Re: >256 ptys (previous subject line was garbage)
Richard Gooch (Richard.Gooch@atnf.CSIRO.AU)
Fri, 5 Jun 1998 21:46:24 +1000
Jakub Jelinek writes:
> >
> > tytso@mit.edu writes:
> > > Date: Mon, 1 Jun 1998 14:56:12 +1000
> > > From: Richard Gooch <Richard.Gooch@atnf.CSIRO.AU>
> > >
> > > > Isn't the devfs supposed to allow major/minor-less devices? I seem to
> > > > recall that for some devices major/minor number were used anywas since
> > > > userland code used it (mounted filesystems e.g.). For ptys this should be
> > > > no problem.
> > >
> > > Devfs does allows major/minor-less devices. You could hack the tty
> > > driver to make use of this. Check out the devfs FAQ at:
> > > http://www.atnf.csiro.au/~rgooch/linux/devfs.txt
> > >
> > > This is actually a bad thing, since there's been a lot of talk about
> > > using major/minor numbers as part of a tty locking scheme.
> >
> > Well, you don't actually *have* to get rid of major/minor
> > numbers. Devfs also allows you to have automatic allocation of device
> > numbers as well as supporting normal hard-wired major/minor numbers.
> > In the devfs FAQ I propose automatic device number allocation for SCSI
> > and tty devices. This would give us up to 64k SCSI partitions and tty
> > devices without breaking userspace.
> >
> > BTW: what is this tty locking scheme?
>
> When most of the systems use glibc at the moment, I think there is no need
> for major/minor-less devices: even glibc2.0 has u_quad_t dev_t, that is
> 64bits... So, the only thing one has to update is glibc to make use of huge
> device numbers, while keeping locking schemes work, fdisk knowing block
> major 8 is SCSI, etc.
There are plenty of good reasons for sticking with libc 5. Automatic
device numbers would allow people with libc 5 systems to access >256
ptys and >16 SCSI discs.
I don't see why autmatic device number allocation would break
anything. I imagine this tty locking scheme (whatever it is) would
still work, as would fdisk.
Regards,
Richard....
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu