diff options
Diffstat (limited to 'contrib/llvm/lib/Target/X86/X86InstrCompiler.td')
| -rw-r--r-- | contrib/llvm/lib/Target/X86/X86InstrCompiler.td | 102 |
1 files changed, 82 insertions, 20 deletions
diff --git a/contrib/llvm/lib/Target/X86/X86InstrCompiler.td b/contrib/llvm/lib/Target/X86/X86InstrCompiler.td index adcc747..da28690 100644 --- a/contrib/llvm/lib/Target/X86/X86InstrCompiler.td +++ b/contrib/llvm/lib/Target/X86/X86InstrCompiler.td @@ -106,6 +106,26 @@ let Defs = [EAX, ESP, EFLAGS], Uses = [ESP] in def WIN_ALLOCA : I<0, Pseudo, (outs), (ins), "# dynamic stack allocation", [(X86WinAlloca)]>; + +// When using segmented stacks these are lowered into instructions which first +// check if the current stacklet has enough free memory. If it does, memory is +// allocated by bumping the stack pointer. Otherwise memory is allocated from +// the heap. + +let Defs = [EAX, ESP, EFLAGS], Uses = [ESP, EAX] in +def SEG_ALLOCA_32 : I<0, Pseudo, (outs GR32:$dst), (ins GR32:$size), + "# variable sized alloca for segmented stacks", + [(set GR32:$dst, + (X86SegAlloca GR32:$size))]>, + Requires<[In32BitMode]>; + +let Defs = [RAX, RSP, EFLAGS], Uses = [RSP, RAX] in +def SEG_ALLOCA_64 : I<0, Pseudo, (outs GR64:$dst), (ins GR64:$size), + "# variable sized alloca for segmented stacks", + [(set GR64:$dst, + (X86SegAlloca GR64:$size))]>, + Requires<[In64BitMode]>; + } @@ -329,18 +349,11 @@ def TLSCall_64 : I<0, Pseudo, (outs), (ins i64mem:$sym), //===----------------------------------------------------------------------===// // Conditional Move Pseudo Instructions -let Constraints = "$src1 = $dst" in { - -// Conditional moves -let Uses = [EFLAGS] in { - // X86 doesn't have 8-bit conditional moves. Use a customInserter to // emit control flow. An alternative to this is to mark i8 SELECT as Promote, // however that requires promoting the operands, and can induce additional -// i8 register pressure. Note that CMOV_GR8 is conservatively considered to -// clobber EFLAGS, because if one of the operands is zero, the expansion -// could involve an xor. -let usesCustomInserter = 1, Constraints = "", Defs = [EFLAGS] in { +// i8 register pressure. +let usesCustomInserter = 1, Uses = [EFLAGS] in { def CMOV_GR8 : I<0, Pseudo, (outs GR8:$dst), (ins GR8:$src1, GR8:$src2, i8imm:$cond), "#CMOV_GR8 PSEUDO!", @@ -380,10 +393,7 @@ def CMOV_RFP80 : I<0, Pseudo, (X86cmov RFP80:$src1, RFP80:$src2, imm:$cond, EFLAGS))]>; } // Predicates = [NoCMov] -} // UsesCustomInserter = 1, Constraints = "", Defs = [EFLAGS] -} // Uses = [EFLAGS] - -} // Constraints = "$src1 = $dst" in +} // UsesCustomInserter = 1, Uses = [EFLAGS] //===----------------------------------------------------------------------===// @@ -532,7 +542,7 @@ def OR32mrLocked : I<0x09, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$zero), let hasSideEffects = 1 in def Int_MemBarrier : I<0, Pseudo, (outs), (ins), "#MEMBARRIER", - [(X86MemBarrier)]>, Requires<[HasSSE2]>; + [(X86MemBarrier)]>; // TODO: Get this to fold the constant into the instruction. let hasSideEffects = 1, Defs = [ESP], isCodeGenOnly = 1 in @@ -630,8 +640,8 @@ def #NAME#64mi8 : RIi8<{ImmOpc8{7}, ImmOpc8{6}, ImmOpc8{5}, ImmOpc8{4}, defm LOCK_ADD : LOCK_ArithBinOp<0x00, 0x80, 0x83, MRM0m, "add">; defm LOCK_SUB : LOCK_ArithBinOp<0x28, 0x80, 0x83, MRM5m, "sub">; defm LOCK_OR : LOCK_ArithBinOp<0x08, 0x80, 0x83, MRM1m, "or">; -defm LOCK_AND : LOCK_ArithBinOp<0x08, 0x80, 0x83, MRM4m, "and">; -defm LOCK_XOR : LOCK_ArithBinOp<0x08, 0x80, 0x83, MRM6m, "xor">; +defm LOCK_AND : LOCK_ArithBinOp<0x20, 0x80, 0x83, MRM4m, "and">; +defm LOCK_XOR : LOCK_ArithBinOp<0x30, 0x80, 0x83, MRM6m, "xor">; // Optimized codegen when the non-memory output is not used. let Defs = [EFLAGS], mayLoad = 1, mayStore = 1, isCodeGenOnly = 1 in { @@ -665,12 +675,20 @@ def LOCK_DEC64m : RI<0xFF, MRM1m, (outs), (ins i64mem:$dst), // Atomic compare and swap. let Defs = [EAX, EDX, EFLAGS], Uses = [EAX, EBX, ECX, EDX], - isCodeGenOnly = 1 in { + isCodeGenOnly = 1 in def LCMPXCHG8B : I<0xC7, MRM1m, (outs), (ins i64mem:$ptr), "lock\n\t" "cmpxchg8b\t$ptr", [(X86cas8 addr:$ptr)]>, TB, LOCK; -} + +let Defs = [RAX, RDX, EFLAGS], Uses = [RAX, RBX, RCX, RDX], + isCodeGenOnly = 1 in +def LCMPXCHG16B : RI<0xC7, MRM1m, (outs), (ins i128mem:$ptr), + "lock\n\t" + "cmpxchg16b\t$ptr", + [(X86cas16 addr:$ptr)]>, TB, LOCK, + Requires<[HasCmpxchg16b]>; + let Defs = [AL, EFLAGS], Uses = [AL], isCodeGenOnly = 1 in { def LCMPXCHG8 : I<0xB0, MRMDestMem, (outs), (ins i8mem:$ptr, GR8:$swap), "lock\n\t" @@ -695,7 +713,7 @@ def LCMPXCHG32 : I<0xB1, MRMDestMem, (outs), (ins i32mem:$ptr, GR32:$swap), let Defs = [RAX, EFLAGS], Uses = [RAX], isCodeGenOnly = 1 in { def LCMPXCHG64 : RI<0xB1, MRMDestMem, (outs), (ins i64mem:$ptr, GR64:$swap), "lock\n\t" - "cmpxchgq\t$swap,$ptr", + "cmpxchg{q}\t{$swap, $ptr|$ptr, $swap}", [(X86cas addr:$ptr, GR64:$swap, 8)]>, TB, LOCK; } @@ -718,11 +736,37 @@ def LXADD32 : I<0xC1, MRMSrcMem, (outs GR32:$dst), (ins GR32:$val, i32mem:$ptr), TB, LOCK; def LXADD64 : RI<0xC1, MRMSrcMem, (outs GR64:$dst), (ins GR64:$val,i64mem:$ptr), "lock\n\t" - "xadd\t$val, $ptr", + "xadd{q}\t{$val, $ptr|$ptr, $val}", [(set GR64:$dst, (atomic_load_add_64 addr:$ptr, GR64:$val))]>, TB, LOCK; } +def ACQUIRE_MOV8rm : I<0, Pseudo, (outs GR8 :$dst), (ins i8mem :$src), + "#ACQUIRE_MOV PSEUDO!", + [(set GR8:$dst, (atomic_load_8 addr:$src))]>; +def ACQUIRE_MOV16rm : I<0, Pseudo, (outs GR16:$dst), (ins i16mem:$src), + "#ACQUIRE_MOV PSEUDO!", + [(set GR16:$dst, (atomic_load_16 addr:$src))]>; +def ACQUIRE_MOV32rm : I<0, Pseudo, (outs GR32:$dst), (ins i32mem:$src), + "#ACQUIRE_MOV PSEUDO!", + [(set GR32:$dst, (atomic_load_32 addr:$src))]>; +def ACQUIRE_MOV64rm : I<0, Pseudo, (outs GR64:$dst), (ins i64mem:$src), + "#ACQUIRE_MOV PSEUDO!", + [(set GR64:$dst, (atomic_load_64 addr:$src))]>; + +def RELEASE_MOV8mr : I<0, Pseudo, (outs), (ins i8mem :$dst, GR8 :$src), + "#RELEASE_MOV PSEUDO!", + [(atomic_store_8 addr:$dst, GR8 :$src)]>; +def RELEASE_MOV16mr : I<0, Pseudo, (outs), (ins i16mem:$dst, GR16:$src), + "#RELEASE_MOV PSEUDO!", + [(atomic_store_16 addr:$dst, GR16:$src)]>; +def RELEASE_MOV32mr : I<0, Pseudo, (outs), (ins i32mem:$dst, GR32:$src), + "#RELEASE_MOV PSEUDO!", + [(atomic_store_32 addr:$dst, GR32:$src)]>; +def RELEASE_MOV64mr : I<0, Pseudo, (outs), (ins i64mem:$dst, GR64:$src), + "#RELEASE_MOV PSEUDO!", + [(atomic_store_64 addr:$dst, GR64:$src)]>; + //===----------------------------------------------------------------------===// // Conditional Move Pseudo Instructions. //===----------------------------------------------------------------------===// @@ -759,6 +803,24 @@ let Uses = [EFLAGS], usesCustomInserter = 1 in { [(set VR128:$dst, (v2i64 (X86cmov VR128:$t, VR128:$f, imm:$cond, EFLAGS)))]>; + def CMOV_V8F32 : I<0, Pseudo, + (outs VR256:$dst), (ins VR256:$t, VR256:$f, i8imm:$cond), + "#CMOV_V8F32 PSEUDO!", + [(set VR256:$dst, + (v8f32 (X86cmov VR256:$t, VR256:$f, imm:$cond, + EFLAGS)))]>; + def CMOV_V4F64 : I<0, Pseudo, + (outs VR256:$dst), (ins VR256:$t, VR256:$f, i8imm:$cond), + "#CMOV_V4F64 PSEUDO!", + [(set VR256:$dst, + (v4f64 (X86cmov VR256:$t, VR256:$f, imm:$cond, + EFLAGS)))]>; + def CMOV_V4I64 : I<0, Pseudo, + (outs VR256:$dst), (ins VR256:$t, VR256:$f, i8imm:$cond), + "#CMOV_V4I64 PSEUDO!", + [(set VR256:$dst, + (v4i64 (X86cmov VR256:$t, VR256:$f, imm:$cond, + EFLAGS)))]>; } |
