On Thu, 2010-06-10 at 15:40 -0400, Stephen Smalley wrote: > On Thu, 2010-06-10 at 23:57 +0500, Shaz wrote: > > Dear all, > > > > I have a mobile use case where TE is kind of too heavy ... if you > > remember from earlier threads that discussed the use case. Can I > > somehow use the SELinux infrastructure ... kernel and userspace to > > only use the rbac and uid models and leave out TE? > > > > How much is it going to take as a technical effort and some fine > > guidelines will be appreciated. > > > > Will it be called SELinux if we get it done in the first place? > > TE (the mechanism) isn't particularly heavy, so I presume you mean the > policy configuration for it. There are multiple options there: > - You can build a subset of refpolicy, similar to what Fedora does in > its selinux-policy-minimum package. That will certainly yield a smaller > policy than a full refpolicy, but there are some fundamental limitations > on what you can achieve using that approach. > > - You can construct your own policy from scratch, generating an initial > working one via scripts/selinux/mdp in the kernel source tree and then > expanding upon it. That will yield the smallest possible policy. > > - You could try the older SEEdit work by Hitachi Soft, although it isn't > being maintained anymore AFAIK. And just to highlight the range we're talking about in policy sizes: 13K /etc/selinux/dummy/policy/policy.24 208K /etc/selinux/seedit/policy/policy.24 808K /etc/selinux/minimum/policy/policy.24 5.5M /etc/selinux/targeted/policy/policy.24 where: dummy = minimal policy generated from kernel source tree, seedit = seedit base policy, minimum = selinux-policy-minimum (stripped down refpolicy) targeted = full targeted policy -- Stephen Smalley National Security Agency -- This message was distributed to subscribers of the selinux mailing list. If you no longer wish to subscribe, send mail to majordomo@xxxxxxxxxxxxx with the words "unsubscribe selinux" without quotes as the message.