>>The LC-II and LC-III were _NOT_ supported by the GENERIC Kernels in 1.2
>>(or 1.2.1 as I recall).  Support for these went into the source tree soon
>>after the code freeze though, and any GENERIC Kernel built nowdays will
>>support these machines.
>
>Not exactly....the only generic kernel my LCIII works with...is the
>GENERICSBC-26, I've tried 27, 31 and 32 I believe...and most of the non-sbc
>kernels and none of them, except SBC-26 works. i believe this is also the
>case with the LCII (My friend's got one). It could be my problem but I
>doubt it very much. ADBTEST #160 works also :)
>
>just to tell you :)

Jens,

Couldn't let this one pass ;-)

If you may recall, the original poster reported a TRAP error during ADB 
probe using the GENERIC 1.2 Kernel.  This is to be expected since the MRG 
code used in the 1.2 GENERIC kernels did not support the LC-II and 
LC-III.  Scott Reynolds used the ROM Sniffer to scope out the LC-III and 
I did the same for an LC-II user at about the same time.  We exchanged 
information and Scott incorporated the ROM Table changes into the source 
tree following the official 1.2 release.  Very little has changed in the 
MRG code since then and any GENERIC Kernel built with these changes 
should get past the ADB probe during boot.  That's not to say that all of 
these Kernels will complete booting or will successfully run on these 
systems as you've already noticed.

The issue of GENERIC vs. GENERICSBC is determined by the type of disk 
used on the system, and has nothing to do with the model Mac which NetBSD 
is being run on.  The general rule of thumb is to try Kernels with the 
standard SCSI drive first and only use the SBC supported kernels if disk 
errors or corruption is encountered.  For the most part LC-II and LC-III 
users who are using Apple or Quantum disk drives will probably be better 
off using the SBC kernels, but that's not an absolute rule.

There are other things that affect the usability of any one Kernel on a 
user's system. Remember, this is a development effort and most of the 
Kernels for the Mac are based on variations of the -current code at 
differing stages.  At times various things become "broken" during 
development that affects one or more models of the Mac. These include 
video mapping, memory mapping, Booter version, ethernet support, video 
card support, etc.  Thankfully this is becoming less and less of a 
problem as more of the code has stabilized.

The ADBTEST and HWDIRECT Kernels also have a sub-set of Mac systems that 
they support. It's best to check the Web pages on puma for a list of 
which Mac models are best supported by the various HWDIRECT kernels.  I'm 
not sure if there's also information on which SCSI driver is used in 
these kernels, but that could be extremely important for LC-II and LC-III 
users.

Finally, the LC-II and the LC-III are sometimes difficult to track due to 
the various models that were built around the basic Motherboard.  It 
seems that there are some LC-II variants that were sold in Europe that 
don't seem to conform to the requirments expected by NetBSD, although I 
was never able to find any variation that caused ADB probe problems in 
the MRG code.  That is not the case with the Quadra 630 family which has 
caused problems with the MRG code that have not been resolved, and 
probably won't be given the tremendous success we've seen in the HWDIRECT 
solution to date.  In fact, I believe the only systems that are currently 
supported by MRG code and not fully functional under the HWDIRECT code 
are some LC variants using the "Cuda" chip, like the Performa-550.

-bob