Do not bother to preserve segment boundaries in the tcp
readahead queues.
* Avoid wasting the tail IOB space for each segments.
Instead, pack the newly received data into the tail space
of the last IOB. Also, advertise the tail space as
a part of the window.
* Use IOB chain directly. Eliminate IOB queue overhead.
* Allow to accept only a part of a segment.
* This change improves the memory efficiency.
And probably more importantly, allows less-confusing
recv window advertisement behavior.
Previously, even when we advertise N bytes window,
we often couldn't actually accept N bytes. Depending on
the segment sizes and IOB configurations, it was causing
segment drops.
Also, the previous code was moving the right edge of the
window back and forth too often, even when nothing in
the system was competing on the IOBs. Shrinking the
window that way is a kinda well known recipe to confuse
the peer stack.
* 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.
I assume this was just an oversight because I couldn't
find any obvious reason to special-case only the first IOB.
The commit message of the original commit is cited below.
```
commit bf210560017b8df1201c02bcf1afee1fabaa37e8
Author: chao.an <anchao@xiaomi.com>
Date: Fri Nov 27 09:50:38 2020 +0800
net/tcp: fallback to unthrottle pool to avoid deadlock
Add a fallback mechanism to ensure that there are still available
iobs for an free connection, Guarantees all connections will have
a minimum threshold iob to keep the connection not be hanged.
Change-Id: I59bed98d135ccd1f16264b9ccacdd1b0d91261de
Signed-off-by: chao.an <anchao@xiaomi.com>
```
* 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
Gregory Nutt is the copyright holder for those files and he has submitted the
SGA as a result we can migrate the licenses to Apache.
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
NuttX provides the UDP_BINDTODEVICE socket option. This is a UDP protocol-specific implementation of the semi-standard Linux SO_BINDTODEVICE socket option: "SO_BINDTODEVICE forces packets on the socket to only egress the bound interface, regardless of what the IP routing table would normally choose. Similarly only packets which ingress the bound interface will be received on the socket, packets from other interfaces will not be delivered to the socket." https://codingrelic.geekhold.com/2009/10/code-snippet-sobindtodevice.html
If CONFIG_NET_UDP_BINDTODEVICE is selected and a UDP socket is bound to the device, then unrecognized packets UDP packets must not be dropped, but must be forwarded along to the bound socket unconditionally.
It the typical case, this should have no impact. It does effect the applications that use DHCP and do select the UDP_BINDTODEVICE socket option.
This PR replace existing improper logic in the code and also the improper attempts to fix problems from PR #3601 and PR #3598. Those changes are improper because they expose DHCP appliction dependencies in the OS, breaking modularity and independence of the OS and application.
Tested with stm32f4discovery:netnsh with CONFIG_NET_UDP_BINDTODEVICE. A proper DHCP test setup is needed, however.
Several licenses were missed in the initial work
David Sidrane has submitted the ICLA and we can migrate the licenses
to Apache.
Gregory Nutt has submitted the SGA and we can migrate the licenses
to Apache.
Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
Summary:
- In arp_wait_setup() and arp_wait_cancel(), g_arp_waiters
is protected by a critical section.
- However, I noticed that arp_notify() does not protect the
g_arp_waiters that would cause memory corruption
- This commit fixes the issue.
Impact:
- None
Testing:
- Tested with spresense:rndis_smp, spresense:rndis
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
