[DEBUGINFO, NVPTX] Render `-no-cuda-debug` LLVM option when required.

[ABataev]

Summary

When emission of the lineinfo is requested for the NVPTX, render the
LLVM -no-cuda-debug option to disable emission of the debug option
in the .target directive. Required for the correct work of the ptxas
tool, which does not allow emission of the optimized code in presence of
the debug option.

[echristo]

So, I'd really prefer not to set options via the backend option path. From here I think we should aim to take all of the options we added and having the asm printer in the backend know how to set them depending on the target. We could also add things to the IR metadata if necessary, but I'd like to avoid that if possible.

Thoughts?

[ABataev]

In D45783#1106019, @echristo wrote:

So, I'd really prefer not to set options via the backend option path. From here I think we should aim to take all of the options we added and having the asm printer in the backend know how to set them depending on the target. We could also add things to the IR metadata if necessary, but I'd like to avoid that if possible.

Thoughts?

This patch is outdated a bit, I have a patch https://reviews.llvm.org/D46061 that does it in a different way, via MCOptionю This is only a backend patch. I planned to update this patch after D46061 is accepted.

[ABataev]

The patch is not required anymore as it is implemented in the different way

lib/Driver/ToolChains/Cuda.cpp

 enum DebugInfoKind {
   NoDebug,       /// No debug info.
   LineTableOnly, /// Line tables only.
   FullDebug      /// Full debug info.
 };
 } // anonymous namespace
 
 static DebugInfoKind mustEmitDebugInfo(const ArgList &Args) {
-  Arg *A = Args.getLastArg(options::OPT_O_Group);
-  if (Args.hasFlag(options::OPT_cuda_noopt_device_debug,
-                   options::OPT_no_cuda_noopt_device_debug,
-                   !A || A->getOption().matches(options::OPT_O0))) {
-    if (const Arg *A = Args.getLastArg(options::OPT_g_Group)) {
-      const Option &Opt = A->getOption();
-      if (Opt.matches(options::OPT_gN_Group)) {
-        if (Opt.matches(options::OPT_g0) || Opt.matches(options::OPT_ggdb0))
-          return NoDebug;
-        if (Opt.matches(options::OPT_gline_tables_only) ||
-            Opt.matches(options::OPT_ggdb1))
-          return LineTableOnly;
-      }
-      return FullDebug;
-    }
-  }
-  return NoDebug;
+  const Arg *OptArg = Args.getLastArg(options::OPT_O_Group);
+  DebugInfoKind DIKind = NoDebug;
+  if (const Arg *DebugArg = Args.getLastArg(options::OPT_g_Group)) {
+    DIKind = FullDebug;
+    const Option &Opt = DebugArg->getOption();
+        DIKind = NoDebug;
+      else if (Opt.matches(options::OPT_gline_tables_only) ||
+        DIKind = LineTableOnly;
+  if (Args.hasFlag(options::OPT_cuda_noopt_device_debug,
+                   options::OPT_no_cuda_noopt_device_debug,
+                   !OptArg || OptArg->getOption().matches(options::OPT_O0)))
+    return DIKind;
+  // If --no-cuda-noopt-device-debug is provided and O>0 and debug info
+  // requested - generate lineinfo.
+  return DIKind == NoDebug ? NoDebug : LineTableOnly;
 }
 
 void NVPTX::Assembler::ConstructJob(Compilation &C, const JobAction &JA,
                                     const InputInfo &Output,
                                     const InputInfoList &Inputs,
                                     const ArgList &Args,
                                     const char *LinkingOutput) const {
   const auto &TC =
void CudaToolChain::addClangTargetOptions(
   } else {
     // Libdevice in CUDA-7.0 requires PTX version that's more recent
     // than LLVM defaults to. Use PTX4.2 which is the PTX version that
     // came with CUDA-7.0.
     CC1Args.push_back("-target-feature");
     CC1Args.push_back("+ptx42");
   }
 
+  // Disable emission of the `debug` option in the `.target` if the lineinfo is
+  // requested.
+  if (mustEmitDebugInfo(DriverArgs) == LineTableOnly) {
+    CC1Args.push_back("-mllvm");
+    CC1Args.push_back("-no-cuda-debug");
+  }
+
   if (DeviceOffloadingKind == Action::OFK_OpenMP) {
     SmallVector<StringRef, 8> LibraryPaths;
     // Add path to lib and/or lib64 folders.
     SmallString<256> DefaultLibPath =
       llvm::sys::path::parent_path(getDriver().Dir);
     llvm::sys::path::append(DefaultLibPath,
         Twine("lib") + CLANG_LIBDIR_SUFFIX);
     LibraryPaths.emplace_back(DefaultLibPath.c_str());

test/Driver/cuda-dwarf-2.cu

 // RUN:   FileCheck %s -check-prefix NO_DEBUG
 // RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -ggdb0 -O3 --cuda-noopt-device-debug 2>&1 | \
 // RUN:   FileCheck %s -check-prefix NO_DEBUG
 // RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -ggdb1 2>&1 | \
 // RUN:   FileCheck %s -check-prefix NO_DEBUG -check-prefix LINE_TABLE
 // RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -gline-tables-only -O2 --cuda-noopt-device-debug 2>&1 | \
 // RUN:   FileCheck %s -check-prefix NO_DEBUG -check-prefix LINE_TABLE
 
+// LINE_TABLE: "-triple" "nvptx64-nvidia-cuda"
+// LINE_TABLE-SAME: "-mllvm" "-no-cuda-debug"
 // NO_DEBUG: ptxas
 // NO_DEBUG-NOT: "-g"
 // LINE_TABLE: "-lineinfo"
 // NO_DEBUG: fatbinary
 // NO_DEBUG-NOT: "-g"
 
 // RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g 2>&1 | \
 // RUN:   FileCheck %s -check-prefix HAS_DEBUG

test/Driver/openmp-offload-gpu.c

 // RUN:   | FileCheck -check-prefix=NO_DEBUG %s
 // RUN:   %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb0 -O3 --cuda-noopt-device-debug 2>&1 \
 // RUN:   | FileCheck -check-prefix=NO_DEBUG %s
 // RUN:   %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -gline-tables-only 2>&1 \
 // RUN:   | FileCheck -check-prefix=NO_DEBUG -check-prefix=LINE_TABLE %s
 // RUN:   %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb1 -O2 --cuda-noopt-device-debug 2>&1 \
 // RUN:   | FileCheck -check-prefix=NO_DEBUG -check-prefix=LINE_TABLE %s
 
+// LINE_TABLE: "-triple" "nvptx64-nvidia-cuda"
+// LINE_TABLE-SAME: "-mllvm" "-no-cuda-debug"
 // NO_DEBUG: ptxas
 // LINE_TABLE: "-lineinfo"
 // NO_DEBUG-NOT: "-g"
 // NO_DEBUG: nvlink
 // NO_DEBUG-NOT: "-g"
 
 // RUN:   %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g 2>&1 \
 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s