Re: [RFA/PATCH] auto-import errors & DWARF 2 (was Re: _cooked_sizeusually set when/where?)

[Brian Ford <ford at vss dot fsi dot com>]

On Wed, 14 Apr 2004, Alan Modra wrote:

> Would you mind trying out this patch?  The reloc_done fudge isn't
> needed any more.

Great!  That's a much nicer patch.

Although I did not test it extensively, I did confirm that it solves my
original problem.  If you are confident it will not cause issues with
other targets, please apply.

> 	* simple.c (bfd_simple_get_relocated_section_contents): Don't
> 	change reloc_done.  Set and restore _cooked_size.
> 	(RETURN): Delete.
>
> Index: bfd/simple.c
> ===================================================================
> RCS file: /cvs/src/src/bfd/simple.c,v
> retrieving revision 1.14
> diff -u -p -r1.14 simple.c
> --- bfd/simple.c	30 Nov 2003 18:40:41 -0000	1.14
> +++ bfd/simple.c	14 Apr 2004 09:08:10 -0000
> @@ -140,28 +140,7 @@ bfd_simple_get_relocated_section_content
>    bfd_byte *contents, *data;
>    int storage_needed;
>    void *saved_offsets;
> -  bfd_boolean saved_reloc_done = sec->reloc_done;
> -
> -#undef RETURN
> -#define RETURN(x)				\
> -  do						\
> -    {						\
> -      sec->reloc_done = saved_reloc_done;	\
> -      return (x);				\
> -    }						\
> -  while (0)
> -
> -  /* Foul hack to prevent bfd_section_size aborts.  The reloc_done flag
> -     only controls that macro (and the related size macros), selecting
> -     between _raw_size and _cooked_size.  We may be called with relocation
> -     done or not, so we need to save the done-flag and mark the section as
> -     not relocated.
> -
> -     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);
> -  sec->reloc_done = 0;
> +  bfd_size_type old_cooked_size;
>
>    if (! (sec->flags & SEC_RELOC))
>      {
> @@ -175,7 +154,7 @@ bfd_simple_get_relocated_section_content
>        if (contents)
>  	bfd_get_section_contents (abfd, sec, contents, 0, size);
>
> -      RETURN (contents);
> +      return contents;
>      }
>
>    /* In order to use bfd_get_relocated_section_contents, we need
> @@ -205,7 +184,7 @@ bfd_simple_get_relocated_section_content
>      {
>        data = bfd_malloc (bfd_section_size (abfd, sec));
>        if (data == NULL)
> -	RETURN (NULL);
> +	return NULL;
>        outbuf = data;
>      }
>
> @@ -224,7 +203,7 @@ bfd_simple_get_relocated_section_content
>      {
>        if (data)
>  	free (data);
> -      RETURN (NULL);
> +      return NULL;
>      }
>    bfd_map_over_sections (abfd, simple_save_output_info, saved_offsets);
>
> @@ -239,6 +218,12 @@ bfd_simple_get_relocated_section_content
>    else
>      storage_needed = 0;
>
> +  /* This function might be called before _cooked_size has been set, and
> +     bfd_perform_relocation needs _cooked_size to be valid.  */
> +  old_cooked_size = sec->_cooked_size;
> +  if (sec->_cooked_size == 0)

I would have used old_coocked_size here for the test since it is an auto.
But then again, I'm a realtime programmer obsessed with these silly little
"efficiencies", and I've learned not to trust the compiler to optimize
anything :).

> +    sec->_cooked_size = sec->_raw_size;
> +
>    contents = bfd_get_relocated_section_contents (abfd,
>  						 &link_info,
>  						 &link_order,
> @@ -262,10 +247,11 @@ bfd_simple_get_relocated_section_content
>      free (symbol_table);
>  #endif
>
> +  sec->_cooked_size = old_cooked_size;
>    bfd_map_over_sections (abfd, simple_restore_output_info, saved_offsets);
>    free (saved_offsets);
>
>    _bfd_generic_link_hash_table_free (link_info.hash);
>
> -  RETURN (contents);
> +  return contents;
>  }

Thanks again.

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