[Power] Use AtomicExpandPass for fence insertion, and use lwsync where appropriate
Summary:
This patch makes use of AtomicExpandPass in Power for inserting fences around
atomic as part of an effort to remove fence insertion from SelectionDAGBuilder.
As a big bonus, it lets us use sync 1 (lightweight sync, often used by the mnemonic
lwsync) instead of sync 0 (heavyweight sync) in many cases.
I also added a test, as there was no test for the barriers emitted by the Power
backend for atomic loads and stores.
Test Plan: new test + make check-all
Reviewers: jfb
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D5180
llvm-svn: 218331
diff --git a/llvm/lib/Target/PowerPC/PPCISelLowering.cpp b/llvm/lib/Target/PowerPC/PPCISelLowering.cpp
index d96cdab..e216a72c 100644
--- a/llvm/lib/Target/PowerPC/PPCISelLowering.cpp
+++ b/llvm/lib/Target/PowerPC/PPCISelLowering.cpp
@@ -6538,6 +6538,38 @@
// Other Lowering Code
//===----------------------------------------------------------------------===//
+static Instruction* callIntrinsic(IRBuilder<> &Builder, Intrinsic::ID Id) {
+ Module *M = Builder.GetInsertBlock()->getParent()->getParent();
+ Function *Func = Intrinsic::getDeclaration(M, Id);
+ return Builder.CreateCall(Func);
+}
+
+// The mappings for emitLeading/TrailingFence is taken from
+// http://www.cl.cam.ac.uk/~pes20/cpp/cpp0xmappings.html
+Instruction* PPCTargetLowering::emitLeadingFence(IRBuilder<> &Builder,
+ AtomicOrdering Ord, bool IsStore,
+ bool IsLoad) const {
+ if (Ord == SequentiallyConsistent)
+ return callIntrinsic(Builder, Intrinsic::ppc_sync);
+ else if (isAtLeastRelease(Ord))
+ return callIntrinsic(Builder, Intrinsic::ppc_lwsync);
+ else
+ return nullptr;
+}
+
+Instruction* PPCTargetLowering::emitTrailingFence(IRBuilder<> &Builder,
+ AtomicOrdering Ord, bool IsStore,
+ bool IsLoad) const {
+ if (IsLoad && isAtLeastAcquire(Ord))
+ return callIntrinsic(Builder, Intrinsic::ppc_lwsync);
+ // FIXME: this is too conservative, a dependent branch + isync is enough.
+ // See http://www.cl.cam.ac.uk/~pes20/cpp/cpp0xmappings.html and
+ // http://www.rdrop.com/users/paulmck/scalability/paper/N2745r.2011.03.04a.html
+ // and http://www.cl.cam.ac.uk/~pes20/cppppc/ for justification.
+ else
+ return nullptr;
+}
+
MachineBasicBlock *
PPCTargetLowering::EmitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB,
bool is64bit, unsigned BinOpcode) const {
diff --git a/llvm/lib/Target/PowerPC/PPCISelLowering.h b/llvm/lib/Target/PowerPC/PPCISelLowering.h
index 5628bc7..7444d41 100644
--- a/llvm/lib/Target/PowerPC/PPCISelLowering.h
+++ b/llvm/lib/Target/PowerPC/PPCISelLowering.h
@@ -413,6 +413,11 @@
const SelectionDAG &DAG,
unsigned Depth = 0) const override;
+ Instruction* emitLeadingFence(IRBuilder<> &Builder, AtomicOrdering Ord,
+ bool IsStore, bool IsLoad) const override;
+ Instruction* emitTrailingFence(IRBuilder<> &Builder, AtomicOrdering Ord,
+ bool IsStore, bool IsLoad) const override;
+
MachineBasicBlock *
EmitInstrWithCustomInserter(MachineInstr *MI,
MachineBasicBlock *MBB) const override;
diff --git a/llvm/lib/Target/PowerPC/PPCInstrInfo.td b/llvm/lib/Target/PowerPC/PPCInstrInfo.td
index 38020ef..5643ed7 100644
--- a/llvm/lib/Target/PowerPC/PPCInstrInfo.td
+++ b/llvm/lib/Target/PowerPC/PPCInstrInfo.td
@@ -1704,8 +1704,13 @@
}
}
-def : Pat<(int_ppc_sync), (SYNC 0)>, Requires<[IsNotBookE]>;
-def : Pat<(int_ppc_sync), (MSYNC)>, Requires<[IsBookE]>;
+// FIXME: We should use a specific flag to check for the presence of the sync
+// instruction instead of relying on IsBookE.
+// For example, the PPC A2 supports SYNC despite being a BookE processor.
+def : Pat<(int_ppc_sync), (SYNC 0)>, Requires<[IsNotBookE]>;
+def : Pat<(int_ppc_lwsync), (SYNC 1)>, Requires<[IsNotBookE]>;
+def : Pat<(int_ppc_sync), (MSYNC)>, Requires<[IsBookE]>;
+def : Pat<(int_ppc_lwsync), (MSYNC)>, Requires<[IsBookE]>;
//===----------------------------------------------------------------------===//
// PPC32 Arithmetic Instructions.
diff --git a/llvm/lib/Target/PowerPC/PPCTargetMachine.cpp b/llvm/lib/Target/PowerPC/PPCTargetMachine.cpp
index 878d136..ed90e27 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetMachine.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetMachine.cpp
@@ -86,6 +86,7 @@
return *getPPCTargetMachine().getSubtargetImpl();
}
+ void addIRPasses() override;
bool addPreISel() override;
bool addILPOpts() override;
bool addInstSelector() override;
@@ -99,6 +100,11 @@
return new PPCPassConfig(this, PM);
}
+void PPCPassConfig::addIRPasses() {
+ addPass(createAtomicExpandPass(&getPPCTargetMachine()));
+ TargetPassConfig::addIRPasses();
+}
+
bool PPCPassConfig::addPreISel() {
if (!DisableCTRLoops && getOptLevel() != CodeGenOpt::None)
addPass(createPPCCTRLoops(getPPCTargetMachine()));