STM32H723 HAL example raises HAL_ETH_ErrorCallback on every packet, but still works!

[MS.9]

Hi!

when I run the HAL example code LwIP_HTTP_Server_Netconn_RTOS I get HAL_ETH_ErrorCallback() called on every packet, but the program seems to work fine.

its this bit that gets called (so something hapenning with DMA)

if((HAL_ETH_GetDMAError(heth) & ETH_DMACSR_RBU) == ETH_DMACSR_RBU)
{
    osSemaphoreRelease(RxPktSemaphore);
}

I've read as much as is humanly possible on this and other forums to see if this is normal or what the issue could be, but I have found nothing.

I've played around with using different LwIP & Ethernet buffer sizes etc, and different memory sections/protection options etc, but it wont go away, yet the program still runs fine.

Can anyone explain what is going on please.

the code I'm using is in

STM32Cube_FW_H7_V1.11.2\Projects\NUCLEO-H723ZG\Applications\LwIP\LwIP_HTTP_Server_Netconn_RTOS

thanks all,

Mark

[LCE]

> I've played around with using different LwIP & Ethernet buffer sizes etc, and different memory
> sections/protection options etc, but it wont go away, yet the program still runs fine.

Have you also increased the number of descriptors?

I don't use the HAL ETH driver and I don't know Netcon or RTOS ...

So what kind of "lwIP RX buffer scheme" is used?

[ASEHST]

Hello @MS.9,

Thank you for your contribution. The RBU problem was resolved by correcting the tail pointer calculation in this fix: [HAL][ETH] Fix the calculation of the tail pointer so that it points … · STMicroelectronics/stm32h7xx_hal_driver@ceda3ce (github.com)

 

With Regards,

[MS.9]

Thanks for the quick response and fix.

I had seen this error for a long while and never really looked into it as everyhing seemed to work ok.

sad to see it go - was becoming a familiar friend :)