Re: [PATCH] Fix size recalculation of fortran arrays

[Andrew Burgess <andrew.burgess@embecosm.com>]

* Hannes Domani via Gdb-patches <gdb-patches@sourceware.org> [2020-05-01 13:46:11 +0000]:

>  Am Freitag, 1. Mai 2020, 15:11:40 MESZ hat Andrew Burgess <andrew.burgess@embecosm.com> Folgendes geschrieben:
> 
> > * Hannes Domani via Gdb-patches <gdb-patches@sourceware.org> [2020-05-01 14:35:18 +0200]:
> >
> > > My recent change regarding size calculation of arrays of stubbed types
> > > didn't take array strides and associated/allocated type properties into
> > > account, which basically broke fortran arrays.
> > >
> > > Fixed by refactoring the array size calculation of
> > > create_array_type_with_stride into a new function, and also use it for
> > > the stubbed array size recalculation.
> > >
> > > gdb/ChangeLog:
> > >
> > > 2020-05-01  Hannes Domani  <ssbssa@yahoo.de>
> > >
> > >     * gdbtypes.c (calculate_static_array_size): New function.
> > >     (create_array_type_with_stride): Use calculate_static_array_size.
> > >     (check_typedef): Likewise.
> > > ---
> > >  gdb/gdbtypes.c | 130 ++++++++++++++++++++++++++-----------------------
> > >  1 file changed, 68 insertions(+), 62 deletions(-)
> > >
> > > diff --git a/gdb/gdbtypes.c b/gdb/gdbtypes.c
> > > index 6648dc4d67..ad5c7ee0af 100644
> > > --- a/gdb/gdbtypes.c
> > > +++ b/gdb/gdbtypes.c
> > > @@ -1177,6 +1177,62 @@ discrete_position (struct type *type, LONGEST val, LONGEST *pos)
> > >      }
> > >  }
> > >
> > > +/* If the array has static bounds, calculate its size.  */
> > > +
> > > +static bool
> > > +calculate_static_array_size (struct type *type)
> >
> > Could you rename this to something like update_static_array_size -
> > calculate to me implies something will be calculated and returned.
> > Also please update the header comment to reference TYPE, and describe
> > the return values, something like:
> >
> >   /* If the array TYPE has static bounds calculate and update its
> >     size, then return true.  Otherwise return false and leave TYPE
> >     unchanged.  */
> >
> > With those changes this looks good.
> 
> Pushed with those changes, thanks.
> 
> 
> > Thanks for the quick fix.
> 
> I'm very sorry for the breakage, but running the testsuite on Windows can
> be very frustrating at times.
> Maybe I'm doing something wrong, but each test only has ~66% chance of even
> staring correctly.

I feel your pain.  Any time I have to do anything on Windows I know
I'm going to have a bad day.

It can be worth mentioning in your mailing list post if you are
worried that your testing might not have gone well, most reviewers
will be happy to apply a patch locally and do a test run to double
check.

Thanks,
Andrew