PROCESSOR-SDK-AM62X: Ethernet MAC address

Hello Cyra,

The subject owner is on vacation until the middle of next week. Please ping the thread on Wednesday of next week to make sure we get back to you.

Regards,

Nick

Is there any update on this?

Is there any update on this?

Hi,

I can only offer an overview of how TI sets the MAC addresses as we do not have written instructions on how to change the MAC IDs. You could implement what TI has done.

This method will require a I2C EEPROM that stores the MAC IDs. The TI EVM schematics show how the I2C EEPROM is attached.

• Currently TI uses U-Boot to pass the MAC addresses to the kernel. You could use the DTS file but requires a unique DTB for each board.
• On the TI EVM there are additional MAC addresses stored in the EEPROM and read by u-boot to set the ethaddr and ethaddr1 u-boot environment variables. These variables are what get passed to the kernel on boot: 
  • This is a link to the u-boot source code on how the eeproms read the info from the I2C:
    • https://git.ti.com/cgit/ti-u-boot/ti-u-boot/tree/board/ti/common/board_detect.c
    • Look at this function: board_ti_set_ethaddr(int index)
• Using the prebuilt SDK on a TI SKEVM the ethaddr and ethaddr1 variables that are passed to the kernel can be viewed by performing the printenv command when in u-boot. The two variables ethaddr and ethaddr1 will be displayed. 

The driver selection for the PHY is done in the kernel menuconfig config. This will bring in the driver. The board dts will need to setup the compatible field for PHY node used by the CPSW node. If the driver is configured in the kernel config there will be a driver sign on message in the boot log indicating that the driver was loaded when the kernel identifies the PHY.

Best Regards,

Schuyler

Hi,

Thank you for the update. I was wondering if there is a possibility to e-fuse the MAC address instead. It would be more efficient than creating multiple DTB files.

Has there been any update on this?

Hi,

I agree on the DTB issue which is probably the main reason why TI currently uses an eeprom to store the MAC addresses for the TI EVMs. Currently there is not customer option to program efuses for MAC addresses on TI Sitara devices. I can try to suggest that for future devices. 

Best Regards,

Schuyler

May I know why is it not possible to use e-fuse on AM62? Is it because or permission from customer side or something?

I wonder why the technical reference of the TI AM62x stated that the default MAC address is from Efuse? is this documentation error if we can't refuse the mac address on TI device?

Is there any update on this for Yen Mei Goh?

Dear Schuyler,

I found an example here: GitHub - u-boot/u-boot: "Das U-Boot" Source Tree

u-boot/eeprom.c at master · u-boot/u-boot · GitHub

But for this option, It will need the image to be biuld for every device and everytime it needs to be changed. My end customer is looking into an option that doesn't need to build the image multiple time. any suggestion how this can be done???

Best Regards,

Felix

yen mei goh said:

I wonder why the technical reference of the TI AM62x stated that the default MAC address is from Efuse? is this documentation error if we can't refuse the mac address on TI device?

We now have the instructions for programming the OTP efuses for the MAC address at https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1235880/faq-how-to-update-ids-usb-pcie-pid-vid-mac-id-by-programming-extended-otp-on-am6x .

The eeprom based approach will also work as well.

  Pekka