This fixes connection closing issues with CONFIG_NET_TCP_WRITE_BUFFERS.
Because TCP_CLOSE is used for both of input and output for tcp_callback,
the close callback and the send callback confuses each other as
the following. As it effectively disposes the connection immediately,
we end up with responding to the consequent ACK and FIN/ACK from the peer
with RSTs.
tcp_timer
-> tcp_close_eventhandler
returns TCP_CLOSE (meaning an active close)
-> psock_send_eventhandler
called with TCP_CLOSE from tcp_close_eventhandler, misinterpet as
a passive close.
-> tcp_lost_connection
-> tcp_shutdown_monitor
-> tcp_callback
-> tcp_close_eventhandler
misinterpret TCP_CLOSE from itself as
a passive close
The current code just leave the window value from the segment
from the peer. It doesn't make sense.
Instead, always use 0.
This matches what NetBSD and Linux do.
(As far as I read their code correctly.)
* 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
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.
I plan to use a simple IOB chain w/o IOB queue for tcp recv buffers.
for large configurations, uint16_t is not large enough to represent
the entire socket buffer.
* 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.
Summary:
- I noticed that NFS over TCP does not work correctly when
it tried to reconnect to the NFS server.
- This commit fixes this issue
Impact:
- gs2200m only
Testing:
- Tested with spresense:wifi_smp
- NOTE: need to update gs2200m_main.c
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
Summary:
- I noticed that an NFS connection is automatically dropped.
- Actually, this happens in TCP mode if no traffic happens
for 600 seconds.
- I confirmed that the NFS client on Ubuntu automatically
reconnects to the NFS server when an RPC call failed.
- This commit adds the same behavior to fix the issue.
Impact:
- NFS client in TCP mode
Testing:
- Tested with spresense:rndis_smp
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
* it wasn't intended.
* it doesn't always work:
(gdb) source tools/esp32/backtrace.gdbscript
(gdb) esp32_bt 0x40139706 0x80139811 0x3ffafd40
Attempt to assign to an unmodifiable value.
(gdb)
needed anymore.
Decouple the IRAM heap from the text allocator since that heap can
still be used as a generic pool of memory.
Implement the up_extraheaps_init function to initialize all of the
additional heaps.
Signed-off-by: Abdelatif Guettouche <abdelatif.guettouche@espressif.com>
needed anymore.
Implement the up_extraheaps_init function to initialize all separate
heaps.
Signed-off-by: Abdelatif Guettouche <abdelatif.guettouche@espressif.com>
chips that have multiple separate heaps.
For now it's used to enable APIs to initialize the different heaps
during the start sequence but can be extended for other purposes that
manage those heaps.
Signed-off-by: Abdelatif Guettouche <abdelatif.guettouche@espressif.com>
Write protection must be disabled before performing changes to the WDT
registers. Furthermore, the routine was resetting the wrong field from
the RTC WDT register.
The RTC_CNTL_WDT_FLASHBOOT_MOD_EN field relates to Flash Boot Protection
and it is enabled by the 1st stage bootloader. The 2nd stage bootloader
takes care of disabling it.
Then the 2nd stage bootloader enables the RTC WDT for checking the
startup sequence of the application image.
Signed-off-by: Gustavo Henrique Nihei <gustavo.nihei@espressif.com>
Summary:
- I noticed that Telnet session is sometimes not closed
when tested with nxplayer + command stress test
- The symptom relates to the change of IOB throttling in TCP
but we are still finding the root cause.
- This commit is just a workaround to avoid the issue by
disabling telnet character mode.
Impact:
- spresense:rndis, spresense:rndis_smp only
Testing:
- Tested with nxplayer (HTTP audio streaming) + command stress test
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
from https://pubs.opengroup.org/onlinepubs/007904875/functions/setpriority.html:
1.The nice value shall in the range [-{NZERO},{NZERO} -1]
2.Lower nice value shall cause more favorable scheduling
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
Change-Id: I5ad60d92abc3b69fbaa406da68cec2e40ca3fa6d
regions.
With this we also don't need two separate linker scripts (one for IRAM
and another one for FLASH). Only one linker script (now called
esp32.ld) is now present.
Signed-off-by: Abdelatif Guettouche <abdelatif.guettouche@espressif.com>
since kernel component should use UTC instead local time
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
Change-Id: Icf939e1ab0af8e577105f539d2553bc67b3b3d10
