i965: Drop i915 register/instruction definitions.
[mesa.git] / src / mesa / drivers / dri / i965 / intel_context.h
1 /**************************************************************************
2 *
3 * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
4 * All Rights Reserved.
5 *
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:
13 *
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
16 * of the Software.
17 *
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 TUNGSTEN GRAPHICS 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.
25 *
26 **************************************************************************/
27
28 #ifndef INTELCONTEXT_INC
29 #define INTELCONTEXT_INC
30
31
32 #include <stdbool.h>
33 #include <string.h>
34 #include "main/mtypes.h"
35 #include "main/mm.h"
36
37 #ifdef __cplusplus
38 extern "C" {
39 /* Evil hack for using libdrm in a c++ compiler. */
40 #define virtual virt
41 #endif
42
43 #include "drm.h"
44 #include "intel_bufmgr.h"
45
46 #include "intel_screen.h"
47 #include "intel_tex_obj.h"
48 #include "i915_drm.h"
49
50 #ifdef __cplusplus
51 #undef virtual
52 #endif
53
54 #include "tnl/t_vertex.h"
55
56 #define TAG(x) intel##x
57 #include "tnl_dd/t_dd_vertex.h"
58 #undef TAG
59
60 struct intel_region;
61 struct intel_context;
62
63 typedef void (*intel_tri_func) (struct intel_context *, intelVertex *,
64 intelVertex *, intelVertex *);
65 typedef void (*intel_line_func) (struct intel_context *, intelVertex *,
66 intelVertex *);
67 typedef void (*intel_point_func) (struct intel_context *, intelVertex *);
68
69 #define INTEL_WRITE_PART 0x1
70 #define INTEL_WRITE_FULL 0x2
71 #define INTEL_READ 0x4
72
73 #define INTEL_MAX_FIXUP 64
74
75 #ifndef likely
76 #ifdef __GNUC__
77 #define likely(expr) (__builtin_expect(expr, 1))
78 #define unlikely(expr) (__builtin_expect(expr, 0))
79 #else
80 #define likely(expr) (expr)
81 #define unlikely(expr) (expr)
82 #endif
83 #endif
84
85 struct intel_sync_object {
86 struct gl_sync_object Base;
87
88 /** Batch associated with this sync object */
89 drm_intel_bo *bo;
90 };
91
92 struct brw_context;
93
94 struct intel_batchbuffer {
95 /** Current batchbuffer being queued up. */
96 drm_intel_bo *bo;
97 /** Last BO submitted to the hardware. Used for glFinish(). */
98 drm_intel_bo *last_bo;
99 /** BO for post-sync nonzero writes for gen6 workaround. */
100 drm_intel_bo *workaround_bo;
101 bool need_workaround_flush;
102
103 struct cached_batch_item *cached_items;
104
105 uint16_t emit, total;
106 uint16_t used, reserved_space;
107 uint32_t *map;
108 uint32_t *cpu_map;
109 #define BATCH_SZ (8192*sizeof(uint32_t))
110
111 uint32_t state_batch_offset;
112 bool is_blit;
113 bool needs_sol_reset;
114
115 struct {
116 uint16_t used;
117 int reloc_count;
118 } saved;
119 };
120
121 /**
122 * intel_context is derived from Mesa's context class: struct gl_context.
123 */
124 struct intel_context
125 {
126 struct gl_context ctx; /**< base class, must be first field */
127
128 struct
129 {
130 void (*destroy) (struct intel_context * intel);
131 void (*finish_batch) (struct intel_context * intel);
132 void (*new_batch) (struct intel_context * intel);
133
134 void (*invalidate_state) (struct intel_context *intel,
135 GLuint new_state);
136
137 void (*debug_batch)(struct intel_context *intel);
138 void (*annotate_aub)(struct intel_context *intel);
139 bool (*render_target_supported)(struct intel_context *intel,
140 struct gl_renderbuffer *rb);
141
142 /** Can HiZ be enabled on a depthbuffer of the given format? */
143 bool (*is_hiz_depth_format)(struct intel_context *intel,
144 gl_format format);
145
146 void (*update_texture_surface)(struct gl_context *ctx,
147 unsigned unit,
148 uint32_t *binding_table,
149 unsigned surf_index);
150 void (*update_renderbuffer_surface)(struct brw_context *brw,
151 struct gl_renderbuffer *rb,
152 bool layered,
153 unsigned unit);
154 void (*update_null_renderbuffer_surface)(struct brw_context *brw,
155 unsigned unit);
156 void (*create_constant_surface)(struct brw_context *brw,
157 drm_intel_bo *bo,
158 uint32_t offset,
159 uint32_t size,
160 uint32_t *out_offset,
161 bool dword_pitch);
162
163 /**
164 * Send the appropriate state packets to configure depth, stencil, and
165 * HiZ buffers (i965+ only)
166 */
167 void (*emit_depth_stencil_hiz)(struct brw_context *brw,
168 struct intel_mipmap_tree *depth_mt,
169 uint32_t depth_offset,
170 uint32_t depthbuffer_format,
171 uint32_t depth_surface_type,
172 struct intel_mipmap_tree *stencil_mt,
173 bool hiz, bool separate_stencil,
174 uint32_t width, uint32_t height,
175 uint32_t tile_x, uint32_t tile_y);
176
177 } vtbl;
178
179 GLuint NewGLState;
180
181 dri_bufmgr *bufmgr;
182 unsigned int maxBatchSize;
183
184 /**
185 * Generation number of the hardware: 2 is 8xx, 3 is 9xx pre-965, 4 is 965.
186 */
187 int gen;
188 int gt;
189 bool needs_ff_sync;
190 bool is_haswell;
191 bool is_baytrail;
192 bool is_g4x;
193 bool is_945;
194 bool has_separate_stencil;
195 bool must_use_separate_stencil;
196 bool has_hiz;
197 bool has_llc;
198 bool has_swizzling;
199
200 int urb_size;
201
202 drm_intel_context *hw_ctx;
203
204 struct intel_batchbuffer batch;
205
206 drm_intel_bo *first_post_swapbuffers_batch;
207 bool need_throttle;
208 bool no_batch_wrap;
209
210 /**
211 * Set if we're either a debug context or the INTEL_DEBUG=perf environment
212 * variable is set, this is the flag indicating to do expensive work that
213 * might lead to a perf_debug() call.
214 */
215 bool perf_debug;
216
217 struct {
218 drm_intel_bo *bo;
219 GLuint offset;
220 uint32_t buffer_len;
221 uint32_t buffer_offset;
222 char buffer[4096];
223 } upload;
224
225 uint32_t max_gtt_map_object_size;
226
227 GLuint stats_wm;
228
229 bool hw_stencil;
230 bool hw_stipple;
231 bool no_rast;
232 bool always_flush_batch;
233 bool always_flush_cache;
234 bool disable_throttling;
235
236 GLenum reduced_primitive;
237
238 /**
239 * Set if rendering has occured to the drawable's front buffer.
240 *
241 * This is used in the DRI2 case to detect that glFlush should also copy
242 * the contents of the fake front buffer to the real front buffer.
243 */
244 bool front_buffer_dirty;
245
246 /**
247 * Track whether front-buffer rendering is currently enabled
248 *
249 * A separate flag is used to track this in order to support MRT more
250 * easily.
251 */
252 bool is_front_buffer_rendering;
253 /**
254 * Track whether front-buffer is the current read target.
255 *
256 * This is closely associated with is_front_buffer_rendering, but may
257 * be set separately. The DRI2 fake front buffer must be referenced
258 * either way.
259 */
260 bool is_front_buffer_reading;
261
262 bool use_early_z;
263
264 int driFd;
265
266 __DRIcontext *driContext;
267 struct intel_screen *intelScreen;
268 void (*saved_viewport)(struct gl_context * ctx,
269 GLint x, GLint y, GLsizei width, GLsizei height);
270
271 /**
272 * Configuration cache
273 */
274 driOptionCache optionCache;
275 };
276
277 #define SUBPIXEL_X 0.125
278 #define SUBPIXEL_Y 0.125
279
280 /**
281 * Align a value down to an alignment value
282 *
283 * If \c value is not already aligned to the requested alignment value, it
284 * will be rounded down.
285 *
286 * \param value Value to be rounded
287 * \param alignment Alignment value to be used. This must be a power of two.
288 *
289 * \sa ALIGN()
290 */
291 #define ROUND_DOWN_TO(value, alignment) ((value) & ~(alignment - 1))
292
293 static INLINE uint32_t
294 U_FIXED(float value, uint32_t frac_bits)
295 {
296 value *= (1 << frac_bits);
297 return value < 0 ? 0 : value;
298 }
299
300 static INLINE uint32_t
301 S_FIXED(float value, uint32_t frac_bits)
302 {
303 return value * (1 << frac_bits);
304 }
305
306 /* ================================================================
307 * From linux kernel i386 header files, copes with odd sizes better
308 * than COPY_DWORDS would:
309 * XXX Put this in src/mesa/main/imports.h ???
310 */
311 #if defined(i386) || defined(__i386__)
312 static INLINE void * __memcpy(void * to, const void * from, size_t n)
313 {
314 int d0, d1, d2;
315 __asm__ __volatile__(
316 "rep ; movsl\n\t"
317 "testb $2,%b4\n\t"
318 "je 1f\n\t"
319 "movsw\n"
320 "1:\ttestb $1,%b4\n\t"
321 "je 2f\n\t"
322 "movsb\n"
323 "2:"
324 : "=&c" (d0), "=&D" (d1), "=&S" (d2)
325 :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
326 : "memory");
327 return (to);
328 }
329 #else
330 #define __memcpy(a,b,c) memcpy(a,b,c)
331 #endif
332
333
334 /* ================================================================
335 * Debugging:
336 */
337 extern int INTEL_DEBUG;
338
339 #define DEBUG_TEXTURE 0x1
340 #define DEBUG_STATE 0x2
341 #define DEBUG_IOCTL 0x4
342 #define DEBUG_BLIT 0x8
343 #define DEBUG_MIPTREE 0x10
344 #define DEBUG_PERF 0x20
345 #define DEBUG_BATCH 0x80
346 #define DEBUG_PIXEL 0x100
347 #define DEBUG_BUFMGR 0x200
348 #define DEBUG_REGION 0x400
349 #define DEBUG_FBO 0x800
350 #define DEBUG_GS 0x1000
351 #define DEBUG_SYNC 0x2000
352 #define DEBUG_PRIMS 0x4000
353 #define DEBUG_VERTS 0x8000
354 #define DEBUG_DRI 0x10000
355 #define DEBUG_SF 0x20000
356 #define DEBUG_STATS 0x100000
357 #define DEBUG_WM 0x400000
358 #define DEBUG_URB 0x800000
359 #define DEBUG_VS 0x1000000
360 #define DEBUG_CLIP 0x2000000
361 #define DEBUG_AUB 0x4000000
362 #define DEBUG_SHADER_TIME 0x8000000
363 #define DEBUG_BLORP 0x10000000
364 #define DEBUG_NO16 0x20000000
365
366 #ifdef HAVE_ANDROID_PLATFORM
367 #define LOG_TAG "INTEL-MESA"
368 #include <cutils/log.h>
369 #ifndef ALOGW
370 #define ALOGW LOGW
371 #endif
372 #define dbg_printf(...) ALOGW(__VA_ARGS__)
373 #else
374 #define dbg_printf(...) printf(__VA_ARGS__)
375 #endif /* HAVE_ANDROID_PLATFORM */
376
377 #define DBG(...) do { \
378 if (unlikely(INTEL_DEBUG & FILE_DEBUG_FLAG)) \
379 dbg_printf(__VA_ARGS__); \
380 } while(0)
381
382 #define perf_debug(...) do { \
383 static GLuint msg_id = 0; \
384 if (unlikely(INTEL_DEBUG & DEBUG_PERF)) \
385 dbg_printf(__VA_ARGS__); \
386 if (intel->perf_debug) \
387 _mesa_gl_debug(&intel->ctx, &msg_id, \
388 MESA_DEBUG_TYPE_PERFORMANCE, \
389 MESA_DEBUG_SEVERITY_MEDIUM, \
390 __VA_ARGS__); \
391 } while(0)
392
393 #define WARN_ONCE(cond, fmt...) do { \
394 if (unlikely(cond)) { \
395 static bool _warned = false; \
396 static GLuint msg_id = 0; \
397 if (!_warned) { \
398 fprintf(stderr, "WARNING: "); \
399 fprintf(stderr, fmt); \
400 _warned = true; \
401 \
402 _mesa_gl_debug(ctx, &msg_id, \
403 MESA_DEBUG_TYPE_OTHER, \
404 MESA_DEBUG_SEVERITY_HIGH, fmt); \
405 } \
406 } \
407 } while (0)
408
409 #define PCI_CHIP_845_G 0x2562
410 #define PCI_CHIP_I830_M 0x3577
411 #define PCI_CHIP_I855_GM 0x3582
412 #define PCI_CHIP_I865_G 0x2572
413 #define PCI_CHIP_I915_G 0x2582
414 #define PCI_CHIP_I915_GM 0x2592
415 #define PCI_CHIP_I945_G 0x2772
416 #define PCI_CHIP_I945_GM 0x27A2
417 #define PCI_CHIP_I945_GME 0x27AE
418 #define PCI_CHIP_G33_G 0x29C2
419 #define PCI_CHIP_Q35_G 0x29B2
420 #define PCI_CHIP_Q33_G 0x29D2
421
422
423 /* ================================================================
424 * intel_context.c:
425 */
426
427 extern bool intelInitContext(struct intel_context *intel,
428 int api,
429 unsigned major_version,
430 unsigned minor_version,
431 const struct gl_config * mesaVis,
432 __DRIcontext * driContextPriv,
433 void *sharedContextPrivate,
434 struct dd_function_table *functions,
435 unsigned *dri_ctx_error);
436
437 extern void intelFinish(struct gl_context * ctx);
438 extern void _intel_flush(struct gl_context * ctx, const char *file, int line);
439
440 #define intel_flush(ctx) _intel_flush(ctx, __FILE__, __LINE__)
441
442 extern void intelInitDriverFunctions(struct dd_function_table *functions);
443
444 void intel_init_syncobj_functions(struct dd_function_table *functions);
445
446
447 /* ================================================================
448 * intel_state.c:
449 */
450
451 #define COMPAREFUNC_ALWAYS 0
452 #define COMPAREFUNC_NEVER 0x1
453 #define COMPAREFUNC_LESS 0x2
454 #define COMPAREFUNC_EQUAL 0x3
455 #define COMPAREFUNC_LEQUAL 0x4
456 #define COMPAREFUNC_GREATER 0x5
457 #define COMPAREFUNC_NOTEQUAL 0x6
458 #define COMPAREFUNC_GEQUAL 0x7
459
460 #define STENCILOP_KEEP 0
461 #define STENCILOP_ZERO 0x1
462 #define STENCILOP_REPLACE 0x2
463 #define STENCILOP_INCRSAT 0x3
464 #define STENCILOP_DECRSAT 0x4
465 #define STENCILOP_INCR 0x5
466 #define STENCILOP_DECR 0x6
467 #define STENCILOP_INVERT 0x7
468
469 #define LOGICOP_CLEAR 0
470 #define LOGICOP_NOR 0x1
471 #define LOGICOP_AND_INV 0x2
472 #define LOGICOP_COPY_INV 0x3
473 #define LOGICOP_AND_RVRSE 0x4
474 #define LOGICOP_INV 0x5
475 #define LOGICOP_XOR 0x6
476 #define LOGICOP_NAND 0x7
477 #define LOGICOP_AND 0x8
478 #define LOGICOP_EQUIV 0x9
479 #define LOGICOP_NOOP 0xa
480 #define LOGICOP_OR_INV 0xb
481 #define LOGICOP_COPY 0xc
482 #define LOGICOP_OR_RVRSE 0xd
483 #define LOGICOP_OR 0xe
484 #define LOGICOP_SET 0xf
485
486 #define BLENDFACT_ZERO 0x01
487 #define BLENDFACT_ONE 0x02
488 #define BLENDFACT_SRC_COLR 0x03
489 #define BLENDFACT_INV_SRC_COLR 0x04
490 #define BLENDFACT_SRC_ALPHA 0x05
491 #define BLENDFACT_INV_SRC_ALPHA 0x06
492 #define BLENDFACT_DST_ALPHA 0x07
493 #define BLENDFACT_INV_DST_ALPHA 0x08
494 #define BLENDFACT_DST_COLR 0x09
495 #define BLENDFACT_INV_DST_COLR 0x0a
496 #define BLENDFACT_SRC_ALPHA_SATURATE 0x0b
497 #define BLENDFACT_CONST_COLOR 0x0c
498 #define BLENDFACT_INV_CONST_COLOR 0x0d
499 #define BLENDFACT_CONST_ALPHA 0x0e
500 #define BLENDFACT_INV_CONST_ALPHA 0x0f
501 #define BLENDFACT_MASK 0x0f
502
503 enum {
504 DRI_CONF_BO_REUSE_DISABLED,
505 DRI_CONF_BO_REUSE_ALL
506 };
507
508 extern int intel_translate_shadow_compare_func(GLenum func);
509 extern int intel_translate_compare_func(GLenum func);
510 extern int intel_translate_stencil_op(GLenum op);
511 extern int intel_translate_logic_op(GLenum opcode);
512
513 void intel_update_renderbuffers(__DRIcontext *context,
514 __DRIdrawable *drawable);
515 void intel_prepare_render(struct intel_context *intel);
516
517 void
518 intel_resolve_for_dri2_flush(struct intel_context *intel,
519 __DRIdrawable *drawable);
520
521 void i915_set_buf_info_for_region(uint32_t *state, struct intel_region *region,
522 uint32_t buffer_id);
523 void intel_init_texture_formats(struct gl_context *ctx);
524
525 /*======================================================================
526 * Inline conversion functions.
527 * These are better-typed than the macros used previously:
528 */
529 static INLINE struct intel_context *
530 intel_context(struct gl_context * ctx)
531 {
532 return (struct intel_context *) ctx;
533 }
534
535 static INLINE bool
536 is_power_of_two(uint32_t value)
537 {
538 return (value & (value - 1)) == 0;
539 }
540
541 #ifdef __cplusplus
542 }
543 #endif
544
545 #endif