Re: [PATCH] forwprop: Fix up rotate pattern matching [PR106523]

[Jakub Jelinek <jakub@redhat.com>]

On Tue, Jan 17, 2023 at 11:59:53AM +0100, Aldy Hernandez wrote:
> 
> 
> On 1/17/23 10:47, Jakub Jelinek wrote:
> 
> > Aldy/Andrew, is the ranger query ok or should I use something different
> > when check_range_stmt is non-NULL and I know on which statement to ask?
> 
> <snip>
> 
> > +	  int_range_max r;
> > +	  if (!get_global_range_query ()->range_of_expr (r, rotcnt,
> > +							 check_range_stmt))
> > +	    return false;
> 
> range_of_expr will work with and without a statement.  If no statement is
> provided, it will return the global range.  So you can use the same
> range_of_expr call with a statement or without one if you don't know it.
> 
> Note that get_global_range_query () will always return a global query object
> (think SSA_NAME_RANGE_INFO).  It will never use an existing ranger (for
> example, if called within VRP or another pass that has an active ranger
> enabled).  If simplify_rotate() may be used from some of these passes you
> *may* want to use get_range_query() which will pick up the active ranger, or
> a global query object if no ranger is active.

This is always in the forwprop pass.
I think it doesn't have any active ranger instance, but I could be wrong.

A question would be if it would be worth to activate it in this spot lazily
if it isn't active yet (and destruct at the end of the pass).

> For that matter, since get_global_range_query() uses a global query, it
> really doesn't matter if you pass a statement or not, since our global range
> store has no context (SSA_NAME_RANGE_INFO).  Although, I personally always
> pass the statement if known, because it's good form, and if things ever
> change to an active ranger, everything will just work.

	Jakub