From: S_TITZ@iravcl.ira.uka.de (Olaf Titz) Subject: Re: RISC approach to OS - Re: GNU kids on the block? Date: 31 Aug 1992 11:17:58 GMT
In <1992Aug28.174743.8140@colorado.edu> drew@ophelia.cs.colorado.edu writes:
> In article <1992Aug28.171744.6460@rz.uni-karlsruhe.de> S_TITZ@iravcl.ira.uka.de (Olaf Titz) writes:
> >
> >While the Linux kernel does its job well, its being monolithic is a
> >problem since all of the parts are interdependent, and to comprehend
> >the work of one of them, you have to know the whole system. This may
> >work for Linux but is unacceptable for bigger systems.
>
> This isn't true - although the kernel is one big chunk of
> code, it is subdivided into smaller modules, such as the VM
> code, each one of the file systems, each device driver, etc. Some of these
> modules have incestous relationships with eachother due to various
> kludges and hacks, but this isn't necessary.
OK, its modularized, but the problem ARE the kludges and hacks. I have
done some patching of the screen blanking code - I thought it was in
console.c but was astonished how many modules actually were involved.
(Btw. console.c is too big and does too many things, IMHO. But there's
nothing to say fundamentally against it, as all is working well :-)
> You do have to understand some basic "common" things used in the
> kernel - ie, how the fs register points into user space, how
> sleep/wakeup work, etc, but to say you have to understand the
> whole kernel to write a device driver is a gross overstatement.
OK, it's an overstatement. And the scheduler is indeed one of the
finest I know - sleep/wakeup can really be /that/ easy...
>...
> I agree too - 1.5 - 3M for a typical vendor's kernel is outrageous,
> when it doesn't do anything that a generic 500K BSD or Linux kernel
> can't do.
By developing ever bigger RAM chips hardware manufacturers unwillingly
(???) spawn new proofs for Parkinsons Law of Data every 2 years :-(
>...
> Blasphemy! There is nothing but the one true EMACS! :-)
Tell me which version :-)
> Boycott AT&T for their absurd anti-BSDI lawsuit.
System V Rel. 4 is a sufficient reason too. Cf. memory consumption.
MfG,
Olaf
-- Olaf Titz - comp.sc.student - Univ of Karlsruhe - s_titz@iravcl.ira.uka.de - uknf@dkauni2.bitnet - praetorius@irc - +49-721-60439 - did i forget something? The only use I can find for vi is editing the emacs sources while porting them to a new machine. - Larry Campbell