Deriving the Mesh Local Prefix from the Extended PAN ID was carried
forward from legacy implementations. Removing this coupling to conform
to the existing Thread Specification.
This commit updates the auto-start behavior in `Srp::Client` which
monitors the Thread Network Data entries to discover and select SRP
sever to register with. There are three types of entries in Network
Data and are selected in the order below:
1) Preferred unicast entries with address in service data,
2) Anycast entries (each having a seq number,
3) Unicast entries with address info included in server data.
This commit also defines `AutoStart` class which includes all the
state and data variables related to auto-start feature. It also
updates `test_srp_auto_start_mode` test to cover the newly added
behaviors.
This commit updates `test_srp_name_conflicts.py` to cover the
situation where different clients try to register services with the
same service instance label but under different services.
This commit updates `Cert_9_2_17_Orphan` packet verification to check
that the two leaders are sending MLE advertisement before checking
for child attach (Child ID response). The current test checked this
after the child attach but leader 1 is turned off 5 seconds after
the child is attached which may not be enough time to ensure rx
of an advertisement (which every 30 seconds on average).
The test failed when verifying that HOST sends a ping packet to BR
while the destination is a link local address. That requires adding a
route on HOST before the ping. In this specific step, HOST added a
route to link local prefix going to BR. However, HOST failed to send
the echo request to BR using this newly added route. Since the impact
of disabling this step is small, I'll skip this step as for now and
look for a long term fix in the future.
Current `Routing Manager` implementation will try to send the failed
RS messages in 60 seconds. RS failures happen often when the infra
interface is re-enabled which is the case of rebooting, and it could
not be acceptable for the border router to take 1 minute to recover
the IPv6 connectivity after reboot.
To resolve this issue, this commit changes the delay to
`kRtrSolicitationInterval` (4 seconds) which is the interval between
two successful RS messages.
I just found that we don't have a test case covers the scenario of
DNS-SD when there are multiple BRs. Specially, one of the BRs starts
with no mDNS cache.
There was a time that our mDNS publisher answered the queries
correctly, even when there was a bug in mDNS publisher that it
cancelled resolving/browsing too early (was fixed in
https://github.com/openthread/ot-br-posix/pull/1120). That bug was
hidden because mDNS cache already contained the needed information.
This commit implements new logging model in OpenThread. Each core
module can specify its own module name using `RegisterLogModule()`.
The registered log module name is then included in the all the log
messages emitted from the specific file. This model replaces and
enhances the log region model.
This commit adds tests for NAT64 prefix
advertisement. BORDER_ROUTING_NAT64 is set to 1 to enable the feature
in tests.
It also adjusted some util functions for prefixes and routes in
netdata.
Testing outbound connectivity to IPv4 hosts is not covered yet. It
will be added after we update OTBR.
- Adds platform API `trelDnssdInitialize` to initialize TREL
DNS-SD module.
- Adds `test_trel_connectivity.py` test.
- Fixed `Border Router` TREL test to really use TREL.
MATN test cases often takes long time. We'd better move them to a
separate job so that we don't have to rerun MATN cases every time a
border router test fails.
This commit caches Border Router test results so that success BR tests
will be skipped when re-running the workflow.
The cache keys are chosen in a way that:
- Each new PR push will trigger cache miss and run the tests
- Each ot-br-posix change will trigger cache miss and run the tests
This commit introduces a number of enhancements to `thread-cert`
Border Router tests to improve reliability:
- Checks `ot-rcp` process status for OTBR tests to help catch errors
earlier
- Make sure `ot-rcp` processes are terminated after one test to avoid
corrupting subsequent tests
- Upload core dump as artifacts if otbr-agent crashed
This commit implements OTBR firewall. This implementation focuses on
ingress filtering. We may also introduce egress filtering when
necessary.
For security purpose, there are some packet forwarding rules to
follow, which were originally introduced in the spec.
- Inbound packets initiated with On-Link addresses source (OMR and
mesh local prefix based addresses) should be blocked.
- Inbound unicast packets whose destination address is not OMR address
or DUA should be blocked.
- Inbound unicast packets whose source address or destination address
is link-local should be blocked. Note that we don’t need to
explicitly add rules for link-local addresses since this should
already be handled by the kernel.
These rules can be easily implemented by iptables and ipset.
Before otbr-agent starts, there is a script creating the iptables
rules. The rules themselves are constant so we don't need to change
them dynamically. During the runtime of otbr-agent, otbr-agent updates
ipsets accordingly whenever there's a change of on-link prefixes.
This commit implements new radio filter mechanism in OpenThread. The
radio filter is mainly intended for testing. It can be used to
temporarily block all tx/rx on the IEEE 802.15.4 radio. When radio
filter is enabled, radio is put to sleep instead of receive (to
ensure that the device does not receive any frame and/or potentially
send ack). Also the frame transmission requests return immediately
without sending the frame over the air (return "no ack" error if ack
is requested, otherwise return success). This commit also add CLI
command `radiofilter` for this feature. It also adds test-case
`test_radio_filter` to validate the behavior of the new feature.
This feature requires `OPENTHREAD_CONFIG_MAC_FILTER_ENABLE`.
This is a workaround that disables the scenario of verifying the
meshcop service when thread is disabled. This scenario sometimes
doesn't work well with mDNSResponder publisher because of the renaming
logic. We are expecting a refactor of advertising proxy & mdns and
this issue should be fixed at that time.
The spec asks border router to publish meshcop service even when
thread interface is inactive. This commit adds test case to verify the
feature.
This commit also introduces boot_delay to add delays between booting
nodes. In this way we can let border routers not starting at the same
time, so that there will be fewer renaming conflicts. This can make
the test more stable.
This commit updates `Dns::ServiceDiscovery::Server` to use
case-insensitive string match when comparing DNS names. It also
updates `test_dnssd.py` to use mixed case names when browsing for or
resolving services. This validates the case-insensitive treatment
of names by SRP client and server and DNS-SD server (resolver) and
DNS client.
HostNode is based on OtbrDocker, which has otbr-agent service started
by default. We should disable this service on HostNode otherwise the
border router test cases won't have the exact expected topology.
This commit fixes a bug that PBBR does not send Address Query and
Backbone Query on behalf of MED Children, and add a test to make sure
DUA forwarding works on PBBR:
- Make sure PBBR sends Address Query and Backbone Query on behalf of
MED Child
- Make sure PBBR does not send Address Query when the destination is
MED Child
- Make sure MED Children do not send Address Query by itself
- Verifiy DUA reachability between PBBR and MED Child
- Make sure PBBR reduce IPv6 TTL by 1 when forwarding DUA destined
messages
Added new API/CLI for reference devices to set allowed ranges of
router IDs. When the allowed ID range is set, the leader will only
assign router IDs in the range. In this way, we can assign
non-overlapping router ID ranges to different thread networks so that
no RLOC16s will collide across multiple thread networks.
- double wait time for advertising proxy top update and multicast the
SRP updates
- adjust waiting time to avoid collision with lease time schedule
- fully remove the service before starting next test case
- add debug logging
Jonathan Hui