Subject: Re: compartmentalization of kernel memory
To: None <>
From: John Gordon <>
List: tech-kern
Date: 04/08/2003 17:32:30

> Unless you are going to limit yourself to 15 processes maximum, then you 
> are going to have to support multiple L1s.   So most of the benefit of 
> domains (fast context switching, since the virtually-addressed cache 
> doesn't need flushing) are lost as soon as you have 3 or 4 L1s that are 
> active in the system.

For many embedded systems (which is my background) 15 "processes" + a kernel is
often more than enough. Even for PDAs that is probably OK. WinCE only supports
16 processes doesn't it? I believe WinCE uses the StrongARM process ID value
and not the domain mechanism though.

As you say though, I don't think that they'd be really useful in something like
NetBSD unless it was in a *very* minimal configuration system. Even the
StrongARM process ID feature is not much good for anything more complex than

> Finally, domains are rarely used and there are rumours that some chips 
> don't implement them properly -- I've no personal experience of using 
> them, so I can't be sure of this.

VxWorks uses them in a very minimal way to protect the page tables on some of
the ARM chips (SA-1100 I think was where this issue first came up IIRC, so it
is probably on the SA-1110 as well, and perhaps even the newer XScale devices).
I don't think we ever used them on the other parts.


Rate Corporate America at

Do you Yahoo!?
Yahoo! Tax Center - File online, calculators, forms, and more