Re: [Patch, fortran] PR105168 - [13/14/15/16 Regression] ICE in gfc_maybe_dereference_var, at fortran/trans-expr.cc:2870 since r9-3522-gd0477233215e37de

Thanks, Jerry.

Pushed to mainline as r16-8107. I will do the backporting in a few
weeks time and will report back when I am done.

Paul

On Sat, 14 Mar 2026 at 17:20, Jerry D <[email protected]> wrote:
>
> On 3/14/26 4:29 AM, Paul Richard Thomas wrote:
> > Hi All,
> >
> > The original error has changed to a segfault in run time.
> > Nevertheless, the fix posted a while back on the PR compiles, runs and
> > regtests on FC42/x86_64.
> >
> > The first chunk gets rid of the segfault since it passes a class
> > container to 'f', rather than the array descriptor.
> >
> > The second chunk resets the temporary descriptor lbounds to one,
> > adjusts the ubounds accordingly and sets the offset.
> >
> > As remarked in comment 8, the version with an allocatable component
> > 'a' is thoroughly broken. This is now PR124503.
> >
> > The other chunks are trivial conversions of NULL to NULL_TREE.
> >
> > OK for mainline and backporting?
> >
> > Paul
>
> All good here Paul, OK for mainline and backport as appropriate.
>
> Regards,
>
> Jerry