On Sat, Feb 8, 2014 at 10:25 AM, Jakub Jelinek <ja...@redhat.com> wrote: > On Thu, Feb 06, 2014 at 10:04:00AM +0100, Richard Biener wrote: >> > PS, just looking at the patch now again, in the light of PR59594 >> > the reordering of datarefs before goto again looks also wrong, we IMHO have >> > to perform a vector ordered remove in that case, ok to handle it as a >> > follow-up? >> >> Yeah... > > Here it is. Bootstrapped/regtested on x86_64-linux and i686-linux, ok for > trunk?
Ok. Thanks, Richard. > 2014-02-08 Jakub Jelinek <ja...@redhat.com> > > * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers > not at the end of datarefs vector use ordered_remove to avoid > reordering datarefs vector. > > --- gcc/tree-vect-data-refs.c.jj 2014-02-06 18:04:15.000000000 +0100 > +++ gcc/tree-vect-data-refs.c 2014-02-07 21:55:01.278701818 +0100 > @@ -3303,7 +3303,8 @@ again: > datarefs.pop (); > break; > } > - datarefs[i] = dr = datarefs.pop (); > + datarefs.ordered_remove (i); > + dr = datarefs[i]; > goto again; > } > > > > Jakub
