On 7/5/06, Peter Seebach <seebs@plethora.net> wrote:
> In message <e8enpv$b26$1@sea.gmane.org>, Christos Zoulas writes:
> >I've been forced in the past to use current in production because I needed
> >features that only existed in current at the time. The short answer is that
> >I tested it, and if it worked for me, I used it. For example, this has been
> >running for a while:
>
> Encouraging.
>
> FWIW, 3.99.21 won't go MP on my hardware.  I have a beautifully and perfectly
> jinxed computer.  It has two processors, SCSI, and firewire.  Any two will
> work.  FreeBSD can do firewire all the time, and either two processors without
> the SCSI controller, or one processor with.  (Turning off ACPI, which is the
> only way to make SCSI work, prevents it from going MP.)  NetBSD can do SCSI
> and MP in 3.0, but not FireWire.  3.99.21 doesn't do MP, and also seems to be
> unable to actually attach a disk in firewire.  Obviously, MP works in general
> in -current, so my most likely option is to build -current and try to debug
> things there, because at least it SEES the second processor, and I can
> probably find some obvious explanation for "cpu1: not started".
>
> *sigh*  At least it's not Linux.

I think -current needs ACPI for MP as well.
Also, I wonder what the differences are between current's firewire and
3.0's.  Maybe they could be easily back-ported?