[6930] in Athena Bugs
vax 7.2L: attach -m [-f]
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Mon Jan 21 14:48:47 1991
To: bugs@ATHENA.MIT.EDU
Date: Mon, 21 Jan 91 14:48:23 EST
From: Ralph Swick <swick@ATHENA.MIT.EDU>
System name: lyre
Type and version: MVAX-II 7.2L
Display type: QVSS
What were you trying to do?
Reattach an in-use fs in order to get write access.
What's wrong:
Two problems. Since the fs was already attached and I'm
unable to detach it ('cause I can't or don't want to kill
the processes w/ open files on it), I know I've got to
do something special.
I try:
% attach -m /mnt -w myfilsys
which fails, reporting
attach: Filesystem "myfilsys" is already attached (mapping)
I argue that this is a bug, 'cause myfilsys is actually mounted
somewhere else; my request was *not* a no-op. I shouldn't have
to use the -f hammer in this case; attach should be wiser, both
since the mountpoint is different and the mode is different.
Ok, so I try:
% attach -m /mnt -w -f myfilsys
which successfully mounts the fs, but munges attachtab so that
it contains two lines both claiming myfilsys is mounted on /mnt.
The old (default) mountpoint is lost and I presume detach
will eventually fail.
What should have happened:
In the case that the fs is already attached but (only) the mode
is different, attach should attempt to change the mode if
sensible/possible. If it fails to change the mode, an error
message indicating the attempt was made and failed should be
issued. If changing the mode is deemed impossible, attach
should report the error as currently worded, with add words
indicating the current mode (and not attempt the mapping?).
In the case that the fs is already attached but the mountpoint
is different, attach should attach the fs without further ado.
No message, no nothing; the user should be presumed to be smart
enough to know why s/he specified -m. Attachtab should show
both attaches with their correct respective mountpoints.
If you disagree with the above, then the man page should
document this particular case; i.e. specifically point out
the necessity of using -f with -m if the fs is already attached.
Note that this is dangerous if the fs is actually already
attached at the specified mountpoint. In any case, attachtab
should always reflect reality.
Please describe any relevant documentation references:
man attach
