CLONE_NAMESPACE, links for dirs and mount(2) for normal users questions

CLONE_NAMESPACE, links for dirs and mount(2) for normal users questions

[Remi Turk]

Hi,

Long long ago, (March 2000) Alexander Viro replied to Pavel Machek:
>> Am I right that from now on each process can have completely different
>> view of filesystem like in plan9?
>
>Almost there ;-) And yes, the only thing we lack for proper namespaces is
>the union-directories (clone() bit is trivial).
Are there any patches already?
If not, where should I start to implement them?

Probably related to the first question, what about allowing mount(2)
(as a CONFIG-option) for normal user processes when they
have a) rw access to the device and b) are the owner/have rw-access
to the mountpoint. (There would be at least one security problem:
A normal user could mount a loopback ext2 filesystem with
panic-on-error (man tune2fs) and then corrupt it)

In April, Al Viro wrote:
> 1.  We should never have more than one dentry for a writable directory.
> 
> Print it and hang it on the wall. It's a fundamental requirement. There is
> no way to work around it in our VFS. I tried to invent a scheme that would
> allow that for more than a year. And I've done most of namespace-related code
> in our VFS since the moment when Bill Hawes stopped working on it, so I suspect
> that right now I have the best working knowledge of that stuff. There is no
> fscking way to survive multiple dentries for writable directory without major
> lossage. Period.
Do I understand correctly that this means hardlinks to directories
(except . and ..) are fundamentally impossible in Linux?
(I'm thinking about trying to write a garbage collected
filesystem with hardlinks to directories.)

-- 
Linux 2.4.0-test11 #1 Mon Nov 20 17:19:26 CET 2000
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: CLONE_NAMESPACE, links for dirs and mount(2) for normal users questions

[Peter Samuelson]

[Remi Turk]
> Do I understand correctly that this means hardlinks to directories
> (except . and ..) are fundamentally impossible in Linux?

Why do you want to be able to do that?  Use symlinks or loopback mounts
and stay out of trouble.

> (I'm thinking about trying to write a garbage collected filesystem
> with hardlinks to directories.)

Sounds like a lot of extra complexity.  Is this academic or do you have
a practical use for it?

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

Re: CLONE_NAMESPACE, links for dirs and mount(2) for normal users questions

[Remi Turk]

Peter Samuelson wrote:
> 
> [Remi Turk]
> > Do I understand correctly that this means hardlinks to directories
> > (except . and ..) are fundamentally impossible in Linux?
> 
> Why do you want to be able to do that?  Use symlinks or loopback mounts
> and stay out of trouble.

Probably just because I'm crazy ;-)
Of course I could use symlinks or loopback mounts, but it itn't nearly
as much fun as a directory which really contains itself :-)

> > (I'm thinking about trying to write a garbage collected filesystem
> > with hardlinks to directories.)
> 
> Sounds like a lot of extra complexity.  Is this academic or do you have
> a practical use for it?

Acedemic.
Well, I might find some practical use for it some day.

(I got inspired by an old thread about garbage collecting filesystems
without a clear answer about whether it was possible to write one in
Linux
(http://www.uwsg.indiana.edu/hypermail/linux/kernel/0001.1/0410.html)
and the recent thread  "[BUG] Inconsistent behaviour of rmdir")

> 
> Peter

-- 
Linux 2.4.0-test11-ac4 #1 Tue Nov 28 15:51:01 CET 2000
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/