STM32H7 LwIP sending raw Ethernet frames with no recognizable protocol

[Akhil1]

I am trying to setup DHCP over Ethernet using LwIP, with an ultimate goal of using TCP/UDP for communication. I have configured everything through Cube and set up the MPU according to the FAQ guidelines for Ethernet on STM32H7 and am finally receiving/sending packets.

Unfortunately the packets I receive are just raw Ethernet frames w/ no recognizable protocol. I searched around and only found one other post about this and the fix for them was to call LwIP_Process() in the main loop which I am already doing.

Here is what wire-shark is capturing on the interface the stm32 is connected to:

Here is my MPU Configuration:

I have also attached my IOC file, any help will be greatly appreciated.

Edit: Providing more context:

Buffer definition:

ETH_DMADescTypeDef DMARxDscrTab[ETH_RX_DESC_CNT]
    __attribute__((section(".RxDecripSection"))); /* Ethernet Rx DMA Descriptors */
ETH_DMADescTypeDef DMATxDscrTab[ETH_TX_DESC_CNT]
    __attribute__((section(".TxDecripSection"))); /* Ethernet Tx DMA Descriptors */
uint8_t Rx_Buff[ETH_RX_DESC_CNT][ETH_RX_BUFFER_SIZE]
    __attribute__((section(".RxArraySection"))); /* Ethernet Receive Buffers */

Linker script:

/* Entry Point */
ENTRY(Reset_Handler)
 
/* Highest address of the user mode stack */
_estack = 0x24080000;    /* end of RAM */
/* Generate a link error if heap and stack don't fit into RAM */
_Min_Heap_Size = 0x400;      /* required amount of heap  */
_Min_Stack_Size = 0x800; /* required amount of stack */
 
/* Specify the memory areas */
MEMORY
{
DTCMRAM (xrw)      : ORIGIN = 0x20000000, LENGTH = 128K
RAM_D1 (xrw)      : ORIGIN = 0x24000000, LENGTH = 512K
RAM_D2 (xrw)      : ORIGIN = 0x30000000, LENGTH = 288K
RAM_D3 (xrw)      : ORIGIN = 0x38000000, LENGTH = 64K
ITCMRAM (xrw)      : ORIGIN = 0x00000000, LENGTH = 64K
FLASH (rx)      : ORIGIN = 0x8000000, LENGTH = 2048K
}
 
/* Define output sections */
SECTIONS
{
  /* The startup code goes first into FLASH */
  .isr_vector :
  {
    . = ALIGN(4);
    KEEP(*(.isr_vector)) /* Startup code */
    . = ALIGN(4);
  } >FLASH
 
  /* The program code and other data goes into FLASH */
  .text :
  {
    . = ALIGN(4);
    *(.text)           /* .text sections (code) */
    *(.text*)          /* .text* sections (code) */
    *(.glue_7)         /* glue arm to thumb code */
    *(.glue_7t)        /* glue thumb to arm code */
    *(.eh_frame)
 
    KEEP (*(.init))
    KEEP (*(.fini))
 
    . = ALIGN(4);
    _etext = .;        /* define a global symbols at end of code */
  } >FLASH
 
  /* Constant data goes into FLASH */
  .rodata :
  {
    . = ALIGN(4);
    *(.rodata)         /* .rodata sections (constants, strings, etc.) */
    *(.rodata*)        /* .rodata* sections (constants, strings, etc.) */
    . = ALIGN(4);
  } >FLASH
 
  .ARM.extab   : { *(.ARM.extab* .gnu.linkonce.armextab.*) } >FLASH
  .ARM : {
    __exidx_start = .;
    *(.ARM.exidx*)
    __exidx_end = .;
  } >FLASH
 
  .preinit_array     :
  {
    PROVIDE_HIDDEN (__preinit_array_start = .);
    KEEP (*(.preinit_array*))
    PROVIDE_HIDDEN (__preinit_array_end = .);
  } >FLASH
 
  .init_array :
  {
    PROVIDE_HIDDEN (__init_array_start = .);
    KEEP (*(SORT(.init_array.*)))
    KEEP (*(.init_array*))
    PROVIDE_HIDDEN (__init_array_end = .);
  } >FLASH
 
  .fini_array :
  {
    PROVIDE_HIDDEN (__fini_array_start = .);
    KEEP (*(SORT(.fini_array.*)))
    KEEP (*(.fini_array*))
    PROVIDE_HIDDEN (__fini_array_end = .);
  } >FLASH
 
  /* used by the startup to initialize data */
  _sidata = LOADADDR(.data);
 
  /* Initialized data sections goes into RAM, load LMA copy after code */
  .data :
  {
    . = ALIGN(4);
    _sdata = .;        /* create a global symbol at data start */
    *(.data)           /* .data sections */
    *(.data*)          /* .data* sections */
 
    . = ALIGN(4);
    _edata = .;        /* define a global symbol at data end */
  } >RAM_D1 AT> FLASH
 
 
  /* Uninitialized data section */
  . = ALIGN(4);
  .bss :
  {
    /* This is used by the startup in order to initialize the .bss secion */
    _sbss = .;         /* define a global symbol at bss start */
    __bss_start__ = _sbss;
    *(.bss)
    *(.bss*)
    *(COMMON)
 
    . = ALIGN(4);
    _ebss = .;         /* define a global symbol at bss end */
    __bss_end__ = _ebss;
  } >RAM_D1
 
  /* User_heap_stack section, used to check that there is enough RAM left */
  ._user_heap_stack :
  {
    . = ALIGN(8);
    PROVIDE ( end = . );
    PROVIDE ( _end = . );
    . = . + _Min_Heap_Size;
    . = . + _Min_Stack_Size;
    . = ALIGN(8);
  } >RAM_D1
 
  .lwip_sec (NOLOAD) : {
    . = ABSOLUTE(0x30040000);
    *(.RxDecripSection)
 
    . = ABSOLUTE(0x30040060);
    *(.TxDecripSection)
 
    . = ABSOLUTE(0x30040200);
    *(.RxArraySection)
  } >RAM_D2 AT> FLASH
 
  /* Remove information from the standard libraries */
  /DISCARD/ :
  {
    libc.a ( * )
    libm.a ( * )
    libgcc.a ( * )
  }
 
  .ARM.attributes 0 : { *(.ARM.attributes) }
}

[Tesla DeLorean · Accepted solution]

And what address is it *actually* using for the buffers, not just the descriptors.

Perhaps make the memory area larger, and make sure you commit any data with a fencing instruction, or explicit flush. ie __DSB, and SCB_CleanDCache_by_Addr()

[Akhil1]

Oh this was it, the Txbuffer in low_level_output was on the stack and was being cached. Adding SCB_CleanInvalidDCache() before calling HAL_ETH_Transmit(....) in ethernetif.c fixed the issue for me.

Just wondering if there is a better way to do this without touching generated files (unfortunately there is no /* USER CODE */ section in low_level_output)

[Akhil1]

NVM got it, just used the MPU to setup the stack/heap area as Write Through RW Allocate and it works

[Piranha]

For these needs __DMB() is enough. In contrast to __DSB(), it leaves the possibility for a CPU to execute non memory access instructions, while waiting for a memory access to complete.

[Piranha]

As Clive (Avogadro) already wrote, for a transmit invalidation is not necessary and clean is enough.

[Piranha]

Be aware of the fact, that output frames can come also from lwIP memory pools and even flash memory. Therefore D-cache management with SCB_CleanDCache_by_Addr() is more universal approach and gives more performance as the memory can be cached when processing it.

[Akhil1]

That makes sense but unfortunately cube doesn't have /* USER CODE */ sections in that function and we are a pretty big team so it would get messy if I modified the generated code, is there a way around this?