Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

net: tcp: Implement Keep-alive support #66316

Merged
merged 3 commits into from Dec 11, 2023
Merged

net: tcp: Implement Keep-alive support #66316

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e988f2f
net: tcp: Implement Keep-alive support
rlubos Dec 8, 2023
091ea1f
net: tcp: Move TCP Kconfig options to separate file
rlubos Dec 8, 2023
7e38282
tests: net: socket: tcp: Add tests for TCP keep alive feature
rlubos Dec 8, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
8 changes: 7 additions & 1 deletion include/zephyr/net/socket.h
View file
Open in desktop
Expand Up @@ -1054,7 +1054,7 @@ struct ifreq {
/** sockopt: Size of socket recv buffer */
#define SO_RCVBUF 8

/** sockopt: Enable sending keep-alive messages on connections (ignored, for compatibility) */
/** Enable sending keep-alive messages on connections */
#define SO_KEEPALIVE 9
/** sockopt: Place out-of-band data into receive stream (ignored, for compatibility) */
#define SO_OOBINLINE 10
Expand Down Expand Up @@ -1095,6 +1095,12 @@ struct ifreq {
/* Socket options for IPPROTO_TCP level */
/** sockopt: Disable TCP buffering (ignored, for compatibility) */
#define TCP_NODELAY 1
/** Start keepalives after this period (seconds) */
#define TCP_KEEPIDLE 2
/** Interval between keepalives (seconds) */
#define TCP_KEEPINTVL 3
/** Number of keepalives before dropping connection */
#define TCP_KEEPCNT 4

/* Socket options for IPPROTO_IP level */
/** sockopt: Set or receive the Type-Of-Service value for an outgoing packet. */
Expand Down
205 changes: 1 addition & 204 deletions subsys/net/ip/Kconfig
View file
Open in desktop
Expand Up @@ -385,210 +385,7 @@ config NET_MAX_MCAST_ROUTES
This determines how many entries can be stored in multicast
routing table.

config NET_TCP
bool "TCP"
depends on NET_IP
help
The value depends on your network needs.

config NET_TCP_CHECKSUM
bool "Check TCP checksum"
default y
depends on NET_TCP
help
Enables TCP handler to check TCP checksum. If the checksum is invalid,
then the packet is discarded.

if NET_TCP
module = NET_TCP
module-dep = NET_LOG
module-str = Log level for TCP
module-help = Enables TCP handler output debug messages
source "subsys/net/Kconfig.template.log_config.net"
endif # NET_TCP

config NET_TCP_TIME_WAIT_DELAY
int "How long to wait in TIME_WAIT state (in milliseconds)"
depends on NET_TCP
default 1500
help
To avoid a (low-probability) issue when delayed packets from
previous connection get delivered to next connection reusing
the same local/remote ports, RFC 793 (TCP) suggests to keep
an old, closed connection in a special "TIME_WAIT" state for
the duration of 2*MSL (Maximum Segment Lifetime). The RFC
suggests to use MSL of 2 minutes, but notes "This is an
engineering choice, and may be changed if experience indicates
it is desirable to do so." For low-resource systems, having
large MSL may lead to quick resource exhaustion (and related
DoS attacks). At the same time, the issue of packet misdelivery
is largely alleviated in the modern TCP stacks by using random,
non-repeating port numbers and initial sequence numbers. Due
to this, Zephyr uses much lower value of 1500ms by default.
Value of 0 disables TIME_WAIT state completely.

config NET_TCP_ACK_TIMEOUT
int "How long to wait for ACK (in milliseconds)"
depends on NET_TCP
default 1000
range 1 2147483647
help
This value affects the timeout when waiting ACK to arrive in
various TCP states. The value is in milliseconds. Note that
having a very low value here could prevent connectivity.

config NET_TCP_INIT_RETRANSMISSION_TIMEOUT
int "Initial value of Retransmission Timeout (RTO) (in milliseconds)"
depends on NET_TCP
default 200
range 100 60000
help
This value affects the timeout between initial retransmission
of TCP data packets. The value is in milliseconds.

config NET_TCP_RETRY_COUNT
int "Maximum number of TCP segment retransmissions"
depends on NET_TCP
default 9
help
The following formula can be used to determine the time (in ms)
that a segment will be be buffered awaiting retransmission:
n=NET_TCP_RETRY_COUNT
Sum((1<<n) * NET_TCP_INIT_RETRANSMISSION_TIMEOUT)
n=0
With the default value of 9, the IP stack will try to
retransmit for up to 1:42 minutes. This is as close as possible
to the minimum value recommended by RFC1122 (1:40 minutes).
Only 5 bits are dedicated for the retransmission count, so accepted
values are in the 0-31 range. It's highly recommended to not go
below 9, though.
Should a retransmission timeout occur, the receive callback is
called with -ECONNRESET error code and the context is dereferenced.

config NET_TCP_RANDOMIZED_RTO
bool "Use a randomized retransmission time"
default y
depends on NET_TCP
help
It can happen that two similar stacks enter a retransmission cycle
due to a packet collision. If the transmission timeout is the same
both stacks will retry at the same moment resulting in another
collision. By introducing a randomized retry timeout, the chance of
a second collision is reduced and it reduces furter the more
retransmissions occur.

config NET_TCP_FAST_RETRANSMIT
bool "Fast-retry algorithm based on the number of duplicated ACKs"
depends on NET_TCP
default y
help
When a packet is lost, the receiver will keep acknowledging the
sequence number of the last correctly received byte. Upon reception
of a sequence of acknowledgements for the same sequence number,
this can be deduced as that the packet afterwards must have been lost.
In that case a retransmission is triggerd to avoid having to wait for
the retransmit timer to elapse.

config NET_TCP_CONGESTION_AVOIDANCE
bool "Implement a congestion avoidance algorithm in TCP"
depends on NET_TCP
default y
help
To avoid overstressing a link reduce the transmission rate as soon as
packets are starting to drop.

config NET_TCP_MAX_SEND_WINDOW_SIZE
int "Maximum sending window size to use"
depends on NET_TCP
default 0
range 0 65535
help
This value affects how the TCP selects the maximum sending window
size. The default value 0 lets the TCP stack select the value
according to amount of network buffers configured in the system.

config NET_TCP_MAX_RECV_WINDOW_SIZE
int "Maximum receive window size to use"
depends on NET_TCP
default 0
range 0 65535
help
This value defines the maximum TCP receive window size. Increasing
this value can improve connection throughput, but requires more
receive buffers available in the system for efficient operation.
The default value 0 lets the TCP stack select the value
according to amount of network buffers configured in the system.

config NET_TCP_RECV_QUEUE_TIMEOUT
int "How long to queue received data (in ms)"
depends on NET_TCP
default 2000
range 0 10000
help
If we receive out-of-order TCP data, we queue it. This value tells
how long the data is kept before it is discarded if we have not been
able to pass the data to the application. If set to 0, then receive
queueing is not enabled. The value is in milliseconds.
Note that we only queue data sequentially in current version i.e.,
there should be no holes in the queue. For example, if we receive
SEQs 5,4,3,6 and are waiting SEQ 2, the data in segments 3,4,5,6 is
queued (in this order), and then given to application when we receive
SEQ 2. But if we receive SEQs 5,4,3,7 then the SEQ 7 is discarded
because the list would not be sequential as number 6 is be missing.

config NET_TCP_PKT_ALLOC_TIMEOUT
int "How long to wait for a TCP packet allocation (in ms)"
depends on NET_TCP
default 100
range 10 1000
help
The TCP network stack allocates packets from the buffers and the
allocation can take some time depending on the situation.
This value indicates how long the stack should wait for the packet to
be allocated, before returning an internal error and trying again.

config NET_TCP_WORKQ_STACK_SIZE
int "TCP work queue thread stack size"
default 1200 if X86
default 1024
depends on NET_TCP
help
Set the TCP work queue thread stack size in bytes.

config NET_TCP_ISN_RFC6528
bool "Use ISN algorithm from RFC 6528"
default y
depends on NET_TCP
select MBEDTLS
select MBEDTLS_MD
select MBEDTLS_MAC_MD5_ENABLED
help
Implement Initial Sequence Number calculation as described in
RFC 6528 chapter 3. https://tools.ietf.org/html/rfc6528
If this is not set, then sys_rand32_get() is used for ISN value.

config NET_TCP_WORKER_PRIO
int "Priority of the TCP work queue"
default 2
depends on NET_TCP
help
Set the priority of the the TCP worker queue, that handles all
transmission and maintenance within the TCP stack.
Value 0 = highest priortity.
When CONFIG_NET_TC_THREAD_COOPERATIVE = y, lowest priority is
CONFIG_NUM_COOP_PRIORITIES-1 else lowest priority is
CONFIG_NUM_PREEMPT_PRIORITIES-1.
Make sure the priority is lower than lower layer TX threads to
avoid the TCP stack consume all net_bufs before transferring
execution to the lower layer network stack, with a high risk of
running out of net_bufs.

config NET_TCP_REJECT_CONN_WITH_RST
bool "Reject connection attempts on unbound TCP ports with RST"
default y
help
If enabled, TCP stack will reject connection attempts on unbound ports
with TCP RST packet.
source "subsys/net/ip/Kconfig.tcp"

config NET_TEST_PROTOCOL
bool "JSON based test protocol (UDP)"
Expand Down