[RFC PATCH] audit: normalize NETFILTER_PKT

[Richard Guy Briggs]

On 2017-02-08 18:11, Paul Moore wrote:
> On Wed, Feb 8, 2017 at 7:32 AM, Richard Guy Briggs <rgb at redhat.com> wrote:
> > On 2017-02-07 23:02, Paul Moore wrote:
> >> On Tue, Feb 7, 2017 at 4:22 PM, Richard Guy Briggs <rgb at redhat.com> wrote:
> >> > On 2017-02-06 14:41, Paul Moore wrote:
> >> >> On Sat, Feb 4, 2017 at 8:25 AM, Steve Grubb <sgrubb at redhat.com> wrote:
> >> >> > On Friday, February 3, 2017 6:44:16 PM EST Paul Moore wrote:
> >> >> >> I'm still trying to understand what purpose this record actually
> >> >> >> serves, and what requirements may exist.  In an earlier thread
> >> >> >> somewhere Steve mentioned some broad requirements around data
> >> >> >> import/export, and I really wonder if the NETFILTER_PKT record
> >> >> >> provides anything useful here when it really isn't connecting the
> >> >> >> traffic to the sender/receiver without a lot of additional logging and
> >> >> >> post-processing smarts.  If you were interested in data import/export
> >> >> >> I think auditing the socket syscalls would provide a much more useful
> >> >> >> set of records in the audit log.
> >> >> >
> >> >> > The problem here is we cannot be selective enough through the syscall
> >> >> > interface to get exactly what we want. For example, any auditing of connect
> >> >> > and accept will also get af_unix traffic which is likely to be uid/gid lookups
> >> >> > through sssd or glibc. Typically we want the IPv4/6 traffic. The netfilter rules
> >> >> > are better suited to describing which packets are of interest.
> >> >>
> >> >> Okay, but how useful are these NETFILTER_PKT records, really?  The
> >> >> only linkage you have back to the process on the local machine is via
> >> >> the addr/proto/port tuple and that seems far from ideal.
> >> >
> >> > And even that could be spoofed easily and gathering more corroborating
> >> > information would seem useful.
> >> >
> >> > Would the presence of the SOCKADDR record in any SYSCALL record be
> >> > useful for somehow tagging a class of fd as being of interest?
> >>
> >> I don't think we want to create a SOCKADDR record for every syscall,
> >> but it seems reasonable that we may want to include it for targeted
> >> syscalls.  Right now it looks like we create a SOCKADDR record
> >> whenever we copy a sockaddr struct across the kernel/userspace
> >> boundary, that should be sufficient, yes?
> >
> > Yes, we certainly don't need it for every syscall.  Since the sockaddr
> > record is only created if it is available we could further flag or check
> > the protocol to further process only the network-based sockaddrs and
> > ignore the unix sockaddrs for this purpose.  I'm picturing adding a flag
> > to the fd, but that is making me a bit nervous about overstepping our
> > usual code area.
> 
> Let's keep it as-is, I would think there are other cases where having
> the address info for AF_UNIX (and others) might be helpful.

I wasn't suggesting removing the existing AUDIT_SOCKADDR support for
AF_UNIX or other types of sockets, but rather when they are encountered
by the audit subsystem flag the fd (if it isn't already identified as
a network socket) as having more interesting information for network
auditing.

> paul moore

- RGB

--
Richard Guy Briggs <rgb at redhat.com>
Kernel Security Engineering, Base Operating Systems, Red Hat
Remote, Ottawa, Canada
Voice: +1.647.777.2635, Internal: (81) 32635