AM625: Where is the EVM u-boot code that reads eth1 MAC from EEPROM?

Hey thank you!  This is not really easy for me to follow but at first glance I don't see anything that would indicate that the EEPROM bus and address are configurable - would you agree?

Also I tried `ethtool -e eth0` on my PhyBoard dev board (not an EVM) and got the response:  `Cannot get EEPROM data: Operation not supported`.  I assume it's because the EEPROM is not at the hardcoded bus/address that the tlan driver expects? 

Hi Thomas,

Thomas Nichols said:

Also I tried `ethtool -e eth0` on my PhyBoard dev board (not an EVM) and got the response:  `Cannot get EEPROM data: Operation not supported`.  I assume it's because the EEPROM is not at the hardcoded bus/address that the tlan driver expects? 

I believe this message is due to the fact that the SK-AM62x EVM eth0 driver (the CPSW ethernet driver) does not support eeprom access. Can you try running "ethtool -i eth0"? I expect that you should see something like the below.

root@am62xx-evm:~# ethtool -i eth0
driver: am65-cpsw-nuss
version: 6.6.32-ti-g6de6e418c80e-dirty
firmware-version: 
expansion-rom-version: 
bus-info: 8000000.ethernet
supports-statistics: yes
supports-test: no
supports-eeprom-access: no
supports-register-dump: yes
supports-priv-flags: yes 

What Linux SDK version are you using?

Thomas Nichols said:

This is not really easy for me to follow but at first glance I don't see anything that would indicate that the EEPROM bus and address are configurable - would you agree?

Here is some information I got from the developer.

• This is how U-boot parses MAC addresses from EEPROM: https://git.ti.com/cgit/ti-u-boot/ti-u-boot/tree/board/ti/am64x/evm.c?h=ti-u-boot-2024.04#n221 
• More information about how U-boot obtains the MAC addresses and how Linux fetches these addresses: https://lore.kernel.org/all/c4b91154-7a8b-4642-a642-6ae93b448115@ti.com/
  • This patch is specified for a different SoC but the process should be similar for AM625 

I will check with the developer if the tlan.c driver is actually used to fetch the MAC address from EEPROM. 

-Daolin

Hey that's great.  In fact my SOM vendor also uses the uboot ethaddr env var, but it wasn't clear to me whether setting it there caused it to be passed through to the kernel.  If I write custom u-boot board.c code to load the MAC from a different EEPROM and `setenv ethaddr` from there, that is perfect that it will solve the problem for the linux side.

Thank you.  This is the u-boot EEPROM code used by my SOM vendor: https://github.com/phytec/u-boot-phytec-ti/blob/v2024.04-10.01.10-phy/board/phytec/common/phytec_som_detection_blocks.c#L55-L88

I don't expect you folks to weigh in on vendor code, but I suspect it is highly inspired by the EVM board detect code.  In either case, I hope between the phytec code and ti-u-boot/tree/board/ti/common/board_detect.c it should provide me enough reference material if I need to implement custom logic i.e. from reading the MAC from a different EEPROM location.  Anyway you have helped me piece together how MAC address handling works in the EVM which is much appreciated!  Thanks again.

Hi Thomas, 

No problem, glad to have helped! If you have additional questions related to this topic feel free to reach out here again.

-Daolin

Hi Daolin,

I am interested in this topic as I had one doubt not cleared for long time yet.

On custom board, there is not EEPROM, customer take it like AM335x which has 2 MAC ID, so wasn't aware of not existing the 2nd MAC ID from effused register.  when use the CPSW in dual MAC mode, what happen to the second MAC address with ifconfig command? a random value of uninitialized variant? and maybe different on each power cycle?