bfd/elf.c : bfd_seek to 0 valid ?

[jeroen dobbelaere]

Richard Earnshaw wrote:
[..]

> 
> 
> Well, since pos is calculated to be hdr->sh_offset + offset, and since it 
> seams reasonable for offset to be zero, the question has to be (assuming 
> offset can't be negative!) is it legitimate for hdr->sh_offset to zero.
> 
> My guess is that the answer to that question should normally be 'no', and 
> the reason for it being so is that the section hasn't been mapped into the 
> output bfd.
> 
> R.
> 

section->name = '.rel.dyn' (output section); the content itself comes from a '.rel.got'.

I have the impression, that for some reason, the .rel.dyn section does not
get a fileposition assigned (resulting in the hdr->sh_offset == 0).

Where should this mapping to the output bfd happen ?

Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes