On 2006-11-06, Laurie Gellatly <[EMAIL PROTECTED]> wrote: > We too will have a form of Redboot stored in flash in the > final product though it won't have ethernet support. We will > have to make each MAC different though and I was expecting > this to be a process in the manufacturing stage of loading the > image. > > Somehow each image will have to be modified to make its MAC > different
That sounds nasty. We load all devices with identical images, then there's a test-station that sends a command to the unit via Ethernet to program its MAC address. On existing products, there's a small serial EEPROM where our verions of RedBoot stores that MAC address. > OR somthing has to program the Ethernet chip duing production > with the desired unique MAC. Our Ethernet chip's MAC registers > are memory addressable so the later should be easy for us. I've never seen an Ethernet chip with a non-volatile MAC. On all of the products I've ever worked on, software had to read the MAC from some non-volatile memory location and program the Ethernet chip on startup. What strikes me as odd is that eCos Ethernet drivers don't seem to have a way for application code to get that MAC address to them. Ethernet drivers appear to be expected to "just know" what MAC to write to the Ethernet controller during initialization. There's an API to _change_ the MAC address after the driver has initialized the interface, but there's no way to tell the driver what the MAC is so that the interface is initially started with the proper MAC address. AFAICT. I think. > Some part of the process has to be unique for each device - I > guess the real question will be which will be the most > convenient? -- Grant Edwards [EMAIL PROTECTED] -- Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss
