Matthew Jacob writes:
>
>
> > Peter Svensson writes:
> > > On Thu, 27 Apr 2000, Richard Gooch wrote:
> > >
> > > > I see no reason we can't have /dev/scsi/identifiers/ as well.
> > >
> > > And even for disks it makes sense. You want to mke2fs (or your
> > > filesystem of choise) the correct physical disk. You cannot really
> > > depend on the label in that situation, but a WWN sounds like just
> > > what is needed.
> >
> > Or just rely on host,bus,target,lun names. It's fair enough to assume
> > that devices aren't being replugged while you're in the middle of
> > typing "mke2fs"!
>
> No, it's not. LIPs can happen any time, people can pull drive
> carrires in an Andataco SCSI box at any time. But I believe it would
> be fair for the system to demand that while it has a device open
> either the address doesn't change (thus requiring virtual addressing
> at some level) or you get an I/O error back while trying to get to
> it when it changed.
It may be theoretically possible, but it's not *sane*. If I can't be
sure of what's plugged into the system while I'm making labels (or
just scanning for labels!), then I can't work effectively.
Even if I knew the serial number on the platter, and could address by
that, I can't be sure someone won't unhook the drive just as I hit
return on "mke2fs". At some point you have to draw the line and say
"I'm configuring: keep your filthy hands off the vaults!".
Regards,
Richard....
Permanent: rgooch@atnf.csiro.au
Current: rgooch@ras.ucalgary.ca
-
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/
This archive was generated by hypermail 2b29 : Sun Apr 30 2000 - 21:00:17 EST