Ronald G Minnich <[EMAIL PROTECTED]> writes: > On Fri, 9 Nov 2001, Ollie Lho wrote: > > > I found that you added IDE init code to southbridge.c for SiS > > 630. Could this give me (us) the illusion that the kernel can handle > > IDE init by itself ?? > > yes. This could give that illusion. Adam Agnew asked me to add that code. > > for all other platforms we still need to add the linux patch. Sorry for > the confusion. >
I do some basic IDE init for the dual Amd, and for the p4dc6 dual Xeon. Though the dual Xeon code still needs to be checked in. My basic feeling is that this isn't really what we want to do long term but setting 2 bits that allow the kernels IDE driver to function makes it a lot easier to sleep at night. Until we can tell the kernel what we need it to do I don't expect it to do too much. Getting the kernel to do what we need it to do looks like a 2.5 thing. Also Adams demo of a polled IDE driver looks very compelling. 26K for a polled IDE driver and the bootrom for a sis900 nic. I think that writing a freestanding bootloader for linuxBIOS that would cover all of the easy cases, looks like one of the next major steps to under take. I really haven't seen any other projects that we can adapt and make work under linuxBIOS, so we probably need to take this step ourselves. A standalone bootloader does not need to be a linuxBIOS only solution, nor does it need to be an x86 only solution. So I can see it being used by other people as well. Which makes it the kind of thing that is much more likely to get developed and stay current. Jeff Garzick actually suggested this a few months ago. Thoughts? Eric
