NetBSD-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: block device lock difference between netbsd-4 and netbsd-5

On Wed, Feb 03, 2010 at 02:59:47AM -0500, Thor Lancelot Simon wrote:
> > Why is it more of a security hole than accessing the raw device?
> I didn't suggest that permitting access to the raw device for mounted
> filesystems (or overlapping mounted filesystems) was not a security hole.

Then this explanation doesn't help to explain why block devices
were limited but raw devices were not.

> It's All Bad.

It's how you do e.g. backups on live file sytems.

After all you could umount and mount and use the block device in between.
Unless of course you forbid mounts with securelevel=2 (or any similar
abstraction available in kauth).

I agree that block and raw device access should be locked out at
securelevel=2. But this has nothing to do with locking out concurrent

There may also be a limitation on who can _write_ to a block or raw
device. Writes to a mounted device may corrupt and crash the
filesystem. Again this has nothing to do with the number of users
but with mounted device vs. other types of access.
Of course, that is a matter of taste. UNIX lets you do smart things
but it also lets you do stupid things. For me this can easily be handled
by proper permissions to the device files.

                                Michael van Elst
                                "A potential Snark may lurk in every tree."

Home | Main Index | Thread Index | Old Index