On Sat, 4 Jan 1997, Herb Peyerl wrote:

>  > btw: do you know where to get the sources for the 4.4BSD libm.so.x.x or
>  > any other 6888x libm ?
> 
> Actually, I think scott reynolds (scottr@netbsd.org, the mac port maintainer)
> was trying to fix an fpsp bug... I don't know if that's what you're looking
> for... You might contact him...
I'll try...
for now: ( to return to the topic ) I tried to hunt down the
XhpBSD-flicker-bug using gdb and a -g in certain Makefiles... the results:
indeed, the crashes occur when some clients try to produce stippled
rectangles or polygones... one prine candidate to crash is
tcPolyFillStippledRect in mit/server/ddx/hpbsd/topcat/tcPoly???.c and
there seem to be some part of the given GC struc incorrect/missing since
this function tries to get a privGC from the devPrivates (or so ) at the 
end of the GC which results in something really strange, since some
pointers in it appear to be NULL what results in a page fault that causes
the crash.
Then: there is a assemler routine that should produce a stipple-pattern
with a notice attached saying that future gcc's ( they used 2.3 ) may
miscompile that for some reason, but I'm really not familiar with 68k
assembler language so I can't fix it... 

further questions: are there differences how certain data types are
defined under 4.4BSD and NetBSD ? is int a 32bit signed under 4.4 ? I
can't think of other ways how these sources produce a working binary under
4.4BSD and a crashing one under NetBSD...

bye
Michael