Add option for emitting dbg info for call site parameters

[djtodoro]

Summary

Generating of all info about call sites and call site parameters will be restricted with the option until all testing and patching is done.

Example of using the option:

clang -g -O2 -Xclang -femit-debug-entry-values test.c

In addition, when the option is set add flag all_call_sites
in a subprogram in order to support GNU extension of DWARF as well.

Authors: @asowda, @NikolaPrica, @djtodoro, @ivanbaev

[djtodoro]

-Rebase
-Add all_call_sites flag in the case of GNU extensions

[aprantl]

include/clang/Driver/Options.td

919

I assume that this is the same -f option that GCC uses?

[djtodoro]

include/clang/Driver/Options.td

919

Actually in GCC production of call_site and call_site_parameters debug info is enabled by default.
Production of entry_values is implemented on top of variable's value tracking system and there is no particular option just for entry_values.

Basically, we introduce this option as experimental one. As soon as we test everything we should get rid of this or turn it ON by default (and maybe add '-fno-emit-param-entry-values' in order to have an option for disabling the functionality). That will be ideal scenario.

[probinson]

I guess I'm not clear what your final goal is for the option. Keep it, even though GCC doesn't have one like it? Eliminate it? Please clearly state what you intend to have in the end, and what you might have in the short term in case that is different.

include/clang/Driver/CC1Options.td

362 ↗

(On Diff #195188)

I think this is not necessary, see comment on Options.td.

include/clang/Driver/Options.td

919

By convention, the name of the option should start with 'f' and match the option spelling (with hyphens changed to underscore).

921

I believe that by marking this with [CC1Option] cc1 will automatically understand it and you won't need to define a separate option in CC1Options.td.

[aprantl]

And if we plan to enable it by default, too, perhaps not adding a driver option (CC1 only) is preferable, since we need to maintain compatibility with driver options indefinitely.

[djtodoro]

@probinson @aprantl Thanks a lot for your comments!

Let's clarify some things. I'm sorry about the confusion.

Initial patch for the functionality can be restricted by this option (like we pushed here), since LLDB doesn't read this type of debug info (yet).
The plan is, when LLDB has all necessary info and we are sure that everything works fine during using this info in debugging sessions, we can turn it to ON by default.

Does that make sense ? :)

include/clang/Driver/Options.td

921

I can't remember what problem we occurred with this on 4.0, but we definitely don't need both definitions.
Thanks for this!

[djtodoro]

-Refactor
-Remove CC1 def

[djtodoro]

-Fix comments

[probinson]

In D58033#1470260, @djtodoro wrote:

@probinson @aprantl Thanks a lot for your comments!

Let's clarify some things. I'm sorry about the confusion.

Initial patch for the functionality can be restricted by this option (like we pushed here), since LLDB doesn't read this type of debug info (yet).
The plan is, when LLDB has all necessary info and we are sure that everything works fine during using this info in debugging sessions, we can turn it to ON by default.

Does that make sense ? :)

Yes. And I agree with Adrian, this should be a CC1-only option; it can be default-off for now and maybe debugger-tuning specific later.

[djtodoro]

-Add only cc1 option
-Set up back end

[aprantl]

Is there some kind of testcase?

[djtodoro]

Is there some kind of testcase?

@aprantl Usage of the option is tested within following patches from the stack. I am not sure if we need some additional test here?

[probinson]

lib/Driver/ToolChains/Clang.cpp

3402

If this is now a cc1-only option, this part goes away right?

[djtodoro]

lib/Driver/ToolChains/Clang.cpp

3402

Yes, thanks for this!

[djtodoro]

-Remove unneeded code

[probinson]

I'm okay with this, but give @aprantl a chance to confirm.

[vsk]

I think a small extension to test/CodeGenCXX/dbg-info-all-calls-described.cpp for the 'Supported: DWARF4 + -femit_param_entry_values' case would be appropriate here.

[djtodoro]

@probinson @aprantl Thanks!

@vsk Thanks for the comment! Sure, we will add that.

[djtodoro]

-Add an input in test/CodeGenCXX/dbg-info-all-calls-described.cpp
-Rename the option

[djtodoro]

-Set EnableDebugEntryValues only for a higher level of optimizations (-O1 and higher)

[aprantl]

LGTM, with an additional testcase.

test/CodeGenCXX/dbg-info-all-calls-described.cpp

23 ↗

(On Diff #200485)

+ a negative test for -O0

[djtodoro]

-Add a negative test for -O0

[djtodoro]

-Rebase

[djtodoro]

-Explicitly enable the option only in the case of X86

[probinson]

lib/Frontend/CompilerInvocation.cpp

754

You want to disable entry-values for all targets other than X86?

[djtodoro]

lib/Frontend/CompilerInvocation.cpp

754

@probinson Thanks for the comment!

You want to disable entry-values for all targets other than X86?

Yes, for now. Because, if we want to support the entry-values feature for an architecture (other than X86), we would need to handle target specifics of call arguments lowering for the architecture.

For the initial set of patches, we have accomplished to test and add support only for X86. We will work on supporting more architectures.

include/clang/Basic/CodeGenOptions.def

CODEGENOPT(DisableLLVMPasses , 1, 0) ///< Don't run any LLVM IR passes to get
                                      ///< frontend.
 CODEGENOPT(DisableLifetimeMarkers, 1, 0) ///< Don't emit any lifetime markers
 CODEGENOPT(DisableO0ImplyOptNone , 1, 0) ///< Don't annonate function with optnone at O0
 CODEGENOPT(ExperimentalNewPassManager, 1, 0) ///< Enables the new, experimental
                                              ///< pass manager.
 CODEGENOPT(DebugPassManager, 1, 0) ///< Prints debug information for the new
                                    ///< pass manager.
 CODEGENOPT(DisableRedZone    , 1, 0) ///< Set when -mno-red-zone is enabled.
+CODEGENOPT(EnableParamEntryValues, 1, 0) ///< Emit call site parameter dbg info
 CODEGENOPT(IndirectTlsSegRefs, 1, 0) ///< Set when -mno-tls-direct-seg-refs
                                      ///< is specified.
 CODEGENOPT(DisableTailCalls  , 1, 0) ///< Do not emit tail calls.
 CODEGENOPT(NoEscapingBlockTailCalls, 1, 0) ///< Do not emit tail calls from
                                            ///< escaping blocks.
 CODEGENOPT(EmitDeclMetadata  , 1, 0) ///< Emit special metadata indicating what
                                      ///< Decl* various IR entities came from.
                                      ///< Only useful when running CodeGen as a

include/clang/Driver/Options.td

def fmath_errno : Flag<["-"], "fmath-errno">, Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Require math functions to indicate errors by setting errno">;
 def fno_math_errno : Flag<["-"], "fno-math-errno">, Group<f_Group>;
 def fbracket_depth_EQ : Joined<["-"], "fbracket-depth=">, Group<f_Group>, Flags<[CoreOption]>;
 def fsignaling_math : Flag<["-"], "fsignaling-math">, Group<f_Group>;
 def fno_signaling_math : Flag<["-"], "fno-signaling-math">, Group<f_Group>;
 def fjump_tables : Flag<["-"], "fjump-tables">, Group<f_Group>;
 def fno_jump_tables : Flag<["-"], "fno-jump-tables">, Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Do not use jump tables for lowering switches">;
+def femit_param_entry_values : Flag<["-"], "femit-param-entry-values">, Group<f_Group>,
+  Flags<[CC1Option]>,
+  HelpText<"Enables debug info about call site parameter's entry values">;
 def fforce_enable_int128 : Flag<["-"], "fforce-enable-int128">,
   Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Enable support for int128_t type">;
 def fno_force_enable_int128 : Flag<["-"], "fno-force-enable-int128">,
   Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Disable support for int128_t type">;
 def fkeep_static_consts : Flag<["-"], "fkeep-static-consts">, Group<f_Group>, Flags<[CC1Option]>,
   HelpText<"Keep static const variables even if unused">;

lib/CodeGen/CGDebugInfo.cpp

if (!CGM.getLangOpts().Optimize || DebugKind == codegenoptions::NoDebugInfo ||
       DebugKind == codegenoptions::LocTrackingOnly)
     return llvm::DINode::FlagZero;
 
   // Call site-related attributes are available in DWARF v5. Some debuggers,
   // while not fully DWARF v5-compliant, may accept these attributes as if they
   // were part of DWARF v4.
   bool SupportsDWARFv4Ext =
       CGM.getCodeGenOpts().DwarfVersion == 4 &&
-      CGM.getCodeGenOpts().getDebuggerTuning() == llvm::DebuggerKind::LLDB;
+      (CGM.getCodeGenOpts().getDebuggerTuning() == llvm::DebuggerKind::LLDB ||
+       (CGM.getCodeGenOpts().EnableParamEntryValues &&
+       CGM.getCodeGenOpts().getDebuggerTuning() == llvm::DebuggerKind::GDB));
+
   if (!SupportsDWARFv4Ext && CGM.getCodeGenOpts().DwarfVersion < 5)
     return llvm::DINode::FlagZero;
 
   return llvm::DINode::FlagAllCallsDescribed;
 }

lib/Driver/ToolChains/Clang.cpp

static void RenderDebugOptions(const ToolChain &TC, const Driver &D,
   if (DebuggerTuning == llvm::DebuggerKind::SCE)
     CmdArgs.push_back("-debug-forward-template-params");
 
   // Do we need to explicitly import anonymous namespaces into the parent
   // scope?
   if (DebuggerTuning == llvm::DebuggerKind::SCE)
     CmdArgs.push_back("-dwarf-explicit-import");
 
+  // Enable param entry values functionality.
+  if (Args.hasArg(options::OPT_femit_param_entry_values))
+    CmdArgs.push_back("-femit-param-entry-values");
+
   RenderDebugInfoCompressionArgs(Args, CmdArgs, D, TC);
 }
 
 void Clang::ConstructJob(Compilation &C, const JobAction &JA,
                          const InputInfo &Output, const InputInfoList &Inputs,
                          const ArgList &Args, const char *LinkingOutput) const {
   const auto &TC = getToolChain();
   const llvm::Triple &RawTriple = TC.getTriple();

lib/Frontend/CompilerInvocation.cpp

for (const auto &Arg : Args.getAllArgValues(OPT_fdebug_prefix_map_EQ))
     Opts.DebugPrefixMap.insert(StringRef(Arg).split('='));
 
   if (const Arg *A =
           Args.getLastArg(OPT_emit_llvm_uselists, OPT_no_emit_llvm_uselists))
     Opts.EmitLLVMUseLists = A->getOption().getID() == OPT_emit_llvm_uselists;
 
   Opts.DisableLLVMPasses = Args.hasArg(OPT_disable_llvm_passes);
   Opts.DisableLifetimeMarkers = Args.hasArg(OPT_disable_lifetimemarkers);
+  Opts.EnableParamEntryValues = Args.hasArg(OPT_femit_param_entry_values);
   Opts.DisableO0ImplyOptNone = Args.hasArg(OPT_disable_O0_optnone);
   Opts.DisableRedZone = Args.hasArg(OPT_disable_red_zone);
   Opts.IndirectTlsSegRefs = Args.hasArg(OPT_mno_tls_direct_seg_refs);
   Opts.ForbidGuardVariables = Args.hasArg(OPT_fforbid_guard_variables);
   Opts.UseRegisterSizedBitfieldAccess = Args.hasArg(
     OPT_fuse_register_sized_bitfield_access);
   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);