Re: [PATCH] bfd robustification (arithmetic overflows on allocationin elf.c and corrupt version section handling)

[Nick Clifton <nickc at redhat dot com>]

Hi Jakub,

> > But ... have you looked at PR binutils/868 ?  It applies particularly to 
> > the use of the realloc() function, but it might be worth considering in 
> > the wider context of the changes that you are making.

> I think that should be orthogonal to those changes.  So bfd could provide
> bfd_realloc, bfd_realloc_or_free, bfd_realloc2 and bfd_realloc2_or_free.

Agreed.  I just wanted to bring it to your attention since you are 
cleaning up code in this area.


> > > +#define HALF_BFD_SIZE_TYPE \
> > > +  (((bfd_size_type) 1) << (8 * sizeof (bfd_size_type) / 2))
> >
> > Hmm, does this 8 assume that a byte is an 8-bit quantity ?

> It does, but I thought we rely on the host char being 8 bit,
> only target char can be different.  I could use CHAR_BIT instead,
> but nothing in bfd/binutils/ld uses CHAR_BIT ATM, so it would
> surprise me if it worked at all with CHAR_BIT != 8.

Yeah me too.  I only raised this issue since I have this paranoia about 
magic constants appearing in the code.

Cheers
  Nick