On Thu, 3 Jun 1999, Bill Studenmund wrote:

> On Thu, 3 Jun 1999, Frederick Bruckman wrote:
> 
> > On Thu, 3 Jun 1999, T@W wrote:
> > 
> > > guy santiglia wrote:
> > > >>
> > > >> -On my LC475 it'll hang once in a while at the adb point (adb at obio0) in
> > > >> bootup (especially after having saved any new settings in the Booter
> > > >> without rebooting my Mac)
> 
> Look at the cpu_delay (sp?) factor. I bet it's different on the boots
> where the kernel hangs.

I first noticed that quite some time ago, but it's not consistent. I also
tried hacking mac68k_calibrate_delay to force a particular value. No luck.

> > causes it to hang in the same place. I found that using Apple-B
> > instead of the menu to boot would take care of that. That doesn't seem
> > to be the fault of the booter. It's not the booter that hangs...
> 
> Weird. I'd bet that what moving the mouse did was change the calculated
> cpu factor. If that factor's off, adb won't get its timings right and all
> will lose.

Moving the mouse doesn't seem to change the delay factor.

It's my suspicion that adb_hw_setup() doesn't really zap the bus the way
it's supposed to. (There's a telling comment in the ADB_HW_CUDA case...)
As long as the bus is fairly quiet, the code that follows that call in
adb_reinit() still works, but there are no guarantees. That would explain
a lot of the weirdness, including the fact that some kernels boot exactly
half the time--at random!