| Submitter | Mikael Morin |
|---|---|
| Date | Aug. 3, 2012, 2:18 p.m. |
| Message ID | <501BDDBD.2060308@sfr.fr> |
| Download | mbox | patch |
| Permalink | /patch/174994/ |
| State | New |
| Headers | show |
Comments
Dear Mikael, This looks to be "obvious" and is certainly OK for trunk. Thanks for the patch. Paul On 3 August 2012 16:18, Mikael Morin <mikael.morin@sfr.fr> wrote: > Hello, > > here is the fix for the regression I have introduced with my assumed > rank bounds patch. > > Will test and commit as obvious. > Mikael > > > 2012-08-02 Mikael Morin <mikael@gcc.gnu.org> > > PR fortran/54166 > * trans-array.c (set_loop_bounds): Access specinfo using spec_dim. > > 2012-08-02 Mikael Morin <mikael@gcc.gnu.org> > > PR fortran/54166 > * gfortran.dg/array_5.f90: New test. > >
On Fri, Aug 3, 2012 at 7:18 AM, Mikael Morin <mikael.morin@sfr.fr> wrote: > Hello, > > here is the fix for the regression I have introduced with my assumed > rank bounds patch. > > Will test and commit as obvious. > Mikael > > > 2012-08-02 Mikael Morin <mikael@gcc.gnu.org> > > PR fortran/54166 > * trans-array.c (set_loop_bounds): Access specinfo using spec_dim. > > 2012-08-02 Mikael Morin <mikael@gcc.gnu.org> > > PR fortran/54166 > * gfortran.dg/array_5.f90: New test. > > Will this also fix http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54175
Patch
Index: trans-array.c =================================================================== --- trans-array.c (révision 190099) +++ trans-array.c (copie de travail) @@ -4503,9 +4503,9 @@ set_loop_bounds (gfc_loopinfo *loop) else if (INTEGER_CST_P (info->start[dim]) && !INTEGER_CST_P (specinfo->start[spec_dim]) && integer_onep (info->stride[dim]) - == integer_onep (specinfo->stride[dim]) + == integer_onep (specinfo->stride[spec_dim]) && INTEGER_CST_P (info->stride[dim]) - == INTEGER_CST_P (specinfo->stride[dim])) + == INTEGER_CST_P (specinfo->stride[spec_dim])) loopspec[n] = ss; /* We don't work out the upper bound. else if (INTEGER_CST_P (info->finish[n])