search for: pr12377

...dent from the arithmetic operations that can be carried through canonicalization, reassociation etc. This could be done today with existing overflow intrinsics, but modeling the trap would require a branch. I suggest filing a bug for your case and continuing discussion there. It may be related to PR12377: Loop trip count not calculated. -Andy

On 10/03/2013 01:22 AM, Andrew Trick wrote: > > I’m not sure how to make sense of an NUW flag coming from a sub. > > NSW is just a misnomer for signed overflow. SCEV canonicalizes sub a,b to add a, (-b). Unfortunately, signed overflow is not the same thing for sub and add... > > sub nsw %a, %b != add nsw %a, (-1 * %b) > > sub nsw i32, -1, INT_MIN is true. > > add

...canonicalization, reassociation etc. This could be done today with existing overflow intrinsics, but modeling the trap would require a branch. Why this? Do we need to handle also trapping arithmetic? > I suggest filing a bug for your case and continuing discussion there. It may be related to PR12377: Loop trip count not calculated. I've done a little rewrite of functions to compute trip count for LT and GT comparisons. However that case is not handled because of the missing NSW flag in the SCEV. So probably it would be better to proposed a patch with my modification before. -Michele