-frepo problems

Jeffrey A Law law@cygnus.com
Tue Jan 12 03:15:00 GMT 1999


  In message < 19990110164644.22037@liafa1.liafa.jussieu.fr >you write:
  > This is quite enough for a diagnostic... It will probably get a while
  > until I fix it, as the precise details of name-mangling are not very
  > thoroughly documented, so I will have to go hunting to find what's going
  > on... pointers on where I should start looking are VERY welcome.
  > 
  > With the `newer' OpenBSD egcs configuration, that uses VTABLE_THUNKS,
  > name-mangling is already different...  it is definitely an internal bug 
  > of egcs, as c++filt should be able to reverse the mangling and give proper 
  > names...
  > 
[ ... ]
  > 
  > So the names are unmangled correctly, apparently, though eg++ is not called
  > ,
  > as it should.
  > The rpo file looks like:
  > M dumbrepo.C
  > D /home/espie/Projects/egcs
  > A '-frepo' '-c'
  > O __cl__Ct7printer1i1
  > O __vt_t7printer1i1
  > and c++filt yields:
  > M dumbrepo.C
  > D /home/espie/Projects/egcs
  > A '-frepo' '-c'
  > O printer<1>::_cl(void) const
  > O __vt_t7printer1i1
  > 
  > which is not very encouraging either.
Hmmm, earlier you said it was an "internal bug of egcs".  I think you meant
the demangler.  Being specific always helps :-)

Yes, thunks changes mangling in case you're wondering.

That's about all I can help with.  This stuff isn't my area of expertise.

jeff



More information about the Gcc-bugs mailing list