[Clang] Add a flag to include GPU startup files (#112025)

Summary:
The C library for GPUs provides the ability to target regular C/C++
programs by providing the C library and a file containing kernels that
call the `main` function. This is mostly used for unit tests, this patch
provides a quick way to add them without needing to know the paths. I
currently do this explicitly, but according to the libc++ contributors
we don't want to need to specify these paths manually. See the
discussion in llvm/llvm-project#104515.

I just default to `lib/` if the target-specific one isn't found because
the linker will handle giving a reasonable error message if it's not
found. Basically the use-case looks like this.

```console
$ clang test.c --target=amdgcn-amd-amdhsa -mcpu=native -startfiles -stdlib
$ amdhsa-loader a.out
PASS!
```

Author: jhuber6

clang/include/clang/Driver/Options.td

@@ -5633,6 +5633,7 @@ def noprebind : Flag<["-"], "noprebind">;
 def noprofilelib : Flag<["-"], "noprofilelib">;
 def noseglinkedit : Flag<["-"], "noseglinkedit">;
 def nostartfiles : Flag<["-"], "nostartfiles">, Group<Link_Group>;
+def startfiles : Flag<["-"], "startfiles">, Group<Link_Group>;
 def nostdinc : Flag<["-"], "nostdinc">,
   Visibility<[ClangOption, CLOption, DXCOption]>, Group<IncludePath_Group>,
   HelpText<"Disable both standard system #include directories and builtin #include directories">;
@@ -5645,6 +5646,9 @@ def nostdincxx : Flag<["-"], "nostdinc++">, Visibility<[ClangOption, CC1Option]>
 def nostdlib : Flag<["-"], "nostdlib">,
   Visibility<[ClangOption, CLOption, FlangOption, DXCOption]>,
   Group<Link_Group>;
+def stdlib : Flag<["-"], "stdlib">,
+  Visibility<[ClangOption, CLOption, FlangOption, DXCOption]>,
+  Group<Link_Group>;
 def nostdlibxx : Flag<["-"], "nostdlib++">;
 def object : Flag<["-"], "object">;
 def o : JoinedOrSeparate<["-"], "o">,

clang/lib/Driver/ToolChains/AMDGPU.cpp

@@ -648,6 +648,17 @@ void amdgpu::Linker::ConstructJob(Compilation &C, const JobAction &JA,
         Args.MakeArgString("-plugin-opt=-mattr=" + llvm::join(Features, ",")));
   }
 
+  if (Args.hasArg(options::OPT_stdlib))
+    CmdArgs.append({"-lc", "-lm"});
+  if (Args.hasArg(options::OPT_startfiles)) {
+    std::optional<std::string> IncludePath = getToolChain().getStdlibPath();
+    if (!IncludePath)
+      IncludePath = "/lib";
+    SmallString<128> P(*IncludePath);
+    llvm::sys::path::append(P, "crt1.o");
+    CmdArgs.push_back(Args.MakeArgString(P));
+  }
+
   CmdArgs.push_back("-o");
   CmdArgs.push_back(Output.getFilename());
   C.addCommand(std::make_unique<Command>(

clang/lib/Driver/ToolChains/Cuda.cpp

@@ -643,6 +643,17 @@ void NVPTX::Linker::ConstructJob(Compilation &C, const JobAction &JA,
   llvm::sys::path::append(DefaultLibPath, CLANG_INSTALL_LIBDIR_BASENAME);
   CmdArgs.push_back(Args.MakeArgString(Twine("-L") + DefaultLibPath));
 
+  if (Args.hasArg(options::OPT_stdlib))
+    CmdArgs.append({"-lc", "-lm"});
+  if (Args.hasArg(options::OPT_startfiles)) {
+    std::optional<std::string> IncludePath = getToolChain().getStdlibPath();
+    if (!IncludePath)
+      IncludePath = "/lib";
+    SmallString<128> P(*IncludePath);
+    llvm::sys::path::append(P, "crt1.o");
+    CmdArgs.push_back(Args.MakeArgString(P));
+  }
+
   C.addCommand(std::make_unique<Command>(
       JA, *this,
       ResponseFileSupport{ResponseFileSupport::RF_Full, llvm::sys::WEM_UTF8,

clang/test/Driver/amdgpu-toolchain.c

@@ -32,3 +32,7 @@
 // RUN: %clang -### --target=amdgcn-amd-amdhsa -mcpu=gfx906 -nogpulib \
 // RUN:   -r %s 2>&1 | FileCheck -check-prefixes=RELO %s
 // RELO-NOT: -shared
+
+// RUN: %clang -target amdgcn-amd-amdhsa -march=gfx90a -stdlib -startfiles \
+// RUN:   -nogpulib -nogpuinc -### %s 2>&1 | FileCheck -check-prefix=STARTUP %s
+// STARTUP: ld.lld{{.*}}"-lc" "-lm" "{{.*}}crt1.o"

clang/test/Driver/cuda-cross-compiling.c

@@ -105,3 +105,11 @@
 // RUN:   | FileCheck -check-prefix=FEATURE %s
 
 // FEATURE: clang-nvlink-wrapper{{.*}}"--plugin-opt=-mattr=+ptx63"
+
+//
+// Test including the libc startup files and libc
+//
+// RUN: %clang -target nvptx64-nvidia-cuda -march=sm_61 -stdlib -startfiles \
+// RUN:   -nogpulib -nogpuinc -### %s 2>&1 | FileCheck -check-prefix=STARTUP %s
+
+// STARTUP: clang-nvlink-wrapper{{.*}}"-lc" "-lm" "{{.*}}crt1.o"