STM32H735 ETH_IRQHandler() is not called

[MStei.4]

Dear Forum,

I am using STM32H735 with RTOS LwIP with a DP83826 PHY, and want to see at least a ping on the network.

From HW point of view the PHY seems to work because I can communicate with it and read config registers. The strap pins are set to a valid default configuration.

On the Rx0+Rx1 pins at the uC I can see some incoming frames (oscilloscope), but there are no output Tx0/Tx1 frames visible.

 

I have checked my configuration against a Nucleo-STM32F439ZI configuration, where a ping works well.

The LWIP_RAM_HEAP_POINTER was configured wrong, but I have set the address in the lwipopts.h file to 0x30004000, so no hard fault anymore.

The function ETH_IRQHandler() is not called event though the NVIC init functions are set:

void HAL_ETH_MspInit(ETH_HandleTypeDef* ethHandle)

{

...

  /* Peripheral interrupt init */

  HAL_NVIC_SetPriority(ETH_IRQn, 5, 0);

  HAL_NVIC_EnableIRQ(ETH_IRQn);

...

}

I did also generate a code sample (just for comparision) with cube with STM32H735 and RTOS and LwIP, but I have no clue where to search or what to check anymore. I have increased the Task Stack a little above the cube-default, but no luck yet.

There is no entry of the MAC in the ARP table of course.

Is there anything else I can check?

[Imen.D]

Hello @MStei.4​ ,

Check the PHY parameter for the Ethernet interface.

Is the PHY Address defined in 0?

Imen

[Piranha]

He said that he can read the PHY registers. And a wrong PHY address will not stop ETH peripheral interrupts from functioning.

[Piranha]

Could be this:

https://community.st.com/s/question/0D53W00001oImr4SAC/bug-stm32-lwip-ethernet-driver-tx-deadlock

Or many other issues...

https://community.st.com/s/question/0D50X0000BOtfhnSQB/how-to-make-ethernet-and-lwip-working-on-stm32

[MStei.4]

Hello, yes reading / writing PHY registers works fine

DP83826_BMCR: 00003000

DP83826_BMSR: 0000786d

DP83826_PHYIDR1: 00002000

DP83826_PHYIDR2: 0000a111

DP83826_PHYSTS: 00004f11

DP83826_PHYSCR: 0000010b

DP83826_PHYCR: 00008401

The PHY Address is in PHYCR Bit0-4 and is 1, I hope also that I do not have configured something wrong, but the strap pins are setup for basic-mode including RMII interface. rxclk = 50MHz, Bias resistor is correct. 100Mbit is set... all according data sheet. I will investigate in the deadlock that you wrote below, thank you!

[MStei.4]

Maybe my problem is more fundamental. It was a good hint with the tx deadlock, I did the change as described below in the low_level_init but did not help so far.

/* create a binary semaphore used for informing ethernetif of frame reception */
  RxPktSemaphore = osSemaphoreNew(1, 0, NULL);
 
  /* create a binary semaphore used for informing ethernetif of frame transmission */
  TxPktSemaphore = osSemaphoreNew(1, 0, NULL);

The ETH_IRQHandler() is hit once, I haven't seen it before because of debugging the wrong code part, shame on me. But the error is still the same.

After the IRQHandler hit I press run and pause and then, the heth structure states show that:

gState: 0x23 /*!< an internal process is started       */

ErrorCode: 0x08 /*!< DMA transfer error */

DMAErrorCode: 0x4080

I still don't know what is the reason. What else I can chreck here?