Subject: NetBSD-1.3.3 observations and questions
To: port-mac68k <port-mac68k@netbsd.org>
From: Bob Nestor <rnestor@metronet.com>
List: port-mac68k
Date: 01/01/1999 20:02:04
Since I haven't had any success in getting -current up and running on my 
system as my system seems to be overly sensitive to the -current bug, so 
I thought I'd at least try moving from 1.3.2 to 1.3.3.  My efforts 
weren't as successful as I'd hoped for and I'm curious if others on this 
list have seen some of the same things I have.

First, a question about the Xserver.  I've been running Ken's OSFA color 
Xserver for some time now and was a little surprised that it, or 
something similar, isn't the default for the 1.3.3 Release.  It took a 
little digging before I realized my X startup problem was in trying to 
start a B&W Xserver on a system booted for color.  Is there any 
good/technical/political reason why we shouldn't be using Ken's OSFA 
Xserver as the default for new installations?  It does seem to work for 
all internal video and I believe it works for the SLOTMAN kernels too.

Second, the 1.3.3 system seems a little less stable than 1.3.2.  I don't 
have any hard and fast figures, but lots of little things seemed to be a 
lot more crash prone.  For instance, doing an fsck on some filesystems 
would almost always produce a kernel panic.  (Yes, I have multiple disks 
with multiple NetBSD partitions, no I don't have more than two NetBSD 
partitions on any one disk.)  Thankfully fsck under 1.3.2 was always 
successful in recovering from any damage caused by 1.3.3, and it never 
caused a kernel panic.

Third, and this one is strange, I can generate an MMU fault and kernel 
panic by just cycling power on my modem!  This is reproducable and I only 
discovered it because my ppp configuation files from 1.3.2 don't seem to 
work under 1.3.3. (That's a minor problem that even I should be able to 
fix, but the kernel panic from power cycling the modem isn't.)

Has anyone else experienced similar problems under 1.3.3?  I know 
-current is still having some problems on some systems like mine, but I'm 
curious if it also exhibits the kernel panic on modem power cycling.  
Maybe this problem I'm seeing in 1.3.3 on my system is the root cause of 
the problem in -current.  If so it might be easier to find by looking at 
the changes that went in between 1.3.2 and 1.3.3.  I'd also be happy to 
do any testing I can on my system to help find these problems.

thanks,
-bob