[PATCH 2/2] ld: Allow symbols from PROVIDE to be use in MEMORY regions

[Alan Modra]

On Thu, Oct 15, 2020 at 01:58:10PM +0100, Andrew Burgess wrote:
> Now at the location of the first call expld.phase is indeed set to
> lang_first_phase_enum, so that's not an issue.  The problem is that
> expld.phase is _still_ set to lang_first_phase_enum at the time of the
> second call.

Oops, indeed it is.  We don't get to lang_mark_phase_enum until later.

> What I did do was move away from using the "phase" parameter, and
> instead just passed a TRUE/FALSE 'update_regions_p' predicate.
> Hopefully this removes possible confusion with "why a parameter and
> not expld.phase?", but also avoids passing parameters through global
> variables.

Yes, this is better.

> What do you think to this?

It's fine to apply.  Thanks for your patience.

-- 
Alan Modra
Australia Development Lab, IBM