[Bug c++/12273] New: demangled symbol names unrecognised in non-debug builds

taviso at cmpxchg8b dot com sourceware-bugzilla@sourceware.org
Mon Nov 29 17:29:00 GMT 2010


           Summary: demangled symbol names unrecognised in non-debug
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned@sourceware.org
        ReportedBy: taviso@cmpxchg8b.com

DSOs that export mangled symbols are unrecognised in their mangled form this
used to work in 7.1, but for some reason doesn't seem to work any more. I'm not
sure what changed that caused it to break.

You can reproduce the problem like this, create an object with an exported
mangled symbol:

$ cat dynsym.cc
#include <stdio.h>

class GDB {
   static int whatever (void) {
     return fprintf(stderr, "Hello World\n");

int main(int argc, char **argv)
   return GDB::whatever();
$ g++ -export-dynamic dynsym.cc -s -o dynsym

Verify the symbol is present:

$ nm -D dynsym | c++filt | grep GDB
00000000004008ca W GDB::whatever()

Try to break on the demangled symbol name:

$ gdb -q ./dynsym
(gdb) b GDB::whatever()
Can't find member of namespace, class, struct, or union named "GDB::whatever"
Hint: try 'GDB::whatever()<TAB> or 'GDB::whatever()<ESC-?>
(Note leading single quote.)

Note that if I use the mangled symbol name directly, GDB accepts it.

(gdb) b _ZN3GDB8whateverEv
Breakpoint 1 at 0x4008ce
(gdb) r
Breakpoint 1, 0x00000000004008ce in GDB::whatever() ()

I don't think this is a problem with not recognising the symbols, as
it's listed as a completion target if I try to complete on symbol
names, and info address works:

(gdb) info address GDB::whatever()
Symbol "GDB::whatever()" is at 0x4008ca in a file compiled without debugging.

This used to work, in a vanilla GDB 7.1:

$ gdb dynsym
GNU gdb (GDB) 7.1
(gdb) b GDB::whatever()
Breakpoint 1 at 0x8048639

Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

More information about the Gdb-prs mailing list