Thanks, Marc, On Nov 15, 2012, at 11:57 PM, Marc Welz wrote:
> So this is a sound workaround. What seems to be happening is that > I2C accesses are failing (as per your previous report), so there is no > way for uboot to retrieve the serial number out of the I2C eeprom and > so derive a mac address. > > Uboot only computes a MAC address if none has been set in flash, so you > are using the same mechanism that one would use to set a > proper/purchased mac address instead of the 0x02:* locally assigned > ones. Interesting. I thought I was imagining things because after running saveenv and rebooting, the MAC address seemed to revert back to 02:44:01:02:00:00. I thought I must have forgotten to run saveenv, but now I wonder if I had run saveenv but that at least some of the i2c stuff happened to work for that boot and so the old MAC address reappeared. Does that sound plausible? Is there anything I could/should look for when trying to fix the i2c problem? If I wanted to try reading the i2c eeprom from u-boot, which i2c address (and bus?) would I use? Is there an i2c address map anywhere? Thanks, Dave
