_cooked_size usually set when/where?

[Brian Ford]

We are trying to add support for DWARF2 to x86 PE/COFF (Cygwin & mingw).

I found the following problem (due to another bug, but anyway...).

failure_report.o(.text+0x10a)/home/ford/local2/lib/gcc/i686-pc-cygwin/3.4.0/../../../../i686-pc-cygwin/bin/ld:
BFD 2.15.90 20040329 assertion fail ../../sources/bfd/simple.c:163

which is:

     Debug sections won't change size while we're only relocating.  There
     may be trouble here someday if it tries to run relaxation
     unexpectedly, so make sure.  */
  BFD_ASSERT (sec->_raw_size == sec->_cooked_size);

_raw_size is the expected size, but _cooked_size is 0.  It goes on to:

/home/ford/local2/lib/gcc/i686-pc-cygwin/3.4.0/../../../../i686-pc-cygwin/bin/ld:
BFD 2.15.90 20040329 internal error, aborting at ../../sources/bfd/reloc.c
line 4280 in bfd_generic_get_relocated_section_contents

due to bfd_perform_relocation returning bfd_reloc_outofrange due to the
above.

/home/ford/local2/lib/gcc/i686-pc-cygwin/3.4.0/../../../../i686-pc-cygwin/bin/ld:
Please report this bug.

Info: resolving _XtStrings by linking to __imp__XtStrings (auto-import)

(Incidentally, I don't know why auto-import is broken yet, nor do I know
if it is even due to my local patches.  One bug at time :-)

So, my question is, when is _cooked_size usually set up and by whom?

Thanks for your help.

-- 
Brian Ford
Senior Realtime Software Engineer
VITAL - Visual Simulation Systems
FlightSafety International
Phone: 314-551-8460
Fax:   314-551-8444