src/mesa/drivers/dri/intel/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 #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
  66 struct intel_region;
  67 struct intel_context;
  68
  69 typedef void (*intel_tri_func) (struct intel_context *, intelVertex *,
  70                                 intelVertex *, intelVertex *);
  71 typedef void (*intel_line_func) (struct intel_context *, intelVertex *,
  72                                  intelVertex *);
  73 typedef 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
  89 extern void intelFallback(struct intel_context *intel, GLbitfield bit,
  90                           bool 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
 110 struct 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 struct brw_context;
 118
 119 struct intel_batchbuffer {
 120    /** Current batchbuffer being queued up. */
 121    drm_intel_bo *bo;
 122    /** Last BO submitted to the hardware.  Used for glFinish(). */
 123    drm_intel_bo *last_bo;
 124    /** BO for post-sync nonzero writes for gen6 workaround. */
 125    drm_intel_bo *workaround_bo;
 126    bool need_workaround_flush;
 127
 128    struct cached_batch_item *cached_items;
 129
 130    uint16_t emit, total;
 131    uint16_t used, reserved_space;
 132    uint32_t *map;
 133    uint32_t *cpu_map;
 134 #define BATCH_SZ (8192*sizeof(uint32_t))
 135
 136    uint32_t state_batch_offset;
 137    bool is_blit;
 138    bool needs_sol_reset;
 139
 140    struct {
 141       uint16_t used;
 142       int reloc_count;
 143    } saved;
 144 };
 145
 146 /**
 147  * intel_context is derived from Mesa's context class: struct gl_context.
 148  */
 149 struct intel_context
 150 {
 151    struct gl_context ctx;  /**< base class, must be first field */
 152
 153    struct
 154    {
 155       void (*destroy) (struct intel_context * intel);
 156       void (*emit_state) (struct intel_context * intel);
 157       void (*finish_batch) (struct intel_context * intel);
 158       void (*new_batch) (struct intel_context * intel);
 159       void (*emit_invarient_state) (struct intel_context * intel);
 160       void (*update_texture_state) (struct intel_context * intel);
 161
 162       void (*render_start) (struct intel_context * intel);
 163       void (*render_prevalidate) (struct intel_context * intel);
 164       void (*set_draw_region) (struct intel_context * intel,
 165                                struct intel_region * draw_regions[],
 166                                struct intel_region * depth_region,
 167                                GLuint num_regions);
 168       void (*update_draw_buffer)(struct intel_context *intel);
 169
 170       void (*reduced_primitive_state) (struct intel_context * intel,
 171                                        GLenum rprim);
 172
 173       bool (*check_vertex_size) (struct intel_context * intel,
 174                                       GLuint expected);
 175       void (*invalidate_state) (struct intel_context *intel,
 176                                 GLuint new_state);
 177
 178       void (*assert_not_dirty) (struct intel_context *intel);
 179
 180       void (*debug_batch)(struct intel_context *intel);
 181       void (*annotate_aub)(struct intel_context *intel);
 182       bool (*render_target_supported)(struct intel_context *intel,
 183                                       struct gl_renderbuffer *rb);
 184
 185       /** Can HiZ be enabled on a depthbuffer of the given format? */
 186       bool (*is_hiz_depth_format)(struct intel_context *intel,
 187                                   gl_format format);
 188
 189       /**
 190        * Surface state operations (i965+ only)
 191        * \{
 192        */
 193       void (*update_texture_surface)(struct gl_context *ctx,
 194                                      unsigned unit,
 195                                      uint32_t *binding_table,
 196                                      unsigned surf_index);
 197       void (*update_renderbuffer_surface)(struct brw_context *brw,
 198                                           struct gl_renderbuffer *rb,
 199                                           unsigned unit);
 200       void (*update_null_renderbuffer_surface)(struct brw_context *brw,
 201                                                unsigned unit);
 202       void (*create_constant_surface)(struct brw_context *brw,
 203                                       drm_intel_bo *bo,
 204                                       uint32_t offset,
 205                                       int width,
 206                                       uint32_t *out_offset);
 207       /** \} */
 208    } vtbl;
 209
 210    GLbitfield Fallback;  /**< mask of INTEL_FALLBACK_x bits */
 211    GLuint NewGLState;
 212
 213    dri_bufmgr *bufmgr;
 214    unsigned int maxBatchSize;
 215
 216    /**
 217     * Generation number of the hardware: 2 is 8xx, 3 is 9xx pre-965, 4 is 965.
 218     */
 219    int gen;
 220    int gt;
 221    bool needs_ff_sync;
 222    bool is_haswell;
 223    bool is_g4x;
 224    bool is_945;
 225    bool has_separate_stencil;
 226    bool must_use_separate_stencil;
 227    bool has_hiz;
 228    bool has_llc;
 229    bool has_swizzling;
 230
 231    int urb_size;
 232
 233    drm_intel_context *hw_ctx;
 234
 235    struct intel_batchbuffer batch;
 236
 237    drm_intel_bo *first_post_swapbuffers_batch;
 238    bool need_throttle;
 239    bool no_batch_wrap;
 240    bool tnl_pipeline_running; /**< Set while i915's _tnl_run_pipeline. */
 241
 242    struct
 243    {
 244       GLuint id;
 245       uint32_t start_ptr; /**< for i8xx */
 246       uint32_t primitive;       /**< Current hardware primitive type */
 247       void (*flush) (struct intel_context *);
 248       drm_intel_bo *vb_bo;
 249       uint8_t *vb;
 250       unsigned int start_offset; /**< Byte offset of primitive sequence */
 251       unsigned int current_offset; /**< Byte offset of next vertex */
 252       unsigned int count;       /**< Number of vertices in current primitive */
 253    } prim;
 254
 255    struct {
 256       drm_intel_bo *bo;
 257       GLuint offset;
 258       uint32_t buffer_len;
 259       uint32_t buffer_offset;
 260       char buffer[4096];
 261    } upload;
 262
 263    GLuint stats_wm;
 264
 265    /* Offsets of fields within the current vertex:
 266     */
 267    GLuint coloroffset;
 268    GLuint specoffset;
 269    GLuint wpos_offset;
 270
 271    struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
 272    GLuint vertex_attr_count;
 273
 274    GLfloat polygon_offset_scale;        /* dependent on depth_scale, bpp */
 275
 276    bool hw_stencil;
 277    bool hw_stipple;
 278    bool no_rast;
 279    bool always_flush_batch;
 280    bool always_flush_cache;
 281
 282    /* State for intelvb.c and inteltris.c.
 283     */
 284    GLuint RenderIndex;
 285    GLmatrix ViewportMatrix;
 286    GLenum render_primitive;
 287    GLenum reduced_primitive; /*< Only gen < 6 */
 288    GLuint vertex_size;
 289    GLubyte *verts;              /* points to tnl->clipspace.vertex_buf */
 290
 291    /* Fallback rasterization functions
 292     */
 293    intel_point_func draw_point;
 294    intel_line_func draw_line;
 295    intel_tri_func draw_tri;
 296
 297    /**
 298     * Set if rendering has occured to the drawable's front buffer.
 299     *
 300     * This is used in the DRI2 case to detect that glFlush should also copy
 301     * the contents of the fake front buffer to the real front buffer.
 302     */
 303    bool front_buffer_dirty;
 304
 305    /**
 306     * Track whether front-buffer rendering is currently enabled
 307     *
 308     * A separate flag is used to track this in order to support MRT more
 309     * easily.
 310     */
 311    bool is_front_buffer_rendering;
 312    /**
 313     * Track whether front-buffer is the current read target.
 314     *
 315     * This is closely associated with is_front_buffer_rendering, but may
 316     * be set separately.  The DRI2 fake front buffer must be referenced
 317     * either way.
 318     */
 319    bool is_front_buffer_reading;
 320
 321    /**
 322     * Count of intel_regions that are mapped.
 323     *
 324     * This allows us to assert that no batch buffer is emitted if a
 325     * region is mapped.
 326     */
 327    int num_mapped_regions;
 328
 329    bool use_texture_tiling;
 330    bool use_early_z;
 331
 332    int driFd;
 333
 334    __DRIcontext *driContext;
 335    struct intel_screen *intelScreen;
 336    void (*saved_viewport)(struct gl_context * ctx,
 337                           GLint x, GLint y, GLsizei width, GLsizei height);
 338
 339    /**
 340     * Configuration cache
 341     */
 342    driOptionCache optionCache;
 343 };
 344
 345 extern char *__progname;
 346
 347
 348 #define SUBPIXEL_X 0.125
 349 #define SUBPIXEL_Y 0.125
 350
 351 /**
 352  * Align a value down to an alignment value
 353  *
 354  * If \c value is not already aligned to the requested alignment value, it
 355  * will be rounded down.
 356  *
 357  * \param value  Value to be rounded
 358  * \param alignment  Alignment value to be used.  This must be a power of two.
 359  *
 360  * \sa ALIGN()
 361  */
 362 #define ROUND_DOWN_TO(value, alignment) ((value) & ~(alignment - 1))
 363
 364 #define IS_POWER_OF_TWO(val) (((val) & (val - 1)) == 0)
 365
 366 static INLINE uint32_t
 367 U_FIXED(float value, uint32_t frac_bits)
 368 {
 369    value *= (1 << frac_bits);
 370    return value < 0 ? 0 : value;
 371 }
 372
 373 static INLINE uint32_t
 374 S_FIXED(float value, uint32_t frac_bits)
 375 {
 376    return value * (1 << frac_bits);
 377 }
 378
 379 #define INTEL_FIREVERTICES(intel)               \
 380 do {                                            \
 381    if ((intel)->prim.flush)                     \
 382       (intel)->prim.flush(intel);               \
 383 } while (0)
 384
 385 /* ================================================================
 386  * From linux kernel i386 header files, copes with odd sizes better
 387  * than COPY_DWORDS would:
 388  * XXX Put this in src/mesa/main/imports.h ???
 389  */
 390 #if defined(i386) || defined(__i386__)
 391 static INLINE void * __memcpy(void * to, const void * from, size_t n)
 392 {
 393    int d0, d1, d2;
 394    __asm__ __volatile__(
 395       "rep ; movsl\n\t"
 396       "testb $2,%b4\n\t"
 397       "je 1f\n\t"
 398       "movsw\n"
 399       "1:\ttestb $1,%b4\n\t"
 400       "je 2f\n\t"
 401       "movsb\n"
 402       "2:"
 403       : "=&c" (d0), "=&D" (d1), "=&S" (d2)
 404       :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
 405       : "memory");
 406    return (to);
 407 }
 408 #else
 409 #define __memcpy(a,b,c) memcpy(a,b,c)
 410 #endif
 411
 412
 413 /* ================================================================
 414  * Debugging:
 415  */
 416 extern int INTEL_DEBUG;
 417
 418 #define DEBUG_TEXTURE   0x1
 419 #define DEBUG_STATE     0x2
 420 #define DEBUG_IOCTL     0x4
 421 #define DEBUG_BLIT      0x8
 422 #define DEBUG_MIPTREE   0x10
 423 #define DEBUG_PERF      0x20
 424 #define DEBUG_VERBOSE   0x40
 425 #define DEBUG_BATCH     0x80
 426 #define DEBUG_PIXEL     0x100
 427 #define DEBUG_BUFMGR    0x200
 428 #define DEBUG_REGION    0x400
 429 #define DEBUG_FBO       0x800
 430 #define DEBUG_GS        0x1000
 431 #define DEBUG_SYNC      0x2000
 432 #define DEBUG_PRIMS     0x4000
 433 #define DEBUG_VERTS     0x8000
 434 #define DEBUG_DRI       0x10000
 435 #define DEBUG_SF        0x20000
 436 #define DEBUG_SANITY    0x40000
 437 #define DEBUG_SLEEP     0x80000
 438 #define DEBUG_STATS     0x100000
 439 #define DEBUG_TILE      0x200000
 440 #define DEBUG_WM        0x400000
 441 #define DEBUG_URB       0x800000
 442 #define DEBUG_VS        0x1000000
 443 #define DEBUG_CLIP      0x2000000
 444 #define DEBUG_AUB       0x4000000
 445 #define DEBUG_SHADER_TIME 0x8000000
 446 #define DEBUG_NO16      0x20000000
 447
 448 #ifdef HAVE_ANDROID_PLATFORM
 449 #define LOG_TAG "INTEL-MESA"
 450 #include <cutils/log.h>
 451 #ifndef ALOGW
 452 #define ALOGW LOGW
 453 #endif
 454 #define dbg_printf(...) ALOGW(__VA_ARGS__)
 455 #else
 456 #define dbg_printf(...) printf(__VA_ARGS__)
 457 #endif /* HAVE_ANDROID_PLATFORM */
 458
 459 #define DBG(...) do {                                           \
 460         if (unlikely(INTEL_DEBUG & FILE_DEBUG_FLAG))            \
 461                 dbg_printf(__VA_ARGS__);                        \
 462 } while(0)
 463
 464 #define fallback_debug(...) do {                                \
 465         if (unlikely(INTEL_DEBUG & DEBUG_PERF))                 \
 466                 dbg_printf(__VA_ARGS__);                        \
 467 } while(0)
 468
 469 #define perf_debug(...) do {                                    \
 470         if (unlikely(INTEL_DEBUG & DEBUG_PERF))                 \
 471                 dbg_printf(__VA_ARGS__);                        \
 472 } while(0)
 473
 474 #define WARN_ONCE(cond, fmt...) do {                            \
 475    if (unlikely(cond)) {                                        \
 476       static bool _warned = false;                              \
 477       static GLuint msg_id = 0;                                 \
 478       if (!_warned) {                                           \
 479          fprintf(stderr, "WARNING: ");                          \
 480          fprintf(stderr, fmt);                                  \
 481          _warned = true;                                        \
 482                                                                 \
 483          _mesa_gl_debug(ctx, &msg_id,                           \
 484                         MESA_DEBUG_TYPE_OTHER,                  \
 485                         MESA_DEBUG_SEVERITY_HIGH, fmt);         \
 486       }                                                         \
 487    }                                                            \
 488 } while (0)
 489
 490 #define PCI_CHIP_845_G                  0x2562
 491 #define PCI_CHIP_I830_M                 0x3577
 492 #define PCI_CHIP_I855_GM                0x3582
 493 #define PCI_CHIP_I865_G                 0x2572
 494 #define PCI_CHIP_I915_G                 0x2582
 495 #define PCI_CHIP_I915_GM                0x2592
 496 #define PCI_CHIP_I945_G                 0x2772
 497 #define PCI_CHIP_I945_GM                0x27A2
 498 #define PCI_CHIP_I945_GME               0x27AE
 499 #define PCI_CHIP_G33_G                  0x29C2
 500 #define PCI_CHIP_Q35_G                  0x29B2
 501 #define PCI_CHIP_Q33_G                  0x29D2
 502
 503
 504 /* ================================================================
 505  * intel_context.c:
 506  */
 507
 508 extern bool intelInitContext(struct intel_context *intel,
 509                              int api,
 510                              unsigned major_version,
 511                              unsigned minor_version,
 512                              const struct gl_config * mesaVis,
 513                              __DRIcontext * driContextPriv,
 514                              void *sharedContextPrivate,
 515                              struct dd_function_table *functions,
 516                              unsigned *dri_ctx_error);
 517
 518 extern void intelFinish(struct gl_context * ctx);
 519 extern void intel_flush_rendering_to_batch(struct gl_context *ctx);
 520 extern void _intel_flush(struct gl_context * ctx, const char *file, int line);
 521
 522 #define intel_flush(ctx) _intel_flush(ctx, __FILE__, __LINE__)
 523
 524 extern void intelInitDriverFunctions(struct dd_function_table *functions);
 525
 526 void intel_init_syncobj_functions(struct dd_function_table *functions);
 527
 528
 529 /* ================================================================
 530  * intel_state.c:
 531  */
 532
 533 #define COMPAREFUNC_ALWAYS              0
 534 #define COMPAREFUNC_NEVER               0x1
 535 #define COMPAREFUNC_LESS                0x2
 536 #define COMPAREFUNC_EQUAL               0x3
 537 #define COMPAREFUNC_LEQUAL              0x4
 538 #define COMPAREFUNC_GREATER             0x5
 539 #define COMPAREFUNC_NOTEQUAL            0x6
 540 #define COMPAREFUNC_GEQUAL              0x7
 541
 542 #define STENCILOP_KEEP                  0
 543 #define STENCILOP_ZERO                  0x1
 544 #define STENCILOP_REPLACE               0x2
 545 #define STENCILOP_INCRSAT               0x3
 546 #define STENCILOP_DECRSAT               0x4
 547 #define STENCILOP_INCR                  0x5
 548 #define STENCILOP_DECR                  0x6
 549 #define STENCILOP_INVERT                0x7
 550
 551 #define LOGICOP_CLEAR                   0
 552 #define LOGICOP_NOR                     0x1
 553 #define LOGICOP_AND_INV                 0x2
 554 #define LOGICOP_COPY_INV                0x3
 555 #define LOGICOP_AND_RVRSE               0x4
 556 #define LOGICOP_INV                     0x5
 557 #define LOGICOP_XOR                     0x6
 558 #define LOGICOP_NAND                    0x7
 559 #define LOGICOP_AND                     0x8
 560 #define LOGICOP_EQUIV                   0x9
 561 #define LOGICOP_NOOP                    0xa
 562 #define LOGICOP_OR_INV                  0xb
 563 #define LOGICOP_COPY                    0xc
 564 #define LOGICOP_OR_RVRSE                0xd
 565 #define LOGICOP_OR                      0xe
 566 #define LOGICOP_SET                     0xf
 567
 568 #define BLENDFACT_ZERO                  0x01
 569 #define BLENDFACT_ONE                   0x02
 570 #define BLENDFACT_SRC_COLR              0x03
 571 #define BLENDFACT_INV_SRC_COLR          0x04
 572 #define BLENDFACT_SRC_ALPHA             0x05
 573 #define BLENDFACT_INV_SRC_ALPHA         0x06
 574 #define BLENDFACT_DST_ALPHA             0x07
 575 #define BLENDFACT_INV_DST_ALPHA         0x08
 576 #define BLENDFACT_DST_COLR              0x09
 577 #define BLENDFACT_INV_DST_COLR          0x0a
 578 #define BLENDFACT_SRC_ALPHA_SATURATE    0x0b
 579 #define BLENDFACT_CONST_COLOR           0x0c
 580 #define BLENDFACT_INV_CONST_COLOR       0x0d
 581 #define BLENDFACT_CONST_ALPHA           0x0e
 582 #define BLENDFACT_INV_CONST_ALPHA       0x0f
 583 #define BLENDFACT_MASK                  0x0f
 584
 585 enum {
 586    DRI_CONF_BO_REUSE_DISABLED,
 587    DRI_CONF_BO_REUSE_ALL
 588 };
 589
 590 extern int intel_translate_shadow_compare_func(GLenum func);
 591 extern int intel_translate_compare_func(GLenum func);
 592 extern int intel_translate_stencil_op(GLenum op);
 593 extern int intel_translate_blend_factor(GLenum factor);
 594 extern int intel_translate_logic_op(GLenum opcode);
 595
 596 void intel_update_renderbuffers(__DRIcontext *context,
 597                                 __DRIdrawable *drawable);
 598 void intel_prepare_render(struct intel_context *intel);
 599
 600 void
 601 intel_downsample_for_dri2_flush(struct intel_context *intel,
 602                                 __DRIdrawable *drawable);
 603
 604 void i915_set_buf_info_for_region(uint32_t *state, struct intel_region *region,
 605                                   uint32_t buffer_id);
 606 void intel_init_texture_formats(struct gl_context *ctx);
 607
 608 /*======================================================================
 609  * Inline conversion functions.
 610  * These are better-typed than the macros used previously:
 611  */
 612 static INLINE struct intel_context *
 613 intel_context(struct gl_context * ctx)
 614 {
 615    return (struct intel_context *) ctx;
 616 }
 617
 618 static INLINE bool
 619 is_power_of_two(uint32_t value)
 620 {
 621    return (value & (value - 1)) == 0;
 622 }
 623
 624 #ifdef __cplusplus
 625 }
 626 #endif
 627
 628 #endif