Subject: XhpBSD once again
To: NetBSD Mailinglist <port-hp300@NetBSD.ORG>
From: Zadok <ml@rz.uni-potsdam.de>
List: port-hp300
Date: 01/12/1997 17:08:03
Hi there...
I still hunt the "flicker-bug" in XhpBSD and came to this: in the sources
they aligned glyph operations to 16 bit boundaries and used padding to
full 32 bit. I changed this to full 32 bit alignment without any padding,
had to do some really minor adjustments and - the server didn't crash
anymore when the openwindows clock tried to draw it's hands, the hands
don't appear at all... but for instance the cmdtool which triggered the
crash too worked fine... without any visible changes in it's appearance,
anything that worked on the buggy server still does so but the things that
caused the crash don't get drawn anymore, I suspect the evil in the
mit/server/ddx/hpbsd/cfb directory where stipple-patterns were created. So
I think the not crashing but feature-lacking server for 4.4BSD was
compiled for a plain 68000 ( because of this the 16 bit alignment ) and I
still suspect that some data types ( still int ) were defined differently
on 4.4BSD. could anyone point me some incompatibilities between 4.4BSD and
NetBSD ? what does gcc different when configured for a 68000 instead of a
68020 or better ? 

bye
Michael