This is the mail archive of the binutils@sourceware.cygnus.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: linker problem with WinCE exe's (2)


   Date: Fri, 04 Jun 1999 23:15:19 +0200
   From: "Richard Wisenöcker" <richard.wisenoecker@gmx.at>

   1) symbol->value:
    Is also 0, it seems that idata$x is not like a symbol (look to the linker output at the
   end of the mail).

The value is 0 is reasonable.  It means that the symbol is defined at
the start of the section.

Most of the appearances of idata$x in the map file you sent are as
names of input sections, not symbol names.  The names on the left are
all section names.  You can also see section names by running objdump
-h on an object file.

   2) symbol->section->output_section->vma:
    Is always 0x14000, independent if it is idata$2, idata$3 or any other

The map file shows that all of the idata$x input sections are being
put into the same output section, so for all of them
symbol->section->output_section should be the same, so of course the
vma field is always the same.

   3) symbol->section->output_offset:
   Is in all cases (idata$2, idata$3,...) equal to 0 too.

This I don't understand.  The map file shows these sections appearing
at different offsets.  I don't know why output_offset would be 0.
This is probably what you should examine.

   4) about _bfd_coff_generic_relocate_section:
   I had just a short look, but I will study it in more detail. But in my case I try to
   generate MIPS code for WinCE.

I assume that WinCE, like other versions of Windows, uses the PE
object file format.  PE is a variant of COFF, and BFD uses the COFF
support as a basis for handling it.

Ian

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]