s0, t0, s1, t1;
481    float x0, y0, x1, y1;
482
483    /* XXX: could put the texcoord scaling calculation into the vertex
484     * shader.
485     */
486    s0 = sx / src_width;
487    s1 = (sx + width) / src_width;
488    t0 = sy / src_height;
489    t1 = (sy + height) / src_height;
490
491    x0 = dx;
492    x1 = dx + width;
493    y0 = dy;
494    y1 = dy + height;
495
496    /* draw quad */
497    renderer_draw_conditional(r, 4 * 8);
498    add_vertex_1tex(r, x0, y0, s0, t0);
499    add_vertex_1tex(r, x1, y0, s1, t0);
500    add_vertex_1tex(r, x1, y1, s1, t1);
501    add_vertex_1tex(r, x0, y1, s0, t1);
502}
503
504void
505renderer_draw_yuv(struct xa_context *r,
506		  float src_x,
507		  float src_y,
508		  float src_w,
509		  float src_h,
510		  int dst_x,
511		  int dst_y, int dst_w, int dst_h, struct xa_surface *srf[])
512{
513   const int num_attribs = 2;	/*pos + tex coord */
514
515   setup_vertex_data_yuv(r,
516                         src_x, src_y, src_w, src_h,
517                         dst_x, dst_y, dst_w, dst_h, srf);
518
519   if (!r->scissor_valid) {
520       r->scissor.minx = 0;
521       r->scissor.miny = 0;
522       r->scissor.maxx = r->dst->tex->width0;
523       r->scissor.maxy = r->dst->tex->height0;
524   }
525
526   r->pipe->set_scissor_states(r->pipe, 0, 1, &r->scissor);
527
528   struct cso_velems_state velems;
529   velems.count = num_attribs;
530   memcpy(velems.velems, r->velems, sizeof(r->velems[0]) * velems.count);
531
532   cso_set_vertex_elements(r->cso, &velems);
533   util_draw_user_vertex_buffer(r->cso, r->buffer, PIPE_PRIM_QUADS,
534                                4,	/* verts */
535                                num_attribs);	/* attribs/vert */
536   r->buffer_size = 0;
537
538   xa_scissor_reset(r);
539}
540
541void
542renderer_begin_solid(struct xa_context *r)
543{
544    r->buffer_size = 0;
545    r->attrs_per_vertex = 1;
546    renderer_set_constants(r, PIPE_SHADER_FRAGMENT, r->solid_color,
547                           4 * sizeof(float));
548}
549
550void
551renderer_solid(struct xa_context *r,
552	       int x0, int y0, int x1, int y1)
553{
554    /*
555     * debug_printf("solid rect[(%d, %d), (%d, %d)], rgba[%f, %f, %f, %f]\n",
556     * x0, y0, x1, y1, color[0], color[1], color[2], color[3]); */
557
558    renderer_draw_conditional(r, 4 * 4);
559
560    /* 1st vertex */
561    add_vertex_none(r, x0, y0);
562    /* 2nd vertex */
563    add_vertex_none(r, x1, y0);
564    /* 3rd vertex */
565    add_vertex_none(r, x1, y1);
566    /* 4th vertex */
567    add_vertex_none(r, x0, y1);
568}
569
570void
571renderer_draw_flush(struct xa_context *r)
572{
573    renderer_draw_conditional(r, 0);
574}
575
576void
577renderer_begin_textures(struct xa_context *r)
578{
579    r->attrs_per_vertex = 1 + r->num_bound_samplers;
580    r->buffer_size = 0;
581    if (r->has_solid_src || r->has_solid_mask)
582       renderer_set_constants(r, PIPE_SHADER_FRAGMENT, r->solid_color,
583                              4 * sizeof(float));
584}
585
586void
587renderer_texture(struct xa_context *r,
588		 int *pos,
589		 int width, int height,
590		 const float *src_matrix,
591		 const float *mask_matrix)
592{
593    struct pipe_sampler_view **sampler_view = r->bound_sampler_views;
594
595#if 0
596    if (src_matrix) {
597	debug_printf("src_matrix = \n");
598	debug_printf("%f, %f, %f\n", src_matrix[0], src_matrix[1], src_matrix[2]);
599	debug_printf("%f, %f, %f\n", src_matrix[3], src_matrix[4], src_matrix[5]);
600	debug_printf("%f, %f, %f\n", src_matrix[6], src_matrix[7], src_matrix[8]);
601    }
602    if (mask_matrix) {
603	debug_printf("mask_matrix = \n");
604	debug_printf("%f, %f, %f\n", mask_matrix[0], mask_matrix[1], mask_matrix[2]);
605	debug_printf("%f, %f, %f\n", mask_matrix[3], mask_matrix[4], mask_matrix[5]);
606	debug_printf("%f, %f, %f\n", mask_matrix[6], mask_matrix[7], mask_matrix[8]);
607    }
608#endif
609
610    switch(r->attrs_per_vertex) {
611    case 2:
612	renderer_draw_conditional(r, 4 * 8);
613        if (!r->has_solid_src) {
614           add_vertex_data1(r,
615                            pos[0], pos[1], /* src */
616                            pos[4], pos[5], /* dst */
617                            width, height,
618                            sampler_view[0]->texture, src_matrix);
619        } else {
620           add_vertex_data1(r,
621                            pos[2], pos[3], /* mask */
622                            pos[4], pos[5], /* dst */
623                            width, height,
624                            sampler_view[0]->texture, mask_matrix);
625        }
626	break;
627    case 3:
628	renderer_draw_conditional(r, 4 * 12);
629	add_vertex_data2(r,
630			 pos[0], pos[1], /* src */
631			 pos[2], pos[3], /* mask */
632			 pos[4], pos[5], /* dst */
633			 width, height,
634			 sampler_view[0]->texture, sampler_view[1]->texture,
635			 src_matrix, mask_matrix);
636	break;
637    default:
638	break;
639    }
640}
641