Commit Graph

6 Commits

Author SHA1 Message Date
6f47602582 pkt-loop-filter: Add optional debugging of dropped packets
Add a debug mode to pkt-loop-filter that outputs debug messages for every
dropped packet (with the reason it was dropped). Also add a small script to
read the kernel trace pipe, after making sure tracing is active (otherwise
there will be no output in the pipe).

The source MAC address+VLAN is squeezed into a single u64 when printing as
a quick workaround to the lack of MAC address printing in BPF printk.

Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
2022-06-29 16:58:34 +02:00
96a5c2ebbf pkt-loop-filter: Set map size from userspace instead of in BPF definition
With this we can add a command line parameter to dynamically size the map
later.

Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
2022-06-14 17:43:39 +02:00
60ed7a8e8b pkt-loop-filter: Filter multicast traffic
This adds filtering of multicast traffic to the set of interfaces. The
filtering works by marking one of the interfaces as "primary" (which is
just the first interface name that is supplied on the command line) and
filtering everything with an all-ones destination MAC address if it's
coming in on any interface that's not the primary one.

To handle interfaces going down, we actually supply all the ifindexes to
the BPF program, and also install a tracing hook that listens to ifdown
events and switches the logic to the next ifindex in the sequence if the
primary one goes down. This is a bit rudimentary but should at least
provide basic filtering.

Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
2022-06-14 17:43:35 +02:00
50f3609658 pkt-loop-filter: Switch to using BPF skeletons
This makes it easier to populate the global variables we'll need for
handling multicast, and also means we don't have to worry about keeping the
BPF object file around (since it'll be statically linked).

Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
2022-06-14 17:43:35 +02:00
c57a5bc169 pkt-loop-filter: Attach a tracing program to catch ifdown events
This is needed to be able to react to interfaces going down so we can
allow multicast on a secondary interface if the primary goes down. We don't
actually react to the event yet, just print it; handling this will be added
in a subsequent commit.

Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
2022-06-14 17:43:31 +02:00
070715cf1c Add pkt-loop-filter example
Add an example to filter looping packets on (for instance) a bond
interface, by recording the egress MAC+VLAN and dropping any packets that
come in on other (related) interfaces with the same MAC+VLAN.

Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
2022-05-20 14:53:14 +02:00