// Use movzbl instead of movzbq when the destination is a register; it's
// equivalent due to implicit zero-extending, and it has a smaller encoding.
def MOVZX64rr8 : I<0xB6, MRMSrcReg, (outs GR64:$dst), (ins GR8 :$src),
- "movz{bl|x}\t{$src, ${dst:subreg32}|${dst:subreg32}, $src}",
- [(set GR64:$dst, (zext GR8:$src))]>, TB;
+ "", [(set GR64:$dst, (zext GR8:$src))]>, TB;
def MOVZX64rm8 : I<0xB6, MRMSrcMem, (outs GR64:$dst), (ins i8mem :$src),
- "movz{bl|x}\t{$src, ${dst:subreg32}|${dst:subreg32}, $src}",
- [(set GR64:$dst, (zextloadi64i8 addr:$src))]>, TB;
+ "", [(set GR64:$dst, (zextloadi64i8 addr:$src))]>, TB;
// Use movzwl instead of movzwq when the destination is a register; it's
// equivalent due to implicit zero-extending, and it has a smaller encoding.
def MOVZX64rr16: I<0xB7, MRMSrcReg, (outs GR64:$dst), (ins GR16:$src),
- "movz{wl|x}\t{$src, ${dst:subreg32}|${dst:subreg32}, $src}",
- [(set GR64:$dst, (zext GR16:$src))]>, TB;
+ "", [(set GR64:$dst, (zext GR16:$src))]>, TB;
def MOVZX64rm16: I<0xB7, MRMSrcMem, (outs GR64:$dst), (ins i16mem:$src),
- "movz{wl|x}\t{$src, ${dst:subreg32}|${dst:subreg32}, $src}",
- [(set GR64:$dst, (zextloadi64i16 addr:$src))]>, TB;
+ "", [(set GR64:$dst, (zextloadi64i16 addr:$src))]>, TB;
// There's no movzlq instruction, but movl can be used for this purpose, using
// implicit zero-extension. The preferred way to do 32-bit-to-64-bit zero
// necessarily all zero. In such cases, we fall back to these explicit zext
// instructions.
def MOVZX64rr32 : I<0x89, MRMDestReg, (outs GR64:$dst), (ins GR32:$src),
- "mov{l}\t{$src, ${dst:subreg32}|${dst:subreg32}, $src}",
- [(set GR64:$dst, (zext GR32:$src))]>;
+ "", [(set GR64:$dst, (zext GR32:$src))]>;
def MOVZX64rm32 : I<0x8B, MRMSrcMem, (outs GR64:$dst), (ins i32mem:$src),
- "mov{l}\t{$src, ${dst:subreg32}|${dst:subreg32}, $src}",
- [(set GR64:$dst, (zextloadi64i32 addr:$src))]>;
+ "", [(set GR64:$dst, (zextloadi64i32 addr:$src))]>;
// Any instruction that defines a 32-bit result leaves the high half of the
// register. Truncate can be lowered to EXTRACT_SUBREG. CopyFromReg may
"add{q}\t{$src2, $dst|$dst, $src2}",
[(set GR64:$dst, (add GR64:$src1, (load addr:$src2))),
(implicit EFLAGS)]>;
+
+// Register-Register Addition - Equivalent to the normal rr form (ADD64rr), but
+// differently encoded.
+def ADD64mrmrr : RI<0x03, MRMSrcReg, (outs GR64:$dst), (ins GR64:$src1, GR64:$src2),
+ "add{l}\t{$src2, $dst|$dst, $src2}", []>;
+
} // isTwoAddress
// Memory-Register Addition
def SHL64ri : RIi8<0xC1, MRM4r, (outs GR64:$dst), (ins GR64:$src1, i8imm:$src2),
"shl{q}\t{$src2, $dst|$dst, $src2}",
[(set GR64:$dst, (shl GR64:$src1, (i8 imm:$src2)))]>;
-// NOTE: We don't use shifts of a register by one, because 'add reg,reg' is
-// cheaper.
+// NOTE: We don't include patterns for shifts of a register by one, because
+// 'add reg,reg' is cheaper.
+def SHL64r1 : RI<0xD1, MRM4r, (outs GR64:$dst), (ins GR64:$src1),
+ "shr{q}\t$dst", []>;
} // isTwoAddress
let Uses = [CL] in
[(store (sra (loadi64 addr:$dst), (i8 1)), addr:$dst)]>;
// Rotate instructions
+
+let isTwoAddress = 1 in {
+def RCL64r1 : RI<0xD1, MRM2r, (outs GR64:$dst), (ins GR64:$src),
+ "rcl{q}\t{1, $dst|$dst, 1}", []>;
+def RCL64m1 : RI<0xD1, MRM2m, (outs i64mem:$dst), (ins i64mem:$src),
+ "rcl{q}\t{1, $dst|$dst, 1}", []>;
+let Uses = [CL] in {
+def RCL64rCL : RI<0xD3, MRM2r, (outs GR64:$dst), (ins GR64:$src),
+ "rcl{q}\t{%cl, $dst|$dst, CL}", []>;
+def RCL64mCL : RI<0xD3, MRM2m, (outs i64mem:$dst), (ins i64mem:$src),
+ "rcl{q}\t{%cl, $dst|$dst, CL}", []>;
+}
+def RCL64ri : RIi8<0xC1, MRM2r, (outs GR64:$dst), (ins GR64:$src, i8imm:$cnt),
+ "rcl{q}\t{$cnt, $dst|$dst, $cnt}", []>;
+def RCL64mi : RIi8<0xC1, MRM2m, (outs i64mem:$dst), (ins i64mem:$src, i8imm:$cnt),
+ "rcl{q}\t{$cnt, $dst|$dst, $cnt}", []>;
+
+def RCR64r1 : RI<0xD1, MRM3r, (outs GR64:$dst), (ins GR64:$src),
+ "rcr{q}\t{1, $dst|$dst, 1}", []>;
+def RCR64m1 : RI<0xD1, MRM3m, (outs i64mem:$dst), (ins i64mem:$src),
+ "rcr{q}\t{1, $dst|$dst, 1}", []>;
+let Uses = [CL] in {
+def RCR64rCL : RI<0xD3, MRM3r, (outs GR64:$dst), (ins GR64:$src),
+ "rcr{q}\t{%cl, $dst|$dst, CL}", []>;
+def RCR64mCL : RI<0xD3, MRM3m, (outs i64mem:$dst), (ins i64mem:$src),
+ "rcr{q}\t{%cl, $dst|$dst, CL}", []>;
+}
+def RCR64ri : RIi8<0xC1, MRM3r, (outs GR64:$dst), (ins GR64:$src, i8imm:$cnt),
+ "rcr{q}\t{$cnt, $dst|$dst, $cnt}", []>;
+def RCR64mi : RIi8<0xC1, MRM3m, (outs i64mem:$dst), (ins i64mem:$src, i8imm:$cnt),
+ "rcr{q}\t{$cnt, $dst|$dst, $cnt}", []>;
+}
+
let isTwoAddress = 1 in {
let Uses = [CL] in
def ROL64rCL : RI<0xD3, MRM0r, (outs GR64:$dst), (ins GR64:$src),
"cmp{q}\t{$src2, $src1|$src1, $src2}",
[(X86cmp GR64:$src1, GR64:$src2),
(implicit EFLAGS)]>;
+def CMP64mrmrr : RI<0x3B, MRMSrcReg, (outs), (ins GR64:$src1, GR64:$src2),
+ "cmp{q}\t{$src2, $src1|$src1, $src2}", []>;
def CMP64mr : RI<0x39, MRMDestMem, (outs), (ins i64mem:$src1, GR64:$src2),
"cmp{q}\t{$src2, $src1|$src1, $src2}",
[(X86cmp (loadi64 addr:$src1), GR64:$src2),
// Materialize i64 constant where top 32-bits are zero.
let AddedComplexity = 1, isReMaterializable = 1, isAsCheapAsAMove = 1 in
def MOV64ri64i32 : Ii32<0xB8, AddRegFrm, (outs GR64:$dst), (ins i64i32imm:$src),
- "mov{l}\t{$src, ${dst:subreg32}|${dst:subreg32}, $src}",
- [(set GR64:$dst, i64immZExt32:$src)]>;
+ "", [(set GR64:$dst, i64immZExt32:$src)]>;
//===----------------------------------------------------------------------===//
// Thread Local Storage Instructions
}
// Optimized codegen when the non-memory output is not used.
+let Defs = [EFLAGS] in {
// FIXME: Use normal add / sub instructions and add lock prefix dynamically.
def LOCK_ADD64mr : RI<0x03, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src2),
"lock\n\t"
def LOCK_DEC64m : RI<0xFF, MRM1m, (outs), (ins i64mem:$dst),
"lock\n\t"
"dec{q}\t$dst", []>, LOCK;
-
+}
// Atomic exchange, and, or, xor
let Constraints = "$val = $dst", Defs = [EFLAGS],
- usesCustomDAGSchedInserter = 1 in {
+ usesCustomInserter = 1 in {
def ATOMAND64 : I<0, Pseudo, (outs GR64:$dst),(ins i64mem:$ptr, GR64:$val),
"#ATOMAND64 PSEUDO!",
[(set GR64:$dst, (atomic_load_and_64 addr:$ptr, GR64:$val))]>;
[(set GR64:$dst, (atomic_load_umax_64 addr:$ptr, GR64:$val))]>;
}
+// Segmentation support instructions
+
+// i16mem operand in LAR64rm and GR32 operand in LAR32rr is not a typo.
+def LAR64rm : RI<0x02, MRMSrcMem, (outs GR64:$dst), (ins i16mem:$src),
+ "lar{q}\t{$src, $dst|$dst, $src}", []>, TB;
+def LAR64rr : RI<0x02, MRMSrcReg, (outs GR64:$dst), (ins GR32:$src),
+ "lar{q}\t{$src, $dst|$dst, $src}", []>, TB;
+
+// String manipulation instructions
+
+def LODSQ : RI<0xAD, RawFrm, (outs), (ins), "lodsq", []>;
+
//===----------------------------------------------------------------------===//
// Non-Instruction Patterns
//===----------------------------------------------------------------------===//
(MOV64ri tglobaladdr :$dst)>, Requires<[FarData]>;
def : Pat<(i64 (X86Wrapper texternalsym:$dst)),
(MOV64ri texternalsym:$dst)>, Requires<[FarData]>;
+def : Pat<(i64 (X86Wrapper tblockaddress:$dst)),
+ (MOV64ri tblockaddress:$dst)>, Requires<[FarData]>;
// In static codegen with small code model, we can get the address of a label
// into a register with 'movl'. FIXME: This is a hack, the 'imm' predicate of
(MOV64ri64i32 tglobaladdr :$dst)>, Requires<[SmallCode]>;
def : Pat<(i64 (X86Wrapper texternalsym:$dst)),
(MOV64ri64i32 texternalsym:$dst)>, Requires<[SmallCode]>;
+def : Pat<(i64 (X86Wrapper tblockaddress:$dst)),
+ (MOV64ri64i32 tblockaddress:$dst)>, Requires<[SmallCode]>;
// In kernel code model, we can get the address of a label
// into a register with 'movq'. FIXME: This is a hack, the 'imm' predicate of
(MOV64ri32 tglobaladdr :$dst)>, Requires<[KernelCode]>;
def : Pat<(i64 (X86Wrapper texternalsym:$dst)),
(MOV64ri32 texternalsym:$dst)>, Requires<[KernelCode]>;
+def : Pat<(i64 (X86Wrapper tblockaddress:$dst)),
+ (MOV64ri32 tblockaddress:$dst)>, Requires<[KernelCode]>;
// If we have small model and -static mode, it is safe to store global addresses
// directly as immediates. FIXME: This is really a hack, the 'imm' predicate
def : Pat<(store (i64 (X86Wrapper texternalsym:$src)), addr:$dst),
(MOV64mi32 addr:$dst, texternalsym:$src)>,
Requires<[NearData, IsStatic]>;
+def : Pat<(store (i64 (X86Wrapper tblockaddress:$src)), addr:$dst),
+ (MOV64mi32 addr:$dst, tblockaddress:$src)>,
+ Requires<[NearData, IsStatic]>;
// Calls
// Direct PC relative function call for small code model. 32-bit displacement
(SUBREG_TO_REG
(i64 0),
(MOVZX32_NOREXrr8
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR64:$src, GR64_ABCD),
+ (EXTRACT_SUBREG (i64 (COPY_TO_REGCLASS GR64:$src, GR64_ABCD)),
x86_subreg_8bit_hi)),
x86_subreg_32bit)>;
def : Pat<(and (srl_su GR32:$src, (i8 8)), (i32 255)),
(MOVZX32_NOREXrr8
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR32:$src, GR32_ABCD),
+ (EXTRACT_SUBREG (i32 (COPY_TO_REGCLASS GR32:$src, GR32_ABCD)),
x86_subreg_8bit_hi))>,
Requires<[In64BitMode]>;
def : Pat<(srl_su GR16:$src, (i8 8)),
(EXTRACT_SUBREG
(MOVZX32_NOREXrr8
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD),
+ (EXTRACT_SUBREG (i16 (COPY_TO_REGCLASS GR16:$src, GR16_ABCD)),
x86_subreg_8bit_hi)),
x86_subreg_16bit)>,
Requires<[In64BitMode]>;
def : Pat<(i32 (zext (srl_su GR16:$src, (i8 8)))),
(MOVZX32_NOREXrr8
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD),
+ (EXTRACT_SUBREG (i16 (COPY_TO_REGCLASS GR16:$src, GR16_ABCD)),
x86_subreg_8bit_hi))>,
Requires<[In64BitMode]>;
def : Pat<(i32 (anyext (srl_su GR16:$src, (i8 8)))),
(MOVZX32_NOREXrr8
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD),
+ (EXTRACT_SUBREG (i16 (COPY_TO_REGCLASS GR16:$src, GR16_ABCD)),
x86_subreg_8bit_hi))>,
Requires<[In64BitMode]>;
def : Pat<(i64 (zext (srl_su GR16:$src, (i8 8)))),
(SUBREG_TO_REG
(i64 0),
(MOVZX32_NOREXrr8
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD),
+ (EXTRACT_SUBREG (i16 (COPY_TO_REGCLASS GR16:$src, GR16_ABCD)),
x86_subreg_8bit_hi)),
x86_subreg_32bit)>;
def : Pat<(i64 (anyext (srl_su GR16:$src, (i8 8)))),
(SUBREG_TO_REG
(i64 0),
(MOVZX32_NOREXrr8
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD),
+ (EXTRACT_SUBREG (i16 (COPY_TO_REGCLASS GR16:$src, GR16_ABCD)),
x86_subreg_8bit_hi)),
x86_subreg_32bit)>;
def : Pat<(store (i8 (trunc_su (srl_su GR64:$src, (i8 8)))), addr:$dst),
(MOV8mr_NOREX
addr:$dst,
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR64:$src, GR64_ABCD),
+ (EXTRACT_SUBREG (i64 (COPY_TO_REGCLASS GR64:$src, GR64_ABCD)),
x86_subreg_8bit_hi))>;
def : Pat<(store (i8 (trunc_su (srl_su GR32:$src, (i8 8)))), addr:$dst),
(MOV8mr_NOREX
addr:$dst,
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR32:$src, GR32_ABCD),
+ (EXTRACT_SUBREG (i32 (COPY_TO_REGCLASS GR32:$src, GR32_ABCD)),
x86_subreg_8bit_hi))>,
Requires<[In64BitMode]>;
def : Pat<(store (i8 (trunc_su (srl_su GR16:$src, (i8 8)))), addr:$dst),
(MOV8mr_NOREX
addr:$dst,
- (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD),
+ (EXTRACT_SUBREG (i16 (COPY_TO_REGCLASS GR16:$src, GR16_ABCD)),
x86_subreg_8bit_hi))>,
Requires<[In64BitMode]>;
(implicit EFLAGS)),
(DEC64m addr:$dst)>;
+// Register-Register Logical Or with EFLAGS result
+def : Pat<(parallel (X86or_flag GR64:$src1, GR64:$src2),
+ (implicit EFLAGS)),
+ (OR64rr GR64:$src1, GR64:$src2)>;
+
+// Register-Integer Logical Or with EFLAGS result
+def : Pat<(parallel (X86or_flag GR64:$src1, i64immSExt8:$src2),
+ (implicit EFLAGS)),
+ (OR64ri8 GR64:$src1, i64immSExt8:$src2)>;
+def : Pat<(parallel (X86or_flag GR64:$src1, i64immSExt32:$src2),
+ (implicit EFLAGS)),
+ (OR64ri32 GR64:$src1, i64immSExt32:$src2)>;
+
+// Register-Memory Logical Or with EFLAGS result
+def : Pat<(parallel (X86or_flag GR64:$src1, (loadi64 addr:$src2)),
+ (implicit EFLAGS)),
+ (OR64rm GR64:$src1, addr:$src2)>;
+
+// Memory-Register Logical Or with EFLAGS result
+def : Pat<(parallel (store (X86or_flag (loadi64 addr:$dst), GR64:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (OR64mr addr:$dst, GR64:$src2)>;
+def : Pat<(parallel (store (X86or_flag (loadi64 addr:$dst), i64immSExt8:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (OR64mi8 addr:$dst, i64immSExt8:$src2)>;
+def : Pat<(parallel (store (X86or_flag (loadi64 addr:$dst), i64immSExt32:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (OR64mi32 addr:$dst, i64immSExt32:$src2)>;
+
+// Register-Register Logical XOr with EFLAGS result
+def : Pat<(parallel (X86xor_flag GR64:$src1, GR64:$src2),
+ (implicit EFLAGS)),
+ (XOR64rr GR64:$src1, GR64:$src2)>;
+
+// Register-Integer Logical XOr with EFLAGS result
+def : Pat<(parallel (X86xor_flag GR64:$src1, i64immSExt8:$src2),
+ (implicit EFLAGS)),
+ (XOR64ri8 GR64:$src1, i64immSExt8:$src2)>;
+def : Pat<(parallel (X86xor_flag GR64:$src1, i64immSExt32:$src2),
+ (implicit EFLAGS)),
+ (XOR64ri32 GR64:$src1, i64immSExt32:$src2)>;
+
+// Register-Memory Logical XOr with EFLAGS result
+def : Pat<(parallel (X86xor_flag GR64:$src1, (loadi64 addr:$src2)),
+ (implicit EFLAGS)),
+ (XOR64rm GR64:$src1, addr:$src2)>;
+
+// Memory-Register Logical XOr with EFLAGS result
+def : Pat<(parallel (store (X86xor_flag (loadi64 addr:$dst), GR64:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (XOR64mr addr:$dst, GR64:$src2)>;
+def : Pat<(parallel (store (X86xor_flag (loadi64 addr:$dst), i64immSExt8:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (XOR64mi8 addr:$dst, i64immSExt8:$src2)>;
+def : Pat<(parallel (store (X86xor_flag (loadi64 addr:$dst), i64immSExt32:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (XOR64mi32 addr:$dst, i64immSExt32:$src2)>;
+
+// Register-Register Logical And with EFLAGS result
+def : Pat<(parallel (X86and_flag GR64:$src1, GR64:$src2),
+ (implicit EFLAGS)),
+ (AND64rr GR64:$src1, GR64:$src2)>;
+
+// Register-Integer Logical And with EFLAGS result
+def : Pat<(parallel (X86and_flag GR64:$src1, i64immSExt8:$src2),
+ (implicit EFLAGS)),
+ (AND64ri8 GR64:$src1, i64immSExt8:$src2)>;
+def : Pat<(parallel (X86and_flag GR64:$src1, i64immSExt32:$src2),
+ (implicit EFLAGS)),
+ (AND64ri32 GR64:$src1, i64immSExt32:$src2)>;
+
+// Register-Memory Logical And with EFLAGS result
+def : Pat<(parallel (X86and_flag GR64:$src1, (loadi64 addr:$src2)),
+ (implicit EFLAGS)),
+ (AND64rm GR64:$src1, addr:$src2)>;
+
+// Memory-Register Logical And with EFLAGS result
+def : Pat<(parallel (store (X86and_flag (loadi64 addr:$dst), GR64:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (AND64mr addr:$dst, GR64:$src2)>;
+def : Pat<(parallel (store (X86and_flag (loadi64 addr:$dst), i64immSExt8:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (AND64mi8 addr:$dst, i64immSExt8:$src2)>;
+def : Pat<(parallel (store (X86and_flag (loadi64 addr:$dst), i64immSExt32:$src2),
+ addr:$dst),
+ (implicit EFLAGS)),
+ (AND64mi32 addr:$dst, i64immSExt32:$src2)>;
+
//===----------------------------------------------------------------------===//
// X86-64 SSE Instructions
//===----------------------------------------------------------------------===//