Treat __sync_synchronize and asm("":::"memory") as stronger fences.

lib/CodeGen/CGStmt.cpp

 //===--- CGStmt.cpp - Emit LLVM Code from Statements ----------------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 // This contains code to emit Stmt nodes as LLVM code.
@@ skipping 1683 matching lines @@
   bool HasSideEffect = S.isVolatile() || S.getNumOutputs() == 0;
   llvm::InlineAsm::AsmDialect AsmDialect = isa<MSAsmStmt>(&S) ?
     llvm::InlineAsm::AD_Intel : llvm::InlineAsm::AD_ATT;
   llvm::InlineAsm *IA =
     llvm::InlineAsm::get(FTy, AsmString, Constraints, HasSideEffect,
                          /* IsAlignStack */ false, AsmDialect);
   llvm::CallInst *Result = Builder.CreateCall(IA, Args);
   Result->addAttribute(llvm::AttributeSet::FunctionIndex,
                        llvm::Attribute::NoUnwind);
 
+  // @LOCALMOD-START
+  // For PNaCl, emit two compiler fences containing a sequentially
+  // consistent fence (just like __sync_synchronize). This should
+  // enforce ordering of more than just atomic memory accesses, though
+  // it won't guarantee that all accesses (e.g. those to non-escaping
+  // objects) will not be reordered.
+  bool isEmptyAsm = AsmString.empty();
+  // Different triples will encode "touch everything" differently, e.g.:
+  //  - le32-unknown-nacl has "~{memory}".
+  //  - x86 "~{memory},~{dirflag},~{fpsr},~{flags}".
+  // The following code therefore only searches for memory: this pass
+  // deals with portable assembly, touching anything else than memory in
+  // an empty assembly statement is meaningless.
+  bool touchesMemory = Constraints.find("~{memory}") != std::string::npos;
+
+  if (getTarget().getTriple().getArch() == llvm::Triple::le32 &&
+      ResultType->isVoidTy() && HasSideEffect && isEmptyAsm && touchesMemory) {
+    Builder.CreateFence(llvm::SequentiallyConsistent);
+    Builder.CreateCall(
+        llvm::InlineAsm::get(FTy, AsmString, Constraints, HasSideEffect))->
+        addAttribute(llvm::AttributeSet::FunctionIndex,
+                     llvm::Attribute::NoUnwind);
+  }
+  // @LOCALMOD-END
+
   // Slap the source location of the inline asm into a !srcloc metadata on the
   // call.  FIXME: Handle metadata for MS-style inline asms.
   if (const GCCAsmStmt *gccAsmStmt = dyn_cast<GCCAsmStmt>(&S))
     Result->setMetadata("srcloc", getAsmSrcLocInfo(gccAsmStmt->getAsmString(),
                                                    *this));
 
   // Extract all of the register value results from the asm.
   std::vector<llvm::Value*> RegResults;
   if (ResultRegTypes.size() == 1) {
     RegResults.push_back(Result);
@@ skipping 33 matching lines @@
       }
     }
 
     EmitStoreThroughLValue(RValue::get(Tmp), ResultRegDests[i]);
   }
 }
 
 void CodeGenFunction::EmitCapturedStmt(const CapturedStmt &S) {
   llvm_unreachable("not implemented yet");
 }