> Date: Wed, 16 Jun 1999 22:58:00 +1000
> From: Richard Gooch <rgooch@atnf.csiro.au>
>
> So here we are, arguing about names, because suddenly Ted admitted
> that a virtual FS might be good. But then he says he doesn't like the
> names. Before that he was denying to the hilt that a virtual FS had
> any merit.
>
> Sigh. Let me try again. The fact that people are arguing about names
> means that it is clear the kernel should *not* be dictating what shows
> up in /dev. That should be done by a user-mode daemon. Like it or not,
> the names in /dev *are* policy, and kernel should not be dictating
> policy. Therefore, the kernel should not be deciding what names appear
> in /dev. Q.E.D.
>
The kernel dictates policy.
That is the way it is, has been, and always will be.
The current policy requires a filesystem with special
attributes to be specially set to access devices
made available by the kernel.
The policy supplied by devfs gives us device names directly
that we may then manipulate freely.
You are used to the current policy, whereas myself, Richard
and many others have tried and find that we prefer the new
policy.
I personally have a system that switches back and forth
regularly for various reasons. It does so quite readily.
The two systems are NOT gratuitously incompatible.
I find that I prefer my system in devfs mode.
If I see a device in /dev with devfs I know that there
is a real device there. This cuts down immensly on
troubleshooting time for device initialization failures.
It also means that I do not have tons of devices hanging
out in /dev that have no purpose on my system.
(Not to mention making life much easier for EPC/X support, 224
devices on one card, you do not get two cards in with 8/8 Major/Minor
without doing some registration tricks that devfs renders more
useful).
Daniel Taylor
-
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/