[Bug tree-optimization/128306] New: `a - a` for real types where a is not a NaN can be assumed to be >= 0

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128306

            Bug ID: 128306
           Summary: `a - a` for real types where a is not a NaN can be
                    assumed to be >= 0
           Product: gcc
           Version: 17.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: pinskia at gcc dot gnu.org
            Blocks: 85316
  Target Milestone: ---

Created attachment 66040
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=66040&action=edit
testcase

The attached testcase should just be optimized to `return 0`.  Without trapping
math enabled for sure.  I am not 100% sure it can be optimized to that with
trapping math.  I am not sure about the rules of fp exceptions for `a - a`.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85316
[Bug 85316] [meta-bug] VRP range propagation missed cases