[Bug tree-optimization/50913] [4.7 Regression] ICE in scan_tree_for_params_right_scev, at graphite-sese-to-poly.c:633 compiling libgfortran with -floop-interchange -m32

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50913

Sebastian Pop <spop at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
                 CC|                            |spop at gcc dot gnu.org
         Resolution|FIXED                       |---

--- Comment #11 from Sebastian Pop <spop at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #8)
> There is a disconnect on how we analyze data-references during SCOP detection
> (outermost_loop is the root of the loop tree) and during SESE-to-poly where
> outermost is determined by outermost_loop_in_sese_1 ().  That influences
> the SCEV result and thus we do not break the SCOP at a stmt we have to break
> it.
> 
> The following patch fixes this using a sledgehammer - require the
> data-ref to be representable if analyzed with respect to all loops
> it can nest in.

This is an effective shortcut to disable graphite on a lot of loops...
see bug 14741.

Let me see if I can find a better solution is for this bug.