Date:	Wed, 29 Mar 2000 18:36:41 +0200
From:	almesber@lrc.di.epfl.ch
To:	Andries.Brouwer@cwi.nl
Subject: Re: Location of shmfs; devfs automagics

Andries.Brouwer@cwi.nl wrote:
> People who enable devfs quickly find out that their system is broken
> and that they have to run a daemon to repair it.

Worse yet, they first have to figure out what broke their system ...
I'm not sure everyone appreciates the "You agreed to the pact. Your
soul is mine now." approach of devfs.

> Enabling it in the kernel should make power available to people
> who desire to use it. It should not kill well-functioning systems.

Fullest agreement.

> So (i) enabling devfs should not cause an automatic mount,
> and (ii) the canonical mount point should be /devfs, not /dev.

(iii) temporarily teach devfs to create entries for devices on the
underlying file system if a lookup doesn't find a match and there's
no devfsd. Later on, add a printk telling the user that they should
get serious with the migration to "clean" devfs. Probably also
printk a warning if some underlying devices were created before
devfsd is started, so people realize that their system isn't "pure"
yet, even if it looks that way. Even later on, remove that code.

With (iii) in place, it should work even if you don't have (i) and
(ii). Plus you gain the ability to eaily go back and forth in your
migration.

- Werner

-- 
  _________________________________________________________________________
 / Werner Almesberger, ICA, EPFL, CH       werner.almesberger@ica.epfl.ch /
/_IN_N_032__Tel_+41_21_693_6621__Fax_+41_21_693_6610_____________________/

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/