Support for 64bit userspace.

[Jayachandran C.]

On Thu, Jul 29, 2010 at 5:11 AM, Juli Mallett <jmallett at freebsd.org> wrote:

> Hi JC,
>
> On Wed, Jul 28, 2010 at 16:32, Jayachandran C. <c.jayachandran at gmail.com>
> wrote:
> > On Wed, Jul 28, 2010 at 2:04 AM, Jayachandran C.
> > <c.jayachandran at gmail.com> wrote:
> >> Here's my initial work to get 64bit user space.  With this set of
> >> changes, I can boot to the single user shell with n64 /sbin/init and
> >> /bin/sh.  There are still issues to fix, and the dynamic loader is
> >> still not working, but I think this would be a good time to get some
> >> initial feedback.
> >
> > The dynamic loader seems to work after some straight-forward changes.
> > The only change is that the got[1] entry needs to be 64-bit.  The
> > changes are attached.
> >
> > There is still an occasional kernel panic in n64, which I need to
> > track down - but I think the changes other than pmap.c/pmap64.c are
> > ready for check in.  I would like some feedback on pmap64.c I can
> > clean it up for check in.
>
> I'm uncomfortable with pmap.c and pmap64.c.  I don't get the
> impression that the ifdefs required to do both in the same file are
> really a great burden, but obviously my efforts are incomplete
> compared to yours.  It seems like there are very few operations that
> should care about how deep the page tables are and that macros and
> ifdefed inline functions are the way to go.  If you check it in as two
> files, I'll probably just go through and ifdef the deltas that can't
> be eliminated.  What do you think are the problem areas that justify
> having two different files?
>

I kept the 32 bit pmap.c code untouched so that I don't have to deal with
the breakage there during the early 64bit changes.

I will check in the other changes now, and will do another round of work on
pmap.c/pmap64.c to see if it can be merged (and eliminate the config
option).

Thanks,
JC.