[CUDA] Place GPU binary into .nv_fatbin section and align it by 8.

This matches the way nvcc encapsulates GPU binaries into host object file.
Now cuobjdump can deal with clang-compiled object files.

Differential Revision: https://reviews.llvm.org/D23429

llvm-svn: 278549

diff --git a/clang/lib/CodeGen/CGCUDANV.cpp b/clang/lib/CodeGen/CGCUDANV.cpp
index 3e1f2b5..dacc53b 100644 (file)
--- a/clang/lib/CodeGen/CGCUDANV.cpp
+++ b/clang/lib/CodeGen/CGCUDANV.cpp
@@ -55,10 +55,18 @@ private:
   /// where the C code specifies const char*.
   llvm::Constant *makeConstantString(const std::string &Str,
                                      const std::string &Name = "",
+                                     const std::string &SectionName = "",
                                      unsigned Alignment = 0) {
     llvm::Constant *Zeros[] = {llvm::ConstantInt::get(SizeTy, 0),
                                llvm::ConstantInt::get(SizeTy, 0)};
     auto ConstStr = CGM.GetAddrOfConstantCString(Str, Name.c_str());
+    llvm::GlobalVariable *GV =
+        cast<llvm::GlobalVariable>(ConstStr.getPointer());
+    if (!SectionName.empty())
+      GV->setSection(SectionName);
+    if (Alignment)
+      GV->setAlignment(Alignment);
+
     return llvm::ConstantExpr::getGetElementPtr(ConstStr.getElementType(),
                                                 ConstStr.getPointer(), Zeros);
  }
@@ -285,7 +293,8 @@ llvm::Function *CGNVCUDARuntime::makeModuleCtorFunction() {
     llvm::Constant *Values[] = {
         llvm::ConstantInt::get(IntTy, 0x466243b1), // Fatbin wrapper magic.
         llvm::ConstantInt::get(IntTy, 1),          // Fatbin version.
-        makeConstantString(GpuBinaryOrErr.get()->getBuffer(), "", 16), // Data.
+        makeConstantString(GpuBinaryOrErr.get()->getBuffer(), // Data.
+                           "", ".nv_fatbin", 8),              //
         llvm::ConstantPointerNull::get(VoidPtrTy)}; // Unused in fatbin v1.
     llvm::GlobalVariable *FatbinWrapper = new llvm::GlobalVariable(
         TheModule, FatbinWrapperTy, true, llvm::GlobalValue::InternalLinkage,

diff --git a/clang/test/CodeGenCUDA/device-stub.cu b/clang/test/CodeGenCUDA/device-stub.cu
index 5979ba3..3376803 100644 (file)
--- a/clang/test/CodeGenCUDA/device-stub.cu
+++ b/clang/test/CodeGenCUDA/device-stub.cu
@@ -45,10 +45,12 @@ void use_pointers() {
 // * constant unnamed string with the kernel name
 // CHECK: private unnamed_addr constant{{.*}}kernelfunc{{.*}}\00"
 // * constant unnamed string with GPU binary
-// CHECK: private unnamed_addr constant{{.*}}\00"
+// CHECK: private unnamed_addr constant{{.*GPU binary would be here.*}}\00"
+// CHECK-SAME: section ".nv_fatbin", align 8
 // * constant struct that wraps GPU binary
 // CHECK: @__cuda_fatbin_wrapper = internal constant { i32, i32, i8*, i8* } 
-// CHECK:       { i32 1180844977, i32 1, {{.*}}, i8* null }
+// CHECK-SAME: { i32 1180844977, i32 1, {{.*}}, i8* null }
+// CHECK-SAME: section ".nvFatBinSegment"
 // * variable to save GPU binary handle after initialization
 // CHECK: @__cuda_gpubin_handle = internal global i8** null
 // * Make sure our constructor/destructor was added to global ctor/dtor list.