================
Comment at: lib/CodeGen/CGBuiltin.cpp:112
@@ +111,3 @@
+ Result = EmitFromInt(CGF, Result, T, ValueType);
+ return Result;
+}
----------------
Don't really need the Result temporary here?
================
Comment at: lib/CodeGen/CGBuiltin.cpp:118
@@ -87,26 +117,3 @@
const CallExpr *E) {
- QualType T = E->getType();
- assert(E->getArg(0)->getType()->isPointerType());
- assert(CGF.getContext().hasSameUnqualifiedType(T,
- E->getArg(0)->getType()->getPointeeType()));
- assert(CGF.getContext().hasSameUnqualifiedType(T, E->getArg(1)->getType()));
-
- llvm::Value *DestPtr = CGF.EmitScalarExpr(E->getArg(0));
- unsigned AddrSpace = DestPtr->getType()->getPointerAddressSpace();
-
- llvm::IntegerType *IntType =
- llvm::IntegerType::get(CGF.getLLVMContext(),
- CGF.getContext().getTypeSize(T));
- llvm::Type *IntPtrType = IntType->getPointerTo(AddrSpace);
-
- llvm::Value *Args[2];
- Args[0] = CGF.Builder.CreateBitCast(DestPtr, IntPtrType);
- Args[1] = CGF.EmitScalarExpr(E->getArg(1));
- llvm::Type *ValueType = Args[1]->getType();
- Args[1] = EmitToInt(CGF, Args[1], T, IntType);
-
- llvm::Value *Result =
- CGF.Builder.CreateAtomicRMW(Kind, Args[0], Args[1],
- llvm::SequentiallyConsistent);
- Result = EmitFromInt(CGF, Result, T, ValueType);
+ llvm::Value *Result = MakeBinaryAtomicValue(CGF, Kind, E);
return RValue::get(Result);
----------------
You can just "return RValue::get(MakeBinary...)" -- similarly to how it's done
below, for consistency
================
Comment at: lib/CodeGen/CGBuiltin.cpp:165
@@ +164,3 @@
+static Value *MakeAtomicCmpXchgValue(CodeGenFunction &CGF, const CallExpr *E,
+ bool ReturnBool) {
+ QualType T = ReturnBool ? E->getArg(1)->getType() : E->getType();
----------------
Please document all function parameters in the comment (especially ReturnBool)
================
Comment at: lib/CodeGen/CGBuiltin.cpp:174
@@ +173,3 @@
+
+ Value *Args[3];
+ Args[0] = CGF.Builder.CreateBitCast(DestPtr, IntPtrType);
----------------
Would universal initialization be simpler here?
Value *Args[3] = {CGF.Builder.CreateBi......, CGF.EmitScalar...}
================
Comment at: lib/CodeGen/CGBuiltin.cpp:6865
@@ +6864,3 @@
+
+Value *CodeGenFunction::EmitNVPTXBuiltinExpr(unsigned BuiltinID,
+ const CallExpr *E) {
----------------
I'm wondering if there's some unwritten rule saying that all target builtins
should be crammed into a single file... It's closing in on 7KLOC now and no end
in sight. Would it be very bad for NVPTX to have its own CGBuiltinNVPTX or
something like that? Clang splits classes to multiple files (Sema,
CodeGenFunction, etc) already...
Eric?
http://reviews.llvm.org/D10666
EMAIL PREFERENCES
http://reviews.llvm.org/settings/panel/emailpreferences/
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
