On Mon, 26 Sep 2005, Matthew Orgass wrote:

> > I believe the data I mention above also contains the jump vector to use when
> > the machine comes back up from a low power sleep (which is why
> > NetBSD/hpcmips and NetBSD/hpcarm can't use low power sleep -- when
> > awakening, they will jump back into the WinCE ROM, see corrupted memory, and
> > hard reset).
>
>   Resets and NMI go to 0xbfc00000.  I've looked at my Clio's ROM, but it
> hit C code before following a hook from RAM and I haven't followed it
> beyond that (yet).  The NEC documentation is very comprehensive.  The
> global necel.com web site is missing documents again, but the local necel
> sites seem to have the VR4121 user manual.

Maybe it's system specific, then, in the WinCE core.  The MobilePro 8x0, for
instance, has a WinCE 2.11 "service pack" that replaces some pretty
low-level components (and more specifically, one driver related patch for
sleep/awaken).  Something in there might be hackable, but at that point it
is no longer a machine independent hook.

<sigh>

-- 
-- Todd Vierling <tv@duh.org> <tv@pobox.com> <todd@vierling.name>