[Bug tree-optimization/71031] [6/7 Regression] ICE in extract_range_from_binary_expr_1, at tree-vrp.c:2535 w/ -Os

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71031

--- Comment #8 from rguenther at suse dot de <rguenther at suse dot de> ---
On Wed, 18 May 2016, mpolacek at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71031
> 
> Marek Polacek <mpolacek at gcc dot gnu.org> changed:
> 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |jakub at gcc dot gnu.org,
>                    |                            |rguenth at gcc dot gnu.org
> 
> --- Comment #7 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
> To expand a bit on Comment 4: we have [1, od_5] + UINT_MAX on type unsigned and
> are in extract_range_from_binary_expr_1.  We combine the lower bounds, which is
> 1 + UINT_MAX = 0(OVF).  We  then combine the upper bounds, because the max_op0
> is not a constant, the result of that is UINT_MAX.  So WMIN is 0(OVF) and WMAX
> is UINT_MAX.  These are equal to TYPE_MIN and TYPE_MAX, respectively.  What to
> do with that, drop to VR_VARYING if we have min overflow or max underflow?
> 
> Richi/Jakub, any ideas?  I'm kinda stuck here :(.

The situation (+ UINT_MAX) is equal to [1, od_5] - 1 which would mean
a result range of ideally [0, od_5 - 1].

I think Eric added this code so he may want to have a look here to see
what general solution is appropriate.

But yes, instead of asserting we can always drop to VR_VARYING ...