Benoit MARTEL wrote:
> 
> Until this week (and since before 1.2), my IIvx had never powered itself 
> off. I just waited for the shutdown to tell me it was safe to power it 
> off and pressed the power button at the back.
> 
> I faintly remember a discussion about it on this list but it was long 
> ago. What I remember (quite possibly incorrectly) from it was that some 
> machines tend to be able to power-off or not from time to time maybe 
> because their inability to do so resulted from a kernel-size related bug.
> End of vague memories.

Ah, well, the part you're forgetting (and the part that is different from
your situation is that the machines which tend not to be able to power-off
or restart due to kernel size also tend to hang when you try to do one of
these).
 
> A few days ago, I compiled a custom kernel and removed lots of unneeded 
> options. I even decided to give HWDIRECT (commenting out MRG_ADB) a try. 
> The machine now powers-off!
> 
> Incidentally, I started peeking around the source lately to familiarize 
> myself with it. I stumbled upon cpu_reboot and it seems to me that wether 
> my machine powers-off or not is simply a matter of wether or not I use 
> MRG_ADB.

This is correct.  According to the source, this situation also applies to
the following machines:

 IIsi
 IIvi
 Performa 600
 IIvx 
 Color Classic
 Color Classic II
 LC 520
 Quadra 840AV
 LC 550, Performa 550 
 LC 575, Performa 575/577/578
 LC 630, Performa 630, Quadra 630

> Are my memories completely off (damn, another bad Brain-SIMM)?

As I said above, not completely off ;-)

> Is there still a recurring problem with powering-off (or was there ever 
> one) or is it really only a question of not using MRG_ADB?

I believe the powering off problem is still present on some machines (I
think the IIcx is the machine hardest hit by this).

In your case, there is currently no MRG-based support for accessing the
necessary part of the ADB/PRAM hardware which controls powering off the
machine.  The MRG ADB support only knows how to do softpower-off using the
VIA chips, which limits it to machines with Mac II-class hardware.  If I
could figure out what the command looks like which powers off the machine,
maybe I could fix it, but I haven't dug around in John's code enough, yet.

I hope this explains things a bit.

-- 
Colin Wood                                 cwood@ichips.intel.com
Component Design Engineer - MD6                 Intel Corporation
-----------------------------------------------------------------
I speak only on my own behalf, not for my employer.