>> The assumption was that not much needs to happen before the SPD I2C >> bus is accessible by the CPU - is that valid? > > Since you need I2C, you need to get parts of the south bridge working. > So, its basically the same amount of work as with the USB debug port.
Most systems I work with have I2C controllers (and typically more than one) in the north bridge, not somewhere south (which is an insane idea). > Question is: Will we see systems without USB (debug port) but with > DDR2 > sockets? Yes. They exist already (mostly in the embedded world, of course). >> If not, what IS easily accessible besides the boot ROM? (Which we >> don't want to rely on since we don't know exactly what it will speak >> when in the future.) We just need one bit that can do kHz signalling. >> SMI#? > > Boot rom is a good start I bet. That will always be there (as long > as we > all do firmware development at least). > > It might be parallel yesterday, LPC today and SPI tomorrow, but that > is only the interface it connects to. Different connector, different > VHDL source and we should be fine, no? SPI buses as implemented on most x86 systems for flash, do _not_ support attaching extra devices on the bus (you'd need a separate chip-select signal, and those either don't exist or aren't readily available on the board). > If LPC goes away, we need to do SPI. Ok. If we get DDR3, we would have > to do that. Technical standards come and go,.. I'd prefer a solution > that is so cheap that I dont mind throwing the whole kit away after > doing a port or two, rather than trying to create something that is > good > forever. The "SPD" solution works over all families of DIMMs. >>> I think that we are going into a world where we have to figure out >>> usb debug port. >> >> It's certainly one good debugging option but maybe not the only good >> one. > > Especially it does not help for reflashing. Finding something nifty > here > would be nice. JTAG into the north bridge, and you can init all system buses, done ;-) On some systems you don't even need such init, the PCI starts running -- so you can DMA from a plug-in card to the flash chip. Again though: none of this is suitable for an "end-user", someone not too hardware savvy to help getting LinuxBIOS going on his board (or to restore flash after an "accident", etc). Segher -- linuxbios mailing list [email protected] http://www.openbios.org/mailman/listinfo/linuxbios
