Original from  Daniel Lynes  to All on 12-01-1996
Original Subject: Protected mode difference

                         ---------------------------------------

DL> Just wondering if anyone would happen to know why Win95 will not allow
DL> read or write access to protected memory, yet OS/2 works as expected (allows
DL> read-only access to protected memory?)

 This is a function of the hardware. The Descriptor Tables and the 
Page Tables. The "Operating System" can/will set up the tables, but
the hardware actually uses them. 

"Protected memory" is memory that is either out of the range of
addressability for your app/process, or memory marked as needing a
higher privledge level than you are attempting to access it with.
Either of the above will make it "unreadable". If you can read it but
not write it, it is not really "protected memory" as such, but instead
just plain "read only" memory as implemented in the "protect
mode" of
the Intel CPU. Your own code segments will have this characteristic. 
 There is also the possibility of having "execute only code" which can
be directly "called" but not "read", but again this is
not really 
"protected memory" as such, but only exploitation of another facet of 
the CPU "protection" mechanisms. 

 OS/2 does have "protected memory". The kernel code is unreadable from
your application for example. You can "call" it through specific APIs
(which end up going through callgates), but you cannot "read" it
directly. 

 Can you give some specific examples of what Win95 is doing and the 
same under OS/2?


   Denis       

 All opinions are my very own, IBM has no claim upon them
. 
. 
.
--- Maximus/2 3.01