[CodeGen] Emit a helper function for __builtin_os_log_format to reduce

code size.

Currently clang expands a call to __builtin_os_log_format into a long
sequence of instructions at the call site, causing code size to
increase in some cases.

This commit attempts to reduce code size by emitting a helper function
that can be shared by calls to __builtin_os_log_format with similar
formats and arguments. The helper function has linkonce_odr linkage to
enable the linker to merge identical functions across translation units.
Attribute 'noinline' is attached to the helper function at -Oz so that
the inliner doesn't inline functions that can potentially be merged.

This commit also fixes a bug where the generated IR writes past the end
of the buffer when "%m" is the last specifier appearing in the format
string passed to __builtin_os_log_format.

Original patch by Duncan Exon Smith.

rdar://problem/34065973
rdar://problem/34196543

Differential Revision: https://reviews.llvm.org/D38606

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@315045 91177308-0d34-0410-b5e6-96231b3b80d8

4 files changed, 580 insertions, 261 deletions

diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index f8ce0ce375..5c8b36cdfa 100644
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -663,6 +663,179 @@ Value *CodeGenFunction::EmitCheckedArgForBuiltin(const Expr *E,
   return ArgValue;
 }
 
+/// Get the argument type for arguments to os_log_helper.
+static CanQualType getOSLogArgType(ASTContext &C, int Size) {
+  QualType UnsignedTy = C.getIntTypeForBitwidth(Size * 8, /*Signed=*/false);
+  return C.getCanonicalType(UnsignedTy);
+}
+
+llvm::Function *CodeGenFunction::generateBuiltinOSLogHelperFunction(
+    const analyze_os_log::OSLogBufferLayout &Layout,
+    CharUnits BufferAlignment) {
+  ASTContext &Ctx = getContext();
+
+  llvm::SmallString<64> Name;
+  {
+    raw_svector_ostream OS(Name);
+    OS << "__os_log_helper";
+    OS << "_" << BufferAlignment.getQuantity();
+    OS << "_" << int(Layout.getSummaryByte());
+    OS << "_" << int(Layout.getNumArgsByte());
+    for (const auto &Item : Layout.Items)
+      OS << "_" << int(Item.getSizeByte()) << "_"
+         << int(Item.getDescriptorByte());
+  }
+
+  if (llvm::Function *F = CGM.getModule().getFunction(Name))
+    return F;
+
+  llvm::SmallVector<ImplicitParamDecl, 4> Params;
+  Params.emplace_back(Ctx, nullptr, SourceLocation(), &Ctx.Idents.get("buffer"),
+                      Ctx.VoidPtrTy, ImplicitParamDecl::Other);
+
+  for (unsigned int I = 0, E = Layout.Items.size(); I < E; ++I) {
+    char Size = Layout.Items[I].getSizeByte();
+    if (!Size)
+      continue;
+
+    Params.emplace_back(Ctx, nullptr, SourceLocation(),
+                        &Ctx.Idents.get(std::string("arg") + std::to_string(I)),
+                        getOSLogArgType(Ctx, Size), ImplicitParamDecl::Other);
+  }
+
+  FunctionArgList Args;
+  for (auto &P : Params)
+    Args.push_back(&P);
+
+  // The helper function has linkonce_odr linkage to enable the linker to merge
+  // identical functions. To ensure the merging always happens, 'noinline' is
+  // attached to the function when compiling with -Oz.
+  const CGFunctionInfo &FI =
+      CGM.getTypes().arrangeBuiltinFunctionDeclaration(Ctx.VoidTy, Args);
+  llvm::FunctionType *FuncTy = CGM.getTypes().GetFunctionType(FI);
+  llvm::Function *Fn = llvm::Function::Create(
+      FuncTy, llvm::GlobalValue::LinkOnceODRLinkage, Name, &CGM.getModule());
+  Fn->setVisibility(llvm::GlobalValue::HiddenVisibility);
+  CGM.SetLLVMFunctionAttributes(nullptr, FI, Fn);
+  CGM.SetLLVMFunctionAttributesForDefinition(nullptr, Fn);
+
+  // Attach 'noinline' at -Oz.
+  if (CGM.getCodeGenOpts().OptimizeSize == 2)
+    Fn->addFnAttr(llvm::Attribute::NoInline);
+
+  auto NL = ApplyDebugLocation::CreateEmpty(*this);
+  IdentifierInfo *II = &Ctx.Idents.get(Name);
+  FunctionDecl *FD = FunctionDecl::Create(
+      Ctx, Ctx.getTranslationUnitDecl(), SourceLocation(), SourceLocation(), II,
+      Ctx.VoidTy, nullptr, SC_PrivateExtern, false, false);
+
+  StartFunction(FD, Ctx.VoidTy, Fn, FI, Args);
+
+  // Create a scope with an artificial location for the body of this function.
+  auto AL = ApplyDebugLocation::CreateArtificial(*this);
+
+  CharUnits Offset;
+  Address BufAddr(Builder.CreateLoad(GetAddrOfLocalVar(&Params[0]), "buf"),
+                  BufferAlignment);
+  Builder.CreateStore(Builder.getInt8(Layout.getSummaryByte()),
+                      Builder.CreateConstByteGEP(BufAddr, Offset++, "summary"));
+  Builder.CreateStore(Builder.getInt8(Layout.getNumArgsByte()),
+                      Builder.CreateConstByteGEP(BufAddr, Offset++, "numArgs"));
+
+  unsigned I = 1;
+  for (const auto &Item : Layout.Items) {
+    Builder.CreateStore(
+        Builder.getInt8(Item.getDescriptorByte()),
+        Builder.CreateConstByteGEP(BufAddr, Offset++, "argDescriptor"));
+    Builder.CreateStore(
+        Builder.getInt8(Item.getSizeByte()),
+        Builder.CreateConstByteGEP(BufAddr, Offset++, "argSize"));
+
+    CharUnits Size = Item.size();
+    if (!Size.getQuantity())
+      continue;
+
+    Address Arg = GetAddrOfLocalVar(&Params[I]);
+    Address Addr = Builder.CreateConstByteGEP(BufAddr, Offset, "argData");
+    Addr = Builder.CreateBitCast(Addr, Arg.getPointer()->getType(),
+                                 "argDataCast");
+    Builder.CreateStore(Builder.CreateLoad(Arg), Addr);
+    Offset += Size;
+    ++I;
+  }
+
+  FinishFunction();
+
+  return Fn;
+}
+
+RValue CodeGenFunction::emitBuiltinOSLogFormat(const CallExpr &E) {
+  assert(E.getNumArgs() >= 2 &&
+         "__builtin_os_log_format takes at least 2 arguments");
+  ASTContext &Ctx = getContext();
+  analyze_os_log::OSLogBufferLayout Layout;
+  analyze_os_log::computeOSLogBufferLayout(Ctx, &E, Layout);
+  Address BufAddr = EmitPointerWithAlignment(E.getArg(0));
+  llvm::SmallVector<llvm::Value *, 4> RetainableOperands;
+
+  // Ignore argument 1, the format string. It is not currently used.
+  CallArgList Args;
+  Args.add(RValue::get(BufAddr.getPointer()), Ctx.VoidPtrTy);
+
+  for (const auto &Item : Layout.Items) {
+    int Size = Item.getSizeByte();
+    if (!Size)
+      continue;
+
+    llvm::Value *ArgVal;
+
+    if (const Expr *TheExpr = Item.getExpr()) {
+      ArgVal = EmitScalarExpr(TheExpr, /*Ignore*/ false);
+
+      // Check if this is a retainable type.
+      if (TheExpr->getType()->isObjCRetainableType()) {
+        assert(getEvaluationKind(TheExpr->getType()) == TEK_Scalar &&
+               "Only scalar can be a ObjC retainable type");
+        // Check if the object is constant, if not, save it in
+        // RetainableOperands.
+        if (!isa<Constant>(ArgVal))
+          RetainableOperands.push_back(ArgVal);
+      }
+    } else {
+      ArgVal = Builder.getInt32(Item.getConstValue().getQuantity());
+    }
+
+    unsigned ArgValSize =
+        CGM.getDataLayout().getTypeSizeInBits(ArgVal->getType());
+    llvm::IntegerType *IntTy = llvm::Type::getIntNTy(getLLVMContext(),
+                                                     ArgValSize);
+    ArgVal = Builder.CreateBitOrPointerCast(ArgVal, IntTy);
+    CanQualType ArgTy = getOSLogArgType(Ctx, Size);
+    // If ArgVal has type x86_fp80, zero-extend ArgVal.
+    ArgVal = Builder.CreateZExtOrBitCast(ArgVal, ConvertType(ArgTy));
+    Args.add(RValue::get(ArgVal), ArgTy);
+  }
+
+  const CGFunctionInfo &FI =
+      CGM.getTypes().arrangeBuiltinFunctionCall(Ctx.VoidTy, Args);
+  llvm::Function *F = CodeGenFunction(CGM).generateBuiltinOSLogHelperFunction(
+      Layout, BufAddr.getAlignment());
+  EmitCall(FI, CGCallee::forDirect(F), ReturnValueSlot(), Args);
+
+  // Push a clang.arc.use cleanup for each object in RetainableOperands. The
+  // cleanup will cause the use to appear after the final log call, keeping
+  // the object valid while it’s held in the log buffer.  Note that if there’s
+  // a release cleanup on the object, it will already be active; since
+  // cleanups are emitted in reverse order, the use will occur before the
+  // object is released.
+  if (!RetainableOperands.empty() && getLangOpts().ObjCAutoRefCount &&
+      CGM.getCodeGenOpts().OptimizationLevel != 0)
+    for (llvm::Value *Object : RetainableOperands)
+      pushFullExprCleanup<CallObjCArcUse>(getARCCleanupKind(), Object);
+
+  return RValue::get(BufAddr.getPointer());
+}
+
 RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
                                         unsigned BuiltinID, const CallExpr *E,
                                         ReturnValueSlot ReturnValue) {
@@ -2801,69 +2974,8 @@ RValue CodeGenFunction::EmitBuiltinExpr(const FunctionDecl *FD,
     // Fall through - it's already mapped to the intrinsic by GCCBuiltin.
     break;
   }
-  case Builtin::BI__builtin_os_log_format: {
-    assert(E->getNumArgs() >= 2 &&
-           "__builtin_os_log_format takes at least 2 arguments");
-    analyze_os_log::OSLogBufferLayout Layout;
-    analyze_os_log::computeOSLogBufferLayout(CGM.getContext(), E, Layout);
-    Address BufAddr = EmitPointerWithAlignment(E->getArg(0));
-    // Ignore argument 1, the format string. It is not currently used.
-    CharUnits Offset;
-    Builder.CreateStore(
-        Builder.getInt8(Layout.getSummaryByte()),
-        Builder.CreateConstByteGEP(BufAddr, Offset++, "summary"));
-    Builder.CreateStore(
-        Builder.getInt8(Layout.getNumArgsByte()),
-        Builder.CreateConstByteGEP(BufAddr, Offset++, "numArgs"));
-
-    llvm::SmallVector<llvm::Value *, 4> RetainableOperands;
-    for (const auto &Item : Layout.Items) {
-      Builder.CreateStore(
-          Builder.getInt8(Item.getDescriptorByte()),
-          Builder.CreateConstByteGEP(BufAddr, Offset++, "argDescriptor"));
-      Builder.CreateStore(
-          Builder.getInt8(Item.getSizeByte()),
-          Builder.CreateConstByteGEP(BufAddr, Offset++, "argSize"));
-      Address Addr = Builder.CreateConstByteGEP(BufAddr, Offset);
-      if (const Expr *TheExpr = Item.getExpr()) {
-        Addr = Builder.CreateElementBitCast(
-            Addr, ConvertTypeForMem(TheExpr->getType()));
-        // Check if this is a retainable type.
-        if (TheExpr->getType()->isObjCRetainableType()) {
-          assert(getEvaluationKind(TheExpr->getType()) == TEK_Scalar &&
-                 "Only scalar can be a ObjC retainable type");
-          llvm::Value *SV = EmitScalarExpr(TheExpr, /*Ignore*/ false);
-          RValue RV = RValue::get(SV);
-          LValue LV = MakeAddrLValue(Addr, TheExpr->getType());
-          EmitStoreThroughLValue(RV, LV);
-          // Check if the object is constant, if not, save it in
-          // RetainableOperands.
-          if (!isa<Constant>(SV))
-            RetainableOperands.push_back(SV);
-        } else {
-          EmitAnyExprToMem(TheExpr, Addr, Qualifiers(), /*isInit*/ true);
-        }
-      } else {
-        Addr = Builder.CreateElementBitCast(Addr, Int32Ty);
-        Builder.CreateStore(
-            Builder.getInt32(Item.getConstValue().getQuantity()), Addr);
-      }
-      Offset += Item.size();
-    }
-
-    // Push a clang.arc.use cleanup for each object in RetainableOperands. The
-    // cleanup will cause the use to appear after the final log call, keeping
-    // the object valid while it's held in the log buffer.  Note that if there's
-    // a release cleanup on the object, it will already be active; since
-    // cleanups are emitted in reverse order, the use will occur before the
-    // object is released.
-    if (!RetainableOperands.empty() && getLangOpts().ObjCAutoRefCount &&
-        CGM.getCodeGenOpts().OptimizationLevel != 0)
-      for (llvm::Value *object : RetainableOperands)
-        pushFullExprCleanup<CallObjCArcUse>(getARCCleanupKind(), object);
-
-    return RValue::get(BufAddr.getPointer());
-  }
+  case Builtin::BI__builtin_os_log_format:
+    return emitBuiltinOSLogFormat(*E);
 
   case Builtin::BI__builtin_os_log_format_buffer_size: {
     analyze_os_log::OSLogBufferLayout Layout;
diff --git a/lib/CodeGen/CodeGenFunction.h b/lib/CodeGen/CodeGenFunction.h
index 50c8ebf7a5..8cee6b64d4 100644
--- a/lib/CodeGen/CodeGenFunction.h
+++ b/lib/CodeGen/CodeGenFunction.h
@@ -76,6 +76,10 @@ class ObjCAtThrowStmt;
 class ObjCAtSynchronizedStmt;
 class ObjCAutoreleasePoolStmt;
 
+namespace analyze_os_log {
+class OSLogBufferLayout;
+}
+
 namespace CodeGen {
 class CodeGenTypes;
 class CGCallee;
@@ -3304,6 +3308,13 @@ public:
                          unsigned BuiltinID, const CallExpr *E,
                          ReturnValueSlot ReturnValue);
 
+  /// Emit IR for __builtin_os_log_format.
+  RValue emitBuiltinOSLogFormat(const CallExpr &E);
+
+  llvm::Function *generateBuiltinOSLogHelperFunction(
+      const analyze_os_log::OSLogBufferLayout &Layout,
+      CharUnits BufferAlignment);
+
   RValue EmitBlockCallExpr(const CallExpr *E, ReturnValueSlot ReturnValue);
 
   /// EmitTargetBuiltinExpr - Emit the given builtin call. Returns 0 if the call
diff --git a/test/CodeGen/builtins.c b/test/CodeGen/builtins.c
index 390c2e35bf..3584585de8 100644
--- a/test/CodeGen/builtins.c
+++ b/test/CodeGen/builtins.c
@@ -378,229 +378,385 @@ long long test_builtin_readcyclecounter() {
 #ifdef __x86_64__
 
 // CHECK-LABEL: define void @test_builtin_os_log
-// CHECK: (i8* [[BUF:%.*]], i32 [[I:%.*]], i8* [[DATA:%.*]])
+// CHECK: (i8* %[[BUF:.*]], i32 %[[I:.*]], i8* %[[DATA:.*]])
 void test_builtin_os_log(void *buf, int i, const char *data) {
   volatile int len;
-  // CHECK: store i8* [[BUF]], i8** [[BUF_ADDR:%.*]], align 8
-  // CHECK: store i32 [[I]], i32* [[I_ADDR:%.*]], align 4
-  // CHECK: store i8* [[DATA]], i8** [[DATA_ADDR:%.*]], align 8
-
-  // CHECK: store volatile i32 34
+  // CHECK: %[[BUF_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[I_ADDR:.*]] = alloca i32, align 4
+  // CHECK: %[[DATA_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[LEN:.*]] = alloca i32, align 4
+  // CHECK: store i8* %[[BUF]], i8** %[[BUF_ADDR]], align 8
+  // CHECK: store i32 %[[I]], i32* %[[I_ADDR]], align 4
+  // CHECK: store i8* %[[DATA]], i8** %[[DATA_ADDR]], align 8
+
+  // CHECK: store volatile i32 34, i32* %[[LEN]]
   len = __builtin_os_log_format_buffer_size("%d %{public}s %{private}.16P", i, data, data);
 
-  // CHECK: [[BUF2:%.*]] = load i8*, i8** [[BUF_ADDR]]
-  // CHECK: [[SUMMARY:%.*]] = getelementptr i8, i8* [[BUF2]], i64 0
-  // CHECK: store i8 3, i8* [[SUMMARY]]
-  // CHECK: [[NUM_ARGS:%.*]] = getelementptr i8, i8* [[BUF2]], i64 1
-  // CHECK: store i8 4, i8* [[NUM_ARGS]]
-  //
-  // CHECK: [[ARG1_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 2
-  // CHECK: store i8 0, i8* [[ARG1_DESC]]
-  // CHECK: [[ARG1_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 3
-  // CHECK: store i8 4, i8* [[ARG1_SIZE]]
-  // CHECK: [[ARG1:%.*]] = getelementptr i8, i8* [[BUF2]], i64 4
-  // CHECK: [[ARG1_INT:%.*]] = bitcast i8* [[ARG1]] to i32*
-  // CHECK: [[I2:%.*]] = load i32, i32* [[I_ADDR]]
-  // CHECK: store i32 [[I2]], i32* [[ARG1_INT]]
-
-  // CHECK: [[ARG2_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 8
-  // CHECK: store i8 34, i8* [[ARG2_DESC]]
-  // CHECK: [[ARG2_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 9
-  // CHECK: store i8 8, i8* [[ARG2_SIZE]]
-  // CHECK: [[ARG2:%.*]] = getelementptr i8, i8* [[BUF2]], i64 10
-  // CHECK: [[ARG2_PTR:%.*]] = bitcast i8* [[ARG2]] to i8**
-  // CHECK: [[DATA2:%.*]] = load i8*, i8** [[DATA_ADDR]]
-  // CHECK: store i8* [[DATA2]], i8** [[ARG2_PTR]]
-
-  // CHECK: [[ARG3_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 18
-  // CHECK: store i8 17, i8* [[ARG3_DESC]]
-  // CHECK: [[ARG3_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 19
-  // CHECK: store i8 4, i8* [[ARG3_SIZE]]
-  // CHECK: [[ARG3:%.*]] = getelementptr i8, i8* [[BUF2]], i64 20
-  // CHECK: [[ARG3_INT:%.*]] = bitcast i8* [[ARG3]] to i32*
-  // CHECK: store i32 16, i32* [[ARG3_INT]]
-
-  // CHECK: [[ARG4_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 24
-  // CHECK: store i8 49, i8* [[ARG4_DESC]]
-  // CHECK: [[ARG4_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 25
-  // CHECK: store i8 8, i8* [[ARG4_SIZE]]
-  // CHECK: [[ARG4:%.*]] = getelementptr i8, i8* [[BUF2]], i64 26
-  // CHECK: [[ARG4_PTR:%.*]] = bitcast i8* [[ARG4]] to i8**
-  // CHECK: [[DATA3:%.*]] = load i8*, i8** [[DATA_ADDR]]
-  // CHECK: store i8* [[DATA3]], i8** [[ARG4_PTR]]
-
+  // CHECK: %[[V1:.*]] = load i8*, i8** %[[BUF_ADDR]]
+  // CHECK: %[[V2:.*]] = load i32, i32* %[[I_ADDR]]
+  // CHECK: %[[V3:.*]] = load i8*, i8** %[[DATA_ADDR]]
+  // CHECK: %[[V4:.*]] = ptrtoint i8* %[[V3]] to i64
+  // CHECK: %[[V5:.*]] = load i8*, i8** %[[DATA_ADDR]]
+  // CHECK: %[[V6:.*]] = ptrtoint i8* %[[V5]] to i64
+  // CHECK: call void @__os_log_helper_1_3_4_4_0_8_34_4_17_8_49(i8* %[[V1]], i32 %[[V2]], i64 %[[V4]], i32 16, i64 %[[V6]])
   __builtin_os_log_format(buf, "%d %{public}s %{private}.16P", i, data, data);
 }
 
-// CHECK-LABEL: define void @test_builtin_os_log_errno
-// CHECK: (i8* [[BUF:%.*]], i8* [[DATA:%.*]])
-void test_builtin_os_log_errno(void *buf, const char *data) {
-  volatile int len;
-  // CHECK: store i8* [[BUF]], i8** [[BUF_ADDR:%.*]], align 8
-  // CHECK: store i8* [[DATA]], i8** [[DATA_ADDR:%.*]], align 8
-
-  // CHECK: store volatile i32 2
-  len = __builtin_os_log_format_buffer_size("%S");
-
-  // CHECK: [[BUF2:%.*]] = load i8*, i8** [[BUF_ADDR]]
-  // CHECK: [[SUMMARY:%.*]] = getelementptr i8, i8* [[BUF2]], i64 0
-  // CHECK: store i8 2, i8* [[SUMMARY]]
-  // CHECK: [[NUM_ARGS:%.*]] = getelementptr i8, i8* [[BUF2]], i64 1
-  // CHECK: store i8 1, i8* [[NUM_ARGS]]
-
-  // CHECK: [[ARG1_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 2
-  // CHECK: store i8 96, i8* [[ARG1_DESC]]
-  // CHECK: [[ARG1_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 3
-  // CHECK: store i8 0, i8* [[ARG1_SIZE]]
-  // CHECK: [[ARG1:%.*]] = getelementptr i8, i8* [[BUF2]], i64 4
-  // CHECK: [[ARG1_INT:%.*]] = bitcast i8* [[ARG1]] to i32*
-  // CHECK: store i32 0, i32* [[ARG1_INT]]
-
-  __builtin_os_log_format(buf, "%m");
-}
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_1_3_4_4_0_8_34_4_17_8_49
+// CHECK: (i8* %[[BUFFER:.*]], i32 %[[ARG0:.*]], i64 %[[ARG1:.*]], i32 %[[ARG2:.*]], i64 %[[ARG3:.*]])
+
+// CHECK: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK: %[[ARG0_ADDR:.*]] = alloca i32, align 4
+// CHECK: %[[ARG1_ADDR:.*]] = alloca i64, align 8
+// CHECK: %[[ARG2_ADDR:.*]] = alloca i32, align 4
+// CHECK: %[[ARG3_ADDR:.*]] = alloca i64, align 8
+// CHECK: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK: store i32 %[[ARG0]], i32* %[[ARG0_ADDR]], align 4
+// CHECK: store i64 %[[ARG1]], i64* %[[ARG1_ADDR]], align 8
+// CHECK: store i32 %[[ARG2]], i32* %[[ARG2_ADDR]], align 4
+// CHECK: store i64 %[[ARG3]], i64* %[[ARG3_ADDR]], align 8
+// CHECK: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK: store i8 3, i8* %[[SUMMARY]], align 1
+// CHECK: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK: store i8 4, i8* %[[NUMARGS]], align 1
+// CHECK: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK: store i8 0, i8* %[[ARGDESCRIPTOR]], align 1
+// CHECK: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK: store i8 4, i8* %[[ARGSIZE]], align 1
+// CHECK: %[[ARGDATA:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+// CHECK: %[[ARGDATACAST:.*]] = bitcast i8* %[[ARGDATA]] to i32*
+// CHECK: %[[V0:.*]] = load i32, i32* %[[ARG0_ADDR]], align 4
+// CHECK: store i32 %[[V0]], i32* %[[ARGDATACAST]], align 1
+// CHECK: %[[ARGDESCRIPTOR1:.*]] = getelementptr i8, i8* %[[BUF]], i64 8
+// CHECK: store i8 34, i8* %[[ARGDESCRIPTOR1]], align 1
+// CHECK: %[[ARGSIZE2:.*]] = getelementptr i8, i8* %[[BUF]], i64 9
+// CHECK: store i8 8, i8* %[[ARGSIZE2]], align 1
+// CHECK: %[[ARGDATA3:.*]] = getelementptr i8, i8* %[[BUF]], i64 10
+// CHECK: %[[ARGDATACAST4:.*]] = bitcast i8* %[[ARGDATA3]] to i64*
+// CHECK: %[[V1:.*]] = load i64, i64* %[[ARG1_ADDR]], align 8
+// CHECK: store i64 %[[V1]], i64* %[[ARGDATACAST4]], align 1
+// CHECK: %[[ARGDESCRIPTOR5:.*]] = getelementptr i8, i8* %[[BUF]], i64 18
+// CHECK: store i8 17, i8* %[[ARGDESCRIPTOR5]], align 1
+// CHECK: %[[ARGSIZE6:.*]] = getelementptr i8, i8* %[[BUF]], i64 19
+// CHECK: store i8 4, i8* %[[ARGSIZE6]], align 1
+// CHECK: %[[ARGDATA7:.*]] = getelementptr i8, i8* %[[BUF]], i64 20
+// CHECK: %[[ARGDATACAST8:.*]] = bitcast i8* %[[ARGDATA7]] to i32*
+// CHECK: %[[V2:.*]] = load i32, i32* %[[ARG2_ADDR]], align 4
+// CHECK: store i32 %[[V2]], i32* %[[ARGDATACAST8]], align 1
+// CHECK: %[[ARGDESCRIPTOR9:.*]] = getelementptr i8, i8* %[[BUF]], i64 24
+// CHECK: store i8 49, i8* %[[ARGDESCRIPTOR9]], align 1
+// CHECK: %[[ARGSIZE10:.*]] = getelementptr i8, i8* %[[BUF]], i64 25
+// CHECK: store i8 8, i8* %[[ARGSIZE10]], align 1
+// CHECK: %[[ARGDATA11:.*]] = getelementptr i8, i8* %[[BUF]], i64 26
+// CHECK: %[[ARGDATACAST12:.*]] = bitcast i8* %[[ARGDATA11]] to i64*
+// CHECK: %[[V3:.*]] = load i64, i64* %[[ARG3_ADDR]], align 8
+// CHECK: store i64 %[[V3]], i64* %[[ARGDATACAST12]], align 1
 
 // CHECK-LABEL: define void @test_builtin_os_log_wide
-// CHECK: (i8* [[BUF:%.*]], i8* [[DATA:%.*]], i32* [[STR:%.*]])
+// CHECK: (i8* %[[BUF:.*]], i8* %[[DATA:.*]], i32* %[[STR:.*]])
 typedef int wchar_t;
 void test_builtin_os_log_wide(void *buf, const char *data, wchar_t *str) {
   volatile int len;
-  // CHECK: store i8* [[BUF]], i8** [[BUF_ADDR:%.*]], align 8
-  // CHECK: store i8* [[DATA]], i8** [[DATA_ADDR:%.*]], align 8
-  // CHECK: store i32* [[STR]], i32** [[STR_ADDR:%.*]],
 
-  // CHECK: store volatile i32 12
+  // CHECK: %[[BUF_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[DATA_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[STR_ADDR:.*]] = alloca i32*, align 8
+  // CHECK: %[[LEN:.*]] = alloca i32, align 4
+  // CHECK: store i8* %[[BUF]], i8** %[[BUF_ADDR]], align 8
+  // CHECK: store i8* %[[DATA]], i8** %[[DATA_ADDR]], align 8
+  // CHECK: store i32* %[[STR]], i32** %[[STR_ADDR]], align 8
+
+  // CHECK: store volatile i32 12, i32* %[[LEN]], align 4
   len = __builtin_os_log_format_buffer_size("%S", str);
 
-  // CHECK: [[BUF2:%.*]] = load i8*, i8** [[BUF_ADDR]]
-  // CHECK: [[SUMMARY:%.*]] = getelementptr i8, i8* [[BUF2]], i64 0
-  // CHECK: store i8 2, i8* [[SUMMARY]]
-  // CHECK: [[NUM_ARGS:%.*]] = getelementptr i8, i8* [[BUF2]], i64 1
-  // CHECK: store i8 1, i8* [[NUM_ARGS]]
-
-  // CHECK: [[ARG1_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 2
-  // CHECK: store i8 80, i8* [[ARG1_DESC]]
-  // CHECK: [[ARG1_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 3
-  // CHECK: store i8 8, i8* [[ARG1_SIZE]]
-  // CHECK: [[ARG1:%.*]] = getelementptr i8, i8* [[BUF2]], i64 4
-  // CHECK: [[ARG1_PTR:%.*]] = bitcast i8* [[ARG1]] to i32**
-  // CHECK: [[STR2:%.*]] = load i32*, i32** [[STR_ADDR]]
-  // CHECK: store i32* [[STR2]], i32** [[ARG1_PTR]]
+  // CHECK: %[[V1:.*]] = load i8*, i8** %[[BUF_ADDR]], align 8
+  // CHECK: %[[V2:.*]] = load i32*, i32** %[[STR_ADDR]], align 8
+  // CHECK: %[[V3:.*]] = ptrtoint i32* %[[V2]] to i64
+  // CHECK: call void @__os_log_helper_1_2_1_8_80(i8* %[[V1]], i64 %[[V3]])
 
   __builtin_os_log_format(buf, "%S", str);
 }
 
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_1_2_1_8_80
+// CHECK: (i8* %[[BUFFER:.*]], i64 %[[ARG0:.*]])
+
+// CHECK: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK: %[[ARG0_ADDR:.*]] = alloca i64, align 8
+// CHECK: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK: store i64 %[[ARG0]], i64* %[[ARG0_ADDR]], align 8
+// CHECK: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK: store i8 2, i8* %[[SUMMARY]], align 1
+// CHECK: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK: store i8 1, i8* %[[NUMARGS]], align 1
+// CHECK: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK: store i8 80, i8* %[[ARGDESCRIPTOR]], align 1
+// CHECK: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK: store i8 8, i8* %[[ARGSIZE]], align 1
+// CHECK: %[[ARGDATA:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+// CHECK: %[[ARGDATACAST:.*]] = bitcast i8* %[[ARGDATA]] to i64*
+// CHECK: %[[V0:.*]] = load i64, i64* %[[ARG0_ADDR]], align 8
+// CHECK: store i64 %[[V0]], i64* %[[ARGDATACAST]], align 1
+
 // CHECK-LABEL: define void @test_builtin_os_log_precision_width
-// CHECK: (i8* [[BUF:%.*]], i8* [[DATA:%.*]], i32 [[PRECISION:%.*]], i32 [[WIDTH:%.*]])
+// CHECK: (i8* %[[BUF:.*]], i8* %[[DATA:.*]], i32 %[[PRECISION:.*]], i32 %[[WIDTH:.*]])
 void test_builtin_os_log_precision_width(void *buf, const char *data,
                                          int precision, int width) {
   volatile int len;
-  // CHECK: store i8* [[BUF]], i8** [[BUF_ADDR:%.*]], align 8
-  // CHECK: store i8* [[DATA]], i8** [[DATA_ADDR:%.*]], align 8
-  // CHECK: store i32 [[PRECISION]], i32* [[PRECISION_ADDR:%.*]], align 4
-  // CHECK: store i32 [[WIDTH]], i32* [[WIDTH_ADDR:%.*]], align 4
-
-  // CHECK: store volatile i32 24,
+  // CHECK: %[[BUF_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[DATA_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[PRECISION_ADDR:.*]] = alloca i32, align 4
+  // CHECK: %[[WIDTH_ADDR:.*]] = alloca i32, align 4
+  // CHECK: %[[LEN:.*]] = alloca i32, align 4
+  // CHECK: store i8* %[[BUF]], i8** %[[BUF_ADDR]], align 8
+  // CHECK: store i8* %[[DATA]], i8** %[[DATA_ADDR]], align 8
+  // CHECK: store i32 %[[PRECISION]], i32* %[[PRECISION_ADDR]], align 4
+  // CHECK: store i32 %[[WIDTH]], i32* %[[WIDTH_ADDR]], align 4
+
+  // CHECK: store volatile i32 24, i32* %[[LEN]], align 4
   len = __builtin_os_log_format_buffer_size("Hello %*.*s World", precision, width, data);
 
-  // CHECK: [[BUF2:%.*]] = load i8*, i8** [[BUF_ADDR]]
-  // CHECK: [[SUMMARY:%.*]] = getelementptr i8, i8* [[BUF2]], i64 0
-  // CHECK: store i8 2, i8* [[SUMMARY]]
-  // CHECK: [[NUM_ARGS:%.*]] = getelementptr i8, i8* [[BUF2]], i64 1
-  // CHECK: store i8 3, i8* [[NUM_ARGS]]
-
-  // CHECK: [[ARG1_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 2
-  // CHECK: store i8 0, i8* [[ARG1_DESC]]
-  // CHECK: [[ARG1_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 3
-  // CHECK: store i8 4, i8* [[ARG1_SIZE]]
-  // CHECK: [[ARG1:%.*]] = getelementptr i8, i8* [[BUF2]], i64 4
-  // CHECK: [[ARG1_INT:%.*]] = bitcast i8* [[ARG1]] to i32*
-  // CHECK: [[ARG1_VAL:%.*]] = load i32, i32* [[PRECISION_ADDR]]
-  // CHECK: store i32 [[ARG1_VAL]], i32* [[ARG1_INT]]
-
-  // CHECK: [[ARG2_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 8
-  // CHECK: store i8 16, i8* [[ARG2_DESC]]
-  // CHECK: [[ARG2_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 9
-  // CHECK: store i8 4, i8* [[ARG2_SIZE]]
-  // CHECK: [[ARG2:%.*]] = getelementptr i8, i8* [[BUF2]], i64 10
-  // CHECK: [[ARG2_INT:%.*]] = bitcast i8* [[ARG2]] to i32*
-  // CHECK: [[ARG2_VAL:%.*]] = load i32, i32* [[WIDTH_ADDR]]
-  // CHECK: store i32 [[ARG2_VAL]], i32* [[ARG2_INT]]
-
-  // CHECK: [[ARG3_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 14
-  // CHECK: store i8 32, i8* [[ARG3_DESC]]
-  // CHECK: [[ARG3_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 15
-  // CHECK: store i8 8, i8* [[ARG3_SIZE]]
-  // CHECK: [[ARG3:%.*]] = getelementptr i8, i8* [[BUF2]], i64 16
-  // CHECK: [[ARG3_PTR:%.*]] = bitcast i8* [[ARG3]] to i8**
-  // CHECK: [[DATA2:%.*]] = load i8*, i8** [[DATA_ADDR]]
-  // CHECK: store i8* [[DATA2]], i8** [[ARG3_PTR]]
-
+  // CHECK: %[[V1:.*]] = load i8*, i8** %[[BUF_ADDR]], align 8
+  // CHECK: %[[V2:.*]] = load i32, i32* %[[PRECISION_ADDR]], align 4
+  // CHECK: %[[V3:.*]] = load i32, i32* %[[WIDTH_ADDR]], align 4
+  // CHECK: %[[V4:.*]] = load i8*, i8** %[[DATA_ADDR]], align 8
+  // CHECK: %[[V5:.*]] = ptrtoint i8* %[[V4]] to i64
+  // CHECK: call void @__os_log_helper_1_2_3_4_0_4_16_8_32(i8* %[[V1]], i32 %[[V2]], i32 %[[V3]], i64 %[[V5]])
   __builtin_os_log_format(buf, "Hello %*.*s World", precision, width, data);
 }
 
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_1_2_3_4_0_4_16_8_32
+// CHECK: (i8* %[[BUFFER:.*]], i32 %[[ARG0:.*]], i32 %[[ARG1:.*]], i64 %[[ARG2:.*]])
+
+// CHECK: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK: %[[ARG0_ADDR:.*]] = alloca i32, align 4
+// CHECK: %[[ARG1_ADDR:.*]] = alloca i32, align 4
+// CHECK: %[[ARG2_ADDR:.*]] = alloca i64, align 8
+// CHECK: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK: store i32 %[[ARG0]], i32* %[[ARG0_ADDR]], align 4
+// CHECK: store i32 %[[ARG1]], i32* %[[ARG1_ADDR]], align 4
+// CHECK: store i64 %[[ARG2]], i64* %[[ARG2_ADDR]], align 8
+// CHECK: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK: store i8 2, i8* %[[SUMMARY]], align 1
+// CHECK: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK: store i8 3, i8* %[[NUMARGS]], align 1
+// CHECK: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK: store i8 0, i8* %[[ARGDESCRIPTOR]], align 1
+// CHECK: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK: store i8 4, i8* %[[ARGSIZE]], align 1
+// CHECK: %[[ARGDATA:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+// CHECK: %[[ARGDATACAST:.*]] = bitcast i8* %[[ARGDATA]] to i32*
+// CHECK: %[[V0:.*]] = load i32, i32* %[[ARG0_ADDR]], align 4
+// CHECK: store i32 %[[V0]], i32* %[[ARGDATACAST]], align 1
+// CHECK: %[[ARGDESCRIPTOR1:.*]] = getelementptr i8, i8* %[[BUF]], i64 8
+// CHECK: store i8 16, i8* %[[ARGDESCRIPTOR1]], align 1
+// CHECK: %[[ARGSIZE2:.*]] = getelementptr i8, i8* %[[BUF]], i64 9
+// CHECK: store i8 4, i8* %[[ARGSIZE2]], align 1
+// CHECK: %[[ARGDATA3:.*]] = getelementptr i8, i8* %[[BUF]], i64 10
+// CHECK: %[[ARGDATACAST4:.*]] = bitcast i8* %[[ARGDATA3]] to i32*
+// CHECK: %[[V1:.*]] = load i32, i32* %[[ARG1_ADDR]], align 4
+// CHECK: store i32 %[[V1]], i32* %[[ARGDATACAST4]], align 1
+// CHECK: %[[ARGDESCRIPTOR5:.*]] = getelementptr i8, i8* %[[BUF]], i64 14
+// CHECK: store i8 32, i8* %[[ARGDESCRIPTOR5]], align 1
+// CHECK: %[[ARGSIZE6:.*]] = getelementptr i8, i8* %[[BUF]], i64 15
+// CHECK: store i8 8, i8* %[[ARGSIZE6]], align 1
+// CHECK: %[[ARGDATA7:.*]] = getelementptr i8, i8* %[[BUF]], i64 16
+// CHECK: %[[ARGDATACAST8:.*]] = bitcast i8* %[[ARGDATA7]] to i64*
+// CHECK: %[[V2:.*]] = load i64, i64* %[[ARG2_ADDR]], align 8
+// CHECK: store i64 %[[V2]], i64* %[[ARGDATACAST8]], align 1
+
 // CHECK-LABEL: define void @test_builtin_os_log_invalid
-// CHECK: (i8* [[BUF:%.*]], i32 [[DATA:%.*]])
+// CHECK: (i8* %[[BUF:.*]], i32 %[[DATA:.*]])
 void test_builtin_os_log_invalid(void *buf, int data) {
   volatile int len;
-  // CHECK: store i8* [[BUF]], i8** [[BUF_ADDR:%.*]], align 8
-  // CHECK: store i32 [[DATA]], i32* [[DATA_ADDR:%.*]]
+  // CHECK: %[[BUF_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[DATA_ADDR:.*]] = alloca i32, align 4
+  // CHECK: %[[LEN:.*]] = alloca i32, align 4
+  // CHECK: store i8* %[[BUF]], i8** %[[BUF_ADDR]], align 8
+  // CHECK: store i32 %[[DATA]], i32* %[[DATA_ADDR]], align 4
 
-  // CHECK: store volatile i32 8,
+  // CHECK: store volatile i32 8, i32* %[[LEN]], align 4
   len = __builtin_os_log_format_buffer_size("invalid specifier %: %d even a trailing one%", data);
 
-  // CHECK: [[BUF2:%.*]] = load i8*, i8** [[BUF_ADDR]]
-  // CHECK: [[SUMMARY:%.*]] = getelementptr i8, i8* [[BUF2]], i64 0
-  // CHECK: store i8 0, i8* [[SUMMARY]]
-  // CHECK: [[NUM_ARGS:%.*]] = getelementptr i8, i8* [[BUF2]], i64 1
-  // CHECK: store i8 1, i8* [[NUM_ARGS]]
-
-  // CHECK: [[ARG1_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 2
-  // CHECK: store i8 0, i8* [[ARG1_DESC]]
-  // CHECK: [[ARG1_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 3
-  // CHECK: store i8 4, i8* [[ARG1_SIZE]]
-  // CHECK: [[ARG1:%.*]] = getelementptr i8, i8* [[BUF2]], i64 4
-  // CHECK: [[ARG1_INT:%.*]] = bitcast i8* [[ARG1]] to i32*
-  // CHECK: [[ARG1_VAL:%.*]] = load i32, i32* [[DATA_ADDR]]
-  // CHECK: store i32 [[ARG1_VAL]], i32* [[ARG1_INT]]
+  // CHECK: %[[V1:.*]] = load i8*, i8** %[[BUF_ADDR]], align 8
+  // CHECK: %[[V2:.*]] = load i32, i32* %[[DATA_ADDR]], align 4
+  // CHECK: call void @__os_log_helper_1_0_1_4_0(i8* %[[V1]], i32 %[[V2]])
 
   __builtin_os_log_format(buf, "invalid specifier %: %d even a trailing one%", data);
 }
 
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_1_0_1_4_0
+// CHECK: (i8* %[[BUFFER:.*]], i32 %[[ARG0:.*]])
+
+// CHECK: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK: %[[ARG0_ADDR:.*]] = alloca i32, align 4
+// CHECK: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK: store i32 %[[ARG0]], i32* %[[ARG0_ADDR]], align 4
+// CHECK: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK: store i8 0, i8* %[[SUMMARY]], align 1
+// CHECK: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK: store i8 1, i8* %[[NUMARGS]], align 1
+// CHECK: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK: store i8 0, i8* %[[ARGDESCRIPTOR]], align 1
+// CHECK: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK: store i8 4, i8* %[[ARGSIZE]], align 1
+// CHECK: %[[ARGDATA:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+// CHECK: %[[ARGDATACAST:.*]] = bitcast i8* %[[ARGDATA]] to i32*
+// CHECK: %[[V0:.*]] = load i32, i32* %[[ARG0_ADDR]], align 4
+// CHECK: store i32 %[[V0]], i32* %[[ARGDATACAST]], align 1
+
 // CHECK-LABEL: define void @test_builtin_os_log_percent
-// CHECK: (i8* [[BUF:%.*]], i8* [[DATA1:%.*]], i8* [[DATA2:%.*]])
+// CHECK: (i8* %[[BUF:.*]], i8* %[[DATA1:.*]], i8* %[[DATA2:.*]])
 // Check that the %% which does not consume any argument is correctly handled
 void test_builtin_os_log_percent(void *buf, const char *data1, const char *data2) {
   volatile int len;
-  // CHECK: store i8* [[BUF]], i8** [[BUF_ADDR:%.*]], align 8
-  // CHECK: store i8* [[DATA1]], i8** [[DATA1_ADDR:%.*]], align 8
-  // CHECK: store i8* [[DATA2]], i8** [[DATA2_ADDR:%.*]], align 8
-  // CHECK: store volatile i32 22
+  // CHECK: %[[BUF_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[DATA1_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[DATA2_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[LEN:.*]] = alloca i32, align 4
+  // CHECK: store i8* %[[BUF]], i8** %[[BUF_ADDR]], align 8
+  // CHECK: store i8* %[[DATA1]], i8** %[[DATA1_ADDR]], align 8
+  // CHECK: store i8* %[[DATA2]], i8** %[[DATA2_ADDR]], align 8
+  // CHECK: store volatile i32 22, i32* %[[LEN]], align 4
+
   len = __builtin_os_log_format_buffer_size("%s %% %s", data1, data2);
 
-  // CHECK: [[BUF2:%.*]] = load i8*, i8** [[BUF_ADDR]]
-  // CHECK: [[SUMMARY:%.*]] = getelementptr i8, i8* [[BUF2]], i64 0
-  // CHECK: store i8 2, i8* [[SUMMARY]]
-  // CHECK: [[NUM_ARGS:%.*]] = getelementptr i8, i8* [[BUF2]], i64 1
-  // CHECK: store i8 2, i8* [[NUM_ARGS]]
-  //
-  // CHECK: [[ARG1_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 2
-  // CHECK: store i8 32, i8* [[ARG1_DESC]]
-  // CHECK: [[ARG1_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 3
-  // CHECK: store i8 8, i8* [[ARG1_SIZE]]
-  // CHECK: [[ARG1:%.*]] = getelementptr i8, i8* [[BUF2]], i64 4
-  // CHECK: [[ARG1_PTR:%.*]] = bitcast i8* [[ARG1]] to i8**
-  // CHECK: [[DATA1:%.*]] = load i8*, i8** [[DATA1_ADDR]]
-  // CHECK: store i8* [[DATA1]], i8** [[ARG1_PTR]]
-  //
-  // CHECK: [[ARG2_DESC:%.*]] = getelementptr i8, i8* [[BUF2]], i64 12
-  // CHECK: store i8 32, i8* [[ARG2_DESC]]
-  // CHECK: [[ARG2_SIZE:%.*]] = getelementptr i8, i8* [[BUF2]], i64 13
-  // CHECK: store i8 8, i8* [[ARG2_SIZE]]
-  // CHECK: [[ARG2:%.*]] = getelementptr i8, i8* [[BUF2]], i64 14
-  // CHECK: [[ARG2_PTR:%.*]] = bitcast i8* [[ARG2]] to i8**
-  // CHECK: [[DATA2:%.*]] = load i8*, i8** [[DATA2_ADDR]]
-  // CHECK: store i8* [[DATA2]], i8** [[ARG2_PTR]]
+  // CHECK: %[[V1:.*]] = load i8*, i8** %[[BUF_ADDR]], align 8
+  // CHECK: %[[V2:.*]] = load i8*, i8** %[[DATA1_ADDR]], align 8
+  // CHECK: %[[V3:.*]] = ptrtoint i8* %[[V2]] to i64
+  // CHECK: %[[V4:.*]] = load i8*, i8** %[[DATA2_ADDR]], align 8
+  // CHECK: %[[V5:.*]] = ptrtoint i8* %[[V4]] to i64
+  // CHECK: call void @__os_log_helper_1_2_2_8_32_8_32(i8* %[[V1]], i64 %[[V3]], i64 %[[V5]])
+
   __builtin_os_log_format(buf, "%s %% %s", data1, data2);
 }
 
-#endif
\ No newline at end of file
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_1_2_2_8_32_8_32
+// CHECK: (i8* %[[BUFFER:.*]], i64 %[[ARG0:.*]], i64 %[[ARG1:.*]])
+
+// CHECK: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK: %[[ARG0_ADDR:.*]] = alloca i64, align 8
+// CHECK: %[[ARG1_ADDR:.*]] = alloca i64, align 8
+// CHECK: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK: store i64 %[[ARG0]], i64* %[[ARG0_ADDR]], align 8
+// CHECK: store i64 %[[ARG1]], i64* %[[ARG1_ADDR]], align 8
+// CHECK: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK: store i8 2, i8* %[[SUMMARY]], align 1
+// CHECK: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK: store i8 2, i8* %[[NUMARGS]], align 1
+// CHECK: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK: store i8 32, i8* %[[ARGDESCRIPTOR]], align 1
+// CHECK: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK: store i8 8, i8* %[[ARGSIZE]], align 1
+// CHECK: %[[ARGDATA:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+// CHECK: %[[ARGDATACAST:.*]] = bitcast i8* %[[ARGDATA]] to i64*
+// CHECK: %[[V0:.*]] = load i64, i64* %[[ARG0_ADDR]], align 8
+// CHECK: store i64 %[[V0]], i64* %[[ARGDATACAST]], align 1
+// CHECK: %[[ARGDESCRIPTOR1:.*]] = getelementptr i8, i8* %[[BUF]], i64 12
+// CHECK: store i8 32, i8* %[[ARGDESCRIPTOR1]], align 1
+// CHECK: %[[ARGSIZE2:.*]] = getelementptr i8, i8* %[[BUF]], i64 13
+// CHECK: store i8 8, i8* %[[ARGSIZE2]], align 1
+// CHECK: %[[ARGDATA3:.*]] = getelementptr i8, i8* %[[BUF]], i64 14
+// CHECK: %[[ARGDATACAST4:.*]] = bitcast i8* %[[ARGDATA3]] to i64*
+// CHECK: %[[V1:.*]] = load i64, i64* %[[ARG1_ADDR]], align 8
+// CHECK: store i64 %[[V1]], i64* %[[ARGDATACAST4]], align 1
+
+// Check that the following two functions call the same helper function.
+
+// CHECK-LABEL: define void @test_builtin_os_log_merge_helper0
+// CHECK: call void @__os_log_helper_1_0_2_4_0_8_0(
+void test_builtin_os_log_merge_helper0(void *buf, int i, double d) {
+  __builtin_os_log_format(buf, "%d %f", i, d);
+}
+
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_1_0_2_4_0_8_0(
+
+// CHECK-LABEL: define void @test_builtin_os_log_merge_helper1
+// CHECK: call void @__os_log_helper_1_0_2_4_0_8_0(
+void test_builtin_os_log_merge_helper1(void *buf, unsigned u, long long ll) {
+  __builtin_os_log_format(buf, "%u %lld", u, ll);
+}
+
+// Check that this function doesn't write past the end of array 'buf'.
+
+// CHECK-LABEL: define void @test_builtin_os_log_errno
+void test_builtin_os_log_errno() {
+  // CHECK: %[[VLA:.*]] = alloca i8, i64 4, align 16
+  // CHECK: call void @__os_log_helper_16_2_1_0_96(i8* %[[VLA]])
+
+  char buf[__builtin_os_log_format_buffer_size("%m")];
+  __builtin_os_log_format(buf, "%m");
+}
+
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_16_2_1_0_96
+// CHECK: (i8* %[[BUFFER:.*]])
+
+// CHECK: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK: store i8 2, i8* %[[SUMMARY]], align 16
+// CHECK: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK: store i8 1, i8* %[[NUMARGS]], align 1
+// CHECK: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK: store i8 96, i8* %[[ARGDESCRIPTOR]], align 2
+// CHECK: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK: store i8 0, i8* %[[ARGSIZE]], align 1
+// CHECK-NEXT: ret void
+
+// CHECK-LABEL: define void @test_builtin_os_log_long_double
+// CHECK: (i8* %[[BUF:.*]], x86_fp80 %[[LD:.*]])
+void test_builtin_os_log_long_double(void *buf, long double ld) {
+  // CHECK: %[[BUF_ADDR:.*]] = alloca i8*, align 8
+  // CHECK: %[[LD_ADDR:.*]] = alloca x86_fp80, align 16
+  // CHECK: %[[COERCE:.*]] = alloca i128, align 16
+  // CHECK: store i8* %[[BUF]], i8** %[[BUF_ADDR]], align 8
+  // CHECK: store x86_fp80 %[[LD]], x86_fp80* %[[LD_ADDR]], align 16
+  // CHECK: %[[V0:.*]] = load i8*, i8** %[[BUF_ADDR]], align 8
+  // CHECK: %[[V1:.*]] = load x86_fp80, x86_fp80* %[[LD_ADDR]], align 16
+  // CHECK: %[[V2:.*]] = bitcast x86_fp80 %[[V1]] to i80
+  // CHECK: %[[V3:.*]] = zext i80 %[[V2]] to i128
+  // CHECK: store i128 %[[V3]], i128* %[[COERCE]], align 16
+  // CHECK: %[[V4:.*]] = bitcast i128* %[[COERCE]] to { i64, i64 }*
+  // CHECK: %[[V5:.*]] = getelementptr inbounds { i64, i64 }, { i64, i64 }* %[[V4]], i32 0, i32 0
+  // CHECK: %[[V6:.*]] = load i64, i64* %[[V5]], align 16
+  // CHECK: %[[V7:.*]] = getelementptr inbounds { i64, i64 }, { i64, i64 }* %[[V4]], i32 0, i32 1
+  // CHECK: %[[V8:.*]] = load i64, i64* %[[V7]], align 8
+  // CHECK: call void @__os_log_helper_1_0_1_16_0(i8* %[[V0]], i64 %[[V6]], i64 %[[V8]])
+
+  __builtin_os_log_format(buf, "%Lf", ld);
+}
+
+// CHECK-LABEL: define linkonce_odr hidden void @__os_log_helper_1_0_1_16_0
+// CHECK: (i8* %[[BUFFER:.*]], i64 %[[ARG0_COERCE0:.*]], i64 %[[ARG0_COERCE1:.*]])
+
+// CHECK: %[[ARG0:.*]] = alloca i128, align 16
+// CHECK: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK: %[[ARG0_ADDR:.*]] = alloca i128, align 16
+// CHECK: %[[V0:.*]] = bitcast i128* %[[ARG0]] to { i64, i64 }*
+// CHECK: %[[V1:.*]] = getelementptr inbounds { i64, i64 }, { i64, i64 }* %[[V0]], i32 0, i32 0
+// CHECK: store i64 %[[ARG0_COERCE0]], i64* %[[V1]], align 16
+// CHECK: %[[V2:.*]] = getelementptr inbounds { i64, i64 }, { i64, i64 }* %[[V0]], i32 0, i32 1
+// CHECK: store i64 %[[ARG0_COERCE1]], i64* %[[V2]], align 8
+// CHECK: %[[ARG01:.*]] = load i128, i128* %[[ARG0]], align 16
+// CHECK: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK: store i128 %[[ARG01]], i128* %[[ARG0_ADDR]], align 16
+// CHECK: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK: store i8 0, i8* %[[SUMMARY]], align 1
+// CHECK: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK: store i8 1, i8* %[[NUMARGS]], align 1
+// CHECK: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK: store i8 0, i8* %[[ARGDESCRIPTOR]], align 1
+// CHECK: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK: store i8 16, i8* %[[ARGSIZE]], align 1
+// CHECK: %[[ARGDATA:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+// CHECK: %[[ARGDATACAST:.*]] = bitcast i8* %[[ARGDATA]] to i128*
+// CHECK: %[[V3:.*]] = load i128, i128* %[[ARG0_ADDR]], align 16
+// CHECK: store i128 %[[V3]], i128* %[[ARGDATACAST]], align 1
+
+#endif
diff --git a/test/CodeGenObjC/os_log.m b/test/CodeGenObjC/os_log.m
index 5d48783de4..7b8398a375 100644
--- a/test/CodeGenObjC/os_log.m
+++ b/test/CodeGenObjC/os_log.m
@@ -1,4 +1,5 @@
 // RUN: %clang_cc1 %s -emit-llvm -o - -triple x86_64-darwin-apple -fobjc-arc -O2 | FileCheck %s
+// RUN: %clang_cc1 %s -emit-llvm -o - -triple x86_64-darwin-apple -fobjc-arc -O0 | FileCheck %s -check-prefix=CHECK-O0
 
 // Make sure we emit clang.arc.use before calling objc_release as part of the
 // cleanup. This way we make sure the object will not be released until the
@@ -12,28 +13,67 @@ extern __attribute__((visibility("default"))) NSString *GenString();
 // Behavior of __builtin_os_log differs between platforms, so only test on X86
 #ifdef __x86_64__
 // CHECK-LABEL: define i8* @test_builtin_os_log
+// CHECK-O0-LABEL: define i8* @test_builtin_os_log
+// CHECK: (i8* returned %[[BUF:.*]])
+// CHECK-O0: (i8* %[[BUF:.*]])
 void *test_builtin_os_log(void *buf) {
   return __builtin_os_log_format(buf, "capabilities: %@", GenString());
 
-  // CHECK: store i8 2, i8*
-  // CHECK: [[NUM_ARGS:%.*]] = getelementptr i8, i8* {{.*}}, i64 1
-  // CHECK: store i8 1, i8* [[NUM_ARGS]]
-  //
-  // CHECK: [[ARG1_DESC:%.*]] = getelementptr i8, i8* {{.*}}, i64 2
-  // CHECK: store i8 64, i8* [[ARG1_DESC]]
-  // CHECK: [[ARG1_SIZE:%.*]] = getelementptr i8, i8* {{.*}}, i64 3
-  // CHECK: store i8 8, i8* [[ARG1_SIZE]]
-  // CHECK: [[ARG1:%.*]] = getelementptr i8, i8* {{.*}}, i64 4
-  // CHECK: [[ARG1_CAST:%.*]] = bitcast i8* [[ARG1]] to
-
-  // CHECK: [[STRING:%.*]] = {{.*}} call {{.*}} @GenString()
-  // CHECK: [[STRING_CAST:%.*]] = bitcast {{.*}} [[STRING]] to
-  // CHECK: call {{.*}} @objc_retainAutoreleasedReturnValue(i8* [[STRING_CAST]])
-  // CHECK: store {{.*}} [[STRING]], {{.*}} [[ARG1_CAST]]
-
-  // CHECK: call void (...) @clang.arc.use({{.*}} [[STRING]])
-  // CHECK: call void @objc_release(i8* [[STRING_CAST]])
-  // CHECK: ret i8*
+  // CHECK: %[[CALL:.*]] = tail call %[[V0:.*]]* (...) @GenString()
+  // CHECK: %[[V0]] = bitcast %[[V0]]* %[[CALL]] to i8*
+  // CHECK: %[[V1:.*]] = tail call i8* @objc_retainAutoreleasedReturnValue(i8* %[[V0]])
+  // CHECK: %[[V2:.*]] = ptrtoint %[[V0]]* %[[CALL]] to i64
+  // CHECK: store i8 2, i8* %[[BUF]], align 1
+  // CHECK: %[[NUMARGS_I:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+  // CHECK: store i8 1, i8* %[[NUMARGS_I]], align 1
+  // CHECK: %[[ARGDESCRIPTOR_I:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+  // CHECK: store i8 64, i8* %[[ARGDESCRIPTOR_I]], align 1
+  // CHECK: %[[ARGSIZE_I:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+  // CHECK: store i8 8, i8* %[[ARGSIZE_I]], align 1
+  // CHECK: %[[ARGDATA_I:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+  // CHECK: %[[ARGDATACAST_I:.*]] = bitcast i8* %[[ARGDATA_I]] to i64*
+  // CHECK: store i64 %[[V2]], i64* %[[ARGDATACAST_I]], align 1
+  // CHECK: tail call void (...) @clang.arc.use(%[[V0]]* %[[CALL]])
+  // CHECK: tail call void @objc_release(i8* %[[V0]])
+  // CHECK: ret i8* %[[BUF]]
+
+  // clang.arc.use is used and removed in IR optimizations. At O0, we should not
+  // emit clang.arc.use, since it will not be removed and we will have a link
+  // error.
+  // CHECK-O0: %[[BUF_ADDR:.*]] = alloca i8*, align 8
+  // CHECK-O0: store i8* %[[BUF]], i8** %[[BUF_ADDR]], align 8
+  // CHECK-O0: %[[V0:.*]] = load i8*, i8** %[[BUF_ADDR]], align 8
+  // CHECK-O0: %[[CALL:.*]] = call %[[V0]]* (...) @GenString()
+  // CHECK-O0: %[[V1:.*]] = bitcast %[[V0]]* %[[CALL]] to i8*
+  // CHECK-O0: %[[V2:.*]] = call i8* @objc_retainAutoreleasedReturnValue(i8* %[[V1]])
+  // CHECK-O0: %[[V3:.*]] = bitcast i8* %[[V2]] to %[[V0]]*
+  // CHECK-O0: %[[V4:.*]] = ptrtoint %[[V0]]* %[[V3]] to i64
+  // CHECK-O0: call void @__os_log_helper_1_2_1_8_64(i8* %[[V0]], i64 %[[V4]])
+  // CHECK-O0: %[[V5:.*]] = bitcast %[[V0]]* %[[V3]] to i8*
+  // CHECK-O0-NOT call void (...) @clang.arc.use({{.*}}
+  // CHECK-O0: call void @objc_release(i8* %[[V5]])
+  // CHECK-O0: ret i8* %[[V0]]
 }
 
+// CHECK-O0-LABEL: define linkonce_odr hidden void @__os_log_helper_1_2_1_8_64
+// CHECK-O0: (i8* %[[BUFFER:.*]], i64 %[[ARG0:.*]])
+
+// CHECK-O0: %[[BUFFER_ADDR:.*]] = alloca i8*, align 8
+// CHECK-O0: %[[ARG0_ADDR:.*]] = alloca i64, align 8
+// CHECK-O0: store i8* %[[BUFFER]], i8** %[[BUFFER_ADDR]], align 8
+// CHECK-O0: store i64 %[[ARG0]], i64* %[[ARG0_ADDR]], align 8
+// CHECK-O0: %[[BUF:.*]] = load i8*, i8** %[[BUFFER_ADDR]], align 8
+// CHECK-O0: %[[SUMMARY:.*]] = getelementptr i8, i8* %[[BUF]], i64 0
+// CHECK-O0: store i8 2, i8* %[[SUMMARY]], align 1
+// CHECK-O0: %[[NUMARGS:.*]] = getelementptr i8, i8* %[[BUF]], i64 1
+// CHECK-O0: store i8 1, i8* %[[NUMARGS]], align 1
+// CHECK-O0: %[[ARGDESCRIPTOR:.*]] = getelementptr i8, i8* %[[BUF]], i64 2
+// CHECK-O0: store i8 64, i8* %[[ARGDESCRIPTOR]], align 1
+// CHECK-O0: %[[ARGSIZE:.*]] = getelementptr i8, i8* %[[BUF]], i64 3
+// CHECK-O0: store i8 8, i8* %[[ARGSIZE]], align 1
+// CHECK-O0: %[[ARGDATA:.*]] = getelementptr i8, i8* %[[BUF]], i64 4
+// CHECK-O0: %[[ARGDATACAST:.*]] = bitcast i8* %[[ARGDATA]] to i64*
+// CHECK-O0: %[[V0:.*]] = load i64, i64* %[[ARG0_ADDR]], align 8
+// CHECK-O0: store i64 %[[V0]], i64* %[[ARGDATACAST]], align 1
+
 #endif