Bill Layer wrote:
> 
> Well,
> 
> > Why the hell can't PalmOS get installed/upgraded from Linux?
> 
> That doesn't make much real sense to me.. the connection to the PalmOS device
> is a simple serial interface. Any task that Win32 can perform with a
> serial interface, can also be done in Linux (only better :). Since the OS
> upgrade is just a serial transfer, I image that the Linux pilot-link suite
> should be fully capable of this, with only a little hacking. Heck, you might be
> able to do it with minicom...
> 
> The question in my mind is, what is the 'magic bullet' that turns a
> run-of-the-mill app install or data fetch into a flash-OS programming
> operation? If this is not already documented, it should be possible to log the
> entire transaction between the PC and the PalmOS device performing an OS
> upgrade (Win32 SoftICE?), and then begin reverse-engineering the procedure from
> that data. I doubt there is any form of encryption used.
> 
> Or maybe I'm oversimplifying it..

As far as I can tell, you're basically correct.  The basic process is
this: the Palm reboots in `debug' mode, the desktop and Palm do some
negotiation over the serial cable and begin transferring the ROM.  Once
the transfer is done, the Palm resets again and (hopefully) starts up..

One problem is that Palm distributes their OS updates (at least 3.3,
dunno about 3.5 (yet)) in executables rather than in zip files.  The
other is that I doubt many people have access to information about how
the debug mode works (at least, not many people who are not under NDA or
something).

-- 
 _  _  _  _ _  ___    _ _  _  ___ _ _  __   Shin: a device for 
/ \/ \(_)| ' // ._\  / - \(_)/ ./| ' /(__   finding furniture in the  
\_||_/|_||_|_\\___/  \_-_/|_|\__\|_|_\ __)  dark. 
[ Mike Hicks | http://umn.edu/~hick0088/ | mailto:hick0088 at tc.umn.edu ]