Rusty Russell <rusty at rustcorp.com.au> writes: > On Mon, 2006-07-31 at 08:45 -0600, Eric W. Biederman wrote: >> I have a fundamental problem with this patch. >> Don't we enter the kernel at: >> arch/i386/boot/compressed/head.S startup_32? >> Doesn't %ebx get stomped? >> >> Otherwise I think the concept is fine, the implementation just >> won't work. > > If you want to boot compressed kernels, yes. I think we do. It would certainly simplify everyones life if we can. I think booting compressed kernels is a natural outgrowth of having a generic kernel that can boot on everything. >> Well I can tell to some extent it doesn't work as you got the >> x86 calling conventions wrong. > > Well, the init function pointer is labelled "fastcall", but as the code > is untested I can easily believe I screwed up in at least one place. Ok. I didn't think I saw that. I was looking for asmlinkage though :) > Since the Xen people seem happy to revise their boot procedure, I'm > going to drop mu proposal and let you, Zach and Jeremy sort it out from > scratch. Sounds like a plan. > I want a single entry point, and I want a new hypervisor ABI > implementation to merely have to add a new paravirt ops. Meet those, > and I'm happy... Sounds like a good criteria. Eric
