i965: Avoid flushing the batch for every blorp op.
[mesa.git] / src / mesa / drivers / dri / i965 / brw_blorp.h
1 /*
2 * Copyright © 2012 Intel Corporation
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
21 * IN THE SOFTWARE.
22 */
23
24 #pragma once
25
26 #include <stdint.h>
27
28 #include "brw_context.h"
29 #include "intel_mipmap_tree.h"
30
31 struct brw_context;
32
33 #ifdef __cplusplus
34 extern "C" {
35 #endif
36
37 void
38 brw_blorp_blit_miptrees(struct brw_context *brw,
39 struct intel_mipmap_tree *src_mt,
40 unsigned src_level, unsigned src_layer,
41 struct intel_mipmap_tree *dst_mt,
42 unsigned dst_level, unsigned dst_layer,
43 float src_x0, float src_y0,
44 float src_x1, float src_y1,
45 float dst_x0, float dst_y0,
46 float dst_x1, float dst_y1,
47 GLenum filter, bool mirror_x, bool mirror_y);
48
49 bool
50 brw_blorp_clear_color(struct brw_context *brw, struct gl_framebuffer *fb,
51 bool partial_clear);
52
53 void
54 brw_blorp_resolve_color(struct brw_context *brw,
55 struct intel_mipmap_tree *mt);
56
57 #ifdef __cplusplus
58 } /* end extern "C" */
59
60 /**
61 * Binding table indices used by BLORP.
62 */
63 enum {
64 BRW_BLORP_TEXTURE_BINDING_TABLE_INDEX,
65 BRW_BLORP_RENDERBUFFER_BINDING_TABLE_INDEX,
66 BRW_BLORP_NUM_BINDING_TABLE_ENTRIES
67 };
68
69
70 class brw_blorp_mip_info
71 {
72 public:
73 brw_blorp_mip_info();
74
75 void set(struct intel_mipmap_tree *mt,
76 unsigned int level, unsigned int layer);
77
78 struct intel_mipmap_tree *mt;
79
80 /**
81 * The miplevel to use.
82 */
83 uint32_t level;
84
85 /**
86 * The 2D layer within the miplevel. Combined, level and layer define the
87 * 2D miptree slice to use.
88 */
89 uint32_t layer;
90
91 /**
92 * Width of the miplevel to be used. For surfaces using
93 * INTEL_MSAA_LAYOUT_IMS, this is measured in samples, not pixels.
94 */
95 uint32_t width;
96
97 /**
98 * Height of the miplevel to be used. For surfaces using
99 * INTEL_MSAA_LAYOUT_IMS, this is measured in samples, not pixels.
100 */
101 uint32_t height;
102
103 /**
104 * X offset within the surface to texture from (or render to). For
105 * surfaces using INTEL_MSAA_LAYOUT_IMS, this is measured in samples, not
106 * pixels.
107 */
108 uint32_t x_offset;
109
110 /**
111 * Y offset within the surface to texture from (or render to). For
112 * surfaces using INTEL_MSAA_LAYOUT_IMS, this is measured in samples, not
113 * pixels.
114 */
115 uint32_t y_offset;
116 };
117
118 class brw_blorp_surface_info : public brw_blorp_mip_info
119 {
120 public:
121 brw_blorp_surface_info();
122
123 void set(struct brw_context *brw,
124 struct intel_mipmap_tree *mt,
125 unsigned int level, unsigned int layer,
126 bool is_render_target);
127
128 uint32_t compute_tile_offsets(uint32_t *tile_x, uint32_t *tile_y) const;
129
130 /* Setting this flag indicates that the buffer's contents are W-tiled
131 * stencil data, but the surface state should be set up for Y tiled
132 * MESA_FORMAT_R8 data (this is necessary because surface states don't
133 * support W tiling).
134 *
135 * Since W tiles are 64 pixels wide by 64 pixels high, whereas Y tiles of
136 * MESA_FORMAT_R8 data are 128 pixels wide by 32 pixels high, the width and
137 * pitch stored in the surface state will be multiplied by 2, and the
138 * height will be halved. Also, since W and Y tiles store their data in a
139 * different order, the width and height will be rounded up to a multiple
140 * of the tile size, to ensure that the WM program can access the full
141 * width and height of the buffer.
142 */
143 bool map_stencil_as_y_tiled;
144
145 unsigned num_samples;
146
147 /* Setting this flag indicates that the surface should be set up in
148 * ARYSPC_LOD0 mode. Ignored prior to Gen7.
149 */
150 bool array_spacing_lod0;
151
152 /**
153 * Format that should be used when setting up the surface state for this
154 * surface. Should correspond to one of the BRW_SURFACEFORMAT_* enums.
155 */
156 uint32_t brw_surfaceformat;
157
158 /**
159 * For MSAA surfaces, MSAA layout that should be used when setting up the
160 * surface state for this surface.
161 */
162 intel_msaa_layout msaa_layout;
163 };
164
165
166 struct brw_blorp_coord_transform_params
167 {
168 void setup(GLfloat src0, GLfloat src1, GLfloat dst0, GLfloat dst1,
169 bool mirror);
170
171 float multiplier;
172 float offset;
173 };
174
175
176 struct brw_blorp_wm_push_constants
177 {
178 uint32_t dst_x0;
179 uint32_t dst_x1;
180 uint32_t dst_y0;
181 uint32_t dst_y1;
182 /* Top right coordinates of the rectangular grid used for scaled blitting */
183 float rect_grid_x1;
184 float rect_grid_y1;
185 brw_blorp_coord_transform_params x_transform;
186 brw_blorp_coord_transform_params y_transform;
187 /* Pad out to an integral number of registers */
188 uint32_t pad[6];
189 };
190
191 /* Every 32 bytes of push constant data constitutes one GEN register. */
192 const unsigned int BRW_BLORP_NUM_PUSH_CONST_REGS =
193 sizeof(brw_blorp_wm_push_constants) / 32;
194
195 struct brw_blorp_prog_data
196 {
197 unsigned int first_curbe_grf;
198
199 /**
200 * True if the WM program should be run in MSDISPMODE_PERSAMPLE with more
201 * than one sample per pixel.
202 */
203 bool persample_msaa_dispatch;
204 };
205
206
207 enum gen7_fast_clear_op {
208 GEN7_FAST_CLEAR_OP_NONE,
209 GEN7_FAST_CLEAR_OP_FAST_CLEAR,
210 GEN7_FAST_CLEAR_OP_RESOLVE,
211 };
212
213
214 class brw_blorp_params
215 {
216 public:
217 brw_blorp_params();
218
219 virtual uint32_t get_wm_prog(struct brw_context *brw,
220 brw_blorp_prog_data **prog_data) const = 0;
221
222 uint32_t x0;
223 uint32_t y0;
224 uint32_t x1;
225 uint32_t y1;
226 brw_blorp_mip_info depth;
227 uint32_t depth_format;
228 brw_blorp_surface_info src;
229 brw_blorp_surface_info dst;
230 enum gen6_hiz_op hiz_op;
231 enum gen7_fast_clear_op fast_clear_op;
232 unsigned num_samples;
233 bool use_wm_prog;
234 brw_blorp_wm_push_constants wm_push_consts;
235 bool color_write_disable[4];
236 };
237
238
239 void
240 brw_blorp_exec(struct brw_context *brw, const brw_blorp_params *params);
241
242
243 /**
244 * Parameters for a HiZ or depth resolve operation.
245 *
246 * For an overview of HiZ ops, see the following sections of the Sandy Bridge
247 * PRM, Volume 1, Part 2:
248 * - 7.5.3.1 Depth Buffer Clear
249 * - 7.5.3.2 Depth Buffer Resolve
250 * - 7.5.3.3 Hierarchical Depth Buffer Resolve
251 */
252 class brw_hiz_op_params : public brw_blorp_params
253 {
254 public:
255 brw_hiz_op_params(struct intel_mipmap_tree *mt,
256 unsigned int level, unsigned int layer,
257 gen6_hiz_op op);
258
259 virtual uint32_t get_wm_prog(struct brw_context *brw,
260 brw_blorp_prog_data **prog_data) const;
261 };
262
263 struct brw_blorp_blit_prog_key
264 {
265 /* Number of samples per pixel that have been configured in the surface
266 * state for texturing from.
267 */
268 unsigned tex_samples;
269
270 /* MSAA layout that has been configured in the surface state for texturing
271 * from.
272 */
273 intel_msaa_layout tex_layout;
274
275 /* Actual number of samples per pixel in the source image. */
276 unsigned src_samples;
277
278 /* Actual MSAA layout used by the source image. */
279 intel_msaa_layout src_layout;
280
281 /* Number of samples per pixel that have been configured in the render
282 * target.
283 */
284 unsigned rt_samples;
285
286 /* MSAA layout that has been configured in the render target. */
287 intel_msaa_layout rt_layout;
288
289 /* Actual number of samples per pixel in the destination image. */
290 unsigned dst_samples;
291
292 /* Actual MSAA layout used by the destination image. */
293 intel_msaa_layout dst_layout;
294
295 /* Type of the data to be read from the texture (one of
296 * BRW_REGISTER_TYPE_{UD,D,F}).
297 */
298 unsigned texture_data_type;
299
300 /* True if the source image is W tiled. If true, the surface state for the
301 * source image must be configured as Y tiled, and tex_samples must be 0.
302 */
303 bool src_tiled_w;
304
305 /* True if the destination image is W tiled. If true, the surface state
306 * for the render target must be configured as Y tiled, and rt_samples must
307 * be 0.
308 */
309 bool dst_tiled_w;
310
311 /* True if all source samples should be blended together to produce each
312 * destination pixel. If true, src_tiled_w must be false, tex_samples must
313 * equal src_samples, and tex_samples must be nonzero.
314 */
315 bool blend;
316
317 /* True if the rectangle being sent through the rendering pipeline might be
318 * larger than the destination rectangle, so the WM program should kill any
319 * pixels that are outside the destination rectangle.
320 */
321 bool use_kill;
322
323 /**
324 * True if the WM program should be run in MSDISPMODE_PERSAMPLE with more
325 * than one sample per pixel.
326 */
327 bool persample_msaa_dispatch;
328
329 /* True for scaled blitting. */
330 bool blit_scaled;
331
332 /* Scale factors between the pixel grid and the grid of samples. We're
333 * using grid of samples for bilinear filetring in multisample scaled blits.
334 */
335 float x_scale;
336 float y_scale;
337
338 /* True for blits with filter = GL_LINEAR. */
339 bool bilinear_filter;
340 };
341
342 class brw_blorp_blit_params : public brw_blorp_params
343 {
344 public:
345 brw_blorp_blit_params(struct brw_context *brw,
346 struct intel_mipmap_tree *src_mt,
347 unsigned src_level, unsigned src_layer,
348 struct intel_mipmap_tree *dst_mt,
349 unsigned dst_level, unsigned dst_layer,
350 GLfloat src_x0, GLfloat src_y0,
351 GLfloat src_x1, GLfloat src_y1,
352 GLfloat dst_x0, GLfloat dst_y0,
353 GLfloat dst_x1, GLfloat dst_y1,
354 GLenum filter, bool mirror_x, bool mirror_y);
355
356 virtual uint32_t get_wm_prog(struct brw_context *brw,
357 brw_blorp_prog_data **prog_data) const;
358
359 private:
360 brw_blorp_blit_prog_key wm_prog_key;
361 };
362
363 /**
364 * \name BLORP internals
365 * \{
366 *
367 * Used internally by gen6_blorp_exec() and gen7_blorp_exec().
368 */
369
370 void
371 gen6_blorp_init(struct brw_context *brw);
372
373 void
374 gen6_blorp_emit_state_base_address(struct brw_context *brw,
375 const brw_blorp_params *params);
376
377 void
378 gen6_blorp_emit_vertices(struct brw_context *brw,
379 const brw_blorp_params *params);
380
381 uint32_t
382 gen6_blorp_emit_blend_state(struct brw_context *brw,
383 const brw_blorp_params *params);
384
385 uint32_t
386 gen6_blorp_emit_cc_state(struct brw_context *brw,
387 const brw_blorp_params *params);
388
389 uint32_t
390 gen6_blorp_emit_wm_constants(struct brw_context *brw,
391 const brw_blorp_params *params);
392
393 void
394 gen6_blorp_emit_vs_disable(struct brw_context *brw,
395 const brw_blorp_params *params);
396
397 uint32_t
398 gen6_blorp_emit_binding_table(struct brw_context *brw,
399 const brw_blorp_params *params,
400 uint32_t wm_surf_offset_renderbuffer,
401 uint32_t wm_surf_offset_texture);
402
403 uint32_t
404 gen6_blorp_emit_depth_stencil_state(struct brw_context *brw,
405 const brw_blorp_params *params);
406
407 void
408 gen6_blorp_emit_gs_disable(struct brw_context *brw,
409 const brw_blorp_params *params);
410
411 void
412 gen6_blorp_emit_clip_disable(struct brw_context *brw,
413 const brw_blorp_params *params);
414
415 void
416 gen6_blorp_emit_drawing_rectangle(struct brw_context *brw,
417 const brw_blorp_params *params);
418 /** \} */
419
420 #endif /* __cplusplus */