Ronald G Minnich <[EMAIL PROTECTED]> writes: > several things I noticed with SMP kernels and SMP linuxbios > 1) linuxbios always notes that CPU #1 is initialized. It never mentions > CPU #0. With two CPUs in it never mentions CPU #2. It should either > mention CPU#0 or CPU#2, I think it should be mentioning CPU#0 and #1. > So something about setup in SMP is not quite there. It looks like > either both CPUs think they are CPU #1, or only one CPU is getting > set up. > > 2) In linux, with SMP, with one CPU, it acts like it has two cpus. > it mentions CPU #0 and #1, and it seems to think that the only CPU > is CPU #1. is it maybe trying to send an IPI from #1 to #0? > > 3) with two CPUs, it seems to think that BOTH CPUs are CPU #1, and > even gets an error concerning CPU #0, and ends up thinking it > has taken it offline. > > This is starting to look like a pretty simple init problem. it almost > seems that whichever CPU should be #0 is somehow getting set as #1 in > linuxbios, and then when there are two CPUs, linuxbios does something > wrong. > > I think this is close ...
Certainly. I'll check my tree shortly and see which SMP bit I left out. I frequently get CPU#1 going first. But I am also seeing CPU#0 being initialized. Though the 1 cpu case sounds interesting... My two working theories. a) I forgot to merge something into the tree. b) I have a the one good L440GX board. So after I verify that we are running the same code, I go on hunting assuming that my hardware is working better. Eric
