Implement -frecord-command-line (-frecord-gcc-switches)

[scott.linder]

Summary

See the parent revision https://reviews.llvm.org/D54487 for differences between this implementation and the equivalent GCC option.

[troyj]

I realize that you're probably striving for option compatibility with gcc, but continuing to name it -frecord-gcc-switches when it actually records Clang switches seems weird to me. It almost sounds like something that would dump gcc equivalents of all Clang options, or maybe let you know which Clang options you've used that match gcc options. Either way, by the name -- if you aren't familiar with the gcc option -- it doesn't read like it records Clang options.

Would it be that bad to name it -frecord-clang-switches? Or just -frecord-switches?

[scott.linder]

In D54489#1297504, @troyj wrote:

I realize that you're probably striving for option compatibility with gcc, but continuing to name it -frecord-gcc-switches when it actually records Clang switches seems weird to me. It almost sounds like something that would dump gcc equivalents of all Clang options, or maybe let you know which Clang options you've used that match gcc options. Either way, by the name -- if you aren't familiar with the gcc option -- it doesn't read like it records Clang options.

Would it be that bad to name it -frecord-clang-switches? Or just -frecord-switches?

I agree, and this was my original plan, but then I noticed that Clang already implements -grecord-gcc-switches and so I decided to mirror the naming for the -f variant as well.

If anything I think dropping the -gcc- altogether would make the most sense. I don't understand why GCC includes it in the first place.

[rjmccall]

In D54489#1297509, @scott.linder wrote:

In D54489#1297504, @troyj wrote:

I realize that you're probably striving for option compatibility with gcc, but continuing to name it -frecord-gcc-switches when it actually records Clang switches seems weird to me. It almost sounds like something that would dump gcc equivalents of all Clang options, or maybe let you know which Clang options you've used that match gcc options. Either way, by the name -- if you aren't familiar with the gcc option -- it doesn't read like it records Clang options.

Would it be that bad to name it -frecord-clang-switches? Or just -frecord-switches?

I agree, and this was my original plan, but then I noticed that Clang already implements -grecord-gcc-switches and so I decided to mirror the naming for the -f variant as well.

If anything I think dropping the -gcc- altogether would make the most sense. I don't understand why GCC includes it in the first place.

Well, having something in the name that make it clear that this is about command line switches and not switch statements seems sensible. It's a little unfortunate to give a useful feature a vendor-specific name, but I think the right approach is probably to accept that for compatibility and then also accept a more neutral and descriptive name.

The code seems fine.

[scott.linder]

Change canonical option name to -frecord-command-line and add an alias from -frecord-gcc-switches

Do the same for -grecord-command-line/-grecord-gcc-switches

[rjmccall]

docs/ClangCommandLineReference.rst

797

1. Is this section always called .LLVM.command.line, or does it differ by target object-file format?
2. For that matter, is this actually supported by all object-file formats?
3. Please describe the format of the section. Is it just a flat C string? Is it nul-terminated? Are different options combined with spaces, and if so, does that mean that interior spaces in command line arguments are impossible to distinguish? If the latter, I assume that this is required behavior for backward compatibility; please at least apologize for that in the documentation. :)

[scott.linder]

docs/ClangCommandLineReference.rst

797

For (1) I think the answer should be that we emit .LLVM.command.line (or whatever name we land on) for every object-file format with the concept of names for sections.

For (2), I have only implemented this for ELF so far in LLVM. I don't know how reasonable that is, and if it isn't I can look at adding it to other common object-file formats that LLVM supports.

For (3), my current proposal for the format is the same as the .comment section for idents: null-surrounded strings. Interior spaces are escaped, in the same manner as for the -g variant. There may still be more thought to put into the format, as the GCC version null-terminates each individual option; the reason I avoided this is that during object linking the options of many command-lines simply get mixed together, which seems less than useful.

As an example, for ELF clang -frecord-command-line -S "foo .c" produces the ASM:

.section        .LLVM.command.line,"MS",@progbits,1
.zero   1
.ascii  "/usr/local/bin/clang-8 -frecord-command-line -S foo\\ .c"
.zero   1

And for multiple command-lines in a single object (e.g. linking three objects with recorded command-lines) this would be:

.section        .LLVM.command.line,"MS",@progbits,1
.zero   1
.ascii  "/usr/local/bin/clang-8 -frecord-command-line -c foo\\ .c"
.zero   1
.ascii  "/usr/local/bin/clang-8 -frecord-command-line -some -unique -options -c bar.c"
.zero   1
.ascii  "/usr/local/bin/clang-8 -frecord-command-line -something -else -c baz.c"
.zero   1

I will try to capture more of this in the documentation.

[scott.linder]

Looking a bit further, it seems some other object-file formats have conventions for naming (e.g. Mach-O sections are _snake_case) and so .LLVM.command.line may not end up being what we want universally.

Additionally it seems things like .ident are not supported by all object-file formats in LLVM; using Mach-O as an example again, it uses the default of "no .ident support", and I don't see the ident in the .S or .o

[rjmccall]

docs/ClangCommandLineReference.rst

797

Okay. I don't particularly need you to implement this on additional targets, but yes, please capture all this layout information in the documentation and make sure that the driver enforces any meaningful restrictions about which targets support it.

[scott.linder]

Update documentation for new option and error in the driver when generating for unsupported object-file format.

[rjmccall]

docs/ClangCommandLineReference.rst

799

How about something like "After linking, the section may contain multiple command lines, which will be individually terminated by null bytes. Separate arguments within a command line are combined with spaces; spaces and backslashes within an argument are escaped with backslashes."

And please write out "(ELF Only)" as something more like "This option is only currently supported on ELF targets."

[scott.linder]

Update documentation for new option

[rjmccall]

Thanks, LGTM.

[scott.linder]

There may be changes to some details in the LLVM patch; once they are finalized I will update the Clang patch.

[scott.linder]

Update documentation to match section naming change, and to explicitly note that the section format differs from the equivalent GCC format.

[rjmccall]

So we're using the same command line option as GCC to produce something in the same section as GCC but formatting that section incompatibly with GCC? That combination of choices does not seem like a good idea.

[scott.linder]

In D54489#1330255, @rjmccall wrote:

So we're using the same command line option as GCC to produce something in the same section as GCC but formatting that section incompatibly with GCC? That combination of choices does not seem like a good idea.

From the GCC man page -frecord-gcc-switches "is only implemented on some targets and the exact format of the recording is target and binary file format dependent, but it usually takes the form of a section containing ASCII text"

In the LLVM portion of this review it was decided that the implications of this is that well-behaved consumers should not make assumptions about the format. They should also not make assumptions about the name of the section, but we are effectively backwards-compatible in that we are a section with ASCII text, so tools which just e.g. dump the section for uses will continue to work with our implementation.

[rjmccall]

In D54489#1330331, @scott.linder wrote:

In D54489#1330255, @rjmccall wrote:

So we're using the same command line option as GCC to produce something in the same section as GCC but formatting that section incompatibly with GCC? That combination of choices does not seem like a good idea.

From the GCC man page -frecord-gcc-switches "is only implemented on some targets and the exact format of the recording is target and binary file format dependent, but it usually takes the form of a section containing ASCII text"

In the LLVM portion of this review it was decided that the implications of this is that well-behaved consumers should not make assumptions about the format. They should also not make assumptions about the name of the section, but we are effectively backwards-compatible in that we are a section with ASCII text, so tools which just e.g. dump the section for uses will continue to work with our implementation.

Ah, okay. I didn't realize the format was target-specific on GCC. That's fine.

[scott.linder]

rL349155

[MaskRay]

In D54489#1331249, @scott.linder wrote:

rL349155

@scott.linder Hi Scott, you may have an opion on https://gcc.gnu.org/ml/gcc-patches/2020-03/msg00230.html . I also started a thread on cfe-dev: http://lists.llvm.org/pipermail/cfe-dev/2020-March/064795.html

docs/ClangCommandLineReference.rst

 Disable auto-generation of preprocessed source files and a script for reproduction during a clang crash
 
 .. option:: -fno-sanitize-blacklist
 
 Don't use blacklist file for sanitizers
 
 .. option:: -fparse-all-comments
 
+.. option:: -frecord-command-line, -frecord-gcc-switches, -fno-record-command-line, -fno-record-gcc-switches
+
+Generate a section named ".GCC.command.line" containing the clang driver
+command-line. After linking, the section may contain multiple command lines,
+which will be individually terminated by null bytes. Separate arguments within
+a command line are combined with spaces; spaces and backslashes within an
+argument are escaped with backslashes. This format differs from the format of
+the equivalent section produced by GCC with the -frecord-gcc-switches flag.
+This option is currently only supported on ELF targets.
+
 .. option:: -fsanitize-address-field-padding=<arg>
 
 Level of field padding for AddressSanitizer
 
 .. option:: -fsanitize-address-globals-dead-stripping
 
 Enable linker dead stripping of globals in AddressSanitizer
 

 .. option:: -gembed-source, -gno-embed-source
 
 Embed source text in DWARF debug sections
 
 .. option:: -ggnu-pubnames, -gno-gnu-pubnames
 
 .. option:: -gpubnames, -gno-pubnames
 
-.. option:: -grecord-gcc-switches, -gno-record-gcc-switches
+.. option:: -grecord-command-line, -grecord-gcc-switches, -gno-record-command-line, -gno-record-gcc-switches
 
 .. option:: -gsplit-dwarf
 
 .. option:: -gstrict-dwarf, -gno-strict-dwarf
 
 .. option:: -gz
 
 DWARF debug sections compression type

include/clang/Basic/CodeGenOptions.h

public:
 
   /// The string to embed in debug information as the current working directory.
   std::string DebugCompilationDir;
 
   /// The string to embed in the debug information for the compile unit, if
   /// non-empty.
   std::string DwarfDebugFlags;
 
+  /// The string containing the commandline for the llvm.commandline metadata,
+  /// if non-empty.
+  std::string RecordCommandLine;
+
   std::map<std::string, std::string> DebugPrefixMap;
 
   /// The ABI to use for passing floating point arguments.
   std::string FloatABI;
 
   /// The floating-point denormal mode to use.
   std::string FPDenormalMode;
 

include/clang/Driver/CC1Options.td

 def debug_info_macro : Flag<["-"], "debug-info-macro">,
   HelpText<"Emit macro debug information">;
 def dwarf_version_EQ : Joined<["-"], "dwarf-version=">;
 def debugger_tuning_EQ : Joined<["-"], "debugger-tuning=">;
 def fdebug_compilation_dir : Separate<["-"], "fdebug-compilation-dir">,
   HelpText<"The compilation directory to embed in the debug info.">;
 def dwarf_debug_flags : Separate<["-"], "dwarf-debug-flags">,
   HelpText<"The string to embed in the Dwarf debug flags record.">;
+def record_command_line : Separate<["-"], "record-command-line">,
+  HelpText<"The string to embed in the .LLVM.command.line section.">;
 def compress_debug_sections : Flag<["-", "--"], "compress-debug-sections">,
     HelpText<"DWARF debug sections compression">;
 def compress_debug_sections_EQ : Joined<["-"], "compress-debug-sections=">,
     HelpText<"DWARF debug sections compression type">;
 def mno_exec_stack : Flag<["-"], "mnoexecstack">,
   HelpText<"Mark the file as not needing an executable stack">;
 def massembler_fatal_warnings : Flag<["-"], "massembler-fatal-warnings">,
   HelpText<"Make assembler warnings fatal">;

include/clang/Driver/Options.td

def fdiagnostics_color : Flag<["-"], "fdiagnostics-color">, Group<f_Group>,
   Flags<[CoreOption, DriverOption]>;
 def fdiagnostics_color_EQ : Joined<["-"], "fdiagnostics-color=">, Group<f_Group>;
 def fansi_escape_codes : Flag<["-"], "fansi-escape-codes">, Group<f_Group>,
   Flags<[CoreOption, CC1Option]>, HelpText<"Use ANSI escape codes for diagnostics">;
 def fcomment_block_commands : CommaJoined<["-"], "fcomment-block-commands=">, Group<f_clang_Group>, Flags<[CC1Option]>,
   HelpText<"Treat each comma separated argument in <arg> as a documentation comment block command">,
   MetaVarName<"<arg>">;
 def fparse_all_comments : Flag<["-"], "fparse-all-comments">, Group<f_clang_Group>, Flags<[CC1Option]>;
+def frecord_command_line : Flag<["-"], "frecord-command-line">,
+  Group<f_clang_Group>;
+def fno_record_command_line : Flag<["-"], "fno-record-command-line">,
+  Group<f_clang_Group>;
+def : Flag<["-"], "frecord-gcc-switches">, Alias<frecord_command_line>;
+def : Flag<["-"], "fno-record-gcc-switches">, Alias<fno_record_command_line>;
 def fcommon : Flag<["-"], "fcommon">, Group<f_Group>;
 def fcompile_resource_EQ : Joined<["-"], "fcompile-resource=">, Group<f_Group>;
 def fcomplete_member_pointers : Flag<["-"], "fcomplete-member-pointers">, Group<f_clang_Group>,
    Flags<[CoreOption, CC1Option]>,
    HelpText<"Require member pointer base types to be complete if they would be significant under the Microsoft ABI">;
 def fno_complete_member_pointers : Flag<["-"], "fno-complete-member-pointers">, Group<f_clang_Group>,
    Flags<[CoreOption]>,
    HelpText<"Do not require member pointer base types to be complete if they would be significant under the Microsoft ABI">;

 
 def gfull : Flag<["-"], "gfull">, Group<g_Group>;
 def gused : Flag<["-"], "gused">, Group<g_Group>;
 def gstabs : Joined<["-"], "gstabs">, Group<g_Group>, Flags<[Unsupported]>;
 def gcoff : Joined<["-"], "gcoff">, Group<g_Group>, Flags<[Unsupported]>;
 def gxcoff : Joined<["-"], "gxcoff">, Group<g_Group>, Flags<[Unsupported]>;
 def gvms : Joined<["-"], "gvms">, Group<g_Group>, Flags<[Unsupported]>;
 def gtoggle : Flag<["-"], "gtoggle">, Group<g_flags_Group>, Flags<[Unsupported]>;
-def grecord_gcc_switches : Flag<["-"], "grecord-gcc-switches">, Group<g_flags_Group>;
-def gno_record_gcc_switches : Flag<["-"], "gno-record-gcc-switches">,
+def grecord_command_line : Flag<["-"], "grecord-command-line">,
+  Group<g_flags_Group>;
+def gno_record_command_line : Flag<["-"], "gno-record-command-line">,
   Group<g_flags_Group>;
+def : Flag<["-"], "grecord-gcc-switches">, Alias<grecord_command_line>;
+def : Flag<["-"], "gno-record-gcc-switches">, Alias<gno_record_command_line>;
 def gstrict_dwarf : Flag<["-"], "gstrict-dwarf">, Group<g_flags_Group>;
 def gno_strict_dwarf : Flag<["-"], "gno-strict-dwarf">, Group<g_flags_Group>;
 def gcolumn_info : Flag<["-"], "gcolumn-info">, Group<g_flags_Group>, Flags<[CoreOption]>;
 def gno_column_info : Flag<["-"], "gno-column-info">, Group<g_flags_Group>, Flags<[CoreOption]>;
 def gsplit_dwarf : Flag<["-"], "gsplit-dwarf">, Group<g_flags_Group>;
 def gsplit_dwarf_EQ : Joined<["-"], "gsplit-dwarf=">, Group<g_flags_Group>,
   HelpText<"Set DWARF fission mode to either 'split' or 'single'">,
   Values<"split,single">;

lib/CodeGen/CodeGenModule.h

private:
   /// linkage specifications, giving them the "expected" name where possible.
   void EmitStaticExternCAliases();
 
   void EmitDeclMetadata();
 
   /// Emit the Clang version as llvm.ident metadata.
   void EmitVersionIdentMetadata();
 
+  /// Emit the Clang commandline as llvm.commandline metadata.
+  void EmitCommandLineMetadata();
+
   /// Emits target specific Metadata for global declarations.
   void EmitTargetMetadata();
 
   /// Emits OpenCL specific Metadata e.g. OpenCL version.
   void EmitOpenCLMetadata();
 
   /// Emit the llvm.gcov metadata used to tell LLVM where to emit the .gcno and
   /// .gcda files in a way that persists in .bc files.

lib/CodeGen/CodeGenModule.cpp

if (getCodeGenOpts().EmitGcovArcs || getCodeGenOpts().EmitGcovNotes)
     EmitCoverageFile();
 
   if (DebugInfo)
     DebugInfo->finalize();
 
   if (getCodeGenOpts().EmitVersionIdentMetadata)
     EmitVersionIdentMetadata();
 
+  if (!getCodeGenOpts().RecordCommandLine.empty())
+    EmitCommandLineMetadata();
+
   EmitTargetMetadata();
 }
 
 void CodeGenModule::EmitOpenCLMetadata() {
   // SPIR v2.0 s2.13 - The OpenCL version used by the module is stored in the
   // opencl.ocl.version named metadata node.
   llvm::Metadata *OCLVerElts[] = {
       llvm::ConstantAsMetadata::get(llvm::ConstantInt::get(
llvm::NamedMDNode *IdentMetadata =
     TheModule.getOrInsertNamedMetadata("llvm.ident");
   std::string Version = getClangFullVersion();
   llvm::LLVMContext &Ctx = TheModule.getContext();
 
   llvm::Metadata *IdentNode[] = {llvm::MDString::get(Ctx, Version)};
   IdentMetadata->addOperand(llvm::MDNode::get(Ctx, IdentNode));
 }
 
+void CodeGenModule::EmitCommandLineMetadata() {
+  llvm::NamedMDNode *CommandLineMetadata =
+    TheModule.getOrInsertNamedMetadata("llvm.commandline");
+  std::string CommandLine = getCodeGenOpts().RecordCommandLine;
+  llvm::LLVMContext &Ctx = TheModule.getContext();
+
+  llvm::Metadata *CommandLineNode[] = {llvm::MDString::get(Ctx, CommandLine)};
+  CommandLineMetadata->addOperand(llvm::MDNode::get(Ctx, CommandLineNode));
+}
+
 void CodeGenModule::EmitTargetMetadata() {
   // Warning, new MangledDeclNames may be appended within this loop.
   // We rely on MapVector insertions adding new elements to the end
   // of the container.
   // FIXME: Move this loop into the one target that needs it, and only
   // loop over those declarations for which we couldn't emit the target
   // metadata when we emitted the declaration.
   for (unsigned I = 0; I != MangledDeclNames.size(); ++I) {

lib/Driver/ToolChains/Clang.cpp

for (const InputInfo &Input : FrontendInputs) {
     else
       Input.getInputArg().renderAsInput(Args, CmdArgs);
   }
 
   Args.AddAllArgs(CmdArgs, options::OPT_undef);
 
   const char *Exec = D.getClangProgramPath();
 
-  // Optionally embed the -cc1 level arguments into the debug info, for build
-  // analysis.
+  // Optionally embed the -cc1 level arguments into the debug info or a
+  // section, for build analysis.
   // Also record command line arguments into the debug info if
   // -grecord-gcc-switches options is set on.
   // By default, -gno-record-gcc-switches is set on and no recording.
-  if (TC.UseDwarfDebugFlags() ||
-      Args.hasFlag(options::OPT_grecord_gcc_switches,
-                   options::OPT_gno_record_gcc_switches, false)) {
+  auto GRecordSwitches =
+      Args.hasFlag(options::OPT_grecord_command_line,
+                   options::OPT_gno_record_command_line, false);
+  auto FRecordSwitches =
+      Args.hasFlag(options::OPT_frecord_command_line,
+                   options::OPT_fno_record_command_line, false);
+  if (FRecordSwitches && !Triple.isOSBinFormatELF())
+    D.Diag(diag::err_drv_unsupported_opt_for_target)
+        << Args.getLastArg(options::OPT_frecord_command_line)->getAsString(Args)
+        << TripleStr;
+  if (TC.UseDwarfDebugFlags() || GRecordSwitches || FRecordSwitches) {
     ArgStringList OriginalArgs;
     for (const auto &Arg : Args)
       Arg->render(Args, OriginalArgs);
 
     SmallString<256> Flags;
     Flags += Exec;
     for (const char *OriginalArg : OriginalArgs) {
       SmallString<128> EscapedArg;
       EscapeSpacesAndBackslashes(OriginalArg, EscapedArg);
       Flags += " ";
       Flags += EscapedArg;
     }
+    auto FlagsArgString = Args.MakeArgString(Flags);
+    if (TC.UseDwarfDebugFlags() || GRecordSwitches) {
       CmdArgs.push_back("-dwarf-debug-flags");
-    CmdArgs.push_back(Args.MakeArgString(Flags));
+      CmdArgs.push_back(FlagsArgString);
+    }
+    if (FRecordSwitches) {
+      CmdArgs.push_back("-record-command-line");
+      CmdArgs.push_back(FlagsArgString);
+    }
   }
 
   // Host-side cuda compilation receives all device-side outputs in a single
   // fatbin as Inputs[1]. Include the binary with -fcuda-include-gpubinary.
   if ((IsCuda || IsHIP) && CudaDeviceInput) {
       CmdArgs.push_back("-fcuda-include-gpubinary");
       CmdArgs.push_back(CudaDeviceInput->getFilename());
       if (Args.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc, false))

lib/Frontend/CompilerInvocation.cpp

static bool ParseCodeGenArgs(CodeGenOptions &Opts, ArgList &Args, InputKind IK,
   Opts.RelaxedAliasing = Args.hasArg(OPT_relaxed_aliasing);
   Opts.StructPathTBAA = !Args.hasArg(OPT_no_struct_path_tbaa);
   Opts.NewStructPathTBAA = !Args.hasArg(OPT_no_struct_path_tbaa) &&
                            Args.hasArg(OPT_new_struct_path_tbaa);
   Opts.FineGrainedBitfieldAccesses =
       Args.hasFlag(OPT_ffine_grained_bitfield_accesses,
                    OPT_fno_fine_grained_bitfield_accesses, false);
   Opts.DwarfDebugFlags = Args.getLastArgValue(OPT_dwarf_debug_flags);
+  Opts.RecordCommandLine = Args.getLastArgValue(OPT_record_command_line);
   Opts.MergeAllConstants = Args.hasArg(OPT_fmerge_all_constants);
   Opts.NoCommon = Args.hasArg(OPT_fno_common);
   Opts.NoImplicitFloat = Args.hasArg(OPT_no_implicit_float);
   Opts.OptimizeSize = getOptimizationLevelSize(Args);
   Opts.SimplifyLibCalls = !(Args.hasArg(OPT_fno_builtin) ||
                             Args.hasArg(OPT_ffreestanding));
   if (Opts.SimplifyLibCalls)
     getAllNoBuiltinFuncValues(Args, Opts.NoBuiltinFuncs);

test/Driver/clang_f_opts.c

 // RUN: %clang -### -S -fno-delete-null-pointer-checks -fdelete-null-pointer-checks %s 2>&1 | FileCheck -check-prefix=CHECK-NULL-POINTER-CHECKS %s
 // CHECK-NO-NULL-POINTER-CHECKS: "-fno-delete-null-pointer-checks"
 // CHECK-NULL-POINTER-CHECKS-NOT: "-fno-delete-null-pointer-checks"
 
 // RUN: %clang -### -S -fomit-frame-pointer -pg %s 2>&1 | FileCheck -check-prefix=CHECK-NO-MIX-OMIT-FP-PG %s
 // RUN: %clang -### -S -fomit-frame-pointer -fno-omit-frame-pointer -pg %s 2>&1 | FileCheck -check-prefix=CHECK-MIX-NO-OMIT-FP-PG %s
 // CHECK-NO-MIX-OMIT-FP-PG: '-fomit-frame-pointer' not allowed with '-pg'
 // CHECK-MIX-NO-OMIT-FP-PG-NOT: '-fomit-frame-pointer' not allowed with '-pg'
+
+// RUN: %clang -### -S -target x86_64-unknown-linux -frecord-gcc-switches %s 2>&1 | FileCheck -check-prefix=CHECK-RECORD-GCC-SWITCHES %s
+// RUN: %clang -### -S -target x86_64-unknown-linux -fno-record-gcc-switches %s 2>&1 | FileCheck -check-prefix=CHECK-NO-RECORD-GCC-SWITCHES %s
+// RUN: %clang -### -S -target x86_64-unknown-linux -fno-record-gcc-switches -frecord-gcc-switches %s 2>&1 | FileCheck -check-prefix=CHECK-RECORD-GCC-SWITCHES %s
+// RUN: %clang -### -S -target x86_64-unknown-linux -frecord-gcc-switches -fno-record-gcc-switches %s 2>&1 | FileCheck -check-prefix=CHECK-NO-RECORD-GCC-SWITCHES %s
+// RUN: %clang -### -S -target x86_64-unknown-linux -frecord-command-line %s 2>&1 | FileCheck -check-prefix=CHECK-RECORD-GCC-SWITCHES %s
+// RUN: %clang -### -S -target x86_64-unknown-linux -fno-record-command-line %s 2>&1 | FileCheck -check-prefix=CHECK-NO-RECORD-GCC-SWITCHES %s
+// RUN: %clang -### -S -target x86_64-unknown-linux -fno-record-command-line -frecord-command-line %s 2>&1 | FileCheck -check-prefix=CHECK-RECORD-GCC-SWITCHES %s
+// RUN: %clang -### -S -target x86_64-unknown-linux -frecord-command-line -fno-record-command-line %s 2>&1 | FileCheck -check-prefix=CHECK-NO-RECORD-GCC-SWITCHES %s
+// Test with a couple examples of non-ELF object file formats
+// RUN: %clang -### -S -target x86_64-unknown-macosx -frecord-command-line %s 2>&1 | FileCheck -check-prefix=CHECK-RECORD-GCC-SWITCHES-ERROR %s
+// RUN: %clang -### -S -target x86_64-unknown-windows -frecord-command-line %s 2>&1 | FileCheck -check-prefix=CHECK-RECORD-GCC-SWITCHES-ERROR %s
+// CHECK-RECORD-GCC-SWITCHES: "-record-command-line"
+// CHECK-NO-RECORD-GCC-SWITCHES-NOT: "-record-command-line"
+// CHECK-RECORD-GCC-SWITCHES-ERROR: error: unsupported option '-frecord-command-line' for target

test/Driver/debug-options.c

 //             | FileCheck -check-prefix=GNO_RECORD %s
 // RUN: %clang -### -c -grecord-gcc-switches -gno-record-gcc-switches %s 2>&1 \
 //             | FileCheck -check-prefix=GNO_RECORD %s/
 // RUN: %clang -### -c -grecord-gcc-switches -o - %s 2>&1 \
 //             | FileCheck -check-prefix=GRECORD_O %s
 // RUN: %clang -### -c -O3 -ffunction-sections -grecord-gcc-switches %s 2>&1 \
 //             | FileCheck -check-prefix=GRECORD_OPT %s
 //
+// RUN: %clang -### -c -grecord-command-line %s 2>&1 \
+//             | FileCheck -check-prefix=GRECORD %s
+// RUN: %clang -### -c -gno-record-command-line %s 2>&1 \
+//             | FileCheck -check-prefix=GNO_RECORD %s
+// RUN: %clang -### -c -grecord-command-line -gno-record-command-line %s 2>&1 \
+//             | FileCheck -check-prefix=GNO_RECORD %s/
+// RUN: %clang -### -c -grecord-command-line -o - %s 2>&1 \
+//             | FileCheck -check-prefix=GRECORD_O %s
+// RUN: %clang -### -c -O3 -ffunction-sections -grecord-command-line %s 2>&1 \
+//             | FileCheck -check-prefix=GRECORD_OPT %s
+//
 // RUN: %clang -### -c -gstrict-dwarf -gno-strict-dwarf %s 2>&1 \
 // RUN:        | FileCheck -check-prefix=GIGNORE %s
 //
 // RUN: %clang -### -c -ggnu-pubnames %s 2>&1 | FileCheck -check-prefix=GPUB %s
 // RUN: %clang -### -c -ggdb %s 2>&1 | FileCheck -check-prefix=NOPUB %s
 // RUN: %clang -### -c -ggnu-pubnames -gno-gnu-pubnames %s 2>&1 | FileCheck -check-prefix=NOPUB %s
 // RUN: %clang -### -c -ggnu-pubnames -gno-pubnames %s 2>&1 | FileCheck -check-prefix=NOPUB %s
 //