[Mat_as] Giovannini wrote:
> I tried the current snapshot kernel with my Q800. Below is my report on
> it to Mr. Steve Allen.
> In summary:
> 
> GENERIC-3 hangs on ADB probe
> GENERICSBC-3 (w/debug) doesn't respond to key presses
> GENERICMGR-3 works fine, better even that GENERIC 1.4 release.
> 
> I'd be glad to try anything short of recompiling a kernel (I don't have
> enough disk space to do it, sadly).

hopefully, the next kernel will work a little bit better.  scott recently
(like in the past 2 or 3 days) committed some changes that have made the
hwdirect ADB driver a little more stable (at least on an 840av and a IIci
;-)

> The ADB probe took quite some time (around 7 seconds) to find two
> targets: keyboard and mouse. The boot process was considerably slower
> than with the stock 1.4 kernel

really?  strange.  like i said, tho, the next set that steve makes will
probably work a lot better.

> and savecore failed saying that it
> couldn't find some device. All services started normally, but at the
> login prompt nothing that I typed appeared. I telnetted and tried to do
> a dmesg to get the messages but dmesg says: "magic number incorrect". I
> attribute it to some out-of-synchronization between kernel and base packages.

sounds like your kernel wasn't named /netbsd to me...

> MRG works. dmesg and savecore complain again, but I can use the keyboard
> just fine (I don't know about the mouse, as I don't have X installed).
> What's more, using the stock kernel, it boots with the keyboard set to
> caps lock ON initially (so when I first type a login, I get ALL CAPS),
> but the MRG kernel sets the keyboard to downcase.

that's really strange.  if the next round of kernels (assuming the sources
are from the 10th or later) doesn't fix this, please let me know.

> The "adb0 at obio0"
> line backspaces a little and overwrites the "(mrg)" identifier.

that's a long-standing problem that results from a hack in the MRG
support.  i recently saw a PR which addresses the compiler warning that
this hack was meant to fix (i.e. printf() with a null argument causes the
compiler to complain, so we're currently using backspaces), so once that's
changed, perhaps we can fix this little annoyance.
 
later.

colin