RE: Misbehaving IBM SCSI Disk on XL366

[Kurt_Ludwig student uml edu (Kurt Ludwig)]

YAY!!! Another person with the same problem as me!!!!!!
have 2 U2W IBM 9.1GB SCSI drives (probably the same revision as yours)
they're on a Datastorm32+

[n1pfc@inferno n1pfc]$ cat /proc/scsi/ncr53c8xx/0
General information:
  Chip NCR53C895, device id 0xc, revision id 0x1
  IO port address 0x8800, IRQ number 27
  Synchronous period factor 10, max commands per lun 32
[n1pfc@inferno n1pfc]$ cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
  Vendor: PIONEER  Model: CD-ROM DR-U06S   Rev: 1.05
  Type:   CD-ROM                           ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 02 Lun: 00
  Vendor: TEAC     Model: CD-R55S          Rev: 1.0F
  Type:   CD-ROM                           ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 03 Lun: 00
  Vendor: IBM      Model: DDRS-39130D      Rev: DC1B
  Type:   Direct-Access                    ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 04 Lun: 00
  Vendor: IBM      Model: DDRS-39130D      Rev: DC1B
  Type:   Direct-Access                    ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 05 Lun: 00
  Vendor: COMPAQ   Model: HD00431731       Rev: 3208
  Type:   Direct-Access                    ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 06 Lun: 00
  Vendor: DEC      Model: RZ25L    (C) DEC Rev: 0006
  Type:   Direct-Access                    ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 12 Lun: 00
  Vendor: COMPAQ   Model: HC0183172A       Rev: 3208
  Type:   Direct-Access                    ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 15 Lun: 00
  Vendor: COMPAQ   Model: HC0183172A       Rev: 3208
  Type:   Direct-Access                    ANSI SCSI revision: 02
[n1pfc@inferno n1pfc]$

THe DEC / Compaq drives NEVER time out but the IBM ones do all the time esp
under heavy load.

have tried this with kernel 2.25, 2.2.12, and 2.2.13 and get the same
messages you talk about.

other specs on my system:

164SX @ 533MHz
384MB RAM
RH 6.0 w/ the newer kernels
SB AWE32 w/ 8MB RAM (this one's another story)
DEC 21140 NIC (bay networks w/ the DEC chipset)
3Com 3c905B NIC
Diamond Stealth video VRAM w/ 2MB video

so Ryan You're not alone with your gripe on IBM SCSI U2W drives =)

anyone have any ideas?
-this problem is what caused me to go and spend $400 for a new controller
only to have the SAME problem-

Kurt

-----Original Message-----
From: Ryan Kirkpatrick [mailto:rkirkpat@nag.cs.colorado.edu]
Sent: Sunday, October 31, 1999 4:43 PM
To: axp-list@lists.redhat.com
Subject: Misbehaving IBM SCSI Disk on XL366



	I am having an odd problem with an IBM 2GB Fast SCSI disk on my
XL366 alpha. This disk is being used as a system disk and is connected to
the onboard NCR53C810 controller. The system is running Debian 2.1 and
kernel 2.2.7. The disk in question is detected as below:

  Vendor: IBM       Model: DORS-32160        Rev: WA6A
  Type:   Direct-Access                      ANSI SCSI revision: 01 CCS

Yea, I know that '01 CCS' does not look all that great, but the disk
is listed as a Fast SCSI-2 disk by IBM.
	Previously, I had been using the NCR53C7xx,8xx driver in the
kernel to support the onboard scsi channel. Under this setup, the disk in
question is detected with no problems and can be used with no timeouts or
other nasties. The kernel gives me this on boot up:

scsi1 : target 1 accepting period 100ns offset 8 10.00MHz FAST SCSI-II
scsi1 : setting target 1 to period 100ns offset 8 10.00MHz FAST SCSI-II
   Vendor: IBM       Model: DORS-32160        Rev: WA6A
   Type:   Direct-Access                      ANSI SCSI revision: 01 CCS
Detected scsi disk sdg at scsi1, channel 0, id 1, lun 0
...
SCSI device sdg: hdwr sector= 512 bytes. Sectors= 4127761 [2015 MB] [2.0 GB]

	Yesterday I installed two Tekram 390 scsi cards and thereby
switched to the NCR53C8xx driver for all controllers (as recommended by
the axp-list archives and the kernel help messages). I recompiled a 2.2.7
kernel with this change and proceeded to boot the sytem. All disks are
correctly detected on all chains, and this is what the kernel tells me
about the disk in question:

   Vendor: IBM       Model: DORS-32160        Rev: WA6A
   Type:   Direct-Access                      ANSI SCSI revision: 01 CCS
Detected scsi disk sda at scsi0, channel 0, id 1, lun 0
...
ncr53c810a-0-<1,*>: FAST-10 SCSI 10.0 MB/s (100 ns, offset 8)
SCSI device sdb: hdwr sector= 512 bytes. Sectors= 4127761 [2015 MB] [2.0 GB]

Only, when the system goes to get partition maps for the disks and then
later load init off of this disk (since it is the system disk), I start
getting SCSI timeouts on this disk for the onboard scsi chain. It recovers
from the reset, prints the same SCSI speed detection messages, and then
tries again, only to get stuck and reset due to timeouts again, slowly but
painfully making progress in the boot up. Basically, the system will not
boot.
	I repeated the above test with 2.2.5 and 2.2.12, same
configurations, same problem.
	Now, I booted with the Debian 2.1 boot floppy's kernel, 2.0.36,
and things were quite different. This is what the kenel told me about the
disk:

   Vendor: IBM       Model: DORS-32160        Rev: WA6A
   Type:   Direct-Access                      ANSI SCSI revision: 01 CCS
Detected scsi disk sda at scsi0, channel 0, id 1, lun 0
...
ncr53c810a-0-<1,*>: SYNC transfers not supported.
SCSI device sda: hdwr sector= 512 bytes. Sectors= 4127761 [2015 MB] [2.0 GB]

Then the system boots just fine, and there are no SCSI timeouts, and all
disks on all controllers are visable. I can then proceed to stress test
(i.e. bonnie -s 500) on all disks, compile kernels, etc... with not even a
single scsi reset. I get decent speed from my fast and ultra disks on the
Tekram controllers, but a rather poor 3.7 MB/sec from the disk in
question.
	So, as far as I can determine, this is a software problem, or at
the very least is not a scsi cable/termination issue. On the same chain as
the disk in question are an Exabyte tape drive (ext), two Toshiba CD-ROMs,
and a Yamaha CD-R (which 2.0.36 does not detect or support very well,
works fine with 2.2.7). This scsi chain is unchanged from before adding
the two Tekram cards.
	It appears that for some reason, while the NCR53C7xx,8xx drivers,
the onboard SCSI controller, and the IBM SCSI disk all worked together
well in the 2.2.7 kernels, synchronous support and all... Going to the
NCR53C8xx drivers, with out changing anything else causes some serious
heartburn in the system.
	Therefore, does anyone have any ideas on how to fix this problem?
I would very much like to escape from the 2.0.x kernels and go back to the
2.2.x kernels, if only for the sake of my CD-R drive. If I could even just
tell the IBM drive to not use sync at all in 2.2.x kernels, I would be
happy. So, would someone out there please give me some advice on this
problem? Thanks in advance!

---------------------------------------------------------------------------
|   "For to me to live is Christ, and to die is gain."                    |
|                                            --- Philippians 1:21 (KJV)   |
---------------------------------------------------------------------------
|   Ryan Kirkpatrick  |  Boulder, Colorado  |  http://www.rkirkpat.net/   |
---------------------------------------------------------------------------

--
To unsubscribe: send e-mail to axp-list-request@redhat.com with
'unsubscribe' as the subject.  Do not send it to axp-list@redhat.com