[Bug c/104115] New: Does not understand that inequality is transitive

[Bug c/104115] New: Does not understand that inequality is transitive

[thvdveld at vub dot be]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104115

            Bug ID: 104115
           Summary: Does not understand that inequality is transitive
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: thvdveld at vub dot be
  Target Milestone: ---

GCC seems to fail to eliminate the third assert for unsigned integers.
This works as expected when integers are used instead of unsigned.

```c
void f(unsigned int a, unsigned int b, unsigned int c) {
    assert(a < b);
    assert(b < c);
    assert(a < c);
}
```

https://gcc.godbolt.org/z/o9ajsMTb7

LLVM seems to be able to optimize the unsigned version, however not the version
with `int`. (https://github.com/llvm/llvm-project/issues/53273)

[Bug c/104115] Does not understand that inequality is transitive

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104115

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 12 handles this fine.

[Bug c/104115] Does not understand that inequality is transitive

[thvdveld at vub dot be]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104115

--- Comment #2 from Thibaut Vandervelden <thvdveld at vub dot be> ---
Sorry, my bad.

[Bug c/104115] Does not understand that inequality is transitive

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104115

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |aldyh at gcc dot gnu.org,
                   |                            |amacleod at redhat dot com,
                   |                            |jakub at gcc dot gnu.org

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Since r12-3119-g675a3e40567e1d0dd6d7e7be3efab74b22731415
Maybe we want the testcase for the testsuite so that we don't regress it?

[Bug c/104115] Does not understand that inequality is transitive

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104115

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Note, gcc.dg/tree-ssa/evrp-trans.c seems to cover it for int, but perhaps it
would be useful to have a variant of the testcase with unsigned types instead
of signed too.

[Bug c/104115] Does not understand that inequality is transitive

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104115

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:f35bb467a01db988d5fe1637db5bff4b4f9e7086

commit r12-6738-gf35bb467a01db988d5fe1637db5bff4b4f9e7086
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Jan 19 15:02:31 2022 +0100

    testsuite: Test evrp-trans.c also with unsigned types [PR104115]

    The testcase from the PR got fixed with r12-3119-g675a3e40567e1d
    and looks quite similar to the evrp-trans.c test, except evrp-trans.c
    is tested on signed integer types.
    I think it would be useful to test it for unsigned comparisons too.

    2022-01-19  Jakub Jelinek  <jakub@redhat.com>

            PR c/104115
            * gcc.dg/tree-ssa/evrp-trans2.c: New test.