This commit ignores unknown TLVs when parsing a message and continues
to parse the rest parts of the message.
An unknown TLV can cause the simulator to drop the whole packet, which
could lead to packet parsing issues. For example, if a Child ID
Response message is dropped, the dissector will fail to add the Short
Address and Extended Address to the Device Descriptors, causing
subsequent secure packets unable to be decrypted.
According to PEP8, constants should be written in all capital letters.
This commit capitalizes these constants:
- thread_cert.TestCase.topology
- thread_cert.TestCase.support_ncp
In Thread 1.2, it is possible to receive an IEEE 802.15.4-2015
packet. Per specification, receiver should acknowledge this packet
with an IEEE 802.15.4-2015 ACK(Enh-ACK). This Enh-ACK can include
header IE with it and requires security enabled bit in FCF be set to
true. It is impractical for the host to generate the Enh-ACK and send
to RCP for transmission within AIFS time(192us). So RCP should prepare
the Enh-ACK by itself, which requires it to fill in the frame counter
and do the encryption/authentication.
This commit tries to address the need of filling auxiliary frame
counter in RCP by including the following changes:
- Move frame counter related functions from MAC layer to SubMac layer,
which is mirrored in RCP.
- Set mMacFrameCounter to RCP using newly added spinel properties.
- RCP reports last used frame counter to host in TxDone and RxDone.
- Add a simulation test for reset verification.
This commit fixes the problem that `whitelist` is not enabled in
commissioning test cases which results in false pass of `Cert_8_2_01`
and `Cert_8_2_02`.
Without this commit, `Cert_8_2_01` and `Cert_8_2_02` will fail if we
turn on `whitelist`, because we now use random extaddress for
discovery requests.
This commit introduces a few enhancements:
- In NCP architecture forward the IPv6 fragments instead of the full
reassembled IPv6 packet (which in the default configuration can't be
forwarder due to SPINEL limitations). Without this fix the
fragmentation does not work if Thread device wants to send more than
MTU data through the Border Router (to the IPv6 address of NCP).
- Use OPENTHREAD_CONFIG_IP6_MAX_ASSEMBLED_DATAGRAM only for the
receiving path. For transmitting path, relay on the message buffers
pool. This is especially important for some of the use-cases we
have, and also reduce a need for regenerating OpenThread libraries.
- In case receiving IPv6 packet is bigger than supported
OPENTHREAD_CONFIG_IP6_MAX_ASSEMBLED_DATAGRAM, exit before trying to
allocate the next message buffer.
This commit fixes key_id_mode handling and added exception handling
when key_id_mode is 0.
According to IEEE Std 802.15.4-2015, when Key Id Mode is 0x00:
Key is determined implicitly from the originator and recipient(s)
of the frame, as indicated in the frame header.
Our test scripts cannot handle such packets. If we leave key_id_mode =
0 case unhandled, we would get some exceptions that may cause
developers to think that something is wrong. However dropping packets
in this case is a normal behavior.
This commit includes
[Y] add `OPENTHREAD_CONFIG_DUA_ENABLE` option for DUA feature
[Y] generate DUA via SLAAC
[Y] update test script to verify communication via DUA
- Update all test cases to inherit thread_cert.TestCase and update
topologies accordingly. The topologies should all be correct because
they are automatically generated, not hand written.
- Use default Thread version: 1.1
- Cleanup node tmp files in tmp directory according to the current
PORT_OFFSET in setUp. This is required by
test_reed_address_solicit_rejected.py and test_coap_observe.py since
they define multiple test functions in one TestCase.
- Removed call to set_extaddr64 because it's not necessary and causing
failures.
Currently, a REED which expects to become Router soon does not send
any SVR_DATA.ntf. This commit handles rejecting ADDR_SOL.rsp to allow
REED to send SVR_DATA.ntf after it is rejected.
This commit includes:
- cache domain prefix in the Thread Network
- subscribe/unsubscribe AllDomainBBRs Multicast address
- update test to cover AllNetworkBBRs and AllDomainBBRs subscription
- Introduce BACKBONE_ROUTER option for Backbone Router function.
- Implement Backbone Router service registration.
- Add basic Backbone Router service.
- Primary Backbone Router restores its Dataset when reattached after
short reset, increases sequence number and re-register to Leader.
- Add configurable jitter for Backbone Router service registration.
- Add Backbone Router service test.
This commit initializes _initialized of Node before doing
initialization. Otherwise, we will have an AttributeError in
desctructor when initialization fails.
This commit adds enhanced frame pending feature of Thread 1.2. This
requires the platform radio driver to correctly set frame pending bit
for Data frames and MAC Data Request frames.
* [coap] Add option scanning functions
In a few places in the OpenThread code-base, the following construct is
seen:
```
otCoapOptionIterator iterator;
otCoapOptionIteratorInit(&iterator, message);
for (const otCoapOption *opt = otCoapOptionIteratorGetFirstOption(&iterator);
opt != NULL;
opt = otCoapOptionIteratorGetNextOption(&iterator))
{
if (opt->mNumber == OT_COAP_OPTION_DESIREDOPTION)
{
doSomethingWith(opt);
break;
}
}
```
or its equivalent. In cases where multiple options are sought, the
above is still the better approach (using a `switch` statement), but
otherwise this seems repetitive.
`otCoapOptionIteratorGetFirstOptionMatching` basically implements the
above loop, returning the first option that matches. This same code can
also be used for testing for presence too by comparing the resultant
pointer to the `NULL` singleton.
For completeness in cases where a software developer may want to
*resume* looking for further matching options,
`otCoapOptionIteratorGetNextOptionMatching` is also provided.
* [coap] Add function for decoding Uint options
To properly implement features such as RFC7641 and RFC7959, one must be
able to decode unsigned integer options. It seems silly to have an
encoder but not a decoder, so here is the missing piece.
Future considerations may be a `uint32_t` version since the majority of
use cases do not call for a `uint64_t`; the type was chosen since it was
the most general case.
* [coap] Reference the RFC for the Observe option.
For the sake of completeness, reference the relevant RFC in the
documentation so that developers can read up more on how it should work.
* [coap] Add mObserve flag to metadata.
When we're implementing observations, a few exceptions to the usual CoAP
conventions apply, notably:
- when we receive the initial reply, the request is _not_ finished, we
will receive subsequent requests until we tell the server to stop.
- when we send a confirmable notification, the observing client will
reply with an `ACK` with no further traffic: we need to differentiate
this from an empty `ACK` to a request meaning we should expect a reply
later.
While we could just repeatedly scan for the `Observe` option on the
request, that is wasteful in terms of processing time. This allows us
to scan the request just once, and set a single-bit flag to save
processing later.
* [coap] Don't expect traffic after ACK to notification
If the client sends an empty `ACK` in reply to a confirmable RFC7641
notification, we will _NOT_ see further replies with additional
information. Consider the notification as acknowledged and pass the
details back to the application.
* [coap] Pass through notifications to handler
Ensure that all notifications sent by the CoAP server are passed back to
the application's CoAP request handler. When we receive the first one,
the request continues -- it does not finish until the server drops the
`Observe` flag or we tell the server to stop.
* [coap] Disable expiry timer once subscribed
If a request carries `Observe=0` and we receive the first reply, ignore
future time-out events as the transaction will finish only when
explicitly terminated.
* [coap] Support eager subscription cancellation
A CoAP client may signal its intention to unsubscribe from a resource by
sending a `GET` request with a token matching that of the original
request and setting the `Observe` option to the value 1.
In the event the application does this, we should pick this up and
cancel the pending transaction locally. The handler will be notified of
this by a final call where `aError=OT_ERROR_NONE` and no message given.
* [cli] Add RFC7641 support to CLI example
As an aide to those developing code that uses RFC7641, implement an
example client and server that supports this feature.
This commit introduces the first Thread 1.2 feature - enhanced keep
alive.
This commit also introduces the build config
OPENTHREAD_CONFIG_THREAD_VERSION to include/exclude Thread 1.2 code.
Always output command result, i.e. "Done" on success, error message
otherwise.
Factory commands also follow the same pattern. Their existing error
messages are not removed for backward compatibility consideration.
Simon Lin