On Tue, 30 Jan 2018, Jakub Jelinek wrote:
> On Tue, Jan 30, 2018 at 08:59:43AM +0100, Richard Biener wrote:
> > The issue is that
> >
> > static bool
> > tree_unroll_loops_completely_1 (bool may_increase_size, bool unroll_outer,
> > bitmap father_bbs, struct loop *
On Tue, Jan 30, 2018 at 08:59:43AM +0100, Richard Biener wrote:
> The issue is that
>
> static bool
> tree_unroll_loops_completely_1 (bool may_increase_size, bool unroll_outer,
> bitmap father_bbs, struct loop *loop)
> {
> struct loop *loop_father;
> bool change
On Mon, Jan 29, 2018 at 11:58:45PM -0700, Jeff Law wrote:
> On 01/29/2018 04:37 PM, Jakub Jelinek wrote:
> > Hi!
> >
> > As mentioned in the PR, cunroll sometimes registers some SSA_NAMEs for
> > renaming and then invokes some SCEV using functions before finally updating
> > the SSA form. On the
On Mon, 29 Jan 2018, Jeff Law wrote:
> On 01/29/2018 04:37 PM, Jakub Jelinek wrote:
> > Hi!
> >
> > As mentioned in the PR, cunroll sometimes registers some SSA_NAMEs for
> > renaming and then invokes some SCEV using functions before finally updating
> > the SSA form. On the testcase we end up w
On 01/29/2018 04:37 PM, Jakub Jelinek wrote:
> Hi!
>
> As mentioned in the PR, cunroll sometimes registers some SSA_NAMEs for
> renaming and then invokes some SCEV using functions before finally updating
> the SSA form. On the testcase we end up with 3 degenerate PHIs pointing at
> each other, so
Hi!
As mentioned in the PR, cunroll sometimes registers some SSA_NAMEs for
renaming and then invokes some SCEV using functions before finally updating
the SSA form. On the testcase we end up with 3 degenerate PHIs pointing at
each other, so follow_copies_to_constant loops forever.
The following
