Emit memmove, not memcpy, for structure copies; this is unfortunately
required for correctness in cases of copying a struct to itself or to an overlapping struct (itself for cases like *a = *a, and overlapping is possible with unions). Hopefully, this won't end up being a perf issue; LLVM *should* be able to optimize memmove to memcpy in a lot of cases, and for small copies the generated code *should* be mostly comparable. (In reality, LLVM is currently horrible at optimizing memmove, but that's a bug, not a fundamental issue.) gcc currently generates wrong code; that's http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32667. llvm-svn: 51566
This commit is contained in:
@@ -35,7 +35,8 @@ CodeGenModule::CodeGenModule(ASTContext &C, const LangOptions &LO,
|
||||
llvm::Module &M, const llvm::TargetData &TD,
|
||||
Diagnostic &diags, bool GenerateDebugInfo)
|
||||
: Context(C), Features(LO), TheModule(M), TheTargetData(TD), Diags(diags),
|
||||
Types(C, M, TD), MemCpyFn(0), MemSetFn(0), CFConstantStringClassRef(0) {
|
||||
Types(C, M, TD), MemCpyFn(0), MemMoveFn(0), MemSetFn(0),
|
||||
CFConstantStringClassRef(0) {
|
||||
//TODO: Make this selectable at runtime
|
||||
Runtime = CreateObjCRuntime(M,
|
||||
getTypes().ConvertType(getContext().IntTy),
|
||||
@@ -589,6 +590,17 @@ llvm::Function *CodeGenModule::getMemCpyFn() {
|
||||
return MemCpyFn = getIntrinsic(IID);
|
||||
}
|
||||
|
||||
llvm::Function *CodeGenModule::getMemMoveFn() {
|
||||
if (MemMoveFn) return MemMoveFn;
|
||||
llvm::Intrinsic::ID IID;
|
||||
switch (Context.Target.getPointerWidth(0)) {
|
||||
default: assert(0 && "Unknown ptr width");
|
||||
case 32: IID = llvm::Intrinsic::memmove_i32; break;
|
||||
case 64: IID = llvm::Intrinsic::memmove_i64; break;
|
||||
}
|
||||
return MemMoveFn = getIntrinsic(IID);
|
||||
}
|
||||
|
||||
llvm::Function *CodeGenModule::getMemSetFn() {
|
||||
if (MemSetFn) return MemSetFn;
|
||||
llvm::Intrinsic::ID IID;
|
||||
|
||||
Reference in New Issue
Block a user