[PATCH] eject -T support

[Rob Landley]

On Saturday 07 October 2006 5:12 pm, Denis Vlasenko wrote:
> > > I tested it and it worked for me as shell script.
> > 
> > With busybox umount!?
> 
> Yes:

I put rather a lot of effort into rewriting umount and making it work in all 
the weird cases I could think of.

I don't remember if I ever actually implemented "umount -a /dev/hdc".  The 
default kernel behavior (18 months ago) _used_ to be to pop the most recent 
mount but leave the rest in the case of multiple mounts.  But apparently, 
umount /dev/hdc went away entirely, possibly to avoid the case where you can 
unmount mounts that have since been mounted over (which mades the kernel sad, 
and its linked lists dangly).

Not quite as bad as the circa 2.6.12 fun where you you could pivot_root rootfs 
and then umount it, and the kernel would lock _hard_ in an endless loop 
trying to find the end umount's doubly linked list (which was, of course, the 
same as the starting point, I.E. the rootfs entry).

Somewhere around here, I still have a mount/umount/remount sequence that makes 
the sucker panic.  Never did figure out what was going on.  Not a big deal, 
only triggerable by root...

Rob
-- 
Never bet against the cheap plastic solution.