Hey,

I've seen sometimes message "data modified in free list... xxx != 0xdeadbeef",
and I'd like to know whether it means real memory or if swap space also
counts.  I have most probably some SCSI problems and if this could be
caused by that, I don't need to suspect my two SIMMs.

Secondly, does any Warp Engine owner know if it provides termination
power to SCSI bus if the appropriate jumper is set?  And especially,
if I remove resistors from WE, is it still supposed to provide term power?
I just measured it (pin 26) and there is no voltage if there are no
resistors (I have WE in the middle and so I had to remove resistors
from it).  I haven't yet checked what it shows if I add those resistors
back.

I just noticed this after I removed one drive (also set to provide
term power) and suddenly the A3000 didn't boot at all (but when I
set another drive to provide term power, boots succeeded again).
So, if this is not how other WE's work, I have to check mine more
accurately...


I just quickly measured that there is connection between pin 26 and that
jumper so either it doesn't provide term power without resistors (if at all)
or then the power source is cut (and if anyone knows where the power
to term line is supposed to come, I'd like to know, too).

Hmm, also relating to this WE, if I remove those resistors, do I get
really a non-terminated bus (to be terminated by an external device)
or does this chip (NCR something) terminate the line automaticly?
Hopefully not, but I just have so strange failures that I don't know
what to do ;-(
System panics, programs just die, ... At least two times those panics
have been caused by filesystem problems (if I can trust to the output
of kernel debugger), but most often it just says MMU fault.  I just
don't know if I should believe that MMU is bad, it's been working
nicely for about six months (and has had CPU fan, and is as it was
when I bought it, no over-clocking or such).  And this could perhaps
to caused by SCSI problems if swapping goes mad (although I don't
know if it could show up as an MMU fault).

One big change is that I changed from A3000's desktop case to a
tower, but it has two fans (in addition to CPU fan) and I don't
think it's hotter than that desktop case.  Of course, the motherboard
is now in vertical position, but I don't know if that could change
anything (perhaps the CPU-fan isn't able to cool CPU enough because
now all the heat doesn't go up thru the fan but some emits from the
side of CPU).  I used parts of that original case so daughterboard
is still well attached.  And these problems didn't start immediately
after I changed the cover (also ADOS works nicely...).


At first I thought that all this is caused by some bad library or
executable somewhere, because re-compiling binaries and libraries
seems to help (but takes much time because of crashes).  But then
something brokes again... now I have about everything re-installed
and still crashes (I will soon try a fresh install from tar balls
but somehow I think it will not help).

However, ADOS sees no problems (except those motherboard memory problems
mentioned several days ago, but I threw those zips away so they don't
affect anymore).


Just in case anyone has had similar problems, I have an A3000 (25MHz
model) with Warp Engine (40MHz), CV64 (2MB) and QuickNet Ethernet
card.  Three HDs (Quantum Saturn, IBM Spitfire and some Seagate which
isn't used by NetBSD) and one CD-ROM (but most of the time switched off
so it won't count).  Normally the SCSI chain is terminated by CD-ROM's
external case (active terminator) and by one of those HD's (also active
termination).  I have checked that the external end is terminated even
if the CD-ROM is off (it gets about 4.7V between pin 26 and ground).
Yesterday I put resistors back to Warp Engine (and also removed that
CD-ROM), but still problems.  All those SCSI devices are connected
to Warp's controller because motherboard's seemed to cause problems
after installing Warp Engine.


I have used the latest sources from sup.netbsd.org to make the kernel
(and most of the other software, too).  I'll try to write down what debugger
says next time I get a panic just in case it could help solving this
problem.


Thanks for reading, and hopefully someone has good ideas on what to do next
(I cannot easily remove Warp Engine because then I wouldn't have any
fast mem to use NetBSD ;-)


	Mika Kortelainen