[384] in linux-scsi channel archive
lockups and other problems -- BT-946C, ATI-AX0, SB16-SCSI
daemon@ATHENA.MIT.EDU (J.H.N.Chin@reading.ac.uk)
Sun Jul 16 18:27:06 1995
From: J.H.N.Chin@reading.ac.uk
Date: Sun, 16 Jul 1995 18:53:31 +0100
To: linux-scsi@vger.rutgers.edu
Haven't had any replies to my previous questions. Perhaps they were
too vague. Anyhow, I have tracked down source of some of the problems
a bit more exactly. Questions are at the end.
----------------------------------------------------------------------
System 1 description:
Gateway 2000 4DX2-66P (~18months old)
Phoenix A486 bios v1.03
IDE card (occupies PCI slot)
ATI AX0 (PCI mach32 graphics card)
BT-946C (PCI SCSI card: requires motherboard CMOS setting adjustments)
Soundblaster-16 SCSI (AHA152x)
WD caviar 405M IDE disk
Micropolis 3243 4G SCSI disk
Fujitsu M2694ESA 1G SCSI disk
System 2 description:
Research Machines `F/A-486P DA SA520 MPC2 ERGO' (new)
AMI bios of some description
IDE disk, CD-ROM and graphics card removed
disks and cards from system 1 added
(BT-946C detected properly by system when jumpers removed from it)
----------------------------------------------------------------------
Problem 1:
with SCSI disk(s) attached to the buslogic.
Replicate by:
1. log in as root on two VCs, and user on another
2. run `dd if=/dev/sd?# of=/dev/null' for some ? and #
on each of the root logins.
3. start X as user.
=> system 1 locks solid somewhere around where the logo appears.
=> system 2 works fine.
----------------------------------------------------------------------
Problem 2:
Replicate by:
a) with SCSI disks attached to the SB16:
1. run `dump 0f - /dev/sda1 > somewhere' where somewhere
is on /dev/sdb?. (the dump is Remy Card's v0.1)
=> system 1: after some time, the bus hangs (/usr is on /dev/sda1
so the whole machine ends up needing rebooting).
On one occasion, syslog recorded:
[...] kernel: scsi0 : resetting for second half of retries.
=> system 2: same symptoms
b) with SCSI disks attached to the buslogic:
1. as before
=> system 1: (not attempted)
=> system 2: similar symptoms, except:
dump generates error messages (not recorded, regrettably).
syslog records (for example):
[...] kernel: scsi1 : resetting for second half of retries.
[...] kernel: SCSI disk error : host 1 id 0 lun 0 return code = 18000002
[...] kernel: Current error sd801: sense key Medium Error
[...] kernel: Additional sense indicates Unrecovered read error
[...] kernel: scsidisk I/O error: dev 0801, sector 838528
[...] kernel: SCSI disk error : host 1 id 0 lun 0 return code = 18000002
[...] kernel: Current error sd801: sense key Medium Error
[...] kernel: Additional sense indicates Unrecovered read error
[...] kernel: scsidisk I/O error: dev 0801, sector 838530
Halting the system and running `verify' from the buslogic autoscsi
programme reveals no problem with the disk.
Note: I'm not sure if these messages were from the just announced
driver or the kernel default one (linux-1.2.11). I think I had
the same problem with both.
----------------------------------------------------------------------
Questions:
1. Help?
2. I guess that problem 1 is caused by PCI teething problems due to the
old motherboard. It a `GTY BAT 496 IP' which I believe is an Intel
`SAT 486 IP' in disguise.
Is there any likelihood that a BIOS upgrade will cure this
or is it time for a new motherboard?
(Gateway says the BIOS upgrade is called `P4DBIO.EXE' on their BBS
(which of course I haven't got a modem to access, hint, hint...))
3. To what are all the errors from problem 2 due?
Is it some problem with the dump program itself (seems unlikely)?
Do I really have a hardware problem with my drive?
----------------------------------------------------------------------
Additional:
One of the FAQs mentions that some Micropolis drives have problems
with Buslogic cards. Which drives? What problems?
A firmware upgrade is suggested. My buslogic has firmware v4.22
and bios v4.86. Are these the most recent?
I'm going to contact Intel on Monday to ask if they know of any
problems with the motherboard and whether a BIOS upgrade will
resolve it.
However, what suggestions are there for replacement boards
(that I can buy here in UK) should it prove necessary?
----------------------------------------------------------------------
Any and all assistance will be gratefully received.
thank you,
-jonathan
--
Jonathan H N Chin, 3 kyu | Cybernetics / CompSci | "Respondeo, etsi mutabor"
| University of Reading |
shrchin@reading.ac.uk | Reading, RG6 6AY, UK | < Rosenstock-Huessy >
cybjhnc@cyber.rdg.ac.uk | TEL:(+44) 1734 318612 |
