[Bug tree-optimization/82776] Unable to optimize the loop when iteration count is unavailable.

[amker at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82776

amker at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amker at gcc dot gnu.org

--- Comment #3 from amker at gcc dot gnu.org ---
(In reply to Jakub Jelinek from comment #2)
> Perhaps at least when seeing an unconditional POINTER_PLUS_EXPR with
> constant increment and no other bumps for the same pointer in the loop, we
> could derive upper niter bound from that, even when the loop has multiple
> exits etc.

Yes.  GCC's niter analyzer only analyzes exit condition with induction variable
involved while it's not the case here.  As a result, niter analysis is skipped
for the loop.  In this code, the no-wrapping information is actually with the
IV, as you suggested, we should be able to refine bound information when niter
analysis of exit condition is skipped.   I will have a look.  Thanks.