Re: [Ltib] Ltib - How to make the MAC address in EEPROM works with kernel 2.6.34 just like kernel 2.6.27?

[Li Yuan-Lung]

Dear Kevin,

Thanks for your kindly reply. I have download it.

assuming that u-boot reads the MAC address from the EEPROM and is setup to pass it to the kernel using ther ‘ethaddr=$ethaddr’ parameter.

I think this functionality approach my guessing. 

Currently the MAC address need to be input manually like "ethaddr=xx:xx:xx:xx:xx:xx" in u-boot. In the later release the u-boot will have the ability to get MAC from EEPROM and automatically generates the string into the kernel parameter(just like "ethaddr=$ethaddr" in the assuming) passed for the kernel.

Am I right with the new u-boot for kernel 2.6.34?

If the above is true, the only one method for MAC address assigning is to use "maddr" in S1L stage, right?

(Seems the CDL v2.0 has a lot of change, I am still reviewing it.)

Thank you very much!

Henry

2010/7/7 Kevin Wells <address@hidden>

 

Hi Henry, Mike,

 

Try ftp://ftp.lpclinux.com/lpc32x0/cdl/ for the latest CDL.

 

I’ll chime in on the decision to remove the use of EEPROM information from later versions of S1L, u-boot, and the kernel.

 

>Maybe the kernel 2.6.34 is just in the beta test stage, the EEPROM is just temporarily empty for developing?

It’s not really for debugging, but there is nothing current in the EEPROM that is needed for the Linux kernel….assuming that u-boot reads the MAC address from the EEPROM and is setup to pass it to the kernel using ther ‘ethaddr=$ethaddr’ parameter.

 

>They also removed writing MAC adress, SDRAM memory configuration information, board hardware id to the EEPROM in the latest S1L version (CDL library v2 at the phytec ftp).
>My question is, how u-boot will know what SDRAM memory is installed on the board?

From what I know, only 64M boards are available, so the use of the memory sizing field was removed. It’s still in the EEPROM from the factory though (I don’t have any new boards to verify this), but I don’t think you’ll find anyone with a non-64M board. (???) Pre-v2.0 versions of S1L populated this data with defaults if it didn’t exist, but Phytec seemed to do good making sure each board had the correct information and a unique MAC address.

 

>Below is my question:

>1) Why the mechanism chaged? Is there any consideration or limitation to make kernel 2.6.34 with EEPROM?

This was changed in 2.6.34 to be similar to the other board’s approach for handling the MAC address of the Ethernet driver. The serial EEPROM read support in the PHY3250 2.6.27.8 kernel version was sort of a hack, I don’t think I would have been able to release that code into the mainline, so I changed/removed it.

 

The EA3250 and FDI3250 boards do not ship with a MAC address, so you need to assign a reasonable address to It manually in u-boot. The PHY3250 version of u-boot *will* read the serial EEPROM and use the MAC address in it if it exists and you can pass this queried address to the kernel using the ‘ethaddr=$ethaddr’ argument. Or you can still set up any address you want by changing the ethaddr environment variable or kernel command line.

 

So for the 2.6.34 release, the ethernet address must be passed on the command line from u-boot. You can use the address on the board sticker(PHY3250) or make one up that won’t interfere with your network.

 

>2) Could I make the mechanism in 2.6.34 become the same with kernel 2.6.27? How?

Yes, this can be done by moving the SPI EEPROM read function in the 2.6.27.8 board_phy3250.c board file into the 2.6.34 phy3250.c board file.  The ethernet driver (lpc_eth.c) will also need some way to get the address specific for the board.

 

I hope this helps,

Kevin

 

 

 

Subject: Re: [Ltib] Ltib - How to make the MAC address in EEPROM works with kernel 2.6.34 just like kernel 2.6.27?

 

Dear Mike,

Well, the latest CDL I got is version 1.22.

Below is the source link:

http://ics.nxp.com/support/software/lpc32xx.cdl.drivers/

 

You said that they have the v2 released, but I think I do not have the access right to the ftp. It's hard for me to find the latest release. (if the above link is not the latest one. :-( )

 

My question is just like you. I though it doesn't make sense to remove such important information from EEPROM. So I just want to know the thought of the NXP developers. Maybe the kernel 2.6.34 is just in the beta test stage, the EEPROM is just temporarily empty for developing?

 

Henry

2010/7/6 Mike RA9FTM <address@hidden>

Hi, Henry

They also removed writing MAC adress, SDRAM memory configuration information, board hardware id to the EEPROM in the latest S1L version (CDL library v2 at the phytec ftp).

My question is, how u-boot will know what SDRAM memory is installed on the board?

--
73! Mike RA9FTM

2010/7/6 Li Yuan-Lung <henry.li@climax.com.tw>

Yes,

Sorry to forget tell you that it is LPC3250 platform.

 

Henry

2010/7/6 Mike RA9FTM <address@hidden>

 

Hi

 

Is your message about lpc3250? 

--
73! Mike RA9FTM

2010/7/6 Li Yuan-Lung <henry.li@climax.com.tw>

Dear NXP guys,

As a new issue, I create a new topic for this problem.

 

I have knew the facts as below:

1. The MAC address was read from EEPROM in kernel 2.6.27. 

2. The MAC address was passed by u-boot through kernel parameter in kernel 2.6.34. 

 

Below is my question:

1) Why the mechanism chaged? Is there any consideration or limitation to make kernel 2.6.34 with EEPROM?

2) Could I make the mechanism in 2.6.34 become the same with kernel 2.6.27? How?

 

Henry

 

_______________________________________________
LTIB home page: http://ltib.org

Ltib mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/ltib

 

--
Best Regards,

Henry Li
SW Engineer of Climax Co.
TEL: +886-2-2794-0001 ext.231
FAX: +886-2-2792-6618

 

--
Best Regards,

Henry Li
SW Engineer of Climax Co.
TEL: +886-2-2794-0001 ext.231
FAX: +886-2-2792-6618

--
Best Regards,

Henry Li
SW Engineer of Climax Co.
TEL: +886-2-2794-0001 ext.231
FAX: +886-2-2792-6618