lkml.org 
[lkml]   [1996]   [May]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: dropping kerneld...
To knock a capability just because it is something you personally don't
use and doesn't fit into your way of doing things is totally
unproductive. I believe the kerneld approach to be an advanced
technigue that should provide a big benefit in the future. Demand
loading of capabilities when needed definately opens up the opportunity
to build driver capabilities with greater freedom. I believe from
reading Linus's original coment it was more directed along the lines I
don't use it so I won't fix it and if it doesn't get fixed by somebody
who wants it, it will be deleted. I looked at that as a challenge to
get it fixed, not as a reason to abondon it.

I personally use kerneld, I find the concept intrigueing and see no
disadvantage in saving Ram and possibly cache space by the automatic
unloading of those items not normally needed. I personally tune my
system by selectively loading some drivers and using kerneld for others.
This automates the use of modules letting the machine remember what is
loaded and what isn't instead of my manually loading and uloading them.
By the way their are modulules that can't exist simultaneously such as
different drivers for the parallel port depending on use at a given
time.

No matter what size machine I have available I always find a way to
give it a bigger job than it is optimized for. Including the 8192
processor MasPar that I ended up writing code to virtulize the number of
proccessors so I could use 16 times the number. Just because you seem
to have plenty of memory now doesn't mean that will always be the case.

Hubert Bahr
hab@ece.engr.ucf.edu

Lauri Tischler wrote:
>
> Anno Domini 30 Apr 96 at 8:01, Matthias Urlichs wrote:
> > In linux.dev.kernel, article <6F5724445CA@espoo.fipower.fi>,
> > "Lauri Tischler" <ltischler@fipower.fi> writes:
> > > Yup, I'm missing a lot of pain and useless complications by not
> > > using kerneld and modules. Quite useless contraptions both.
> > >
> > Before modules and initrd, I've had to build a kernel for each of these
> > beasts. Ugh. Now, I build the kerneld 100% modularized (OK, OK, binfmt_elf
> > and ext2 aren't, but they're used on every system anyway). Initrd then just
> > loads one SCSO adapter after the other until one is found...
> Yech.. sounds nutty and dangerous.
> Why dont you just load appropiate modules directly without kerneld ?
>
> I have said before that kerneld _might_ have a place somewhere where
> underpowered hardware is the _only_ choise. (compare travelling 3000
> km's with bicycle instead by airplane).
>
> Saving 100-300 kilobytes of memory is truly not worth the hassle in
> any 'normal' systems.
>
> --
> Thought for the day:
> Communist (n): one who has given up all hope
> of becoming a Capitalist.
>
> Lauri Tischler, Network Admin Tel: +358-0-47846331
> Finlandia Interface Oy Fax: +358-0-47846500
> Piispanportti 12 (PL 61) GSM: +358-40-5569010
> 02211 Espoo FINLAND Internet: ltischler@fipower.fi


\
 
 \ /
  Last update: 2005-03-22 13:37    [W:0.048 / U:0.280 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site