Bug#263601: m68k: workaround for binutils problem
At Thu, 05 Aug 2004 12:32:51 +0200,
Roman Zippel wrote:
> The new binutils 2.15 package combined with newer gcc releases might
> expose a longstanding binutils problem, it's discussed here:
> http://sources.redhat.com/ml/binutils/2002-08/msg00535.html
> I'm running tests with newer gcc and since the upgrade to binutils 2.15
> some of the tests fail because of this.
> It's rather unlikely that this problem is properly fixed in binutils
> soon, so I think we should include this patch to avoid further problems
> with the pending release.
At Fri, 6 Aug 2004 06:34:37 +0200,
Matthias Klose wrote:
> according to Dan binutils won't be fixed/cannot be fixed for
> sarge. Please apply this workaround.
OK... I put it.
I discussed with this issues and taught some things from Daniel on IRC
(thanks Daniel!). AFAIK:
- R_***_NONE is used to fill NULL out for not fixed relocations.
It's old binutils problem, newer binutils can handle it. m68k is
the last architecture to be updated.
- R_68K_NONE is sometimes used. No one is quite sure what is
changed.
- ia64, alpha, sparc have R_***_NONE code in elf_machine_lazy_rel,
but it seems being for backward compatibility.
- No one has been worked for this problem. But adding this code
should be harmless. If we fix it, we need to rewrite some
relocation routines.
- We may be able to have this code for a while.
Regards,
-- gotom
Reply to: