OSDev.org

this is not really how it works exactly( i know..i know)...but I wanted you to have a really simple example:

rings in the processor are stuff that let your OS do things without the processor...protected mode - protects the computer

Ring 0 lets your kernel do whatever it wants.

Ring 3 protects whatever you want ( your OS's apps for example ) from messing up things. ( I wonder if Windows Apps run in ring 0 ? )

The higher the ring number...the more protection.

Hope that helps...if you want a more detailed example I can get links...

>Hope that helps...if you want a more detailed example I can get links...

yeah sure it will be a great help

and what about ring 1 and 2 ? do they exist or not ?

thanks

suhmaamiar

don't know what those rings do...i asked here but no one answerd

here are links:

http://cwdixon.com/support/win98_support/apps_ring_architecture.htm
http://www.wikipedia.org/wiki/Ring_0
http://www.goldencode.com/atlos2/notes/theseus/memoryanalysis.html

Hope that helps,

Most processors only have 2 rings, intel x86 processors beyond 286 have 4.

Ring 0 : Most privileged, mostly used for system code
Ring 1: Mostly used for driver code
Ring 2: Mostly used for system libraries
Ring 3: Least privileged, mostly used for user applications

2 level processors can be equate to only having a ring 0 and ring 3. Some OS (Eg Linux afaik) only use ring 0 and ring 3 even on x86 architecture for portability reasons.

What this means. A processor operating in ring 3 can't access ring 0 code/data without passing through a call gate, etc. Basically most of the hardware protection hangs off the protection level the code is currently running in.

The intel docs give a pretty good explanation of how these work, so have a read of those if you're confused.

Curufir

Curufir, do you know any valid link for the documents
describing the ring structures ?

http://x86.ddj.com/intel.doc/386manuals.htm

The one you want is "Intel Architecture Software Developer's Manual, Volume 3: System Programming Guide". Look under protection.

Curufir

stilll not clear

i need some expert's opinion here or some technical
papers on this topics

for the last 5 or 6 days i have been searching all around
the web and found a very little amount of relevant
data

gone through all of my text books, and still not clear

after all that time i have finally found some thing
very informative on this subject, here is the link ;D

http://www.cs.berkeley.edu/projects/spr ... apers.html

..

the question was

which one is a better design for a distributed operating
system ?

A monolithic OR A Micro Kernel ?

do you really think this matter is discussed in intel
manual ??? :-\

the site i have refered, has a paper which compares
a monolithic DOS (Sprite) with a micro DOS (Amoeba)

In a distributed OS, you usually don't know where the services like display, database, applications and filesystems are running on the network, and you don't want the applications to see the difference between opening a local file or a remote file.

You don't want either the application to see whether the display is remote or local, etc.
In that way, most communications will use a message-based approach (because you can hardly rely on shared memory in the case of distributed OSes :-/ ), which is usually available in micro-kernels and widely used.

So i would say micro-kernels are better suited to Distributed OSes ... though other people could tell you that Unix+X-Window+NFS looks like a distributed system and that it has monolithic kernel ...