[lld] Implement --dynamic-list-cpp-{new,typeinfo}

[zatrazz]

Summary

This patch implements both the --dynamic-list-cpp-new and
--dynamic-list-cpp-typeinfo options. Former add all the
C++ operator new and delete symbols on the dynamic list
while latter add all the runtime type identification.

This patch depends on http://reviews.llvm.org/D18771,
http://reviews.llvm.org/D18772, and http://reviews.llvm.org/D18773

[ruiu]

Do you need this feature? This feature seems to be designed for a very specific purpose, so I think you want to first explain why we must support these options.

[zatrazz]

In D18774#391687, @ruiu wrote:

Do you need this feature? This feature seems to be designed for a very specific purpose, so I think you want to first explain why we must support these options.

This is mainly used when building shared libraries with some -Bsymbolic or -Bsymbol-function while still allowing the programs to override new/delete if it this case. Indeed this is a very narrow cara and have added it only add parity check with ld/gold. I can drop the patch if the usercase is not strong enough.

[ruiu]

I'd drop this because this is probably a bit too much to be added to LLD if the reason is just for the sake of compatibility with gnu ld/gold. We can discuss that later if someone comes to us and why they need this option.

ELF/Config.h

struct Configuration {
   bool Verbose;
   bool WarnCommon;
   bool ZExecStack;
   bool ZNodelete;
   bool ZNow;
   bool ZOrigin;
   bool ZRelro;
   bool DynamicListData;
+  bool DynamicListCppNew;
+  bool DynamicListCppTypeinfo;
   ELFKind EKind = ELFNoneKind;
   uint16_t EMachine = llvm::ELF::EM_NONE;
   uint64_t EntryAddr = -1;
   unsigned LtoO;
   unsigned Optimize;
 };
 
 // The only instance of Configuration struct.
 extern Configuration *Config;
 
 } // namespace elf
 } // namespace lld
 
 #endif

ELF/Driver.cpp

void LinkerDriver::readConfigs(opt::InputArgList &Args) {
   Config->Init = getString(Args, OPT_init, "_init");
   Config->OutputFile = getString(Args, OPT_o);
   Config->SoName = getString(Args, OPT_soname);
   Config->Sysroot = getString(Args, OPT_sysroot);
   StringRef DynamicListPath = getString(Args, OPT_dynamic_list);
   if (!DynamicListPath.empty())
     readDynamicList(DynamicListPath);
   Config->DynamicListData = Args.hasArg(OPT_dynamic_list_data);
+  Config->DynamicListCppNew = Args.hasArg(OPT_dynamic_list_cpp_new);
+  Config->DynamicListCppTypeinfo = Args.hasArg(OPT_dynamic_list_cpp_typeinfo);
 
   Config->Optimize = getInteger(Args, OPT_O, 0);
   Config->LtoO = getInteger(Args, OPT_lto_O, 2);
   if (Config->LtoO > 3)
     error("invalid optimization level for LTO: " + getString(Args, OPT_lto_O));
 
   Config->ZExecStack = hasZOption(Args, "execstack");
   Config->ZNodelete = hasZOption(Args, "nodelete");

ELF/Options.td

def z : JoinedOrSeparate<["-"], "z">, MetaVarName<"<option>">,
   HelpText<"Linker option extensions">;
 
 def dynamic_list : Separate<["--", "-"], "dynamic-list">,
   HelpText<"Dynamic list to use">;
 
 def dynamic_list_data : Flag<["--", "-"], "dynamic-list-data">,
   HelpText<"Add all global data to the dynamic symbol list">;
 
+def dynamic_list_cpp_new : Flag<["--"], "dynamic-list-cpp-new">,
+  HelpText<"Add the builtin C++ operator new and delete on the dynamic symbol list">;
+
+def dynamic_list_cpp_typeinfo : Flag<["--"], "dynamic-list-cpp-typeinfo">,
+  HelpText<"Add the builtin C++ runtime type indentification on the dynamic symbol list">;
+
 // Aliases
 def alias_Bdynamic_call_shared: Flag<["-"], "call_shared">, Alias<Bdynamic>;
 def alias_Bdynamic_dy: Flag<["-"], "dy">, Alias<Bdynamic>;
 def alias_Bstatic_dn: Flag<["-"], "dn">, Alias<Bstatic>;
 def alias_Bstatic_non_shared: Flag<["-"], "non_shared">, Alias<Bstatic>;
 def alias_Bstatic_static: Flag<["-"], "static">, Alias<Bstatic>;
 def alias_L__library_path : Joined<["--"], "library-path=">, Alias<L>;
 def alias_discard_all_x: Flag<["-"], "x">, Alias<discard_all>;

ELF/Writer.cpp

template <class ELFT> static bool includeInSymtab(const SymbolBody &B) {
   if (auto *D = dyn_cast<DefinedRegular<ELFT>>(&B)) {
     // Exclude symbols pointing to garbage-collected sections.
     if (D->Section && !D->Section->Live)
       return false;
   }
   return true;
 }
 
+#define STRCOMPARE(__str, __cstr) \
+  (__str.compare(0, sizeof(__cstr)-1, __cstr) == 0)
+
+static inline bool CheckNewPrefix(const std::string &N) {
+  return STRCOMPARE (N, "operator new") || STRCOMPARE (N, "operator delete");
+}
+
+static inline bool CheckTypeInfoPrefix(const std::string &N) {
+  return STRCOMPARE (N, "typeinfo name for") ||
+         STRCOMPARE (N, "typeinfo for");
+}
+
 template <class ELFT>
 static bool includeInDynsym(const SymbolBody &B) {
   uint8_t V = B.getVisibility();
   if (V != STV_DEFAULT && V != STV_PROTECTED)
     return false;
   if (Config->ExportDynamic || Config->Shared)
     return true;
   if (DynList->isInDynamicList(B.getName()))
     return true;
   // Handle --dynamic-list-data.
   if (Config->DynamicListData)
     if (B.Type == STT_OBJECT)
       return true;
+  // Handle --dynamic-list-cpp-new and --dynamic-list-cpp-typeinfo
+  if (Config->DynamicListCppNew) {
+    if (CheckNewPrefix(demangle(B.getName())))
+      return true;
+  }
+  if (Config->DynamicListCppTypeinfo) {
+    if (CheckTypeInfoPrefix(demangle(B.getName())))
+      return true;
+  }
   return B.MustBeInDynSym;
 }
 
 // This class knows how to create an output section for a given
 // input section. Output section type is determined by various
 // factors, including input section's sh_flags, sh_type and
 // linker scripts.
 namespace {

test/ELF/dynamic-list-new-typeinfo.s

+# REQUIRES: x86
+
+## This object is build in linked against to force dynamic loading without
+## the use of full paths in linking command.
+# RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %p/Inputs/shared.s -o %t2.o
+# RUN: ld.lld -shared %t2.o -soname shared -o %t2.so
+
+# Check export new and delete symbols.
+# RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %s -o %t
+# RUN: ld.lld --dynamic-list-cpp-new %t %t2.so -o %t.exe
+# RUN: llvm-readobj -dyn-symbols %t.exe | FileCheck %s
+
+# CHECK:      DynamicSymbols [
+# CHECK-NEXT:   Symbol {
+# CHECK-NEXT:     Name: @ (0)
+# CHECK-NEXT:     Value: 0x0
+# CHECK-NEXT:     Size: 0
+# CHECK-NEXT:     Binding: Local (0x0)
+# CHECK-NEXT:     Type: None (0x0)
+# CHECK-NEXT:     Other: 0
+# CHECK-NEXT:     Section: Undefined (0x0)
+# CHECK-NEXT:   }
+# CHECK-NEXT:   Symbol {
+# CHECK-NEXT:     Name: _ZdaPv@ (1)
+# CHECK-NEXT:     Value: 0x11003
+# CHECK-NEXT:     Size: 0
+# CHECK-NEXT:     Binding: Global (0x1)
+# CHECK-NEXT:     Type: None (0x0)
+# CHECK-NEXT:     Other: 0
+# CHECK-NEXT:     Section: .text (0x4)
+# CHECK-NEXT:   }
+# CHECK-NEXT:   Symbol {
+# CHECK-NEXT:     Name: _ZdlPv@ (8)
+# CHECK-NEXT:     Value: 0x11002
+# CHECK-NEXT:     Size: 0
+# CHECK-NEXT:     Binding: Global (0x1)
+# CHECK-NEXT:     Type: None (0x0)
+# CHECK-NEXT:     Other: 0
+# CHECK-NEXT:     Section: .text (0x4)
+# CHECK-NEXT:   }
+# CHECK-NEXT:   Symbol {
+# CHECK-NEXT:     Name: _Znam@ (15)
+# CHECK-NEXT:     Value: 0x11001
+# CHECK-NEXT:     Size: 0
+# CHECK-NEXT:     Binding: Global (0x1)
+# CHECK-NEXT:     Type: None (0x0)
+# CHECK-NEXT:     Other: 0
+# CHECK-NEXT:     Section: .text (0x4)
+# CHECK-NEXT:   }
+# CHECK-NEXT:   Symbol {
+# CHECK-NEXT:     Name: _Znwm@ (21)
+# CHECK-NEXT:     Value: 0x11000
+# CHECK-NEXT:     Size: 0
+# CHECK-NEXT:     Binding: Global (0x1)
+# CHECK-NEXT:     Type: None (0x0)
+# CHECK-NEXT:     Other: 0
+# CHECK-NEXT:     Section: .text (0x4)
+# CHECK-NEXT:   }
+# CHECK-NEXT: ] 
+
+# Check export typeinfo symbols.
+# RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %s -o %t
+# RUN: ld.lld --dynamic-list-cpp-typeinfo %t %t2.so -o %t.exe
+# RUN: llvm-readobj -dyn-symbols %t.exe | FileCheck -check-prefix=CHECK2 %s
+
+# CHECK2:      DynamicSymbols [
+#CHECK2-NEXT:    Symbol {
+#CHECK2-NEXT:      Name: @ (0)
+#CHECK2-NEXT:      Value: 0x0
+#CHECK2-NEXT:      Size: 0
+#CHECK2-NEXT:      Binding: Local (0x0)
+#CHECK2-NEXT:      Type: None (0x0)
+#CHECK2-NEXT:      Other: 0
+#CHECK2-NEXT:      Section: Undefined (0x0)
+#CHECK2-NEXT:    }
+#CHECK2-NEXT:    Symbol {
+#CHECK2-NEXT:      Name: _ZTI3Bar@ (1)
+#CHECK2-NEXT:      Value: 0x11004
+#CHECK2-NEXT:      Size: 8
+#CHECK2-NEXT:      Binding: Global (0x1)
+#CHECK2-NEXT:      Type: Object (0x1)
+#CHECK2-NEXT:      Other: 0
+#CHECK2-NEXT:      Section: .text (0x4)
+#CHECK2-NEXT:    }
+#CHECK2-NEXT:    Symbol {
+#CHECK2-NEXT:      Name: _ZTI3Foo@ (10)
+#CHECK2-NEXT:      Value: 0x11004
+#CHECK2-NEXT:      Size: 8
+#CHECK2-NEXT:      Binding: Global (0x1)
+#CHECK2-NEXT:      Type: Object (0x1)
+#CHECK2-NEXT:      Other: 0
+#CHECK2-NEXT:      Section: .text (0x4)
+#CHECK2-NEXT:    }
+#CHECK2-NEXT:    Symbol {
+#CHECK2-NEXT:      Name: _ZTS3Bar@ (19)
+#CHECK2-NEXT:      Value: 0x11004
+#CHECK2-NEXT:      Size: 5
+#CHECK2-NEXT:      Binding: Global (0x1)
+#CHECK2-NEXT:      Type: Object (0x1)
+#CHECK2-NEXT:      Other: 0
+#CHECK2-NEXT:      Section: .text (0x4)
+#CHECK2-NEXT:    }
+#CHECK2-NEXT:    Symbol {
+#CHECK2-NEXT:      Name: _ZTS3Foo@ (28)
+#CHECK2-NEXT:      Value: 0x11009
+#CHECK2-NEXT:      Size: 5
+#CHECK2-NEXT:      Binding: Global (0x1)
+#CHECK2-NEXT:      Type: Object (0x1)
+#CHECK2-NEXT:      Other: 0
+#CHECK2-NEXT:      Section: .text (0x4)
+#CHECK2-NEXT:    }
+# CHECK2-NEXT: ] 
+
+## operator new (size_t)
+.globl _Znwm
+_Znwm:
+  ret
+
+## operator new[] (size_t)
+.globl _Znam
+_Znam:
+  ret
+
+## operator delete (void *)
+.globl _ZdlPv
+_ZdlPv:
+  ret
+
+## operator delete[] (void *)
+.globl _ZdaPv
+_ZdaPv:
+  ret
+
+## typeinfo for Bar
+.type  _ZTI3Bar, @object
+.size  _ZTI3Bar, 8
+.globl _ZTI3Bar
+_ZTI3Bar:
+.quad
+
+## typeinfo for Foo
+.type  _ZTI3Foo, @object
+.size  _ZTI3Foo, 8
+.globl _ZTI3Foo
+_ZTI3Foo:
+.quad
+
+## typeinfo name for Bar
+.type  _ZTS3Bar, @object
+.size  _ZTS3Bar, 5
+.globl _ZTS3Bar
+_ZTS3Bar:
+.string "3Bar"
+
+## typeinfo name for Foo
+.type  _ZTS3Foo, @object
+.size  _ZTS3Foo, 5
+.globl _ZTS3Foo
+_ZTS3Foo:
+.string "3Foo"
+
+.globl _start
+_start:
+  ret