Jeff,
I can provide a short explanation, but if you'd like to get a better
understanding a good textbook on operating systems would be a good
idea. (A book coverying theory rather than any specific OS.)
When an operating system uses 'protected memory' each process gets
it's own address space and the kernel gets it's own memory space.
A process cannot access memory outside it's space. When a process
needs to use kernel services a 'context switch' occurs and the
the kernel code gets executed in the kernel's context. If Netscape
does something stupid Netscape may crash, but the kernel and
all the other processes keep right on going.
A shared-memory architecture allows proceses to share some/all
memory. If, for example, Netscape writes to a location used by
the kernel the system will likely crash. Eventually. Windows 98
and it's predecessors use a shared memory model. There are some
protections provided by Intel processors that Windows takes
advantage of, but, in the end, a buggy program can access memory
outside it's space and can crash the system.
Windows 2000/NT is architected by the same people that did VMS and
like Unix uses protected memory. While I don't know much about
Microsoft products, I do feel this is the most significant change
to come from them in many years. I think it's a big deal because
if MS can make the kernel reliable then Windows will have the same
reliability that those of us using Unix/Linux now enjoy. (by the
way, I don't use any Microsoft products and have no financial
interest in the company.)
So, hopefully, this lame explanation is useful. If I missed something
let me know and I'll try to explain privately (I doubt everyone wants
to read 'Mike on OS theory.' A good textbook on the subject would be
more informative than me also.
Mike
> Mike,
>
> Could you spare a few ones and zeroes and explain the difference between a
>"protected memory model"
> and a "shared memory model"? When I went to Windows/95 and Windows/NT school, much
>was made out of
> the fact that all processes within the OS had their own unique memory address space,
>but that the
> upper half of memory was devoted to the kernel and was shared. VMS uses the same
>model (Microsoft's
> innovation: moving the VMS memory model from the VAX to the 80386. Big whoop), and
>VMS is
> considered a rock solid OS. But now you tell me that Windows is actually sharing
>memory, I assume
> between processes. I don't understand.
>
> >
> > In any case, Linux has proven itself to be much more reliable than the
> > operating systems produced at Microsoft. (And Apple, but most folks
> > don't care about Apple.) It should be interesting to see how reliable
> > Windows 2000/NT ends up being as it is a protected-memory architecture
> > and _should_ be as reliable as Linux. We'll see.
> This is true.
> >
> > Mike
>
> Many thanks,
>
>
> --
> Jeff Silverman, PC guy, Linux wannabe, Java wannabe, Software engineer, husband,
>father etc.
> See my website: http://www.commercialventvac.com/~jeffs
> [EMAIL PROTECTED]
>
> ========================================================================
> Contributions/Posts To: [EMAIL PROTECTED]
> To Unsubscribe: [EMAIL PROTECTED], "unsubscribe" in message body
> Report Problems to: [EMAIL PROTECTED]
> List archive at: http://www.ssc.com/mailing-lists/
-
To unsubscribe from this list: send the line "unsubscribe linux-diald" in
the body of a message to [EMAIL PROTECTED]
