If we only enable one of `CONFIG_NET_ARP_SEND` and
`CONFIG_NET_ICMPv6_NEIGHBOR`, both IPv4 and IPv6 traffic will send
ARP or NDP, which causes problem.
Example:
`CONFIG_NET_ARP_SEND=n`
`CONFIG_NET_ICMPv6_NEIGHBOR=y`
Wrong:
IPv4 traffic (`PF_INET`) goes into `icmpv6_neighbor`, which
definitely causes problem.
Correct:
IPv4 traffic doesn't call anything, IPv6 traffic calls `icmpv6_neighbor`
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
When CONFIG_NET_TCP_WRITE_BUFFERS is enabled, iobs are used for
both queuing data from application, and for assembling packets
for sending. If there is a system-wide shortage of iobs, it could
happen that there is not enough free space to form any packets
to send. The buffers allocated for TCP data also can't be released
until the packet is sent.
Normally this should be avoided by setting suitable values for
CONFIG_IOB_NBUFFERS and CONFIG_IOB_THROTTLE. The default values
are ok for light usage, but can run out when using multiple
simultaneous TCP streams.
Before this commit, iob shortage would cause TCP connections to
get stuck and eventually timeout. With this change, TCP stack
sends smaller packets, eventually freeing some buffers from the
write queue.
with nonblocking sockets, POLLOUT is returned when the application
calls poll even if the send buffer is full
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
Non-blocking sockets are almost never in the connected state when app try to
get result of connect success through poll, to avoid confusion because of
this error print, so downgrade the print level to warning.
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
The error log is as follows:
tcp/tcp_send_buffered.c:376:57: runtime error: member access within misaligned
address 0x10075942 for type 'struct tcp_sack_s', which requires 4 byte alignment
Signed-off-by: liqinhui <liqinhui@xiaomi.com>
The netdev of link-local address cannot be auto decided, and the link-local address should always be reguarded as address on local network.
The problem we met:
When using `icmpv6_autoconfig` with multiple netdev, the `icmpv6_neighbor` may take out wrong netdev with ip address already set, then it may send solicitation with wrong address (`dev->d_ipv6draddr`) on wrong device, and regard the link-local address as conflict (because `dev->d_ipv6draddr` exists on this network).
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
When a task needs to send data, a callback is allocated and the
transmission is happening in a worker task through devif_send.
Synchronization between the two tasks (sender & worker) is
achieved by a semaphore.
If devif_send fails, this semaphore was never posted, leaving
the sending task blocked indefinitely. This commit fixes this
by checking the return code of netif_send, and posting this
semaphore in case of failure.
Polling then stops, and execution is resumed on the sending
task.
- 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>
This reverts commit fbe641a916.
This issue already fixed by below change:
| commit 715785245c
| Author: chao an <anchao@xiaomi.com>
| Date: Mon Jan 16 12:37:44 2023 +0800
|
| net/tcp: fix potential busy loop in tcp_send_buffered.c
|
| if the wrbuffer does not have enough space to send the rest of
| the data, then the send function will loop infinitely in nonblock
| mode, add send timeout check to avoid this issue.
|
| Signed-off-by: chao an <anchao@xiaomi.com>
Signed-off-by: chao an <anchao@xiaomi.com>
psock_tcp_send will enter busyloop state when no IOB keeps at unavailable state
when the socket is blocking socket and this will cause WDT.
According to socket's manpage, -1 should be returned while errno set to EAGAIN
if send/recv timeout was set.
Here's the description:
SO_RCVTIMEO and SO_SNDTIMEO
Specify the receiving or sending timeouts until reporting an error. The argument is a struct timeval. If an input or output function blocks for this period of time, and data has been sent or received, the return value of that function will be the amount of data transferred; if no data has been transferred and the timeout has been reached then -1 is returned with errno set to EAGAIN or EWOULDBLOCK, or EINPROGRESS (for connect(2)) just as if the socket was specified to be nonblocking. If the timeout is set to zero (the default) then the operation will never timeout. Timeouts only have effect for system calls that perform socket I/O (e.g., read(2), recvmsg(2), send(2), sendmsg(2)); timeouts have no effect for select(2), poll(2), epoll_wait(2), and so on.
Signed-off-by: liangchaozhong <liangchaozhong@xiaomi.com>
if the wrbuffer does not have enough space to send the rest of
the data, then the send function will loop infinitely in nonblock
mode, add send timeout check to avoid this issue.
Signed-off-by: chao an <anchao@xiaomi.com>
fix build break if enable CONFIG_NET_IPv6 only
In file included from tcp/tcp_sendfile.c:38:
tcp/tcp_sendfile.c: In function ‘sendfile_eventhandler’:
tcp/tcp_sendfile.c:173:27: error: ‘struct tcp_conn_s’ has no member named ‘domain’
173 | DEBUGASSERT(conn->domain == PF_INET6);
| ^~
Signed-off-by: chao an <anchao@xiaomi.com>
Fixed by commit:
| net: remove pvconn reference from all devif callback
|
| Do not use 'pvconn' argument to get the connection pointer since
| pvconn is normally NULL for some events like NETDEV_DOWN.
| Instead, the connection pointer can be reliably obtained from the
| corresponding private pointer.
Signed-off-by: chao an <anchao@xiaomi.com>
since it is impossible to track producer and consumer
correctly if TCP/IP stack pass IOB directly to netdev
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
fix regression of invalid update the rexmit_seq in buffer mode
rexmit_seq should not be used instead of sndseq in fast retransmission,
sndseq of retransmission in the packet does not need to be re-updated
Signed-off-by: chao.an <anchao@xiaomi.com>
Retransmit only one the earliest not acknowledged segment
(according to RFC 6298 (5.4)). The issue is the same as it was
in tcp_send_unbuffered.c and tcp_sendfile.c.
Signed-off-by: chao.an <anchao@xiaomi.com>
According to RFC 5681 (3.2) the TCP Fast Retransmit algorithm should start
if the threshold of 3 duplicate ACKs is reached.
Thus the threshold should be a constant, not an integer option.
Do not use pvconn argument to get the TCP connection pointer because pvconn is
normally NULL for some events like NETDEV_DOWN. Instead, the TCP connection pointer
can be reliably obtained from the corresponding TCP socket.
Fix a wrong assertion in:
```
commit 98ec46d726
Author: YAMAMOTO Takashi <yamamoto@midokura.com>
Date: Tue Jul 20 09:10:43 2021 +0900
tcp_send_buffered.c: fix iob allocation deadlock
Ensure to put the wrb back onto the write_q when blocking
on iob allocation. Otherwise, it can deadlock with other
threads doing the same thing.
```
I forget to submit this with https://github.com/apache/incubator-nuttx/pull/4257
With an applictation using mbedtls, I observed retransmitted segments
with corrupted user data, detected by the peer tls during mac processing.
Looking at the packet dump, I suspect that a wrb which has been put back
onto the write_q for retransmission was partially sent but fully acked.
Note: it's normal for a retransmission to be acked before sent.
In that case, the bug fixed in this commit would cause the wrb have
a wrong sequence number, possibly the same as the next wrb. It matches
what I saw in the packet dump. That is, the broken segments contain the
payload identical to one of the previous segment.
Consider a bi-directional TCP connection:
1. we use all IOBs for tx queue
2. we advertize zero recv window because we have no free IOBs
3. if the peer tcp does the same thing,
both sides advertize zero window and can not drain the tx queue.
For a similar stall to happen, the peer doesn't need to be
a naive tcp implementation like nuttx. A naive application blocking
on send() without draining its read buffer is enough.
(Probably such an application should be fixed to drain rx even
when tx is full. However, it's another story.)
This commit avoids the situation by prevent tx from grabbing
the all IOBs in the first place. (assuming CONFIG_IOB_THROTTLE > 0)
