prog_execute.c

Mesa is an open-source implementation of the OpenGL specification - a system for rendering interacti

            /* do if/else */
            if (cond) {
               /* do if-clause (just continue execution) */
            }
            else {
               /* go to the instruction after ELSE or ENDIF */
               assert(inst->BranchTarget >= 0);
               pc = inst->BranchTarget - 1;
            }
         }
         break;
      case OPCODE_ELSE:
         /* goto ENDIF */
         assert(inst->BranchTarget >= 0);
         pc = inst->BranchTarget - 1;
         break;
      case OPCODE_ENDIF:
         /* nothing */
         break;
      case OPCODE_INT:         /* float to int */
         {
            GLfloat a[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            result[0] = (GLfloat) (GLint) a[0];
            result[1] = (GLfloat) (GLint) a[1];
            result[2] = (GLfloat) (GLint) a[2];
            result[3] = (GLfloat) (GLint) a[3];
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_KIL_NV:      /* NV_f_p only (conditional) */
         if (eval_condition(machine, inst)) {
            return GL_FALSE;
         }
         break;
      case OPCODE_KIL:         /* ARB_f_p only */
         {
            GLfloat a[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            if (a[0] < 0.0F || a[1] < 0.0F || a[2] < 0.0F || a[3] < 0.0F) {
               return GL_FALSE;
            }
         }
         break;
      case OPCODE_LG2:         /* log base 2 */
         {
            GLfloat a[4], result[4];
            fetch_vector1(&inst->SrcReg[0], machine, a);
            result[0] = result[1] = result[2] = result[3] = LOG2(a[0]);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_LIT:
         {
            const GLfloat epsilon = 1.0F / 256.0F;      /* from NV VP spec */
            GLfloat a[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            a[0] = MAX2(a[0], 0.0F);
            a[1] = MAX2(a[1], 0.0F);
            /* XXX ARB version clamps a[3], NV version doesn't */
            a[3] = CLAMP(a[3], -(128.0F - epsilon), (128.0F - epsilon));
            result[0] = 1.0F;
            result[1] = a[0];
            /* XXX we could probably just use pow() here */
            if (a[0] > 0.0F) {
               if (a[1] == 0.0 && a[3] == 0.0)
                  result[2] = 1.0;
               else
                  result[2] = EXPF(a[3] * LOGF(a[1]));
            }
            else {
               result[2] = 0.0;
            }
            result[3] = 1.0F;
            store_vector4(inst, machine, result);
            if (DEBUG_PROG) {
               printf("LIT (%g %g %g %g) : (%g %g %g %g)\n",
                      result[0], result[1], result[2], result[3],
                      a[0], a[1], a[2], a[3]);
            }
         }
         break;
      case OPCODE_LOG:
         {
            GLfloat t[4], q[4], abs_t0;
            fetch_vector1(&inst->SrcReg[0], machine, t);
            abs_t0 = FABSF(t[0]);
            if (abs_t0 != 0.0F) {
               /* Since we really can't handle infinite values on VMS
                * like other OSes we'll use __MAXFLOAT to represent
                * infinity.  This may need some tweaking.
                */
#ifdef VMS
               if (abs_t0 == __MAXFLOAT)
#else
               if (IS_INF_OR_NAN(abs_t0))
#endif
               {
                  SET_POS_INFINITY(q[0]);
                  q[1] = 1.0F;
                  SET_POS_INFINITY(q[2]);
               }
               else {
                  int exponent;
                  GLfloat mantissa = FREXPF(t[0], &exponent);
                  q[0] = (GLfloat) (exponent - 1);
                  q[1] = (GLfloat) (2.0 * mantissa); /* map [.5, 1) -> [1, 2) */
                  q[2] = (GLfloat) (q[0] + LOG2(q[1]));
               }
            }
            else {
               SET_NEG_INFINITY(q[0]);
               q[1] = 1.0F;
               SET_NEG_INFINITY(q[2]);
            }
            q[3] = 1.0;
            store_vector4(inst, machine, q);
         }
         break;
      case OPCODE_LRP:
         {
            GLfloat a[4], b[4], c[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            fetch_vector4(&inst->SrcReg[1], machine, b);
            fetch_vector4(&inst->SrcReg[2], machine, c);
            result[0] = a[0] * b[0] + (1.0F - a[0]) * c[0];
            result[1] = a[1] * b[1] + (1.0F - a[1]) * c[1];
            result[2] = a[2] * b[2] + (1.0F - a[2]) * c[2];
            result[3] = a[3] * b[3] + (1.0F - a[3]) * c[3];
            store_vector4(inst, machine, result);
            if (DEBUG_PROG) {
               printf("LRP (%g %g %g %g) = (%g %g %g %g), "
                      "(%g %g %g %g), (%g %g %g %g)\n",
                      result[0], result[1], result[2], result[3],
                      a[0], a[1], a[2], a[3],
                      b[0], b[1], b[2], b[3], c[0], c[1], c[2], c[3]);
            }
         }
         break;
      case OPCODE_MAD:
         {
            GLfloat a[4], b[4], c[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            fetch_vector4(&inst->SrcReg[1], machine, b);
            fetch_vector4(&inst->SrcReg[2], machine, c);
            result[0] = a[0] * b[0] + c[0];
            result[1] = a[1] * b[1] + c[1];
            result[2] = a[2] * b[2] + c[2];
            result[3] = a[3] * b[3] + c[3];
            store_vector4(inst, machine, result);
            if (DEBUG_PROG) {
               printf("MAD (%g %g %g %g) = (%g %g %g %g) * "
                      "(%g %g %g %g) + (%g %g %g %g)\n",
                      result[0], result[1], result[2], result[3],
                      a[0], a[1], a[2], a[3],
                      b[0], b[1], b[2], b[3], c[0], c[1], c[2], c[3]);
            }
         }
         break;
      case OPCODE_MAX:
         {
            GLfloat a[4], b[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            fetch_vector4(&inst->SrcReg[1], machine, b);
            result[0] = MAX2(a[0], b[0]);
            result[1] = MAX2(a[1], b[1]);
            result[2] = MAX2(a[2], b[2]);
            result[3] = MAX2(a[3], b[3]);
            store_vector4(inst, machine, result);
            if (DEBUG_PROG) {
               printf("MAX (%g %g %g %g) = (%g %g %g %g), (%g %g %g %g)\n",
                      result[0], result[1], result[2], result[3],
                      a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]);
            }
         }
         break;
      case OPCODE_MIN:
         {
            GLfloat a[4], b[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            fetch_vector4(&inst->SrcReg[1], machine, b);
            result[0] = MIN2(a[0], b[0]);
            result[1] = MIN2(a[1], b[1]);
            result[2] = MIN2(a[2], b[2]);
            result[3] = MIN2(a[3], b[3]);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_MOV:
         {
            GLfloat result[4];
            fetch_vector4(&inst->SrcReg[0], machine, result);
            store_vector4(inst, machine, result);
            if (DEBUG_PROG) {
               printf("MOV (%g %g %g %g)\n",
                      result[0], result[1], result[2], result[3]);
            }
         }
         break;
      case OPCODE_MUL:
         {
            GLfloat a[4], b[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            fetch_vector4(&inst->SrcReg[1], machine, 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);
            if (DEBUG_PROG) {
               printf("MUL (%g %g %g %g) = (%g %g %g %g) * (%g %g %g %g)\n",
                      result[0], result[1], result[2], result[3],
                      a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]);
            }
         }
         break;
      case OPCODE_NOISE1:
         {
            GLfloat a[4], result[4];
            fetch_vector1(&inst->SrcReg[0], machine, a);
            result[0] =
               result[1] =
               result[2] = result[3] = _slang_library_noise1(a[0]);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_NOISE2:
         {
            GLfloat a[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            result[0] =
               result[1] =
               result[2] = result[3] = _slang_library_noise2(a[0], a[1]);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_NOISE3:
         {
            GLfloat a[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            result[0] =
               result[1] =
               result[2] =
               result[3] = _slang_library_noise3(a[0], a[1], a[2]);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_NOISE4:
         {
            GLfloat a[4], result[4];
            fetch_vector4(&inst->SrcReg[0], machine, a);
            result[0] =
               result[1] =
               result[2] =
               result[3] = _slang_library_noise4(a[0], a[1], a[2], a[3]);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_NOP:
         break;
      case OPCODE_PK2H:        /* pack two 16-bit floats in one 32-bit float */
         {
            GLfloat a[4], result[4];
            GLhalfNV hx, hy;
            GLuint *rawResult = (GLuint *) result;
            GLuint twoHalves;
            fetch_vector4(&inst->SrcReg[0], machine, a);
            hx = _mesa_float_to_half(a[0]);
            hy = _mesa_float_to_half(a[1]);
            twoHalves = hx | (hy << 16);
            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
               = twoHalves;
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_PK2US:       /* pack two GLushorts into one 32-bit float */
         {
            GLfloat a[4], result[4];
            GLuint usx, usy, *rawResult = (GLuint *) result;
            fetch_vector4(&inst->SrcReg[0], machine, a);
            a[0] = CLAMP(a[0], 0.0F, 1.0F);
            a[1] = CLAMP(a[1], 0.0F, 1.0F);
            usx = IROUND(a[0] * 65535.0F);
            usy = IROUND(a[1] * 65535.0F);
            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
               = usx | (usy << 16);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_PK4B:        /* pack four GLbytes into one 32-bit float */
         {
            GLfloat a[4], result[4];
            GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
            fetch_vector4(&inst->SrcReg[0], machine, a);
            a[0] = CLAMP(a[0], -128.0F / 127.0F, 1.0F);
            a[1] = CLAMP(a[1], -128.0F / 127.0F, 1.0F);
            a[2] = CLAMP(a[2], -128.0F / 127.0F, 1.0F);
            a[3] = CLAMP(a[3], -128.0F / 127.0F, 1.0F);
            ubx = IROUND(127.0F * a[0] + 128.0F);
            uby = IROUND(127.0F * a[1] + 128.0F);
            ubz = IROUND(127.0F * a[2] + 128.0F);
            ubw = IROUND(127.0F * a[3] + 128.0F);
            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
               = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_PK4UB:       /* pack four GLubytes into one 32-bit float */
         {
            GLfloat a[4], result[4];
            GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
            fetch_vector4(&inst->SrcReg[0], machine, a);
            a[0] = CLAMP(a[0], 0.0F, 1.0F);
            a[1] = CLAMP(a[1], 0.0F, 1.0F);
            a[2] = CLAMP(a[2], 0.0F, 1.0F);
            a[3] = CLAMP(a[3], 0.0F, 1.0F);
            ubx = IROUND(255.0F * a[0]);
            uby = IROUND(255.0F * a[1]);
            ubz = IROUND(255.0F * a[2]);
            ubw = IROUND(255.0F * a[3]);
            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
               = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_POW:
         {
            GLfloat a[4], b[4], result[4];
            fetch_vector1(&inst->SrcReg[0], machine, a);
            fetch_vector1(&inst->SrcReg[1], machine, b);
            result[0] = result[1] = result[2] = result[3]
               = (GLfloat) _mesa_pow(a[0], b[0]);
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_RCP:
         {
            GLfloat a[4], result[4];
            fetch_vector1(&inst->SrcReg[0], machine, a);
            if (DEBUG_PROG) {
               if (a[0] == 0)
                  printf("RCP(0)\n");
               else if (IS_INF_OR_NAN(a[0]))
                  printf("RCP(inf)\n");
            }
            result[0] = result[1] = result[2] = result[3] = 1.0F / a[0];
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_RET:         /* return from subroutine (conditional) */
         if (eval_condition(machine, inst)) {
            if (machine->StackDepth == 0) {
               return GL_TRUE;  /* Per GL_NV_vertex_program2 spec */
            }
            /* subtract one because of pc++ in the for loop */
            pc = machine->CallStack[--machine->StackDepth] - 1;
         }
         break;
      case OPCODE_RFL:         /* reflection vector */
         {
            GLfloat axis[4], dir[4], result[4], tmpX, tmpW;
            fetch_vector4(&inst->SrcReg[0], machine, axis);
            fetch_vector4(&inst->SrcReg[1], machine, dir);
            tmpW = DOT3(axis, axis);
            tmpX = (2.0F * DOT3(axis, dir)) / tmpW;
            result[0] = tmpX * axis[0] - dir[0];
            result[1] = tmpX * axis[1] - dir[1];
            result[2] = tmpX * axis[2] - dir[2];
            /* result[3] is never written! XXX enforce in parser! */
            store_vector4(inst, machine, result);
         }
         break;
      case OPCODE_RSQ:         /* 1 / sqrt() */
         {
            GLfloat a[4], result[4];
            fetch_vector1(&inst->SrcReg[0], machine, a);
            a[0] = FABSF(a[0]);
            result[0] = result[1] = result[2] = result[3] = INV_SQRTF(a[0]);
            store_vector4(inst, machine, result);
            if (DEBUG_PROG) {
               printf("RSQ %g = 1/sqrt(|%g|)\n", result[0], a[0]);
            }
         }
         break;
      case OPCODE_SCS:         /* sine and cos */
         {