"Nick French" <[EMAIL PROTECTED]> writes: > All we want is to load a kernel off an IDE drive why do things get so > complicated. I think it would be fair to say we just need a basic way to > read a kernel image of a hard disk due to the fact it wont fit in flash. If > we over engineer this we miss the point that what we want is to be able to > boot from flash but until this is a reality (bigger standard flash) we need > a short fix to get the image off an IDE drive, am I wrong? If we put too > much work into this and find that flash parts have become the right size we > will all be going back to flash. > > It just seemed to me if we had a bootable harddisk with LILO could we not > use the same method as LILO to get the image.
As long as we don't use the same code base we can. > I know I am repeating myself but the reason we are doing this is because the > kernel wont fit in the flash. > > I hope I am not treading on anyone's toes with what I have said All we have to do is put an ELF header in the first couple of sectors, so that we can scan for it (any fixed location will collied with someones partition table scheme). And have the ELF header specify where the rest of the ELF image is. This is equivalent to what lilo does. And it only takes an extra for loop over the normal elfboot code. Eric
