src/mesa/drivers/dri/i965/gen7_sf_state.c

   1 /*
   2  * Copyright © 2011 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 #include "brw_context.h"
  25 #include "brw_state.h"
  26 #include "brw_defines.h"
  27 #include "brw_util.h"
  28 #include "main/macros.h"
  29 #include "intel_batchbuffer.h"
  30
  31 static void
  32 upload_sbe_state(struct brw_context *brw)
  33 {
  34    struct intel_context *intel = &brw->intel;
  35    struct gl_context *ctx = &intel->ctx;
  36    uint32_t urb_entry_read_length;
  37    /* BRW_NEW_FRAGMENT_PROGRAM */
  38    uint32_t num_outputs = _mesa_bitcount_64(brw->fragment_program->Base.InputsRead);
  39    /* _NEW_LIGHT */
  40    bool shade_model_flat = ctx->Light.ShadeModel == GL_FLAT;
  41    uint32_t dw1, dw10, dw11;
  42    int i;
  43    int attr = 0, input_index = 0;
  44    int urb_entry_read_offset = 1;
  45    uint16_t attr_overrides[FRAG_ATTRIB_MAX];
  46    /* _NEW_BUFFERS */
  47    bool render_to_fbo = ctx->DrawBuffer->Name != 0;
  48    uint32_t point_sprite_origin;
  49
  50    /* CACHE_NEW_VS_PROG */
  51    urb_entry_read_length = ((brw->vs.prog_data->vue_map.num_slots + 1) / 2 -
  52                             urb_entry_read_offset);
  53    if (urb_entry_read_length == 0) {
  54       /* Setting the URB entry read length to 0 causes undefined behavior, so
  55        * if we have no URB data to read, set it to 1.
  56        */
  57       urb_entry_read_length = 1;
  58    }
  59
  60    /* FINISHME: Attribute Swizzle Control Mode? */
  61    dw1 =
  62       GEN7_SBE_SWIZZLE_ENABLE |
  63       num_outputs << GEN7_SBE_NUM_OUTPUTS_SHIFT |
  64       urb_entry_read_length << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT |
  65       urb_entry_read_offset << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT;
  66
  67    /* _NEW_POINT
  68     *
  69     * Window coordinates in an FBO are inverted, which means point
  70     * sprite origin must be inverted.
  71     */
  72    if ((ctx->Point.SpriteOrigin == GL_LOWER_LEFT) != render_to_fbo) {
  73       point_sprite_origin = GEN6_SF_POINT_SPRITE_LOWERLEFT;
  74    } else {
  75       point_sprite_origin = GEN6_SF_POINT_SPRITE_UPPERLEFT;
  76    }
  77    dw1 |= point_sprite_origin;
  78
  79
  80    dw10 = 0;
  81    dw11 = 0;
  82
  83    /* Create the mapping from the FS inputs we produce to the VS outputs
  84     * they source from.
  85     */
  86    for (; attr < FRAG_ATTRIB_MAX; attr++) {
  87       enum glsl_interp_qualifier interp_qualifier =
  88          brw->fragment_program->InterpQualifier[attr];
  89       bool is_gl_Color = attr == FRAG_ATTRIB_COL0 || attr == FRAG_ATTRIB_COL1;
  90
  91       if (!(brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr)))
  92          continue;
  93
  94       if (ctx->Point.PointSprite &&
  95           attr >= FRAG_ATTRIB_TEX0 && attr <= FRAG_ATTRIB_TEX7 &&
  96           ctx->Point.CoordReplace[attr - FRAG_ATTRIB_TEX0]) {
  97          dw10 |= (1 << input_index);
  98       }
  99
 100       if (attr == FRAG_ATTRIB_PNTC)
 101          dw10 |= (1 << input_index);
 102
 103       /* flat shading */
 104       if (interp_qualifier == INTERP_QUALIFIER_FLAT ||
 105           (shade_model_flat && is_gl_Color &&
 106            interp_qualifier == INTERP_QUALIFIER_NONE))
 107          dw11 |= (1 << input_index);
 108
 109       /* The hardware can only do the overrides on 16 overrides at a
 110        * time, and the other up to 16 have to be lined up so that the
 111        * input index = the output index.  We'll need to do some
 112        * tweaking to make sure that's the case.
 113        */
 114       assert(input_index < 16 || attr == input_index);
 115
 116       /* CACHE_NEW_VS_PROG | _NEW_LIGHT | _NEW_PROGRAM */
 117       attr_overrides[input_index++] =
 118          get_attr_override(&brw->vs.prog_data->vue_map,
 119                            urb_entry_read_offset, attr,
 120                            ctx->VertexProgram._TwoSideEnabled);
 121    }
 122
 123    for (; input_index < FRAG_ATTRIB_MAX; input_index++)
 124       attr_overrides[input_index] = 0;
 125
 126    BEGIN_BATCH(14);
 127    OUT_BATCH(_3DSTATE_SBE << 16 | (14 - 2));
 128    OUT_BATCH(dw1);
 129
 130    /* Output dwords 2 through 9 */
 131    for (i = 0; i < 8; i++) {
 132       OUT_BATCH(attr_overrides[i * 2] | attr_overrides[i * 2 + 1] << 16);
 133    }
 134
 135    OUT_BATCH(dw10); /* point sprite texcoord bitmask */
 136    OUT_BATCH(dw11); /* constant interp bitmask */
 137    OUT_BATCH(0); /* wrapshortest enables 0-7 */
 138    OUT_BATCH(0); /* wrapshortest enables 8-15 */
 139    ADVANCE_BATCH();
 140 }
 141
 142 const struct brw_tracked_state gen7_sbe_state = {
 143    .dirty = {
 144       .mesa  = (_NEW_LIGHT |
 145                 _NEW_POINT |
 146                 _NEW_PROGRAM),
 147       .brw   = (BRW_NEW_CONTEXT |
 148                 BRW_NEW_FRAGMENT_PROGRAM),
 149       .cache = CACHE_NEW_VS_PROG
 150    },
 151    .emit = upload_sbe_state,
 152 };
 153
 154 static void
 155 upload_sf_state(struct brw_context *brw)
 156 {
 157    struct intel_context *intel = &brw->intel;
 158    struct gl_context *ctx = &intel->ctx;
 159    uint32_t dw1, dw2, dw3;
 160    float point_size;
 161    /* _NEW_BUFFERS */
 162    bool render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0;
 163
 164    dw1 = GEN6_SF_STATISTICS_ENABLE |
 165          GEN6_SF_VIEWPORT_TRANSFORM_ENABLE;
 166
 167    /* _NEW_BUFFERS */
 168    dw1 |= (brw_depthbuffer_format(brw) << GEN7_SF_DEPTH_BUFFER_SURFACE_FORMAT_SHIFT);
 169
 170    /* _NEW_POLYGON */
 171    if ((ctx->Polygon.FrontFace == GL_CCW) ^ render_to_fbo)
 172       dw1 |= GEN6_SF_WINDING_CCW;
 173
 174    if (ctx->Polygon.OffsetFill)
 175        dw1 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_SOLID;
 176
 177    if (ctx->Polygon.OffsetLine)
 178        dw1 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_WIREFRAME;
 179
 180    if (ctx->Polygon.OffsetPoint)
 181        dw1 |= GEN6_SF_GLOBAL_DEPTH_OFFSET_POINT;
 182
 183    switch (ctx->Polygon.FrontMode) {
 184    case GL_FILL:
 185        dw1 |= GEN6_SF_FRONT_SOLID;
 186        break;
 187
 188    case GL_LINE:
 189        dw1 |= GEN6_SF_FRONT_WIREFRAME;
 190        break;
 191
 192    case GL_POINT:
 193        dw1 |= GEN6_SF_FRONT_POINT;
 194        break;
 195
 196    default:
 197        assert(0);
 198        break;
 199    }
 200
 201    switch (ctx->Polygon.BackMode) {
 202    case GL_FILL:
 203        dw1 |= GEN6_SF_BACK_SOLID;
 204        break;
 205
 206    case GL_LINE:
 207        dw1 |= GEN6_SF_BACK_WIREFRAME;
 208        break;
 209
 210    case GL_POINT:
 211        dw1 |= GEN6_SF_BACK_POINT;
 212        break;
 213
 214    default:
 215        assert(0);
 216        break;
 217    }
 218
 219    dw2 = 0;
 220
 221    if (ctx->Polygon.CullFlag) {
 222       switch (ctx->Polygon.CullFaceMode) {
 223       case GL_FRONT:
 224          dw2 |= GEN6_SF_CULL_FRONT;
 225          break;
 226       case GL_BACK:
 227          dw2 |= GEN6_SF_CULL_BACK;
 228          break;
 229       case GL_FRONT_AND_BACK:
 230          dw2 |= GEN6_SF_CULL_BOTH;
 231          break;
 232       default:
 233          assert(0);
 234          break;
 235       }
 236    } else {
 237       dw2 |= GEN6_SF_CULL_NONE;
 238    }
 239
 240    /* _NEW_SCISSOR */
 241    if (ctx->Scissor.Enabled)
 242       dw2 |= GEN6_SF_SCISSOR_ENABLE;
 243
 244    /* _NEW_LINE */
 245    dw2 |= U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7) <<
 246       GEN6_SF_LINE_WIDTH_SHIFT;
 247    if (ctx->Line.SmoothFlag) {
 248       dw2 |= GEN6_SF_LINE_AA_ENABLE;
 249       dw2 |= GEN6_SF_LINE_AA_MODE_TRUE;
 250       dw2 |= GEN6_SF_LINE_END_CAP_WIDTH_1_0;
 251    }
 252
 253    /* FINISHME: Last Pixel Enable?  Vertex Sub Pixel Precision Select?
 254     * FINISHME: AA Line Distance Mode?
 255     */
 256
 257    dw3 = 0;
 258
 259    /* _NEW_PROGRAM | _NEW_POINT */
 260    if (!(ctx->VertexProgram.PointSizeEnabled || ctx->Point._Attenuated))
 261       dw3 |= GEN6_SF_USE_STATE_POINT_WIDTH;
 262
 263    /* Clamp to ARB_point_parameters user limits */
 264    point_size = CLAMP(ctx->Point.Size, ctx->Point.MinSize, ctx->Point.MaxSize);
 265
 266    /* Clamp to the hardware limits and convert to fixed point */
 267    dw3 |= U_FIXED(CLAMP(point_size, 0.125, 255.875), 3);
 268
 269    /* _NEW_LIGHT */
 270    if (ctx->Light.ProvokingVertex != GL_FIRST_VERTEX_CONVENTION) {
 271       dw3 |=
 272          (2 << GEN6_SF_TRI_PROVOKE_SHIFT) |
 273          (2 << GEN6_SF_TRIFAN_PROVOKE_SHIFT) |
 274          (1 << GEN6_SF_LINE_PROVOKE_SHIFT);
 275    } else {
 276       dw3 |= (1 << GEN6_SF_TRIFAN_PROVOKE_SHIFT);
 277    }
 278
 279    BEGIN_BATCH(7);
 280    OUT_BATCH(_3DSTATE_SF << 16 | (7 - 2));
 281    OUT_BATCH(dw1);
 282    OUT_BATCH(dw2);
 283    OUT_BATCH(dw3);
 284    OUT_BATCH_F(ctx->Polygon.OffsetUnits * 2); /* constant.  copied from gen4 */
 285    OUT_BATCH_F(ctx->Polygon.OffsetFactor); /* scale */
 286    OUT_BATCH_F(0.0); /* XXX: global depth offset clamp */
 287    ADVANCE_BATCH();
 288 }
 289
 290 const struct brw_tracked_state gen7_sf_state = {
 291    .dirty = {
 292       .mesa  = (_NEW_LIGHT |
 293                 _NEW_PROGRAM |
 294                 _NEW_POLYGON |
 295                 _NEW_LINE |
 296                 _NEW_SCISSOR |
 297                 _NEW_BUFFERS |
 298                 _NEW_POINT),
 299       .brw   = BRW_NEW_CONTEXT,
 300       .cache = CACHE_NEW_VS_PROG
 301    },
 302    .emit = upload_sf_state,
 303 };