[PATCH] X86: rework expansion of atomic instructions

[Andrea Di Biagio]

Hi Tim,

I really like this patch and in general it looks very good to me!

However, since I am not the code owner, you might want to ask for more feedback from Nadav or Jim.

Cheers,
Andrea

================
Comment at: lib/Target/X86/X86AtomicExpandPass.cpp:108
@@ +107,3 @@
+    return true;
+  else if (AI->getType()->getPrimitiveSizeInBits() > NativeWidth)
+    return false;
----------------
Please remove the 'else' after return.

================
Comment at: lib/Target/X86/X86AtomicExpandPass.cpp:147
@@ +146,3 @@
+    return shouldExpandAtomicRMW(AI);
+  else if (StoreInst *SI = dyn_cast<StoreInst>(Inst))
+    return shouldExpandStore(SI);
----------------
Please remove the 'else' after return.

http://reviews.llvm.org/D4160