Xref: utzoo comp.graphics:9393 comp.os.vms:21630
Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!rutgers!mcnc!rti!bcw
From: bcw@rti.UUCP (Bruce Wright)
Newsgroups: comp.graphics,comp.os.vms
Subject: Re: Help with Iris 1200 XNS driver on VAX/VMS
Summary: Patching system version number in driver ... DON'T DO IT!
Keywords: Iris 1200, XNS driver, VAX/VMS
Message-ID: <3452@rti.UUCP>
Date: 17 Jan 90 05:40:17 GMT
References: <1393@ssc-bee.ssc-vax.UUCP> <288@dino.cs.iastate.edu>
Organization: Research Triangle Institute, RTP, NC
Lines: 40

In article <288@dino.cs.iastate.edu>, hascall@cs.iastate.edu (John Hascall) writes:
>       A possible solution is to patch the driver image so that it
>       looks like it was compiled under VMS 5.x.  The version number 
>       is stored at offset DPT$W_VERSION (x18) in the Driver Prologue
>       Table (which just happens to be the first thing in the driver :-)
>       So what you need to do is:
>
>		[procedure deleted]
>
>       and give it a try (might work, might not, depends on if the
>       driver does any VMS 4.x specific things).

This procedure is HIGHLY unlikely to work - it has about a 99.5% chance
of crashing your system.

One problem is that VMS V5.0 introduced the modular Executive, and ALL
the driver entry point addresses changed (whether they needed to or not),
so that they now point into a jump table rather than to the actual routine
in the Executive.  So all the little routines that drivers are so fond of
calling (the EXE$, COM$, and IOC$ routines in particular) are no longer
found in the same locations.

Another problem is that numerous entry points were REMOVED from VMS V5.0.
The functions they performed are now done differently - and the old
routines simply don't exist.

Yet another problem is that some of the system data structures grew with 
VMS V5.0, and a driver linked against the old system will refer to offsets
whose meanings have changed and expect things like the UCBs to be smaller
than they are on V5.0.  Since many drives allocate extra space in a UCB
extension at the end of their UCBs, this could cause fatal corruption.

It's barely possible that there exist drivers which could be patched in
this manner, but it's difficult to imagine them performing any useful
functions.  Don't try the procedure outlined above - it's unlikely to
work and at best will probably crash the system;  at worst it may cause
widespread corruption (which would hopefully not find its way into the
file system ...)

						Bruce C. Wright