1 /**************************************************************************
3 * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
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:
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
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.
26 **************************************************************************/
28 #ifndef INTELCONTEXT_INC
29 #define INTELCONTEXT_INC
34 #include "main/mtypes.h"
39 /* Evil hack for using libdrm in a c++ compiler. */
44 #include "intel_bufmgr.h"
46 #include "intel_screen.h"
47 #include "intel_tex_obj.h"
55 #include "tnl/t_vertex.h"
57 #define TAG(x) intel##x
58 #include "tnl_dd/t_dd_vertex.h"
61 #define DV_PF_555 (1<<8)
62 #define DV_PF_565 (2<<8)
63 #define DV_PF_8888 (3<<8)
64 #define DV_PF_4444 (8<<8)
65 #define DV_PF_1555 (9<<8)
70 typedef void (*intel_tri_func
) (struct intel_context
*, intelVertex
*,
71 intelVertex
*, intelVertex
*);
72 typedef void (*intel_line_func
) (struct intel_context
*, intelVertex
*,
74 typedef void (*intel_point_func
) (struct intel_context
*, intelVertex
*);
77 * Bits for intel->Fallback field
80 #define INTEL_FALLBACK_DRAW_BUFFER 0x1
81 #define INTEL_FALLBACK_READ_BUFFER 0x2
82 #define INTEL_FALLBACK_DEPTH_BUFFER 0x4
83 #define INTEL_FALLBACK_STENCIL_BUFFER 0x8
84 #define INTEL_FALLBACK_USER 0x10
85 #define INTEL_FALLBACK_RENDERMODE 0x20
86 #define INTEL_FALLBACK_TEXTURE 0x40
87 #define INTEL_FALLBACK_DRIVER 0x1000 /**< first for drivers */
90 extern void intelFallback(struct intel_context
*intel
, GLbitfield bit
,
92 #define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
95 #define INTEL_WRITE_PART 0x1
96 #define INTEL_WRITE_FULL 0x2
97 #define INTEL_READ 0x4
99 #define INTEL_MAX_FIXUP 64
103 #define likely(expr) (__builtin_expect(expr, 1))
104 #define unlikely(expr) (__builtin_expect(expr, 0))
106 #define likely(expr) (expr)
107 #define unlikely(expr) (expr)
111 struct intel_sync_object
{
112 struct gl_sync_object Base
;
114 /** Batch associated with this sync object */
121 * intel_context is derived from Mesa's context class: struct gl_context.
125 struct gl_context ctx
; /**< base class, must be first field */
129 void (*destroy
) (struct intel_context
* intel
);
130 void (*emit_state
) (struct intel_context
* intel
);
131 void (*finish_batch
) (struct intel_context
* intel
);
132 void (*new_batch
) (struct intel_context
* intel
);
133 void (*emit_invarient_state
) (struct intel_context
* intel
);
134 void (*update_texture_state
) (struct intel_context
* intel
);
136 void (*render_start
) (struct intel_context
* intel
);
137 void (*render_prevalidate
) (struct intel_context
* intel
);
138 void (*set_draw_region
) (struct intel_context
* intel
,
139 struct intel_region
* draw_regions
[],
140 struct intel_region
* depth_region
,
142 void (*update_draw_buffer
)(struct intel_context
*intel
);
144 void (*reduced_primitive_state
) (struct intel_context
* intel
,
147 bool (*check_vertex_size
) (struct intel_context
* intel
,
149 void (*invalidate_state
) (struct intel_context
*intel
,
152 void (*assert_not_dirty
) (struct intel_context
*intel
);
154 void (*debug_batch
)(struct intel_context
*intel
);
155 bool (*render_target_supported
)(struct intel_context
*intel
,
156 struct gl_renderbuffer
*rb
);
158 /** Can HiZ be enabled on a depthbuffer of the given format? */
159 bool (*is_hiz_depth_format
)(struct intel_context
*intel
,
163 * \name HiZ operations
165 * See the following sections of the Sandy Bridge PRM, Volume 1, Part2:
166 * - 7.5.3.1 Depth Buffer Clear
167 * - 7.5.3.2 Depth Buffer Resolve
168 * - 7.5.3.3 Hierarchical Depth Buffer Resolve
171 void (*resolve_hiz_slice
)(struct intel_context
*intel
,
172 struct intel_mipmap_tree
*mt
,
176 void (*resolve_depth_slice
)(struct intel_context
*intel
,
177 struct intel_mipmap_tree
*mt
,
183 * Surface state operations (i965+ only)
186 void (*update_texture_surface
)(struct gl_context
*ctx
, unsigned unit
);
187 void (*update_renderbuffer_surface
)(struct brw_context
*brw
,
188 struct gl_renderbuffer
*rb
,
190 void (*update_null_renderbuffer_surface
)(struct brw_context
*brw
,
192 void (*create_constant_surface
)(struct brw_context
*brw
,
195 uint32_t *out_offset
);
199 GLbitfield Fallback
; /**< mask of INTEL_FALLBACK_x bits */
203 unsigned int maxBatchSize
;
206 * Generation number of the hardware: 2 is 8xx, 3 is 9xx pre-965, 4 is 965.
214 bool has_separate_stencil
;
215 bool must_use_separate_stencil
;
222 struct intel_batchbuffer
{
223 /** Current batchbuffer being queued up. */
225 /** Last BO submitted to the hardware. Used for glFinish(). */
226 drm_intel_bo
*last_bo
;
227 /** BO for post-sync nonzero writes for gen6 workaround. */
228 drm_intel_bo
*workaround_bo
;
229 bool need_workaround_flush
;
231 struct cached_batch_item
*cached_items
;
233 uint16_t emit
, total
;
234 uint16_t used
, reserved_space
;
236 #define BATCH_SZ (8192*sizeof(uint32_t))
238 uint32_t state_batch_offset
;
240 bool needs_sol_reset
;
248 drm_intel_bo
*first_post_swapbuffers_batch
;
251 bool tnl_pipeline_running
; /**< Set while i915's _tnl_run_pipeline. */
256 uint32_t start_ptr
; /**< for i8xx */
257 uint32_t primitive
; /**< Current hardware primitive type */
258 void (*flush
) (struct intel_context
*);
261 unsigned int start_offset
; /**< Byte offset of primitive sequence */
262 unsigned int current_offset
; /**< Byte offset of next vertex */
263 unsigned int count
; /**< Number of vertices in current primitive */
270 uint32_t buffer_offset
;
276 /* Offsets of fields within the current vertex:
282 struct tnl_attr_map vertex_attrs
[VERT_ATTRIB_MAX
];
283 GLuint vertex_attr_count
;
285 GLfloat polygon_offset_scale
; /* dependent on depth_scale, bpp */
290 bool always_flush_batch
;
291 bool always_flush_cache
;
293 /* State for intelvb.c and inteltris.c.
296 GLmatrix ViewportMatrix
;
297 GLenum render_primitive
;
298 GLenum reduced_primitive
; /*< Only gen < 6 */
300 GLubyte
*verts
; /* points to tnl->clipspace.vertex_buf */
302 /* Fallback rasterization functions
304 intel_point_func draw_point
;
305 intel_line_func draw_line
;
306 intel_tri_func draw_tri
;
309 * Set if rendering has occured to the drawable's front buffer.
311 * This is used in the DRI2 case to detect that glFlush should also copy
312 * the contents of the fake front buffer to the real front buffer.
314 bool front_buffer_dirty
;
317 * Track whether front-buffer rendering is currently enabled
319 * A separate flag is used to track this in order to support MRT more
322 bool is_front_buffer_rendering
;
324 * Track whether front-buffer is the current read target.
326 * This is closely associated with is_front_buffer_rendering, but may
327 * be set separately. The DRI2 fake front buffer must be referenced
330 bool is_front_buffer_reading
;
333 * Count of intel_regions that are mapped.
335 * This allows us to assert that no batch buffer is emitted if a
338 int num_mapped_regions
;
340 bool use_texture_tiling
;
345 __DRIcontext
*driContext
;
346 struct intel_screen
*intelScreen
;
347 void (*saved_viewport
)(struct gl_context
* ctx
,
348 GLint x
, GLint y
, GLsizei width
, GLsizei height
);
351 * Configuration cache
353 driOptionCache optionCache
;
356 extern char *__progname
;
359 #define SUBPIXEL_X 0.125
360 #define SUBPIXEL_Y 0.125
362 #define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
365 * Align a value up to an alignment value
367 * If \c value is not already aligned to the requested alignment value, it
368 * will be rounded up.
370 * \param value Value to be rounded
371 * \param alignment Alignment value to be used. This must be a power of two.
373 * \sa ROUND_DOWN_TO()
375 #define ALIGN(value, alignment) (((value) + alignment - 1) & ~(alignment - 1))
378 * Align a value down to an alignment value
380 * If \c value is not already aligned to the requested alignment value, it
381 * will be rounded down.
383 * \param value Value to be rounded
384 * \param alignment Alignment value to be used. This must be a power of two.
388 #define ROUND_DOWN_TO(value, alignment) ((value) & ~(alignment - 1))
390 #define IS_POWER_OF_TWO(val) (((val) & (val - 1)) == 0)
392 static INLINE
uint32_t
393 U_FIXED(float value
, uint32_t frac_bits
)
395 value
*= (1 << frac_bits
);
396 return value
< 0 ? 0 : value
;
399 static INLINE
uint32_t
400 S_FIXED(float value
, uint32_t frac_bits
)
402 return value
* (1 << frac_bits
);
405 #define INTEL_FIREVERTICES(intel) \
407 if ((intel)->prim.flush) \
408 (intel)->prim.flush(intel); \
411 /* ================================================================
412 * From linux kernel i386 header files, copes with odd sizes better
413 * than COPY_DWORDS would:
414 * XXX Put this in src/mesa/main/imports.h ???
416 #if defined(i386) || defined(__i386__)
417 static INLINE
void * __memcpy(void * to
, const void * from
, size_t n
)
420 __asm__
__volatile__(
425 "1:\ttestb $1,%b4\n\t"
429 : "=&c" (d0
), "=&D" (d1
), "=&S" (d2
)
430 :"0" (n
/4), "q" (n
),"1" ((long) to
),"2" ((long) from
)
435 #define __memcpy(a,b,c) memcpy(a,b,c)
439 /* ================================================================
442 extern int INTEL_DEBUG
;
444 #define DEBUG_TEXTURE 0x1
445 #define DEBUG_STATE 0x2
446 #define DEBUG_IOCTL 0x4
447 #define DEBUG_BLIT 0x8
448 #define DEBUG_MIPTREE 0x10
449 #define DEBUG_FALLBACKS 0x20
450 #define DEBUG_VERBOSE 0x40
451 #define DEBUG_BATCH 0x80
452 #define DEBUG_PIXEL 0x100
453 #define DEBUG_BUFMGR 0x200
454 #define DEBUG_REGION 0x400
455 #define DEBUG_FBO 0x800
456 #define DEBUG_GS 0x1000
457 #define DEBUG_SYNC 0x2000
458 #define DEBUG_PRIMS 0x4000
459 #define DEBUG_VERTS 0x8000
460 #define DEBUG_DRI 0x10000
461 #define DEBUG_SF 0x20000
462 #define DEBUG_SANITY 0x40000
463 #define DEBUG_SLEEP 0x80000
464 #define DEBUG_STATS 0x100000
465 #define DEBUG_TILE 0x200000
466 #define DEBUG_WM 0x400000
467 #define DEBUG_URB 0x800000
468 #define DEBUG_VS 0x1000000
469 #define DEBUG_CLIP 0x2000000
470 #define DEBUG_AUB 0x4000000
472 #define DBG(...) do { \
473 if (unlikely(INTEL_DEBUG & FILE_DEBUG_FLAG)) \
474 printf(__VA_ARGS__); \
477 #define fallback_debug(...) do { \
478 if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS)) \
479 printf(__VA_ARGS__); \
482 #define PCI_CHIP_845_G 0x2562
483 #define PCI_CHIP_I830_M 0x3577
484 #define PCI_CHIP_I855_GM 0x3582
485 #define PCI_CHIP_I865_G 0x2572
486 #define PCI_CHIP_I915_G 0x2582
487 #define PCI_CHIP_I915_GM 0x2592
488 #define PCI_CHIP_I945_G 0x2772
489 #define PCI_CHIP_I945_GM 0x27A2
490 #define PCI_CHIP_I945_GME 0x27AE
491 #define PCI_CHIP_G33_G 0x29C2
492 #define PCI_CHIP_Q35_G 0x29B2
493 #define PCI_CHIP_Q33_G 0x29D2
496 /* ================================================================
500 extern bool intelInitContext(struct intel_context
*intel
,
502 const struct gl_config
* mesaVis
,
503 __DRIcontext
* driContextPriv
,
504 void *sharedContextPrivate
,
505 struct dd_function_table
*functions
);
507 extern void intelFinish(struct gl_context
* ctx
);
508 extern void intel_flush_rendering_to_batch(struct gl_context
*ctx
);
509 extern void _intel_flush(struct gl_context
* ctx
, const char *file
, int line
);
511 #define intel_flush(ctx) _intel_flush(ctx, __FILE__, __LINE__)
513 extern void intelInitDriverFunctions(struct dd_function_table
*functions
);
515 void intel_init_syncobj_functions(struct dd_function_table
*functions
);
518 /* ================================================================
521 extern void intelInitStateFuncs(struct dd_function_table
*functions
);
523 #define COMPAREFUNC_ALWAYS 0
524 #define COMPAREFUNC_NEVER 0x1
525 #define COMPAREFUNC_LESS 0x2
526 #define COMPAREFUNC_EQUAL 0x3
527 #define COMPAREFUNC_LEQUAL 0x4
528 #define COMPAREFUNC_GREATER 0x5
529 #define COMPAREFUNC_NOTEQUAL 0x6
530 #define COMPAREFUNC_GEQUAL 0x7
532 #define STENCILOP_KEEP 0
533 #define STENCILOP_ZERO 0x1
534 #define STENCILOP_REPLACE 0x2
535 #define STENCILOP_INCRSAT 0x3
536 #define STENCILOP_DECRSAT 0x4
537 #define STENCILOP_INCR 0x5
538 #define STENCILOP_DECR 0x6
539 #define STENCILOP_INVERT 0x7
541 #define LOGICOP_CLEAR 0
542 #define LOGICOP_NOR 0x1
543 #define LOGICOP_AND_INV 0x2
544 #define LOGICOP_COPY_INV 0x3
545 #define LOGICOP_AND_RVRSE 0x4
546 #define LOGICOP_INV 0x5
547 #define LOGICOP_XOR 0x6
548 #define LOGICOP_NAND 0x7
549 #define LOGICOP_AND 0x8
550 #define LOGICOP_EQUIV 0x9
551 #define LOGICOP_NOOP 0xa
552 #define LOGICOP_OR_INV 0xb
553 #define LOGICOP_COPY 0xc
554 #define LOGICOP_OR_RVRSE 0xd
555 #define LOGICOP_OR 0xe
556 #define LOGICOP_SET 0xf
558 #define BLENDFACT_ZERO 0x01
559 #define BLENDFACT_ONE 0x02
560 #define BLENDFACT_SRC_COLR 0x03
561 #define BLENDFACT_INV_SRC_COLR 0x04
562 #define BLENDFACT_SRC_ALPHA 0x05
563 #define BLENDFACT_INV_SRC_ALPHA 0x06
564 #define BLENDFACT_DST_ALPHA 0x07
565 #define BLENDFACT_INV_DST_ALPHA 0x08
566 #define BLENDFACT_DST_COLR 0x09
567 #define BLENDFACT_INV_DST_COLR 0x0a
568 #define BLENDFACT_SRC_ALPHA_SATURATE 0x0b
569 #define BLENDFACT_CONST_COLOR 0x0c
570 #define BLENDFACT_INV_CONST_COLOR 0x0d
571 #define BLENDFACT_CONST_ALPHA 0x0e
572 #define BLENDFACT_INV_CONST_ALPHA 0x0f
573 #define BLENDFACT_MASK 0x0f
576 DRI_CONF_BO_REUSE_DISABLED
,
577 DRI_CONF_BO_REUSE_ALL
580 extern int intel_translate_shadow_compare_func(GLenum func
);
581 extern int intel_translate_compare_func(GLenum func
);
582 extern int intel_translate_stencil_op(GLenum op
);
583 extern int intel_translate_blend_factor(GLenum factor
);
584 extern int intel_translate_logic_op(GLenum opcode
);
586 void intel_update_renderbuffers(__DRIcontext
*context
,
587 __DRIdrawable
*drawable
);
588 void intel_prepare_render(struct intel_context
*intel
);
590 void i915_set_buf_info_for_region(uint32_t *state
, struct intel_region
*region
,
592 void intel_init_texture_formats(struct gl_context
*ctx
);
594 /*======================================================================
595 * Inline conversion functions.
596 * These are better-typed than the macros used previously:
598 static INLINE
struct intel_context
*
599 intel_context(struct gl_context
* ctx
)
601 return (struct intel_context
*) ctx
;
605 is_power_of_two(uint32_t value
)
607 return (value
& (value
- 1)) == 0;