Mention that it's "data offset" in 32-bit words. Helpful when doing
code review and using search for "th_off".
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
We need to make sure that when listening a connection establishment,
the connection gets cleared if we do not receive final ACK.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If we receive data that is out-of-order, queue sequential
TCP segments until we have received earlier segment or a timeout
happens.
Note that we only queue data sequentially in current version i.e.,
there should be no holes in the queue. For example, if we receive
SEQs 5,4,3,6 and are waiting SEQ 2, the data in segments 3,4,5,6 is
queued (in this order), and then given to application when we receive
SEQ 2. But if we receive SEQs 5,4,3,7 then the SEQ 7 is discarded
because the list would not be sequential as number 6 is be missing.
Fixes#30364
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
As the tcp header struct can be cast to unaligned memory, mark
it as packed and access fields using UNALIGNED_GET/PUT when
needed.
Fixes#31145
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Comment out TCP1 specific stuff when TCP2 is enabled. This means
shuffling the code around a bit so that common code is placed at
the end. Introduce also net_tcp_foreach() function to TCP2 so that
it can be used from net-shell.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
When receiving data that needs to be passed the data to application,
queue it for short time so that we do not have TCP connection lock
held. This way if the application wants to send data, there is no
possibility that the connection lock would prevent sending data.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Windows 10 sends ECN-Echo and Congestion Window Reduced (CWR) flags
together with SYN flag in the connection establishment but the code
did not ignore these flags and send just SYN back (instead of SYN|ACK).
This caused the connection establishement in application level to
fail as the application was never notified about it.
Fixes#29258
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The TCP2 was calling accept callback before actually finalizing
the connection attempt.
Fixes#29164
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
We need to have timer that closes the connection for good if
we do not get the FIN and ACK reponse from the peer.
If there is any pending data when application does close(),
send them before sending FIN.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The TCP2 stack does operations directly on the packet data which may
or may not be aligned. The unaligned access causes a fault on the
Cortex-M0+ so use the UNALIGNED_* macros instead.
Signed-off-by: Michael Hope <mlhx@google.com>
In order to implement a blocking connect, add a semaphore
and block on it in net_tcp_connect().
The semaphore is released when ESTABLISHED state is reached.
In case tcp_conn_unref() is called while waiting on the semaphore,
defer the unreference, tcp_conn_unref() will be called from
net_tcp_connect().
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to support the retransmission for the outgoing data:
1. The outgoing data packet is appended to the send_data queue
in net_tcp_queue_data().
2. tcp_send_queued_data() is called and will use tcp_send_data()
to sends queued but unsent data packet by packet
until there's an unsent data and the receiver's window isn't full.
tcp_send_queued_data() subscribes send_data_timer
that will handle retrasmissions with tcp_resend_data().
3. tcp_send_data() peeks a single chunk of data from the send_data
queue that will not exceed the maximum segment size
until the the receiver's window is full.
tcp_send_data() uses conn->seq and conn->unack_len as the sequence
number for the TCP packet.
conn->unacked_len is advanced on each send.
4. On data acknowledgment:
- acknowledged amount of data is removed from the beginning
of the send_data queue
- conn->seq is advanced by the acknowledged amount
- conn->unacked_len is decremented by the acknowledged amount
- send_data_timer is cancelled
- tcp_send_queued_data() is called to send queued but
prevoiusly unsent data
5. On timeout, tcp_resend_data() will reset conn->unack_len,
peek one packet from the beginning of the send_queue and resend,
terminating the connection on retries exceeded.
Meanwhile the outgoing data tcp_send_queued_data() is just
appended to the send_data but not sent.
In case of the acknowledgement, tcp_send_queued_data() will
start sending multiple packets until the receiver's window
is full.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to support the retransmission for the outgoind data,
add the following into the TCP connection:
- struct net_pkt *send_data (used as a queue)
- send_data_total
- unack_len
- send_data_timer
- send_data_retries
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to support the send window, add send_win into
the TCP connection.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to use struct net_pkt for the outgoing data queue,
update tcp_pkt_alloc() macros, so they support allocation
without buffer.
Additionally, merge tcp_pkt_alloc() macros macros into one.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to fix the line tracking of the TCP packet allocation
with the test protocol enabled, refactor tcp_pkt_alloc(),
so the line of the allocation can be tracked properly.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
Make sure we only parse the received TCP options only once. Store
the options to tcp conn struct for later use.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Avoid dynamic allocations and all the issues if we run out of
memory, by placing the connection endpoint directly to TCP
connection struct.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Some other part of Zephyr has similar defines (SRC)
causing build failures.
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
In order to support TIME_WAIT state during the TCP connection
termination, add a TIME_WAIT timer and the corresponding state.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to improve readability, rename TCP_FIN_WAIT
states into TCP_FIN_WAIT_1, TCP_FIN_WAIT_2.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
Use existing network stack calls to create net pkt,
ip header and udp header creation. Also simplify
finalize api calls.
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
Use existing network stack calls to create net pkt,
ip header and tcp header creation. Also simplify
finalize api calls.
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
According to RFC 793 and IANA "TCP Option Kind Numbers" an option with
kind 0 is an End of Option List option and not a PAD.
Signed-off-by: Ruslan Mstoi <ruslan.mstoi@intel.com>
The TCP connection might be concurrently modified from the
TR/TX threads, so add a mutex to protect from the concurrent
modification.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
After removing the test windows, update test protocol
functions for the TTCN-3 sanity check.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
These test receive windows were used by TTCN-3 sanity check,
but it's possible to do without them.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to drop dependency on a heap, drop the
test send window.
Use net_tcp_queue_data() to receive the outgoing data
from the socket layer.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to drop dependency on a heap, drop the test
receive window.
The receive window was needed for TTCN-3 sanity check,
but it's possible to do without it.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
k_timer callback is executed from ISR context, which isn't
currenty compatible with Zephyr's shell implementation.
This problem was found while testing TCP2 client side.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
In order to properly remove struct tcp and its data, net_context
needs to be supplied as k_timer user data. Modify net_tcp_unref()
to take a net_context argument when removing, and add ifdefs
around code that after this will only be used by TCP testing
functionality.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
Add a net_context backpointer and receiver user data to the tcp
structure. Once the desired callback and user data is set in
net_tcp_recv(), net_context_packet_received() can be called at
TCP reception in tcp_data_get(), moving the TCP data to the
recipient. IP and TCP protocol headers are sent as NULL, they
are not used by e.g. the socket code.
Signed-off-by: Patrik Flykt <patrik.flykt@intel.com>
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
Register the TCP connection when accept() and connect() is called.
With the connection registration net_context will have the necessary
callback pointer set up, whereby net_context can call the proper
function when receiving packets for the TCP connecton.
With the new TCP stack this callback is always the same function.
Signed-off-by: Patrik Flykt <patrik.flykt@intel.com>
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
