Subject: Re: partition bookkeeping
To: Greywolf <greywolf@starwolf.com>
From: Oleg Polyanski <luke@eed.miee.ru>
List: tech-kern
Date: 09/23/1999 09:08:22
>>>>> "G" == Greywolf  writes:

 G> # 	Consider my disk has name `sd0' (because it was detected first during
 G> # 	the kernel booting). Later  I will add ZIP  drive and IT will get the
 G> # 	name `sd0' and hard  disk will  become  `sd1'.  Bad  surprise, manual
 G> # 	intervention required.  Something like `/dev/scsi/dsk/c0d0p0' as root
 G>                                                 ^^^^^^^^^^^^^^^^^^^^
 G> 	AAAUUUUGGGHHHHHHHH!   -------------------'

 G> /me smacks head against brick wall

 G> "Can't sleep, clowns will eat me....can't sleep, clowns will eat me..."

 G> Are we going to continue dumbing down this thing, or what?  I agree
 G> that we will need a /dev/disk or something similar directory, but
 G> I do NOT think that we should go further down than that -- it's just too
 G> cumbersome.  /dev/$BUSTYPE/$PERIPHTYPE/cXtXpX is just fscking
 G> ludicrous.  Not to mention, hard to type repetitively (and I've
 G> had to do that).

	Huh?

 G> If you're worried about devices switching locations, WIRE DOWN YOUR
 G> DAMNED DEVICES and be *done* with it.  [Profuse apologies here

	Your forgot about `the human factor'.

 G> for assuming that everyone uses source -- I'm sure that this is
 G> a patent lie; nonetheless, I find it hard to believe that there
 G> are that many people incapable of compiling a new kernel for reasons
 G> of disk space.]

 G> As far as the root device goes, using Ignatios' idea of mounting /kern and
 G> referencing /kern/rootdev as a valid device makes a WHOLE lot of sense.  
 G> You'll at least be able to put yourself back together without too much
 G> grief.

 G> Should we (do we?) have a peripheral scanner which will configure
 G> a kernel with wired down devices followed by the generic devices
 G> and (optionally) attempt a rebuild?

	I think we   really need it.  Without  peripheral  scanner  that maps
	current  hardware hierarchy into kernel  and file system any new disk
	naming scheme is unnatural.

 G> If we don't, is someone already working on it?

 G> If nobody's working on it, I'll be glad to take a stab at it...

	I will glad to help/cooperate.

 G> # 	fs unique identifier will help.