If we are not in ESTABLISHED state, then there is no need to
try to resend any pending data packets.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If the expire send timer expires, then it sends the packet.
If that happens, then we must not try to send the same packet
again if we receive ACK etc. which can cause re-sends to happen.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If the packet sending is slow then we must NOT increment the ref
count when re-sending it. This is unlikely but can happen if there
are lot of debug prints etc. extra activities that prevent the driver
to actually send the packet fast enough.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Make sure that the sent flag is only set after we have really
sent the packet and the driver has verified that.
If the net_pkt_set_sent() is called while still in tcp.c, then
depending on how fast the device is, it might happen that the
retry timer expires before the packet is actually sent. This was
seen in frdm-k64f with ethernet and various debug prints activated.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The second 'const' is misguided, indicating that the returns pointer
value itself cannot be changed, but since pointers are passed by value
anyway this is not useful and was generating warnings with XCC.
The leading 'const' indicates that the memory pointed to is constant,
which is all we needed.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This is needed in order to get information which function is
doing the ref. With inline function this was not possible.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The API name space for Bluetooth is bt_* and BT_* so it makes sense to
align the Kconfig name space with this. The additional benefit is that
this also makes the names shorter. It is also in line with what Linux
uses for Bluetooth Kconfig entries.
Some Bluetooth-related Networking Kconfig defines are renamed as well
in order to be consistent, such as NET_L2_BLUETOOTH.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This fixes the existing situation that "if application buffers data,
it's the problem of application". It's actually the problem of the
stack, as it doesn't allow application to control receive window,
and without this control, any buffer will overflow, peer packets
will be dropped, peer won't receive acks for them, and will employ
exponential backoff, the connection will crawl to a halt.
This patch adds net_context_tcp_recved() function which an
application must explicitly call when it *processes* data, to
advance receive window.
Jira: ZEP-1999
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
For IPv6 header compressed packet, the IP header offsets will
be wrong. In this case there is no need to print error when
trying to print TCP packet information.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The sequence number validator was checking the seq numbers
incorrectly. This caused some valid RST packets to be dropped
and the TCP stream to hang.
Added also a TCP test case that tests the seq validator.
Jira: ZEP-2289
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The commit 210c30805b ("net: context: Close connection fast
if TIME_WAIT support is off") was not a proper way of closing
the connection. So if Zephyr closes the connection (active close),
then send FIN and install a timer that makes sure that if the peer
FIN + ACK is lost, we close the connection properly after a timeout.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The commit 00ac0487b0 ("net: context: Remove tcp struct SYN-ACK
timer handling") removed also the passive close ACK timer.
Adding that ACK timer back so that we can close the connection
properly even if the last ACK from peer is lost.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
If the TCP data packet needs to be re-sent after the packet is lost,
then the acknowledgment number will be changed. This then means that
the TCP checksum needs to be recalculated too.
Signed-off-by: june li <junelizh@foxmail.com>
Remove NET_TCP_HDR() macro as we cannot safely access TCP header
via it if the network packet header spans over multiple net_buf
fragments.
Fixed also the TCP unit tests so that they pass correctly.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Instead of waiting forever for a free net_buf, set a timeout to
the allocations (500 ms). This way the application will not be
blocked by memory exhaustion.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Defines a new tunable, CONFIG_NET_TCP_RETRY_COUNT, that determines the
number of segment retransmissions that the IP stack will attempt to
perform before resetting the connection.
The default value is 9 retransmissions, which amounts to 1:42 minutes,
as close as possible to the minimum recommended by RFC1122.
Jira: ZEP-1956, ZEP-1957
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
Without change to add ACK to FIN, invalid TCP packet is generated,
where ack sequence number is non-zero. Without adjusting sequence
number as done, ACK which we send in response to peer's FIN/ACK is
not recognized by peer, and peer keeps retransmitting its FIN/ACK.
Jira: ZEP-2104
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
We must check if we receive RST in any of the TCP states.
If we do not do this, then the net_context might leak as it
would never be released in some of the states. Receiving RST
in any TCP state is not described in TCP state diagram but is
described in RFC 793 which says in chapter "Reset Processing"
that system "...aborts the connection and advises the user and
goes to the CLOSED state."
We need to also validate the received RST and accept only those
TCP reset packets that contain valid sequence number.
The validate_state_transitions() function is also changed to
accept CLOSED state transition from various other states.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
RFC793, "Transmission Control Protocol", defines sequence numbers
just as 32-bit numbers without a sign. It doesn't specify any adhoc
rules for comparing them, so standard modular arithmetic should be
used.
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
This is related to commit "net: tcp: Make sure ACK timer is not
run if cancelled" which did not set the cancel flag when the timer
was cancelled from tcp.c.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The TCP trace values were not printed because of incorrect
config option used. Print also seq and ack values in decimal
in order to make it easier to correlate the values in other
prints in tcp.c.
Change-Id: I44d1535a84dcba8c6c937d348516ba801193ca23
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Convert code to use u{8,16,32,64}_t and s{8,16,32,64}_t instead of C99
integer types.
Jira: ZEP-2051
Change-Id: I4ec03eb2183d59ef86ea2c20d956e5d272656837
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
If TCP debugging is enabled but if the loglevel is set to lower
than 4, then compiler prints warning about unused flags variable
in net_tcp_trace().
Change-Id: I2e663644b50fe97b75088202e21b286aa010953e
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Most of these macros are not exactly exposing a buffer, but a specific
header pointer (ipv6, ivp4, ethernet and so on), so it relevant to
rename them accordingly.
Change-Id: I66e32f7c3f2bc75994befb28d823e24299a53f5c
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
- net_pkt becomes a stand-alone structure with network packet meta
information.
- network packet data is still managed through net_buf, mostly named
'frag'.
- net_pkt memory management is done through k_mem_slab
- function got introduced or relevantly renamed to target eithe net_pkt
or net_buf fragments.
- net_buf's sent_list ends up in net_pkt now, and thus helps to save
memory when TCP is enabled.
Change-Id: Ibd5c17df4f75891dec79db723a4c9fc704eb843d
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
There have been long lasting confusion between net_buf and net_nbuf.
While the first is actually a buffer, the second one is not. It's a
network buffer descriptor. More precisely it provides meta data about a
network packet, and holds the chain of buffer fragments made of net_buf.
Thus renaming net_nbuf to net_pkt and all names around it as well
(function, Kconfig option, ..).
Though net_pkt if the new name, it still inherit its logic from net_buf.
'
This patch is the first of a serie that will separate completely net_pkt
from net_buf.
Change-Id: Iecb32d2a0d8f4647692e5328e54b5c35454194cd
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
We need to check whether buf_sent was true when resending the TCP
segment, and do a buf ref if needed. If this is not done, the buf
will be unref after send, which will cause unpredictable results.
Change-Id: Ibd4490305de88ac6ffd04ec42bba196e57da5c10
Signed-off-by: june li <junelizh@foxmail.com>
After failing to send the buf we need to release it.
This is not done for Bluetooth or IEEE 802.15.4 links which
create a copy of the sent buf and the failure case is already
checked by net_tcp_send_buf().
Change-Id: Ia556376b58ad74f68accb64eb2221a78d59dc2ec
Signed-off-by: june li <junelizh@foxmail.com>
Use UNALIGNED_PUT() to store the MSS value into network packet
because the memory location cannot be guaranteed to be properly
aligned.
Change-Id: I77fd7a70ef45eedb657cac29457b0239b0a1d4c2
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
When sending TCP data, check if the retry timer needs
to be started.
Change-Id: Iea90716e918dec0b22e60bf32467b11c0d1a296f
Signed-off-by: june li <junelizh@foxmail.com>
The default timeout (4 min) is very long. Allow tweaking the
value via Kconfig option.
Change-Id: Iddfd48b96f3612b9bba7caa4d64357505df9644d
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
No functionality changes, use separate variable so that we do
not need to do big-endian conversion multiple times.
Change-Id: I8874b427bd39dfa2d952034a2623c47544a644fc
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Print TCP header information in one line as there is really no
need to use multiple lines. Also use debug level when printing
the header info so that it is only seen if debugging is activated.
Change-Id: I27f314ca060239545769dec07148897da3426436
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Current finalize api's takes buf as input parameter and returns
the finalized buf. But if there are any issue while finalizing,
it failed to throw an error.
Change-Id: I6db54b7453eec41a8051fab50d5c0dc937debd54
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
Due to recent TCP fixes for 6lo, we are making a copy of buffers
sent to net_tcp_send_buf() so that TCP retransmit can send the original
(unmodified) buffer. This original buffer is freed via the TCP
sent_list when the related ACK packet is received.
However, there are users of the net_tcp_send_buf() function which
will never get a corresponding ACK (and do not add the buffer to the
TCP sent_list). An example is send_ack() in net_context.c. In this
case, we leak the original buffer.
To fix this leak in the 6lo specific block of net_tcp_send_buf(),
let's check to see if the original buffer was added to the TCP sent_list
and if not, then avoid the buffer copy process entirely.
Change-Id: If99e0e5bf266d33dd3466dc5d74443eaa39d10a8
Signed-off-by: Michael Scott <michael.scott@linaro.org>
The tcphdr->offset was not set when tcp options were added.
Change-Id: I19fe97983ce81948a9a84893183e5c9000f12767
Signed-off-by: june li <junelizh@foxmail.com>
There are no users of net_tcp_set_state() left outside of
subsys/net/ip/tcp.c.
And the naming of this function is confusing -- it could easily
be mistaken for net_tcp_change_state() which contains additional
logic for certain tcp states.
Let's remove it entirely and fix the remaining uses to set
tcp->state directly.
Change-Id: I92855ad180e8682780fcff11e50af06adcbc177c
Signed-off-by: Michael Scott <michael.scott@linaro.org>
In net_tcp_release() when a TCP connection is being closed, we
should call net_tcp_change_state instead of net_tcp_set_state.
net_tcp_change_state() will call into net_tcp_set_state() but
also contains logic specific to NET_TCP_CLOSED which unregisters
the context's conn_handler and sends an accept callback (if
present) with -ENETRESET error for user code to handle.
This fixes an EISCONN error returned by net_context_bind() when
a TCP-based net_context was reused. Due to the conn_handler
not being cleaned up in the TCP code.
Change-Id: I8439a028a1c7ae5fd2a50d11caa9947a0ac6c7d4
Signed-off-by: Michael Scott <michael.scott@linaro.org>
When sending TCP data using 802.15.4 or Bluetooth technologies,
the 6lo code modified the original IPv6 header. This caused
issue when acknowledgment was waited to the sent packet as
the code could never match the sequence and ack numbers in
TCP header.
This commit changes this and the packet is cloned when sending
it so the 6lo code will modify a copy of the data and will not
touch the original packet.
JIRA: ZEP-1719
Change-Id: Iae51f35d5b5ada0d2543b58a29abbf10f146777e
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
TCP maintains 'sent_list' for retransmission if it doesn't get ACK for it.
Same list is not freed on net_tcp_release() call. This causes memory leak.
Change-Id: I2b2def1ea19487cc48ea4fbb6343ef0c773f288f
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
finalize_segment() will call net_ipv4_finalize() or
net_ipv6_finalize(). Both the functions perform net_nbuf_compact().
But after finalize_segment(), net_nbuf_compact() called again, which
is unnecessary.
Change-Id: I9fab63bcc44eec87061a4b55edd5053cf6556a75
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
Protocol family is checked in prepare_segment() and in same function
it's again verified by finalize_segment(). So remove the double checking
in finalize_segment().
Change-Id: I17123ab8741d017d7e3ff1ef3fb07371b0d4aa66
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
Using net_buf_ref() technically works but debugging the network buffer
allocations is more difficult if done like that.
Change-Id: Iac81bd3ab95547741d49f32763baaa54e97b4877
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
