src/mesa/swrast/s_fragprog.c

   1 /*
   2  * Mesa 3-D graphics library
   3  * Version:  7.0.3
   4  *
   5  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
   6  *
   7  * Permission is hereby granted, free of charge, to any person obtaining a
   8  * copy of this software and associated documentation files (the "Software"),
   9  * to deal in the Software without restriction, including without limitation
  10  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
  11  * and/or sell copies of the Software, and to permit persons to whom the
  12  * Software is furnished to do so, subject to the following conditions:
  13  *
  14  * The above copyright notice and this permission notice shall be included
  15  * in all copies or substantial portions of the Software.
  16  *
  17  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  18  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  19  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  20  * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
  21  * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
  22  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  23  */
  24
  25 #include "main/glheader.h"
  26 #include "main/colormac.h"
  27 #include "main/context.h"
  28 #include "shader/prog_instruction.h"
  29
  30 #include "s_fragprog.h"
  31 #include "s_span.h"
  32
  33
  34 /**
  35  * Apply texture object's swizzle (X/Y/Z/W/0/1) to incoming 'texel'
  36  * and return results in 'colorOut'.
  37  */
  38 static INLINE void
  39 swizzle_texel(const GLfloat texel[4], GLfloat colorOut[4], GLuint swizzle)
  40 {
  41    if (swizzle == SWIZZLE_NOOP) {
  42       COPY_4V(colorOut, texel);
  43    }
  44    else {
  45       GLfloat vector[6];
  46       vector[SWIZZLE_X] = texel[0];
  47       vector[SWIZZLE_Y] = texel[1];
  48       vector[SWIZZLE_Z] = texel[2];
  49       vector[SWIZZLE_W] = texel[3];
  50       vector[SWIZZLE_ZERO] = 0.0F;
  51       vector[SWIZZLE_ONE] = 1.0F;
  52       colorOut[0] = vector[GET_SWZ(swizzle, 0)];
  53       colorOut[1] = vector[GET_SWZ(swizzle, 1)];
  54       colorOut[2] = vector[GET_SWZ(swizzle, 2)];
  55       colorOut[3] = vector[GET_SWZ(swizzle, 3)];
  56    }
  57 }
  58
  59
  60 /**
  61  * Fetch a texel with given lod.
  62  * Called via machine->FetchTexelLod()
  63  */
  64 static void
  65 fetch_texel_lod( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda,
  66                  GLuint unit, GLfloat color[4] )
  67 {
  68    const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
  69
  70    if (texObj) {
  71       SWcontext *swrast = SWRAST_CONTEXT(ctx);
  72       GLfloat rgba[4];
  73
  74       lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
  75
  76       swrast->TextureSample[unit](ctx, texObj, 1,
  77                                   (const GLfloat (*)[4]) texcoord,
  78                                   &lambda, &rgba);
  79       swizzle_texel(rgba, color, texObj->_Swizzle);
  80    }
  81    else {
  82       ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
  83    }
  84 }
  85
  86
  87 /**
  88  * Fetch a texel with the given partial derivatives to compute a level
  89  * of detail in the mipmap.
  90  * Called via machine->FetchTexelDeriv()
  91  * \param lodBias  the lod bias which may be specified by a TXB instruction,
  92  *                 otherwise zero.
  93  */
  94 static void
  95 fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4],
  96                    const GLfloat texdx[4], const GLfloat texdy[4],
  97                    GLfloat lodBias, GLuint unit, GLfloat color[4] )
  98 {
  99    SWcontext *swrast = SWRAST_CONTEXT(ctx);
 100    const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
 101    const struct gl_texture_object *texObj = texUnit->_Current;
 102
 103    if (texObj) {
 104       const struct gl_texture_image *texImg =
 105          texObj->Image[0][texObj->BaseLevel];
 106       const GLfloat texW = (GLfloat) texImg->WidthScale;
 107       const GLfloat texH = (GLfloat) texImg->HeightScale;
 108       GLfloat lambda;
 109       GLfloat rgba[4];
 110
 111       lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
 112                                       texdx[1], texdy[1], /* dt/dx, dt/dy */
 113                                       texdx[3], texdy[3], /* dq/dx, dq/dy */
 114                                       texW, texH,
 115                                       texcoord[0], texcoord[1], texcoord[3],
 116                                       1.0F / texcoord[3]);
 117
 118       lambda += lodBias + texUnit->LodBias + texObj->LodBias;
 119
 120       lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
 121
 122       swrast->TextureSample[unit](ctx, texObj, 1,
 123                                   (const GLfloat (*)[4]) texcoord,
 124                                   &lambda, &rgba);
 125       swizzle_texel(rgba, color, texObj->_Swizzle);
 126    }
 127    else {
 128       ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
 129    }
 130 }
 131
 132
 133 /**
 134  * Initialize the virtual fragment program machine state prior to running
 135  * fragment program on a fragment.  This involves initializing the input
 136  * registers, condition codes, etc.
 137  * \param machine  the virtual machine state to init
 138  * \param program  the fragment program we're about to run
 139  * \param span  the span of pixels we'll operate on
 140  * \param col  which element (column) of the span we'll operate on
 141  */
 142 static void
 143 init_machine(GLcontext *ctx, struct gl_program_machine *machine,
 144              const struct gl_fragment_program *program,
 145              const SWspan *span, GLuint col)
 146 {
 147    if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {
 148       /* Clear temporary registers (undefined for ARB_f_p) */
 149       _mesa_bzero(machine->Temporaries,
 150                   MAX_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
 151    }
 152
 153    /* Setup pointer to input attributes */
 154    machine->Attribs = span->array->attribs;
 155
 156    machine->DerivX = (GLfloat (*)[4]) span->attrStepX;
 157    machine->DerivY = (GLfloat (*)[4]) span->attrStepY;
 158    machine->NumDeriv = FRAG_ATTRIB_MAX;
 159
 160    machine->Samplers = program->Base.SamplerUnits;
 161
 162    /* if running a GLSL program (not ARB_fragment_program) */
 163    if (ctx->Shader.CurrentProgram) {
 164       /* Store front/back facing value */
 165       machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0 - span->facing;
 166    }
 167
 168    machine->CurElement = col;
 169
 170    /* init condition codes */
 171    machine->CondCodes[0] = COND_EQ;
 172    machine->CondCodes[1] = COND_EQ;
 173    machine->CondCodes[2] = COND_EQ;
 174    machine->CondCodes[3] = COND_EQ;
 175
 176    /* init call stack */
 177    machine->StackDepth = 0;
 178
 179    machine->FetchTexelLod = fetch_texel_lod;
 180    machine->FetchTexelDeriv = fetch_texel_deriv;
 181 }
 182
 183
 184 /**
 185  * Run fragment program on the pixels in span from 'start' to 'end' - 1.
 186  */
 187 static void
 188 run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
 189 {
 190    SWcontext *swrast = SWRAST_CONTEXT(ctx);
 191    const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
 192    const GLbitfield64 outputsWritten = program->Base.OutputsWritten;
 193    struct gl_program_machine *machine = &swrast->FragProgMachine;
 194    GLuint i;
 195
 196    for (i = start; i < end; i++) {
 197       if (span->array->mask[i]) {
 198          init_machine(ctx, machine, program, span, i);
 199
 200          if (_mesa_execute_program(ctx, &program->Base, machine)) {
 201
 202             /* Store result color */
 203             if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
 204                COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0][i],
 205                        machine->Outputs[FRAG_RESULT_COLOR]);
 206             }
 207             else {
 208                /* Multiple drawbuffers / render targets
 209                 * Note that colors beyond 0 and 1 will overwrite other
 210                 * attributes, such as FOGC, TEX0, TEX1, etc.  That's OK.
 211                 */
 212                GLuint buf;
 213                for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {
 214                   if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0 + buf)) {
 215                      COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0 + buf][i],
 216                              machine->Outputs[FRAG_RESULT_DATA0 + buf]);
 217                   }
 218                }
 219             }
 220
 221             /* Store result depth/z */
 222             if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
 223                const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPTH][2];
 224                if (depth <= 0.0)
 225                   span->array->z[i] = 0;
 226                else if (depth >= 1.0)
 227                   span->array->z[i] = ctx->DrawBuffer->_DepthMax;
 228                else
 229                   span->array->z[i] = IROUND(depth * ctx->DrawBuffer->_DepthMaxF);
 230             }
 231          }
 232          else {
 233             /* killed fragment */
 234             span->array->mask[i] = GL_FALSE;
 235             span->writeAll = GL_FALSE;
 236          }
 237       }
 238    }
 239 }
 240
 241
 242 /**
 243  * Execute the current fragment program for all the fragments
 244  * in the given span.
 245  */
 246 void
 247 _swrast_exec_fragment_program( GLcontext *ctx, SWspan *span )
 248 {
 249    const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
 250
 251    /* incoming colors should be floats */
 252    if (program->Base.InputsRead & FRAG_BIT_COL0) {
 253       ASSERT(span->array->ChanType == GL_FLOAT);
 254    }
 255
 256    run_program(ctx, span, 0, span->end);
 257
 258    if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
 259       span->interpMask &= ~SPAN_RGBA;
 260       span->arrayMask |= SPAN_RGBA;
 261    }
 262
 263    if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
 264       span->interpMask &= ~SPAN_Z;
 265       span->arrayMask |= SPAN_Z;
 266    }
 267 }
 268