Daniel Jacobowitz escreveu:
> On Fri, Nov 17, 2006 at 01:13:14AM +0000, Pedro Alves wrote:
>   
>> * symtab.h (prim_record_minimal_symbol_and_bfd_section): Declare.
>>     
>
> I don't think you need yet another variant of this function.  Can't you
> use prim_record_minimal_symbol_and_info?  You're getting the section
> from the cs, and there's cs_to_section, so you ought to have the
> "section" integer available too.
>
>
>   
You are right. My thinking was that prim_record_minimal_symbol will only 
pass
.text .data or .bss section to prim_record_minimal_symbol_and_info, and
the way you say, and in the patch attached, we are passing a wider range 
of values.
I didn't want to change more than needed, but looking again, I don't 
think it harms
in this case.

The attached patch does that, and also, I took the opportunity to a 
little extra cleanup
on coffread.c by making all minimal symbol creations through 
record_minimal_symbol.

Cheers,
Pedro Alves

---

2006-11-19  Pedro Alves  <pedro_alves@portugalmail.pt>

        * coffread.c (cs_to_bfd_section): New function.
        (cs_to_section): Use cs_to_bfd_section.
        (record_minimal_symbol): Take the coff_symbol* parameter instead
        of the symbol's name as a char*.
        Add 'int section' parameter. Call 
prim_record_minimal_symbol_and_info
        instead of prim_record_minimal_symbol_and_info.
        Change return type to struct minimal_symbol *.
        (coff_symtab_read): Adapt to new record_minimal_symbol's signature.
        Make all minimal symbol creations go through record_minimal_symbol.