Skip to main content
Visitor II
March 25, 2020
Question

STM32H7 HAL 1.7 Update ETH

  • March 25, 2020
  • 6 replies
  • 1886 views

Hey,

i saw they updated the HAL to 1.7 . Did they update the Ethernet HAL, so it works (properly)?

    This topic has been closed for replies.

    6 replies

    Technical Moderator
    March 30, 2020

    Hello @BengofEtech​ ,

    Which issue are you facing and expect it to be fixed?

    Please note that the ethernet drivers will be reworked to fix the bugs already highlighted by several members on our Community.

    Also, in HAL 1.7, we still embed LwIP 2.0.3. It is expected to switch to 2.1.2 in the coming releases.

    -Amel

    Visitor II
    March 30, 2020

    Hello @Amel NASRI​ 

    After using the NETIFAPI for Link Thread, dhcp etc. im able to optimize the programm in -O2 .

    When switching to O3, the STM32H7 is not receiving any RX Data Interrupts from the Ethernet Phy (LAN Phy which is also used on the eval Boards by STM). The LWIP Stack is working fine, he tries continously to send UDP Packets via the Ethernet output function. Currently im examinating with a probe, whether its beeing send to the Phy, since the hal does not return an error while doing this.

    Currently we need to get Perfomance out of the STM32H7 for optimizing TLS Speed, and the Product is partly delayed party by being not able to optimize on higher levels. Since the Stack works fine, i assume it must be some Error in the Init of the Phy or the HAL at higher optimization Levels. At those levels debugging gets very hard, so i cant really say, what the error is.

    I already switched manually to the 2.1.2 stack, but the problems are more related with the hal than the stack.

    regards from Stuttgart, Germany

    Super User
    March 30, 2020

    There is no significant update to ETH in v.1.7

    Visitor II
    March 31, 2020

    Edit:

    Doesnt work reliably in -O2 . When chanching code on certain locations, the HAL ETH doesnt run anymore, seems to be something with memory allignment

    Graduate II
    March 31, 2020

    You should be aware of these:

    https://community.st.com/s/question/0D50X0000C6eNNSSQ2/bug-fixes-stm32h7-ethernet

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

    And judging by this, I wouldn't hope on them fixing anything, because HAL team is a complete kindergarten incapable of doing it.

    Visitor II
    March 31, 2020

    well, then probably letting it running at o1 seems to be the best alternative. My customer wont pay hundreds of working hours to fix this.

    Graduate II
    April 1, 2020

    Well, then the customer just won't have working Ethernet, just some ever-hanging j*nk which will destroy the customer's reputation. There is no magic around it. Changing optimization levels can't fix code which is non-working by design. ST's kindergarten code for F1, F2, F4 is very bad, for F7 even worse, but for H7 it's completely non-working. In a first link I gave, Alister shares his code for H7 - probably you can try integrating that one.