esp-idf: WiFi stops transmitting data in long-term operation test (IDFGH-7373)

Environment

• Development Kit: N/A
• Kit version (for WroverKit/PicoKit/DevKitC): N/A
• Module or chip used: ESP32-WROOM-32D
• IDF version (run git describe --tags to find it): v4.4.1 (also v4.3.2)
• Build System: Make
• Compiler version (run xtensa-esp32-elf-gcc --version to find it): xtensa-esp32-elf-gcc (crosstool-NG esp-2021r2-patch3) 8.4.0
• Operating System: Linux
• Using an IDE?: No
• Power Supply: USB

Problem Description

In our long-term operation test, WiFi stack seems to stop transmitting data while keeping the connection to the access point… Our application uses esp-tls to connect the server and UDP to communicate with devices in local area network. Both of them stops the operation.

Expected Behavior

WiFi stack continue to transmit data.

Actual Behavior

WiFi stack perhaps fails to flush its buffer.

Steps to reproduce

This problem occurs 2 or 3 weeks after when we started long-term operation test, it’s hard to reproduce and debug. Our application uses both BLE and WiFi.

Debug Logs

getaddrinfo in esp-tls failed

log from esp-tls

2022-05-14 14:00:14,922 : E (1989386792) esp-tls: couldn't get hostname for :xxx.yyy.zzz: getaddrinfo() returns 202, addrinfo=0x0

additional log from our app showing esp-tls internal error and errno (EAGAIN).

2022-05-14 14:00:39,913 : W (1989411792) esp_util: esp error: 32769, ESP_ERR_ESP_TLS_CANNOT_RESOLVE_HOSTNAME
2022-05-14 14:00:39,913 : W (1989411802) esp_util: errno: 11

We just call esp_tls_conn_new_async().

        r = esp_tls_conn_new_async(url + u->field_data[UF_HOST].off, u->field_data[UF_HOST].len,
                                   get_port(url, u), &cfg, tls);

UDP send failed

Our app send UDP multicast to local area network.

2022-05-14 14:00:51,500 : E (1989423402) el-udp: sendto failed: 12
2022-05-14 14:00:51,595 : W (1989423502) wifi:m f null
2022-05-14 14:00:51,643 : W (1989423562) wifi:m f null

Implementation is just like

    r = sendto(s_udp->fd, buf, req_len, 0, (struct sockaddr *)&ip4_multi_addr,
                sizeof(ip4_multi_addr));
    if (-1 == r) {
        ESP_LOGE(TAG, "sendto failed: %d", errno);
    }

Other items if possible

sdkconfig.defaults

Related configurations are like this:

#
# Bluetooth
#
CONFIG_BT_ENABLED=y
CONFIG_BT_NIMBLE_ENABLED=y
CONFIG_BT_NIMBLE_TASK_STACK_SIZE=8192

#
# Wi-Fi
#
CONFIG_ESP32_WIFI_NVS_ENABLED=n

#
# PHY
#
CONFIG_ESP_PHY_CALIBRATION_AND_DATA_STORAGE=n
CONFIG_ESP_PHY_INIT_DATA_IN_PARTITION=y

#
# DHCP server
#
CONFIG_LWIP_IPV6_AUTOCONFIG=y

#
# TCP
#
CONFIG_LWIP_TCP_SYNMAXRTX=6
CONFIG_LWIP_TCP_MSS=1436
CONFIG_LWIP_TCP_RTO_TIME=3000
# end of TCP

thanks in advance,