Re: Journaled FS
Richard Gooch (Richard.Gooch@atnf.CSIRO.AU)
Fri, 19 Jun 1998 21:31:25 +1000
david parsons writes:
> In article <linux.kernel.Pine.GSO.3.96.980617121154.8969C-100000@adina.cs.unibo.it>,
> Andrea Arcangeli <arcangel@cs.unibo.it> wrote:
>
> >I hope fs developers will remeber to use at least 32bit to store major and
> >minor numbers (instead of only 16bit) while projecting the newfs (to
> >remove the need of devfs).
>
> The size of major and minor numbers is not, IMO, a significant reason
> to have a devfs or not. Having device drivers be able to tell you
> that they exist and what devices they're handling -- in a consistant
> fashion, btw -- *IS* a significant reason, and this won't be stopped
> even if the kernel goes to the glibc 64 or 128 or 256 bit major or
> minor numbers.
In fact, larger device numbers are going to cause performance problems
(think about those tables of fops[NUM_MAJORS] when NUM_MAJORS jumps
from 256 to 65536 or beyond).
> ____
> david parsons \bi/ In the grand scheme of things, major and minor
> \/ numbers are a massive kludge.
No kidding. In retrospect, it's one of the sillier ideas in UNIX.
Regards,
Richard....
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu