update to ida 7.6, add builds

2021-10-31 21:20:46 +02:00
parent e0e0f2be99
commit b1809fe2d9
1408 changed files with 279193 additions and 302468 deletions
--- a/idasdk76/module/pdp11/float.c
+++ b/idasdk76/module/pdp11/float.c
@@ -0,0 +1,86 @@
+#include <ieee.h>
+#define E IEEE_E
+#define M IEEE_M
+#define EXONE IEEE_EXONE
+
+// one of the many possible NaNs,
+// the one produced by the base converter
+const fpvalue_t ieee_nan = {{ 0, 0, 0, 0, 0xc000, 0xffff }};
+
+fpvalue_error_t idaapi realcvt(void *m, fpvalue_t *e, uint16 swt)
+{
+  eNI x;
+  long exp;
+  uint16 r, msk, *p = (uint16 *)m;
+
+  switch ( swt )
+  {
+    case 1:            // in float
+    case 0:            // in trunc. float
+    case 3:            // in double
+      ecleaz(x);
+      x[0] = ((r = *p) & 0x8000) ? 0xffff : 0;  /* fill in our sign */
+      r &= (0377 << 7);
+      if ( r == 0 )
+      {
+        if ( x[0] ) // negative and zero exponent = Undefined -> transformed to NaN
+        {
+          memcpy(e, &ieee_nan, sizeof(fpvalue_t));
+          break;
+        }
+        // complain on dirty zero
+        for ( msk = 0x7fff, r = swt + 1; r; r--, msk = 0xffff )
+          if ( *p++ & msk )
+            return REAL_ERROR_BADDATA; // signed zero accepted
+      }
+      else
+      {
+        x[E] = (r >> 7) + (EXONE - 0201); /* DEC exponent offset */
+        x[M] = (*p++ & 0177) | 0200; /* now do the high order mantissa */
+        if ( swt )
+          memcpy(&x[M+1], p, swt*sizeof(uint16));
+        eshift(x, -8);  /* shift our mantissa down 8 bits */
+      }
+      emovo(x, e);  // convert to IEEE format
+      break;
+
+    case 011:          // out float
+    case 010:          // out trunc. float
+    case 013:          // out double
+      emovi(*e, x);
+      r = swt & 3;
+      if ( !x[E] )
+        goto clear;
+      if ( x[E] == E_SPECIAL_EXP )
+      {
+        // all of NaN, +/-Inf transformed to "Undefined value"
+        x[0] = 0xffff;
+        x[E] = 0;
+        goto clear;
+      }
+      exp = (long )x[E] - (EXONE - 0201);
+      if ( !emdnorm(x, 0, 0, exp, 56) || x[E] >= 0377 )
+        return REAL_ERROR_RANGE; // overfloat
+      if ( !x[E] )
+      {
+clear:
+        memset(p, 0, (r + 1)*sizeof(uint16));
+        if ( x[0] )
+          *p = 0x8000; // signed zero accepted
+      }
+      else
+      {
+        *p = (x[0] ? 0100000 : 0) | (x[E] << 7);  /* sign & exp */
+        eshift(x, 8);
+        *p++ |= (x[M] & 0177);
+        if ( r != 0 )
+          memcpy(p, &x[M+1], r*sizeof(uint16));
+      }
+      break;
+
+    default:
+      return REAL_ERROR_FORMAT;
+  }
+
+  return REAL_ERROR_OK;
+}