[Info-vax] VSI: "Official 8.4-1H1 Launch"

[David Froble]

JF Mezei wrote:
> I apologize for having questions in advance. Having deffective genes is
> not my fault :-)

Yes it is   :-)


> At what point in the boot sequence does VMS establish virtual memory,
> and the various modes for code execution (and memory protection) ?
> 
> Would it be correct that this section of the boot is where the most work
> is needed to tailor to an architecture such as x86 ?
> 
> When VMS implements the various memory protection schemes, is this a one
> time x86 specific setup early in boot, or do those x86 specific changes
> propagate well beyond initial boot all the way (for instance) to image
> activator etc ?

I haven't been near this stuff for many, many years.  I can only answer 
from long ago experience.  Most from RSTS, and I've probably forgotten 
90% of that.

The first thing that happens is a rather simple task of copying 
essential code into memory.  At this time there is no virtual memory, 
because the code to support it is not present.

What is essential code?  That can vary.  I don't know exactly how VMS 
works at this stage.  What I will guess is that you don't have VM until 
the code that handles VM is loaded and running.

As for the autoconfig, I'm guessing the code at that point probes for 
various HW.  That's part of the reason for HW releases, along with 
drivers.  Can't probe for something you don't know exists.  If a device 
is found, then requirements for that device are loaded, and whatever 
device tables are updated.

But enough guessing.  The short answer, get and study the Internals and 
Data Structures book.