X86InstrInfo.td

                     [(set R32:$dst, (mul R32:$src1, i32immSExt8:$src2))]>;

def IMUL16rmi  : Ii16<0x69, MRMSrcMem,                      // R16 = [mem16]*I16
                      (ops R16:$dst, i16mem:$src1, i16imm:$src2),
                      "imul{w} {$src2, $src1, $dst|$dst, $src1, $src2}",
                      [(set R16:$dst, (mul (load addr:$src1), imm:$src2))]>,
                      OpSize;
def IMUL32rmi  : Ii32<0x69, MRMSrcMem,                      // R32 = [mem32]*I32
                      (ops R32:$dst, i32mem:$src1, i32imm:$src2),
                      "imul{l} {$src2, $src1, $dst|$dst, $src1, $src2}",
                      [(set R32:$dst, (mul (load addr:$src1), imm:$src2))]>;
def IMUL16rmi8 : Ii8<0x6B, MRMSrcMem,                       // R16 = [mem16]*I8
                     (ops R16:$dst, i16mem:$src1, i16i8imm :$src2),
                     "imul{w} {$src2, $src1, $dst|$dst, $src1, $src2}",
                  [(set R16:$dst, (mul (load addr:$src1), i16immSExt8:$src2))]>,
                     OpSize;
def IMUL32rmi8 : Ii8<0x6B, MRMSrcMem,                       // R32 = [mem32]*I8
                     (ops R32:$dst, i32mem:$src1, i32i8imm: $src2),
                     "imul{l} {$src2, $src1, $dst|$dst, $src1, $src2}",
                  [(set R32:$dst, (mul (load addr:$src1), i32immSExt8:$src2))]>;

//===----------------------------------------------------------------------===//
// Test instructions are just like AND, except they don't generate a result.
//
let isCommutable = 1 in {   // TEST X, Y   --> TEST Y, X
def TEST8rr  : I<0x84, MRMDestReg, (ops R8:$src1, R8:$src2),
                 "test{b} {$src2, $src1|$src1, $src2}",
                 [(X86test R8:$src1, R8:$src2)]>;
def TEST16rr : I<0x85, MRMDestReg, (ops R16:$src1, R16:$src2),
                 "test{w} {$src2, $src1|$src1, $src2}",
                 [(X86test R16:$src1, R16:$src2)]>, OpSize;
def TEST32rr : I<0x85, MRMDestReg, (ops R32:$src1, R32:$src2),
                 "test{l} {$src2, $src1|$src1, $src2}",
                 [(X86test R32:$src1, R32:$src2)]>;
}
def TEST8mr  : I<0x84, MRMDestMem, (ops i8mem :$src1, R8 :$src2),
                 "test{b} {$src2, $src1|$src1, $src2}",
                 [(X86test (loadi8 addr:$src1), R8:$src2)]>;
def TEST16mr : I<0x85, MRMDestMem, (ops i16mem:$src1, R16:$src2),
                 "test{w} {$src2, $src1|$src1, $src2}",
                 [(X86test (loadi16 addr:$src1), R16:$src2)]>,
               OpSize;
def TEST32mr : I<0x85, MRMDestMem, (ops i32mem:$src1, R32:$src2),
                 "test{l} {$src2, $src1|$src1, $src2}",
                 [(X86test (loadi32 addr:$src1), R32:$src2)]>;
def TEST8rm  : I<0x84, MRMSrcMem, (ops R8 :$src1, i8mem :$src2),
                 "test{b} {$src2, $src1|$src1, $src2}",
                 [(X86test R8:$src1, (loadi8 addr:$src2))]>;
def TEST16rm : I<0x85, MRMSrcMem, (ops R16:$src1, i16mem:$src2),
                 "test{w} {$src2, $src1|$src1, $src2}",
                 [(X86test R16:$src1, (loadi16 addr:$src2))]>,
               OpSize;
def TEST32rm : I<0x85, MRMSrcMem, (ops R32:$src1, i32mem:$src2),
                 "test{l} {$src2, $src1|$src1, $src2}",
                 [(X86test R32:$src1, (loadi32 addr:$src2))]>;

def TEST8ri  : Ii8 <0xF6, MRM0r,                     // flags = R8  & imm8
                    (ops R8:$src1, i8imm:$src2),
                    "test{b} {$src2, $src1|$src1, $src2}",
                    [(X86test R8:$src1, imm:$src2)]>;
def TEST16ri : Ii16<0xF7, MRM0r,                     // flags = R16 & imm16
                    (ops R16:$src1, i16imm:$src2),
                    "test{w} {$src2, $src1|$src1, $src2}",
                    [(X86test R16:$src1, imm:$src2)]>, OpSize;
def TEST32ri : Ii32<0xF7, MRM0r,                     // flags = R32 & imm32
                    (ops R32:$src1, i32imm:$src2),
                    "test{l} {$src2, $src1|$src1, $src2}",
                    [(X86test R32:$src1, imm:$src2)]>;
def TEST8mi  : Ii8 <0xF6, MRM0m,                     // flags = [mem8]  & imm8
                    (ops i8mem:$src1, i8imm:$src2),
                    "test{b} {$src2, $src1|$src1, $src2}",
                    [(X86test (loadi8 addr:$src1), imm:$src2)]>;
def TEST16mi : Ii16<0xF7, MRM0m,                     // flags = [mem16] & imm16
                    (ops i16mem:$src1, i16imm:$src2),
                    "test{w} {$src2, $src1|$src1, $src2}",
                    [(X86test (loadi16 addr:$src1), imm:$src2)]>,
               OpSize;
def TEST32mi : Ii32<0xF7, MRM0m,                     // flags = [mem32] & imm32
                    (ops i32mem:$src1, i32imm:$src2),
                    "test{l} {$src2, $src1|$src1, $src2}",
                    [(X86test (loadi32 addr:$src1), imm:$src2)]>;


// Condition code ops, incl. set if equal/not equal/...
def SAHF     : I<0x9E, RawFrm, (ops), "sahf", []>, Imp<[AH],[]>;  // flags = AH
def LAHF     : I<0x9F, RawFrm, (ops), "lahf", []>, Imp<[],[AH]>;  // AH = flags

def SETEr    : I<0x94, MRM0r, 
                 (ops R8   :$dst),
                 "sete $dst",
                 [(set R8:$dst, (X86setcc X86_COND_E))]>,
               TB;                        // R8 = ==
def SETEm    : I<0x94, MRM0m, 
                 (ops i8mem:$dst),
                 "sete $dst",
                 [(store (X86setcc X86_COND_E), addr:$dst)]>,
               TB;                        // [mem8] = ==
def SETNEr   : I<0x95, MRM0r, 
                 (ops R8   :$dst),
                 "setne $dst",
                 [(set R8:$dst, (X86setcc X86_COND_NE))]>,
               TB;                        // R8 = !=
def SETNEm   : I<0x95, MRM0m, 
                 (ops i8mem:$dst),
                 "setne $dst",
                 [(store (X86setcc X86_COND_NE), addr:$dst)]>,
               TB;                        // [mem8] = !=
def SETLr    : I<0x9C, MRM0r, 
                 (ops R8   :$dst),
                 "setl $dst",
                 [(set R8:$dst, (X86setcc X86_COND_L))]>,
               TB;                        // R8 = <  signed
def SETLm    : I<0x9C, MRM0m, 
                 (ops i8mem:$dst),
                 "setl $dst",
                 [(store (X86setcc X86_COND_L), addr:$dst)]>,
               TB;                        // [mem8] = <  signed
def SETGEr   : I<0x9D, MRM0r, 
                 (ops R8   :$dst),
                 "setge $dst",
                 [(set R8:$dst, (X86setcc X86_COND_GE))]>,
               TB;                        // R8 = >= signed
def SETGEm   : I<0x9D, MRM0m, 
                 (ops i8mem:$dst),
                 "setge $dst",
                 [(store (X86setcc X86_COND_GE), addr:$dst)]>,
               TB;                        // [mem8] = >= signed
def SETLEr   : I<0x9E, MRM0r, 
                 (ops R8   :$dst),
                 "setle $dst",
                 [(set R8:$dst, (X86setcc X86_COND_LE))]>,
               TB;                        // R8 = <= signed
def SETLEm   : I<0x9E, MRM0m, 
                 (ops i8mem:$dst),
                 "setle $dst",
                 [(store (X86setcc X86_COND_LE), addr:$dst)]>,
               TB;                        // [mem8] = <= signed
def SETGr    : I<0x9F, MRM0r, 
                 (ops R8   :$dst),
                 "setg $dst",
                 [(set R8:$dst, (X86setcc X86_COND_G))]>,
               TB;                        // R8 = >  signed
def SETGm    : I<0x9F, MRM0m, 
                 (ops i8mem:$dst),
                 "setg $dst",
                 [(store (X86setcc X86_COND_G), addr:$dst)]>,
               TB;                        // [mem8] = >  signed

def SETBr    : I<0x92, MRM0r,
                 (ops R8   :$dst),
                 "setb $dst",
                 [(set R8:$dst, (X86setcc X86_COND_B))]>,
               TB;                        // R8 = <  unsign
def SETBm    : I<0x92, MRM0m,
                 (ops i8mem:$dst),
                 "setb $dst",
                 [(store (X86setcc X86_COND_B), addr:$dst)]>,
               TB;                        // [mem8] = <  unsign
def SETAEr   : I<0x93, MRM0r, 
                 (ops R8   :$dst),
                 "setae $dst",
                 [(set R8:$dst, (X86setcc X86_COND_AE))]>,
               TB;                        // R8 = >= unsign
def SETAEm   : I<0x93, MRM0m, 
                 (ops i8mem:$dst),
                 "setae $dst",
                 [(store (X86setcc X86_COND_AE), addr:$dst)]>,
               TB;                        // [mem8] = >= unsign
def SETBEr   : I<0x96, MRM0r, 
                 (ops R8   :$dst),
                 "setbe $dst",
                 [(set R8:$dst, (X86setcc X86_COND_BE))]>,
               TB;                        // R8 = <= unsign
def SETBEm   : I<0x96, MRM0m, 
                 (ops i8mem:$dst),
                 "setbe $dst",
                 [(store (X86setcc X86_COND_BE), addr:$dst)]>,
               TB;                        // [mem8] = <= unsign
def SETAr    : I<0x97, MRM0r, 
                 (ops R8   :$dst),
                 "seta $dst",
                 [(set R8:$dst, (X86setcc X86_COND_A))]>,
               TB;                        // R8 = >  signed
def SETAm    : I<0x97, MRM0m, 
                 (ops i8mem:$dst),
                 "seta $dst",
                 [(store (X86setcc X86_COND_A), addr:$dst)]>,
               TB;                        // [mem8] = >  signed

def SETSr    : I<0x98, MRM0r, 
                 (ops R8   :$dst),
                 "sets $dst",
                 [(set R8:$dst, (X86setcc X86_COND_S))]>,
               TB;                        // R8 = <sign bit>
def SETSm    : I<0x98, MRM0m, 
                 (ops i8mem:$dst),
                 "sets $dst",
                 [(store (X86setcc X86_COND_S), addr:$dst)]>,
               TB;                        // [mem8] = <sign bit>
def SETNSr   : I<0x99, MRM0r, 
                 (ops R8   :$dst),
                 "setns $dst",
                 [(set R8:$dst, (X86setcc X86_COND_NS))]>,
               TB;                        // R8 = !<sign bit>
def SETNSm   : I<0x99, MRM0m, 
                 (ops i8mem:$dst),
                 "setns $dst",
                 [(store (X86setcc X86_COND_NS), addr:$dst)]>,
               TB;                        // [mem8] = !<sign bit>
def SETPr    : I<0x9A, MRM0r, 
                 (ops R8   :$dst),
                 "setp $dst",
                 [(set R8:$dst, (X86setcc X86_COND_P))]>,
               TB;                        // R8 = parity
def SETPm    : I<0x9A, MRM0m, 
                 (ops i8mem:$dst),
                 "setp $dst",
                 [(store (X86setcc X86_COND_P), addr:$dst)]>,
               TB;                        // [mem8] = parity
def SETNPr   : I<0x9B, MRM0r, 
                 (ops R8   :$dst),
                 "setnp $dst",
                 [(set R8:$dst, (X86setcc X86_COND_NP))]>,
               TB;                        // R8 = not parity
def SETNPm   : I<0x9B, MRM0m, 
                 (ops i8mem:$dst),
                 "setnp $dst",
                 [(store (X86setcc X86_COND_NP), addr:$dst)]>,
               TB;                        // [mem8] = not parity

// Integer comparisons
def CMP8rr  : I<0x38, MRMDestReg,
                (ops R8 :$src1, R8 :$src2),
                "cmp{b} {$src2, $src1|$src1, $src2}",
                [(X86cmp R8:$src1, R8:$src2)]>;
def CMP16rr : I<0x39, MRMDestReg,
                (ops R16:$src1, R16:$src2),
                "cmp{w} {$src2, $src1|$src1, $src2}",
                [(X86cmp R16:$src1, R16:$src2)]>, OpSize;
def CMP32rr : I<0x39, MRMDestReg,
                (ops R32:$src1, R32:$src2),
                "cmp{l} {$src2, $src1|$src1, $src2}",
                [(X86cmp R32:$src1, R32:$src2)]>;
def CMP8mr  : I<0x38, MRMDestMem,
                (ops i8mem :$src1, R8 :$src2),
                "cmp{b} {$src2, $src1|$src1, $src2}",
                [(X86cmp (loadi8 addr:$src1), R8:$src2)]>;
def CMP16mr : I<0x39, MRMDestMem,
                (ops i16mem:$src1, R16:$src2),
                "cmp{w} {$src2, $src1|$src1, $src2}",
                [(X86cmp (loadi16 addr:$src1), R16:$src2)]>, OpSize;
def CMP32mr : I<0x39, MRMDestMem,
                (ops i32mem:$src1, R32:$src2),
                "cmp{l} {$src2, $src1|$src1, $src2}",
                [(X86cmp (loadi32 addr:$src1), R32:$src2)]>;
def CMP8rm  : I<0x3A, MRMSrcMem,
                (ops R8 :$src1, i8mem :$src2),
                "cmp{b} {$src2, $src1|$src1, $src2}",
                [(X86cmp R8:$src1, (loadi8 addr:$src2))]>;
def CMP16rm : I<0x3B, MRMSrcMem,
                (ops R16:$src1, i16mem:$src2),
                "cmp{w} {$src2, $src1|$src1, $src2}",
                [(X86cmp R16:$src1, (loadi16 addr:$src2))]>, OpSize;
def CMP32rm : I<0x3B, MRMSrcMem,
                (ops R32:$src1, i32mem:$src2),
                "cmp{l} {$src2, $src1|$src1, $src2}",
                [(X86cmp R32:$src1, (loadi32 addr:$src2))]>;
def CMP8ri  : Ii8<0x80, MRM7r,
                  (ops R8:$src1, i8imm:$src2),
                  "cmp{b} {$src2, $src1|$src1, $src2}",
                  [(X86cmp R8:$src1, imm:$src2)]>;
def CMP16ri : Ii16<0x81, MRM7r,
                   (ops R16:$src1, i16imm:$src2),
                   "cmp{w} {$src2, $src1|$src1, $src2}",
                   [(X86cmp R16:$src1, imm:$src2)]>, OpSize;
def CMP32ri : Ii32<0x81, MRM7r,
                   (ops R32:$src1, i32imm:$src2),
                   "cmp{l} {$src2, $src1|$src1, $src2}",
                   [(X86cmp R32:$src1, imm:$src2)]>;
def CMP8mi  : Ii8 <0x80, MRM7m,
                   (ops i8mem :$src1, i8imm :$src2),
                   "cmp{b} {$src2, $src1|$src1, $src2}",
                   [(X86cmp (loadi8 addr:$src1), imm:$src2)]>;
def CMP16mi : Ii16<0x81, MRM7m,
                   (ops i16mem:$src1, i16imm:$src2),
                   "cmp{w} {$src2, $src1|$src1, $src2}",
                   [(X86cmp (loadi16 addr:$src1), imm:$src2)]>, OpSize;
def CMP32mi : Ii32<0x81, MRM7m,
                   (ops i32mem:$src1, i32imm:$src2),
                   "cmp{l} {$src2, $src1|$src1, $src2}",
                   [(X86cmp (loadi32 addr:$src1), imm:$src2)]>;

// Sign/Zero extenders
def MOVSX16rr8 : I<0xBE, MRMSrcReg, (ops R16:$dst, R8 :$src),
                   "movs{bw|x} {$src, $dst|$dst, $src}",
                   [(set R16:$dst, (sext R8:$src))]>, TB, OpSize;
def MOVSX16rm8 : I<0xBE, MRMSrcMem, (ops R16:$dst, i8mem :$src),
                   "movs{bw|x} {$src, $dst|$dst, $src}",
                   [(set R16:$dst, (sextloadi16i8 addr:$src))]>, TB, OpSize;
def MOVSX32rr8 : I<0xBE, MRMSrcReg, (ops R32:$dst, R8 :$src),
                   "movs{bl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (sext R8:$src))]>, TB;
def MOVSX32rm8 : I<0xBE, MRMSrcMem, (ops R32:$dst, i8mem :$src),
                   "movs{bl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (sextloadi32i8 addr:$src))]>, TB;
def MOVSX32rr16: I<0xBF, MRMSrcReg, (ops R32:$dst, R16:$src),
                   "movs{wl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (sext R16:$src))]>, TB;
def MOVSX32rm16: I<0xBF, MRMSrcMem, (ops R32:$dst, i16mem:$src),
                   "movs{wl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (sextloadi32i16 addr:$src))]>, TB;

def MOVZX16rr8 : I<0xB6, MRMSrcReg, (ops R16:$dst, R8 :$src),
                   "movz{bw|x} {$src, $dst|$dst, $src}",
                   [(set R16:$dst, (zext R8:$src))]>, TB, OpSize;
def MOVZX16rm8 : I<0xB6, MRMSrcMem, (ops R16:$dst, i8mem :$src),
                   "movz{bw|x} {$src, $dst|$dst, $src}",
                   [(set R16:$dst, (zextloadi16i8 addr:$src))]>, TB, OpSize;
def MOVZX32rr8 : I<0xB6, MRMSrcReg, (ops R32:$dst, R8 :$src),
                   "movz{bl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (zext R8:$src))]>, TB;
def MOVZX32rm8 : I<0xB6, MRMSrcMem, (ops R32:$dst, i8mem :$src),
                   "movz{bl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (zextloadi32i8 addr:$src))]>, TB;
def MOVZX32rr16: I<0xB7, MRMSrcReg, (ops R32:$dst, R16:$src),
                   "movz{wl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (zext R16:$src))]>, TB;
def MOVZX32rm16: I<0xB7, MRMSrcMem, (ops R32:$dst, i16mem:$src),
                   "movz{wl|x} {$src, $dst|$dst, $src}",
                   [(set R32:$dst, (zextloadi32i16 addr:$src))]>, TB;

//===----------------------------------------------------------------------===//
// Miscellaneous Instructions
//===----------------------------------------------------------------------===//

def RDTSC : I<0x31, RawFrm, (ops), "rdtsc", [(X86rdtsc)]>,
            TB, Imp<[],[EAX,EDX]>;

//===----------------------------------------------------------------------===//
// Alias Instructions
//===----------------------------------------------------------------------===//

// Alias instructions that map movr0 to xor.
// FIXME: remove when we can teach regalloc that xor reg, reg is ok.
def MOV8r0   : I<0x30, MRMInitReg, (ops R8 :$dst),
                 "xor{b} $dst, $dst",
                 [(set R8:$dst, 0)]>;
def MOV16r0  : I<0x31, MRMInitReg,  (ops R16:$dst), 
                 "xor{w} $dst, $dst",
                 [(set R16:$dst, 0)]>, OpSize;
def MOV32r0  : I<0x31, MRMInitReg,  (ops R32:$dst), 
                 "xor{l} $dst, $dst",
                 [(set R32:$dst, 0)]>;

//===----------------------------------------------------------------------===//
// Non-Instruction Patterns
//===----------------------------------------------------------------------===//

// ConstantPool GlobalAddress, ExternalSymbol
def : Pat<(i32 (X86Wrapper tconstpool  :$dst)), (MOV32ri tconstpool  :$dst)>;
def : Pat<(i32 (X86Wrapper tglobaladdr :$dst)), (MOV32ri tglobaladdr :$dst)>;
def : Pat<(i32 (X86Wrapper texternalsym:$dst)), (MOV32ri texternalsym:$dst)>;

def : Pat<(add R32:$src1, (X86Wrapper tconstpool:$src2)),
          (ADD32ri R32:$src1, tconstpool:$src2)>;
def : Pat<(add R32:$src1, (X86Wrapper tglobaladdr :$src2)),
          (ADD32ri R32:$src1, tglobaladdr:$src2)>;
def : Pat<(add R32:$src1, (X86Wrapper texternalsym:$src2)),
          (ADD32ri R32:$src1, texternalsym:$src2)>;

def : Pat<(store (X86Wrapper tconstpool:$src), addr:$dst),
          (MOV32mi addr:$dst, tconstpool:$src)>;
def : Pat<(store (X86Wrapper tglobaladdr:$src), addr:$dst),
          (MOV32mi addr:$dst, tglobaladdr:$src)>;
def : Pat<(store (X86Wrapper texternalsym:$src), addr:$dst),
          (MOV32mi addr:$dst, texternalsym:$src)>;

// Calls
def : Pat<(X86call tglobaladdr:$dst),
          (CALLpcrel32 tglobaladdr:$dst)>;
def : Pat<(X86call texternalsym:$dst),
          (CALLpcrel32 texternalsym:$dst)>;

// X86 specific add which produces a flag.
def : Pat<(addc R32:$src1, R32:$src2),
          (ADD32rr R32:$src1, R32:$src2)>;
def : Pat<(addc R32:$src1, (load addr:$src2)),
          (ADD32rm R32:$src1, addr:$src2)>;
def : Pat<(addc R32:$src1, imm:$src2),
          (ADD32ri R32:$src1, imm:$src2)>;
def : Pat<(addc R32:$src1, i32immSExt8:$src2),
          (ADD32ri8 R32:$src1, i32immSExt8:$src2)>;

def : Pat<(subc R32:$src1, R32:$src2),
          (SUB32rr R32:$src1, R32:$src2)>;
def : Pat<(subc R32:$src1, (load addr:$src2)),
          (SUB32rm R32:$src1, addr:$src2)>;
def : Pat<(subc R32:$src1, imm:$src2),
          (SUB32ri R32:$src1, imm:$src2)>;
def : Pat<(subc R32:$src1, i32immSExt8:$src2),
          (SUB32ri8 R32:$src1, i32immSExt8:$src2)>;

def : Pat<(truncstore (i8 imm:$src), addr:$dst, i1), 
          (MOV8mi addr:$dst, imm:$src)>;
def : Pat<(truncstore R8:$src, addr:$dst, i1), 
          (MOV8mr addr:$dst, R8:$src)>;

// {s|z}extload bool -> {s|z}extload byte
def : Pat<(sextloadi16i1 addr:$src), (MOVSX16rm8 addr:$src)>;
def : Pat<(sextloadi32i1 addr:$src), (MOVSX32rm8 addr:$src)>;
def : Pat<(zextloadi8i1  addr:$src), (MOV8rm     addr:$src)>;
def : Pat<(zextloadi16i1 addr:$src), (MOVZX16rm8 addr:$src)>;
def : Pat<(zextloadi32i1 addr:$src), (MOVZX32rm8 addr:$src)>;

// extload bool -> extload byte
def : Pat<(extloadi8i1 addr:$src), (MOV8rm addr:$src)>;

// anyext -> zext
def : Pat<(i16 (anyext R8 :$src)), (MOVZX16rr8  R8 :$src)>;
def : Pat<(i32 (anyext R8 :$src)), (MOVZX32rr8  R8 :$src)>;
def : Pat<(i32 (anyext R16:$src)), (MOVZX32rr16 R16:$src)>;

//===----------------------------------------------------------------------===//
// Some peepholes
//===----------------------------------------------------------------------===//

// (shl x, 1) ==> (add x, x)
def : Pat<(shl R8 :$src1, (i8 1)), (ADD8rr  R8 :$src1, R8 :$src1)>;
def : Pat<(shl R16:$src1, (i8 1)), (ADD16rr R16:$src1, R16:$src1)>;
def : Pat<(shl R32:$src1, (i8 1)), (ADD32rr R32:$src1, R32:$src1)>;

// (or (x >> c) | (y << (32 - c))) ==> (shrd32 x, y, c)
def : Pat<(or (srl R32:$src1, CL:$amt),
              (shl R32:$src2, (sub 32, CL:$amt))),
          (SHRD32rrCL R32:$src1, R32:$src2)>;

def : Pat<(store (or (srl (loadi32 addr:$dst), CL:$amt),
                     (shl R32:$src2, (sub 32, CL:$amt))), addr:$dst),
          (SHRD32mrCL addr:$dst, R32:$src2)>;

// (or (x << c) | (y >> (32 - c))) ==> (shld32 x, y, c)
def : Pat<(or (shl R32:$src1, CL:$amt),
              (srl R32:$src2, (sub 32, CL:$amt))),
          (SHLD32rrCL R32:$src1, R32:$src2)>;

def : Pat<(store (or (shl (loadi32 addr:$dst), CL:$amt),
                     (srl R32:$src2, (sub 32, CL:$amt))), addr:$dst),
          (SHLD32mrCL addr:$dst, R32:$src2)>;

// (or (x >> c) | (y << (16 - c))) ==> (shrd16 x, y, c)
def : Pat<(or (srl R16:$src1, CL:$amt),
              (shl R16:$src2, (sub 16, CL:$amt))),
          (SHRD16rrCL R16:$src1, R16:$src2)>;

def : Pat<(store (or (srl (loadi16 addr:$dst), CL:$amt),
                     (shl R16:$src2, (sub 16, CL:$amt))), addr:$dst),
          (SHRD16mrCL addr:$dst, R16:$src2)>;

// (or (x << c) | (y >> (16 - c))) ==> (shld16 x, y, c)
def : Pat<(or (shl R16:$src1, CL:$amt),
              (srl R16:$src2, (sub 16, CL:$amt))),
          (SHLD16rrCL R16:$src1, R16:$src2)>;

def : Pat<(store (or (shl (loadi16 addr:$dst), CL:$amt),
                     (srl R16:$src2, (sub 16, CL:$amt))), addr:$dst),
          (SHLD16mrCL addr:$dst, R16:$src2)>;


//===----------------------------------------------------------------------===//
// Floating Point Stack Support
//===----------------------------------------------------------------------===//

include "X86InstrFPStack.td"

//===----------------------------------------------------------------------===//
// MMX and XMM Packed Integer support (requires MMX, SSE, and SSE2)
//===----------------------------------------------------------------------===//

include "X86InstrMMX.td"

//===----------------------------------------------------------------------===//
// XMM Floating point support (requires SSE / SSE2)
//===----------------------------------------------------------------------===//

include "X86InstrSSE.td"