Re: [PATCH] alpha: add -mlra and enable LRA by default on BWX targets

On Tue, Jul 21, 2026 at 7:40 AM Oleg Endo <[email protected]> wrote:
>
>
>
> > On Jul 21, 2026, at 1:33, Andreas Schwab <[email protected]> wrote:
> >
> > On Jul 20 2026, Matt Turner wrote:
> >
> >> diff --git ./gcc/config/alpha/alpha.opt ./gcc/config/alpha/alpha.opt
> >> index 1c1fbbb69f1..c97c31a865d 100644
> >> --- ./gcc/config/alpha/alpha.opt
> >> +++ ./gcc/config/alpha/alpha.opt
> >> @@ -136,3 +136,7 @@ Tune expected memory latency.
> >> mtls-size=
> >> Target RejectNegative Joined UInteger Var(alpha_tls_size) Init(32)
> >> Specify bit size of immediate TLS offsets.
> >> +
> >> +mlra
> >> +Target Var(alpha_lra_p) Init(-1) Undocumented
> >> +Use LRA instead of reload (transitional).
> >
> > You also need to update alpha.opt.urls.
> >
>
> Why is this needed and why is this not done automatically? I might have missed the memo with the rationale, but this seems like just another annoyance.

Because it requires python which is not currently a requirement for
building in general.  Also we don't write ever back into the src tree.


>
> Best regards,
> Oleg Endo
>