x86: limiting the use of No_*Suf

H.J. Lu hjl.tools@gmail.com
Mon Aug 15 18:10:23 GMT 2022


On Fri, Aug 12, 2022 at 6:19 AM Jan Beulich <jbeulich@suse.com> wrote:
>
> On 12.08.2022 11:30, Jan Beulich via Binutils wrote:
> > since you didn't like "x86: imply all No_*Suf when none is set in a
> > template" I've been thinking of other ways to limit the excessive use
> > of these attributes. One observation is that templates with IsPrefix,
> > Broadcast, or Masking set always have all No_<x>Suf also set.
> > Therefore the latter could be dropped from those templates (improving
> > readability), compensated by
> > 1) a respective adjustment to match_template(), or
> > 2) a change to i386-gen.c, or
> > 3) respective pre-processor macros.
> > Do you have any opinion on the general direction of this and, if in
> > favor, any preference for one of the three options?
>
> Actually I had forgotten that we already have precedent of 3), for
> AddrPrefixOpReg. So perhaps this can be a hint as to further moves
> in that direction ...
>
> Jan

I prefer #3.

Thanks.

-- 
H.J.


More information about the Binutils mailing list