Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 30 Aug 1995 10:06:14 -0400 (EDT)
From:      "Charles N. Owens" <owensc@enc.edu>
To:        bugs list FreeBSD <freebsd-bugs@freebsd.org>
Subject:   NCR 53c825 WIDE problems
Message-ID:  <Pine.A32.3.91.950830081910.19401A-100000@alpha.enc.edu>

next in thread | raw e-mail | index | archive | help
Greetings,
I'm sending this to the bugs list because I've become convinced that 
there is a bug in the NCR controller code:

I've used the NCR driver for several months now with great success but 
am having trouble with a new machine.

My machine (called delta) is as follows:
	Pentium 100, 32 megs ram
	53c825 controller
	Seagate 4 gig SCSI-2 WIDE hard drive
	dumb isa VGA card

But first...
I have another box (called beta), virtually identical except that it has
P/90, a non-WIDE SCSI-2 hard drive, and a cdrom drive.  On both machines I
run FreeBSD 2.0.5-RELEASE. On "beta" I kept getting these messages: 

/kernel: ncr0 targ 6?: ERROR (81:40) (8-28-0) (8/13) @ (1008:0).
/kernel:        reg: da 10 c0 13 47 8 6 f 1 8 6 28 80 0 8 0.
/kernel: ncr0: restart (fatal error).
/kernel: sd0(ncr0:6:0): COMMAND FAILED (9 ff) @f09a3e00.
/kernel: sd0(ncr0:6:0): COMMAND FAILED (9 ff) @f08b5400.
/kernel: cd0(ncr0:2:0): COMMAND FAILED (9 ff) @f08b5a00.
/kernel: ncr0: timeout ccb=f08b5a00 (skip)
/kernel: ncr0: reset by timeout.
/kernel: sd0(ncr0:6:0): FAST SCSI-2 100ns (10 Mb/sec) offset 8.
/kernel: sd0(ncr0:6:0): UNIT ATTENTION asc:29,0
sd0(ncr0:6:0):  Power on, reset, or bus device reset occurred field 
        replaceable unit: 14

This occasionally resulted in an abrupt system reset!  After some poking, 
I discovered the 2.1-STABLE tree and noticed that the ncr driver source 
seems to have been altered since 2.0.5-RELEASE.  I compiled a new kernel 
from -STABLE, and the above problem seems to have disappeared!  Thanks to 
whoever did the driver fix!

My next hurdle was to get FreeBSD installed on "delta", the P/100 WIDE 
system.  Since the only install disks available are 2.0.5-RELEASE, I 
used them.  For all of my machines, I do install via NFS.  About half-way 
through the extraction of one of the distributions (it varies each time I 
tried), "delta" bombed.  Usually, it spit out "page fault," or "vm_fault" 
and rebooted.  From time to time it simply reset (no warning).  Once, it 
spit out bunch of stuff and paused, and I was able to scroll back on the 
debug screen and see that the FIRST error messages were ncr0 problems,
very similar to the once I'd been seeing earlier on "beta".

At this point, I figured that the problem was primarily with the fact 
that the install kernel contained the unfixed driver, and that perhaps 
the problem was made worse with a faster CPU and PCI bus.

The only way I was able to get FreeBSD installed was to put delta's 4 gig 
drive into beta and use it to do the install.  I copied the 2.1-STABLE 
kernel on it and stuck it back in delta.  It booted fine, and now I'm 
trying to beat the crap out of it to see if I can trust it.  I've been 
running iozone repeatedly, and have had one, completely abrupt reset, and 
a several "panic: page_fault" messages followed by reboot.

Since I need to get this system into production in a hurry, my solution, 
I think, is to slow the CPU down to 90Mhz, and hope that this makes 
it at least as stable as beta is.  Since doing this I've run iozone about 
30 times (command:  iozone 51), and it hasn't crashed yet!  (This, 
compared to it crashing on average on the second run of iozone, when the 
cpu speed was 100MHz).

So, this message is part bug report and part plea for help.
While it seems to be getting ever-more stable, there seems to be some 
instability remaining that becomes visible on faster machines.

If there's any way that I can assist in tracking down this problem, 
please let me know.

BTW,
Are these iozone results correct for my Wide 4gig drive (Seagate ST15230W)?
(this is with my CPU at 90Mhz)

  command		read 		write    (M bytes/sec)
  iozone		12.22		3.27
  iozone 49		3.95		3.23

thanks,
---
-------------------------------------------------------------------------
  Charles Owens					 Email:  owensc@enc.edu
                                       "I read somewhere to learn is to
  Information Technology Services     remember... and I've learned that
  Eastern Nazarene College            we've all forgot..."   - King's X
-------------------------------------------------------------------------





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.A32.3.91.950830081910.19401A-100000>