[mlir][gpu][nvptx] Remove null terminator when outputting PTX

mlir/lib/Target/LLVM/NVVM/Target.cpp

@@ -722,12 +722,8 @@ NVPTXSerializer::moduleToObject(llvm::Module &llvmModule) {
 #undef DEBUG_TYPE
 
   // Return PTX if the compilation target is `assembly`.
-  if (targetOptions.getCompilationTarget() ==
-      gpu::CompilationTarget::Assembly) {
-    // Make sure to include the null terminator.
-    StringRef bin(serializedISA->c_str(), serializedISA->size() + 1);
-    return SmallVector<char, 0>(bin.begin(), bin.end());
-  }
+  if (targetOptions.getCompilationTarget() == gpu::CompilationTarget::Assembly)
+    return SmallVector<char, 0>(serializedISA->begin(), serializedISA->end());
 
   std::optional<SmallVector<char, 0>> result;
   moduleToObjectTimer.startTimer();

mlir/lib/Target/LLVMIR/Dialect/GPU/SelectObjectAttr.cpp

@@ -116,8 +116,11 @@ LogicalResult SelectObjectAttrImpl::embedBinary(
   llvm::Module *module = moduleTranslation.getLLVMModule();
 
   // Embed the object as a global string.
+  // Add null for assembly output for JIT paths that expect null-terminated
+  // strings.
+  bool addNull = (object.getFormat() == gpu::CompilationTarget::Assembly);
   llvm::Constant *binary = llvm::ConstantDataArray::getString(
-      builder.getContext(), object.getObject().getValue(), false);
+      builder.getContext(), object.getObject().getValue(), addNull);
   llvm::GlobalVariable *serializedObj =
       new llvm::GlobalVariable(*module, binary->getType(), true,
                                llvm::GlobalValue::LinkageTypes::InternalLinkage,

mlir/unittests/Target/LLVM/SerializeNVVMTarget.cpp

@@ -130,6 +130,7 @@ TEST_F(MLIRTargetLLVMNVVM, SKIP_WITHOUT_NVPTX(SerializeNVVMToPTX)) {
 
     ASSERT_TRUE(
         StringRef(object->data(), object->size()).contains("nvvm_kernel"));
+    ASSERT_TRUE(StringRef(object->data(), object->size()).count('\0') == 0);
   }
 }