Re: [arm-gnu] Segmentation fault using --be8 option

[Nick Clifton <nickc@xxxxxxxxxx>]

Hi Carlos,

> I reproduced the problem with the following testcase:
>
> echo "EOF" >> test.c
> int _start (void)
> {
>   return 0;
> }
> EOF
> arm-none-linux-gnueabi-gcc -mbig-endian -c test.c
> arm-none-linux-gnueabi-ld -EB --be8 -o test test.o
> Segmentation fault.

Thanks for the test case.  I was able to reproduce the problem using 
this and it turned out to be a simple thinko in the code to handle the 
byte swapping of mapping symbols.  It is fixed by the attached patch.

Cheers
  Nick

bfd/ChangeLog
2006-01-11  Nick Clifton  <nickc@xxxxxxxxxx>

	* elf32-arm.c (elf32_arm_output_symbol_hook): Install new entry
	into the newly (re)allocated map array.

Index: bfd/elf32-arm.c
===================================================================
RCS file: /cvs/src/src/bfd/elf32-arm.c,v
retrieving revision 1.61
diff -c -3 -p -r1.61 elf32-arm.c
*** bfd/elf32-arm.c	12 Dec 2005 17:03:39 -0000	1.61
--- bfd/elf32-arm.c	11 Jan 2006 18:04:32 -0000
*************** elf32_arm_output_symbol_hook (struct bfd
*** 7459,7464 ****
--- 7459,7465 ----
  
    mapcount = arm_data->mapcount + 1;
    map = arm_data->map;
+ 
    /* TODO: This may be inefficient, but we probably don't usually have many
       mapping symbols per section.  */
    newmap = bfd_realloc (map, mapcount * sizeof (* map));
*************** elf32_arm_output_symbol_hook (struct bfd
*** 7467,7474 ****
        arm_data->map = newmap;
        arm_data->mapcount = mapcount;
  
!       map[mapcount - 1].vma = elfsym->st_value;
!       map[mapcount - 1].type = name[1];
      }
  
    return TRUE;
--- 7468,7475 ----
        arm_data->map = newmap;
        arm_data->mapcount = mapcount;
  
!       newmap[mapcount - 1].vma = elfsym->st_value;
!       newmap[mapcount - 1].type = name[1];
      }
  
    return TRUE;