w5500_tsk not deleted when esp_eth driver initialization fails (IDFGH-16721)

[szerwi]

Checklist

• Checked the issue tracker for similar issues to ensure this is not a duplicate
• Read the documentation to confirm the issue is not addressed there and your configuration is set correctly
• Tested with the latest version to ensure the issue hasn't been fixed

How often does this bug occurs?

always

Expected behavior

During esp_eth_driver_install(), there is a w5500_tsk task created for W5500 module. If the initialization fails, the task should be destroyed.

Actual behavior (suspected bug)

The task is not destroyed, resulting in constantly reporting w5500.mac: w5500_get_rx_received_size(238): read RX RSR failed error.

When the W5500 is initialized for the second time, another w5500_tsk is created resulting in memory leak.

Error logs or terminal output

Errors during initialization (W5500 is physically not connected, which results in initialization error):

14:42:09.623 > E (13:42:08.164) w5500.mac: w5500_reset(308): reset timeout
14:42:09.630 > E (13:42:08.165) w5500.mac: w5500_send_command(210): send command timeout
14:42:09.637 > E (13:42:08.165) w5500.mac: emac_w5500_init(893): reset w5500 failed
14:42:09.644 > E (13:42:08.175) w5500.mac: emac_w5500_receive(757): issue RECV command failed
14:42:09.652 > E (13:42:08.176) esp_eth: esp_eth_driver_install(250): init mac failed
14:42:09.659 > E (13:42:08.186) w5500.mac: frame read from module failed
14:42:09.665 > E (13:42:08.197) ARDUINO: SPI Ethernet driver install failed: 263

After the failure, w5500_tsk keeps running resulting in errors every 1 second:

14:42:23.708 > E (14:42:22.233) w5500.mac: w5500_get_rx_received_size(238): read RX RSR failed
14:42:24.708 > E (14:42:23.233) w5500.mac: w5500_get_rx_received_size(238): read RX RSR failed
14:42:25.708 > E (14:42:24.233) w5500.mac: w5500_get_rx_received_size(238): read RX RSR failed
14:42:26.709 > E (14:42:25.233) w5500.mac: w5500_get_rx_received_size(238): read RX RSR failed

Steps to reproduce the behavior

Try to initialize W5500 driver when module is not connected. I use arduino-esp32 and ETH.begin(ETH_PHY_TYPE, ETH_PHY_ADDR, ETH_CSN_PIN, ETH_IRQ_PIN, -1, eth_spi, ETH_SPI_CLOCK_FREQ) function.

Project release version

ESP-IDF v5.5.1.250929

System architecture

Intel/AMD 64-bit (modern PC, older Mac)

Operating system

Windows

Operating system version

Windows 11

Shell

CMD

Additional context

I am using arduino-esp32 3.3.2 framework.

I have W5500 Ethernet module connected via SPI. Examining the driver initialization functions, I have narrowed down the issue and I am convencied that the bug is inside ESP-IDF W5500 driver.

The issue happens when communication with W5500 chip is not working, eg. when the chip is not powered.

[kostaond]

Hi @szerwi, thank you for reporting potential issue. I made a quick investigation and I think that issue is not in the ESP-IDF code. The w5500 task is created when the MAC instance is created by esp_eth_mac_new_w5500. esp_eth_driver_install takes separately created PHY and MAC instances and "combine" them to run the Ethernet driver. However, when esp_eth_driver_install fails, it cannot delete the MAC or PHY instance since it didn't create them. They need to be deleted by entity which originally created them. See example at https://github.com/espressif/esp-idf/blob/master/examples/ethernet/basic/main/ethernet_example_main.c#L101. Please investigate if Arduino code do so and if not, please close this issue and create a new issue in Arduino project.

[szerwi]

Hi @kostaond,
Thank you for the explanation - you are right about the source of the issue. At first I thought that esp_eth_driver_install is responsible for creating the mac and phy, as it allocates the space for entire esp_eth_handle_t, but after going through it one more time I saw that mac and phy are allocated before by other functions.

Now I see that the issue is in arduino-esp32 library, as both ETHClass::begin() and ETHClass::end() functions return on any error encountered, so the end() function stops when eth_handle is null and does not call del() for mac and phy. I will look at it more closely in a few days and submit a PR in arduino-esp32 repo.

[szerwi]

I have submitted PR in arduino-esp32 repo fixing this issue: espressif/arduino-esp32#11982