src/intel/compiler/brw_vec4_vs_visitor.cpp

   1 /*
   2  * Copyright © 2013 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
  21  * DEALINGS IN THE SOFTWARE.
  22  */
  23
  24
  25 #include "brw_vec4_vs.h"
  26 #include "common/gen_debug.h"
  27
  28 namespace brw {
  29
  30 void
  31 vec4_vs_visitor::emit_prolog()
  32 {
  33 }
  34
  35
  36 dst_reg *
  37 vec4_vs_visitor::make_reg_for_system_value(int location)
  38 {
  39    /* VertexID is stored by the VF as the last vertex element, but
  40     * we don't represent it with a flag in inputs_read, so we call
  41     * it VERT_ATTRIB_MAX, which setup_attributes() picks up on.
  42     */
  43    dst_reg *reg = new(mem_ctx) dst_reg(ATTR, VERT_ATTRIB_MAX);
  44
  45    switch (location) {
  46    case SYSTEM_VALUE_BASE_VERTEX:
  47       reg->writemask = WRITEMASK_X;
  48       break;
  49    case SYSTEM_VALUE_BASE_INSTANCE:
  50       reg->writemask = WRITEMASK_Y;
  51       break;
  52    case SYSTEM_VALUE_VERTEX_ID:
  53    case SYSTEM_VALUE_VERTEX_ID_ZERO_BASE:
  54       reg->writemask = WRITEMASK_Z;
  55       break;
  56    case SYSTEM_VALUE_INSTANCE_ID:
  57       reg->writemask = WRITEMASK_W;
  58       break;
  59    case SYSTEM_VALUE_DRAW_ID:
  60       reg = new(mem_ctx) dst_reg(ATTR, VERT_ATTRIB_MAX + 1);
  61       reg->writemask = WRITEMASK_X;
  62       break;
  63    default:
  64       unreachable("not reached");
  65    }
  66
  67    return reg;
  68 }
  69
  70
  71 void
  72 vec4_vs_visitor::emit_urb_write_header(int mrf)
  73 {
  74    /* No need to do anything for VS; an implied write to this MRF will be
  75     * performed by VS_OPCODE_URB_WRITE.
  76     */
  77    (void) mrf;
  78 }
  79
  80
  81 vec4_instruction *
  82 vec4_vs_visitor::emit_urb_write_opcode(bool complete)
  83 {
  84    /* For VS, the URB writes end the thread. */
  85    if (complete) {
  86       if (INTEL_DEBUG & DEBUG_SHADER_TIME)
  87          emit_shader_time_end();
  88    }
  89
  90    vec4_instruction *inst = emit(VS_OPCODE_URB_WRITE);
  91    inst->urb_write_flags = complete ?
  92       BRW_URB_WRITE_EOT_COMPLETE : BRW_URB_WRITE_NO_FLAGS;
  93
  94    return inst;
  95 }
  96
  97
  98 void
  99 vec4_vs_visitor::emit_urb_slot(dst_reg reg, int varying)
 100 {
 101    reg.type = BRW_REGISTER_TYPE_F;
 102    output_reg[varying][0].type = reg.type;
 103
 104    switch (varying) {
 105    case VARYING_SLOT_COL0:
 106    case VARYING_SLOT_COL1:
 107    case VARYING_SLOT_BFC0:
 108    case VARYING_SLOT_BFC1: {
 109       /* These built-in varyings are only supported in compatibility mode,
 110        * and we only support GS in core profile.  So, this must be a vertex
 111        * shader.
 112        */
 113       vec4_instruction *inst = emit_generic_urb_slot(reg, varying, 0);
 114       if (inst && key->clamp_vertex_color)
 115          inst->saturate = true;
 116       break;
 117    }
 118    default:
 119       return vec4_visitor::emit_urb_slot(reg, varying);
 120    }
 121 }
 122
 123
 124 void
 125 vec4_vs_visitor::emit_clip_distances(dst_reg reg, int offset)
 126 {
 127    /* From the GLSL 1.30 spec, section 7.1 (Vertex Shader Special Variables):
 128     *
 129     *     "If a linked set of shaders forming the vertex stage contains no
 130     *     static write to gl_ClipVertex or gl_ClipDistance, but the
 131     *     application has requested clipping against user clip planes through
 132     *     the API, then the coordinate written to gl_Position is used for
 133     *     comparison against the user clip planes."
 134     *
 135     * This function is only called if the shader didn't write to
 136     * gl_ClipDistance.  Accordingly, we use gl_ClipVertex to perform clipping
 137     * if the user wrote to it; otherwise we use gl_Position.
 138     */
 139    gl_varying_slot clip_vertex = VARYING_SLOT_CLIP_VERTEX;
 140    if (!(prog_data->vue_map.slots_valid & VARYING_BIT_CLIP_VERTEX)) {
 141       clip_vertex = VARYING_SLOT_POS;
 142    }
 143
 144    for (int i = 0; i + offset < key->nr_userclip_plane_consts && i < 4;
 145         ++i) {
 146       reg.writemask = 1 << i;
 147       emit(DP4(reg,
 148                src_reg(output_reg[clip_vertex][0]),
 149                src_reg(this->userplane[i + offset])));
 150    }
 151 }
 152
 153
 154 void
 155 vec4_vs_visitor::setup_uniform_clipplane_values()
 156 {
 157    for (int i = 0; i < key->nr_userclip_plane_consts; ++i) {
 158       this->userplane[i] = dst_reg(UNIFORM, this->uniforms);
 159       this->userplane[i].type = BRW_REGISTER_TYPE_F;
 160       for (int j = 0; j < 4; ++j) {
 161          stage_prog_data->param[this->uniforms * 4 + j] =
 162             (gl_constant_value *) &clip_planes[i][j];
 163       }
 164       ++this->uniforms;
 165    }
 166 }
 167
 168
 169 void
 170 vec4_vs_visitor::emit_thread_end()
 171 {
 172    setup_uniform_clipplane_values();
 173
 174    /* Lower legacy ff and ClipVertex clipping to clip distances */
 175    if (key->nr_userclip_plane_consts > 0) {
 176       current_annotation = "user clip distances";
 177
 178       output_reg[VARYING_SLOT_CLIP_DIST0][0] =
 179          dst_reg(this, glsl_type::vec4_type);
 180       output_reg[VARYING_SLOT_CLIP_DIST1][0] =
 181          dst_reg(this, glsl_type::vec4_type);
 182       output_num_components[VARYING_SLOT_CLIP_DIST0][0] = 4;
 183       output_num_components[VARYING_SLOT_CLIP_DIST1][0] = 4;
 184
 185       emit_clip_distances(output_reg[VARYING_SLOT_CLIP_DIST0][0], 0);
 186       emit_clip_distances(output_reg[VARYING_SLOT_CLIP_DIST1][0], 4);
 187    }
 188
 189    /* For VS, we always end the thread by emitting a single vertex.
 190     * emit_urb_write_opcode() will take care of setting the eot flag on the
 191     * SEND instruction.
 192     */
 193    emit_vertex();
 194 }
 195
 196
 197 vec4_vs_visitor::vec4_vs_visitor(const struct brw_compiler *compiler,
 198                                  void *log_data,
 199                                  const struct brw_vs_prog_key *key,
 200                                  struct brw_vs_prog_data *vs_prog_data,
 201                                  const nir_shader *shader,
 202                                  gl_clip_plane *clip_planes,
 203                                  void *mem_ctx,
 204                                  int shader_time_index,
 205                                  bool use_legacy_snorm_formula)
 206    : vec4_visitor(compiler, log_data, &key->tex, &vs_prog_data->base, shader,
 207                   mem_ctx, false /* no_spills */, shader_time_index),
 208      key(key),
 209      vs_prog_data(vs_prog_data),
 210      clip_planes(clip_planes),
 211      use_legacy_snorm_formula(use_legacy_snorm_formula)
 212 {
 213 }
 214
 215
 216 } /* namespace brw */