Re: Auditing nftables changes
Thanks Richard and also thanks Paul. Yes, that was me asking in IRC. I didn't see a reply for a while (at least an hour I thought) plus no other conversations so I wasn't sure how often people looked at it. I think I got disconnected and reconnected and thus probably completely missed your reply. I wondered if there were IRC logs but wasn't sure how to look for them; it sounds like Libera doesn't log and I didn't see any mention about logging of #audit so I assume it is not. I think I'm starting to understand but I have to admit my brain tends to get hung up on trivial things until it gets to an "aha!' moment then I have a working model in my head. I haven't quite got there. For example in some contexts it seems that an "event" is equivalent to one line in the audit logs or in the output of ausearch but in other contexts it seems that event refers to a series of lines that are related as a sequence of related events with the same event id. What happens is that in some contexts in the docs when the word event is used it isn't clear to me whether it is referring one line of the audit output or the group of lines. For me it would be much clearer if the sequence of events with the same id were called, say, an event cluster and then the id would be an event cluster id. I think what confuses me is much the same confusion I would have had decades ago when learning awk. If someone had drilled into me the conceptual mechanic of surrounding the entire awk script in a "foreach line of input" loops and then the pattern drilled in as an "if this pattern matches" and the action as "then do this sequence of commands" then I would have saved some time. I see current awk books drill that in better than I recall. So for audit rules I haven't quite internalized the equivalent conceptual mechanic. I've got pieces of it but it is not 100% there. One thing I haven't got is the concept of the flow of time (say from fork/clone+exec when a new process is created to when that process might call setuid to later when that process makes a system call (like sendmsg) to later when the process is in the kernel and decodes the message structure and determines it is a NETFILTER_CFG change message. In my head I picture that flow as time-based but then I get confused about "when" and "where" the rules apply. I know the "first rule that matches" applies but I haven't grokked at what point(s) that happens (i.e. what is the implicit loop a la awk's implicit loop that I'm missing). Up until Paul clarified about the exclusion list I don't think I had fully understood the "list,action" and how that interacted with the other options. Specifically I see that a rule added to the syscall exit list then "requires" a '-S' option (requires in the sense that it is implicitly there as a default "-S all". Up until then I'd been blindly copying examples. I need to understand "when/where" each list applies in system time and kernel/user space. Anyway, I think I need to spend some time playing until that "aha!" moment comes. It's feels a lot closer thanks to both of your responses and I really apprecaite the time you've taken to read my emails and respond to them. Thanks, Bruce On Thu, Mar 9, 2023 at 2:17 PM Richard Guy Briggs wrote: > > On 2023-03-09 15:14, Paul Moore wrote: > > On Thu, Mar 9, 2023 at 11:33 AM Bruce Elrick > > wrote: > > > > > > I think I need to clarify where I'm confused ;-) > > > > > > With iptables you could write a rule that would only catch system > > > calls that were for iptables changes. That is, you didn't need to > > > capture *all* setsockopt calls (not that there would be lots of > > > *those*) but rather you could add the a2=64 to only get the > > > op=IPT_SO_SET_REPLACE ones. > > > > > > With netfilter, however, since the control interface is netlink and > > > netlink requires a message to a socket and messages are structs, there > > > is no way to have a similarly narrow audit rule as in the case of > > > iptables. > > > > > > That's the first thing I want to confirm: whether my understanding > > > above is correct? > > > > Yes, you are correct. > > > > > I'm confused because your answer implies I'm correct > > > but you didn't explicitly confirm that my interpretation of how it > > > works was correct. > > > > > > You talk about having an exclude filter on NETFILTER_CFG (or rather > > > exclude everything except NETFILTER_CFG??) but my understanding is > > > that you can only do that filtering after the fact using ausearch or > > > writing some sort of correlation code using the auparse library. > > > > The kernel implements an exclude filter which is described in the > > auditctl(8) manpage: > > > > "Add a rule to the event type exclusion filter list. > > This list is used to filter events that you do not > > want to see. For example, if you do not want to see > > any avc messages, you would using this list to > > record that. Events can be excluded by process ID, > > user ID, group ID, login user ID,
Re: Auditing nftables changes
On 2023-03-09 15:14, Paul Moore wrote:
> On Thu, Mar 9, 2023 at 11:33 AM Bruce Elrick
> wrote:
> >
> > I think I need to clarify where I'm confused ;-)
> >
> > With iptables you could write a rule that would only catch system
> > calls that were for iptables changes. That is, you didn't need to
> > capture *all* setsockopt calls (not that there would be lots of
> > *those*) but rather you could add the a2=64 to only get the
> > op=IPT_SO_SET_REPLACE ones.
> >
> > With netfilter, however, since the control interface is netlink and
> > netlink requires a message to a socket and messages are structs, there
> > is no way to have a similarly narrow audit rule as in the case of
> > iptables.
> >
> > That's the first thing I want to confirm: whether my understanding
> > above is correct?
>
> Yes, you are correct.
>
> > I'm confused because your answer implies I'm correct
> > but you didn't explicitly confirm that my interpretation of how it
> > works was correct.
> >
> > You talk about having an exclude filter on NETFILTER_CFG (or rather
> > exclude everything except NETFILTER_CFG??) but my understanding is
> > that you can only do that filtering after the fact using ausearch or
> > writing some sort of correlation code using the auparse library.
>
> The kernel implements an exclude filter which is described in the
> auditctl(8) manpage:
>
> "Add a rule to the event type exclusion filter list.
> This list is used to filter events that you do not
> want to see. For example, if you do not want to see
> any avc messages, you would using this list to
> record that. Events can be excluded by process ID,
> user ID, group ID, login user ID, message type,
> subject context, or executable name. The action is
> ignored and uses its default of "never".
>
> Taken from https://man7.org/linux/man-pages/man8/auditctl.8.html
>
> However, in my last reply I wasn't advocating for this use of the
> exclude filter, I was simply trying to explain that unless you are
> explicitly excluding the creation of NETFILTER_CFG records via the
> exclude filter you should be seeing NETFILTER_CFG in your audit stream
> with basic auditing enabled.
Bruce, it appears you asked this question on IRC Libera #audit as
"virtuous-sloth". I replied there as Paul has clarified here that all
those configuration changes for iptables and nft should be in the audit
log by default if audit is enabled, because they are considered system
configuration changes which are required by certifications to be
audited by default. This wasn't always the behaviour, and as you asked
about "GHAK124" (https://github.com/linux-audit/audit-kernel/issues/124)
has been fixed and updated. At first, they were dependant on audit
configuration and were missing some iptables events (ghak 25, 35, 43,
44), and nft wasn't monitored at all, but as of upstream v5.8 the
iptables issues are resolved, as of v5.9 nft support was added and as of
upstream v5.13 nft support was tamed a bit due to the unnecessarily
large volume of records produced in the initial nft support. If these
events are not showing up in your logs you may have an older kernel.
Have I understood your problem and does this help clarify things?
> > It just seemed surprising that there is a non-trivial loss of audit
> > functionality but that I could not find any obvious discussion about
> > that. By obvious discussion I mean as explicitly as what I'm trying to
> > say here.
>
> Unfortunately it is a fairly common practice for kernel features to be
> added, and removed, without consulting with the various Linux Kernel
> security developers, e.g. audit, SELinux, LSM, etc. Sometimes we are
> successful in retrofitting the necessary security and/or auditing
> hooks, sometimes we are limited due to design choices.
>
> > The other thing I'm trying to understand is how heavy an audit load
> > would it be to have an audit rule that captures *all* sendmsg calls
> > (well, all except where auid=-1 or auid=${serviceuser_uid}). I don't
> > have a good enough understanding of systems programming to know where
> > and how often the sendmsg is called. Of course I know this is highly
> > dependent on workload, but my knowledge is limited enough that I I can
> > convince myself both that the audit load would be not trivial but
> > still manageable in most cases but also I can convince myself that no
> > same sysadmin would consider running such an audit rule. With file IO
> > it's easy to distinguish that file opens are worth auditing but file
> > reads and writes would be insane to audit. It's not so clear for me
> > for sockets.
>
> This is going to be dependent on both the workloads and applications
> used on the system, there is no one "right" answer here.
>
> --
> paul-moore.com
>
> --
> Linux-audit mailing list
> Linux-audit@redhat.com
> https://listman.redhat.com/mailman/listinfo/linux-audit
- RGB
--
Richard Guy Briggs
Sr. S/W Engineer, Kernel Security, Base Operating Systems
Remote, Ottawa, Red H
Re: Auditing nftables changes
On Thu, Mar 9, 2023 at 11:33 AM Bruce Elrick wrote:
>
> I think I need to clarify where I'm confused ;-)
>
> With iptables you could write a rule that would only catch system
> calls that were for iptables changes. That is, you didn't need to
> capture *all* setsockopt calls (not that there would be lots of
> *those*) but rather you could add the a2=64 to only get the
> op=IPT_SO_SET_REPLACE ones.
>
> With netfilter, however, since the control interface is netlink and
> netlink requires a message to a socket and messages are structs, there
> is no way to have a similarly narrow audit rule as in the case of
> iptables.
>
> That's the first thing I want to confirm: whether my understanding
> above is correct?
Yes, you are correct.
> I'm confused because your answer implies I'm correct
> but you didn't explicitly confirm that my interpretation of how it
> works was correct.
>
> You talk about having an exclude filter on NETFILTER_CFG (or rather
> exclude everything except NETFILTER_CFG??) but my understanding is
> that you can only do that filtering after the fact using ausearch or
> writing some sort of correlation code using the auparse library.
The kernel implements an exclude filter which is described in the
auditctl(8) manpage:
"Add a rule to the event type exclusion filter list.
This list is used to filter events that you do not
want to see. For example, if you do not want to see
any avc messages, you would using this list to
record that. Events can be excluded by process ID,
user ID, group ID, login user ID, message type,
subject context, or executable name. The action is
ignored and uses its default of "never".
Taken from https://man7.org/linux/man-pages/man8/auditctl.8.html
However, in my last reply I wasn't advocating for this use of the
exclude filter, I was simply trying to explain that unless you are
explicitly excluding the creation of NETFILTER_CFG records via the
exclude filter you should be seeing NETFILTER_CFG in your audit stream
with basic auditing enabled.
> It just seemed surprising that there is a non-trivial loss of audit
> functionality but that I could not find any obvious discussion about
> that. By obvious discussion I mean as explicitly as what I'm trying to
> say here.
Unfortunately it is a fairly common practice for kernel features to be
added, and removed, without consulting with the various Linux Kernel
security developers, e.g. audit, SELinux, LSM, etc. Sometimes we are
successful in retrofitting the necessary security and/or auditing
hooks, sometimes we are limited due to design choices.
> The other thing I'm trying to understand is how heavy an audit load
> would it be to have an audit rule that captures *all* sendmsg calls
> (well, all except where auid=-1 or auid=${serviceuser_uid}). I don't
> have a good enough understanding of systems programming to know where
> and how often the sendmsg is called. Of course I know this is highly
> dependent on workload, but my knowledge is limited enough that I I can
> convince myself both that the audit load would be not trivial but
> still manageable in most cases but also I can convince myself that no
> same sysadmin would consider running such an audit rule. With file IO
> it's easy to distinguish that file opens are worth auditing but file
> reads and writes would be insane to audit. It's not so clear for me
> for sockets.
This is going to be dependent on both the workloads and applications
used on the system, there is no one "right" answer here.
--
paul-moore.com
--
Linux-audit mailing list
Linux-audit@redhat.com
https://listman.redhat.com/mailman/listinfo/linux-audit
Re: Auditing nftables changes
I think I need to clarify where I'm confused ;-)
With iptables you could write a rule that would only catch system
calls that were for iptables changes. That is, you didn't need to
capture *all* setsockopt calls (not that there would be lots of
*those*) but rather you could add the a2=64 to only get the
op=IPT_SO_SET_REPLACE ones.
With netfilter, however, since the control interface is netlink and
netlink requires a message to a socket and messages are structs, there
is no way to have a similarly narrow audit rule as in the case of
iptables.
That's the first thing I want to confirm: whether my understanding
above is correct? I'm confused because your answer implies I'm correct
but you didn't explicitly confirm that my interpretation of how it
works was correct.
You talk about having an exclude filter on NETFILTER_CFG (or rather
exclude everything except NETFILTER_CFG??) but my understanding is
that you can only do that filtering after the fact using ausearch or
writing some sort of correlation code using the auparse library. But
you are then, in this case, still capturing a haystack and, after the
fact, searching for the needle afterwards. Actually, that's a bad
analogy because ausearch easily finds the events of type=NETFILTER_CFG
very easily and then backtracks and gives you the proctitle, sockaddr
call, and sendmsg syscall associated with the type=netfilter_cfg at
which point you can look at the auid and decide what to do then.
But this is very different from what was possible with iptables where
the rule itself can filter just the iptables-related setsockopt
syscalls.
It just seemed surprising that there is a non-trivial loss of audit
functionality but that I could not find any obvious discussion about
that. By obvious discussion I mean as explicitly as what I'm trying to
say here.
The other thing I'm trying to understand is how heavy an audit load
would it be to have an audit rule that captures *all* sendmsg calls
(well, all except where auid=-1 or auid=${serviceuser_uid}). I don't
have a good enough understanding of systems programming to know where
and how often the sendmsg is called. Of course I know this is highly
dependent on workload, but my knowledge is limited enough that I I can
convince myself both that the audit load would be not trivial but
still manageable in most cases but also I can convince myself that no
same sysadmin would consider running such an audit rule. With file IO
it's easy to distinguish that file opens are worth auditing but file
reads and writes would be insane to audit. It's not so clear for me
for sockets.
Cheers...
Bruce
On Wed, Mar 8, 2023 at 8:34 PM Paul Moore wrote:
>
> On Wed, Mar 8, 2023 at 7:13 PM Bruce Elrick
> wrote:
> > Hello all,
> >
> > I'm not sure if this list is appropriate for questions so please let
> > me know and otherwise ignore if this message is not appropriate.
> >
> > I'm trying to help someone who is finally migrating from iptables to
> > nftables on the back-end and needs to therefore migrate their audit
> > capability.
> >
> > Currently they have a single simple audit rule to detect when there is
> > a iptable change from any audit user apart from their service user
> > using a rule like the accepted answer given in this[0] StackExchange
> > question, although with added filters on the auid (I have to admit I
> > don't know the origin of auid=-1 events):
> >
> > auditctl -a exit,always -F arch=b64 -F a2=64 -F auid!=-1 -F
> > auid!=${serviceuser_uid} -S setsockopt -k iptablesChange
> >
> > They are migrating from Ubuntu bionic to jammy and still using the
> > iptables front-end but since the back-end changes from default
> > iptables to default nftables they need to change their audit rules
> >
> > They did strace testing and noted the syscall changing from
> >
> > setsockopt(4, SOL_IP, IPT_SO_SET_REPLACE,
> > "filter\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...,
> > 80952) = 0
> >
> > to
> >
> > sendto(3, [{nlmsg_len=20,
> > nlmsg_type=NFNL_SUBSYS_NFTABLES<<8|NFT_MSG_GETGEN,
> > nlmsg_flags=NLM_F_REQUEST, nlmsg_seq=0, nlmsg_pid=0},
> > {nfgen_family=AF_UNSPEC, version=NFNETLINK_V0, res_id=htons(0)}], 20,
> > 0, {sa_family=AF_NETLINK, nl_pid=0, nl_groups=}, 12) = 20
> >
> > between the two versions.
> >
> > In my own testing, I decided to approach from the audit tools
> > perspective so I created a broad rule to capture all system call
> > related to a test user:
> >
> > auditctl -a always,exit -S all -F auid=1001 # 1001 is uid of testuser
> >
> > Then I tried various operations using my testuser such as
> > iptables-restore of either a default-accept rule set with no rules or
> > with one or two simple drop rules. I also tested adding just a single
> > iptables rule. I then used ausearch to discover what the audit system
> > captured:
> >
> > # ausearch -i -m NETFILTER_CFG
> > ...
> >
> > type=PROCTITLE msg=audit(03/07/2023 17:18:55.152:143044) :
> > proctitle=iptables-restore
Re: How to define audit rule for one bit *not* set for a syscall argument?
Yes, the case is that we would like to filter out the "thread creation" and only keep the "process creation", by excluding the CLONE_THREAD flag bit of a0 of clone() syscall. Without the audit comparator support for this case, we have to filter out the "thread creation" in user space which introduces a performance penalty. Regards, Richard On Thu, Mar 9, 2023 at 6:22 AM Steve Grubb wrote: > Hello, > > On Wednesday, March 8, 2023 8:46:57 AM EST Richard Du wrote: > > I'm trying to define an audit rule with auditctl for clone() syscall, > and I > > would expect that the a0 of clone() syscall (i.e. the clone_flags > > argument) without the CLONE_THREAD flag bit being set. > > > > int clone(int (*fn)(void *), void *stack, int flags, void *arg, ... > > /* pid_t *parent_tid, void *tls, pid_t *child_tid */ ); > > > > From man page of auditctl, -F option build a rule file: name, operation, > > value. > > -F [n=v | n!=v | nv | n<=v | n>=v | n&v | n&=v] > > > > I can understand that, the n&v (Audit_bitmask) means any bit of a bitmast > > is set, and the n&=v (Audit_bittest) means all bits of a bitmask are set. > > > > While my question is, how to build a rule which means "none of bit of a > > bitmask is set", i.e. ( ! n&=v ). If the current audit comparator dosen't > > support this, can we add the support in furture? > > The comparator does not support this. This is a corner case in which this > is > the first time someone ever needed it. > > -Steve > > > -- Linux-audit mailing list Linux-audit@redhat.com https://listman.redhat.com/mailman/listinfo/linux-audit
