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) (fabs(x - y) <= 0.00001f * MIN2(fabs(x), fabs(y))) 38 #define floatIsZero(x) (floatsEqual((x) + 1, 1)) 39 40 #define NUM_COMPONENTS 4 41 42 void 43 44 45 renderer_set_constants(struct xa_context *r, 46 int shader_type, const float *params, int param_bytes); 47 48 static INLINE boolean 49 is_affine(float *matrix) 50 { 51 return floatIsZero(matrix[2]) && floatIsZero(matrix[5]) 52 && floatsEqual(matrix[8], 1); 53 } 54 55 static INLINE void 56 map_point(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 74 static INLINE void 75 renderer_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 cso_set_vertex_elements(r->cso, r->attrs_per_vertex, r->velems); 83 util_draw_user_vertex_buffer(r->cso, r->buffer, PIPE_PRIM_QUADS, 84 num_verts, /* verts */ 85 r->attrs_per_vertex); /* attribs/vert */ 86 r->buffer_size = 0; 87 } 88 89 static INLINE void 90 renderer_draw_conditional(struct xa_context *r, int next_batch) 91 { 92 if (r->buffer_size + next_batch >= XA_VB_SIZE || 93 (next_batch == 0 && r->buffer_size)) { 94 renderer_draw(r); 95 } 96 } 97 98 void 99 renderer_init_state(struct xa_context *r) 100 { 101 struct pipe_depth_stencil_alpha_state dsa; 102 struct pipe_rasterizer_state raster; 103 unsigned i; 104 105 /* set common initial clip state */ 106 memset(&dsa, 0, sizeof(struct pipe_depth_stencil_alpha_state)); 107 cso_set_depth_stencil_alpha(r->cso, &dsa); 108 109 /* XXX: move to renderer_init_state? */ 110 memset(&raster, 0, sizeof(struct pipe_rasterizer_state)); 111 raster.gl_rasterization_rules = 1; 112 raster.depth_clip = 1; 113 cso_set_rasterizer(r->cso, &raster); 114 115 /* vertex elements state */ 116 memset(&r->velems[0], 0, sizeof(r->velems[0]) * 3); 117 for (i = 0; i < 3; i++) { 118 r->velems[i].src_offset = i * 4 * sizeof(float); 119 r->velems[i].instance_divisor = 0; 120 r->velems[i].vertex_buffer_index = 0; 121 r->velems[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT; 122 } 123 } 124 125 static INLINE void 126 add_vertex_color(struct xa_context *r, float x, float y, float color[4]) 127 { 128 float *vertex = r->buffer + r->buffer_size; 129 130 vertex[0] = x; 131 vertex[1] = y; 132 vertex[2] = 0.f; /*z */ 133 vertex[3] = 1.f; /*w */ 134 135 vertex[4] = color[0]; /*r */ 136 vertex[5] = color[1]; /*g */ 137 vertex[6] = color[2]; /*b */ 138 vertex[7] = color[3]; /*a */ 139 140 r->buffer_size += 8; 141 } 142 143 static INLINE void 144 add_vertex_1tex(struct xa_context *r, float x, float y, float s, float t) 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 vertex[4] = s; /*s */ 154 vertex[5] = t; /*t */ 155 vertex[6] = 0.f; /*r */ 156 vertex[7] = 1.f; /*q */ 157 158 r->buffer_size += 8; 159 } 160 161 static INLINE void 162 add_vertex_2tex(struct xa_context *r, 163 float x, float y, float s0, float t0, float s1, float t1) 164 { 165 float *vertex = r->buffer + r->buffer_size; 166 167 vertex[0] = x; 168 vertex[1] = y; 169 vertex[2] = 0.f; /*z */ 170 vertex[3] = 1.f; /*w */ 171 172 vertex[4] = s0; /*s */ 173 vertex[5] = t0; /*t */ 174 vertex[6] = 0.f; /*r */ 175 vertex[7] = 1.f; /*q */ 176 177 vertex[8] = s1; /*s */ 178 vertex[9] = t1; /*t */ 179 vertex[10] = 0.f; /*r */ 180 vertex[11] = 1.f; /*q */ 181 182 r->buffer_size += 12; 183 } 184 185 static void 186 add_vertex_data1(struct xa_context *r, 187 float srcX, float srcY, float dstX, float dstY, 188 float width, float height, 189 struct pipe_resource *src, const float *src_matrix) 190 { 191 float s0, t0, s1, t1, s2, t2, s3, t3; 192 float pt0[2], pt1[2], pt2[2], pt3[2]; 193 194 pt0[0] = srcX; 195 pt0[1] = srcY; 196 pt1[0] = (srcX + width); 197 pt1[1] = srcY; 198 pt2[0] = (srcX + width); 199 pt2[1] = (srcY + height); 200 pt3[0] = srcX; 201 pt3[1] = (srcY + height); 202 203 if (src_matrix) { 204 map_point((float *)src_matrix, pt0[0], pt0[1], &pt0[0], &pt0[1]); 205 map_point((float *)src_matrix, pt1[0], pt1[1], &pt1[0], &pt1[1]); 206 map_point((float *)src_matrix, pt2[0], pt2[1], &pt2[0], &pt2[1]); 207 map_point((float *)src_matrix, pt3[0], pt3[1], &pt3[0], &pt3[1]); 208 } 209 210 s0 = pt0[0] / src->width0; 211 s1 = pt1[0] / src->width0; 212 s2 = pt2[0] / src->width0; 213 s3 = pt3[0] / src->width0; 214 t0 = pt0[1] / src->height0; 215 t1 = pt1[1] / src->height0; 216 t2 = pt2[1] / src->height0; 217 t3 = pt3[1] / src->height0; 218 219 /* 1st vertex */ 220 add_vertex_1tex(r, dstX, dstY, s0, t0); 221 /* 2nd vertex */ 222 add_vertex_1tex(r, dstX + width, dstY, s1, t1); 223 /* 3rd vertex */ 224 add_vertex_1tex(r, dstX + width, dstY + height, s2, t2); 225 /* 4th vertex */ 226 add_vertex_1tex(r, dstX, dstY + height, s3, t3); 227 } 228 229 static void 230 add_vertex_data2(struct xa_context *r, 231 float srcX, float srcY, float maskX, float maskY, 232 float dstX, float dstY, float width, float height, 233 struct pipe_resource *src, 234 struct pipe_resource *mask, 235 const float *src_matrix, const float *mask_matrix) 236 { 237 float src_s0, src_t0, src_s1, src_t1; 238 float mask_s0, mask_t0, mask_s1, mask_t1; 239 float spt0[2], spt1[2]; 240 float mpt0[2], mpt1[2]; 241 242 spt0[0] = srcX; 243 spt0[1] = srcY; 244 spt1[0] = srcX + width; 245 spt1[1] = srcY + height; 246 247 mpt0[0] = maskX; 248 mpt0[1] = maskY; 249 mpt1[0] = maskX + width; 250 mpt1[1] = maskY + height; 251 252 if (src_matrix) { 253 map_point((float *)src_matrix, spt0[0], spt0[1], &spt0[0], &spt0[1]); 254 map_point((float *)src_matrix, spt1[0], spt1[1], &spt1[0], &spt1[1]); 255 } 256 257 if (mask_matrix) { 258 map_point((float *)mask_matrix, mpt0[0], mpt0[1], &mpt0[0], &mpt0[1]); 259 map_point((float *)mask_matrix, mpt1[0], mpt1[1], &mpt1[0], &mpt1[1]); 260 } 261 262 src_s0 = spt0[0] / src->width0; 263 src_t0 = spt0[1] / src->height0; 264 src_s1 = spt1[0] / src->width0; 265 src_t1 = spt1[1] / src->height0; 266 267 mask_s0 = mpt0[0] / mask->width0; 268 mask_t0 = mpt0[1] / mask->height0; 269 mask_s1 = mpt1[0] / mask->width0; 270 mask_t1 = mpt1[1] / mask->height0; 271 272 /* 1st vertex */ 273 add_vertex_2tex(r, dstX, dstY, 274 src_s0, src_t0, mask_s0, mask_t0); 275 /* 2nd vertex */ 276 add_vertex_2tex(r, dstX + width, dstY, 277 src_s1, src_t0, mask_s1, mask_t0); 278 /* 3rd vertex */ 279 add_vertex_2tex(r, dstX + width, dstY + height, 280 src_s1, src_t1, mask_s1, mask_t1); 281 /* 4th vertex */ 282 add_vertex_2tex(r, dstX, dstY + height, 283 src_s0, src_t1, mask_s0, mask_t1); 284 } 285 286 static void 287 setup_vertex_data_yuv(struct xa_context *r, 288 float srcX, 289 float srcY, 290 float srcW, 291 float srcH, 292 float dstX, 293 float dstY, 294 float dstW, float dstH, struct xa_surface *srf[]) 295 { 296 float s0, t0, s1, t1; 297 float spt0[2], spt1[2]; 298 struct pipe_resource *tex; 299 300 spt0[0] = srcX; 301 spt0[1] = srcY; 302 spt1[0] = srcX + srcW; 303 spt1[1] = srcY + srcH; 304 305 tex = srf[0]->tex; 306 s0 = spt0[0] / tex->width0; 307 t0 = spt0[1] / tex->height0; 308 s1 = spt1[0] / tex->width0; 309 t1 = spt1[1] / tex->height0; 310 311 /* 1st vertex */ 312 add_vertex_1tex(r, dstX, dstY, s0, t0); 313 /* 2nd vertex */ 314 add_vertex_1tex(r, dstX + dstW, dstY, s1, t0); 315 /* 3rd vertex */ 316 add_vertex_1tex(r, dstX + dstW, dstY + dstH, s1, t1); 317 /* 4th vertex */ 318 add_vertex_1tex(r, dstX, dstY + dstH, s0, t1); 319 } 320 321 /* Set up framebuffer, viewport and vertex shader constant buffer 322 * state for a particular destinaton surface. In all our rendering, 323 * these concepts are linked. 324 */ 325 void 326 renderer_bind_destination(struct xa_context *r, 327 struct pipe_surface *surface, int width, int height) 328 { 329 330 struct pipe_framebuffer_state fb; 331 struct pipe_viewport_state viewport; 332 333 /* Framebuffer uses actual surface width/height 334 */ 335 memset(&fb, 0, sizeof fb); 336 fb.width = surface->width; 337 fb.height = surface->height; 338 fb.nr_cbufs = 1; 339 fb.cbufs[0] = surface; 340 fb.zsbuf = 0; 341 342 /* Viewport just touches the bit we're interested in: 343 */ 344 viewport.scale[0] = width / 2.f; 345 viewport.scale[1] = height / 2.f; 346 viewport.scale[2] = 1.0; 347 viewport.scale[3] = 1.0; 348 viewport.translate[0] = width / 2.f; 349 viewport.translate[1] = height / 2.f; 350 viewport.translate[2] = 0.0; 351 viewport.translate[3] = 0.0; 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 } 371 372 void 373 renderer_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(r->pipe->screen, 382 PIPE_BIND_CONSTANT_BUFFER, PIPE_USAGE_STATIC, 383 param_bytes); 384 385 if (*cbuf) { 386 pipe_buffer_write(r->pipe, *cbuf, 0, param_bytes, params); 387 } 388 pipe_set_constant_buffer(r->pipe, shader_type, 0, *cbuf); 389 } 390 391 void 392 renderer_copy_prepare(struct xa_context *r, 393 struct pipe_surface *dst_surface, 394 struct pipe_resource *src_texture, 395 const enum xa_formats src_xa_format, 396 const enum xa_formats dst_xa_format) 397 { 398 struct pipe_context *pipe = r->pipe; 399 struct pipe_screen *screen = pipe->screen; 400 struct xa_shader shader; 401 uint32_t fs_traits = FS_COMPOSITE; 402 403 assert(screen->is_format_supported(screen, dst_surface->format, 404 PIPE_TEXTURE_2D, 0, 405 PIPE_BIND_RENDER_TARGET)); 406 (void)screen; 407 408 /* set misc state we care about */ 409 { 410 struct pipe_blend_state blend; 411 412 memset(&blend, 0, sizeof(blend)); 413 blend.rt[0].rgb_src_factor = PIPE_BLENDFACTOR_ONE; 414 blend.rt[0].alpha_src_factor = PIPE_BLENDFACTOR_ONE; 415 blend.rt[0].rgb_dst_factor = PIPE_BLENDFACTOR_ZERO; 416 blend.rt[0].alpha_dst_factor = PIPE_BLENDFACTOR_ZERO; 417 blend.rt[0].colormask = PIPE_MASK_RGBA; 418 cso_set_blend(r->cso, &blend); 419 } 420 421 /* sampler */ 422 { 423 struct pipe_sampler_state sampler; 424 425 memset(&sampler, 0, sizeof(sampler)); 426 sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; 427 sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; 428 sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; 429 sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; 430 sampler.min_img_filter = PIPE_TEX_FILTER_NEAREST; 431 sampler.mag_img_filter = PIPE_TEX_FILTER_NEAREST; 432 sampler.normalized_coords = 1; 433 cso_single_sampler(r->cso, PIPE_SHADER_FRAGMENT, 0, &sampler); 434 cso_single_sampler_done(r->cso, PIPE_SHADER_FRAGMENT); 435 } 436 437 renderer_bind_destination(r, dst_surface, 438 dst_surface->width, dst_surface->height); 439 440 /* texture/sampler view */ 441 { 442 struct pipe_sampler_view templ; 443 struct pipe_sampler_view *src_view; 444 445 u_sampler_view_default_template(&templ, 446 src_texture, src_texture->format); 447 src_view = pipe->create_sampler_view(pipe, src_texture, &templ); 448 cso_set_sampler_views(r->cso, PIPE_SHADER_FRAGMENT, 1, &src_view); 449 pipe_sampler_view_reference(&src_view, NULL); 450 } 451 452 /* shaders */ 453 if (src_texture->format == PIPE_FORMAT_L8_UNORM) 454 fs_traits |= FS_SRC_LUMINANCE; 455 if (dst_surface->format == PIPE_FORMAT_L8_UNORM) 456 fs_traits |= FS_DST_LUMINANCE; 457 if (xa_format_a(dst_xa_format) != 0 && 458 xa_format_a(src_xa_format) == 0) 459 fs_traits |= FS_SRC_SET_ALPHA; 460 461 shader = xa_shaders_get(r->shaders, VS_COMPOSITE, fs_traits); 462 cso_set_vertex_shader_handle(r->cso, shader.vs); 463 cso_set_fragment_shader_handle(r->cso, shader.fs); 464 465 r->buffer_size = 0; 466 r->attrs_per_vertex = 2; 467 } 468 469 void 470 renderer_copy(struct xa_context *r, 471 int dx, 472 int dy, 473 int sx, 474 int sy, 475 int width, int height, float src_width, float src_height) 476 { 477 float s0, t0, s1, t1; 478 float x0, y0, x1, y1; 479 480 /* XXX: could put the texcoord scaling calculation into the vertex 481 * shader. 482 */ 483 s0 = sx / src_width; 484 s1 = (sx + width) / src_width; 485 t0 = sy / src_height; 486 t1 = (sy + height) / src_height; 487 488 x0 = dx; 489 x1 = dx + width; 490 y0 = dy; 491 y1 = dy + height; 492 493 /* draw quad */ 494 renderer_draw_conditional(r, 4 * 8); 495 add_vertex_1tex(r, x0, y0, s0, t0); 496 add_vertex_1tex(r, x1, y0, s1, t0); 497 add_vertex_1tex(r, x1, y1, s1, t1); 498 add_vertex_1tex(r, x0, y1, s0, t1); 499 } 500 501 void 502 renderer_draw_yuv(struct xa_context *r, 503 float src_x, 504 float src_y, 505 float src_w, 506 float src_h, 507 int dst_x, 508 int dst_y, int dst_w, int dst_h, struct xa_surface *srf[]) 509 { 510 const int num_attribs = 2; /*pos + tex coord */ 511 512 setup_vertex_data_yuv(r, 513 src_x, src_y, src_w, src_h, 514 dst_x, dst_y, dst_w, dst_h, srf); 515 516 cso_set_vertex_elements(r->cso, num_attribs, r->velems); 517 util_draw_user_vertex_buffer(r->cso, r->buffer, PIPE_PRIM_QUADS, 518 4, /* verts */ 519 num_attribs); /* attribs/vert */ 520 r->buffer_size = 0; 521 } 522 523 void 524 renderer_begin_solid(struct xa_context *r) 525 { 526 r->buffer_size = 0; 527 r->attrs_per_vertex = 2; 528 } 529 530 void 531 renderer_solid(struct xa_context *r, 532 int x0, int y0, int x1, int y1, float *color) 533 { 534 /* 535 * debug_printf("solid rect[(%d, %d), (%d, %d)], rgba[%f, %f, %f, %f]\n", 536 * x0, y0, x1, y1, color[0], color[1], color[2], color[3]); */ 537 538 renderer_draw_conditional(r, 4 * 8); 539 540 /* 1st vertex */ 541 add_vertex_color(r, x0, y0, color); 542 /* 2nd vertex */ 543 add_vertex_color(r, x1, y0, color); 544 /* 3rd vertex */ 545 add_vertex_color(r, x1, y1, color); 546 /* 4th vertex */ 547 add_vertex_color(r, x0, y1, color); 548 } 549 550 void 551 renderer_draw_flush(struct xa_context *r) 552 { 553 renderer_draw_conditional(r, 0); 554 } 555 556 void 557 renderer_begin_textures(struct xa_context *r) 558 { 559 r->attrs_per_vertex = 1 + r->num_bound_samplers; 560 r->buffer_size = 0; 561 } 562 563 void 564 renderer_texture(struct xa_context *r, 565 int *pos, 566 int width, int height, 567 const float *src_matrix, 568 const float *mask_matrix) 569 { 570 struct pipe_sampler_view **sampler_view = r->bound_sampler_views; 571 572 #if 0 573 if (src_matrix) { 574 debug_printf("src_matrix = \n"); 575 debug_printf("%f, %f, %f\n", src_matrix[0], src_matrix[1], src_matrix[2]); 576 debug_printf("%f, %f, %f\n", src_matrix[3], src_matrix[4], src_matrix[5]); 577 debug_printf("%f, %f, %f\n", src_matrix[6], src_matrix[7], src_matrix[8]); 578 } 579 if (mask_matrix) { 580 debug_printf("mask_matrix = \n"); 581 debug_printf("%f, %f, %f\n", mask_matrix[0], mask_matrix[1], mask_matrix[2]); 582 debug_printf("%f, %f, %f\n", mask_matrix[3], mask_matrix[4], mask_matrix[5]); 583 debug_printf("%f, %f, %f\n", mask_matrix[6], mask_matrix[7], mask_matrix[8]); 584 } 585 #endif 586 587 switch(r->attrs_per_vertex) { 588 case 2: 589 renderer_draw_conditional(r, 4 * 8); 590 add_vertex_data1(r, 591 pos[0], pos[1], /* src */ 592 pos[4], pos[5], /* dst */ 593 width, height, 594 sampler_view[0]->texture, src_matrix); 595 break; 596 case 3: 597 renderer_draw_conditional(r, 4 * 12); 598 add_vertex_data2(r, 599 pos[0], pos[1], /* src */ 600 pos[2], pos[3], /* mask */ 601 pos[4], pos[5], /* dst */ 602 width, height, 603 sampler_view[0]->texture, sampler_view[1]->texture, 604 src_matrix, mask_matrix); 605 break; 606 default: 607 break; 608 } 609 } 610
