On Tue, 24 Apr 2018, Michael Matz wrote:
> What is lost here (it wasn't explicit before, but is the case and must
> continue to work) is that function calls and returns count as needing the
> observable value in the specified register (function calls also count as
> setters of them). I think that should be made more explicit.
It's no different from a normal variable in global scope. If the compiler sees
gvar = 0;
bar();
gvar = 1;
it can delete the first store iff it can prove that it's dead ('bar' does
not read it), gvar being a register variable making no difference. So I
don't think it needs additional clarification.
> Especially
> because you make it explicit that the register remains available for
> allocation (where it before said it's not) we need to make sure that
> everybody understand that it's only during tiny windows like:
>
> regvar = ...
> ... code not accessing regvar and containing no calls ...
> regvar = ...
No, the first store is not needed. If the regvar is dead, the register is
freely allocatable (if not dead, it's allocatable at a cost of 1 spill).
> The return of functions also must remain making the regvar setter live, to
> give us the important parts of the former "stores aren't deleted even if
> appearing dead" guarantee.
This also follows from the variable being a global variable in the first place.
Alexander
