[RFA, ARM] Fix line number data for PIC register setup code

[Ulrich Weigand]

Richard Earnshaw wrote:
> On 20 Dec 2010, at 21:06, "Richard Henderson" <rth@redhat.com> wrote:
> > On 12/20/2010 10:26 AM, Richard Earnshaw wrote:
> >>> +          for (insn =3D seq; insn; insn =3D NEXT_INSN (insn))
> >>> +        if (INSN_P (insn))
> >>> +          INSN_LOCATOR (insn) =3D prologue_locator;
> >>> +
> >>=20
> >> Hmm, what if get_insns() doesn't return an INSN_P(), which is presumably
> >> what the test in the 'if' clause is for?  Won't it then abort in
> >> NEXT_INSN()?
> >=20
> > No, it won't.  INSN_P does not include notes, for example.
> >=20
> >=20
> > r~
> 
> Ah! I think I must be confusing this with some historical other behaviour.
> 
> Anyway, the other aspects of this patch were fine, so if this bit is OK then
> I'm happy with it all now.

OK, thanks.  I've checked the patch in now.

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com