In message: <Pine.NEB.4.58.0404291909560.10837@server.duh.org>
            Todd Vierling <tv@duh.org> writes:
: On Thu, 29 Apr 2004, M. Warner Losh wrote:
: 
: : You are right that it doesn't put things into the lowest CPU mode, but
: : that requires new ROMs...
: 
: I still have trouble believing this argument.
: 
: Has anyone disassembled the ROM starting from the target of the power-on
: jump vector to see if it has a "patch hook" that can be intercepted and
: routed back to NetBSD?  Typically such a patch hook requires that a certain
: magic word be at a specific location in memory, and/or that some sort of
: in-memory startup structure be populated.

Yes.  I've disassembled at least one ROM and it appears to do a bunch
of things in the ROM then jump back to a hook.  The other ROM I looked
at seemed to be somewhat different.  I'll have to take a closer look
when I get a few minutes.  Now that I have larger CF parts, that makes
the box more attractive to me.

: I've *never* seen an embedded OS in masked ROM without such a hook, to allow
: for -- you guessed it -- power management and/or warm reset hardware fixups.
: I'm fairly certain that the WinCE "service pack" as distributed by NEC
: contains such fixups for machines like ... the MobilePro 800/MC-R700[A].

Part of the problem is that the contents of RAM aren't guaranteed when
coming out of the deep suspend mode until certain things are done,
iirc.  I forget the exact details, since it has been a while since
I've done the leg work to investigate.

There doesn't appear to be a standard location for this, so each
different hpc would need some TLC to find out where the magic location
for it would be...

Again, WinCE runs out of the ROMs, with minimal information in RAM.

Warner