On January 24, 2017 5:02:39 PM GMT+01:00, Marc Glisse <marc.gli...@inria.fr> wrote: >On Tue, 24 Jan 2017, Jeff Law wrote: > >> But that would assume that match.pd is relying on range information >and could >> thus use the improved range information. *If* match.pd is using the >range >> information generated by VRP, it's not terribly pervasive. > >Oh, I thought we already had some explicit calls to get_range_info in >there, but apparently not. We can use VRP info through >get_nonzero_bits, >expr_not_equal_to and maybe one or two more like tree_expr_nonzero_p or
We have the latter already. >tree_expr_nonnegative_p. If we called into match.pd from VRP, I assume >several of the simplify_*_using_ranges could be moved to match.pd. Yes, that will hopefully happen at some point but it also requires folding more stmts to not regress (I'd like to remove the stmt folding callback from the SSA propagator) Richard. But >I >agree that most transformations do not look at ranges at all.
