Demangling of C++ names of Linux Binaries does not work on Mac OS

[cwgreene]

Steps to reproduce:

Compare output of tests/testdata/linux-x86_64/oldbloaty.bin on Linux and Mac.

Linux:

     VM SIZE                                                                                        FILE SIZE
 --------------                                                                                  --------------
  20.0%  99.0Ki [Unmapped]                                                                        6.55Mi  94.4%
  62.1%   307Ki [Other]                                                                            308Ki   4.3%
   2.5%  12.3Ki re2::RE2::Match(re2::StringPiece const&, int, int, re2::RE2::Anchor, re2::String  12.3Ki   0.2%
   1.6%  7.83Ki re2::unicode_groups                                                               7.83Ki   0.1%
   1.5%  7.56Ki re2::NFA::Search                                                                  7.56Ki   0.1%
   1.2%  5.76Ki re2::BitState::TrySearch                                                          5.76Ki   0.1%
   1.0%  4.81Ki _dwarf_exec_frame_instr                                                           4.81Ki   0.1%
   0.9%  4.49Ki re2::DFA::DFA                                                                     4.49Ki   0.1%
   0.9%  4.34Ki re2::Regexp::Parse                                                                4.34Ki   0.1%
   0.8%  4.20Ki re2::RE2::Init                                                                    4.20Ki   0.1%
   0.8%  4.09Ki re2::Prog::IsOnePass                                                              4.09Ki   0.1%
   0.8%  4.04Ki re2::Compiler::PostVisit                                                          4.04Ki   0.1%
   0.8%  4.01Ki _dwarf_internal_srclines                                                          4.01Ki   0.1%
   0.8%  3.91Ki re2::Regexp::FactorAlternationRecursive                                           3.91Ki   0.1%
   0.8%  3.86Ki bool bloaty::(anonymous namespace)::ForEachElf<bloaty::(anonymous namespace)::Do  3.86Ki   0.1%
   0.8%  3.77Ki re2::DFA::RunStateOnByte                                                          3.77Ki   0.1%
   0.7%  3.68Ki re2::unicode_casefold                                                             3.68Ki   0.1%
   0.7%  3.40Ki re2::DFA::InlinedSearchLoop(re2::DFA::SearchParams*, bool, bool, bool) [clone .c  3.40Ki   0.0%
   0.7%  3.38Ki re2::DFA::InlinedSearchLoop(re2::DFA::SearchParams*, bool, bool, bool) [clone .c  3.38Ki   0.0%
   0.7%  3.35Ki re2::Regexp::Walker<int>::WalkInternal(re2::Regexp*, int, bool) [clone .constpro  3.35Ki   0.0%
   0.0%     165 [None]                                                                                 0   0.0%
 100.0%   495Ki TOTAL                                                                             6.93Mi 100.0%

Mac:

     VM SIZE                                                                                        FILE SIZE
 --------------                                                                                  --------------
  20.0%  99.0Ki [Unmapped]                                                                        6.55Mi  94.4%
  62.1%   307Ki [Other]                                                                            308Ki   4.3%
   2.5%  12.3Ki _ZNK3re23RE25MatchERKNS_11StringPieceEiiNS0_6AnchorEPS1_i                         12.3Ki   0.2%
   1.6%  7.83Ki _ZN3re214unicode_groupsE                                                          7.83Ki   0.1%
   1.5%  7.56Ki _ZN3re23NFA6SearchERKNS_11StringPieceES3_bbPS1_i                                  7.56Ki   0.1%
   1.2%  5.76Ki _ZN3re28BitState9TrySearchEiPKc                                                   5.76Ki   0.1%
   1.0%  4.81Ki _dwarf_exec_frame_instr                                                           4.81Ki   0.1%
   0.9%  4.49Ki _ZN3re23DFAC1EPNS_4ProgENS1_9MatchKindEl                                          4.49Ki   0.1%
   0.9%  4.34Ki _ZN3re26Regexp5ParseERKNS_11StringPieceENS0_10ParseFlagsEPNS_12RegexpStatusE      4.34Ki   0.1%
   0.8%  4.20Ki _ZN3re23RE24InitERKNS_11StringPieceERKNS0_7OptionsE                               4.20Ki   0.1%
   0.8%  4.09Ki _ZN3re24Prog9IsOnePassEv                                                          4.09Ki   0.1%
   0.8%  4.04Ki _ZN3re28Compiler9PostVisitEPNS_6RegexpENS_4FragES3_PS3_i                          4.04Ki   0.1%
   0.8%  4.01Ki _dwarf_internal_srclines                                                          4.01Ki   0.1%
   0.8%  3.91Ki _ZN3re26Regexp26FactorAlternationRecursiveEPPS0_iNS0_10ParseFlagsEi               3.91Ki   0.1%
   0.8%  3.86Ki _ZN6bloaty12_GLOBAL__N_110ForEachElfIZNS0_L17DoReadELFSectionsEPNS_9RangeSinkEbE  3.86Ki   0.1%
   0.8%  3.77Ki _ZN3re23DFA14RunStateOnByteEPNS0_5StateEi                                         3.77Ki   0.1%
   0.7%  3.68Ki _ZN3re216unicode_casefoldE                                                        3.68Ki   0.1%
   0.7%  3.40Ki _ZN3re23DFA17InlinedSearchLoopEPNS0_12SearchParamsEbbb.constprop.134              3.40Ki   0.0%
   0.7%  3.38Ki _ZN3re23DFA17InlinedSearchLoopEPNS0_12SearchParamsEbbb.constprop.133              3.38Ki   0.0%
   0.7%  3.35Ki _ZN3re26Regexp6WalkerIiE12WalkInternalEPS0_ib.constprop.183                       3.35Ki   0.0%
   0.0%     165 [None]                                                                                 0   0.0%
 100.0%   495Ki TOTAL                                                                             6.93Mi 100.0%

Note that mac binaries built on mac are unaffected.

Mac binaries (MachO) on linux, of course, can't be processed because bloaty uses otool to process MachO binaries.

[cwgreene]

So, bloaty shells out to c++filt. It appears that when you create a MachO binary (clang++ -c test.cc --target x86_64-pc-linux-macho), clang will generate an additional _ in the mangled name (huzzah for ABI inconsistency!).

So the quick hack is to set a flag on the binary, and if the binary is not MachO, pass in -n to c++filt.

Ubuntu:

$ c++filt --version 
GNU c++filt (GNU Binutils for Ubuntu) 2.24
Copyright 2013 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public License version 3 or (at your option) any later version.
This program has absolutely no warranty.
$ c++filt --help
Usage: c++filt [options] [mangled names]
Options are:
  [-_|--strip-underscore]     Ignore first leading underscore
  [-n|--no-strip-underscore]  Do not ignore a leading underscore (default)
  [-p|--no-params]            Do not display function arguments
  [-i|--no-verbose]           Do not show implementation details (if any)
  [-t|--types]                Also attempt to demangle type encodings
  [-s|--format {none,auto,gnu,lucid,arm,hp,edg,gnu-v3,java,gnat}]
  [@]                   Read extra options from 
  [-h|--help]                 Display this information
  [-v|--version]              Show the version information
Demangled names are displayed to stdout.
If a name cannot be demangled it is just echoed to stdout.
If no names are provided on the command line, stdin is read.
Report bugs to .

On Mac:

$ c++filt --version
GNU c++filt 070207 20070207
Copyright 2005 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public License.  This program has absolutely no warranty.
$ c++filt --help
Usage: c++filt [options] [mangled names]
Options are:
  [-_|--strip-underscore]     Ignore first leading underscore (default)
  [-n|--no-strip-underscore]  Do not ignore a leading underscore
  [-p|--no-params]            Do not display function arguments
  [-i|--no-verbose]           Do not show implementation details (if any)
  [-t|--types]                Also attempt to demangle type encodings
  [-s|--format {none,auto,gnu,lucid,arm,hp,edg,gnu-v3,java,gnat}]
  [@]                   Read extra options from 
  [-h|--help]                 Display this information
  [-v|--version]              Show the version information
Demangled names are displayed to stdout.
If a name cannot be demangled it is just echoed to stdout.
If no names are provided on the command line, stdin is read.

[haberman]

Interesting. Since we are depending on a platform-specific tool, it makes sense to me that we pass it platform-specific flags. The knowledge that Linux binaries need the extra flag on OS X is pretty weird and arbitrary, but since a more principled solution is not presenting itself, let's go for the narrow/arbitrary solution for now.

[haberman]

This should be fixed by #78.