[Bug tree-optimization/82402] [6/7/8 Regression] error: SSA_NAME_OCCURS_IN_ABNORMAL_PHI should be set

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82402

--- Comment #7 from rguenther at suse dot de <rguenther at suse dot de> ---
On Wed, 22 Nov 2017, jakub at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82402
> 
> Jakub Jelinek <jakub at gcc dot gnu.org> changed:
> 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |jakub at gcc dot gnu.org
> 
> --- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
> So shall we ignore it on them in the checking code, or make sure it is never
> set?

Neither.  We should fix whoever fails to set it.

Whether we want to at some point ignore it / remove it is another 
question.  Looks like the culprit is somehow the vectorizer pass
which ends up with a spurious MEM_23(ab) for me.  I guess it doesn't
really expect the loop exit to end in a call with just an abnormal
outgoing edge...

Let me have a look.