[LLD] [COFF] Demangle itanium symbols in mingw mode

[mstorsjo]

Summary

Maybe the msvc demangling should be within if (!mingw)? Performance wise it's probably irrelevant, but the itanium demangling, in combination with dllimport and i386 leading underscores, ended up as a few more lines, so an if (mingw) around it makes it a bit clearer.

[ruiu]

I'm not very familiar with MinGW, but is Itanium name mangling in use on MinGW? MinGW is intended to be a native development environment for Windows, so I thought that they naturally use the platform way of mangling symbols.

[mstorsjo]

In D67051#1654132, @ruiu wrote:

I'm not very familiar with MinGW, but is Itanium name mangling in use on MinGW? MinGW is intended to be a native development environment for Windows, so I thought that they naturally use the platform way of mangling symbols.

Yes, MinGW uses Itanium name mangling only.

Yes, MinGW is intended to be a native development environment, but despite this (for historical reasons) it has done a lot of things very differently, due to being based on the GNU tools originally, and generally doing unixy things in COFF files. GCC doesn't support the MSVC C++ ABI at all, and the name mangling is only a very small part of the whole C++ ABI.

Generally the level of compatibility is that you can call between DLLs built by different tools, as long as it only has a pure C API, but you shouldn't expect to mix object files (or static libraries) between MSVC/link.exe and GCC/ld.bfd.

[ruiu]

LGTM

Thank you for the explanation!

[rnk]

lld/trunk/COFF/Symbols.cpp

31

It is trivial to disambiguate Itanium and MSVC symbols just by looking at the prefix. For both mangling types, we should ignore __imp_ prefixes in the same way. Then, Itanium names match _+Z, and Microsoft names start with ?.

I think we should get rid of the two lld::demangleItanium/MSVC variants, add a check for ? in llvm::demangle, and standardize on that. The COFF demangle helper can do the __imp_ handling. Make sense?

[mstorsjo]

lld/trunk/COFF/Symbols.cpp

31

Sounds sensible, I'll put it on my backlog and give it a shot.

lld/trunk/COFF/Symbols.cpp

 
 namespace lld {
 
 static_assert(sizeof(SymbolUnion) <= 48,
               "symbols should be optimized for memory usage");
 
 // Returns a symbol name for an error message.
 static std::string demangle(StringRef symName) {
-  if (config->demangle)
+  if (config->demangle) {
     if (Optional<std::string> s = demangleMSVC(symName))
       return *s;
+    if (config->mingw) {
+      StringRef demangleInput = symName;
+      std::string prefix;
+      if (demangleInput.consume_front("__imp_"))
+        prefix = "__declspec(dllimport) ";
+      if (config->machine == I386)
+        demangleInput.consume_front("_");
+      if (Optional<std::string> s = demangleItanium(demangleInput))
+        return prefix + *s;
+    }
+  }
   return symName;
 }
 std::string toString(coff::Symbol &b) { return demangle(b.getName()); }
 std::string toCOFFString(const Archive::Symbol &b) {
   return demangle(b.getName());
 }
 
 namespace coff {

lld/trunk/test/COFF/undefined-symbol-itanium-i386.s

+# REQUIRES: x86
+# RUN: llvm-mc -triple=i386-windows-gnu -filetype=obj -o %t.o %s
+# RUN: not lld-link /lldmingw /out:%t.exe %t.o 2>&1 | FileCheck %s
+# RUN: not lld-link /lldmingw /out:%t.exe /demangle %t.o 2>&1 | FileCheck %s
+# RUN: not lld-link /lldmingw /out:%t.exe /demangle:no %t.o 2>&1 | FileCheck --check-prefix=NODEMANGLE %s
+
+# NODEMANGLE: error: undefined symbol: __Z3fooi
+# NODEMANGLE: error: undefined symbol: __Z3barPKc
+# NODEMANGLE: error: undefined symbol: __imp___Z3bazv
+
+# CHECK: error: undefined symbol: foo(int)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(_main)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(_main)
+# CHECK-EMPTY:
+# CHECK-NEXT: error: undefined symbol: bar(char const*)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(_main)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(_f1)
+# CHECK-EMPTY:
+# CHECK-NEXT: error: undefined symbol: __declspec(dllimport) baz()
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(_f2)
+
+        .section        .text,"xr",one_only,_main
+.globl _main
+_main:
+  call  __Z3fooi
+  call  __Z3fooi
+  call  __Z3barPKc
+
+_f1:
+  call  __Z3barPKc
+.Lfunc_end1:
+
+        .section        .text,"xr",one_only,_f2
+.globl _f2
+_f2:
+  call  *__imp___Z3bazv

lld/trunk/test/COFF/undefined-symbol-itanium.s

+# REQUIRES: x86
+# RUN: llvm-mc -triple=x86_64-windows-gnu -filetype=obj -o %t.o %s
+# RUN: not lld-link /lldmingw /out:%t.exe %t.o 2>&1 | FileCheck %s
+# RUN: not lld-link /lldmingw /out:%t.exe /demangle %t.o 2>&1 | FileCheck %s
+# RUN: not lld-link /lldmingw /out:%t.exe /demangle:no %t.o 2>&1 | FileCheck --check-prefix=NODEMANGLE %s
+
+# NODEMANGLE: error: undefined symbol: _Z3fooi
+# NODEMANGLE: error: undefined symbol: _Z3barPKc
+# NODEMANGLE: error: undefined symbol: __imp__Z3bazv
+
+# CHECK: error: undefined symbol: foo(int)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(main)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(main)
+# CHECK-EMPTY:
+# CHECK-NEXT: error: undefined symbol: bar(char const*)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(main)
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(f1)
+# CHECK-EMPTY:
+# CHECK-NEXT: error: undefined symbol: __declspec(dllimport) baz()
+# CHECK-NEXT: >>> referenced by {{.*}}.o:(f2)
+
+        .section        .text,"xr",one_only,main
+.globl main
+main:
+  call  _Z3fooi
+  call  _Z3fooi
+  call  _Z3barPKc
+
+f1:
+  call  _Z3barPKc
+.Lfunc_end1:
+
+        .section        .text,"xr",one_only,f2
+.globl f2
+f2:
+  callq *__imp__Z3bazv(%rip)