Commit Graph

117 Commits

Author SHA1 Message Date
liqinhui
f61dc72892 net/tcp:Add NewReno congestion control.
- NewReno congestion control algorithm is used to solve the problem
  of network congestion breakdown. NewReno congestion control includes
  slow start, collision avoidance, fast retransmission, and fast
  recovery. The implementation refers to RFC6582 and RFC5681.

- In addition, we optimize the congestion algorithm. In the conflict
  avoidance stage, the maximum congestion window max_cwnd is used to
  limit the excessive growth of cwnd and prevent network jitter
  caused by congestion. Maximum congestion window max_cwnd is updated
  with the current congestion window cwnd and the update weight is
  0.875 when an RTO timeout occurs.

Signed-off-by: liqinhui <liqinhui@xiaomi.com>
2023-05-16 12:35:01 -03:00
zhanghongyu
9162800255 tcp_input: drop SYN when no free node in the backlog
Wait for the client to retransmit the ack before trying to allocate the
backlog

Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
2023-04-24 01:52:33 +08:00
Zhe Weng
1aceb1d872 net/tcp: Fix clear condition in ofoseg input
We have a case that an http server gives out-of-ordered ACKs, and NuttX client makes `ofoseg`s with length 0, trying to rebuild / put them into `ofosegs` array, which is not intended (no available data and should be skipped). This breaks later logic and finally crashed in `tcp_ofoseg_bufsize` (`ofosegs[i].data` is `NULL`, which should never happen in normal logic).

Note:
- `iob_trimhead` won't return `NULL` when it's applying on normal IOB.
  - Keep `dev->d_iob == NULL` to avoid `iob_trimhead` changed.
- `iob_free_chain` will do nothing when applied to `NULL`.

Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
2023-04-22 19:26:22 +08:00
zhanghongyu
93c3b8f19e tcp: add TCP_MAXSEG support
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
2023-04-22 01:42:47 +08:00
zhanghongyu
bd4d7a1b76 tcp: accept conn inherits some properties of listener conn
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
2023-04-22 01:42:47 +08:00
Zhe Weng
b058f37353 net/tcp: Reply RST when we cannot receive data
According to RFC 2525, Section 2.17:
"When an application closes a connection in such a way that it can no longer read any received data, the TCP SHOULD, per section 4.2.2.13 of RFC 1122, send a RST if there is any unread received data, or if any new data is received."

When our TCP socket is closed (even when the thread has exited), the peer can keep sending data and NuttX keeps replying ACK (we've tried for ~12h). This is not a good behavior (also different from Linux), so send RST instead of ACK for data receiving in FIN_WAIT.

Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
2023-03-29 09:22:00 -03:00
Zhe Weng
d423992988 net/tcp: Regard snd_wnd update as ACKDATA
Receiving an ACK indicating TCP Window Update will not set ACKDATA flag (because tx_unacked is 0) in our TCP stack. Then this ACK won't let us send anything after receiving it, even if it updates snd_wnd. So we need to check whether we can send data immediately when our snd_wnd is updated (especially from 0), otherwise we will only send next data after timer expiry.

Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
2023-02-03 13:43:39 +02:00
梁超众
5012195bde support ipv4 ToS and ipv6 TrafficClass
Signed-off-by: 梁超众 <liangchaozhong@xiaomi.com>
2023-01-29 13:43:44 +08:00
chao an
64dd7e6376 net/tcp: add Selective-ACK support
Reference:
https://datatracker.ietf.org/doc/html/rfc2018

Iperf2 client/server test on esp32c3:

Drop(1/50):
CONFIG_NET_TCP_DEBUG_DROP_SEND=y
CONFIG_NET_TCP_DEBUG_DROP_SEND_PROBABILITY=50  // Drop probability: 1/50
CONFIG_NET_TCP_DEBUG_DROP_RECV=y
CONFIG_NET_TCP_DEBUG_DROP_RECV_PROBABILITY=50  // Drop probability: 1/50

Drop(1/50) + OFO/SACK:
CONFIG_NET_TCP_DEBUG_DROP_SEND=y
CONFIG_NET_TCP_DEBUG_DROP_SEND_PROBABILITY=50  // Drop probability: 1/50
CONFIG_NET_TCP_DEBUG_DROP_RECV=y
CONFIG_NET_TCP_DEBUG_DROP_RECV_PROBABILITY=50  // Drop probability: 1/50

CONFIG_NET_TCP_OUT_OF_ORDER=y
CONFIG_NET_TCP_SELECTIVE_ACK=y

---------------------------------------------------------
|  TCP Config            | Server | Client |            |
|-------------------------------------------------------|
|  Original              |   12   |     9  |  Mbits/sec |
|  Drop(1/50)            |  0.6   |   0.3  |  Mbits/sec |
|  Drop(1/50) + OFO/SACK |    8   |     8  |  Mbits/sec |
---------------------------------------------------------

Signed-off-by: chao an <anchao@xiaomi.com>
2023-01-18 16:24:09 +08:00
chao an
c581cc5f9b net/tcp: parse tcp options in common function
Signed-off-by: chao an <anchao@xiaomi.com>
2023-01-18 16:24:09 +08:00
chao an
d175f50f01 net/tcp: add out-of-order segment support
Signed-off-by: chao an <anchao@xiaomi.com>
2023-01-18 16:24:09 +08:00
Zhe Weng
1cf3147626 net/netdev: Avoid hardcoded guardsize when using d_iob
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
2023-01-18 14:41:07 +08:00
chao an
22348c890b net/tcp: debug feature to drop the tx/rx packet
Add 2 configurations
1. Config to drop recived packet
CONFIG_NET_TCP_DEBUG_DROP_RECV=y
CONFIG_NET_TCP_DEBUG_DROP_RECV_PROBABILITY=50 /* Default drop probability is 1/50 */

2. Config to drop sent packet
CONFIG_NET_TCP_DEBUG_DROP_SEND=y
CONFIG_NET_TCP_DEBUG_DROP_SEND_PROBABILITY=50 /* Default drop probability is 1/50 */

Iperf2 client/server test on esp32c3:

---------------------------------------------------------
|  TCP Config            | Server | Client |            |
|-------------------------------------------------------|
|  Original              |   12   |     9  |  Mbits/sec |
|  Drop(1/50)            |  0.6   |   0.3  |  Mbits/sec |
|  Drop(1/50) + OFO/SACK |    8   |     8  |  Mbits/sec |
---------------------------------------------------------

Signed-off-by: chao an <anchao@xiaomi.com>
2023-01-14 14:05:26 +08:00
chao an
62004a28a6 net/d_buf: remove d_buf reference from l3/l4
l3/l4 stack will decouple the reference of d_buf gradually, Only legacy
devices still retain d_buf support, new net devices will use d_iob

Signed-off-by: chao an <anchao@xiaomi.com>
2022-12-04 20:37:14 +08:00
chao an
a8d3286258 net: move device buffer define to common header
Signed-off-by: chao an <anchao@xiaomi.com>
2022-10-28 00:32:16 -04:00
Xiang Xiao
bdeaea3742 Remove the unnessary empty line after label
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
2022-09-30 17:54:56 +02:00
zhanghongyu
e03c2c321a tcp: reset conn->nrtx when ack received
Otherwise, when a long test triggers multiple timeout retransmissions,
the late timeout retransmissions are always delayed between 24 and 48 seconds

Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
2022-08-17 21:35:09 +03:00
zhanghongyu
ef660083c8 tcp: check option length before d_len update
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
2022-07-26 12:05:06 +03:00
chao.an
845e259ac7 net/tcp: d_appdata should remove the tcp specific option field
applicate data field should not touch data of IP layer

Signed-off-by: chao.an <anchao@xiaomi.com>
2022-06-15 20:28:10 +08:00
zhanghongyu
3f8b71924f tcp: move wd_timer from wifi driver to tcp stack
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
2022-05-28 16:29:51 +08:00
Xiang Xiao
d8b97d7ae8 net/tcp: Use the relative value for keep alive timer
unify the timer process logic as other tcp state

Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
2022-05-18 18:40:41 +03:00
Xiang Xiao
2d3ee157ce net/tcp: Use the decrease timer in TCP_TIME_WAIT/TCP_FIN_WAIT_2
unify the timer process logic as other tcp state

Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
2022-05-18 18:40:41 +03:00
Alexander Lunev
eec94132c4 net/tcp/sendfile: removed excessive overwrites of conn->sndseq
(conn->sndseq was updated in multiple places that was unreasonable and complicated).
This optimization is the same as it was done for tcp_send_unbuffered.
2022-01-22 00:43:53 +08:00
Alexander Lunev
338b122b2b net/tcp/sendfile: fixed an issue with unackseq calculation.
Wrong unackseq calculation locked conn->tx_unacked at non-zero values
even if all ACKs were received.
This issue is the same as it was with tcp_send_unbuffered.
2022-01-22 00:42:29 +08:00
Petro Karashchenko
9551de7115 net: use HTONS, NTOHS, HTONL, NTOHL macro in kernel code
Signed-off-by: Petro Karashchenko <petro.karashchenko@gmail.com>
2022-01-18 10:59:47 +01:00
Alexander Lunev
e9ab3adf23 net/tcp(unbuffered): advance sndseq by +1 because SYN and FIN occupy one sequence number (RFC 793) 2022-01-03 12:18:44 +09:00
Alexander Lunev
2b60468845 net/tcp(unbuffered): removed excessive overwrites of conn->sndseq
(conn->sndseq was updated in multiple places that was unreasonable and complicated).
2021-12-29 05:35:23 -06:00
Alexander Lunev
e68ffb9f99 net/tcp(unbuffered): fixed an issue with unackseq calculation.
Wrong unackseq calculation locked conn->tx_unacked at non-zero values
even if all ACKs were received. Thus unbuffered psock_tcp_send() never completed.
2021-12-27 20:59:48 -06:00
chao.an
c132e5bed4 net/tcp: sanity check for the listen address
Signed-off-by: chao.an <anchao@xiaomi.com>
2021-09-23 23:07:57 -07:00
YAMAMOTO Takashi
1b82f1c749 tcp_input: snd_wnd processing
* Do not accept the window in old segments.
  Implement SND.WL1/WL2 things in the RFC.

* Do not accept the window in the segment w/o ACK bit set.
  The window is an offset from the ack seq.
  (maybe it's simpler to just drop segments w/o ACK though)

* Subtract snd_wnd by the amount of the ack advancement.
2021-08-25 20:56:05 +08:00
YAMAMOTO Takashi
b815a2c3a8 tcp_input: Don't put back sndseq on an old ack 2021-08-06 21:17:25 -07:00
chao.an
aab03ef86d net/tcp: add window scale support
Reference here:
https://tools.ietf.org/html/rfc1323

Signed-off-by: chao.an <anchao@xiaomi.com>
2021-07-07 03:55:41 -05:00
chao.an
87bffc190c net/tcp: remove the invalid break during tcp option loop
Signed-off-by: chao.an <anchao@xiaomi.com>
2021-07-07 03:55:41 -05:00
YAMAMOTO Takashi
98e7c6924d tcp: always responds to keep-alive segments
* It doesn't make sense to have this conditional on our own
  SO_KEEPALIVE support. (CONFIG_NET_TCP_KEEPALIVE)
  Actually we don't have a control on the peer tcp stack,
  who decides to send us keep-alive probes.

* We should respond them for non ESTABLISHED states. eg. FIN_WAIT_2
  See also:
  https://github.com/apache/incubator-nuttx/pull/3919#issuecomment-868248576
2021-06-30 11:52:08 +09:00
YAMAMOTO Takashi
0886257eb4 tcp_input: Accept segments spanning over rcvseq 2021-06-30 06:22:14 +09:00
YAMAMOTO Takashi
022a2490d1 tcp: Change the way to advance rcvseq
* Move the code to advance rcvseq for user data from tcp_input
  to receive handlers.
  Motivation: allow partial ack.

* If we drop a segment, ignore FIN as well. Note than tcp FIN bit is
  logically after the user data in the same segment.
2021-06-30 06:22:14 +09:00
YAMAMOTO Takashi
14ec75e7fc tcp: window update improvements
* Fixes the case where the window was small but not zero.

* tcp_recvfrom: Remove tcp_ackhandler. Instead, simply schedule TX for
  a possible window update and make tcp_appsend decide.

* Replace rcv_wnd (the last advertized window size value) with
  rcv_adv. (the window edge sequence number advertized to the peer)
  rcv_wnd was complicated to deal with because its base (rcvseq) is
  also moving.

* tcp_appsend: Send a window update even if there are no other reasons
  to send an ack.
  Namely, send an update if it increases the window by
    * 2 * mss
    * or the half of the max possible window size
2021-06-13 21:20:24 -05:00
YAMAMOTO Takashi
7d82e7a7c4 tcp_input: fix a confusing variable name and a comment
It looks like a copy-and-paste mistake.
2021-06-10 22:47:04 -05:00
YAMAMOTO Takashi
eb00e00e48 tcp: Use the tcp seq macros in some obvious places 2021-06-10 22:47:04 -05:00
chao.an
794a6ec23d net/tcp: rename the winszie to snd_wnd to make the semantics more accurate
Change-Id: I8fdc7cf78a7f2cd53a30ef1de702b1a697c43238
Signed-off-by: chao.an <anchao@xiaomi.com>
2020-12-10 12:23:47 +09:00
Gregory Nutt
45699e2701 net/tcp/tcp_input.c: Correct bad check of urgent data length
Urgent data preceded "normal" data in the TCP payload.  If the urgent data is larger than the size of the TCP payload, this indicates that the entire payload is urgent data and that urgent data continues in the next packet.

This case was handled correctly for the case where urgent data was present but was not being handled correctly in the case where the urgent data was NOT present.
2020-12-09 07:25:14 -08:00
YAMAMOTO Takashi
cbdbb369c0 net/tcp/tcp_input.c: Fix syslog formats
Note: tx_unacked is either uint16_t or uint32_t depending on the config.
2020-11-27 05:18:57 -06:00
YAMAMOTO Takashi
40d4c03a8c net/tcp/tcp_input.c: Fix syslog formats 2020-11-25 05:11:26 -08:00
YAMAMOTO Takashi
ee1c914277 net/tcp/tcp_input.c: Fix syslog formats 2020-11-23 05:00:10 -08:00
Masayuki Ishikawa
c7500c5c9d net: tcp: Fix port info in warning message in tcp_input.c
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
2020-05-19 13:36:45 +08:00
Xiang Xiao
517974787f Rename clock_systime[r|spec] to clock_systime_[ticks|timespec]
follow up the new naming convention:
https://cwiki.apache.org/confluence/display/NUTTX/Naming+of+OS+Internal+Functions
2020-05-10 14:35:50 -06:00
Xiang Xiao
b7d922960f Fix nxstyle issue
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
2020-05-08 07:20:49 -06:00
GAEHWILER Reto
2ba2846ced net/tcp/tcp_input.c: Fix of faulty TCP sequence increments
Sequence numbers must not be incremented in the case of retransmitted packets
2020-03-03 09:15:18 -06:00
Gregory Nutt
98f431d8d8 net/: Run all .c files under net/ through tools/nxstyle and fix all resulting complaints. 2020-01-11 13:41:33 -03:00
Xiang Xiao
5c5c08efcd network: simplify the timeout process logic
1.Consolidate absolute to relative timeout conversion into one place(_net_timedwait)
2.Drive the wait timeout logic by net_timedwait instead of devif_timer
This patch help us remove devif_timer(period tick) to save the power in the future.

Change-Id: I534748a5d767ca6da8a7843c3c2f993ed9ea77d4
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
2020-01-11 08:24:49 -06:00