>>>>> "murray" == Murray J Root <[EMAIL PROTECTED]> writes:
murray> When I first boot I get a SMP error and a complaint about NULL address murray> for APIC that looks scary to a newbie - I tested with a few newbie friends. murray> That really needs to not appear for a UP machine, which all P4s are, murray> since P4 isn't built for SMP. murray> I do hope that's only a cooker thing, and not visible to real users. Problem is: - some machines need APIC tables to boot (i.e. they don't setup well the pirq table and normal setup will not work). - other machines have wrong APIC implementation -> i.e. it will not work with ACPI. 1st group use to be new machines (and almost all new VIA boards with Promise controller on board as tertiary/cuaternary IDE) 2nd group used to be older machines (ok, new dell laptops also has this problem). Reasons for the change: - APIC is better than PIC (that is technical argument). - APIC works in the mayority of the machines (this is not determinant, either 1st & 2nd group are small for what I know). - new machines are more probably to _need_ APIC. To fix this problem, in -21mdk (released 1h ago), there is a patch what makes that if you boot with the command line option noapic, it means that you don't want apic (vanilla 2.4 will try _anyways_ to use the APIC if compiled in). I am trying to fix the kernel to detect the machines where APIC don't work and put them not try APIC, problem so far is that APIC is initialized too soon to be able to read dmi at that point. I will like that people that are having problems with APIC that send me a mail indicating: what is your machine. boot messages of failed init (if possible) the output of this program, to be able to detect the BIOS http://people.redhat.com/arjanv/dmidecode.c report of people who need that feature are also welcome to help convince the people for who it don't work that this is needed. Hope this helps. Later, Juan. -- In theory, practice and theory are the same, but in practice they are different -- Larry McVoy
