[asan] Do not return void expression.

[tetsuok]

Summary

Although clang++ does not warn this, it's better to avoid returning void
expression.

[kcc]

what's so bad in returning void expression from a function returning void?
Not that I assert it's a great practice, but it makes the code shorter and I can't imagine where it can lead to problems.
If this is indeed a bad pattern, let's implement a clang warning.

[tetsuok]

Sorry. You're right. This is valid in C++ standard. I thought that this should be avoided because clang warns in C code. I personally avoid this pattern in C++. I should have checked the standard before sending this patch.

lib/Transforms/Instrumentation/AddressSanitizer.cpp

void AddressSanitizer::instrumentMop(Instruction *I, bool UseCalls) {
 
   if (IsWrite)
     NumInstrumentedWrites++;
   else
     NumInstrumentedReads++;
 
   // Instrument a 1-, 2-, 4-, 8-, or 16- byte access with one check.
   if (TypeSize == 8  || TypeSize == 16 ||
-      TypeSize == 32 || TypeSize == 64 || TypeSize == 128)
-    return instrumentAddress(I, I, Addr, TypeSize, IsWrite, 0, UseCalls);
+      TypeSize == 32 || TypeSize == 64 || TypeSize == 128) {
+    instrumentAddress(I, I, Addr, TypeSize, IsWrite, 0, UseCalls);
+    return;
+  }
   // Instrument unusual size (but still multiple of 8).
   // We can not do it with a single check, so we do 1-byte check for the first
   // and the last bytes. We call __asan_report_*_n(addr, real_size) to be able
   // to report the actual access size.
   IRBuilder<> IRB(I);
   Value *LastByte =  IRB.CreateIntToPtr(
       IRB.CreateAdd(IRB.CreatePointerCast(Addr, IntptrTy),
                     ConstantInt::get(IntptrTy, TypeSize / 8 - 1)),