Re: Linux Security Standards
Tobias Burnus <[email protected]> Thu, 22 Jan 2004 10:34:06 +0100
| Newsgroups | gmane.linux.lsb.discuss |
|---|---|
| Message-ID | <[email protected]> |
Hello, On Wed, Jan 21, 2004 at 04:08:01PM -0800, Christian Avery Bryant wrote: > I am curious to know what stress the LSB will be placing on Linux security > in future versions. There is a standard out that is called the Common Criteria > for Information Technology Security Evaluation. While wholly implementing > such a standard for a Linux distribution is not fully feasible (although, > one could argue otherwise), Well, what do you mean by 'wholly'? EAL 7? Then it is probably not feasible. On the other hand, SLES8 on certain IBM hardware has EAL3+. What is the advantage of having CC certificate? It does not increase the security, costs a lot of money and is only valid for a subset of a specific Linux version on a restricted number of hardware. (Well, actually it increases security by providing some test cases and forcing an audit of the structures of the given linux distribution and the distributer.) > it does beg the question: What IS the Linux community doing to insure the > security of a base system, on a level that is accessible to users who are > not fully versed in the art of Kernel hacking, or system administration? Note that the LSB as of now provides 'only' a execution environment, the underlaying system could be e.g. a SUN running Solaris. Therefore, some of the security aspects are really out of scope for the LSB. Those are rather covered by e.g. the Carrier Grade Linux and Data Center Linux as defined by the OSDL (http://www.osdl.org/). And test cases, including those used for the EAL2+ are also available http://ltp.sourceforge.net/ thus there is something which the 'community' has done. Regarding the LSB, it requires PAM and things like password ageing (which in turn de facto require shadow passwords). The question is which other means of security are feasible and make sense within the scope of the LSB. One think I would like to see are Access Control Lists, but this will probably take some time. (The LSB prefers to standardize existing common practice rather than requiring new methods.) If you have things -- security related or not --, which should be included in the LSB, go ahead and suggest them. Regards, Tobias Burnus