[llvm-cxxfilt] Enable support for D programming language demangling

[ljmf00]

Summary

This is part of https://github.com/dlang/projects/issues/81

Enable support for demangling D symbols using llvm-cxxfilt tool

Signed-off-by: Luís Ferreira <[e-mail]>

[jhenderson]

One nit, otherwise LGTM.

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

21

I don't think you need to include this header - strncmp is already used in this file, so it's clearly being included through the existing include hierarchy. See also https://llvm.org/docs/CodingStandards.html#include-as-little-as-possible.

[ljmf00]

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

21

Yeah, my editor LSP plugin probably added it automatically by mistake, and no apparent duplication was detected. I'm going to remove it!

[jhenderson]

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

21

No problem!

[dblaikie]

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

21

Generally we should include anything we depend on - not rely on indirect inclusions. (the style guide gives some allowance for this - like if some other header just has to have provided a definition/inclusion of a thing, that's fine - but in this case there's no strong connection between whatever header includes something for strncmp so it's suitable to include it here, so if that other header is refactored to remove the need/use of the header it doesn't break this code)

So this #include is probably good/proper here.

The advice in the style guide is about not including unused headers and not including headers when a forward declaration would suffice.

79–100

Perhaps this code could be refactored to be shared with llvm-symbolizer? (eg: the code being updated in D110664 here: llvm/lib/DebugInfo/Symbolize/Symbolize.cpp)

79–100

Maybe the refactoring would be better done in D110664 and then that could be the basis for this patch to make one change that'd apply to both llvm-cxxfilt and llvm-symbolizer/symbolizing library.

[ljmf00]

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

21

I can agree on that, I just also agreed with @jhenderson more likely because itanium and rust code parts are also using strncmp without such include.

79–100

I can wait for that refactor on https://reviews.llvm.org/D110664 and make a patch to symbolize code

[dblaikie]

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

21

Yeah, I think good tooling probably trumps whatever people did before - most of us in the absence of tooling just add #includes whenever we notice they're needed, I think. So good tooling probably is doing a better job than us mere humans. It sounds like the include is appropriate to me.

But I'm not super fussed.

79–100

Thanks!

[ljmf00]

Updated the diff to add the requested header. Also added a simpler test to be able to change the parent diff and stack with the diff implementing simple symbols.

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

21

I added the header again.

[jhenderson]

LGTM again.

[ljmf00]

Since support for llvm-cxxfilt was synced with Demangle.cpp logic on https://reviews.llvm.org/D110664, this doesn't makes sense anymore. Tests are going to be added on https://reviews.llvm.org/D111414 instead.

llvm/test/Demangle/dlang.test

+RUN: llvm-cxxfilt -n  < %s | FileCheck --match-full-lines %s
+
+; Full test suite for dlang demangling at
+; llvm/unittests/Demangle/DLangDemangleTest.cpp
+
+CHECK: D main
+       _Dmain
+
+CHECK: demangle.test
+       _D8demangle4testZ

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

 #include "llvm/Option/ArgList.h"
 #include "llvm/Option/Option.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Host.h"
 #include "llvm/Support/InitLLVM.h"
 #include "llvm/Support/WithColor.h"
 #include "llvm/Support/raw_ostream.h"
 #include <cstdlib>
 #include <iostream>
 
 using namespace llvm;
 
 namespace {
 enum ID {
   OPT_INVALID = 0, // This is not an option ID.
 #define OPTION(PREFIX, NAME, ID, KIND, GROUP, ALIAS, ALIASARGS, FLAGS, PARAM,  \
                HELPTEXT, METAVAR, VALUES)                                      \
static std::string demangle(const std::string &Mangled) {
   const char *DecoratedStr = Mangled.c_str();
   if (StripUnderscore)
     if (DecoratedStr[0] == '_')
       ++DecoratedStr;
   size_t DecoratedLength = strlen(DecoratedStr);
 
   char *Undecorated = nullptr;
 
   if (Types ||
       ((DecoratedLength >= 2 && strncmp(DecoratedStr, "_Z", 2) == 0) ||
        (DecoratedLength >= 4 && strncmp(DecoratedStr, "___Z", 4) == 0)))
     Undecorated = itaniumDemangle(DecoratedStr, nullptr, nullptr, &Status);
 
   if (!Undecorated &&
       (DecoratedLength > 6 && strncmp(DecoratedStr, "__imp_", 6) == 0)) {
     Prefix = "import thunk for ";
     Undecorated = itaniumDemangle(DecoratedStr + 6, nullptr, nullptr, &Status);
   }
 
   if (!Undecorated &&
       (DecoratedLength >= 2 && strncmp(DecoratedStr, "_R", 2) == 0)) {
     Undecorated = rustDemangle(DecoratedStr, nullptr, nullptr, &Status);
   }
 
+  if (!Undecorated &&
+      (DecoratedLength >= 2 && strncmp(DecoratedStr, "_D", 2) == 0)) {
+    Undecorated = dlangDemangle(DecoratedStr);
+  }
+
   std::string Result(Undecorated ? Prefix + Undecorated : Mangled);
   free(Undecorated);
   return Result;
 }
 
 // Split 'Source' on any character that fails to pass 'IsLegalChar'.  The
 // returned vector consists of pairs where 'first' is the delimited word, and
 // 'second' are the delimiters following that word.
 static void SplitStringDelims(