The folllowing functionsa are not being used anywhere in the code and
have been removed:
mesh_net_flush()
mesh_net_prov_caps_get()
mesh_net_priv_key_get()
mesh_net_priv_key_set()
mesh_net_prov_rand()
mesh_net_prov_uni()
mesh_net_id_uuid_set()
mesh_net_test_addr()
mesh_net_test_mode()
mesh_net_uni_range_set()
mesh_net_uni_range_get()
mesh_net_set_window_accuracy()
This removes callback for sending every single access layer packet
since the callback does nothing but print debug satement. This
functionality is mature enough not to require such heavy debugging.
When daemon receives beacon with IV=n+1, IVU=False it will
start sending messages with new IV and set sequence to 0.
However if daemon receives another beacon with IV=n+1, IVU=True it
will go back to sending messages with old IV=n (IVU set to True).
Because sequence number has been reset those messages will be dropped
by replay protection and node will lose communication.
Once IV is updated daemon should not go back to using the old value.
This patch adds beacon rejection if IV has already been updated.
4.4.1.2.9 of Mesh Profile Bluetooth specification:
When an element receives a Config NetKey Delete message that
identifies a NetKey that is not in the NetKey List, it
responds with Success, because the result of deleting the
key that does not exist in the NetKey List will be the same
as if the key was deleted from the NetKey List.
For added reliability, it is legal to send short messages as "single
segment" segmented messages, which require transport layer
acknowledgement. This feature is intended for heavy usage in the future
so I am adding it now.
Further, to support this functionality, an additional queue has been
added to allow multiple SAR messages to the same DST to be queued and
sent serially.
This removes unused ttl parameter from mesh_model_rx(). The TTL value
is not processed at the access layer, so there is no need to pass it
to a model.
Also, remove "uint32_t dst" parameter from the typedef of
mesh_model_recv_cb: providing a just destination address is sufficient
for internally implemented models (currently, it's only Config Server)
Future versions of Mesh will introduce new advertising packets, which
do not fit in the limited and rigid filtering currently used. This minor
rewrite allows registering and receiving of *any* AD types, including
the filtering on multiple octets of the incoming AD parts.
Re-arranged for efficiency. Replay Protection was set up as an atomic
check-and-add operation. Now we check the message early so we can
discard it without taking further action, and only add it to the RPL
once fully verified that it was authorized and addressed to us.
Mesh specification requires that Replay Protection be preserved
across node restarts. This adds that storage in
<node_uuid>/rpl/<iv_index>/<src>
Realtime access remains in an l_queue structure, and stored as
messages are processed.
Scrub of Sequence Number handling of OB messages to account for in-node
delivery of segmented messages, so that each discrete message has a
unique sequence number for the RPL.
The specification calls for a flatter Replay Protection List that
applies to all processed messages, regardless of which credentials
were used to secure them. So storage and checking is now centralized
in mesh/net.c
Any packet that may be handled internally by the daemon must be sent in
it's own idle_oneshot context, to prevent multiple nodes from handling
and responding in the same context, eventually corrupting memory.
This addresses the following crash:
Program terminated with signal SIGSEGV, Segmentation fault.
0 tcache_get (tc_idx=0) at malloc.c:2951
2951 tcache->entries[tc_idx] = e->next;
(gdb) bt
0 tcache_get (tc_idx=0) at malloc.c:2951
1 __GI___libc_malloc (bytes=bytes@entry=16) at malloc.c:3058
2 0x0000564cff9bc1de in l_malloc (size=size@entry=16) at ell/util.c:62
3 0x0000564cff9bd46b in l_queue_push_tail (queue=0x564d000c9710, data=data@entry=0x564d000d0d60) at ell/queue.c:136
4 0x0000564cff9beabd in idle_add (callback=callback@entry=0x564cff9be4e0 <oneshot_callback>, user_data=user_data@entry=0x564d000d4700,
flags=flags@entry=268435456, destroy=destroy@entry=0x564cff9be4c0 <idle_destroy>) at ell/main.c:292
5 0x0000564cff9be5f7 in l_idle_oneshot (callback=callback@entry=0x564cff998bc0 <tx_worker>, user_data=user_data@entry=0x564d000d83f0,
destroy=destroy@entry=0x0) at ell/idle.c:144
6 0x0000564cff998326 in send_tx (io=<optimized out>, info=0x7ffd035503f4, data=<optimized out>, len=<optimized out>)
at mesh/mesh-io-generic.c:637
7 0x0000564cff99675a in send_network_beacon (key=0x564d000cfee0) at mesh/net-keys.c:355
8 snb_timeout (timeout=0x564d000dd730, user_data=0x564d000cfee0) at mesh/net-keys.c:364
9 0x0000564cff9bdca2 in timeout_callback (fd=<optimized out>, events=<optimized out>, user_data=0x564d000dd730) at ell/timeout.c:81
10 timeout_callback (fd=<optimized out>, events=<optimized out>, user_data=0x564d000dd730) at ell/timeout.c:70
11 0x0000564cff9bedcd in l_main_iterate (timeout=<optimized out>) at ell/main.c:473
12 0x0000564cff9bee7c in l_main_run () at ell/main.c:520
13 l_main_run () at ell/main.c:502
14 0x0000564cff9bf08c in l_main_run_with_signal (callback=<optimized out>, user_data=0x0) at ell/main.c:642
15 0x0000564cff994b64 in main (argc=<optimized out>, argv=0x7ffd03550668) at mesh/main.c:268
When responding with NetKey List Status, packed NetKey indices into
3 octets per pair. If number of NetKeys is odd, append the last key
index as a 2-octet value.
First valid SNB received from the network should cause the node to
switch into IV_UPD_NORMAL state.
Otherwise, it will never try to enter IV Update procedure when sequence
number approaches the IV_UPDATE_SEQ_TRIGGER, because that's only allowed
in IV_UPD_NORMAL.
This patch fixes saving IV received in SNB to storage.
Previously after creating new node with IV 0 (or loading node with
IV > 0 but after long inactivity) first received SNB should update IV
(and reset sequence number to 0 if necessary).
The bug would prevent new IV being saved in storage which resulted in
sequence number being set to 0 on first SNB received after every daemon
reset but IV never being updated.
The daemon handles multiple nodes, that may or may not be on the same
mesh network. While each node my be seperately configured to beacon or
not beacon, there is nothing gained (except redundent traffic) for each
node to beacon seperately. Beaconing is therefore centralized with the
Network Key the SNB represents, with each *received* beacon delivered
to each node. But for SNBs generated, we keep a count of how many nodes
want beacons sent for a specific key. If 1 or more, we beacon, if 0
nodes want the beacon sent, then we do not beacon.
This correctly initializes net settings related to node features
based on node configuration: either defaults in the case of
a newly node created/provisioned/imported node or the configured
values read from stored existing node.
In certain circumstances, reception of PB-ADV messages may trigger
session close. This rearrange ACKs new verified messages before
performing call-back to avoid situations where the session no longer
exists after the message has been handled.
This caused static analysis errors during provisioning, which are now
addressed.
Friendship support re-written such that it can now support multiple
nodes (on multiple mesh networks) as friends to remote Low Power Nodes
(LPNs). Validated to properly respond to Friend Requests when enabled,
and a hard coded Friend Queue size of 32 (FRND_CACHE_MAX).
This is a major rewrite of Secure Network Beacon (SNB) handling
that includes:
* Seperating Key Refresh from IV_Index handling
This is a clearer handling of the two features. Although both features
are represented in SNB's, they run independantly.
* Creating a Seperate IV_Index initialization and updating state
distinct from the current values sent and received in SNBs.
If a restart occured during an IV Update procedure (96 hours long)
the IVU bit got lost, and Sequence number resetting was not done
correctly.
* Assuring that all Nodes handled by daemon receive each incoming
beacon. SNB handling previously stopped after the first node
successfuly handled it, although the SNB may be valid for many local
nodes.
This confines sequence overcommit logic in mesh-config-json, as other
storages might use a different mechanism to ensure reliability.
Also, refactored logic to calculate overcommit value to avoid division
by zero when messages are sent too fast.
When a model receives a message, it is required by the spec
to respond using the same credentials. When an App Key is used,
this is trivial because App keys are bound to Net keys, so only
the App Index is required. Messages received on a Device key
however, need the Net Index preserved from original message for
re-use during the response.
This eliminates storage_set...() routines as an intermediate
layer between mesh_config layer and the rest of the daemon when
storing updated node configuration values.
For the JSON based implementation, each call to mesh_config_write...()
routines results in writing to the node configuration file.
Since IV Index is used in application nonces, we need to honor IVI flag
not only in network layer crypto, but also in application layer.
This means that if IVI field of incoming packet is different than in
current IV Index, try to decode *both* net and app layers using IV Index
decreased by one.
This patch cleans up redundant checks in add_key() and mesh_net_set_key():
no need to check the result of l_queue_push_tail() and no need to check
if subnet is valid after it was successfully created.
This separates mesh_db_net_key_add() into distinct functions:
mesh_db_net_key_add() and mesh_db_net_key_update() which will be called
based on whether a network key was newly added or updated.
This adds implementation for saving the key refresh phase to
a node configuration file in JSON format. When the key refresh
procedure is finished, the old network keys are remove from the
configuration file.
