Need to use tcpip_api_call() instead of tcpip_callback(), since the
former waits for the tcpip task to complete and thus prevents potential
data races with subsequent TCP/IP tasks.
This fixes a bug where we ping our own IP and the request itself
bounces back to the raw receive function and is incorrectly treated as
reply. (this bug was discovered when fixing ICMPv6 pings with incorrect
checksums, while the ping request was dropped in icmpv6.c due to wrong
checksum, but was also fed to raw layers where it was treated as
"correct" response, so the PINGv6 to ourselves still worked)
Adds backward compatibity option CONFIG_LWIP_DHCPS_ADD_DNS, which
prevents breaking changes.
This option should be removed in IDF v6.0
Merges https://github.com/espressif/esp-idf/pull/14132
* Makes LWIP_HOOK_IP6_INPUT default to LWIP_HOOK_IP6_INPUT_DEFAULT
* Updated the stub hook implementation to actually filter out all IPv6
packets if the input netif has no link local address.
In lwip-2.1.3 tcp_output_control_segment() is used to send RST, empty
ACK, keepalive or zero window probe.
In lwip-2.2.0 this works the same, but the function is used when we
don't know the interface to send to (and if used it finds it and calls)
tcp_output_control_segment_netif().
When `LWIP_EXTRA_IRAM_OPTIMIZATION=y` we would ideally move both
variants to IRAM, but the pcb variant (tcp_output_control_segment) could
be optimized away on some targets and toolchain versions.
(esp32c6 with riscv32-esp-elf/esp-14.2.0_20240906)
It should be sufficient to keep tcp_output_control_segment_netif in IRAM
for most common performance usecases.
Introducing config option `CONFIG_ESP_NETIF_SET_DNS_PER_DEFAULT_NETIF`
to overcome LWIP limitation of using global DNS server info.
This config option enables LWIP callbacks to collect per netif DNS
server info and then restores global DNS servers of whichever network
interface is selected as default.
LWIP submodule update: git log --oneline aa4f6e78..3a3d1fb3
- dns: Allow storing dnsserver per netif (espressif/esp-lwip@3a3d1fb3)
Tested under ESP32S3, running 50 tasks, each with a client, and a server
listening to one client, totalizing 150 sockets/files open, forwarding
data from PPP to Ethernet and vice versa.
Reasoning for 255 total sockets/files is VFS local_fd_t being a uint8_t.
Since FD_SETSIZE is defined at tool level, a CMakeLists user definition
is needed in cases above 64 (the default value).
Merges https://github.com/espressif/esp-idf/pull/13866
It also removes incorrect comment that's not valid any more (leftover from 6933c103e54)
It also removes ESP_THREAD_SAFE macro and sys_mbox_set_owner() which were not used
It also fixes CID 460613: Uninitialized pointer read in api_lib.c, netconn_gethostbyname_addrtype
Updated lwip submodule: git log --oneline 4297782b..aa4f6e78
* api_msg: Fix unused local variable if LWIP_NETCONN_SEM_PER_THREAD=1 (espressif/esp-lwip@aa4f6e78)
* Update submodule: git log --oneline f7922143..a1bd9e44
Detailed description of the changes:
- api_msg: Fix crash to fail-safe error if cannot get semaphore (espressif/esp-lwip@a1bd9e44)
DHCP Option 114 provides a modern method of indicating a captive
portal redirect to DHCP client. This introduces Option 114 to
the DHCPS component as well as provides examples for usage.
