Force owner for all filesystems

Bug #27417 reported by Nikolaus Rath
8
Affects Status Importance Assigned to Milestone
linux-source-2.6.15 (Ubuntu)
Invalid
Wishlist
Ben Collins

Bug Description

- User creates files on xfs/ext3/reiserfs formatted USB disk on system A
 - User needs this files on system B
 - After mounting the disk on system B, the files can't be accessed because user
has different UIDs on systems A and B
 - Even worse: Some other on system b has now full access to the files

--> force all files on device to be owned by pmount caller (like for fat), no
matter what filesystem is used?

Revision history for this message
Martin Pitt (pitti) wrote :

The kernel currently does not support overriding owners for file systems that
properly support permissions (like ext2).

pmount will never be made to override permissions on these file systems by
default. It might get an option to do that if the kernel supports it, though.

However, I do not understand the use case: If you specifically want to disable
permissions, they why do you use ext2, and not vfat in the first place?

Ben, please feel free to close this as wontfix if it's totally unreasonable to
add such functionality to the mount call.

Thanks

Revision history for this message
Ben Collins (ben-collins) wrote :

(In reply to comment #1)
> Ben, please feel free to close this as wontfix if it's totally unreasonable to
> add such functionality to the mount call.

Yeah, I just don't think this is feasible. For vfat, this is done transparently,
because the filesystem itself doesn't support uid/gid. So it already has a
mechanism to emulate them (they don't really exist). Because it can do that, it
can emulate them at arbitrary values. It does not support changing the file
modes and permissions. So it has to have this emulation (e.g., you cannot change
the owner/group of a file on vfat, it's _always_ set at whatever it was mounted
with).

For things like ext2/ext3/xfs/etc, the file perms are specifically there for a
reason. Plus, these filesystems support other modes (suid, sgid) that could
prove dangerous in this use case.

I wont ever support it, but if upstream ever does, we'll get it sooner or later.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.