On Mon, 10 Dec 2018, Kyrill Tkachov wrote:
> Hi Richard,
>
> On 10/12/18 13:44, Richard Biener wrote:
> >
> > When working on PR63184 I noticed that tree-affine.c is still one of
> > the gimple_assign_rhs_to_tree callers and does so quite aggressively
> > for no good reason. The following restr
Hi Richard,
On 10/12/18 13:44, Richard Biener wrote:
When working on PR63184 I noticed that tree-affine.c is still one of
the gimple_assign_rhs_to_tree callers and does so quite aggressively
for no good reason. The following restricts it to the cases we
handle in tree_to_aff_combination, axing
When working on PR63184 I noticed that tree-affine.c is still one of
the gimple_assign_rhs_to_tree callers and does so quite aggressively
for no good reason. The following restricts it to the cases we
handle in tree_to_aff_combination, axing one unreachable case.
It also makes sure to cache expa
