Re: Strange reporting of symbol addresses for a Windows DLL

[Nick Clifton <nickc at redhat dot com>]

Hi Chris,

: ftp://ftp.cygnus.com/pub/home/cgf/new-cygwin1.dll.bz2

Thanks.

It appears that the second addition is happening in
_bfd_pe_get_symbol_info.  The patch below gives the correct numbers
and does not appear to introduce any bew failures.  Please give it a
try and let me know how it goes.

Cheers
	Nick

Index: bfd/peigen.c
===================================================================
RCS file: /cvs/src//src/bfd/peigen.c,v
retrieving revision 1.15
diff -p -r1.15 peigen.c
*** peigen.c	2000/12/14 21:38:30	1.15
--- peigen.c	2001/02/08 22:50:56
*************** _bfd_pe_get_symbol_info (abfd, symbol, r
*** 1958,1969 ****
       symbol_info *ret;
  {
    coff_get_symbol_info (abfd, symbol, ret);
! 
    if (pe_data (abfd) != NULL
        && ((symbol->flags & BSF_DEBUGGING) == 0
  	  || (symbol->flags & BSF_DEBUGGING_RELOC) != 0)
        && ! bfd_is_abs_section (symbol->section))
      ret->value += pe_data (abfd)->pe_opthdr.ImageBase;
  }
  
  /* Handle the .idata section and other things that need symbol table
--- 1958,1971 ----
       symbol_info *ret;
  {
    coff_get_symbol_info (abfd, symbol, ret);
! #if 0 /* This code no longer appears to be necessary.
! 	 ImageBase has already been added in by coff_swap_scnhdr_in.  */
    if (pe_data (abfd) != NULL
        && ((symbol->flags & BSF_DEBUGGING) == 0
  	  || (symbol->flags & BSF_DEBUGGING_RELOC) != 0)
        && ! bfd_is_abs_section (symbol->section))
      ret->value += pe_data (abfd)->pe_opthdr.ImageBase;
+ #endif
  }
  
  /* Handle the .idata section and other things that need symbol table