Need help trying to debug a Pari segfault

[kcrisman]

See #11551. Basically, we now more or less can (yet again) get Sage
to build on Cygwin, but in starting up there is a segfault caused by
Pari when creating a Pari float from an MPFR number

pari_float = cgetr(2 + rounded_prec / wordsize)

and cgetr is where the problem occurs.

I can get around this to get Sage to start by commenting out the
initialization of I in symbolic/pynac.pyx and by commenting out the
initialization of qqbar in sage/all.py, but obviously that is not very
good long-term!

It would be really, really helpful for any Pari experts to just take a
look at the error reports at that ticket (including real_mpfr.c) just
to *suggest* any ideas for further debugging or fixes. I realize that
very few will actually have access to this platform.

Thanks,
- kcrisman