Add MMX shift-by-vector (not correct yet).

git-svn-id: svn://svn.valgrind.org/vex/trunk@572

diff --git a/VEX/priv/guest-x86/gdefs.h b/VEX/priv/guest-x86/gdefs.h
index b859915b336c9a768ac7201460f5c16d22e05ffb..d5c7483c7a636c030966b295ff59f7a3aec876ab 100644 (file)
--- a/VEX/priv/guest-x86/gdefs.h
+++ b/VEX/priv/guest-x86/gdefs.h
@@ -134,6 +134,17 @@ extern ULong calculate_punpcklwd ( ULong, ULong );
 extern ULong calculate_punpckhdq ( ULong, ULong );
 extern ULong calculate_punpckldq ( ULong, ULong );
 
+extern ULong calculate_shl16x4 ( ULong, ULong );
+extern ULong calculate_shl32x2 ( ULong, ULong );
+extern ULong calculate_shl64x1 ( ULong, ULong );
+
+extern ULong calculate_shr16Ux4 ( ULong, ULong );
+extern ULong calculate_shr32Ux2 ( ULong, ULong );
+extern ULong calculate_shr64Ux1 ( ULong, ULong );
+
+extern ULong calculate_shr16Sx4 ( ULong, ULong );
+extern ULong calculate_shr32Sx2 ( ULong, ULong );
+
 
 /* --- DIRTY HELPERS --- */
 

diff --git a/VEX/priv/guest-x86/ghelpers.c b/VEX/priv/guest-x86/ghelpers.c
index b0f4c16d696a2c1b14430a5499de623b7b5f1bfa..875860180c1804d23800936e6ac2fdafd0e98146 100644 (file)
--- a/VEX/priv/guest-x86/ghelpers.c
+++ b/VEX/priv/guest-x86/ghelpers.c
@@ -1836,6 +1836,40 @@ static inline UChar qnarrow16Uto8 ( Int xx )
    return (UChar)xx;
 }
 
+static inline UShort shl16 ( UShort v, UShort n )
+{
+   return n > 15 ? 0 : v << n;
+}
+
+static inline UShort shr16U ( UShort v, UShort n )
+{
+   return n > 15 ? 0 : (((UShort)v) >> n);
+}
+
+static inline UShort shr16S ( UShort v, UShort n )
+{
+   if (n <= 15) 
+      return ((Short)v) >> n;
+   return (v & 0x8000) ? 0xFFFF : 0;
+}
+
+static inline UInt shl32 ( UInt v, UInt n )
+{
+   return n > 31 ? 0 : v << n;
+}
+
+static inline UInt shr32U ( UInt v, UInt n )
+{
+   return n > 31 ? 0 : (((UInt)v) >> n);
+}
+
+static inline UInt shr32S ( UInt v, UInt n )
+{
+   if (n <= 31) 
+      return ((Int)v) >> n;
+   return (v & 0x80000000) ? 0xFFFFFFFF : 0;
+}
+
 
 /* ------------ Normal addition ------------ */
 
@@ -2228,6 +2262,77 @@ ULong calculate_punpckldq ( ULong dst, ULong src )
          );
 }
 
+/* ------------ Shifting ------------ */
+
+ULong calculate_shl16x4 ( ULong xx, ULong yy )
+{
+   return mk16x4(
+             shl16( sel16x4_3(xx), sel16x4_3(yy) ),
+             shl16( sel16x4_2(xx), sel16x4_2(yy) ),
+             shl16( sel16x4_1(xx), sel16x4_1(yy) ),
+             shl16( sel16x4_0(xx), sel16x4_0(yy) )
+          );
+}
+
+ULong calculate_shl32x2 ( ULong xx, ULong yy )
+{
+   return mk32x2(
+             shl32( sel32x2_1(xx), sel32x2_1(yy) ),
+             shl32( sel32x2_0(xx), sel32x2_0(yy) )
+          );
+}
+
+
+ULong calculate_shl64x1 ( ULong xx, ULong yy )
+{
+   if (yy > 63) return 0;
+   return xx << yy;
+}
+
+ULong calculate_shr16Ux4 ( ULong xx, ULong yy )
+{
+   return mk16x4(
+             shr16U( sel16x4_3(xx), sel16x4_3(yy) ),
+             shr16U( sel16x4_2(xx), sel16x4_2(yy) ),
+             shr16U( sel16x4_1(xx), sel16x4_1(yy) ),
+             shr16U( sel16x4_0(xx), sel16x4_0(yy) )
+          );
+}
+
+ULong calculate_shr32Ux2 ( ULong xx, ULong yy )
+{
+   return mk32x2(
+             shr32U( sel32x2_1(xx), sel32x2_1(yy) ),
+             shr32U( sel32x2_0(xx), sel32x2_0(yy) )
+          );
+}
+
+ULong calculate_shr64Ux1 ( ULong xx, ULong yy )
+{
+   if (yy > 63) return 0;
+   return xx >> yy;
+}
+
+
+ULong calculate_shr16Sx4 ( ULong xx, ULong yy )
+{
+   return mk16x4(
+             shr16S( sel16x4_3(xx), sel16x4_3(yy) ),
+             shr16S( sel16x4_2(xx), sel16x4_2(yy) ),
+             shr16S( sel16x4_1(xx), sel16x4_1(yy) ),
+             shr16S( sel16x4_0(xx), sel16x4_0(yy) )
+          );
+}
+
+ULong calculate_shr32Sx2 ( ULong xx, ULong yy )
+{
+   return mk32x2(
+             shr32S( sel32x2_1(xx), sel32x2_1(yy) ),
+             shr32S( sel32x2_0(xx), sel32x2_0(yy) )
+          );
+}
+
+
 
 /*-----------------------------------------------------------*/
 /*--- Describing the x86 guest state, for the benefit     ---*/

diff --git a/VEX/priv/guest-x86/toIR.c b/VEX/priv/guest-x86/toIR.c
index 1d4c541b07faa11087f22cc8934ee2b8e21ca853..a3fbdd24067fbce3b78cf6b407e570cd9dfa2f32 100644 (file)
--- a/VEX/priv/guest-x86/toIR.c
+++ b/VEX/priv/guest-x86/toIR.c
@@ -4692,6 +4692,17 @@ UInt dis_MMXop_regmem_to_reg ( UChar sorb,
       case 0x61: XXX(calculate_punpcklwd); break;
       case 0x62: XXX(calculate_punpckldq); break;
 
+      case 0xF1: XXX(calculate_shl16x4); break;
+      case 0xF2: XXX(calculate_shl32x2); break;
+      case 0xF3: XXX(calculate_shl64x1); break;
+
+      case 0xD1: XXX(calculate_shr16Ux4); break;
+      case 0xD2: XXX(calculate_shr32Ux2); break;
+      case 0xD3: XXX(calculate_shr64Ux1); break;
+
+      case 0xE1: XXX(calculate_shr16Sx4); break;
+      case 0xE2: XXX(calculate_shr32Sx2); break;
+
       case 0xDB: break; /* AND */
       case 0xDF: break; /* ANDN */
       case 0xEB: break; /* OR */
@@ -4929,6 +4940,26 @@ UInt dis_MMX ( Bool* decode_ok, UChar sorb, Int sz, UInt delta )
          delta = dis_MMXop_regmem_to_reg ( sorb, delta, opc, "pxor", False );
          break;
 
+      case 0xF1:
+      case 0xF2:
+      case 0xF3: /* PSLLgg (src)mmxreg-or-mem, (dst)mmxreg */
+         vassert(sz == 4);
+         delta = dis_MMXop_regmem_to_reg ( sorb, delta, opc, "psll", True );
+         break;
+
+      case 0xD1:
+      case 0xD2:
+      case 0xD3: /* PSRLgg (src)mmxreg-or-mem, (dst)mmxreg */
+         vassert(sz == 4);
+         delta = dis_MMXop_regmem_to_reg ( sorb, delta, opc, "psrl", True );
+         break;
+
+      case 0xE1: 
+      case 0xE2: /* PSRAgg (src)mmxreg-or-mem, (dst)mmxreg */
+         vassert(sz == 4);
+         delta = dis_MMXop_regmem_to_reg ( sorb, delta, opc, "psra", True );
+         break;
+
       default:
          *decode_ok = False;
          return delta; /* ignored */
@@ -9478,12 +9509,6 @@ static DisResult disInstr ( /*IN*/  Bool    resteerOK,
 //--          break;
 //--       }
 //-- 
-//--       case 0x77: /* EMMS */
-//--          vg_assert(sz == 4);
-//--          uInstr1(cb, MMX1, 0, Lit16, ((UShort)(opc)) );
-//--          DIP("emms\n");
-//--          break;
-//-- 
 //--       case 0x7E: /* MOVD (src)mmxreg, (dst)ireg-or-mem */
 //--          vg_assert(sz == 4);
 //--          modrm = getUChar(eip);
@@ -9589,6 +9614,17 @@ static DisResult disInstr ( /*IN*/  Bool    resteerOK,
       case 0xEB: /* POR (src)mmxreg-or-mem, (dst)mmxreg */
       case 0xEF: /* PXOR (src)mmxreg-or-mem, (dst)mmxreg */
 
+      case 0xF1: 
+      case 0xF2: 
+      case 0xF3: /* PSLLgg (src)mmxreg-or-mem, (dst)mmxreg */
+
+      case 0xD1: 
+      case 0xD2: 
+      case 0xD3: /* PSRLgg (src)mmxreg-or-mem, (dst)mmxreg */
+
+      case 0xE1: 
+      case 0xE2: /* PSRAgg (src)mmxreg-or-mem, (dst)mmxreg */
+
       case 0x6F: /* MOVQ (src)mmxreg-or-mem, (dst)mmxreg */
       {
          UInt delta0    = delta-1;
@@ -9601,6 +9637,12 @@ static DisResult disInstr ( /*IN*/  Bool    resteerOK,
          break;
       }
 
+      case 0x77: /* EMMS */
+         vassert(sz == 4);
+         do_EMMS_boilerplate();
+         DIP("emms\n");
+         break;
+
 //--       case 0x7F: /* MOVQ (src)mmxreg, (dst)mmxreg-or-mem */
 //--       case 0xE7: /* MOVNTQ (src)mmxreg, (dst)mmxreg-or-mem */
 //--          vg_assert(sz == 4);