intel_context.h revision 9559ca600dde0877fe0abd04dd789bd5a3cdfbde 
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 "main/mtypes.h"
34#include "main/mm.h"
35
36#ifdef __cplusplus
37extern "C" {
38	/* Evil hack for using libdrm in a c++ compiler. */
39	#define virtual virt
40#endif
41
42#include "drm.h"
43#include "intel_bufmgr.h"
44
45#include "intel_screen.h"
46#include "intel_tex_obj.h"
47#include "i915_drm.h"
48
49#ifdef __cplusplus
50	#undef virtual
51}
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#define DV_PF_555  (1<<8)
61#define DV_PF_565  (2<<8)
62#define DV_PF_8888 (3<<8)
63#define DV_PF_4444 (8<<8)
64#define DV_PF_1555 (9<<8)
65
66struct intel_region;
67struct intel_context;
68
69typedef void (*intel_tri_func) (struct intel_context *, intelVertex *,
70                                intelVertex *, intelVertex *);
71typedef void (*intel_line_func) (struct intel_context *, intelVertex *,
72                                 intelVertex *);
73typedef void (*intel_point_func) (struct intel_context *, intelVertex *);
74
75/**
76 * Bits for intel->Fallback field
77 */
78/*@{*/
79#define INTEL_FALLBACK_DRAW_BUFFER	 0x1
80#define INTEL_FALLBACK_READ_BUFFER	 0x2
81#define INTEL_FALLBACK_DEPTH_BUFFER      0x4
82#define INTEL_FALLBACK_STENCIL_BUFFER    0x8
83#define INTEL_FALLBACK_USER		 0x10
84#define INTEL_FALLBACK_RENDERMODE	 0x20
85#define INTEL_FALLBACK_TEXTURE   	 0x40
86#define INTEL_FALLBACK_DRIVER            0x1000  /**< first for drivers */
87/*@}*/
88
89extern void intelFallback(struct intel_context *intel, GLbitfield bit,
90                          GLboolean mode);
91#define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
92
93
94#define INTEL_WRITE_PART  0x1
95#define INTEL_WRITE_FULL  0x2
96#define INTEL_READ        0x4
97
98#define INTEL_MAX_FIXUP 64
99
100#ifndef likely
101#ifdef __GNUC__
102#define likely(expr) (__builtin_expect(expr, 1))
103#define unlikely(expr) (__builtin_expect(expr, 0))
104#else
105#define likely(expr) (expr)
106#define unlikely(expr) (expr)
107#endif
108#endif
109
110struct intel_sync_object {
111   struct gl_sync_object Base;
112
113   /** Batch associated with this sync object */
114   drm_intel_bo *bo;
115};
116
117/**
118 * intel_context is derived from Mesa's context class: struct gl_context.
119 */
120struct intel_context
121{
122   struct gl_context ctx;  /**< base class, must be first field */
123
124   struct
125   {
126      void (*destroy) (struct intel_context * intel);
127      void (*emit_state) (struct intel_context * intel);
128      void (*finish_batch) (struct intel_context * intel);
129      void (*new_batch) (struct intel_context * intel);
130      void (*emit_invarient_state) (struct intel_context * intel);
131      void (*update_texture_state) (struct intel_context * intel);
132
133      void (*render_start) (struct intel_context * intel);
134      void (*render_prevalidate) (struct intel_context * intel);
135      void (*set_draw_region) (struct intel_context * intel,
136                               struct intel_region * draw_regions[],
137                               struct intel_region * depth_region,
138			       GLuint num_regions);
139      void (*update_draw_buffer)(struct intel_context *intel);
140
141      void (*reduced_primitive_state) (struct intel_context * intel,
142                                       GLenum rprim);
143
144      GLboolean (*check_vertex_size) (struct intel_context * intel,
145				      GLuint expected);
146      void (*invalidate_state) (struct intel_context *intel,
147				GLuint new_state);
148
149      void (*assert_not_dirty) (struct intel_context *intel);
150
151      void (*debug_batch)(struct intel_context *intel);
152      bool (*render_target_supported)(gl_format format);
153
154      /** Can HiZ be enabled on a depthbuffer of the given format? */
155      bool (*is_hiz_depth_format)(struct intel_context *intel,
156	                          gl_format format);
157   } vtbl;
158
159   GLbitfield Fallback;  /**< mask of INTEL_FALLBACK_x bits */
160   GLuint NewGLState;
161
162   dri_bufmgr *bufmgr;
163   unsigned int maxBatchSize;
164
165   /**
166    * Generation number of the hardware: 2 is 8xx, 3 is 9xx pre-965, 4 is 965.
167    */
168   int gen;
169   int gt;
170   GLboolean needs_ff_sync;
171   GLboolean is_g4x;
172   GLboolean is_945;
173   GLboolean has_separate_stencil;
174   GLboolean must_use_separate_stencil;
175   GLboolean has_hiz;
176
177   int urb_size;
178
179   struct intel_batchbuffer {
180      /** Current batchbuffer being queued up. */
181      drm_intel_bo *bo;
182      /** Last BO submitted to the hardware.  Used for glFinish(). */
183      drm_intel_bo *last_bo;
184      /** BO for post-sync nonzero writes for gen6 workaround. */
185      drm_intel_bo *workaround_bo;
186      bool need_workaround_flush;
187
188      struct cached_batch_item *cached_items;
189
190      uint16_t emit, total;
191      uint16_t used, reserved_space;
192      uint32_t map[8192];
193#define BATCH_SZ (8192*sizeof(uint32_t))
194
195      uint32_t state_batch_offset;
196      bool is_blit;
197   } batch;
198
199   drm_intel_bo *first_post_swapbuffers_batch;
200   GLboolean need_throttle;
201   GLboolean no_batch_wrap;
202   bool tnl_pipeline_running; /**< Set while i915's _tnl_run_pipeline. */
203
204   struct
205   {
206      GLuint id;
207      uint32_t start_ptr; /**< for i8xx */
208      uint32_t primitive;	/**< Current hardware primitive type */
209      void (*flush) (struct intel_context *);
210      drm_intel_bo *vb_bo;
211      uint8_t *vb;
212      unsigned int start_offset; /**< Byte offset of primitive sequence */
213      unsigned int current_offset; /**< Byte offset of next vertex */
214      unsigned int count;	/**< Number of vertices in current primitive */
215   } prim;
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   GLuint stats_wm;
226
227   /* Offsets of fields within the current vertex:
228    */
229   GLuint coloroffset;
230   GLuint specoffset;
231   GLuint wpos_offset;
232
233   struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
234   GLuint vertex_attr_count;
235
236   GLfloat polygon_offset_scale;        /* dependent on depth_scale, bpp */
237
238   GLboolean hw_stencil;
239   GLboolean hw_stipple;
240   GLboolean depth_buffer_is_float;
241   GLboolean no_rast;
242   GLboolean always_flush_batch;
243   GLboolean always_flush_cache;
244
245   /* 0 - nonconformant, best performance;
246    * 1 - fallback to sw for known conformance bugs
247    * 2 - always fallback to sw
248    */
249   GLuint conformance_mode;
250
251   /* State for intelvb.c and inteltris.c.
252    */
253   GLuint RenderIndex;
254   GLmatrix ViewportMatrix;
255   GLenum render_primitive;
256   GLenum reduced_primitive; /*< Only gen < 6 */
257   GLuint vertex_size;
258   GLubyte *verts;              /* points to tnl->clipspace.vertex_buf */
259
260   /* Fallback rasterization functions
261    */
262   intel_point_func draw_point;
263   intel_line_func draw_line;
264   intel_tri_func draw_tri;
265
266   /**
267    * Set if rendering has occured to the drawable's front buffer.
268    *
269    * This is used in the DRI2 case to detect that glFlush should also copy
270    * the contents of the fake front buffer to the real front buffer.
271    */
272   GLboolean front_buffer_dirty;
273
274   /**
275    * Track whether front-buffer rendering is currently enabled
276    *
277    * A separate flag is used to track this in order to support MRT more
278    * easily.
279    */
280   GLboolean is_front_buffer_rendering;
281   /**
282    * Track whether front-buffer is the current read target.
283    *
284    * This is closely associated with is_front_buffer_rendering, but may
285    * be set separately.  The DRI2 fake front buffer must be referenced
286    * either way.
287    */
288   GLboolean is_front_buffer_reading;
289
290   GLboolean use_texture_tiling;
291   GLboolean use_early_z;
292
293   int driFd;
294
295   __DRIcontext *driContext;
296   struct intel_screen *intelScreen;
297   void (*saved_viewport)(struct gl_context * ctx,
298			  GLint x, GLint y, GLsizei width, GLsizei height);
299
300   /**
301    * Configuration cache
302    */
303   driOptionCache optionCache;
304};
305
306extern char *__progname;
307
308
309#define SUBPIXEL_X 0.125
310#define SUBPIXEL_Y 0.125
311
312#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
313
314/**
315 * Align a value up to an alignment value
316 *
317 * If \c value is not already aligned to the requested alignment value, it
318 * will be rounded up.
319 *
320 * \param value  Value to be rounded
321 * \param alignment  Alignment value to be used.  This must be a power of two.
322 *
323 * \sa ROUND_DOWN_TO()
324 */
325#define ALIGN(value, alignment)  (((value) + alignment - 1) & ~(alignment - 1))
326
327/**
328 * Align a value down to an alignment value
329 *
330 * If \c value is not already aligned to the requested alignment value, it
331 * will be rounded down.
332 *
333 * \param value  Value to be rounded
334 * \param alignment  Alignment value to be used.  This must be a power of two.
335 *
336 * \sa ALIGN()
337 */
338#define ROUND_DOWN_TO(value, alignment) ((value) & ~(alignment - 1))
339
340#define IS_POWER_OF_TWO(val) (((val) & (val - 1)) == 0)
341
342static INLINE uint32_t
343U_FIXED(float value, uint32_t frac_bits)
344{
345   value *= (1 << frac_bits);
346   return value < 0 ? 0 : value;
347}
348
349static INLINE uint32_t
350S_FIXED(float value, uint32_t frac_bits)
351{
352   return value * (1 << frac_bits);
353}
354
355#define INTEL_FIREVERTICES(intel)		\
356do {						\
357   if ((intel)->prim.flush)			\
358      (intel)->prim.flush(intel);		\
359} while (0)
360
361/* ================================================================
362 * From linux kernel i386 header files, copes with odd sizes better
363 * than COPY_DWORDS would:
364 * XXX Put this in src/mesa/main/imports.h ???
365 */
366#if defined(i386) || defined(__i386__)
367static INLINE void * __memcpy(void * to, const void * from, size_t n)
368{
369   int d0, d1, d2;
370   __asm__ __volatile__(
371      "rep ; movsl\n\t"
372      "testb $2,%b4\n\t"
373      "je 1f\n\t"
374      "movsw\n"
375      "1:\ttestb $1,%b4\n\t"
376      "je 2f\n\t"
377      "movsb\n"
378      "2:"
379      : "=&c" (d0), "=&D" (d1), "=&S" (d2)
380      :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
381      : "memory");
382   return (to);
383}
384#else
385#define __memcpy(a,b,c) memcpy(a,b,c)
386#endif
387
388
389/* ================================================================
390 * Debugging:
391 */
392extern int INTEL_DEBUG;
393
394#define DEBUG_TEXTURE	0x1
395#define DEBUG_STATE	0x2
396#define DEBUG_IOCTL	0x4
397#define DEBUG_BLIT	0x8
398#define DEBUG_MIPTREE   0x10
399#define DEBUG_FALLBACKS	0x20
400#define DEBUG_VERBOSE	0x40
401#define DEBUG_BATCH     0x80
402#define DEBUG_PIXEL     0x100
403#define DEBUG_BUFMGR    0x200
404#define DEBUG_REGION    0x400
405#define DEBUG_FBO       0x800
406#define DEBUG_GS        0x1000
407#define DEBUG_SYNC	0x2000
408#define DEBUG_PRIMS	0x4000
409#define DEBUG_VERTS	0x8000
410#define DEBUG_DRI       0x10000
411#define DEBUG_SF        0x20000
412#define DEBUG_SANITY    0x40000
413#define DEBUG_SLEEP     0x80000
414#define DEBUG_STATS     0x100000
415#define DEBUG_TILE      0x200000
416#define DEBUG_SINGLE_THREAD   0x400000
417#define DEBUG_WM        0x800000
418#define DEBUG_URB       0x1000000
419#define DEBUG_VS        0x2000000
420#define DEBUG_CLIP      0x8000000
421
422#define DBG(...) do {						\
423	if (unlikely(INTEL_DEBUG & FILE_DEBUG_FLAG))		\
424		printf(__VA_ARGS__);			\
425} while(0)
426
427#define fallback_debug(...) do {				\
428	if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS))		\
429		printf(__VA_ARGS__);				\
430} while(0)
431
432#define PCI_CHIP_845_G			0x2562
433#define PCI_CHIP_I830_M			0x3577
434#define PCI_CHIP_I855_GM		0x3582
435#define PCI_CHIP_I865_G			0x2572
436#define PCI_CHIP_I915_G			0x2582
437#define PCI_CHIP_I915_GM		0x2592
438#define PCI_CHIP_I945_G			0x2772
439#define PCI_CHIP_I945_GM		0x27A2
440#define PCI_CHIP_I945_GME		0x27AE
441#define PCI_CHIP_G33_G			0x29C2
442#define PCI_CHIP_Q35_G			0x29B2
443#define PCI_CHIP_Q33_G			0x29D2
444
445
446/* ================================================================
447 * intel_context.c:
448 */
449
450extern GLboolean intelInitContext(struct intel_context *intel,
451				  int api,
452                                  const struct gl_config * mesaVis,
453                                  __DRIcontext * driContextPriv,
454                                  void *sharedContextPrivate,
455                                  struct dd_function_table *functions);
456
457extern void intelFinish(struct gl_context * ctx);
458extern void intel_flush(struct gl_context * ctx);
459
460extern void intelInitDriverFunctions(struct dd_function_table *functions);
461
462void intel_init_syncobj_functions(struct dd_function_table *functions);
463
464
465/* ================================================================
466 * intel_state.c:
467 */
468extern void intelInitStateFuncs(struct dd_function_table *functions);
469
470#define COMPAREFUNC_ALWAYS		0
471#define COMPAREFUNC_NEVER		0x1
472#define COMPAREFUNC_LESS		0x2
473#define COMPAREFUNC_EQUAL		0x3
474#define COMPAREFUNC_LEQUAL		0x4
475#define COMPAREFUNC_GREATER		0x5
476#define COMPAREFUNC_NOTEQUAL		0x6
477#define COMPAREFUNC_GEQUAL		0x7
478
479#define STENCILOP_KEEP			0
480#define STENCILOP_ZERO			0x1
481#define STENCILOP_REPLACE		0x2
482#define STENCILOP_INCRSAT		0x3
483#define STENCILOP_DECRSAT		0x4
484#define STENCILOP_INCR			0x5
485#define STENCILOP_DECR			0x6
486#define STENCILOP_INVERT		0x7
487
488#define LOGICOP_CLEAR			0
489#define LOGICOP_NOR			0x1
490#define LOGICOP_AND_INV 		0x2
491#define LOGICOP_COPY_INV		0x3
492#define LOGICOP_AND_RVRSE		0x4
493#define LOGICOP_INV			0x5
494#define LOGICOP_XOR			0x6
495#define LOGICOP_NAND			0x7
496#define LOGICOP_AND			0x8
497#define LOGICOP_EQUIV			0x9
498#define LOGICOP_NOOP			0xa
499#define LOGICOP_OR_INV			0xb
500#define LOGICOP_COPY			0xc
501#define LOGICOP_OR_RVRSE		0xd
502#define LOGICOP_OR			0xe
503#define LOGICOP_SET			0xf
504
505#define BLENDFACT_ZERO			0x01
506#define BLENDFACT_ONE			0x02
507#define BLENDFACT_SRC_COLR		0x03
508#define BLENDFACT_INV_SRC_COLR 		0x04
509#define BLENDFACT_SRC_ALPHA		0x05
510#define BLENDFACT_INV_SRC_ALPHA 	0x06
511#define BLENDFACT_DST_ALPHA		0x07
512#define BLENDFACT_INV_DST_ALPHA 	0x08
513#define BLENDFACT_DST_COLR		0x09
514#define BLENDFACT_INV_DST_COLR		0x0a
515#define BLENDFACT_SRC_ALPHA_SATURATE	0x0b
516#define BLENDFACT_CONST_COLOR		0x0c
517#define BLENDFACT_INV_CONST_COLOR	0x0d
518#define BLENDFACT_CONST_ALPHA		0x0e
519#define BLENDFACT_INV_CONST_ALPHA	0x0f
520#define BLENDFACT_MASK          	0x0f
521
522enum {
523   DRI_CONF_BO_REUSE_DISABLED,
524   DRI_CONF_BO_REUSE_ALL
525};
526
527extern int intel_translate_shadow_compare_func(GLenum func);
528extern int intel_translate_compare_func(GLenum func);
529extern int intel_translate_stencil_op(GLenum op);
530extern int intel_translate_blend_factor(GLenum factor);
531extern int intel_translate_logic_op(GLenum opcode);
532
533void intel_update_renderbuffers(__DRIcontext *context,
534				__DRIdrawable *drawable);
535void intel_prepare_render(struct intel_context *intel);
536
537void i915_set_buf_info_for_region(uint32_t *state, struct intel_region *region,
538				  uint32_t buffer_id);
539
540/*======================================================================
541 * Inline conversion functions.
542 * These are better-typed than the macros used previously:
543 */
544static INLINE struct intel_context *
545intel_context(struct gl_context * ctx)
546{
547   return (struct intel_context *) ctx;
548}
549
550static INLINE GLboolean
551is_power_of_two(uint32_t value)
552{
553   return (value & (value - 1)) == 0;
554}
555
556#endif
557