[comp.unix.microport] lots of panics in uPort V/386?

det@hawkmoon.UUCP (Derek E. Terveer) (02/18/88)

Well, well, well -- comp.unix.microport is now unmoderated and root@<somewhere>
said to "post your heads off", so i thought that i would interject a question
to the assembled heads...

I have recently purchased an ACER 1100 80386, 71M drive (miniscribe), 2.6M of
memory (intel ps/286 above board) and am running 3.0-L2.2.  The problem is that
i seem to be having lots of (relatively) unexplainable panics and i was
wondering if anyone else with the 386 version was also having numbers of these
panics, like "kernel mode traps" (type e), "user mode traps" (type 2 and 8),
and "iupdat - iaddress >2^24" panics.  Plus i keep getting a number of "NMI in
system mode" messages.  

Has anyone seen these before -- i seem to be getting about one every day or
every other day, and some of them are bad enough that i have to reload portions
of the distribution.  For example, the other day i had vi poof into dust (most
annoying) and also the active file for News (even more annoying!).  I'm not
sure if *I'm* doing something wrong or what????  I don't think i'm doing any-
thing really bogus..  Most of these panics seem to happen while i'm compiling
stuff, but i'm not sure if i was also receiving large news files from my news
feed.  I've heard stories about serial ports in the 286 version, and i had
hoped that the rewrite (?? is this true?) of uPort for the 386 version had
cured these reported problems.

any help would be a help!
derek

karl@ddsw1.UUCP (Karl Denninger) (02/28/88)

In article <115@hawkmoon.UUCP> det@hawkmoon.UUCP (Derek E. Terveer) writes:
>Well, well, well -- comp.unix.microport is now unmoderated and root@<somewhere>
>said to "post your heads off", so i thought that i would interject a question
>to the assembled heads...
>
>I have recently purchased an ACER 1100 80386, 71M drive (miniscribe), 2.6M of
>memory (intel ps/286 above board) and am running 3.0-L2.2.  The problem is that
>i seem to be having lots of (relatively) unexplainable panics and i was
>wondering if anyone else with the 386 version was also having numbers of these
>panics, like "kernel mode traps" (type e), "user mode traps" (type 2 and 8),
>and "iupdat - iaddress >2^24" panics.  Plus i keep getting a number of "NMI in
>system mode" messages.  

The key one is the NMI message.

This can only be generated one way -- if your memory board(s) generate a
parity fault.  

16-bit boards have major problems in many of the 80386 systems.  Seems as
though some manufacturers didn't bother to insure DMA compatibility in the
extended memory region on 16-bit boards..... you can guess the result.

If you *do* have memory problems, the system will be flakey no matter what
you do.  It's quite possible that the only possible cure will be to replace
the 16-bit memory with 32-bit.....

Also note that the PS/286 above-board is not certified for operation 
at > 8Mhz on the bus.  Since I don't know the speed that the ACER clocks at
on the bus (note: CPU speed is not relavent here, it's BUS speed) I can't
speculate if the PS/286 card is simply incompatible with your I/O bus, or if
your '386 is one of the ones with the DMA problems.

If you can, try running with all 32-bit memory (even if you don't have a lot
-- slow is better than crashing!)

----
Karl Denninger		       |  Data: +1 312 566-8912
Macro Computer Solutions, Inc. | Voice: +1 312 566-8910
...ihnp4!ddsw1!karl	       | "Quality solutions for work or play"

root@elric.UUCP (root) (03/04/88)

In article <829@ddsw1.UUCP>, karl@ddsw1.UUCP (Karl Denninger) writes:
> In article <115@hawkmoon.UUCP> det@hawkmoon.UUCP (Derek E. Terveer) writes:
> >I have recently purchased an ACER 1100 80386, 71M drive (miniscribe), 2.6M of
> >memory (intel ps/286 above board) and am running 3.0-L2.2.  The problem is that
> >i seem to be having lots of (relatively) unexplainable panics and i was
> >wondering if anyone else with the 386 version was also having numbers of these
> >panics, like "kernel mode traps" (type e), "user mode traps" (type 2 and 8),
> >and "iupdat - iaddress >2^24" panics.  Plus i keep getting a number of "NMI in
> >system mode" messages.  
>
> The key one is the NMI message.
> This can only be generated one way -- if your memory board(s) generate a
> parity fault.  
>
> 16-bit boards have major problems in many of the 80386 systems.  Seems as
> though some manufacturers didn't bother to insure DMA compatibility in the
> extended memory region on 16-bit boards..... you can guess the result.
> If you *do* have memory problems, the system will be flakey no matter what
> you do.  It's quite possible that the only possible cure will be to replace
> the 16-bit memory with 32-bit.....
>
> Also note that the PS/286 above-board is not certified for operation 
> at > 8Mhz on the bus.  Since I don't know the speed that the ACER clocks at
> on the bus (note: CPU speed is not relavent here, it's BUS speed) I can't
> speculate if the PS/286 card is simply incompatible with your I/O bus, or if
> your '386 is one of the ones with the DMA problems.

I called up acer before i purchased the board and asked them at what speed their
bus operated, and they replied 8Mhz, so it *should* be compatible as far as the
bus speed.  My system eventually became so unstable that it *really* trashed
the file system badly enough that it was all i could do to do an emergency bug
out and try and save as many of my files as possible (silly me, i hadn't done a
backup of my stuff within the few weeks of initially bringing up my system).  I
couldn't access the floppy with anything except tar -- and that would hang the
system after it had finished writing the last file.  (Actually, later i found
out that of the 5 floppies i tried to restore, only one had a few recoverable
files on it,,,, sigh)

So i reinstalled the entire system and the system crashed right away after
installing the run-time portion.  Right there i decided that it was *NOT* me or
uport that was having a problem, but it just *had* to be either a configuration
problem or hardware -- i've heard a lot of bad things about uport (but i bought
them anyway! (:-) and i've heard somewhat less good things too, but i simply
couldn't believe that they would release a system that would crash during
installation.  So i called up uport and explained the situation.  I got a
somewhat helpful person who thought it was a memory problem and said he really
couldn't help me beyond that.  

What *I* did was re-run the install program for the above board and disable the
extra parallel (i can never spell that word correctly!) and the serial port
(which is what the "PS" stands for in the PS/286).  This seemed to help
tremendously.  I guess the i/o addresses that i had selected during initial
installation had conflicted in some way with uport.  I had run many tests
(>60) the first time i installed and then ran about 15 more the second time and
none of them came up with any errors.  But i don't know exactly what they test.
Perhaps they don't adequately test the parity error you're talking about.
Anyway, i was able to run for 3 or 4 days in a row doing some extensive
compiling, etc (reinstalling all my lost software, like news!) without crashing.

Unfortunately, whenever i reboot i *still* get a single NMI message within
about 2 hours of the reboot (most often within 5 minutes, but it varies).  And
i have crashed twice since then as well...  So it looks like i still have a
problem.

> If you can, try running with all 32-bit memory (even if you don't have a lot
> -- slow is better than crashing!)

The reason i opted for the 16 bit memory in the first place was 'cause the $1800
that acer was going to charge me for the (proprietary) 32-bit board and the
2Meg of extra memory was a whole heck of a lot more expensive than the $650 or
so for the 16 bit board + 2M of memory from intel.  Perhaps though i was penny
wise and pound foolish.  What do *y'all* think?  I just couldn't afford another
1800 dollars...!

> Karl Denninger ...ihnp4!ddsw1!karl
-- 
Derek Terveer	root@hawkmoon.MN.ORG	..!uunet!rosevax!elric!hawkmoon!root