Skip to contentSkip to conversation

OK to mount multiple FS in one dir?

DISCUSSION Michael D. Crawford 7 participants 10 messages Feb 5, 2001 – Feb 25, 2001

lore
Share & tools
lore
#0 lore raw

I was groping around my FAT/NTFS directories from Linux, mounting and unmounting them into /mnt, and was suprised at some point that I got the message "/dev/sda5 already mounted or /mnt busy". (I'm using a SCSI disk, use hda* for IDE).

Upon further examination, I found that I'd accidentally mounted /dev/sda1 (VFAT) on /mnt while /dev/sda5 (NTFS) was still mounted there. The NTFS files remained invisible until I'd unmounted /dev/sda1 and then I could see them again.

This is with the 2.4.1 kernel on a Pentium III machine with an Adaptec 29160 SCSI controller.

I found I could mount three partitions on /mnt, and they'd all show up as mounted at /mnt in the "mount" command, but if I unmounted one of them (only tried with the currently visible one), then it appeared that there were no filesystems mounted there, but I could continue umounting until the other two were gone.

I'm suprised this works. Note that the kernel rejected an attempt to mount a filesystem that was already mounted, but not to mount a filesystem at a point that was already in use. It looks like there is a stack of mounts on the mount point.

Looking at Documentation/Changes, I see that I need util-linux 2.10o. I had 2.10l. But I had the 2.10r util-linux sources on my machine and installed mount and umount from it, and I find that it gets it right mostly when I mount and unmount multiple things, with the exception that if /dev/sda5 was mounted before /dev/sda1, then if I give the command "umount /dev/sda5", sda1 is the one that gets unmounted rather than sda5, so it takes the most recently mounted filesystem rather than the one you specify.

Michael D. Crawford
GoingWare Inc. - Expert Software Development and Consulting
http://www.goingware.com/
[email protected]

Tilting at Windmills for a Better Tomorrow.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

#1 lore raw

[Michael D. Crawford]

Quoted 1 line

I found I could mount three partitions on /mnt

Yes. New feature, appeared in the 2.4.0test series, or shortly before.

Quoted 4 lines

and they'd all show up as mounted at /mnt in the "mount" command, but if I unmounted one of them (only tried with the currently visible one), then it appeared that there were no filesystems mounted there, but I could continue umounting until the other two were gone.

util-linux gets rather confused by this feature. They say newer versions fix this.

Quoted 7 lines

But I had the 2.10r util-linux sources on my machine and installed mount and umount from it, and I find that it gets it right mostly when I mount and unmount multiple things, with the exception that if /dev/sda5 was mounted before /dev/sda1, then if I give the command "umount /dev/sda5", sda1 is the one that gets unmounted rather than sda5, so it takes the most recently mounted filesystem rather than the one you specify.

I think this is a kernel limitation. 'umount' takes '/dev/sda5' and turns it into '/mnt/test' and calls umount("/mnt/test"). The kernel then unmounts whatever is on "top" of /mnt/test.

I don't think there's anything umount can do about this behavior.

Peter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

#2 lore raw
In reply to Michael D. Crawford

----- Original Message -----
From: "Peter Samuelson" <[email protected]>
To: "Michael D. Crawford" <[email protected]>
Cc: <[email protected]>
Sent: Tuesday, 6. February 2001 00:06
Subject: Re: OK to mount multiple FS in one dir?

Quoted 22 lines

[Michael D. Crawford]

Quoted 1 line

I found I could mount three partitions on /mnt

Yes. New feature, appeared in the 2.4.0test series, or shortly before.

Quoted 4 lines

and they'd all show up as mounted at /mnt in the "mount" command, but if I unmounted one of them (only tried with the currently visible one), then it appeared that there were no filesystems mounted there, but I could continue umounting until the other two were gone.

util-linux gets rather confused by this feature. They say newer versions fix this.

Quoted 7 lines

But I had the 2.10r util-linux sources on my machine and installed mount and umount from it, and I find that it gets it right mostly when I mount and unmount multiple things, with the exception that if /dev/sda5 was mounted before /dev/sda1, then if I give the command "umount /dev/sda5", sda1 is the one that gets unmounted rather than sda5, so it takes the most recently mounted filesystem rather than the one you specify.

Quoted 5 lines

I think this is a kernel limitation. 'umount' takes '/dev/sda5' and turns it into '/mnt/test' and calls umount("/mnt/test"). The kernel then unmounts whatever is on "top" of /mnt/test.

I don't think there's anything umount can do about this behavior.

What about userland umount checking which device is umounted and refusing to umount it or at least issuing a printf warning?

Quoted 2 lines

Peter

-mirabilos

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

#3 lore raw
In reply to Michael D. Crawford
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

I found I could mount three partitions on /mnt

Yes. New feature, appeared in the 2.4.0test series, or shortly before.

I have a question, why was this idea even considered?

Signature · 5 lines
 Lab tests show that use of micro$oft causes cancer in lab animals
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
#5 lore raw
In reply to Michael D. Crawford

[Wakko Warner]

Quoted 1 line

I have a question, why was this idea even considered?

Al Viro likes Plan9 process-local namespaces. He seems to be trying to move Linux in that direction. In the past year he has been hacking the semantics of filesystems and mounting, probably with namespaces as an eventual goal, and this is one of the things that has fallen out of the implementation.

A more useful thing to fall out of the same hacking is loopback mounting -- i.e. the same filesystem mounted multiple places. In Linux-land I guess we call it 'mount --bind'.

Peter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/

#6 lore raw
In reply to Michael D. Crawford
Quoted 9 lines · Peter Samuelson

[Wakko Warner]

Quoted 1 line

I have a question, why was this idea even considered?

Al Viro likes Plan9 process-local namespaces. He seems to be trying to move Linux in that direction. In the past year he has been hacking the semantics of filesystems and mounting, probably with namespaces as an eventual goal, and this is one of the things that has fallen out of the implementation.

Aren't "translucid" mounts the idea behind this?

Signature · 7 lines
John Lenton ([email protected]) -- Random fortune:
For courage mounteth with occasion.
		-- William Shakespeare, "King John"
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
#8 lore raw
In reply to Michael D. Crawford
Quoted 5 lines · Peter Samuelson

A more useful thing to fall out of the same hacking is loopback mounting -- i.e. the same filesystem mounted multiple places. In Linux-land I guess we call it 'mount --bind'.

Peter

Does this kind of thing play nice with nfs and coda, in terms of change notifications and write-backs? In distributed FS we've got the same thing mounted multiple places, of course, but not on the same machine

Signature · 7 lines
                      David Nicol 816.235.1187 [email protected]
                           Pedestrians always have the right of way

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
#4 lore raw
In reply to Wakko Warner
Quoted 5 lines · Wakko Warner
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

I found I could mount three partitions on /mnt

Yes. New feature, appeared in the 2.4.0test series, or shortly before.

I have a question, why was this idea even considered?

Direct request from HPA. Autofs can win from having that (mounting
atop of mountpoint). I'd rather live without that stuff, but back then it looked like an OK idea - we could do that. There is a better solution for original problem, but...

#7 lore raw
In reply to John R Lenton
Quoted 5 lines · John R Lenton
Quoted 4 lines · Peter Samuelson

[Wakko Warner]

Quoted 1 line

I have a question, why was this idea even considered?

Quoted 7 lines
Quoted 5 lines

Al Viro likes Plan9 process-local namespaces. He seems to be trying to move Linux in that direction. In the past year he has been hacking the semantics of filesystems and mounting, probably with namespaces as an eventual goal, and this is one of the things that has fallen out of the implementation.

Aren't "translucid" mounts the idea behind this?

Nope. Completely different beast - bindings have nothing to layered filesystems. I.e. if we bind /foo to /bar then /foo/barf and /bar/barf are the same object. Translucent-type would have one of them redirecting all requests to another.

#9 lore raw
In reply to David L. Nicol
Quoted 6 lines · David L. Nicol
Quoted 4 lines · Peter Samuelson

A more useful thing to fall out of the same hacking is loopback mounting -- i.e. the same filesystem mounted multiple places. In Linux-land I guess we call it 'mount --bind'.

Quoted 6 lines
Quoted 1 line

Peter

Does this kind of thing play nice with nfs and coda, in terms of change notifications and write-backs? In distributed FS we've got the same thing mounted multiple places, of course, but not on the same machine

There is no cache coherency problems since we have no copies to keep in sync ;-) Dentry tree is shared by all instances.