[unios] Re: Generic design. More comments

Pat Wendorf beholder@ican.net
Thu, 17 Dec 1998 13:31:29 -0500


From: Pat Wendorf <beholder@ican.net>

> > Give an opcode timing example if you can... I'd like to know how much it would
> > slow it down.  This would be interesting to find out, if it is only a few hundred
> > instruction per task switch (vs. about twenty or thirty with a kernel), then I
> > don't think it will be too bad.  But if we are talking thousands of instructions
> > to handle it, we are gonna run into a massive speed hit.
> >
>
> It's not related to instruction timing or extra instructions. But if you
> put every part of the kernel in a seperate memory space, every time you
> have to access a part of the kernel, memory spaces must be switched. On
> the x86, this means a page directory switch wich might invalidate the
> cache if it is not big enough, and it means a too large amount of segment
> permission checking.
> If you put the whole kernel (microkernel) in every user process's memory
> space, only a code segment and a data segment must be switched on the x86.

How about on other systems?

--
-----------------------------
Pat Wendorf
UniOS Group
http://members.xoom.com/unios
beholder@ican.net
ICQ: 1503733
-----------------------------



------------------------------------------------------------------------
To unsubscribe from this mailing list, or to change your subscription
to digest, go to the ONElist web site, at http://www.onelist.com and
select the User Center link from the menu bar on the left.
------------------------------------------------------------------------
UniOS Group
http://members.xoom.com/unios