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

Re: securelevel=3



Tobias Weingartner <weingart@wolfram.com> writes:

> On Thursday, May 14, stanislav shalunov wrote:
> > 
> > It might be useful to have some sort of securemask on per process
> > basis (with a special system call that can _set_ some bits of
> > it--there should be no way of clearing a bit, and it should be
> > inherited by children).
[snip]
> 
> It would be nice to see this implemented.  I'm not sure, but there is a
> "class" field in master.passwd, which could be used to point to a database
> of sorts, which could be used by login,xdm,etc for initializing this
> per-process field. Of course, then some substantial changes to the kernel...

I was looking a few weeks ago at something like this implemented for Linux.
There is a standard (posix.6) that specifies how this kind of fine-grained
control should look like. But in that case it applies more to super-user
privilegies, but it could easily be extended to cover normal users.

http://csrc.ncsl.nist.gov/nistpubs/800-7/node17.html#posix6

I'm still not conviced that you can improve security by making the control
mechanism more complicated.

//art
--
Standards are good. Everybody should have one.