Hello,

> > The changes for mesh, interrupts, PCI-PCI bridges ( which shouldn't 
> > concern you, as far as I know the 9500 doesn't have one ) were committed 
> > a couple of days ago, so cvs update -d -P should do the trick :)
> 
> So the PCI-PCI bridges are different than the bandits? Oh, Ok.
Yes. with two bandits all PCI slots should have different interrupts but you can't directly transfer data from the upper three slots to the lower three. With a bridge and only one bandit like in my S900 you can move data between all PCI devices but the lower 4 slots share the same IRQ.

> My drive still comes up as offline. Was this fixed, or is this a patch 
> that wasn't committed yet?

> sd0 at scsibus1 target 0 lun 0: <DEC, RZ1CB-BA (C) DEC, LYE0> disk fixed
> sd0: drive offline
I have no idea what causes this. Maybe there's some timeout set too tight...
Does the drive work?

> Who can say which patches are the best ones to test on an MP machine? Tim?
That's Tim's area, I don't have a MP machine (anymore. sadly there are no G3 boards for the 2nd CPU slot in the S900)

> Also, which patches make the serial work properly? I can still get my 
> serial console to lock. Were they committed?
The interrupt handling is (mainly) in macppc/macppc/extintr.c, the patch to make the serials and other things work properly ( for me at least, 604s seem to have problems for some weird reason ) was committed a few days ago by Allen.

have fun
Michael