Declares __cpu_model as dso local

[hhb]

Summary

__builtin_cpu_supports and __builtin_cpu_is use information in __cpu_model to decide cpu features. Before this change, __cpu_model was not declared as dso local. The generated code looks up the address in GOT when reading __cpu_model. This makes it impossible to use these functions in ifunc, because at that time GOT entries have not been relocated. This change makes it dso local.

[pirama]

Adding reviewers suggested by 'arc cover'.

[craig.topper]

lib/CodeGen/CGBuiltin.cpp

9101

This code looks to be out of date. It's missing the changes from r344832 that added another runtime variable that presumably has the same issue.

[hhb]

Rebase. Change to SetDSOLocal.

[hhb]

lib/CodeGen/CGBuiltin.cpp

9101

By the way, what is the reason for gcc to add a separate cpu_feature2, but not adding an element to cpu_model.__cpu_features array?

[hhb]

Ping..

[hhb]

Anyone can have a look at this change?

[srhines]

Craig, does this look ok now?

[craig.topper]

LGTM

lib/CodeGen/CGBuiltin.cpp

Value *CodeGenFunction::EmitX86CpuIs(StringRef CPUStr) {
   // unsigned int __cpu_type;
   // unsigned int __cpu_subtype;
   // unsigned int __cpu_features[1];
   llvm::Type *STy = llvm::StructType::get(Int32Ty, Int32Ty, Int32Ty,
                                           llvm::ArrayType::get(Int32Ty, 1));
 
   // Grab the global __cpu_model.
   llvm::Constant *CpuModel = CGM.CreateRuntimeVariable(STy, "__cpu_model");
+  cast<llvm::GlobalValue>(CpuModel)->setVisibility(
+      llvm::GlobalValue::HiddenVisibility);
 
   // Calculate the index needed to access the correct field based on the
   // range. Also adjust the expected value.
   unsigned Index;
   unsigned Value;
   std::tie(Index, Value) = StringSwitch<std::pair<unsigned, unsigned>>(CPUStr)
 #define X86_VENDOR(ENUM, STRING)                                               \
   .Case(STRING, {0u, static_cast<unsigned>(llvm::X86::ENUM)})
llvm::Value *CodeGenFunction::EmitX86CpuSupports(uint32_t FeaturesMask) {
   // unsigned int __cpu_type;
   // unsigned int __cpu_subtype;
   // unsigned int __cpu_features[1];
   llvm::Type *STy = llvm::StructType::get(Int32Ty, Int32Ty, Int32Ty,
                                           llvm::ArrayType::get(Int32Ty, 1));
 
   // Grab the global __cpu_model.
   llvm::Constant *CpuModel = CGM.CreateRuntimeVariable(STy, "__cpu_model");
+  cast<llvm::GlobalValue>(CpuModel)->setVisibility(
+      llvm::GlobalValue::HiddenVisibility);
 
   // Grab the first (0th) element from the field __cpu_features off of the
   // global in the struct STy.
   Value *Idxs[] = {ConstantInt::get(Int32Ty, 0), ConstantInt::get(Int32Ty, 3),
                    ConstantInt::get(Int32Ty, 0)};
   Value *CpuFeatures = Builder.CreateGEP(STy, CpuModel, Idxs);
   Value *Features =
       Builder.CreateAlignedLoad(CpuFeatures, CharUnits::fromQuantity(4));
 
   // Check the value of the bit corresponding to the feature requested.
   Value *Bitset = Builder.CreateAnd(
       Features, llvm::ConstantInt::get(Int32Ty, FeaturesMask));
   return Builder.CreateICmpNE(Bitset, llvm::ConstantInt::get(Int32Ty, 0));
 }
 
 Value *CodeGenFunction::EmitX86CpuInit() {
   llvm::FunctionType *FTy = llvm::FunctionType::get(VoidTy,
                                                     /*Variadic*/ false);
   llvm::Constant *Func = CGM.CreateRuntimeFunction(FTy, "__cpu_indicator_init");
   return Builder.CreateCall(Func);
 }

test/CodeGen/builtin-cpu-is.c

 // RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm < %s| FileCheck %s
 
 // Test that we have the structure definition, the gep offsets, the name of the
 // global, the bit grab, and the icmp correct.
 extern void a(const char *);
 
+// CHECK: @__cpu_model = external hidden global { i32, i32, i32, [1 x i32] }
+
 void intel() {
   if (__builtin_cpu_is("intel"))
     a("intel");
 
   // CHECK: [[LOAD:%[^ ]+]] = load i32, i32* getelementptr inbounds ({ i32, i32, i32, [1 x i32] }, { i32, i32, i32, [1 x i32] }* @__cpu_model, i32 0, i32 0)
   // CHECK: = icmp eq i32 [[LOAD]], 1
 }
 

test/CodeGen/builtin-cpu-supports.c

 // RUN: %clang_cc1 -triple x86_64-pc-linux-gnu -emit-llvm < %s| FileCheck %s
 
 // Test that we have the structure definition, the gep offsets, the name of the
 // global, the bit grab, and the icmp correct.
 extern void a(const char *);
 
+// CHECK: @__cpu_model = external hidden global { i32, i32, i32, [1 x i32] }
+
 int main() {
   __builtin_cpu_init();
 
   // CHECK: call void @__cpu_indicator_init
 
   if (__builtin_cpu_supports("sse4.2"))
     a("sse4.2");
 
   // CHECK: [[LOAD:%[^ ]+]] = load i32, i32* getelementptr inbounds ({ i32, i32, i32, [1 x i32] }, { i32, i32, i32, [1 x i32] }* @__cpu_model, i32 0, i32 3, i32 0)
   // CHECK: [[AND:%[^ ]+]] = and i32 [[LOAD]], 256
   // CHECK: = icmp ne i32 [[AND]], 0
 
   return 0;
 }