SPECS: kernel.spec - mount /boot (see -devel-en ml for more inform...

[Paweł Sakowski]

Translating, this should really be an English discussion for the
interested developer who doesn't speak Polish.

On Tue, 2005-01-04 at 17:26 +0100, Witold Krecicki wrote:
> Dnia wtorek 04 styczeń 2005 16:53, Adam Gołębiowski napisał:
> > On Tue, Jan 04, 2005 at 04:26:22PM +0100, Paweł Gołaszewski wrote:
> > > On Tue, 4 Jan 2005, Charles wrote:
> > > > > +%pre
> > > > > +# on target system the /boot might not be mounted (it's not required
> > > > > for system +# to run properly), it could be also mounted ro
> > > > > +mount /boot >/dev/null 2>&1
> > > > > +mount /boot -o remount,rw >/dev/null 2>&1
> > > > > +exit 0
> > > >
> > > > I don't believe it's the best solution. Automatically mount
> > > > disks/partitions in an rpm session?
> > >
> > > Especially because if I do have it unmounted... that's for a
> > > reason. And that state should be retained after rpm's operation
> > > (meaning, if it's not mounted, it should be unmounted after
> > > installation)
> >
> > Is is possible to pass some value from %pre to %post? There's no
> > problem then - in %pre invoke grep twice (for /etc/fstab
> > and /proc/mounts) and we know if we're actually going to
> > mount /boot and should unmount it in %post.
> It would also be useful to detect 'noauto' - somebody might not want
> to use /boot (even though it's in fstab). And the system should be
> installed on the system as it is - among others with the partitions
> mounted. IMHO the solution is mistaken, at the least.

-- 
Paweł Sakowski <saq at pld-linux.org>
PLD Linux Distribution