[LLVMdev] [RFC] Add second "failure" AtomicOrdering to cmpxchg instruction
Tim Northover
t.p.northover at gmail.com
Fri Mar 7 05:47:12 PST 2014
Hi all,
The C++11 (& C11) compare_exchange functions with explicit memory
order allow you to specify two sets of semantics, one for when the
exchange actually happens and one for when it fails. Unfortunately, at
the moment the LLVM IR "cmpxchg" instruction only has one ordering,
which means we get sub-optimal codegen.
This probably affects all architectures which use
load-linked/store-conditional instructions for atomic operations and
don't have versions with built-in acquire/release semantics (and so
need barriers). For example on ARMv7, an "acquire, relaxed" cmpxchg
currently generates:
mov r9, #52
mov r3, #42
LBB0_1: @ =>This Inner Loop Header: Depth=1
ldrexb r1, [r0]
cmp r1, r3
bne LBB0_3
@ BB#2: @ in Loop: Header=BB0_1 Depth=1
strexb r2, r9, [r0]
cmp r2, #0
bne LBB0_1
LBB0_3:
dmb ish
[...]
That second barrier could be skipped by the first "bne" if we knew
that the user didn't need any ordering in case of failure.
The other atomic operations are not affected, so I'd like to add a
second ordering operand to "cmpxchg" to support the full range of
possible requests. The suggested syntax would be
cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new>
[singlethread] <success ordering> <failure ordering>
That is, syntactically I propose:
+ No comma between the two
+ Both operands are required, even if "failure" is the natural pair
of "success"
Semantically, I would like to enforce the constraints in the standards:
+ failure <= success
+ failure != Release
+ failure != AcquireRelease.
Before the DAG, I suggest completely removing getOrdering from
CmpXchgInst, in favour of getSuccessOrdering and getFailureOrdering to
avoid errors. In the DAG, getOrdering would still exist and return the
success ordering, which I believe would make existing targets
conservatively correct. AtomicSDNode would add getFailureOrdering,
used only by ATOMIC_CMP_SWAP nodes.
Existing bitcode files would obviously be supported, and retain
current semantics by BitcodeReader dropping the "release" part of any
single ordering, and using the result as the "failure ordering" (as
currently documented in the LanguageRef).
I have attached initial patches for reference, but I'm not asking for
thorough review at this stage.
What do people think?
Cheers.
Tim.
-------------- next part --------------
diff --git a/lib/CodeGen/CGAtomic.cpp b/lib/CodeGen/CGAtomic.cpp
index cb92f9a..96595df 100644
--- a/lib/CodeGen/CGAtomic.cpp
+++ b/lib/CodeGen/CGAtomic.cpp
@@ -205,8 +205,9 @@ EmitAtomicOp(CodeGenFunction &CGF, AtomicExpr *E, llvm::Value *Dest,
LoadVal1->setAlignment(Align);
llvm::LoadInst *LoadVal2 = CGF.Builder.CreateLoad(Val2);
LoadVal2->setAlignment(Align);
+ // FIXME: The failure order needs demoting if it's AcquireRelease or Release.
llvm::AtomicCmpXchgInst *CXI =
- CGF.Builder.CreateAtomicCmpXchg(Ptr, LoadVal1, LoadVal2, Order);
+ CGF.Builder.CreateAtomicCmpXchg(Ptr, LoadVal1, LoadVal2, Order, Order);
CXI->setVolatile(E->isVolatile());
llvm::StoreInst *StoreVal1 = CGF.Builder.CreateStore(CXI, Val1);
StoreVal1->setAlignment(Align);
diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index 3966419..7649325 100644
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -964,6 +964,7 @@ RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
Args[2] = EmitToInt(*this, EmitScalarExpr(E->getArg(2)), T, IntType);
Value *Result = Builder.CreateAtomicCmpXchg(Args[0], Args[1], Args[2],
+ llvm::SequentiallyConsistent,
llvm::SequentiallyConsistent);
Result = EmitFromInt(*this, Result, T, ValueType);
return RValue::get(Result);
@@ -990,6 +991,7 @@ RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
Value *OldVal = Args[1];
Value *PrevVal = Builder.CreateAtomicCmpXchg(Args[0], Args[1], Args[2],
+ llvm::SequentiallyConsistent,
llvm::SequentiallyConsistent);
Value *Result = Builder.CreateICmpEQ(PrevVal, OldVal);
// zext bool to int.
@@ -1504,6 +1506,7 @@ RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
EmitScalarExpr(E->getArg(0)),
EmitScalarExpr(E->getArg(2)),
EmitScalarExpr(E->getArg(1)),
+ SequentiallyConsistent,
SequentiallyConsistent);
CXI->setVolatile(true);
return RValue::get(CXI);
diff --git a/lib/CodeGen/CGExprScalar.cpp b/lib/CodeGen/CGExprScalar.cpp
index 580cb15..5f932b0 100644
--- a/lib/CodeGen/CGExprScalar.cpp
+++ b/lib/CodeGen/CGExprScalar.cpp
@@ -1735,8 +1735,9 @@ ScalarExprEmitter::EmitScalarPrePostIncDec(const UnaryOperator *E, LValue LV,
if (atomicPHI) {
llvm::BasicBlock *opBB = Builder.GetInsertBlock();
llvm::BasicBlock *contBB = CGF.createBasicBlock("atomic_cont", CGF.CurFn);
- llvm::Value *old = Builder.CreateAtomicCmpXchg(LV.getAddress(), atomicPHI,
- CGF.EmitToMemory(value, type), llvm::SequentiallyConsistent);
+ llvm::Value *old = Builder.CreateAtomicCmpXchg(
+ LV.getAddress(), atomicPHI, CGF.EmitToMemory(value, type),
+ llvm::SequentiallyConsistent, llvm::SequentiallyConsistent);
atomicPHI->addIncoming(old, opBB);
llvm::Value *success = Builder.CreateICmpEQ(old, atomicPHI);
Builder.CreateCondBr(success, contBB, opBB);
@@ -2077,8 +2078,9 @@ LValue ScalarExprEmitter::EmitCompoundAssignLValue(
if (atomicPHI) {
llvm::BasicBlock *opBB = Builder.GetInsertBlock();
llvm::BasicBlock *contBB = CGF.createBasicBlock("atomic_cont", CGF.CurFn);
- llvm::Value *old = Builder.CreateAtomicCmpXchg(LHSLV.getAddress(), atomicPHI,
- CGF.EmitToMemory(Result, LHSTy), llvm::SequentiallyConsistent);
+ llvm::Value *old = Builder.CreateAtomicCmpXchg(
+ LHSLV.getAddress(), atomicPHI, CGF.EmitToMemory(Result, LHSTy),
+ llvm::SequentiallyConsistent, llvm::SequentiallyConsistent);
atomicPHI->addIncoming(old, opBB);
llvm::Value *success = Builder.CreateICmpEQ(old, atomicPHI);
Builder.CreateCondBr(success, contBB, opBB);
-------------- next part --------------
diff --git a/docs/LangRef.rst b/docs/LangRef.rst
index e38184c..942caf0 100644
--- a/docs/LangRef.rst
+++ b/docs/LangRef.rst
@@ -4990,7 +4990,7 @@ Syntax:
::
- cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <ordering> ; yields {ty}
+ cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <success ordering> <failure ordering> ; yields {ty}
Overview:
"""""""""
@@ -5013,8 +5013,11 @@ type, and the type of '<pointer>' must be a pointer to that type. If the
to modify the number or order of execution of this ``cmpxchg`` with
other :ref:`volatile operations <volatile>`.
-The :ref:`ordering <ordering>` argument specifies how this ``cmpxchg``
-synchronizes with other atomic operations.
+The success and failure :ref:`ordering <ordering>` arguments specify how this
+``cmpxchg`` synchronizes with other atomic operations. The both ordering
+parameters must be at least ``monotonic``, the ordering constraint on failure
+must be no stronger than that on success, and the failure ordering cannot be
+either ``release`` or ``acq_rel``.
The optional "``singlethread``" argument declares that the ``cmpxchg``
is only atomic with respect to code (usually signal handlers) running in
@@ -5032,10 +5035,9 @@ operand is read and compared to '``<cmp>``'; if the read value is the
equal, '``<new>``' is written. The original value at the location is
returned.
-A successful ``cmpxchg`` is a read-modify-write instruction for the purpose
-of identifying release sequences. A failed ``cmpxchg`` is equivalent to an
-atomic load with an ordering parameter determined by dropping any
-``release`` part of the ``cmpxchg``'s ordering.
+A successful ``cmpxchg`` is a read-modify-write instruction for the purpose of
+identifying release sequences. A failed ``cmpxchg`` is equivalent to an atomic
+load with an ordering parameter determined the second ordering parameter.
Example:
""""""""
diff --git a/include/llvm/CodeGen/SelectionDAG.h b/include/llvm/CodeGen/SelectionDAG.h
index adba296..a30656a 100644
--- a/include/llvm/CodeGen/SelectionDAG.h
+++ b/include/llvm/CodeGen/SelectionDAG.h
@@ -692,12 +692,14 @@ public:
SDValue getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT, SDValue Chain,
SDValue Ptr, SDValue Cmp, SDValue Swp,
MachinePointerInfo PtrInfo, unsigned Alignment,
- AtomicOrdering Ordering,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
SynchronizationScope SynchScope);
SDValue getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT, SDValue Chain,
SDValue Ptr, SDValue Cmp, SDValue Swp,
MachineMemOperand *MMO,
- AtomicOrdering Ordering,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
SynchronizationScope SynchScope);
/// getAtomic - Gets a node for an atomic op, produces result (if relevant)
@@ -726,9 +728,13 @@ public:
/// getAtomic - Gets a node for an atomic op, produces result and chain and
/// takes N operands.
SDValue getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT, SDVTList VTList,
- SDValue* Ops, unsigned NumOps, MachineMemOperand *MMO,
- AtomicOrdering Ordering,
+ SDValue *Ops, unsigned NumOps, MachineMemOperand *MMO,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
SynchronizationScope SynchScope);
+ SDValue getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT, SDVTList VTList,
+ SDValue *Ops, unsigned NumOps, MachineMemOperand *MMO,
+ AtomicOrdering Ordering, SynchronizationScope SynchScope);
/// getMemIntrinsicNode - Creates a MemIntrinsicNode that may produce a
/// result and takes a list of operands. Opcode may be INTRINSIC_VOID,
diff --git a/include/llvm/CodeGen/SelectionDAGNodes.h b/include/llvm/CodeGen/SelectionDAGNodes.h
index 072b6af..eac3eec 100644
--- a/include/llvm/CodeGen/SelectionDAGNodes.h
+++ b/include/llvm/CodeGen/SelectionDAGNodes.h
@@ -974,6 +974,7 @@ public:
/// Abstact virtual class for operations for memory operations
class MemSDNode : public SDNode {
private:
+
// MemoryVT - VT of in-memory value.
EVT MemoryVT;
@@ -1013,9 +1014,11 @@ public:
bool isNonTemporal() const { return (SubclassData >> 6) & 1; }
bool isInvariant() const { return (SubclassData >> 7) & 1; }
+
AtomicOrdering getOrdering() const {
return AtomicOrdering((SubclassData >> 8) & 15);
}
+
SynchronizationScope getSynchScope() const {
return SynchronizationScope((SubclassData >> 12) & 1);
}
@@ -1089,15 +1092,28 @@ public:
class AtomicSDNode : public MemSDNode {
SDUse Ops[4];
- void InitAtomic(AtomicOrdering Ordering, SynchronizationScope SynchScope) {
+ /// For cmpxchg instructions, the ordering requirements when a store does not
+ /// occur.
+ AtomicOrdering FailureOrdering;
+
+
+ void InitAtomic(AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope) {
// This must match encodeMemSDNodeFlags() in SelectionDAG.cpp.
- assert((Ordering & 15) == Ordering &&
+ assert((SuccessOrdering & 15) == SuccessOrdering &&
+ "Ordering may not require more than 4 bits!");
+ assert((FailureOrdering & 15) == FailureOrdering &&
"Ordering may not require more than 4 bits!");
assert((SynchScope & 1) == SynchScope &&
"SynchScope may not require more than 1 bit!");
- SubclassData |= Ordering << 8;
+ SubclassData |= SuccessOrdering << 8;
SubclassData |= SynchScope << 12;
- assert(getOrdering() == Ordering && "Ordering encoding error!");
+ this->FailureOrdering = FailureOrdering;
+ assert(getSuccessOrdering() == SuccessOrdering &&
+ "Ordering encoding error!");
+ assert(getFailureOrdering() == FailureOrdering &&
+ "Ordering encoding error!");
assert(getSynchScope() == SynchScope && "Synch-scope encoding error!");
}
@@ -1111,12 +1127,11 @@ public:
// SrcVal: address to update as a Value (used for MemOperand)
// Align: alignment of memory
AtomicSDNode(unsigned Opc, unsigned Order, DebugLoc dl, SDVTList VTL,
- EVT MemVT,
- SDValue Chain, SDValue Ptr,
- SDValue Cmp, SDValue Swp, MachineMemOperand *MMO,
- AtomicOrdering Ordering, SynchronizationScope SynchScope)
- : MemSDNode(Opc, Order, dl, VTL, MemVT, MMO) {
- InitAtomic(Ordering, SynchScope);
+ EVT MemVT, SDValue Chain, SDValue Ptr, SDValue Cmp, SDValue Swp,
+ MachineMemOperand *MMO, AtomicOrdering Ordering,
+ SynchronizationScope SynchScope)
+ : MemSDNode(Opc, Order, dl, VTL, MemVT, MMO) {
+ InitAtomic(Ordering, Ordering, SynchScope);
InitOperands(Ops, Chain, Ptr, Cmp, Swp);
}
AtomicSDNode(unsigned Opc, unsigned Order, DebugLoc dl, SDVTList VTL,
@@ -1125,7 +1140,7 @@ public:
SDValue Val, MachineMemOperand *MMO,
AtomicOrdering Ordering, SynchronizationScope SynchScope)
: MemSDNode(Opc, Order, dl, VTL, MemVT, MMO) {
- InitAtomic(Ordering, SynchScope);
+ InitAtomic(Ordering, Ordering, SynchScope);
InitOperands(Ops, Chain, Ptr, Val);
}
AtomicSDNode(unsigned Opc, unsigned Order, DebugLoc dl, SDVTList VTL,
@@ -1134,15 +1149,16 @@ public:
MachineMemOperand *MMO,
AtomicOrdering Ordering, SynchronizationScope SynchScope)
: MemSDNode(Opc, Order, dl, VTL, MemVT, MMO) {
- InitAtomic(Ordering, SynchScope);
+ InitAtomic(Ordering, Ordering, SynchScope);
InitOperands(Ops, Chain, Ptr);
}
AtomicSDNode(unsigned Opc, unsigned Order, DebugLoc dl, SDVTList VTL, EVT MemVT,
SDValue* AllOps, SDUse *DynOps, unsigned NumOps,
MachineMemOperand *MMO,
- AtomicOrdering Ordering, SynchronizationScope SynchScope)
+ AtomicOrdering SuccessOrdering, AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope)
: MemSDNode(Opc, Order, dl, VTL, MemVT, MMO) {
- InitAtomic(Ordering, SynchScope);
+ InitAtomic(SuccessOrdering, FailureOrdering, SynchScope);
assert((DynOps || NumOps <= array_lengthof(Ops)) &&
"Too many ops for internal storage!");
InitOperands(DynOps ? DynOps : Ops, AllOps, NumOps);
@@ -1151,6 +1167,16 @@ public:
const SDValue &getBasePtr() const { return getOperand(1); }
const SDValue &getVal() const { return getOperand(2); }
+ AtomicOrdering getSuccessOrdering() const {
+ return getOrdering();
+ }
+
+ // Not quite enough room in SubclassData for everything, so failure gets its
+ // own field.
+ AtomicOrdering getFailureOrdering() const {
+ return FailureOrdering;
+ }
+
bool isCompareAndSwap() const {
unsigned Op = getOpcode();
return Op == ISD::ATOMIC_CMP_SWAP;
diff --git a/include/llvm/IR/IRBuilder.h b/include/llvm/IR/IRBuilder.h
index d28652f..79ee7b7 100644
--- a/include/llvm/IR/IRBuilder.h
+++ b/include/llvm/IR/IRBuilder.h
@@ -934,10 +934,13 @@ public:
const Twine &Name = "") {
return Insert(new FenceInst(Context, Ordering, SynchScope), Name);
}
- AtomicCmpXchgInst *CreateAtomicCmpXchg(Value *Ptr, Value *Cmp, Value *New,
- AtomicOrdering Ordering,
- SynchronizationScope SynchScope = CrossThread) {
- return Insert(new AtomicCmpXchgInst(Ptr, Cmp, New, Ordering, SynchScope));
+ AtomicCmpXchgInst *
+ CreateAtomicCmpXchg(Value *Ptr, Value *Cmp, Value *New,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope = CrossThread) {
+ return Insert(new AtomicCmpXchgInst(Ptr, Cmp, New, SuccessOrdering,
+ FailureOrdering, SynchScope));
}
AtomicRMWInst *CreateAtomicRMW(AtomicRMWInst::BinOp Op, Value *Ptr, Value *Val,
AtomicOrdering Ordering,
diff --git a/include/llvm/IR/Instructions.h b/include/llvm/IR/Instructions.h
index c249f87..bf39b21 100644
--- a/include/llvm/IR/Instructions.h
+++ b/include/llvm/IR/Instructions.h
@@ -463,7 +463,8 @@ private:
class AtomicCmpXchgInst : public Instruction {
void *operator new(size_t, unsigned) LLVM_DELETED_FUNCTION;
void Init(Value *Ptr, Value *Cmp, Value *NewVal,
- AtomicOrdering Ordering, SynchronizationScope SynchScope);
+ AtomicOrdering SuccessOrdering, AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope);
protected:
AtomicCmpXchgInst *clone_impl() const override;
public:
@@ -472,10 +473,14 @@ public:
return User::operator new(s, 3);
}
AtomicCmpXchgInst(Value *Ptr, Value *Cmp, Value *NewVal,
- AtomicOrdering Ordering, SynchronizationScope SynchScope,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope,
Instruction *InsertBefore = 0);
AtomicCmpXchgInst(Value *Ptr, Value *Cmp, Value *NewVal,
- AtomicOrdering Ordering, SynchronizationScope SynchScope,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope,
BasicBlock *InsertAtEnd);
/// isVolatile - Return true if this is a cmpxchg from a volatile memory
@@ -496,13 +501,20 @@ public:
DECLARE_TRANSPARENT_OPERAND_ACCESSORS(Value);
/// Set the ordering constraint on this cmpxchg.
- void setOrdering(AtomicOrdering Ordering) {
+ void setSuccessOrdering(AtomicOrdering Ordering) {
assert(Ordering != NotAtomic &&
"CmpXchg instructions can only be atomic.");
- setInstructionSubclassData((getSubclassDataFromInstruction() & 3) |
+ setInstructionSubclassData((getSubclassDataFromInstruction() & ~0x1c) |
(Ordering << 2));
}
+ void setFailureOrdering(AtomicOrdering Ordering) {
+ assert(Ordering != NotAtomic &&
+ "CmpXchg instructions can only be atomic.");
+ setInstructionSubclassData((getSubclassDataFromInstruction() & ~0xe0) |
+ (Ordering << 5));
+ }
+
/// Specify whether this cmpxchg is atomic and orders other operations with
/// respect to all concurrently executing threads, or only with respect to
/// signal handlers executing in the same thread.
@@ -512,8 +524,13 @@ public:
}
/// Returns the ordering constraint on this cmpxchg.
- AtomicOrdering getOrdering() const {
- return AtomicOrdering(getSubclassDataFromInstruction() >> 2);
+ AtomicOrdering getSuccessOrdering() const {
+ return AtomicOrdering((getSubclassDataFromInstruction() >> 2) & 7);
+ }
+
+ /// Returns the ordering constraint on this cmpxchg.
+ AtomicOrdering getFailureOrdering() const {
+ return AtomicOrdering((getSubclassDataFromInstruction() >> 5) & 7);
}
/// Returns whether this cmpxchg is atomic between threads or only within a
diff --git a/lib/Analysis/AliasAnalysis.cpp b/lib/Analysis/AliasAnalysis.cpp
index 36ed40d..9583bbe 100644
--- a/lib/Analysis/AliasAnalysis.cpp
+++ b/lib/Analysis/AliasAnalysis.cpp
@@ -338,7 +338,7 @@ AliasAnalysis::getModRefInfo(const VAArgInst *V, const Location &Loc) {
AliasAnalysis::ModRefResult
AliasAnalysis::getModRefInfo(const AtomicCmpXchgInst *CX, const Location &Loc) {
// Acquire/Release cmpxchg has properties that matter for arbitrary addresses.
- if (CX->getOrdering() > Monotonic)
+ if (CX->getSuccessOrdering() > Monotonic)
return ModRef;
// If the cmpxchg address does not alias the location, it does not access it.
diff --git a/lib/AsmParser/LLParser.cpp b/lib/AsmParser/LLParser.cpp
index a4bbcfc..fddada9 100644
--- a/lib/AsmParser/LLParser.cpp
+++ b/lib/AsmParser/LLParser.cpp
@@ -1517,6 +1517,15 @@ bool LLParser::ParseScopeAndOrdering(bool isAtomic, SynchronizationScope &Scope,
Scope = CrossThread;
if (EatIfPresent(lltok::kw_singlethread))
Scope = SingleThread;
+
+ return ParseOrdering(Ordering);
+}
+
+/// ParseOrdering
+/// ::= AtomicOrdering
+///
+/// This sets Ordering to the parsed value.
+bool LLParser::ParseOrdering(AtomicOrdering &Ordering) {
switch (Lex.getKind()) {
default: return TokError("Expected ordering on atomic instruction");
case lltok::kw_unordered: Ordering = Unordered; break;
@@ -1529,7 +1538,6 @@ bool LLParser::ParseScopeAndOrdering(bool isAtomic, SynchronizationScope &Scope,
Lex.Lex();
return false;
}
-
/// ParseOptionalStackAlignment
/// ::= /* empty */
/// ::= 'alignstack' '(' 4 ')'
@@ -4198,11 +4206,12 @@ int LLParser::ParseStore(Instruction *&Inst, PerFunctionState &PFS) {
/// ParseCmpXchg
/// ::= 'cmpxchg' 'volatile'? TypeAndValue ',' TypeAndValue ',' TypeAndValue
-/// 'singlethread'? AtomicOrdering
+/// 'singlethread'? AtomicOrdering AtomicOrdering
int LLParser::ParseCmpXchg(Instruction *&Inst, PerFunctionState &PFS) {
Value *Ptr, *Cmp, *New; LocTy PtrLoc, CmpLoc, NewLoc;
bool AteExtraComma = false;
- AtomicOrdering Ordering = NotAtomic;
+ AtomicOrdering SuccessOrdering = NotAtomic;
+ AtomicOrdering FailureOrdering = NotAtomic;
SynchronizationScope Scope = CrossThread;
bool isVolatile = false;
@@ -4214,11 +4223,16 @@ int LLParser::ParseCmpXchg(Instruction *&Inst, PerFunctionState &PFS) {
ParseTypeAndValue(Cmp, CmpLoc, PFS) ||
ParseToken(lltok::comma, "expected ',' after cmpxchg cmp operand") ||
ParseTypeAndValue(New, NewLoc, PFS) ||
- ParseScopeAndOrdering(true /*Always atomic*/, Scope, Ordering))
+ ParseScopeAndOrdering(true /*Always atomic*/, Scope, SuccessOrdering) ||
+ ParseOrdering(FailureOrdering))
return true;
- if (Ordering == Unordered)
+ if (SuccessOrdering == Unordered || FailureOrdering == Unordered)
return TokError("cmpxchg cannot be unordered");
+ if (SuccessOrdering < FailureOrdering)
+ return TokError("cmpxchg must be at least as ordered on success as failure");
+ if (FailureOrdering == Release || FailureOrdering == AcquireRelease)
+ return TokError("cmpxchg failure ordering cannot include release semantics");
if (!Ptr->getType()->isPointerTy())
return Error(PtrLoc, "cmpxchg operand must be a pointer");
if (cast<PointerType>(Ptr->getType())->getElementType() != Cmp->getType())
@@ -4232,8 +4246,8 @@ int LLParser::ParseCmpXchg(Instruction *&Inst, PerFunctionState &PFS) {
return Error(NewLoc, "cmpxchg operand must be power-of-two byte-sized"
" integer");
- AtomicCmpXchgInst *CXI =
- new AtomicCmpXchgInst(Ptr, Cmp, New, Ordering, Scope);
+ AtomicCmpXchgInst *CXI = new AtomicCmpXchgInst(Ptr, Cmp, New, SuccessOrdering,
+ FailureOrdering, Scope);
CXI->setVolatile(isVolatile);
Inst = CXI;
return AteExtraComma ? InstExtraComma : InstNormal;
diff --git a/lib/AsmParser/LLParser.h b/lib/AsmParser/LLParser.h
index 790ffd2..294a1e1 100644
--- a/lib/AsmParser/LLParser.h
+++ b/lib/AsmParser/LLParser.h
@@ -209,6 +209,7 @@ namespace llvm {
bool ParseOptionalAlignment(unsigned &Alignment);
bool ParseScopeAndOrdering(bool isAtomic, SynchronizationScope &Scope,
AtomicOrdering &Ordering);
+ bool ParseOrdering(AtomicOrdering &Ordering);
bool ParseOptionalStackAlignment(unsigned &Alignment);
bool ParseOptionalCommaAlign(unsigned &Alignment, bool &AteExtraComma);
bool ParseOptionalCommaInAlloca(bool &IsInAlloca);
diff --git a/lib/Bitcode/Reader/BitcodeReader.cpp b/lib/Bitcode/Reader/BitcodeReader.cpp
index 527fd25..75dd56f 100644
--- a/lib/Bitcode/Reader/BitcodeReader.cpp
+++ b/lib/Bitcode/Reader/BitcodeReader.cpp
@@ -203,6 +203,24 @@ static SynchronizationScope GetDecodedSynchScope(unsigned Val) {
}
}
+static AtomicOrdering UpgradeFailureOrdering(AtomicOrdering SuccessOrdering) {
+ switch (SuccessOrdering) {
+ case NotAtomic:
+ case Unordered:
+ llvm_unreachable("invalid ordering should already have been checked");
+ // case Consume: (not specified yet)
+ case Acquire:
+ case AcquireRelease:
+ return Acquire;
+ case Monotonic:
+ case Release:
+ return Monotonic;
+ case SequentiallyConsistent:
+ return SequentiallyConsistent;
+ }
+ llvm_unreachable("completely invalid atomic ordering");
+}
+
static void UpgradeDLLImportExportLinkage(llvm::GlobalValue *GV, unsigned Val) {
switch (Val) {
case 5: GV->setDLLStorageClass(GlobalValue::DLLImportStorageClass); break;
@@ -210,6 +228,7 @@ static void UpgradeDLLImportExportLinkage(llvm::GlobalValue *GV, unsigned Val) {
}
}
+
namespace llvm {
namespace {
/// @brief A class for maintaining the slot number definition
@@ -2882,7 +2901,8 @@ error_code BitcodeReader::ParseFunctionBody(Function *F) {
break;
}
case bitc::FUNC_CODE_INST_CMPXCHG: {
- // CMPXCHG:[ptrty, ptr, cmp, new, vol, ordering, synchscope]
+ // CMPXCHG:[ptrty, ptr, cmp, new, vol, successordering, synchscope,
+ // failureordering]
unsigned OpNum = 0;
Value *Ptr, *Cmp, *New;
if (getValueTypePair(Record, OpNum, NextValueNo, Ptr) ||
@@ -2890,13 +2910,21 @@ error_code BitcodeReader::ParseFunctionBody(Function *F) {
cast<PointerType>(Ptr->getType())->getElementType(), Cmp) ||
popValue(Record, OpNum, NextValueNo,
cast<PointerType>(Ptr->getType())->getElementType(), New) ||
- OpNum+3 != Record.size())
+ (OpNum + 3 != Record.size() && OpNum + 4 != Record.size()))
return Error(InvalidRecord);
- AtomicOrdering Ordering = GetDecodedOrdering(Record[OpNum+1]);
- if (Ordering == NotAtomic || Ordering == Unordered)
+ AtomicOrdering SuccessOrdering = GetDecodedOrdering(Record[OpNum+1]);
+ if (SuccessOrdering == NotAtomic || SuccessOrdering == Unordered)
return Error(InvalidRecord);
SynchronizationScope SynchScope = GetDecodedSynchScope(Record[OpNum+2]);
- I = new AtomicCmpXchgInst(Ptr, Cmp, New, Ordering, SynchScope);
+
+ AtomicOrdering FailureOrdering;
+ if (Record.size() < 7)
+ FailureOrdering = UpgradeFailureOrdering(SuccessOrdering);
+ else
+ FailureOrdering = GetDecodedOrdering(Record[OpNum+3]);
+
+ I = new AtomicCmpXchgInst(Ptr, Cmp, New, SuccessOrdering, FailureOrdering,
+ SynchScope);
cast<AtomicCmpXchgInst>(I)->setVolatile(Record[OpNum]);
InstructionList.push_back(I);
break;
diff --git a/lib/Bitcode/Writer/BitcodeWriter.cpp b/lib/Bitcode/Writer/BitcodeWriter.cpp
index 718cd12..0093c0f 100644
--- a/lib/Bitcode/Writer/BitcodeWriter.cpp
+++ b/lib/Bitcode/Writer/BitcodeWriter.cpp
@@ -1441,9 +1441,11 @@ static void WriteInstruction(const Instruction &I, unsigned InstID,
pushValue(I.getOperand(2), InstID, Vals, VE); // newval.
Vals.push_back(cast<AtomicCmpXchgInst>(I).isVolatile());
Vals.push_back(GetEncodedOrdering(
- cast<AtomicCmpXchgInst>(I).getOrdering()));
+ cast<AtomicCmpXchgInst>(I).getSuccessOrdering()));
Vals.push_back(GetEncodedSynchScope(
cast<AtomicCmpXchgInst>(I).getSynchScope()));
+ Vals.push_back(GetEncodedOrdering(
+ cast<AtomicCmpXchgInst>(I).getFailureOrdering()));
break;
case Instruction::AtomicRMW:
Code = bitc::FUNC_CODE_INST_ATOMICRMW;
diff --git a/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp b/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
index 174d78d..0a1315d 100644
--- a/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
+++ b/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
@@ -2878,6 +2878,7 @@ void SelectionDAGLegalize::ExpandNode(SDNode *Node) {
Node->getOperand(1), Zero, Zero,
cast<AtomicSDNode>(Node)->getMemOperand(),
cast<AtomicSDNode>(Node)->getOrdering(),
+ cast<AtomicSDNode>(Node)->getOrdering(),
cast<AtomicSDNode>(Node)->getSynchScope());
Results.push_back(Swap.getValue(0));
Results.push_back(Swap.getValue(1));
diff --git a/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp b/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp
index cd7c496..18b2376 100644
--- a/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp
+++ b/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp
@@ -193,10 +193,10 @@ SDValue DAGTypeLegalizer::PromoteIntRes_Atomic1(AtomicSDNode *N) {
SDValue DAGTypeLegalizer::PromoteIntRes_Atomic2(AtomicSDNode *N) {
SDValue Op2 = GetPromotedInteger(N->getOperand(2));
SDValue Op3 = GetPromotedInteger(N->getOperand(3));
- SDValue Res = DAG.getAtomic(N->getOpcode(), SDLoc(N),
- N->getMemoryVT(), N->getChain(), N->getBasePtr(),
- Op2, Op3, N->getMemOperand(), N->getOrdering(),
- N->getSynchScope());
+ SDValue Res = DAG.getAtomic(N->getOpcode(), SDLoc(N), N->getMemoryVT(),
+ N->getChain(), N->getBasePtr(), Op2, Op3,
+ N->getMemOperand(), N->getSuccessOrdering(),
+ N->getFailureOrdering(), N->getSynchScope());
// Legalized the chain result - switch anything that used the old chain to
// use the new one.
ReplaceValueWith(SDValue(N, 1), Res.getValue(1));
@@ -2448,6 +2448,7 @@ void DAGTypeLegalizer::ExpandIntRes_ATOMIC_LOAD(SDNode *N,
N->getOperand(1), Zero, Zero,
cast<AtomicSDNode>(N)->getMemOperand(),
cast<AtomicSDNode>(N)->getOrdering(),
+ cast<AtomicSDNode>(N)->getOrdering(),
cast<AtomicSDNode>(N)->getSynchScope());
ReplaceValueWith(SDValue(N, 0), Swap.getValue(0));
ReplaceValueWith(SDValue(N, 1), Swap.getValue(1));
diff --git a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index eac5512..91dfa73 100644
--- a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -4223,9 +4223,10 @@ SDValue SelectionDAG::getMemset(SDValue Chain, SDLoc dl, SDValue Dst,
}
SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
- SDVTList VTList, SDValue* Ops, unsigned NumOps,
+ SDVTList VTList, SDValue *Ops, unsigned NumOps,
MachineMemOperand *MMO,
- AtomicOrdering Ordering,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
SynchronizationScope SynchScope) {
FoldingSetNodeID ID;
ID.AddInteger(MemVT.getRawBits());
@@ -4247,18 +4248,29 @@ SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
SDNode *N = new (NodeAllocator) AtomicSDNode(Opcode, dl.getIROrder(),
dl.getDebugLoc(), VTList, MemVT,
Ops, DynOps, NumOps, MMO,
- Ordering, SynchScope);
+ SuccessOrdering, FailureOrdering,
+ SynchScope);
CSEMap.InsertNode(N, IP);
AllNodes.push_back(N);
return SDValue(N, 0);
}
SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
- SDValue Chain, SDValue Ptr, SDValue Cmp,
- SDValue Swp, MachinePointerInfo PtrInfo,
- unsigned Alignment,
+ SDVTList VTList, SDValue *Ops, unsigned NumOps,
+ MachineMemOperand *MMO,
AtomicOrdering Ordering,
SynchronizationScope SynchScope) {
+ return getAtomic(Opcode, dl, MemVT, VTList, Ops, NumOps, MMO, Ordering,
+ Ordering, SynchScope);
+}
+
+SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
+ SDValue Chain, SDValue Ptr, SDValue Cmp,
+ SDValue Swp, MachinePointerInfo PtrInfo,
+ unsigned Alignment,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope) {
if (Alignment == 0) // Ensure that codegen never sees alignment 0
Alignment = getEVTAlignment(MemVT);
@@ -4278,15 +4290,16 @@ SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
MF.getMachineMemOperand(PtrInfo, Flags, MemVT.getStoreSize(), Alignment);
return getAtomic(Opcode, dl, MemVT, Chain, Ptr, Cmp, Swp, MMO,
- Ordering, SynchScope);
+ SuccessOrdering, FailureOrdering, SynchScope);
}
SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
- SDValue Chain,
- SDValue Ptr, SDValue Cmp,
- SDValue Swp, MachineMemOperand *MMO,
- AtomicOrdering Ordering,
- SynchronizationScope SynchScope) {
+ SDValue Chain,
+ SDValue Ptr, SDValue Cmp,
+ SDValue Swp, MachineMemOperand *MMO,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope) {
assert(Opcode == ISD::ATOMIC_CMP_SWAP && "Invalid Atomic Op");
assert(Cmp.getValueType() == Swp.getValueType() && "Invalid Atomic Op Types");
@@ -4294,7 +4307,8 @@ SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
SDVTList VTs = getVTList(VT, MVT::Other);
SDValue Ops[] = {Chain, Ptr, Cmp, Swp};
- return getAtomic(Opcode, dl, MemVT, VTs, Ops, 4, MMO, Ordering, SynchScope);
+ return getAtomic(Opcode, dl, MemVT, VTs, Ops, 4, MMO, SuccessOrdering,
+ FailureOrdering, SynchScope);
}
SDValue SelectionDAG::getAtomic(unsigned Opcode, SDLoc dl, EVT MemVT,
diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
index 1d82d3e..3cf1b1c 100644
--- a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
@@ -3605,14 +3605,15 @@ static SDValue InsertFenceForAtomic(SDValue Chain, AtomicOrdering Order,
void SelectionDAGBuilder::visitAtomicCmpXchg(const AtomicCmpXchgInst &I) {
SDLoc dl = getCurSDLoc();
- AtomicOrdering Order = I.getOrdering();
+ AtomicOrdering SuccessOrder = I.getSuccessOrdering();
+ AtomicOrdering FailureOrder = I.getFailureOrdering();
SynchronizationScope Scope = I.getSynchScope();
SDValue InChain = getRoot();
const TargetLowering *TLI = TM.getTargetLowering();
if (TLI->getInsertFencesForAtomic())
- InChain = InsertFenceForAtomic(InChain, Order, Scope, true, dl,
+ InChain = InsertFenceForAtomic(InChain, SuccessOrder, Scope, true, dl,
DAG, *TLI);
SDValue L =
@@ -3623,13 +3624,14 @@ void SelectionDAGBuilder::visitAtomicCmpXchg(const AtomicCmpXchgInst &I) {
getValue(I.getCompareOperand()),
getValue(I.getNewValOperand()),
MachinePointerInfo(I.getPointerOperand()), 0 /* Alignment */,
- TLI->getInsertFencesForAtomic() ? Monotonic : Order,
+ TLI->getInsertFencesForAtomic() ? Monotonic : SuccessOrder,
+ TLI->getInsertFencesForAtomic() ? Monotonic : FailureOrder,
Scope);
SDValue OutChain = L.getValue(1);
if (TLI->getInsertFencesForAtomic())
- OutChain = InsertFenceForAtomic(OutChain, Order, Scope, false, dl,
+ OutChain = InsertFenceForAtomic(OutChain, SuccessOrder, Scope, false, dl,
DAG, *TLI);
setValue(&I, L);
diff --git a/lib/IR/AsmWriter.cpp b/lib/IR/AsmWriter.cpp
index 8a25ff8..22fa969 100644
--- a/lib/IR/AsmWriter.cpp
+++ b/lib/IR/AsmWriter.cpp
@@ -1226,6 +1226,37 @@ void AssemblyWriter::writeAtomic(AtomicOrdering Ordering,
}
}
+void AssemblyWriter::writeAtomicCmpXchg(AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope) {
+ assert(SuccessOrdering != NotAtomic && FailureOrdering != NotAtomic);
+
+ switch (SynchScope) {
+ case SingleThread: Out << " singlethread"; break;
+ case CrossThread: break;
+ }
+
+ switch (SuccessOrdering) {
+ default: Out << " <bad ordering " << int(SuccessOrdering) << ">"; break;
+ case Unordered: Out << " unordered"; break;
+ case Monotonic: Out << " monotonic"; break;
+ case Acquire: Out << " acquire"; break;
+ case Release: Out << " release"; break;
+ case AcquireRelease: Out << " acq_rel"; break;
+ case SequentiallyConsistent: Out << " seq_cst"; break;
+ }
+
+ switch (FailureOrdering) {
+ default: Out << " <bad ordering " << int(FailureOrdering) << ">"; break;
+ case Unordered: Out << " unordered"; break;
+ case Monotonic: Out << " monotonic"; break;
+ case Acquire: Out << " acquire"; break;
+ case Release: Out << " release"; break;
+ case AcquireRelease: Out << " acq_rel"; break;
+ case SequentiallyConsistent: Out << " seq_cst"; break;
+ }
+}
+
void AssemblyWriter::writeParamOperand(const Value *Operand,
AttributeSet Attrs, unsigned Idx) {
if (Operand == 0) {
@@ -2018,7 +2049,8 @@ void AssemblyWriter::printInstruction(const Instruction &I) {
if (SI->getAlignment())
Out << ", align " << SI->getAlignment();
} else if (const AtomicCmpXchgInst *CXI = dyn_cast<AtomicCmpXchgInst>(&I)) {
- writeAtomic(CXI->getOrdering(), CXI->getSynchScope());
+ writeAtomicCmpXchg(CXI->getSuccessOrdering(), CXI->getFailureOrdering(),
+ CXI->getSynchScope());
} else if (const AtomicRMWInst *RMWI = dyn_cast<AtomicRMWInst>(&I)) {
writeAtomic(RMWI->getOrdering(), RMWI->getSynchScope());
} else if (const FenceInst *FI = dyn_cast<FenceInst>(&I)) {
diff --git a/lib/IR/AsmWriter.h b/lib/IR/AsmWriter.h
index 222d3a4..b4ce6de 100644
--- a/lib/IR/AsmWriter.h
+++ b/lib/IR/AsmWriter.h
@@ -90,6 +90,9 @@ public:
void writeOperand(const Value *Op, bool PrintType);
void writeParamOperand(const Value *Operand, AttributeSet Attrs,unsigned Idx);
void writeAtomic(AtomicOrdering Ordering, SynchronizationScope SynchScope);
+ void writeAtomicCmpXchg(AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
+ SynchronizationScope SynchScope);
void writeAllMDNodes();
void writeMDNode(unsigned Slot, const MDNode *Node);
diff --git a/lib/IR/Instruction.cpp b/lib/IR/Instruction.cpp
index 4bcf748..47b15e9 100644
--- a/lib/IR/Instruction.cpp
+++ b/lib/IR/Instruction.cpp
@@ -314,7 +314,10 @@ bool Instruction::isIdenticalToWhenDefined(const Instruction *I) const {
FI->getSynchScope() == cast<FenceInst>(FI)->getSynchScope();
if (const AtomicCmpXchgInst *CXI = dyn_cast<AtomicCmpXchgInst>(this))
return CXI->isVolatile() == cast<AtomicCmpXchgInst>(I)->isVolatile() &&
- CXI->getOrdering() == cast<AtomicCmpXchgInst>(I)->getOrdering() &&
+ CXI->getSuccessOrdering() ==
+ cast<AtomicCmpXchgInst>(I)->getSuccessOrdering() &&
+ CXI->getFailureOrdering() ==
+ cast<AtomicCmpXchgInst>(I)->getFailureOrdering() &&
CXI->getSynchScope() == cast<AtomicCmpXchgInst>(I)->getSynchScope();
if (const AtomicRMWInst *RMWI = dyn_cast<AtomicRMWInst>(this))
return RMWI->getOperation() == cast<AtomicRMWInst>(I)->getOperation() &&
@@ -388,7 +391,10 @@ bool Instruction::isSameOperationAs(const Instruction *I,
FI->getSynchScope() == cast<FenceInst>(I)->getSynchScope();
if (const AtomicCmpXchgInst *CXI = dyn_cast<AtomicCmpXchgInst>(this))
return CXI->isVolatile() == cast<AtomicCmpXchgInst>(I)->isVolatile() &&
- CXI->getOrdering() == cast<AtomicCmpXchgInst>(I)->getOrdering() &&
+ CXI->getSuccessOrdering() ==
+ cast<AtomicCmpXchgInst>(I)->getSuccessOrdering() &&
+ CXI->getFailureOrdering() ==
+ cast<AtomicCmpXchgInst>(I)->getFailureOrdering() &&
CXI->getSynchScope() == cast<AtomicCmpXchgInst>(I)->getSynchScope();
if (const AtomicRMWInst *RMWI = dyn_cast<AtomicRMWInst>(this))
return RMWI->getOperation() == cast<AtomicRMWInst>(I)->getOperation() &&
diff --git a/lib/IR/Instructions.cpp b/lib/IR/Instructions.cpp
index c3a34a2..0e812b7 100644
--- a/lib/IR/Instructions.cpp
+++ b/lib/IR/Instructions.cpp
@@ -1216,12 +1216,14 @@ void StoreInst::setAlignment(unsigned Align) {
//===----------------------------------------------------------------------===//
void AtomicCmpXchgInst::Init(Value *Ptr, Value *Cmp, Value *NewVal,
- AtomicOrdering Ordering,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
SynchronizationScope SynchScope) {
Op<0>() = Ptr;
Op<1>() = Cmp;
Op<2>() = NewVal;
- setOrdering(Ordering);
+ setSuccessOrdering(SuccessOrdering);
+ setFailureOrdering(FailureOrdering);
setSynchScope(SynchScope);
assert(getOperand(0) && getOperand(1) && getOperand(2) &&
@@ -1234,30 +1236,36 @@ void AtomicCmpXchgInst::Init(Value *Ptr, Value *Cmp, Value *NewVal,
assert(getOperand(2)->getType() ==
cast<PointerType>(getOperand(0)->getType())->getElementType()
&& "Ptr must be a pointer to NewVal type!");
- assert(Ordering != NotAtomic &&
+ assert(SuccessOrdering != NotAtomic &&
+ "AtomicCmpXchg instructions must be atomic!");
+ assert(FailureOrdering != NotAtomic &&
"AtomicCmpXchg instructions must be atomic!");
+ assert(SuccessOrdering >= FailureOrdering &&
+ "AtomicCmpXchg success ordering must be at least as strong as fail");
}
AtomicCmpXchgInst::AtomicCmpXchgInst(Value *Ptr, Value *Cmp, Value *NewVal,
- AtomicOrdering Ordering,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
SynchronizationScope SynchScope,
Instruction *InsertBefore)
: Instruction(Cmp->getType(), AtomicCmpXchg,
OperandTraits<AtomicCmpXchgInst>::op_begin(this),
OperandTraits<AtomicCmpXchgInst>::operands(this),
InsertBefore) {
- Init(Ptr, Cmp, NewVal, Ordering, SynchScope);
+ Init(Ptr, Cmp, NewVal, SuccessOrdering, FailureOrdering, SynchScope);
}
AtomicCmpXchgInst::AtomicCmpXchgInst(Value *Ptr, Value *Cmp, Value *NewVal,
- AtomicOrdering Ordering,
+ AtomicOrdering SuccessOrdering,
+ AtomicOrdering FailureOrdering,
SynchronizationScope SynchScope,
BasicBlock *InsertAtEnd)
: Instruction(Cmp->getType(), AtomicCmpXchg,
OperandTraits<AtomicCmpXchgInst>::op_begin(this),
OperandTraits<AtomicCmpXchgInst>::operands(this),
InsertAtEnd) {
- Init(Ptr, Cmp, NewVal, Ordering, SynchScope);
+ Init(Ptr, Cmp, NewVal, SuccessOrdering, FailureOrdering, SynchScope);
}
//===----------------------------------------------------------------------===//
@@ -3597,7 +3605,8 @@ StoreInst *StoreInst::clone_impl() const {
AtomicCmpXchgInst *AtomicCmpXchgInst::clone_impl() const {
AtomicCmpXchgInst *Result =
new AtomicCmpXchgInst(getOperand(0), getOperand(1), getOperand(2),
- getOrdering(), getSynchScope());
+ getSuccessOrdering(), getFailureOrdering(),
+ getSynchScope());
Result->setVolatile(isVolatile());
return Result;
}
diff --git a/lib/IR/Verifier.cpp b/lib/IR/Verifier.cpp
index deb708d..4e1ad30 100644
--- a/lib/IR/Verifier.cpp
+++ b/lib/IR/Verifier.cpp
@@ -1829,10 +1829,23 @@ void Verifier::visitAllocaInst(AllocaInst &AI) {
}
void Verifier::visitAtomicCmpXchgInst(AtomicCmpXchgInst &CXI) {
- Assert1(CXI.getOrdering() != NotAtomic,
+
+ // FIXME: more conditions???
+ Assert1(CXI.getSuccessOrdering() != NotAtomic,
+ "cmpxchg instructions must be atomic.", &CXI);
+ Assert1(CXI.getFailureOrdering() != NotAtomic,
"cmpxchg instructions must be atomic.", &CXI);
- Assert1(CXI.getOrdering() != Unordered,
+ Assert1(CXI.getSuccessOrdering() != Unordered,
"cmpxchg instructions cannot be unordered.", &CXI);
+ Assert1(CXI.getFailureOrdering() != Unordered,
+ "cmpxchg instructions cannot be unordered.", &CXI);
+ Assert1(CXI.getSuccessOrdering() >= CXI.getFailureOrdering(),
+ "cmpxchg instructions be at least as constrained on success as fail",
+ &CXI);
+ Assert1(CXI.getFailureOrdering() != Release &&
+ CXI.getFailureOrdering() != AcquireRelease,
+ "cmpxchg failure ordering cannot include release semantics", &CXI);
+
PointerType *PTy = dyn_cast<PointerType>(CXI.getOperand(0)->getType());
Assert1(PTy, "First cmpxchg operand must be a pointer.", &CXI);
Type *ElTy = PTy->getElementType();
diff --git a/lib/Target/ARM/ARMISelLowering.cpp b/lib/Target/ARM/ARMISelLowering.cpp
index 691961e..aa7ca7f 100644
--- a/lib/Target/ARM/ARMISelLowering.cpp
+++ b/lib/Target/ARM/ARMISelLowering.cpp
@@ -6054,10 +6054,10 @@ ReplaceATOMIC_OP_64(SDNode *Node, SmallVectorImpl<SDValue>& Results,
Node->getOperand(i), DAG.getIntPtrConstant(1)));
}
SDVTList Tys = DAG.getVTList(MVT::i32, MVT::i32, MVT::Other);
- SDValue Result =
- DAG.getAtomic(Node->getOpcode(), dl, MVT::i64, Tys, Ops.data(), Ops.size(),
- cast<MemSDNode>(Node)->getMemOperand(), AN->getOrdering(),
- AN->getSynchScope());
+ SDValue Result = DAG.getAtomic(
+ Node->getOpcode(), dl, MVT::i64, Tys, Ops.data(), Ops.size(),
+ cast<MemSDNode>(Node)->getMemOperand(), AN->getSuccessOrdering(),
+ AN->getFailureOrdering(), AN->getSynchScope());
SDValue OpsF[] = { Result.getValue(0), Result.getValue(1) };
Results.push_back(DAG.getNode(ISD::BUILD_PAIR, dl, MVT::i64, OpsF, 2));
Results.push_back(Result.getValue(2));
diff --git a/lib/Target/CppBackend/CPPBackend.cpp b/lib/Target/CppBackend/CPPBackend.cpp
index b137099..31585d9 100644
--- a/lib/Target/CppBackend/CPPBackend.cpp
+++ b/lib/Target/CppBackend/CPPBackend.cpp
@@ -1567,12 +1567,16 @@ void CppWriter::printInstruction(const Instruction *I,
}
case Instruction::AtomicCmpXchg: {
const AtomicCmpXchgInst *cxi = cast<AtomicCmpXchgInst>(I);
- StringRef Ordering = ConvertAtomicOrdering(cxi->getOrdering());
+ StringRef SuccessOrdering =
+ ConvertAtomicOrdering(cxi->getSuccessOrdering());
+ StringRef FailureOrdering =
+ ConvertAtomicOrdering(cxi->getFailureOrdering());
StringRef CrossThread = ConvertAtomicSynchScope(cxi->getSynchScope());
Out << "AtomicCmpXchgInst* " << iName
<< " = new AtomicCmpXchgInst("
<< opNames[0] << ", " << opNames[1] << ", " << opNames[2] << ", "
- << Ordering << ", " << CrossThread << ", " << bbname
+ << SuccessOrdering << ", " << FailureOrdering << ", "
+ << CrossThread << ", " << bbname
<< ");";
nl(Out) << iName << "->setName(\"";
printEscapedString(cxi->getName());
diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 951d3de..4f8a788 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -13803,6 +13803,7 @@ static void ReplaceATOMIC_LOAD(SDNode *Node,
Node->getOperand(1), Zero, Zero,
cast<AtomicSDNode>(Node)->getMemOperand(),
cast<AtomicSDNode>(Node)->getOrdering(),
+ cast<AtomicSDNode>(Node)->getOrdering(),
cast<AtomicSDNode>(Node)->getSynchScope());
Results.push_back(Swap.getValue(0));
Results.push_back(Swap.getValue(1));
diff --git a/lib/Transforms/IPO/MergeFunctions.cpp b/lib/Transforms/IPO/MergeFunctions.cpp
index 0ab1c76..a3c3f0f 100644
--- a/lib/Transforms/IPO/MergeFunctions.cpp
+++ b/lib/Transforms/IPO/MergeFunctions.cpp
@@ -341,7 +341,10 @@ bool FunctionComparator::isEquivalentOperation(const Instruction *I1,
FI->getSynchScope() == cast<FenceInst>(I2)->getSynchScope();
if (const AtomicCmpXchgInst *CXI = dyn_cast<AtomicCmpXchgInst>(I1))
return CXI->isVolatile() == cast<AtomicCmpXchgInst>(I2)->isVolatile() &&
- CXI->getOrdering() == cast<AtomicCmpXchgInst>(I2)->getOrdering() &&
+ CXI->getSuccessOrdering() ==
+ cast<AtomicCmpXchgInst>(I2)->getSuccessOrdering() &&
+ CXI->getFailureOrdering() ==
+ cast<AtomicCmpXchgInst>(I2)->getFailureOrdering() &&
CXI->getSynchScope() == cast<AtomicCmpXchgInst>(I2)->getSynchScope();
if (const AtomicRMWInst *RMWI = dyn_cast<AtomicRMWInst>(I1))
return RMWI->getOperation() == cast<AtomicRMWInst>(I2)->getOperation() &&
diff --git a/lib/Transforms/Instrumentation/MemorySanitizer.cpp b/lib/Transforms/Instrumentation/MemorySanitizer.cpp
index b4ae443..b158f1f 100644
--- a/lib/Transforms/Instrumentation/MemorySanitizer.cpp
+++ b/lib/Transforms/Instrumentation/MemorySanitizer.cpp
@@ -1088,7 +1088,7 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
void visitAtomicCmpXchgInst(AtomicCmpXchgInst &I) {
handleCASOrRMW(I);
- I.setOrdering(addReleaseOrdering(I.getOrdering()));
+ I.setSuccessOrdering(addReleaseOrdering(I.getSuccessOrdering()));
}
// Vector manipulation.
diff --git a/lib/Transforms/Instrumentation/ThreadSanitizer.cpp b/lib/Transforms/Instrumentation/ThreadSanitizer.cpp
index fed7508..6ad5b6f 100644
--- a/lib/Transforms/Instrumentation/ThreadSanitizer.cpp
+++ b/lib/Transforms/Instrumentation/ThreadSanitizer.cpp
@@ -556,8 +556,8 @@ bool ThreadSanitizer::instrumentAtomic(Instruction *I) {
Value *Args[] = {IRB.CreatePointerCast(Addr, PtrTy),
IRB.CreateIntCast(CASI->getCompareOperand(), Ty, false),
IRB.CreateIntCast(CASI->getNewValOperand(), Ty, false),
- createOrdering(&IRB, CASI->getOrdering()),
- createFailOrdering(&IRB, CASI->getOrdering())};
+ createOrdering(&IRB, CASI->getSuccessOrdering()),
+ createFailOrdering(&IRB, CASI->getFailureOrdering())};
CallInst *C = CallInst::Create(TsanAtomicCAS[Idx], ArrayRef<Value*>(Args));
ReplaceInstWithInst(I, C);
} else if (FenceInst *FI = dyn_cast<FenceInst>(I)) {
diff --git a/test/Assembler/atomic.ll b/test/Assembler/atomic.ll
index b245cde..9d80a83 100644
--- a/test/Assembler/atomic.ll
+++ b/test/Assembler/atomic.ll
@@ -10,10 +10,10 @@ define void @f(i32* %x) {
store atomic i32 3, i32* %x release, align 4
; CHECK: store atomic volatile i32 3, i32* %x singlethread monotonic, align 4
store atomic volatile i32 3, i32* %x singlethread monotonic, align 4
- ; CHECK: cmpxchg i32* %x, i32 1, i32 0 singlethread monotonic
- cmpxchg i32* %x, i32 1, i32 0 singlethread monotonic
- ; CHECK: cmpxchg volatile i32* %x, i32 0, i32 1 acq_rel
- cmpxchg volatile i32* %x, i32 0, i32 1 acq_rel
+ ; CHECK: cmpxchg i32* %x, i32 1, i32 0 singlethread monotonic monotonic
+ cmpxchg i32* %x, i32 1, i32 0 singlethread monotonic monotonic
+ ; CHECK: cmpxchg volatile i32* %x, i32 0, i32 1 acq_rel acquire
+ cmpxchg volatile i32* %x, i32 0, i32 1 acq_rel acquire
; CHECK: atomicrmw add i32* %x, i32 10 seq_cst
atomicrmw add i32* %x, i32 10 seq_cst
; CHECK: atomicrmw volatile xchg i32* %x, i32 10 monotonic
diff --git a/test/Bitcode/memInstructions.3.2.ll b/test/Bitcode/memInstructions.3.2.ll
index 868e4b5..21c3deb 100644
--- a/test/Bitcode/memInstructions.3.2.ll
+++ b/test/Bitcode/memInstructions.3.2.ll
@@ -223,69 +223,69 @@ define void @cmpxchg(i32* %ptr,i32 %cmp,i32 %new){
entry:
;cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <ordering>
-; CHECK: %res1 = cmpxchg i32* %ptr, i32 %cmp, i32 %new monotonic
- %res1 = cmpxchg i32* %ptr, i32 %cmp, i32 %new monotonic
+; CHECK: %res1 = cmpxchg i32* %ptr, i32 %cmp, i32 %new monotonic monotonic
+ %res1 = cmpxchg i32* %ptr, i32 %cmp, i32 %new monotonic monotonic
-; CHECK-NEXT: %res2 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new monotonic
- %res2 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new monotonic
+; CHECK-NEXT: %res2 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new monotonic monotonic
+ %res2 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new monotonic monotonic
-; CHECK-NEXT: %res3 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread monotonic
- %res3 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread monotonic
+; CHECK-NEXT: %res3 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread monotonic monotonic
+ %res3 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread monotonic monotonic
-; CHECK-NEXT: %res4 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread monotonic
- %res4 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread monotonic
+; CHECK-NEXT: %res4 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread monotonic monotonic
+ %res4 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread monotonic monotonic
-; CHECK-NEXT: %res5 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acquire
- %res5 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acquire
+; CHECK-NEXT: %res5 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acquire acquire
+ %res5 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acquire acquire
-; CHECK-NEXT: %res6 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acquire
- %res6 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acquire
+; CHECK-NEXT: %res6 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acquire acquire
+ %res6 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acquire acquire
-; CHECK-NEXT: %res7 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acquire
- %res7 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acquire
+; CHECK-NEXT: %res7 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acquire acquire
+ %res7 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acquire acquire
-; CHECK-NEXT: %res8 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acquire
- %res8 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acquire
+; CHECK-NEXT: %res8 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acquire acquire
+ %res8 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acquire acquire
-; CHECK-NEXT: %res9 = cmpxchg i32* %ptr, i32 %cmp, i32 %new release
- %res9 = cmpxchg i32* %ptr, i32 %cmp, i32 %new release
+; CHECK-NEXT: %res9 = cmpxchg i32* %ptr, i32 %cmp, i32 %new release monotonic
+ %res9 = cmpxchg i32* %ptr, i32 %cmp, i32 %new release monotonic
-; CHECK-NEXT: %res10 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new release
- %res10 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new release
+; CHECK-NEXT: %res10 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new release monotonic
+ %res10 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new release monotonic
-; CHECK-NEXT: %res11 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread release
- %res11 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread release
+; CHECK-NEXT: %res11 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread release monotonic
+ %res11 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread release monotonic
-; CHECK-NEXT: %res12 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread release
- %res12 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread release
+; CHECK-NEXT: %res12 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread release monotonic
+ %res12 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread release monotonic
-; CHECK-NEXT: %res13 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acq_rel
- %res13 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acq_rel
+; CHECK-NEXT: %res13 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acq_rel acquire
+ %res13 = cmpxchg i32* %ptr, i32 %cmp, i32 %new acq_rel acquire
-; CHECK-NEXT: %res14 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acq_rel
- %res14 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acq_rel
+; CHECK-NEXT: %res14 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acq_rel acquire
+ %res14 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new acq_rel acquire
-; CHECK-NEXT: %res15 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel
- %res15 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel
+; CHECK-NEXT: %res15 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel acquire
+ %res15 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel acquire
-; CHECK-NEXT: %res16 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel
- %res16 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel
+; CHECK-NEXT: %res16 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel acquire
+ %res16 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread acq_rel acquire
-; CHECK-NEXT: %res17 = cmpxchg i32* %ptr, i32 %cmp, i32 %new seq_cst
- %res17 = cmpxchg i32* %ptr, i32 %cmp, i32 %new seq_cst
+; CHECK-NEXT: %res17 = cmpxchg i32* %ptr, i32 %cmp, i32 %new seq_cst seq_cst
+ %res17 = cmpxchg i32* %ptr, i32 %cmp, i32 %new seq_cst seq_cst
-; CHECK-NEXT: %res18 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new seq_cst
- %res18 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new seq_cst
+; CHECK-NEXT: %res18 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new seq_cst seq_cst
+ %res18 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new seq_cst seq_cst
-; CHECK-NEXT: %res19 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst
- %res19 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst
+; CHECK-NEXT: %res19 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst seq_cst
+ %res19 = cmpxchg i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst seq_cst
-; CHECK-NEXT: %res20 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst
- %res20 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst
+; CHECK-NEXT: %res20 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst seq_cst
+ %res20 = cmpxchg volatile i32* %ptr, i32 %cmp, i32 %new singlethread seq_cst seq_cst
ret void
}
diff --git a/test/CodeGen/AArch64/atomic-ops.ll b/test/CodeGen/AArch64/atomic-ops.ll
index 5857faf..5fe2936 100644
--- a/test/CodeGen/AArch64/atomic-ops.ll
+++ b/test/CodeGen/AArch64/atomic-ops.ll
@@ -897,7 +897,7 @@ define i64 @test_atomic_load_umax_i64(i64 %offset) nounwind {
define i8 @test_atomic_cmpxchg_i8(i8 %wanted, i8 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i8:
- %old = cmpxchg i8* @var8, i8 %wanted, i8 %new acquire
+ %old = cmpxchg i8* @var8, i8 %wanted, i8 %new acquire acquire
; CHECK-NOT: dmb
; CHECK: adrp [[TMPADDR:x[0-9]+]], var8
; CHECK: add x[[ADDR:[0-9]+]], [[TMPADDR]], #:lo12:var8
@@ -920,7 +920,7 @@ define i8 @test_atomic_cmpxchg_i8(i8 %wanted, i8 %new) nounwind {
define i16 @test_atomic_cmpxchg_i16(i16 %wanted, i16 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i16:
- %old = cmpxchg i16* @var16, i16 %wanted, i16 %new seq_cst
+ %old = cmpxchg i16* @var16, i16 %wanted, i16 %new seq_cst seq_cst
; CHECK-NOT: dmb
; CHECK: adrp [[TMPADDR:x[0-9]+]], var16
; CHECK: add x[[ADDR:[0-9]+]], [[TMPADDR]], #:lo12:var16
@@ -943,7 +943,7 @@ define i16 @test_atomic_cmpxchg_i16(i16 %wanted, i16 %new) nounwind {
define i32 @test_atomic_cmpxchg_i32(i32 %wanted, i32 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i32:
- %old = cmpxchg i32* @var32, i32 %wanted, i32 %new release
+ %old = cmpxchg i32* @var32, i32 %wanted, i32 %new release monotonic
; CHECK-NOT: dmb
; CHECK: adrp [[TMPADDR:x[0-9]+]], var32
; CHECK: add x[[ADDR:[0-9]+]], [[TMPADDR]], #:lo12:var32
@@ -966,7 +966,7 @@ define i32 @test_atomic_cmpxchg_i32(i32 %wanted, i32 %new) nounwind {
define i64 @test_atomic_cmpxchg_i64(i64 %wanted, i64 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i64:
- %old = cmpxchg i64* @var64, i64 %wanted, i64 %new monotonic
+ %old = cmpxchg i64* @var64, i64 %wanted, i64 %new monotonic monotonic
; CHECK-NOT: dmb
; CHECK: adrp [[TMPADDR:x[0-9]+]], var64
; CHECK: add x[[ADDR:[0-9]+]], [[TMPADDR]], #:lo12:var64
diff --git a/test/CodeGen/ARM/atomic-64bit.ll b/test/CodeGen/ARM/atomic-64bit.ll
index 0477d4f..7cf45eb 100644
--- a/test/CodeGen/ARM/atomic-64bit.ll
+++ b/test/CodeGen/ARM/atomic-64bit.ll
@@ -171,7 +171,7 @@ define i64 @test7(i64* %ptr, i64 %val1, i64 %val2) {
; CHECK-THUMB: bne
; CHECK-THUMB: dmb {{ish$}}
- %r = cmpxchg i64* %ptr, i64 %val1, i64 %val2 seq_cst
+ %r = cmpxchg i64* %ptr, i64 %val1, i64 %val2 seq_cst seq_cst
ret i64 %r
}
diff --git a/test/CodeGen/ARM/atomic-cmp.ll b/test/CodeGen/ARM/atomic-cmp.ll
index 51ada69..a473807 100644
--- a/test/CodeGen/ARM/atomic-cmp.ll
+++ b/test/CodeGen/ARM/atomic-cmp.ll
@@ -10,6 +10,6 @@ define i8 @t(i8* %a, i8 %b, i8 %c) nounwind {
; T2-LABEL: t:
; T2: ldrexb
; T2: strexb
- %tmp0 = cmpxchg i8* %a, i8 %b, i8 %c monotonic
+ %tmp0 = cmpxchg i8* %a, i8 %b, i8 %c monotonic monotonic
ret i8 %tmp0
}
diff --git a/test/CodeGen/ARM/atomic-ops-v8.ll b/test/CodeGen/ARM/atomic-ops-v8.ll
index 3f93929..87e76a4 100644
--- a/test/CodeGen/ARM/atomic-ops-v8.ll
+++ b/test/CodeGen/ARM/atomic-ops-v8.ll
@@ -987,7 +987,7 @@ define i64 @test_atomic_load_umax_i64(i64 %offset) nounwind {
define i8 @test_atomic_cmpxchg_i8(i8 %wanted, i8 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i8:
- %old = cmpxchg i8* @var8, i8 %wanted, i8 %new acquire
+ %old = cmpxchg i8* @var8, i8 %wanted, i8 %new acquire acquire
; CHECK-NOT: dmb
; CHECK-NOT: mcr
; CHECK: movw r[[ADDR:[0-9]+]], :lower16:var8
@@ -1013,7 +1013,7 @@ define i8 @test_atomic_cmpxchg_i8(i8 %wanted, i8 %new) nounwind {
define i16 @test_atomic_cmpxchg_i16(i16 %wanted, i16 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i16:
- %old = cmpxchg i16* @var16, i16 %wanted, i16 %new seq_cst
+ %old = cmpxchg i16* @var16, i16 %wanted, i16 %new seq_cst seq_cst
; CHECK-NOT: dmb
; CHECK-NOT: mcr
; CHECK: movw r[[ADDR:[0-9]+]], :lower16:var16
@@ -1039,7 +1039,7 @@ define i16 @test_atomic_cmpxchg_i16(i16 %wanted, i16 %new) nounwind {
define i32 @test_atomic_cmpxchg_i32(i32 %wanted, i32 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i32:
- %old = cmpxchg i32* @var32, i32 %wanted, i32 %new release
+ %old = cmpxchg i32* @var32, i32 %wanted, i32 %new release monotonic
; CHECK-NOT: dmb
; CHECK-NOT: mcr
; CHECK: movw r[[ADDR:[0-9]+]], :lower16:var32
@@ -1065,7 +1065,7 @@ define i32 @test_atomic_cmpxchg_i32(i32 %wanted, i32 %new) nounwind {
define i64 @test_atomic_cmpxchg_i64(i64 %wanted, i64 %new) nounwind {
; CHECK-LABEL: test_atomic_cmpxchg_i64:
- %old = cmpxchg i64* @var64, i64 %wanted, i64 %new monotonic
+ %old = cmpxchg i64* @var64, i64 %wanted, i64 %new monotonic monotonic
; CHECK-NOT: dmb
; CHECK-NOT: mcr
; CHECK: movw r[[ADDR:[0-9]+]], :lower16:var64
diff --git a/test/CodeGen/Mips/atomic.ll b/test/CodeGen/Mips/atomic.ll
index 0e60fe1..77d7bf3 100644
--- a/test/CodeGen/Mips/atomic.ll
+++ b/test/CodeGen/Mips/atomic.ll
@@ -77,7 +77,7 @@ entry:
%newval.addr = alloca i32, align 4
store i32 %newval, i32* %newval.addr, align 4
%tmp = load i32* %newval.addr, align 4
- %0 = cmpxchg i32* @x, i32 %oldval, i32 %tmp monotonic
+ %0 = cmpxchg i32* @x, i32 %oldval, i32 %tmp monotonic monotonic
ret i32 %0
; CHECK-EL-LABEL: AtomicCmpSwap32:
@@ -333,7 +333,7 @@ entry:
define signext i8 @AtomicCmpSwap8(i8 signext %oldval, i8 signext %newval) nounwind {
entry:
- %0 = cmpxchg i8* @y, i8 %oldval, i8 %newval monotonic
+ %0 = cmpxchg i8* @y, i8 %oldval, i8 %newval monotonic monotonic
ret i8 %0
; CHECK-EL-LABEL: AtomicCmpSwap8:
@@ -429,7 +429,7 @@ entry:
define i32 @zeroreg() nounwind {
entry:
- %0 = cmpxchg i32* @a, i32 1, i32 0 seq_cst
+ %0 = cmpxchg i32* @a, i32 1, i32 0 seq_cst seq_cst
%1 = icmp eq i32 %0, 1
%conv = zext i1 %1 to i32
ret i32 %conv
diff --git a/test/CodeGen/Mips/atomicops.ll b/test/CodeGen/Mips/atomicops.ll
index 0f0f01a..dc07c63 100644
--- a/test/CodeGen/Mips/atomicops.ll
+++ b/test/CodeGen/Mips/atomicops.ll
@@ -20,7 +20,7 @@ entry:
%add.i = add nsw i32 %0, 2
%1 = load volatile i32* %x, align 4
%call1 = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([8 x i8]* @.str, i32 0, i32 0), i32 %add.i, i32 %1) nounwind
- %2 = cmpxchg i32* %x, i32 1, i32 2 seq_cst
+ %2 = cmpxchg i32* %x, i32 1, i32 2 seq_cst seq_cst
%3 = load volatile i32* %x, align 4
%call2 = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([8 x i8]* @.str, i32 0, i32 0), i32 %2, i32 %3) nounwind
%4 = atomicrmw xchg i32* %x, i32 1 seq_cst
diff --git a/test/CodeGen/PowerPC/Atomics-32.ll b/test/CodeGen/PowerPC/Atomics-32.ll
index 64f1495..b5c03e2 100644
--- a/test/CodeGen/PowerPC/Atomics-32.ll
+++ b/test/CodeGen/PowerPC/Atomics-32.ll
@@ -529,63 +529,63 @@ define void @test_compare_and_swap() nounwind {
entry:
%0 = load i8* @uc, align 1
%1 = load i8* @sc, align 1
- %2 = cmpxchg i8* @sc, i8 %0, i8 %1 monotonic
+ %2 = cmpxchg i8* @sc, i8 %0, i8 %1 monotonic monotonic
store i8 %2, i8* @sc, align 1
%3 = load i8* @uc, align 1
%4 = load i8* @sc, align 1
- %5 = cmpxchg i8* @uc, i8 %3, i8 %4 monotonic
+ %5 = cmpxchg i8* @uc, i8 %3, i8 %4 monotonic monotonic
store i8 %5, i8* @uc, align 1
%6 = load i8* @uc, align 1
%7 = zext i8 %6 to i16
%8 = load i8* @sc, align 1
%9 = sext i8 %8 to i16
%10 = bitcast i8* bitcast (i16* @ss to i8*) to i16*
- %11 = cmpxchg i16* %10, i16 %7, i16 %9 monotonic
+ %11 = cmpxchg i16* %10, i16 %7, i16 %9 monotonic monotonic
store i16 %11, i16* @ss, align 2
%12 = load i8* @uc, align 1
%13 = zext i8 %12 to i16
%14 = load i8* @sc, align 1
%15 = sext i8 %14 to i16
%16 = bitcast i8* bitcast (i16* @us to i8*) to i16*
- %17 = cmpxchg i16* %16, i16 %13, i16 %15 monotonic
+ %17 = cmpxchg i16* %16, i16 %13, i16 %15 monotonic monotonic
store i16 %17, i16* @us, align 2
%18 = load i8* @uc, align 1
%19 = zext i8 %18 to i32
%20 = load i8* @sc, align 1
%21 = sext i8 %20 to i32
%22 = bitcast i8* bitcast (i32* @si to i8*) to i32*
- %23 = cmpxchg i32* %22, i32 %19, i32 %21 monotonic
+ %23 = cmpxchg i32* %22, i32 %19, i32 %21 monotonic monotonic
store i32 %23, i32* @si, align 4
%24 = load i8* @uc, align 1
%25 = zext i8 %24 to i32
%26 = load i8* @sc, align 1
%27 = sext i8 %26 to i32
%28 = bitcast i8* bitcast (i32* @ui to i8*) to i32*
- %29 = cmpxchg i32* %28, i32 %25, i32 %27 monotonic
+ %29 = cmpxchg i32* %28, i32 %25, i32 %27 monotonic monotonic
store i32 %29, i32* @ui, align 4
%30 = load i8* @uc, align 1
%31 = zext i8 %30 to i32
%32 = load i8* @sc, align 1
%33 = sext i8 %32 to i32
%34 = bitcast i8* bitcast (i32* @sl to i8*) to i32*
- %35 = cmpxchg i32* %34, i32 %31, i32 %33 monotonic
+ %35 = cmpxchg i32* %34, i32 %31, i32 %33 monotonic monotonic
store i32 %35, i32* @sl, align 4
%36 = load i8* @uc, align 1
%37 = zext i8 %36 to i32
%38 = load i8* @sc, align 1
%39 = sext i8 %38 to i32
%40 = bitcast i8* bitcast (i32* @ul to i8*) to i32*
- %41 = cmpxchg i32* %40, i32 %37, i32 %39 monotonic
+ %41 = cmpxchg i32* %40, i32 %37, i32 %39 monotonic monotonic
store i32 %41, i32* @ul, align 4
%42 = load i8* @uc, align 1
%43 = load i8* @sc, align 1
- %44 = cmpxchg i8* @sc, i8 %42, i8 %43 monotonic
+ %44 = cmpxchg i8* @sc, i8 %42, i8 %43 monotonic monotonic
%45 = icmp eq i8 %44, %42
%46 = zext i1 %45 to i32
store i32 %46, i32* @ui, align 4
%47 = load i8* @uc, align 1
%48 = load i8* @sc, align 1
- %49 = cmpxchg i8* @uc, i8 %47, i8 %48 monotonic
+ %49 = cmpxchg i8* @uc, i8 %47, i8 %48 monotonic monotonic
%50 = icmp eq i8 %49, %47
%51 = zext i1 %50 to i32
store i32 %51, i32* @ui, align 4
@@ -594,7 +594,7 @@ entry:
%54 = load i8* @sc, align 1
%55 = sext i8 %54 to i16
%56 = bitcast i8* bitcast (i16* @ss to i8*) to i16*
- %57 = cmpxchg i16* %56, i16 %53, i16 %55 monotonic
+ %57 = cmpxchg i16* %56, i16 %53, i16 %55 monotonic monotonic
%58 = icmp eq i16 %57, %53
%59 = zext i1 %58 to i32
store i32 %59, i32* @ui, align 4
@@ -603,7 +603,7 @@ entry:
%62 = load i8* @sc, align 1
%63 = sext i8 %62 to i16
%64 = bitcast i8* bitcast (i16* @us to i8*) to i16*
- %65 = cmpxchg i16* %64, i16 %61, i16 %63 monotonic
+ %65 = cmpxchg i16* %64, i16 %61, i16 %63 monotonic monotonic
%66 = icmp eq i16 %65, %61
%67 = zext i1 %66 to i32
store i32 %67, i32* @ui, align 4
@@ -612,7 +612,7 @@ entry:
%70 = load i8* @sc, align 1
%71 = sext i8 %70 to i32
%72 = bitcast i8* bitcast (i32* @si to i8*) to i32*
- %73 = cmpxchg i32* %72, i32 %69, i32 %71 monotonic
+ %73 = cmpxchg i32* %72, i32 %69, i32 %71 monotonic monotonic
%74 = icmp eq i32 %73, %69
%75 = zext i1 %74 to i32
store i32 %75, i32* @ui, align 4
@@ -621,7 +621,7 @@ entry:
%78 = load i8* @sc, align 1
%79 = sext i8 %78 to i32
%80 = bitcast i8* bitcast (i32* @ui to i8*) to i32*
- %81 = cmpxchg i32* %80, i32 %77, i32 %79 monotonic
+ %81 = cmpxchg i32* %80, i32 %77, i32 %79 monotonic monotonic
%82 = icmp eq i32 %81, %77
%83 = zext i1 %82 to i32
store i32 %83, i32* @ui, align 4
@@ -630,7 +630,7 @@ entry:
%86 = load i8* @sc, align 1
%87 = sext i8 %86 to i32
%88 = bitcast i8* bitcast (i32* @sl to i8*) to i32*
- %89 = cmpxchg i32* %88, i32 %85, i32 %87 monotonic
+ %89 = cmpxchg i32* %88, i32 %85, i32 %87 monotonic monotonic
%90 = icmp eq i32 %89, %85
%91 = zext i1 %90 to i32
store i32 %91, i32* @ui, align 4
@@ -639,7 +639,7 @@ entry:
%94 = load i8* @sc, align 1
%95 = sext i8 %94 to i32
%96 = bitcast i8* bitcast (i32* @ul to i8*) to i32*
- %97 = cmpxchg i32* %96, i32 %93, i32 %95 monotonic
+ %97 = cmpxchg i32* %96, i32 %93, i32 %95 monotonic monotonic
%98 = icmp eq i32 %97, %93
%99 = zext i1 %98 to i32
store i32 %99, i32* @ui, align 4
diff --git a/test/CodeGen/PowerPC/Atomics-64.ll b/test/CodeGen/PowerPC/Atomics-64.ll
index d35b848..122b54e 100644
--- a/test/CodeGen/PowerPC/Atomics-64.ll
+++ b/test/CodeGen/PowerPC/Atomics-64.ll
@@ -536,64 +536,64 @@ define void @test_compare_and_swap() nounwind {
entry:
%0 = load i8* @uc, align 1
%1 = load i8* @sc, align 1
- %2 = cmpxchg i8* @sc, i8 %0, i8 %1 monotonic
+ %2 = cmpxchg i8* @sc, i8 %0, i8 %1 monotonic monotonic
store i8 %2, i8* @sc, align 1
%3 = load i8* @uc, align 1
%4 = load i8* @sc, align 1
- %5 = cmpxchg i8* @uc, i8 %3, i8 %4 monotonic
+ %5 = cmpxchg i8* @uc, i8 %3, i8 %4 monotonic monotonic
store i8 %5, i8* @uc, align 1
%6 = load i8* @uc, align 1
%7 = zext i8 %6 to i16
%8 = load i8* @sc, align 1
%9 = sext i8 %8 to i16
%10 = bitcast i8* bitcast (i16* @ss to i8*) to i16*
- %11 = cmpxchg i16* %10, i16 %7, i16 %9 monotonic
+ %11 = cmpxchg i16* %10, i16 %7, i16 %9 monotonic monotonic
store i16 %11, i16* @ss, align 2
%12 = load i8* @uc, align 1
%13 = zext i8 %12 to i16
%14 = load i8* @sc, align 1
%15 = sext i8 %14 to i16
%16 = bitcast i8* bitcast (i16* @us to i8*) to i16*
- %17 = cmpxchg i16* %16, i16 %13, i16 %15 monotonic
+ %17 = cmpxchg i16* %16, i16 %13, i16 %15 monotonic monotonic
store i16 %17, i16* @us, align 2
%18 = load i8* @uc, align 1
%19 = zext i8 %18 to i32
%20 = load i8* @sc, align 1
%21 = sext i8 %20 to i32
%22 = bitcast i8* bitcast (i32* @si to i8*) to i32*
- %23 = cmpxchg i32* %22, i32 %19, i32 %21 monotonic
+ %23 = cmpxchg i32* %22, i32 %19, i32 %21 monotonic monotonic
store i32 %23, i32* @si, align 4
%24 = load i8* @uc, align 1
%25 = zext i8 %24 to i32
%26 = load i8* @sc, align 1
%27 = sext i8 %26 to i32
%28 = bitcast i8* bitcast (i32* @ui to i8*) to i32*
- %29 = cmpxchg i32* %28, i32 %25, i32 %27 monotonic
+ %29 = cmpxchg i32* %28, i32 %25, i32 %27 monotonic monotonic
store i32 %29, i32* @ui, align 4
%30 = load i8* @uc, align 1
%31 = zext i8 %30 to i64
%32 = load i8* @sc, align 1
%33 = sext i8 %32 to i64
%34 = bitcast i8* bitcast (i64* @sl to i8*) to i64*
- %35 = cmpxchg i64* %34, i64 %31, i64 %33 monotonic
+ %35 = cmpxchg i64* %34, i64 %31, i64 %33 monotonic monotonic
store i64 %35, i64* @sl, align 8
%36 = load i8* @uc, align 1
%37 = zext i8 %36 to i64
%38 = load i8* @sc, align 1
%39 = sext i8 %38 to i64
%40 = bitcast i8* bitcast (i64* @ul to i8*) to i64*
- %41 = cmpxchg i64* %40, i64 %37, i64 %39 monotonic
+ %41 = cmpxchg i64* %40, i64 %37, i64 %39 monotonic monotonic
store i64 %41, i64* @ul, align 8
%42 = load i8* @uc, align 1
%43 = load i8* @sc, align 1
- %44 = cmpxchg i8* @sc, i8 %42, i8 %43 monotonic
+ %44 = cmpxchg i8* @sc, i8 %42, i8 %43 monotonic monotonic
%45 = icmp eq i8 %44, %42
%46 = zext i1 %45 to i8
%47 = zext i8 %46 to i32
store i32 %47, i32* @ui, align 4
%48 = load i8* @uc, align 1
%49 = load i8* @sc, align 1
- %50 = cmpxchg i8* @uc, i8 %48, i8 %49 monotonic
+ %50 = cmpxchg i8* @uc, i8 %48, i8 %49 monotonic monotonic
%51 = icmp eq i8 %50, %48
%52 = zext i1 %51 to i8
%53 = zext i8 %52 to i32
@@ -603,7 +603,7 @@ entry:
%56 = load i8* @sc, align 1
%57 = sext i8 %56 to i16
%58 = bitcast i8* bitcast (i16* @ss to i8*) to i16*
- %59 = cmpxchg i16* %58, i16 %55, i16 %57 monotonic
+ %59 = cmpxchg i16* %58, i16 %55, i16 %57 monotonic monotonic
%60 = icmp eq i16 %59, %55
%61 = zext i1 %60 to i8
%62 = zext i8 %61 to i32
@@ -613,7 +613,7 @@ entry:
%65 = load i8* @sc, align 1
%66 = sext i8 %65 to i16
%67 = bitcast i8* bitcast (i16* @us to i8*) to i16*
- %68 = cmpxchg i16* %67, i16 %64, i16 %66 monotonic
+ %68 = cmpxchg i16* %67, i16 %64, i16 %66 monotonic monotonic
%69 = icmp eq i16 %68, %64
%70 = zext i1 %69 to i8
%71 = zext i8 %70 to i32
@@ -623,7 +623,7 @@ entry:
%74 = load i8* @sc, align 1
%75 = sext i8 %74 to i32
%76 = bitcast i8* bitcast (i32* @si to i8*) to i32*
- %77 = cmpxchg i32* %76, i32 %73, i32 %75 monotonic
+ %77 = cmpxchg i32* %76, i32 %73, i32 %75 monotonic monotonic
%78 = icmp eq i32 %77, %73
%79 = zext i1 %78 to i8
%80 = zext i8 %79 to i32
@@ -633,7 +633,7 @@ entry:
%83 = load i8* @sc, align 1
%84 = sext i8 %83 to i32
%85 = bitcast i8* bitcast (i32* @ui to i8*) to i32*
- %86 = cmpxchg i32* %85, i32 %82, i32 %84 monotonic
+ %86 = cmpxchg i32* %85, i32 %82, i32 %84 monotonic monotonic
%87 = icmp eq i32 %86, %82
%88 = zext i1 %87 to i8
%89 = zext i8 %88 to i32
@@ -643,7 +643,7 @@ entry:
%92 = load i8* @sc, align 1
%93 = sext i8 %92 to i64
%94 = bitcast i8* bitcast (i64* @sl to i8*) to i64*
- %95 = cmpxchg i64* %94, i64 %91, i64 %93 monotonic
+ %95 = cmpxchg i64* %94, i64 %91, i64 %93 monotonic monotonic
%96 = icmp eq i64 %95, %91
%97 = zext i1 %96 to i8
%98 = zext i8 %97 to i32
@@ -653,7 +653,7 @@ entry:
%101 = load i8* @sc, align 1
%102 = sext i8 %101 to i64
%103 = bitcast i8* bitcast (i64* @ul to i8*) to i64*
- %104 = cmpxchg i64* %103, i64 %100, i64 %102 monotonic
+ %104 = cmpxchg i64* %103, i64 %100, i64 %102 monotonic monotonic
%105 = icmp eq i64 %104, %100
%106 = zext i1 %105 to i8
%107 = zext i8 %106 to i32
diff --git a/test/CodeGen/PowerPC/atomic-1.ll b/test/CodeGen/PowerPC/atomic-1.ll
index 1737916..083df47 100644
--- a/test/CodeGen/PowerPC/atomic-1.ll
+++ b/test/CodeGen/PowerPC/atomic-1.ll
@@ -11,7 +11,7 @@ define i32 @exchange_and_add(i32* %mem, i32 %val) nounwind {
define i32 @exchange_and_cmp(i32* %mem) nounwind {
; CHECK-LABEL: exchange_and_cmp:
; CHECK: lwarx
- %tmp = cmpxchg i32* %mem, i32 0, i32 1 monotonic
+ %tmp = cmpxchg i32* %mem, i32 0, i32 1 monotonic monotonic
; CHECK: stwcx.
; CHECK: stwcx.
ret i32 %tmp
diff --git a/test/CodeGen/PowerPC/atomic-2.ll b/test/CodeGen/PowerPC/atomic-2.ll
index e56a779..261335e 100644
--- a/test/CodeGen/PowerPC/atomic-2.ll
+++ b/test/CodeGen/PowerPC/atomic-2.ll
@@ -11,7 +11,7 @@ define i64 @exchange_and_add(i64* %mem, i64 %val) nounwind {
define i64 @exchange_and_cmp(i64* %mem) nounwind {
; CHECK-LABEL: exchange_and_cmp:
; CHECK: ldarx
- %tmp = cmpxchg i64* %mem, i64 0, i64 1 monotonic
+ %tmp = cmpxchg i64* %mem, i64 0, i64 1 monotonic monotonic
; CHECK: stdcx.
; CHECK: stdcx.
ret i64 %tmp
diff --git a/test/CodeGen/SPARC/atomics.ll b/test/CodeGen/SPARC/atomics.ll
index b10336c..4e3e7ae 100644
--- a/test/CodeGen/SPARC/atomics.ll
+++ b/test/CodeGen/SPARC/atomics.ll
@@ -38,7 +38,7 @@ entry:
define i32 @test_cmpxchg_i32(i32 %a, i32* %ptr) {
entry:
- %b = cmpxchg i32* %ptr, i32 %a, i32 123 monotonic
+ %b = cmpxchg i32* %ptr, i32 %a, i32 123 monotonic monotonic
ret i32 %b
}
@@ -48,7 +48,7 @@ entry:
define i64 @test_cmpxchg_i64(i64 %a, i64* %ptr) {
entry:
- %b = cmpxchg i64* %ptr, i64 %a, i64 123 monotonic
+ %b = cmpxchg i64* %ptr, i64 %a, i64 123 monotonic monotonic
ret i64 %b
}
diff --git a/test/CodeGen/SystemZ/cmpxchg-01.ll b/test/CodeGen/SystemZ/cmpxchg-01.ll
index d5ea977..bb0b18a 100644
--- a/test/CodeGen/SystemZ/cmpxchg-01.ll
+++ b/test/CodeGen/SystemZ/cmpxchg-01.ll
@@ -32,7 +32,7 @@ define i8 @f1(i8 %dummy, i8 *%src, i8 %cmp, i8 %swap) {
; CHECK-SHIFT: lcr [[NEGSHIFT:%r[1-9]+]], [[SHIFT]]
; CHECK-SHIFT: rll
; CHECK-SHIFT: rll {{%r[0-9]+}}, %r5, -8([[NEGSHIFT]])
- %res = cmpxchg i8 *%src, i8 %cmp, i8 %swap seq_cst
+ %res = cmpxchg i8 *%src, i8 %cmp, i8 %swap seq_cst seq_cst
ret i8 %res
}
@@ -50,6 +50,6 @@ define i8 @f2(i8 *%src) {
; CHECK-SHIFT: risbg
; CHECK-SHIFT: risbg [[SWAP]], {{%r[0-9]+}}, 32, 55, 0
; CHECK-SHIFT: br %r14
- %res = cmpxchg i8 *%src, i8 42, i8 88 seq_cst
+ %res = cmpxchg i8 *%src, i8 42, i8 88 seq_cst seq_cst
ret i8 %res
}
diff --git a/test/CodeGen/SystemZ/cmpxchg-02.ll b/test/CodeGen/SystemZ/cmpxchg-02.ll
index 08c79d7..8d46a8c 100644
--- a/test/CodeGen/SystemZ/cmpxchg-02.ll
+++ b/test/CodeGen/SystemZ/cmpxchg-02.ll
@@ -32,7 +32,7 @@ define i16 @f1(i16 %dummy, i16 *%src, i16 %cmp, i16 %swap) {
; CHECK-SHIFT: lcr [[NEGSHIFT:%r[1-9]+]], [[SHIFT]]
; CHECK-SHIFT: rll
; CHECK-SHIFT: rll {{%r[0-9]+}}, %r5, -16([[NEGSHIFT]])
- %res = cmpxchg i16 *%src, i16 %cmp, i16 %swap seq_cst
+ %res = cmpxchg i16 *%src, i16 %cmp, i16 %swap seq_cst seq_cst
ret i16 %res
}
@@ -50,6 +50,6 @@ define i16 @f2(i16 *%src) {
; CHECK-SHIFT: risbg
; CHECK-SHIFT: risbg [[SWAP]], {{%r[0-9]+}}, 32, 47, 0
; CHECK-SHIFT: br %r14
- %res = cmpxchg i16 *%src, i16 42, i16 88 seq_cst
+ %res = cmpxchg i16 *%src, i16 42, i16 88 seq_cst seq_cst
ret i16 %res
}
diff --git a/test/CodeGen/SystemZ/cmpxchg-03.ll b/test/CodeGen/SystemZ/cmpxchg-03.ll
index 3917979..f6a2ad0 100644
--- a/test/CodeGen/SystemZ/cmpxchg-03.ll
+++ b/test/CodeGen/SystemZ/cmpxchg-03.ll
@@ -7,7 +7,7 @@ define i32 @f1(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK-LABEL: f1:
; CHECK: cs %r2, %r3, 0(%r4)
; CHECK: br %r14
- %val = cmpxchg i32 *%src, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%src, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -17,7 +17,7 @@ define i32 @f2(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK: cs %r2, %r3, 4092(%r4)
; CHECK: br %r14
%ptr = getelementptr i32 *%src, i64 1023
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -27,7 +27,7 @@ define i32 @f3(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK: csy %r2, %r3, 4096(%r4)
; CHECK: br %r14
%ptr = getelementptr i32 *%src, i64 1024
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -37,7 +37,7 @@ define i32 @f4(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK: csy %r2, %r3, 524284(%r4)
; CHECK: br %r14
%ptr = getelementptr i32 *%src, i64 131071
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -49,7 +49,7 @@ define i32 @f5(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK: cs %r2, %r3, 0(%r4)
; CHECK: br %r14
%ptr = getelementptr i32 *%src, i64 131072
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -59,7 +59,7 @@ define i32 @f6(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK: csy %r2, %r3, -4(%r4)
; CHECK: br %r14
%ptr = getelementptr i32 *%src, i64 -1
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -69,7 +69,7 @@ define i32 @f7(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK: csy %r2, %r3, -524288(%r4)
; CHECK: br %r14
%ptr = getelementptr i32 *%src, i64 -131072
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -81,7 +81,7 @@ define i32 @f8(i32 %cmp, i32 %swap, i32 *%src) {
; CHECK: cs %r2, %r3, 0(%r4)
; CHECK: br %r14
%ptr = getelementptr i32 *%src, i64 -131073
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -93,7 +93,7 @@ define i32 @f9(i32 %cmp, i32 %swap, i64 %src, i64 %index) {
; CHECK: br %r14
%add1 = add i64 %src, %index
%ptr = inttoptr i64 %add1 to i32 *
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -106,7 +106,7 @@ define i32 @f10(i32 %cmp, i32 %swap, i64 %src, i64 %index) {
%add1 = add i64 %src, %index
%add2 = add i64 %add1, 4096
%ptr = inttoptr i64 %add2 to i32 *
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -116,7 +116,7 @@ define i32 @f11(i32 %dummy, i32 %swap, i32 *%ptr) {
; CHECK: lhi %r2, 1001
; CHECK: cs %r2, %r3, 0(%r4)
; CHECK: br %r14
- %val = cmpxchg i32 *%ptr, i32 1001, i32 %swap seq_cst
+ %val = cmpxchg i32 *%ptr, i32 1001, i32 %swap seq_cst seq_cst
ret i32 %val
}
@@ -126,6 +126,6 @@ define i32 @f12(i32 %cmp, i32 *%ptr) {
; CHECK: lhi [[SWAP:%r[0-9]+]], 1002
; CHECK: cs %r2, [[SWAP]], 0(%r3)
; CHECK: br %r14
- %val = cmpxchg i32 *%ptr, i32 %cmp, i32 1002 seq_cst
+ %val = cmpxchg i32 *%ptr, i32 %cmp, i32 1002 seq_cst seq_cst
ret i32 %val
}
diff --git a/test/CodeGen/SystemZ/cmpxchg-04.ll b/test/CodeGen/SystemZ/cmpxchg-04.ll
index f58868f..069bad6 100644
--- a/test/CodeGen/SystemZ/cmpxchg-04.ll
+++ b/test/CodeGen/SystemZ/cmpxchg-04.ll
@@ -7,7 +7,7 @@ define i64 @f1(i64 %cmp, i64 %swap, i64 *%src) {
; CHECK-LABEL: f1:
; CHECK: csg %r2, %r3, 0(%r4)
; CHECK: br %r14
- %val = cmpxchg i64 *%src, i64 %cmp, i64 %swap seq_cst
+ %val = cmpxchg i64 *%src, i64 %cmp, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -17,7 +17,7 @@ define i64 @f2(i64 %cmp, i64 %swap, i64 *%src) {
; CHECK: csg %r2, %r3, 524280(%r4)
; CHECK: br %r14
%ptr = getelementptr i64 *%src, i64 65535
- %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst
+ %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -29,7 +29,7 @@ define i64 @f3(i64 %cmp, i64 %swap, i64 *%src) {
; CHECK: csg %r2, %r3, 0(%r4)
; CHECK: br %r14
%ptr = getelementptr i64 *%src, i64 65536
- %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst
+ %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -39,7 +39,7 @@ define i64 @f4(i64 %cmp, i64 %swap, i64 *%src) {
; CHECK: csg %r2, %r3, -8(%r4)
; CHECK: br %r14
%ptr = getelementptr i64 *%src, i64 -1
- %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst
+ %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -49,7 +49,7 @@ define i64 @f5(i64 %cmp, i64 %swap, i64 *%src) {
; CHECK: csg %r2, %r3, -524288(%r4)
; CHECK: br %r14
%ptr = getelementptr i64 *%src, i64 -65536
- %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst
+ %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -61,7 +61,7 @@ define i64 @f6(i64 %cmp, i64 %swap, i64 *%src) {
; CHECK: csg %r2, %r3, 0(%r4)
; CHECK: br %r14
%ptr = getelementptr i64 *%src, i64 -65537
- %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst
+ %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -73,7 +73,7 @@ define i64 @f7(i64 %cmp, i64 %swap, i64 %src, i64 %index) {
; CHECK: br %r14
%add1 = add i64 %src, %index
%ptr = inttoptr i64 %add1 to i64 *
- %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst
+ %val = cmpxchg i64 *%ptr, i64 %cmp, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -83,7 +83,7 @@ define i64 @f8(i64 %dummy, i64 %swap, i64 *%ptr) {
; CHECK: lghi %r2, 1001
; CHECK: csg %r2, %r3, 0(%r4)
; CHECK: br %r14
- %val = cmpxchg i64 *%ptr, i64 1001, i64 %swap seq_cst
+ %val = cmpxchg i64 *%ptr, i64 1001, i64 %swap seq_cst seq_cst
ret i64 %val
}
@@ -93,6 +93,6 @@ define i64 @f9(i64 %cmp, i64 *%ptr) {
; CHECK: lghi [[SWAP:%r[0-9]+]], 1002
; CHECK: csg %r2, [[SWAP]], 0(%r3)
; CHECK: br %r14
- %val = cmpxchg i64 *%ptr, i64 %cmp, i64 1002 seq_cst
+ %val = cmpxchg i64 *%ptr, i64 %cmp, i64 1002 seq_cst seq_cst
ret i64 %val
}
diff --git a/test/CodeGen/X86/2010-10-08-cmpxchg8b.ll b/test/CodeGen/X86/2010-10-08-cmpxchg8b.ll
index 0e4118a..f69cedc 100644
--- a/test/CodeGen/X86/2010-10-08-cmpxchg8b.ll
+++ b/test/CodeGen/X86/2010-10-08-cmpxchg8b.ll
@@ -18,7 +18,7 @@ entry:
loop:
; CHECK: lock
; CHECK-NEXT: cmpxchg8b
- %r = cmpxchg i64* %ptr, i64 0, i64 1 monotonic
+ %r = cmpxchg i64* %ptr, i64 0, i64 1 monotonic monotonic
%stored1 = icmp eq i64 %r, 0
br i1 %stored1, label %loop, label %continue
continue:
diff --git a/test/CodeGen/X86/Atomics-64.ll b/test/CodeGen/X86/Atomics-64.ll
index 8b0a349..c274688 100644
--- a/test/CodeGen/X86/Atomics-64.ll
+++ b/test/CodeGen/X86/Atomics-64.ll
@@ -704,7 +704,7 @@ entry:
%3 = zext i8 %2 to i32
%4 = trunc i32 %3 to i8
%5 = trunc i32 %1 to i8
- %6 = cmpxchg i8* @sc, i8 %4, i8 %5 monotonic
+ %6 = cmpxchg i8* @sc, i8 %4, i8 %5 monotonic monotonic
store i8 %6, i8* @sc, align 1
%7 = load i8* @sc, align 1
%8 = zext i8 %7 to i32
@@ -712,7 +712,7 @@ entry:
%10 = zext i8 %9 to i32
%11 = trunc i32 %10 to i8
%12 = trunc i32 %8 to i8
- %13 = cmpxchg i8* @uc, i8 %11, i8 %12 monotonic
+ %13 = cmpxchg i8* @uc, i8 %11, i8 %12 monotonic monotonic
store i8 %13, i8* @uc, align 1
%14 = load i8* @sc, align 1
%15 = sext i8 %14 to i16
@@ -722,7 +722,7 @@ entry:
%19 = bitcast i8* bitcast (i16* @ss to i8*) to i16*
%20 = trunc i32 %18 to i16
%21 = trunc i32 %16 to i16
- %22 = cmpxchg i16* %19, i16 %20, i16 %21 monotonic
+ %22 = cmpxchg i16* %19, i16 %20, i16 %21 monotonic monotonic
store i16 %22, i16* @ss, align 2
%23 = load i8* @sc, align 1
%24 = sext i8 %23 to i16
@@ -732,49 +732,49 @@ entry:
%28 = bitcast i8* bitcast (i16* @us to i8*) to i16*
%29 = trunc i32 %27 to i16
%30 = trunc i32 %25 to i16
- %31 = cmpxchg i16* %28, i16 %29, i16 %30 monotonic
+ %31 = cmpxchg i16* %28, i16 %29, i16 %30 monotonic monotonic
store i16 %31, i16* @us, align 2
%32 = load i8* @sc, align 1
%33 = sext i8 %32 to i32
%34 = load i8* @uc, align 1
%35 = zext i8 %34 to i32
%36 = bitcast i8* bitcast (i32* @si to i8*) to i32*
- %37 = cmpxchg i32* %36, i32 %35, i32 %33 monotonic
+ %37 = cmpxchg i32* %36, i32 %35, i32 %33 monotonic monotonic
store i32 %37, i32* @si, align 4
%38 = load i8* @sc, align 1
%39 = sext i8 %38 to i32
%40 = load i8* @uc, align 1
%41 = zext i8 %40 to i32
%42 = bitcast i8* bitcast (i32* @ui to i8*) to i32*
- %43 = cmpxchg i32* %42, i32 %41, i32 %39 monotonic
+ %43 = cmpxchg i32* %42, i32 %41, i32 %39 monotonic monotonic
store i32 %43, i32* @ui, align 4
%44 = load i8* @sc, align 1
%45 = sext i8 %44 to i64
%46 = load i8* @uc, align 1
%47 = zext i8 %46 to i64
%48 = bitcast i8* bitcast (i64* @sl to i8*) to i64*
- %49 = cmpxchg i64* %48, i64 %47, i64 %45 monotonic
+ %49 = cmpxchg i64* %48, i64 %47, i64 %45 monotonic monotonic
store i64 %49, i64* @sl, align 8
%50 = load i8* @sc, align 1
%51 = sext i8 %50 to i64
%52 = load i8* @uc, align 1
%53 = zext i8 %52 to i64
%54 = bitcast i8* bitcast (i64* @ul to i8*) to i64*
- %55 = cmpxchg i64* %54, i64 %53, i64 %51 monotonic
+ %55 = cmpxchg i64* %54, i64 %53, i64 %51 monotonic monotonic
store i64 %55, i64* @ul, align 8
%56 = load i8* @sc, align 1
%57 = sext i8 %56 to i64
%58 = load i8* @uc, align 1
%59 = zext i8 %58 to i64
%60 = bitcast i8* bitcast (i64* @sll to i8*) to i64*
- %61 = cmpxchg i64* %60, i64 %59, i64 %57 monotonic
+ %61 = cmpxchg i64* %60, i64 %59, i64 %57 monotonic monotonic
store i64 %61, i64* @sll, align 8
%62 = load i8* @sc, align 1
%63 = sext i8 %62 to i64
%64 = load i8* @uc, align 1
%65 = zext i8 %64 to i64
%66 = bitcast i8* bitcast (i64* @ull to i8*) to i64*
- %67 = cmpxchg i64* %66, i64 %65, i64 %63 monotonic
+ %67 = cmpxchg i64* %66, i64 %65, i64 %63 monotonic monotonic
store i64 %67, i64* @ull, align 8
%68 = load i8* @sc, align 1
%69 = zext i8 %68 to i32
@@ -782,7 +782,7 @@ entry:
%71 = zext i8 %70 to i32
%72 = trunc i32 %71 to i8
%73 = trunc i32 %69 to i8
- %74 = cmpxchg i8* @sc, i8 %72, i8 %73 monotonic
+ %74 = cmpxchg i8* @sc, i8 %72, i8 %73 monotonic monotonic
%75 = icmp eq i8 %74, %72
%76 = zext i1 %75 to i8
%77 = zext i8 %76 to i32
@@ -793,7 +793,7 @@ entry:
%81 = zext i8 %80 to i32
%82 = trunc i32 %81 to i8
%83 = trunc i32 %79 to i8
- %84 = cmpxchg i8* @uc, i8 %82, i8 %83 monotonic
+ %84 = cmpxchg i8* @uc, i8 %82, i8 %83 monotonic monotonic
%85 = icmp eq i8 %84, %82
%86 = zext i1 %85 to i8
%87 = zext i8 %86 to i32
@@ -805,7 +805,7 @@ entry:
%92 = zext i8 %91 to i32
%93 = trunc i32 %92 to i8
%94 = trunc i32 %90 to i8
- %95 = cmpxchg i8* bitcast (i16* @ss to i8*), i8 %93, i8 %94 monotonic
+ %95 = cmpxchg i8* bitcast (i16* @ss to i8*), i8 %93, i8 %94 monotonic monotonic
%96 = icmp eq i8 %95, %93
%97 = zext i1 %96 to i8
%98 = zext i8 %97 to i32
@@ -817,7 +817,7 @@ entry:
%103 = zext i8 %102 to i32
%104 = trunc i32 %103 to i8
%105 = trunc i32 %101 to i8
- %106 = cmpxchg i8* bitcast (i16* @us to i8*), i8 %104, i8 %105 monotonic
+ %106 = cmpxchg i8* bitcast (i16* @us to i8*), i8 %104, i8 %105 monotonic monotonic
%107 = icmp eq i8 %106, %104
%108 = zext i1 %107 to i8
%109 = zext i8 %108 to i32
@@ -828,7 +828,7 @@ entry:
%113 = zext i8 %112 to i32
%114 = trunc i32 %113 to i8
%115 = trunc i32 %111 to i8
- %116 = cmpxchg i8* bitcast (i32* @si to i8*), i8 %114, i8 %115 monotonic
+ %116 = cmpxchg i8* bitcast (i32* @si to i8*), i8 %114, i8 %115 monotonic monotonic
%117 = icmp eq i8 %116, %114
%118 = zext i1 %117 to i8
%119 = zext i8 %118 to i32
@@ -839,7 +839,7 @@ entry:
%123 = zext i8 %122 to i32
%124 = trunc i32 %123 to i8
%125 = trunc i32 %121 to i8
- %126 = cmpxchg i8* bitcast (i32* @ui to i8*), i8 %124, i8 %125 monotonic
+ %126 = cmpxchg i8* bitcast (i32* @ui to i8*), i8 %124, i8 %125 monotonic monotonic
%127 = icmp eq i8 %126, %124
%128 = zext i1 %127 to i8
%129 = zext i8 %128 to i32
@@ -850,7 +850,7 @@ entry:
%133 = zext i8 %132 to i64
%134 = trunc i64 %133 to i8
%135 = trunc i64 %131 to i8
- %136 = cmpxchg i8* bitcast (i64* @sl to i8*), i8 %134, i8 %135 monotonic
+ %136 = cmpxchg i8* bitcast (i64* @sl to i8*), i8 %134, i8 %135 monotonic monotonic
%137 = icmp eq i8 %136, %134
%138 = zext i1 %137 to i8
%139 = zext i8 %138 to i32
@@ -861,7 +861,7 @@ entry:
%143 = zext i8 %142 to i64
%144 = trunc i64 %143 to i8
%145 = trunc i64 %141 to i8
- %146 = cmpxchg i8* bitcast (i64* @ul to i8*), i8 %144, i8 %145 monotonic
+ %146 = cmpxchg i8* bitcast (i64* @ul to i8*), i8 %144, i8 %145 monotonic monotonic
%147 = icmp eq i8 %146, %144
%148 = zext i1 %147 to i8
%149 = zext i8 %148 to i32
@@ -872,7 +872,7 @@ entry:
%153 = zext i8 %152 to i64
%154 = trunc i64 %153 to i8
%155 = trunc i64 %151 to i8
- %156 = cmpxchg i8* bitcast (i64* @sll to i8*), i8 %154, i8 %155 monotonic
+ %156 = cmpxchg i8* bitcast (i64* @sll to i8*), i8 %154, i8 %155 monotonic monotonic
%157 = icmp eq i8 %156, %154
%158 = zext i1 %157 to i8
%159 = zext i8 %158 to i32
@@ -883,7 +883,7 @@ entry:
%163 = zext i8 %162 to i64
%164 = trunc i64 %163 to i8
%165 = trunc i64 %161 to i8
- %166 = cmpxchg i8* bitcast (i64* @ull to i8*), i8 %164, i8 %165 monotonic
+ %166 = cmpxchg i8* bitcast (i64* @ull to i8*), i8 %164, i8 %165 monotonic monotonic
%167 = icmp eq i8 %166, %164
%168 = zext i1 %167 to i8
%169 = zext i8 %168 to i32
diff --git a/test/CodeGen/X86/atomic16.ll b/test/CodeGen/X86/atomic16.ll
index ec2887e..45d3ff4 100644
--- a/test/CodeGen/X86/atomic16.ll
+++ b/test/CodeGen/X86/atomic16.ll
@@ -217,7 +217,7 @@ define void @atomic_fetch_umin16(i16 %x) nounwind {
}
define void @atomic_fetch_cmpxchg16() nounwind {
- %t1 = cmpxchg i16* @sc16, i16 0, i16 1 acquire
+ %t1 = cmpxchg i16* @sc16, i16 0, i16 1 acquire acquire
; X64: lock
; X64: cmpxchgw
; X32: lock
diff --git a/test/CodeGen/X86/atomic32.ll b/test/CodeGen/X86/atomic32.ll
index 3cb9ca1..474c0e6 100644
--- a/test/CodeGen/X86/atomic32.ll
+++ b/test/CodeGen/X86/atomic32.ll
@@ -243,7 +243,7 @@ define void @atomic_fetch_umin32(i32 %x) nounwind {
}
define void @atomic_fetch_cmpxchg32() nounwind {
- %t1 = cmpxchg i32* @sc32, i32 0, i32 1 acquire
+ %t1 = cmpxchg i32* @sc32, i32 0, i32 1 acquire acquire
; X64: lock
; X64: cmpxchgl
; X32: lock
diff --git a/test/CodeGen/X86/atomic64.ll b/test/CodeGen/X86/atomic64.ll
index aa00045..4f55edc 100644
--- a/test/CodeGen/X86/atomic64.ll
+++ b/test/CodeGen/X86/atomic64.ll
@@ -183,7 +183,7 @@ define void @atomic_fetch_umin64(i64 %x) nounwind {
}
define void @atomic_fetch_cmpxchg64() nounwind {
- %t1 = cmpxchg i64* @sc64, i64 0, i64 1 acquire
+ %t1 = cmpxchg i64* @sc64, i64 0, i64 1 acquire acquire
; X64: lock
; X64: cmpxchgq
; X32: lock
diff --git a/test/CodeGen/X86/atomic6432.ll b/test/CodeGen/X86/atomic6432.ll
index 31e66c8..c0f7267 100644
--- a/test/CodeGen/X86/atomic6432.ll
+++ b/test/CodeGen/X86/atomic6432.ll
@@ -184,7 +184,7 @@ define void @atomic_fetch_umin64(i64 %x) nounwind {
}
define void @atomic_fetch_cmpxchg64() nounwind {
- %t1 = cmpxchg i64* @sc64, i64 0, i64 1 acquire
+ %t1 = cmpxchg i64* @sc64, i64 0, i64 1 acquire acquire
; X32: lock
; X32: cmpxchg8b
ret void
diff --git a/test/CodeGen/X86/atomic8.ll b/test/CodeGen/X86/atomic8.ll
index 3278ed1..203b26f 100644
--- a/test/CodeGen/X86/atomic8.ll
+++ b/test/CodeGen/X86/atomic8.ll
@@ -217,7 +217,7 @@ define void @atomic_fetch_umin8(i8 %x) nounwind {
}
define void @atomic_fetch_cmpxchg8() nounwind {
- %t1 = cmpxchg i8* @sc8, i8 0, i8 1 acquire
+ %t1 = cmpxchg i8* @sc8, i8 0, i8 1 acquire acquire
; X64: lock
; X64: cmpxchgb
; X32: lock
diff --git a/test/CodeGen/X86/atomic_op.ll b/test/CodeGen/X86/atomic_op.ll
index a378d6e..b3045ed 100644
--- a/test/CodeGen/X86/atomic_op.ll
+++ b/test/CodeGen/X86/atomic_op.ll
@@ -101,11 +101,11 @@ entry:
%neg1 = sub i32 0, 10 ; <i32> [#uses=1]
; CHECK: lock
; CHECK: cmpxchgl
- %16 = cmpxchg i32* %val2, i32 %neg1, i32 1 monotonic
+ %16 = cmpxchg i32* %val2, i32 %neg1, i32 1 monotonic monotonic
store i32 %16, i32* %old
; CHECK: lock
; CHECK: cmpxchgl
- %17 = cmpxchg i32* %val2, i32 1976, i32 1 monotonic
+ %17 = cmpxchg i32* %val2, i32 1976, i32 1 monotonic monotonic
store i32 %17, i32* %old
; CHECK: movl [[R17atomic:.*]], %eax
; CHECK: movl $1401, %[[R17mask:[a-z]*]]
@@ -133,6 +133,6 @@ entry:
; CHECK: lock
; CHECK: cmpxchgl %{{.*}}, %gs:(%{{.*}})
- %0 = cmpxchg i32 addrspace(256)* %P, i32 0, i32 1 monotonic
+ %0 = cmpxchg i32 addrspace(256)* %P, i32 0, i32 1 monotonic monotonic
ret void
}
diff --git a/test/CodeGen/X86/cmpxchg16b.ll b/test/CodeGen/X86/cmpxchg16b.ll
index edbd0bc..1d5bb85 100644
--- a/test/CodeGen/X86/cmpxchg16b.ll
+++ b/test/CodeGen/X86/cmpxchg16b.ll
@@ -6,7 +6,7 @@ entry:
; CHECK: movl $1, %ebx
; CHECK: lock
; CHECK-NEXT: cmpxchg16b
- %r = cmpxchg i128* %p, i128 0, i128 1 seq_cst
+ %r = cmpxchg i128* %p, i128 0, i128 1 seq_cst seq_cst
ret void
}
diff --git a/test/CodeGen/X86/coalescer-remat.ll b/test/CodeGen/X86/coalescer-remat.ll
index eb7b7a8..468b70b 100644
--- a/test/CodeGen/X86/coalescer-remat.ll
+++ b/test/CodeGen/X86/coalescer-remat.ll
@@ -5,7 +5,7 @@
define i32 @main() nounwind {
entry:
- %0 = cmpxchg i64* @val, i64 0, i64 1 monotonic
+ %0 = cmpxchg i64* @val, i64 0, i64 1 monotonic monotonic
%1 = tail call i32 (i8*, ...)* @printf(i8* getelementptr ([7 x i8]* @"\01LC", i32 0, i64 0), i64 %0) nounwind
ret i32 0
}
diff --git a/test/CodeGen/X86/nocx16.ll b/test/CodeGen/X86/nocx16.ll
index cceaac4..8b995da 100644
--- a/test/CodeGen/X86/nocx16.ll
+++ b/test/CodeGen/X86/nocx16.ll
@@ -2,7 +2,7 @@
define void @test(i128* %a) nounwind {
entry:
; CHECK: __sync_val_compare_and_swap_16
- %0 = cmpxchg i128* %a, i128 1, i128 1 seq_cst
+ %0 = cmpxchg i128* %a, i128 1, i128 1 seq_cst seq_cst
; CHECK: __sync_lock_test_and_set_16
%1 = atomicrmw xchg i128* %a, i128 1 seq_cst
; CHECK: __sync_fetch_and_add_16
diff --git a/test/Instrumentation/AddressSanitizer/test64.ll b/test/Instrumentation/AddressSanitizer/test64.ll
index 5b30fae..6390644 100644
--- a/test/Instrumentation/AddressSanitizer/test64.ll
+++ b/test/Instrumentation/AddressSanitizer/test64.ll
@@ -26,7 +26,7 @@ entry:
define void @example_cmpxchg(i64* %ptr, i64 %compare_to, i64 %new_value) nounwind uwtable sanitize_address {
entry:
- %0 = cmpxchg i64* %ptr, i64 %compare_to, i64 %new_value seq_cst
+ %0 = cmpxchg i64* %ptr, i64 %compare_to, i64 %new_value seq_cst seq_cst
ret void
}
diff --git a/test/Instrumentation/MemorySanitizer/atomics.ll b/test/Instrumentation/MemorySanitizer/atomics.ll
index ff02452..98697d7 100644
--- a/test/Instrumentation/MemorySanitizer/atomics.ll
+++ b/test/Instrumentation/MemorySanitizer/atomics.ll
@@ -37,7 +37,7 @@ entry:
define i32 @Cmpxchg(i32* %p, i32 %a, i32 %b) sanitize_memory {
entry:
- %0 = cmpxchg i32* %p, i32 %a, i32 %b seq_cst
+ %0 = cmpxchg i32* %p, i32 %a, i32 %b seq_cst seq_cst
ret i32 %0
}
@@ -46,16 +46,16 @@ entry:
; CHECK: icmp
; CHECK: br
; CHECK: @__msan_warning
-; CHECK: cmpxchg {{.*}} seq_cst
+; CHECK: cmpxchg {{.*}} seq_cst seq_cst
; CHECK: store i32 0, {{.*}} @__msan_retval_tls
; CHECK: ret i32
-; relaxed cmpxchg: bump up to "release"
+; relaxed cmpxchg: bump up to "release monotonic"
define i32 @CmpxchgMonotonic(i32* %p, i32 %a, i32 %b) sanitize_memory {
entry:
- %0 = cmpxchg i32* %p, i32 %a, i32 %b monotonic
+ %0 = cmpxchg i32* %p, i32 %a, i32 %b monotonic monotonic
ret i32 %0
}
@@ -64,7 +64,7 @@ entry:
; CHECK: icmp
; CHECK: br
; CHECK: @__msan_warning
-; CHECK: cmpxchg {{.*}} release
+; CHECK: cmpxchg {{.*}} release monotonic
; CHECK: store i32 0, {{.*}} @__msan_retval_tls
; CHECK: ret i32
diff --git a/test/Instrumentation/ThreadSanitizer/atomic.ll b/test/Instrumentation/ThreadSanitizer/atomic.ll
index 70b6cbb..e40268f 100644
--- a/test/Instrumentation/ThreadSanitizer/atomic.ll
+++ b/test/Instrumentation/ThreadSanitizer/atomic.ll
@@ -348,7 +348,7 @@ entry:
define void @atomic8_cas_monotonic(i8* %a) nounwind uwtable {
entry:
- cmpxchg i8* %a, i8 0, i8 1 monotonic
+ cmpxchg i8* %a, i8 0, i8 1 monotonic monotonic
ret void
}
; CHECK: atomic8_cas_monotonic
@@ -356,7 +356,7 @@ entry:
define void @atomic8_cas_acquire(i8* %a) nounwind uwtable {
entry:
- cmpxchg i8* %a, i8 0, i8 1 acquire
+ cmpxchg i8* %a, i8 0, i8 1 acquire acquire
ret void
}
; CHECK: atomic8_cas_acquire
@@ -364,7 +364,7 @@ entry:
define void @atomic8_cas_release(i8* %a) nounwind uwtable {
entry:
- cmpxchg i8* %a, i8 0, i8 1 release
+ cmpxchg i8* %a, i8 0, i8 1 release monotonic
ret void
}
; CHECK: atomic8_cas_release
@@ -372,7 +372,7 @@ entry:
define void @atomic8_cas_acq_rel(i8* %a) nounwind uwtable {
entry:
- cmpxchg i8* %a, i8 0, i8 1 acq_rel
+ cmpxchg i8* %a, i8 0, i8 1 acq_rel acquire
ret void
}
; CHECK: atomic8_cas_acq_rel
@@ -380,7 +380,7 @@ entry:
define void @atomic8_cas_seq_cst(i8* %a) nounwind uwtable {
entry:
- cmpxchg i8* %a, i8 0, i8 1 seq_cst
+ cmpxchg i8* %a, i8 0, i8 1 seq_cst seq_cst
ret void
}
; CHECK: atomic8_cas_seq_cst
@@ -732,7 +732,7 @@ entry:
define void @atomic16_cas_monotonic(i16* %a) nounwind uwtable {
entry:
- cmpxchg i16* %a, i16 0, i16 1 monotonic
+ cmpxchg i16* %a, i16 0, i16 1 monotonic monotonic
ret void
}
; CHECK: atomic16_cas_monotonic
@@ -740,7 +740,7 @@ entry:
define void @atomic16_cas_acquire(i16* %a) nounwind uwtable {
entry:
- cmpxchg i16* %a, i16 0, i16 1 acquire
+ cmpxchg i16* %a, i16 0, i16 1 acquire acquire
ret void
}
; CHECK: atomic16_cas_acquire
@@ -748,7 +748,7 @@ entry:
define void @atomic16_cas_release(i16* %a) nounwind uwtable {
entry:
- cmpxchg i16* %a, i16 0, i16 1 release
+ cmpxchg i16* %a, i16 0, i16 1 release monotonic
ret void
}
; CHECK: atomic16_cas_release
@@ -756,7 +756,7 @@ entry:
define void @atomic16_cas_acq_rel(i16* %a) nounwind uwtable {
entry:
- cmpxchg i16* %a, i16 0, i16 1 acq_rel
+ cmpxchg i16* %a, i16 0, i16 1 acq_rel acquire
ret void
}
; CHECK: atomic16_cas_acq_rel
@@ -764,7 +764,7 @@ entry:
define void @atomic16_cas_seq_cst(i16* %a) nounwind uwtable {
entry:
- cmpxchg i16* %a, i16 0, i16 1 seq_cst
+ cmpxchg i16* %a, i16 0, i16 1 seq_cst seq_cst
ret void
}
; CHECK: atomic16_cas_seq_cst
@@ -1116,7 +1116,7 @@ entry:
define void @atomic32_cas_monotonic(i32* %a) nounwind uwtable {
entry:
- cmpxchg i32* %a, i32 0, i32 1 monotonic
+ cmpxchg i32* %a, i32 0, i32 1 monotonic monotonic
ret void
}
; CHECK: atomic32_cas_monotonic
@@ -1124,7 +1124,7 @@ entry:
define void @atomic32_cas_acquire(i32* %a) nounwind uwtable {
entry:
- cmpxchg i32* %a, i32 0, i32 1 acquire
+ cmpxchg i32* %a, i32 0, i32 1 acquire acquire
ret void
}
; CHECK: atomic32_cas_acquire
@@ -1132,7 +1132,7 @@ entry:
define void @atomic32_cas_release(i32* %a) nounwind uwtable {
entry:
- cmpxchg i32* %a, i32 0, i32 1 release
+ cmpxchg i32* %a, i32 0, i32 1 release monotonic
ret void
}
; CHECK: atomic32_cas_release
@@ -1140,7 +1140,7 @@ entry:
define void @atomic32_cas_acq_rel(i32* %a) nounwind uwtable {
entry:
- cmpxchg i32* %a, i32 0, i32 1 acq_rel
+ cmpxchg i32* %a, i32 0, i32 1 acq_rel acquire
ret void
}
; CHECK: atomic32_cas_acq_rel
@@ -1148,7 +1148,7 @@ entry:
define void @atomic32_cas_seq_cst(i32* %a) nounwind uwtable {
entry:
- cmpxchg i32* %a, i32 0, i32 1 seq_cst
+ cmpxchg i32* %a, i32 0, i32 1 seq_cst seq_cst
ret void
}
; CHECK: atomic32_cas_seq_cst
@@ -1500,7 +1500,7 @@ entry:
define void @atomic64_cas_monotonic(i64* %a) nounwind uwtable {
entry:
- cmpxchg i64* %a, i64 0, i64 1 monotonic
+ cmpxchg i64* %a, i64 0, i64 1 monotonic monotonic
ret void
}
; CHECK: atomic64_cas_monotonic
@@ -1508,7 +1508,7 @@ entry:
define void @atomic64_cas_acquire(i64* %a) nounwind uwtable {
entry:
- cmpxchg i64* %a, i64 0, i64 1 acquire
+ cmpxchg i64* %a, i64 0, i64 1 acquire acquire
ret void
}
; CHECK: atomic64_cas_acquire
@@ -1516,7 +1516,7 @@ entry:
define void @atomic64_cas_release(i64* %a) nounwind uwtable {
entry:
- cmpxchg i64* %a, i64 0, i64 1 release
+ cmpxchg i64* %a, i64 0, i64 1 release monotonic
ret void
}
; CHECK: atomic64_cas_release
@@ -1524,7 +1524,7 @@ entry:
define void @atomic64_cas_acq_rel(i64* %a) nounwind uwtable {
entry:
- cmpxchg i64* %a, i64 0, i64 1 acq_rel
+ cmpxchg i64* %a, i64 0, i64 1 acq_rel acquire
ret void
}
; CHECK: atomic64_cas_acq_rel
@@ -1532,7 +1532,7 @@ entry:
define void @atomic64_cas_seq_cst(i64* %a) nounwind uwtable {
entry:
- cmpxchg i64* %a, i64 0, i64 1 seq_cst
+ cmpxchg i64* %a, i64 0, i64 1 seq_cst seq_cst
ret void
}
; CHECK: atomic64_cas_seq_cst
@@ -1884,7 +1884,7 @@ entry:
define void @atomic128_cas_monotonic(i128* %a) nounwind uwtable {
entry:
- cmpxchg i128* %a, i128 0, i128 1 monotonic
+ cmpxchg i128* %a, i128 0, i128 1 monotonic monotonic
ret void
}
; CHECK: atomic128_cas_monotonic
@@ -1892,7 +1892,7 @@ entry:
define void @atomic128_cas_acquire(i128* %a) nounwind uwtable {
entry:
- cmpxchg i128* %a, i128 0, i128 1 acquire
+ cmpxchg i128* %a, i128 0, i128 1 acquire acquire
ret void
}
; CHECK: atomic128_cas_acquire
@@ -1900,7 +1900,7 @@ entry:
define void @atomic128_cas_release(i128* %a) nounwind uwtable {
entry:
- cmpxchg i128* %a, i128 0, i128 1 release
+ cmpxchg i128* %a, i128 0, i128 1 release monotonic
ret void
}
; CHECK: atomic128_cas_release
@@ -1908,7 +1908,7 @@ entry:
define void @atomic128_cas_acq_rel(i128* %a) nounwind uwtable {
entry:
- cmpxchg i128* %a, i128 0, i128 1 acq_rel
+ cmpxchg i128* %a, i128 0, i128 1 acq_rel acquire
ret void
}
; CHECK: atomic128_cas_acq_rel
@@ -1916,7 +1916,7 @@ entry:
define void @atomic128_cas_seq_cst(i128* %a) nounwind uwtable {
entry:
- cmpxchg i128* %a, i128 0, i128 1 seq_cst
+ cmpxchg i128* %a, i128 0, i128 1 seq_cst seq_cst
ret void
}
; CHECK: atomic128_cas_seq_cst
diff --git a/test/Transforms/LowerAtomic/atomic-swap.ll b/test/Transforms/LowerAtomic/atomic-swap.ll
index 4331677..c319834 100644
--- a/test/Transforms/LowerAtomic/atomic-swap.ll
+++ b/test/Transforms/LowerAtomic/atomic-swap.ll
@@ -3,7 +3,7 @@
define i8 @cmpswap() {
; CHECK-LABEL: @cmpswap(
%i = alloca i8
- %j = cmpxchg i8* %i, i8 0, i8 42 monotonic
+ %j = cmpxchg i8* %i, i8 0, i8 42 monotonic monotonic
; CHECK: [[INST:%[a-z0-9]+]] = load
; CHECK-NEXT: icmp
; CHECK-NEXT: select
diff --git a/test/Transforms/SimplifyCFG/trapping-load-unreachable.ll b/test/Transforms/SimplifyCFG/trapping-load-unreachable.ll
index e9d93e8..5ae62af 100644
--- a/test/Transforms/SimplifyCFG/trapping-load-unreachable.ll
+++ b/test/Transforms/SimplifyCFG/trapping-load-unreachable.ll
@@ -65,7 +65,7 @@ define void @test5(i1 %C, i32* %P) {
entry:
br i1 %C, label %T, label %F
T:
- cmpxchg volatile i32* %P, i32 0, i32 1 seq_cst
+ cmpxchg volatile i32* %P, i32 0, i32 1 seq_cst seq_cst
unreachable
F:
ret void
More information about the llvm-dev
mailing list