Fix bug when atomic load is fused with an arith op (and not in the entry BB)

src/IceTargetLoweringX8632.cpp

 //===- subzero/src/IceTargetLoweringX8632.cpp - x86-32 lowering -----------===//
 //
 //                        The Subzero Code Generator
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 // This file implements the TargetLoweringX8632 class, which
@@ skipping 2709 matching lines @@
       // normally does (split the load into two).
       // Thus, this skips load/arithmetic op folding. Load/arithmetic folding
       // can't happen anyway, since this is x86-32 and integer arithmetic only
       // happens on 32-bit quantities.
       Variable *T = makeReg(IceType_f64);
       OperandX8632Mem *Addr = FormMemoryOperand(Instr->getArg(0), IceType_f64);
       _movq(T, Addr);
       // Then cast the bits back out of the XMM register to the i64 Dest.
       InstCast *Cast = InstCast::create(Func, InstCast::Bitcast, Dest, T);
       lowerCast(Cast);
-      // Make sure that the atomic load isn't elided.
+      // Make sure that the atomic load isn't elided when unused.
       Context.insert(InstFakeUse::create(Func, Dest->getLo()));
       Context.insert(InstFakeUse::create(Func, Dest->getHi()));
       return;
     }
     InstLoad *Load = InstLoad::create(Func, Dest, Instr->getArg(0));
     lowerLoad(Load);
-    // Make sure the atomic load isn't elided.
-    Context.insert(InstFakeUse::create(Func, Dest));
+    // Make sure the atomic load isn't elided when unused, by adding a FakeUse.
+    // Since lowerLoad may fuse the load w/ an arithmetic instruction,
+    // insert the FakeUse on the last-inserted instruction's dest.
+    Context.insert(InstFakeUse::create(Func,
+                                       Context.getLastInserted()->getDest()));
     return;
   }
   case Intrinsics::AtomicRMW:
     if (!Intrinsics::VerifyMemoryOrder(
              llvm::cast<ConstantInteger>(Instr->getArg(3))->getValue())) {
       Func->setError("Unexpected memory ordering for AtomicRMW");
       return;
     }
     lowerAtomicRMW(Instr->getDest(),
                    static_cast<uint32_t>(llvm::cast<ConstantInteger>(
@@ skipping 1268 matching lines @@
     for (SizeT i = 0; i < Size; ++i) {
       Str << "\t.byte\t" << (((unsigned)Data[i]) & 0xff) << "\n";
     }
     Str << "\t.size\t" << MangledName << ", " << Size << "\n";
   }
   Str << "\t" << (IsInternal ? ".local" : ".global") << "\t" << MangledName
       << "\n";
 }
 
 } // end of namespace Ice