Re: Deprecating setlocaldefs, preservebools support in libselinux

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

 



On Thu, 2008-01-24 at 16:12 -0500, Chad Sellers wrote:
> On 1/24/08 4:07 PM, "Stephen Smalley" <sds@xxxxxxxxxxxxx> wrote:
> 
> > 
> > On Thu, 2008-01-24 at 16:02 -0500, Joshua Brindle wrote:
> >> Stephen Smalley wrote:
> >>> I'd still like to deprecate setlocaldefs support and preservebools
> >>> support in libselinux in the trunk (i.e. libselinux 2.x).  I posted
> >>> patches for completely removing such support a long while ago, but those
> >>> particular patches would require an ABI change (as they include API
> >>> removal) and thus I held off on them, but we could also take the more
> >>> intermediate approach of just turning off the functionality by default
> >>> in libselinux without disturbing the ABI.
> >>> 
> >>> As a refresher, setlocaldefs support refers to the support for pulling
> >>> in local boolean and user definitions at policy load time w/o managed
> >>> policy, i.e. the approach used in RHEL4 and Fedora 3 and 4 (but not in
> >>> Fedora 5 and later or RHEL5).  By default, libselinux still checks for
> >>> such definitions and patches them into the in-memory policy at load time
> >>> unless /etc/selinux/config has SETLOCALDEFS=0.  I'd like to make
> >>> SETLOCALDEFS=0 the default in the trunk and require SETLOCALDEFS=1
> >>> in /etc/selinux/config to enable the old behavior.
> >>> 
> >>> preservebools support refers to the support for preserving active
> >>> boolean values across a policy reload by having libselinux patch the
> >>> active values into the in-memory policy at policy load time.  As of
> >>> Linux 2.6.22 and later, this is now handled automatically by the kernel
> >>> as part of the policy reload and isn't needed in userspace.  I'd like to
> >>> also disable this by default in libselinux and perhaps allow it to be
> >>> enabled via some /etc/selinux/config setting.
> >>> 
> >>> Thoughts?
> >>>   
> >> 
> >> I'm fine saying its deprecated but CLIP currently uses an updated
> >> toolchain for both RHEL5 and RHEL4 (adds policy management capabilities
> >> to RHEL4) so removing the boolean preservation functionality would be
> >> detrimental. setlocaldefs isn't used very often afaik but we sometimes
> >> build systems where the use of 'managed policy' is objected to, in which
> >> case the only way to add users is via users.local. With this in mind
> >> we'll just have to be careful when upgrading the CLIP toolchain not to
> >> use a version that eventually removes this support.
> > 
> > When you say "uses an updated toolchain", do you mean that it replaces
> > the system libraries or just that it uses a private copy of the updated
> > userland for managing and generating the kernel policy file?  If the
> > former, then yes, this means that you'd have to at least set values
> > in /etc/selinux/config to enable the legacy behavior, but if the latter,
> > then it shouldn't affect you at all - init and load_policy would still
> > use the system libselinux library for loading the policy, and thus still
> > have the legacy behavior.
> 
> It replaces the system libraries. That's the only way to get certain
> functionality (such as local users on RHEL4).

Ok, well, do you object to changing the defaults as long as we provide a
setting in /etc/selinux/config to provide the legacy compatibility, e.g.
	SETLOCALDEFS=1
	PRESERVEBOOLS=1

-- 
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.

[Index of Archives]     [Selinux Refpolicy]     [Linux SGX]     [Fedora Users]     [Fedora Desktop]     [Yosemite Photos]     [Yosemite Camping]     [Yosemite Campsites]     [KDE Users]     [Gnome Users]

  Powered by Linux