[PATCH v2 3/4] x86: break gas dependency on libopcodes

[H.J. Lu]

On Mon, Nov 21, 2022 at 11:15 PM Jan Beulich <jbeulich@suse.com> wrote:
>
> On 21.11.2022 18:04, H.J. Lu wrote:
> > On Mon, Nov 21, 2022 at 8:52 AM Jan Beulich <jbeulich@suse.com> wrote:
> >>
> >> On 21.11.2022 17:43, H.J. Lu wrote:
> >>> On Mon, Nov 21, 2022 at 8:32 AM Jan Beulich <jbeulich@suse.com> wrote:
> >>>>
> >>>> On 21.11.2022 17:26, H.J. Lu wrote:
> >>>>> On Fri, Nov 18, 2022 at 1:13 AM Jan Beulich <jbeulich@suse.com> wrote:
> >>>>>>
> >>>>>> gas doesn't use anything from libopcodes anymore - suppress linking in
> >>>>>> that library.
> >>>>>> ---
> >>>>>> v2: New, split off from larger earlier patch.
> >>>>>>
> >>>>>> --- a/gas/configure
> >>>>>> +++ b/gas/configure
> >>>>>> @@ -12263,7 +12263,7 @@ _ACEOF
> >>>>>>
> >>>>>>      # Do we need the opcodes library?
> >>>>>>      case ${cpu_type} in
> >>>>>> -      vax | tic30)
> >>>>>> +      vax | tic30 | i386)
> >>>>>>         ;;
> >>>>>>
> >>>>>>        *)
> >>>>>> --- a/gas/configure.ac
> >>>>>> +++ b/gas/configure.ac
> >>>>>> @@ -420,7 +420,7 @@ changequote([,])dnl
> >>>>>>
> >>>>>>      # Do we need the opcodes library?
> >>>>>>      case ${cpu_type} in
> >>>>>> -      vax | tic30)
> >>>>>> +      vax | tic30 | i386)
> >>>>>>         ;;
> >>>>>>
> >>>>>>        *)
> >>>>>>
> >>>>>
> >>>>> We need to add some dependencies on i386-opc.tbl and i386-reg.tbl for
> >>>>> --enable-maintainer-mode to check if i386-tbl.h and i386-init.h are up to
> >>>>> date.  It doesn't need to regenerate i386-tbl.h.  An error message should be
> >>>>> sufficient.
> >>>>
> >>>> This isn't needed here, but in patch 2, where I'm already adjusting
> >>>> existing dependencies. Since I'm not modifying any toplevel files, the
> >>>> building of opcodes/ still ought to be happening before the building of
> >>>> gas/, so I don't see why any further changes should be necessary. Please
> >>>> clarify if you see any such reason.
> >>>>
> >>>
> >>> Since gas no longer depends on libopcodes, one may change i386-opc.tbl and
> >>> run "make" in gas.
> >>
> >> Is running make in gas/ a supported operation?
> >>
> >>>   It is more reliable than the fake dependency in opcodes.
> >>
> >> If the answer to the above is yes, then I may agree here. Except that then
> >> I don't see how this dependency is being enforced prior to this series.
> >> IOW - aren't you asking to address an unrelated issue?
> >>
> >
> > It didn't work before.  Now we have a chance.
>
> But then please in a separate change, on top of this series. I'm willing
> to make such a patch (whether that then finds your approval is a separate
> question), but I'd please like to see things unblocked here.
>

Then please don't add the fake dependency.


-- 
H.J.