patch: honor volatile bitfield types

[Hans-Peter Nilsson]

(I wrote:)
> > Can we similarly promise or say something for accesses of the
> > containing struct as a whole?

No takers?

> Date: Wed, 23 Jun 2010 11:34:04 -0400
> From: DJ Delorie <dj@redhat.com>

> Should be the same as before, I would think.

Primarily I want them similarly defined.  I wasn't expecting
those access to be actually changed by your patches.  Of course
it'd be nice if they could tag along. :)  The first step is this:
to check if anyone is against them being well-defined (in GNU C
terms, of course not in ISO C terms), and perhaps whether people
believe that it's obvious one way (like I do) or the other (like
it seemed other people did).

Thanks BTW, for pushing through the subject as well as the patches. :)

brgds, H-P
PS. Happy midsummer!