Hauke Fath wrote:
> Hi,
> 
> a few observations from a kernel compile session this evening...
> 
> [The machine in question is a Quadra 700/33/68MB]
> 
> 
> 1) With the ADB direct-to-metal driver, the box locks up solid during the
> initial probe. This ran fine until 1.3B.

What kind of ADB devices do you have?  I'm kinda surprised that the
hardware driver would give you problems on your machine...I've got a
similarly configured Q700, and I'm running what I believe is a 1.3 kernel
(custom, hwdirect ADB support) on it.  Of course, I've got John's latest
II-series patch added to it b/c otherwise it doesn't like my Logitech
MouseMan.

> 2) For whatever reason, mac68k/dev/{adb_direct.c,pm_direct.c} link
> ADB_DEBUG unconditionally to DEBUG with the construct
> 
> #ifdef DEBUG
> #ifndef ADB_DEBUG
> #define ADB_DEBUG
> #endif
> #endif
> 
> -- whereas the other adb_* files do not. This breaks builds with DEBUG but
> not ADB_DEBUG defined -- Ick.
> 
> Fix: Don't do this, kick out the above construct.

I'm pretty sure that Scott has cleaned up a _lot_ of the debug code,
although I don't know if he's looked into the direct hardware support
files.  Both files are in serious need of some KNF-ifying, anyway, I
believe.

BTW, is there any reason why you turned on DEBUG?  There is a lot of code
turned on by this define which can be somewhat annoying......

> 3) The ROM based (MRG) ADB driver unconditionally disables the adb device
> if you boot with serial console -- no local X server. I can see situations
> where this comes in handy, but please make it an option (for the booter?).

I'm working on a major change to the ADB front-end (if Scott likes it), so
I think this will be worked around.  I think I should be more or less
finished with it in the next month or so.

Later.

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