s_nvfragprog.c

winNT技术操作系统,国外开放的原代码和LIUX一样

               result[0] = (a[0] != b[0]) ? 1.0F : 0.0F;
               result[1] = (a[1] != b[1]) ? 1.0F : 0.0F;
               result[2] = (a[2] != b[2]) ? 1.0F : 0.0F;
               result[3] = (a[3] != b[3]) ? 1.0F : 0.0F;
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_STR: /* set true, operands ignored */
            {
               static const GLfloat result[4] = { 1.0F, 1.0F, 1.0F, 1.0F };
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_SUB:
            {
               GLfloat a[4], b[4], result[4];
               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
               fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b );
               result[0] = a[0] - b[0];
               result[1] = a[1] - b[1];
               result[2] = a[2] - b[2];
               result[3] = a[3] - b[3];
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_SWZ:
            {
               const struct fp_src_register *source = &inst->SrcReg[0];
               const GLfloat *src = get_register_pointer(ctx, source,
                                                         machine, program);
               GLfloat result[4];
               GLuint i;

               /* do extended swizzling here */
               for (i = 0; i < 4; i++) {
                  if (GET_SWZ(source->Swizzle, i) == SWIZZLE_ZERO)
                     result[i] = 0.0;
                  else if (GET_SWZ(source->Swizzle, i) == SWIZZLE_ONE)
                     result[i] = 1.0;
                  else
                     result[i] = src[GET_SWZ(source->Swizzle, i)];

                  if (source->NegateBase & (1 << i))
                     result[i] = -result[i];
               }
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_TEX: /* Both ARB and NV frag prog */
            /* Texel lookup */
            {
               GLfloat texcoord[4], color[4];
               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
               /* Note: we pass 0 for LOD.  The ARB extension requires it
                * while the NV extension says it's implementation dependant.
                */
               /* KW: Previously lambda was passed as zero, but I
		* believe this is incorrect, the spec seems to
		* indicate rather that lambda should not be
		* changed/biased, unlike TXB where texcoord[3] is
		* added to the lambda calculations.  The lambda should
		* still be calculated normally for TEX & TXP though,
		* not set to zero.  Otherwise it's very difficult to
		* implement normal GL semantics through the fragment
		* shader.
		*/
               fetch_texel( ctx, texcoord, 
			    span->array->lambda[inst->TexSrcUnit][column],
			    inst->TexSrcUnit, color );
               store_vector4( inst, machine, color );
            }
            break;
         case FP_OPCODE_TXB: /* GL_ARB_fragment_program only */
            /* Texel lookup with LOD bias */
            {
               GLfloat texcoord[4], color[4], bias, lambda;

               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
               /* texcoord[3] is the bias to add to lambda */
               bias = ctx->Texture.Unit[inst->TexSrcUnit].LodBias
                    + ctx->Texture.Unit[inst->TexSrcUnit]._Current->LodBias
                    + texcoord[3];
               lambda = span->array->lambda[inst->TexSrcUnit][column] + bias;
               fetch_texel( ctx, texcoord, lambda,
                            inst->TexSrcUnit, color );
               store_vector4( inst, machine, color );
            }
            break;
         case FP_OPCODE_TXD: /* GL_NV_fragment_program only */
            /* Texture lookup w/ partial derivatives for LOD */
            {
               GLfloat texcoord[4], dtdx[4], dtdy[4], color[4];
               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
               fetch_vector4( ctx, &inst->SrcReg[1], machine, program, dtdx );
               fetch_vector4( ctx, &inst->SrcReg[2], machine, program, dtdy );
               fetch_texel_deriv( ctx, texcoord, dtdx, dtdy, inst->TexSrcUnit,
                                  color );
               store_vector4( inst, machine, color );
            }
            break;
         case FP_OPCODE_TXP: /* GL_ARB_fragment_program only */
            /* Texture lookup w/ projective divide */
            {
               GLfloat texcoord[4], color[4];
               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
	       /* Not so sure about this test - if texcoord[3] is
		* zero, we'd probably be fine except for an ASSERT in
		* IROUND_POS() which gets triggered by the inf values created.
		*/
	       if (texcoord[3] != 0.0) {
		  texcoord[0] /= texcoord[3];
		  texcoord[1] /= texcoord[3];
		  texcoord[2] /= texcoord[3];
	       }
               /* KW: Previously lambda was passed as zero, but I
		* believe this is incorrect, the spec seems to
		* indicate rather that lambda should not be
		* changed/biased, unlike TXB where texcoord[3] is
		* added to the lambda calculations.  The lambda should
		* still be calculated normally for TEX & TXP though,
		* not set to zero.
		*/
               fetch_texel( ctx, texcoord, 
			    span->array->lambda[inst->TexSrcUnit][column],
			    inst->TexSrcUnit, color );
               store_vector4( inst, machine, color );
            }
            break;
         case FP_OPCODE_TXP_NV: /* GL_NV_fragment_program only */
            /* Texture lookup w/ projective divide */
            {
               GLfloat texcoord[4], color[4];
               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
               if (inst->TexSrcIdx != TEXTURE_CUBE_INDEX &&
		   texcoord[3] != 0.0) {
                  texcoord[0] /= texcoord[3];
                  texcoord[1] /= texcoord[3];
                  texcoord[2] /= texcoord[3];
               }
               fetch_texel( ctx, texcoord,
                            span->array->lambda[inst->TexSrcUnit][column],
                            inst->TexSrcUnit, color );
               store_vector4( inst, machine, color );
            }
            break;
         case FP_OPCODE_UP2H: /* unpack two 16-bit floats */
            {
               GLfloat a[4], result[4];
               const GLuint *rawBits = (const GLuint *) a;
               GLhalfNV hx, hy;
               fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
               hx = rawBits[0] & 0xffff;
               hy = rawBits[0] >> 16;
               result[0] = result[2] = _mesa_half_to_float(hx);
               result[1] = result[3] = _mesa_half_to_float(hy);
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_UP2US: /* unpack two GLushorts */
            {
               GLfloat a[4], result[4];
               const GLuint *rawBits = (const GLuint *) a;
               GLushort usx, usy;
               fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
               usx = rawBits[0] & 0xffff;
               usy = rawBits[0] >> 16;
               result[0] = result[2] = usx * (1.0f / 65535.0f);
               result[1] = result[3] = usy * (1.0f / 65535.0f);
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_UP4B: /* unpack four GLbytes */
            {
               GLfloat a[4], result[4];
               const GLuint *rawBits = (const GLuint *) a;
               fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
               result[0] = (((rawBits[0] >>  0) & 0xff) - 128) / 127.0F;
               result[1] = (((rawBits[0] >>  8) & 0xff) - 128) / 127.0F;
               result[2] = (((rawBits[0] >> 16) & 0xff) - 128) / 127.0F;
               result[3] = (((rawBits[0] >> 24) & 0xff) - 128) / 127.0F;
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_UP4UB: /* unpack four GLubytes */
            {
               GLfloat a[4], result[4];
               const GLuint *rawBits = (const GLuint *) a;
               fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
               result[0] = ((rawBits[0] >>  0) & 0xff) / 255.0F;
               result[1] = ((rawBits[0] >>  8) & 0xff) / 255.0F;
               result[2] = ((rawBits[0] >> 16) & 0xff) / 255.0F;
               result[3] = ((rawBits[0] >> 24) & 0xff) / 255.0F;
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_XPD: /* cross product */
            {
               GLfloat a[4], b[4], result[4];
               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
               fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b );
               result[0] = a[1] * b[2] - a[2] * b[1];
               result[1] = a[2] * b[0] - a[0] * b[2];
               result[2] = a[0] * b[1] - a[1] * b[0];
               result[3] = 1.0;
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_X2D: /* 2-D matrix transform */
            {
               GLfloat a[4], b[4], c[4], result[4];
               fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
               fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b );
               fetch_vector4( ctx, &inst->SrcReg[2], machine, program, c );
               result[0] = a[0] + b[0] * c[0] + b[1] * c[1];
               result[1] = a[1] + b[0] * c[2] + b[1] * c[3];
               result[2] = a[2] + b[0] * c[0] + b[1] * c[1];
               result[3] = a[3] + b[0] * c[2] + b[1] * c[3];
               store_vector4( inst, machine, result );
            }
            break;
         case FP_OPCODE_PRINT:
            {
               if (inst->SrcReg[0].File != -1) {
                  GLfloat a[4];
                  fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a);
                  _mesa_printf("%s%g, %g, %g, %g\n", (const char *) inst->Data,
                               a[0], a[1], a[2], a[3]);
               }
               else {
                  _mesa_printf("%s\n", (const char *) inst->Data);
               }
            }
            break;
         case FP_OPCODE_END:
            return GL_TRUE;
         default:
            _mesa_problem(ctx, "Bad opcode %d in _mesa_exec_fragment_program",
                          inst->Opcode);
            return GL_TRUE; /* return value doesn't matter */
      }
   }
   return GL_TRUE;
}


static void
init_machine( GLcontext *ctx, struct fp_machine *machine,
              const struct fragment_program *program,
              const struct sw_span *span, GLuint col )
{
   GLuint inputsRead = program->InputsRead;
   GLuint u;

   if (ctx->FragmentProgram.CallbackEnabled)
      inputsRead = ~0;

   if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {
      /* Clear temporary registers (undefined for ARB_f_p) */
      _mesa_bzero(machine->Temporaries,
                  MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
   }

   /* Load input registers */
   if (inputsRead & (1 << FRAG_ATTRIB_WPOS)) {
      GLfloat *wpos = machine->Inputs[FRAG_ATTRIB_WPOS];
      ASSERT(span->arrayMask & SPAN_Z);
      wpos[0] = (GLfloat) span->x + col;
      wpos[1] = (GLfloat) span->y;
      wpos[2] = (GLfloat) span->array->z[col] / ctx->DrawBuffer->_DepthMaxF;
      wpos[3] = span->w + col * span->dwdx;
   }
   if (inputsRead & (1 << FRAG_ATTRIB_COL0)) {
      GLfloat *col0 = machine->Inputs[FRAG_ATTRIB_COL0];
      ASSERT(span->arrayMask & SPAN_RGBA);
      col0[0] = CHAN_TO_FLOAT(span->array->rgba[col][RCOMP]);
      col0[1] = CHAN_TO_FLOAT(span->array->rgba[col][GCOMP]);
      col0[2] = CHAN_TO_FLOAT(span->array->rgba[col][BCOMP]);
      col0[3] = CHAN_TO_FLOAT(span->array->rgba[col][ACOMP]);
   }
   if (inputsRead & (1 << FRAG_ATTRIB_COL1)) {
      GLfloat *col1 = machine->Inputs[FRAG_ATTRIB_COL1];
      col1[0] = CHAN_TO_FLOAT(span->array->spec[col][RCOMP]);
      col1[1] = CHAN_TO_FLOAT(span->array->spec[col][GCOMP]);
      col1[2] = CHAN_TO_FLOAT(span->array->spec[col][BCOMP]);
      col1[3] = CHAN_TO_FLOAT(span->array->spec[col][ACOMP]);
   }
   if (inputsRead & (1 << FRAG_ATTRIB_FOGC)) {
      GLfloat *fogc = machine->Inputs[FRAG_ATTRIB_FOGC];
      ASSERT(span->arrayMask & SPAN_FOG);
      fogc[0] = span->array->fog[col];
      fogc[1] = 0.0F;
      fogc[2] = 0.0F;
      fogc[3] = 0.0F;
   }
   for (u = 0; u < ctx->Const.MaxTextureCoordUnits; u++) {
      if (inputsRead & (1 << (FRAG_ATTRIB_TEX0 + u))) {
         GLfloat *tex = machine->Inputs[FRAG_ATTRIB_TEX0 + u];
         /*ASSERT(ctx->Texture._EnabledCoordUnits & (1 << u));*/
         COPY_4V(tex, span->array->texcoords[u][col]);
         /*ASSERT(tex[0] != 0 || tex[1] != 0 || tex[2] != 0);*/
      }
   }

   /* init condition codes */
   machine->CondCodes[0] = COND_EQ;
   machine->CondCodes[1] = COND_EQ;
   machine->CondCodes[2] = COND_EQ;
   machine->CondCodes[3] = COND_EQ;
}



/**
 * Execute the current fragment program, operating on the given span.
 */
void
_swrast_exec_fragment_program( GLcontext *ctx, struct sw_span *span )
{
   const struct fragment_program *program = ctx->FragmentProgram._Current;
   GLuint i;

   ctx->_CurrentProgram = GL_FRAGMENT_PROGRAM_ARB; /* or NV, doesn't matter */

   if (program->Parameters) {
      _mesa_load_state_parameters(ctx, program->Parameters);
   }   

   for (i = 0; i < span->end; i++) {
      if (span->array->mask[i]) {
         init_machine(ctx, &ctx->FragmentProgram.Machine,
                      ctx->FragmentProgram._Current, span, i);

#ifdef USE_TCC
         if (!_swrast_execute_codegen_program(ctx, program, ~0,
					      &ctx->FragmentProgram.Machine,
					      span, i)) {
            span->array->mask[i] = GL_FALSE;  /* killed fragment */
            span->writeAll = GL_FALSE;
         }
#else
         if (!execute_program(ctx, program, ~0,
                              &ctx->FragmentProgram.Machine, span, i)) {
            span->array->mask[i] = GL_FALSE;  /* killed fragment */
            span->writeAll = GL_FALSE;
         }
#endif

         /* Store output registers */
         {
            const GLfloat *colOut
               = ctx->FragmentProgram.Machine.Outputs[FRAG_OUTPUT_COLR];
            UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][RCOMP], colOut[0]);
            UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][GCOMP], colOut[1]);
            UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][BCOMP], colOut[2]);
            UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][ACOMP], colOut[3]);
         }
         /* depth value */
         if (program->OutputsWritten & (1 << FRAG_OUTPUT_DEPR)) {
            const GLfloat depth
               = ctx->FragmentProgram.Machine.Outputs[FRAG_OUTPUT_DEPR][2];
            span->array->z[i] = IROUND(depth * ctx->DrawBuffer->_DepthMaxF);
         }
      }
   }

   if (program->OutputsWritten & (1 << FRAG_OUTPUT_DEPR)) {
      span->interpMask &= ~SPAN_Z;
      span->arrayMask |= SPAN_Z;
   }

   ctx->_CurrentProgram = 0;
}