esp-idf/components/lwip/Kconfig

681 lines
27 KiB
Plaintext

menu "LWIP"
config L2_TO_L3_COPY
bool "Enable copy between Layer2 and Layer3 packets"
default n
help
If this feature is enabled, all traffic from layer2(WIFI Driver) will be
copied to a new buffer before sending it to layer3(LWIP stack), freeing
the layer2 buffer.
Please be notified that the total layer2 receiving buffer is fixed and
ESP32 currently supports 25 layer2 receiving buffer, when layer2 buffer
runs out of memory, then the incoming packets will be dropped in hardware.
The layer3 buffer is allocated from the heap, so the total layer3 receiving
buffer depends on the available heap size, when heap runs out of memory,
no copy will be sent to layer3 and packet will be dropped in layer2.
Please make sure you fully understand the impact of this feature before
enabling it.
config ETHARP_SUPPORT_VLAN
bool "Support receiving and sending ethernet packets with VLAN header"
default n
help
ETHARP_SUPPORT_VLAN==1: support receiving and sending ethernet packets with
VLAN header. See the description of LWIP_HOOK_VLAN_CHECK and LWIP_HOOK_VLAN_SET
hooks to check/set VLAN headers.
If ETHARP_VLAN_CHECK is defined, only VLAN-traffic for this VLAN is accepted.
If ETHARP_VLAN_CHECK is not defined, all traffic is accepted.
config LWIP_IRAM_OPTIMIZATION
bool "Enable LWIP IRAM optimization"
default n
help
If this feature is enabled, some functions relating to RX/TX in LWIP will be
put into IRAM, it can improve UDP/TCP throughput by >10% for single core mode,
it doesn't help too much for dual core mode. On the other hand, it needs about
10KB IRAM for these optimizations.
If this feature is disabled, all lwip functions will be put into FLASH.
config LWIP_MAX_SOCKETS
int "Max number of open sockets"
range 1 16
default 10
help
Sockets take up a certain amount of memory, and allowing fewer
sockets to be open at the same time conserves memory. Specify
the maximum amount of sockets here. The valid value is from 1
to 16.
config LWIP_RANDOMIZE_INITIAL_LOCAL_PORTS
bool "Randomize the local port for the first"
default y
help
LWIP_RANDOMIZE_INITIAL_LOCAL_PORTS==1: randomize the local port for the first
local TCP/UDP pcb (default==0). This can prevent creating predictable port
numbers after booting a device.
config USE_ONLY_LWIP_SELECT
bool "Support LWIP socket select() only"
default n
help
The virtual filesystem layer of select() redirects sockets to
lwip_select() and non-socket file descriptors to their respective driver
implementations. If this option is enabled then all calls of select()
will be redirected to lwip_select(), therefore, select can be used
for sockets only.
config LWIP_SO_LINGER
bool "Enable SO_LINGER processing"
default n
help
Enabling this option allows SO_LINGER processing.
l_onoff = 1,l_linger can set the timeout.
If l_linger=0, When a connection is closed, TCP will terminate the connection.
This means that TCP will discard any data packets stored in the socket send buffer
and send an RST to the peer.
If l_linger!=0,Then closesocket() calls to block the process until
the remaining data packets has been sent or timed out.
config LWIP_SO_REUSE
bool "Enable SO_REUSEADDR option"
default y
help
Enabling this option allows binding to a port which remains in
TIME_WAIT.
config LWIP_SO_REUSE_RXTOALL
bool "SO_REUSEADDR copies broadcast/multicast to all matches"
depends on LWIP_SO_REUSE
default y
help
Enabling this option means that any incoming broadcast or multicast
packet will be copied to all of the local sockets that it matches
(may be more than one if SO_REUSEADDR is set on the socket.)
This increases memory overhead as the packets need to be copied,
however they are only copied per matching socket. You can safely
disable it if you don't plan to receive broadcast or multicast
traffic on more than one socket at a time.
config LWIP_SO_RCVBUF
bool "Enable SO_RCVBUF option"
default n
help
Enabling this option allows checking for available data on a netconn.
config LWIP_IP4_FRAG
bool "Enable fragment outgoing IP4 packets"
default y
help
Enabling this option allows fragmenting outgoing IP4 packets if their size
exceeds MTU.
config LWIP_IP6_FRAG
bool "Enable fragment outgoing IP6 packets"
default y
help
Enabling this option allows fragmenting outgoing IP6 packets if their size
exceeds MTU.
config LWIP_IP4_REASSEMBLY
bool "Enable reassembly incoming fragmented IP4 packets"
default n
help
Enabling this option allows reassemblying incoming fragmented IP4 packets.
config LWIP_IP6_REASSEMBLY
bool "Enable reassembly incoming fragmented IP6 packets"
default n
help
Enabling this option allows reassemblying incoming fragmented IP6 packets.
config LWIP_STATS
bool "Enable LWIP statistics"
default n
help
Enabling this option allows LWIP statistics
config LWIP_ETHARP_TRUST_IP_MAC
bool "Enable LWIP ARP trust"
default n
help
Enabling this option allows ARP table to be updated.
If this option is enabled, the incoming IP packets cause the ARP table to be
updated with the source MAC and IP addresses supplied in the packet.
You may want to disable this if you do not trust LAN peers to have the
correct addresses, or as a limited approach to attempt to handle
spoofing. If disabled, lwIP will need to make a new ARP request if
the peer is not already in the ARP table, adding a little latency.
The peer *is* in the ARP table if it requested our address before.
Also notice that this slows down input processing of every IP packet!
There are two known issues in real application if this feature is enabled:
- The LAN peer may have bug to update the ARP table after the ARP entry is aged out.
If the ARP entry on the LAN peer is aged out but failed to be updated, all IP packets
sent from LWIP to the LAN peer will be dropped by LAN peer.
- The LAN peer may not be trustful, the LAN peer may send IP packets to LWIP with
two different MACs, but the same IP address. If this happens, the LWIP has problem
to receive IP packets from LAN peer.
So the recommendation is to disable this option.
Here the LAN peer means the other side to which the ESP station or soft-AP is connected.
config ESP_GRATUITOUS_ARP
bool "Send gratuitous ARP periodically"
default y
help
Enable this option allows to send gratuitous ARP periodically.
This option solve the compatibility issues.If the ARP table of the AP is old, and the AP
doesn't send ARP request to update it's ARP table, this will lead to the STA sending IP packet fail.
Thus we send gratuitous ARP periodically to let AP update it's ARP table.
config GARP_TMR_INTERVAL
int "GARP timer interval(seconds)"
default 60
depends on ESP_GRATUITOUS_ARP
help
Set the timer interval for gratuitous ARP. The default value is 60s
config TCPIP_RECVMBOX_SIZE
int "TCPIP task receive mail box size"
default 32
range 6 64 if !LWIP_WND_SCALE
range 6 1024 if LWIP_WND_SCALE
help
Set TCPIP task receive mail box size. Generally bigger value means higher throughput
but more memory. The value should be bigger than UDP/TCP mail box size.
config LWIP_DHCP_DOES_ARP_CHECK
bool "DHCP: Perform ARP check on any offered address"
default y
help
Enabling this option performs a check (via ARP request) if the offered IP address
is not already in use by another host on the network.
config LWIP_DHCP_RESTORE_LAST_IP
bool "DHCP: Restore last IP obtained from DHCP server"
default n
help
When this option is enabled, DHCP client tries to re-obtain last valid IP address obtained from DHCP
server. Last valid DHCP configuration is stored in nvs and restored after reset/power-up. If IP is still
available, there is no need for sending discovery message to DHCP server and save some time.
menu "DHCP server"
config LWIP_DHCPS_LEASE_UNIT
int "Multiplier for lease time, in seconds"
range 1 3600
default 60
help
The DHCP server is calculating lease time multiplying the sent
and received times by this number of seconds per unit.
The default is 60, that equals one minute.
config LWIP_DHCPS_MAX_STATION_NUM
int "Maximum number of stations"
range 1 64
default 8
help
The maximum number of DHCP clients that are connected to the server.
After this number is exceeded, DHCP server removes of the oldest device
from it's address pool, without notification.
endmenu # DHCPS
menuconfig LWIP_AUTOIP
bool "Enable IPV4 Link-Local Addressing (AUTOIP)"
default n
help
Enabling this option allows the device to self-assign an address
in the 169.256/16 range if none is assigned statically or via DHCP.
See RFC 3927.
config LWIP_AUTOIP_TRIES
int "DHCP Probes before self-assigning IPv4 LL address"
range 1 100
default 2
depends on LWIP_AUTOIP
help
DHCP client will send this many probes before self-assigning a
link local address.
From LWIP help: "This can be set as low as 1 to get an AutoIP
address very quickly, but you should be prepared to handle a
changing IP address when DHCP overrides AutoIP." (In the case of
ESP-IDF, this means multiple SYSTEM_EVENT_STA_GOT_IP events.)
config LWIP_AUTOIP_MAX_CONFLICTS
int "Max IP conflicts before rate limiting"
range 1 100
default 9
depends on LWIP_AUTOIP
help
If the AUTOIP functionality detects this many IP conflicts while
self-assigning an address, it will go into a rate limited mode.
config LWIP_AUTOIP_RATE_LIMIT_INTERVAL
int "Rate limited interval (seconds)"
range 5 120
default 20
depends on LWIP_AUTOIP
help
If rate limiting self-assignment requests, wait this long between
each request.
config LWIP_IPV6_AUTOCONFIG
bool "Enable IPV6 stateless address autoconfiguration"
default n
help
Enabling this option allows the devices to IPV6 stateless address autoconfiguration.
See RFC 4862.
menuconfig LWIP_NETIF_LOOPBACK
bool "Support per-interface loopback"
default y
help
Enabling this option means that if a packet is sent with a destination
address equal to the interface's own IP address, it will "loop back" and
be received by this interface.
config LWIP_LOOPBACK_MAX_PBUFS
int "Max queued loopback packets per interface"
range 0 16
default 8
depends on LWIP_NETIF_LOOPBACK
help
Configure the maximum number of packets which can be queued for
loopback on a given interface. Reducing this number may cause packets
to be dropped, but will avoid filling memory with queued packet data.
menu "TCP"
config LWIP_TCP_ISN_HOOK
bool "Enable TCP ISN Hook"
default y
help
Enables custom TCP ISN hook to randomize initial sequence
number in TCP connection. This is recommended as default
lwIP implementation (`tcp_next_iss`) is not very strong,
as it does not take into consideration any platform
specific entropy source.
config LWIP_MAX_ACTIVE_TCP
int "Maximum active TCP Connections"
range 1 1024
default 16
help
The maximum number of simultaneously active TCP
connections. The practical maximum limit is
determined by available heap memory at runtime.
Changing this value by itself does not substantially
change the memory usage of LWIP, except for preventing
new TCP connections after the limit is reached.
config LWIP_MAX_LISTENING_TCP
int "Maximum listening TCP Connections"
range 1 1024
default 16
help
The maximum number of simultaneously listening TCP
connections. The practical maximum limit is
determined by available heap memory at runtime.
Changing this value by itself does not substantially
change the memory usage of LWIP, except for preventing
new listening TCP connections after the limit is reached.
config LWIP_TCP_HIGH_SPEED_RETRANSMISSION
bool "TCP high speed retransmissions"
default y
help
Speed up the TCP retransmission interval. If disabled,
it is recommended to change the number of SYN retransmissions to 6,
and TCP initial rto time to 3000.
config TCP_MAXRTX
int "Maximum number of retransmissions of data segments"
default 12
range 3 12
help
Set maximum number of retransmissions of data segments.
config TCP_SYNMAXRTX
int "Maximum number of retransmissions of SYN segments"
default 6 if !LWIP_TCP_HIGH_SPEED_RETRANSMISSION
default 12 if LWIP_TCP_HIGH_SPEED_RETRANSMISSION
range 3 12
help
Set maximum number of retransmissions of SYN segments.
config TCP_MSS
int "Maximum Segment Size (MSS)"
default 1440
range 536 1460
help
Set maximum segment size for TCP transmission.
Can be set lower to save RAM, the default value 1460(ipv4)/1440(ipv6) will give best throughput.
IPv4 TCP_MSS Range: 576 <= TCP_MSS <= 1460
IPv6 TCP_MSS Range: 1220<= TCP_MSS <= 1440
config TCP_MSL
int "Maximum segment lifetime (MSL)"
default 60000
help
Set maximum segment lifetime in in milliseconds.
config TCP_SND_BUF_DEFAULT
int "Default send buffer size"
default 5744 # 4 * default MSS
range 2440 65535 if !LWIP_WND_SCALE
range 2440 1024000 if LWIP_WND_SCALE
help
Set default send buffer size for new TCP sockets.
Per-socket send buffer size can be changed at runtime
with lwip_setsockopt(s, TCP_SNDBUF, ...).
This value must be at least 2x the MSS size, and the default
is 4x the default MSS size.
Setting a smaller default SNDBUF size can save some RAM, but
will decrease performance.
config TCP_WND_DEFAULT
int "Default receive window size"
default 5744 # 4 * default MSS
range 2440 65535 if !LWIP_WND_SCALE
range 2440 1024000 if LWIP_WND_SCALE
help
Set default TCP receive window size for new TCP sockets.
Per-socket receive window size can be changed at runtime
with lwip_setsockopt(s, TCP_WINDOW, ...).
Setting a smaller default receive window size can save some RAM,
but will significantly decrease performance.
config TCP_RECVMBOX_SIZE
int "Default TCP receive mail box size"
default 6
range 6 64 if !LWIP_WND_SCALE
range 6 1024 if LWIP_WND_SCALE
help
Set TCP receive mail box size. Generally bigger value means higher throughput
but more memory. The recommended value is: TCP_WND_DEFAULT/TCP_MSS + 2, e.g. if
TCP_WND_DEFAULT=14360, TCP_MSS=1436, then the recommended receive mail box size is
(14360/1436 + 2) = 12.
TCP receive mail box is a per socket mail box, when the application receives packets
from TCP socket, LWIP core firstly posts the packets to TCP receive mail box and the
application then fetches the packets from mail box. It means LWIP can caches maximum
TCP_RECCVMBOX_SIZE packets for each TCP socket, so the maximum possible cached TCP packets
for all TCP sockets is TCP_RECCVMBOX_SIZE multiples the maximum TCP socket number. In other
words, the bigger TCP_RECVMBOX_SIZE means more memory.
On the other hand, if the receiv mail box is too small, the mail box may be full. If the
mail box is full, the LWIP drops the packets. So generally we need to make sure the TCP
receive mail box is big enough to avoid packet drop between LWIP core and application.
config TCP_QUEUE_OOSEQ
bool "Queue incoming out-of-order segments"
default y
help
Queue incoming out-of-order segments for later use.
Disable this option to save some RAM during TCP sessions, at the expense
of increased retransmissions if segments arrive out of order.
config ESP_TCP_KEEP_CONNECTION_WHEN_IP_CHANGES
bool "Keep TCP connections when IP changed"
default n
help
This option is enabled when the following scenario happen:
network dropped and reconnected, IP changes is like: 192.168.0.2->0.0.0.0->192.168.0.2
Disable this option to keep consistent with the original LWIP code behavior.
choice TCP_OVERSIZE
prompt "Pre-allocate transmit PBUF size"
default TCP_OVERSIZE_MSS
help
Allows enabling "oversize" allocation of TCP transmission pbufs ahead of time,
which can reduce the length of pbuf chains used for transmission.
This will not make a difference to sockets where Nagle's algorithm
is disabled.
Default value of MSS is fine for most applications, 25% MSS may save
some RAM when only transmitting small amounts of data. Disabled will
have worst performance and fragmentation characteristics, but uses
least RAM overall.
config TCP_OVERSIZE_MSS
bool "MSS"
config TCP_OVERSIZE_QUARTER_MSS
bool "25% MSS"
config TCP_OVERSIZE_DISABLE
bool "Disabled"
endchoice
config LWIP_WND_SCALE
bool "Support TCP window scale"
depends on WIFI_LWIP_ALLOCATION_FROM_SPIRAM_FIRST
default n
help
Enable this feature to support TCP window scaling.
config TCP_RCV_SCALE
int "Set TCP receiving window scaling factor"
depends on LWIP_WND_SCALE
range 0 14
default 0
help
Enable this feature to support TCP window scaling.
config LWIP_TCP_RTO_TIME
int "Default TCP rto time"
default 3000 if !LWIP_TCP_HIGH_SPEED_RETRANSMISSION
default 1500 if LWIP_TCP_HIGH_SPEED_RETRANSMISSION
help
Set default TCP rto time for a reasonable initial rto.
In bad network environment, recommend set value of rto time to 1500.
endmenu # TCP
menu "UDP"
config LWIP_MAX_UDP_PCBS
int "Maximum active UDP control blocks"
range 1 1024
default 16
help
The maximum number of active UDP "connections" (ie
UDP sockets sending/receiving data).
The practical maximum limit is determined by available
heap memory at runtime.
config UDP_RECVMBOX_SIZE
int "Default UDP receive mail box size"
default 6
range 6 64
help
Set UDP receive mail box size. The recommended value is 6.
UDP receive mail box is a per socket mail box, when the application receives packets
from UDP socket, LWIP core firstly posts the packets to UDP receive mail box and the
application then fetches the packets from mail box. It means LWIP can caches maximum
UDP_RECCVMBOX_SIZE packets for each UDP socket, so the maximum possible cached UDP packets
for all UDP sockets is UDP_RECCVMBOX_SIZE multiples the maximum UDP socket number. In other
words, the bigger UDP_RECVMBOX_SIZE means more memory.
On the other hand, if the receiv mail box is too small, the mail box may be full. If the
mail box is full, the LWIP drops the packets. So generally we need to make sure the UDP
receive mail box is big enough to avoid packet drop between LWIP core and application.
endmenu # UDP
config TCPIP_TASK_STACK_SIZE
int "TCP/IP Task Stack Size"
default 3072
# for high log levels, tcpip_adapter API calls can end up
# a few calls deep and logging there can trigger a stack overflow
range 2048 65536 if LOG_DEFAULT_LEVEL < 4
range 2560 65536 if LOG_DEFAULT_LEVEL >= 4
help
Configure TCP/IP task stack size, used by LWIP to process multi-threaded TCP/IP operations.
Setting this stack too small will result in stack overflow crashes.
choice TCPIP_TASK_AFFINITY
prompt "TCP/IP task affinity"
default TCPIP_TASK_AFFINITY_NO_AFFINITY
help
Allows setting LwIP tasks affinity, i.e. whether the task is pinned to
CPU0, pinned to CPU1, or allowed to run on any CPU.
Currently this applies to "TCP/IP" task and "Ping" task.
config TCPIP_TASK_AFFINITY_NO_AFFINITY
bool "No affinity"
config TCPIP_TASK_AFFINITY_CPU0
bool "CPU0"
config TCPIP_TASK_AFFINITY_CPU1
bool "CPU1"
depends on !FREERTOS_UNICORE
endchoice
config TCPIP_TASK_AFFINITY
hex
default FREERTOS_NO_AFFINITY if TCPIP_TASK_AFFINITY_NO_AFFINITY
default 0x0 if TCPIP_TASK_AFFINITY_CPU0
default 0x1 if TCPIP_TASK_AFFINITY_CPU1
config LWIP_IPV6_MEMP_NUM_ND6_QUEUE
int "Max number of IPv6 packets to queue during MAC resolution"
range 3 20
default 3
help
Config max number of IPv6 packets to queue during MAC resolution.
config LWIP_IPV6_ND6_NUM_NEIGHBORS
int "Max number of entries in IPv6 neighbor cache"
range 3 10
default 5
help
Config max number of entries in IPv6 neighbor cache.
menuconfig PPP_SUPPORT
bool "Enable PPP support (new/experimental)"
default n
help
Enable PPP stack. Now only PPP over serial is possible.
PPP over serial support is experimental and unsupported.
config PPP_NOTIFY_PHASE_SUPPORT
bool "Enable Notify Phase Callback"
depends on PPP_SUPPORT
default n
help
Enable to set a callback which is called on change of the internal PPP state machine.
config PPP_PAP_SUPPORT
bool "Enable PAP support"
depends on PPP_SUPPORT
default n
help
Enable Password Authentication Protocol (PAP) support
config PPP_CHAP_SUPPORT
bool "Enable CHAP support"
depends on PPP_SUPPORT
default n
help
Enable Challenge Handshake Authentication Protocol (CHAP) support
config PPP_MSCHAP_SUPPORT
bool "Enable MSCHAP support"
depends on PPP_SUPPORT
default n
help
Enable Microsoft version of the Challenge-Handshake Authentication Protocol (MSCHAP) support
config PPP_MPPE_SUPPORT
bool "Enable MPPE support"
depends on PPP_SUPPORT
default n
help
Enable Microsoft Point-to-Point Encryption (MPPE) support
config PPP_DEBUG_ON
bool "Enable PPP debug log output"
depends on PPP_SUPPORT
default n
help
Enable PPP debug log output
menu "ICMP"
config LWIP_MULTICAST_PING
bool "Respond to multicast pings"
default n
config LWIP_BROADCAST_PING
bool "Respond to broadcast pings"
default n
endmenu # ICMP
menu "LWIP RAW API"
config LWIP_MAX_RAW_PCBS
int "Maximum LWIP RAW PCBs"
range 1 1024
default 16
help
The maximum number of simultaneously active LWIP
RAW protocol control blocks. The practical maximum
limit is determined by available heap memory at runtime.
endmenu # LWIP RAW API
menu "SNTP"
config LWIP_DHCP_MAX_NTP_SERVERS
int "Maximum number of NTP servers"
default 1
range 1 16
help
Set maximum number of NTP servers used by LwIP SNTP module.
First argument of sntp_setserver/sntp_setservername functions
is limited to this value.
config LWIP_SNTP_UPDATE_DELAY
int "Request interval to update time (ms)"
range 15000 4294967295
default 3600000
help
This option allows you to set the time update period via SNTP.
Default is 1 hour. Must not be below 15 seconds by specification.
(SNTPv4 RFC 4330 enforces a minimum update time of 15 seconds).
endmenu # SNTP
config LWIP_ESP_LWIP_ASSERT
bool "Enable LWIP ASSERT checks"
default y
help
Enable this option allows lwip to check assert.
It is recommended to keep it open, do not close it.
endmenu