Re: [PATCH 0/6] x86: drop L1OM/K1OM from gas plus associated tidying

["H.J. Lu" <hjl.tools@gmail.com>]

On Thu, Mar 17, 2022 at 1:32 AM Jan Beulich <jbeulich@suse.com> wrote:
>
> On 16.03.2022 18:11, H.J. Lu wrote:
> > On Wed, Mar 16, 2022 at 1:42 AM Jan Beulich <jbeulich@suse.com> wrote:
> >>
> >> It was the bugs / shortcomings dealt with here as a "side effect"
> >> which made me raise the question of the utility of having the
> >> rudimentary support for the two sub-architectures. Oddly enough
> >> some of the bugs need fixing _before_ removing the support, or else
> >> IAMCU would regress. Furthermore a new IAMCU test can be put in
> >> place only _after_ removing the support.
> >>
> >> 1: assorted IAMCU CPU checking fixes
> >> 2: drop L1OM/K1OM support from gas
> >> 3: add another IAMCU testcase
> >> 4: unify CPU flag on/off processing
> >> 5: never set i386_cpu_flags' "unused" field
> >> 6: don't accept base architectures as extensions
> >>
> >
> > OK to all.
>
> Thanks. I realize only now that I should also update gas/doc/ in
> patch 2. I'll assume that's fine to slip in, for only being
> consistent.

Yes, please.

> > Can you also remove L1OM/K1OM from ld and binutils?
>
> I have to admit that I wasn't sure how far to go: Removing support
> from ld certainly makes sense (and I'll see to find time), but
> wouldn't it be reasonable to keep the minimal support there is in
> binutils/?

I don't think we need to support more than EM_L1OM and EM_K1OM,
similar to EM_INTELGT.

> I've noticed there's one piece of special casing behavior for L1OM
> in i386-dis.c. I'd be inclined to drop that too - thoughts?

Yes.

-- 
H.J.