SUMMARY: Sun3/Sun4 and Sun4c SCSI addressing (was Maxtor 207 MB disk)

From: John Turner (jxt@hayworth.lanl.gov)
Date: Sat Jan 09 1993 - 00:32:22 CST


The list came through very quickly for me. I should have posted my
question sooner, as it would have saved me quite a bit of time.

My question dealt with jumper settings on a Maxtor LXT-213SY, one of
the 3 1/2" 207 MB disks Sun shipped with various SPARCs. The twist is
that I'm hanging a couple of these, along with a 424 MB embedded SCSI
disk, on a Sun 3/60. The reason I modified the subject was that the
problem is not specific to this disk. The problem is actually due to
differences between how Sun3 (and I think Sun4) architecture machines
map SCSI addresses to SCSI devices and how Sun4c architecture machines
do it. See below.

Here are excerpts from my original post:

> The Field Engineer's Handbook shows a block of five jumpers:
> A0, A1, A2, WS, and EP. The first three supposedly set the SCSI
> address, and WS and EP are Wait Spin and Parity.

[stuff edited out]

> The drives come with jumpers A0 and A1 in, which makes sense,
> because that would be sd0. Fine. That also agrees with what the
> Field Engineer's Handbook says about how to set the jumpers:
>
> Target 0 ==> all jumpers out (000)
> Target 1 ==> A0 in, A1 and A2 out (001)
> Target 2 ==> A1 in, A0 and A2 out (010)
> Target 3 ==> A2 out, A0 and A1 in (011)

[stuff edited out]

> A summary and other information that may be useful:
>
> - Sun 3/60 running 4.1.1_U1
> - kernel configured for up to four disks
> - System already has a 424 MB disk as sd0 and a 60 MB tape as st0. No
> ESDI disks or anything.
> - I have three Maxtor LXT-213SY 207 MB disks that all behave the same
> way. That is, any one of the three disks works fine as sd2 (with A0
> in and both A1 and A2 out). But that jumper configuration should
> come up as sd1, and with any other jumper configuration the disk
> isn't recognized
> - I poked around the EEPROM, and nothing in the SCSI board configuration
> block is set. I doubt this matters, but don't really know.

Although I received many correct answers, I'm only including one, from
zeke@mpl.ucsd.edu (Rob Scott), which is a nice explanation of the problem
and provides one solution:

>
> The key factor here is that you are trying to put the disk on a 3/60
> instead of a Sparcstation. Sun3 machines and Sun4 (not Sparc) machines
> define disk drives a bit differently in the kernel than do the Sparcstations.
> The Sun3 and Sun4 machines assume that each physical disk consists of
> 2 logical units with the subunit addressing imbedded in the disk. Here
> are the relevant lines from the /sys/sun3/conf/GENERIC file:
>
> disk sd0 at si0 drive 000 flags 0
> disk sd1 at si0 drive 001 flags 0
> disk sd2 at si0 drive 010 flags 0
> disk sd3 at si0 drive 011 flags 0
> disk sd4 at si0 drive 020 flags 0
> disk sd6 at si0 drive 030 flags 0
>
> These lines map the (octal) "drive" number to SCSI ID, with each drive
> number consisting of 8 subdrives. Because no commonly used SCSI disks today
> use logical subunits, physical units map to disks as:
> Physical Unit: 0 Drive: \000 (0) Disk: sd0
> Physical Unit: 1 Drive: \010 (8) Disk: sd2
> Physical Unit: 2 Drive: \020 (16) Disk: sd4
> Physical Unit: 3 Drive: \030 (24) Disk: sd6
>
>
> This differs from the Sparcstation kernel definitions for the first SCSI
> controller, which are as follows:
>
> disk sd0 at scsibus0 target 3 lun 0 # first hard SCSI disk
> disk sd1 at scsibus0 target 1 lun 0 # second hard SCSI disk
> disk sd2 at scsibus0 target 2 lun 0 # third hard SCSI disk
> disk sd3 at scsibus0 target 0 lun 0 # fourth hard SCSI disk
>
> These mapping are as you described in your request for help.
>
>
> The answer (as I have done here for my 4/490 servers) is to change the
> disk "drive" (target) definitions in the kernel file to something like
> the following (if you want to match the Sparcstation mapping):
>
> disk sd0 at si0 drive 030 flags 0
> disk sd1 at si0 drive 010 flags 0
> disk sd2 at si0 drive 020 flags 0
> disk sd3 at si0 drive 000 flags 0

The other solution, as suggested by several people, is to just rebuild
the kernel with only the even drives included (sd0, sd2, sd4, sd6).

Many, many thanks to all who responded:

leclerc@eps.slb.com (Leclerc Francois) [who wins the double prize for being
                                         the first to respond with the correct
                                         answer and also for being the respondent
                                         from farthest away (France)]
Perry_Hutchison.Portland@xerox.com
Tim Beyea <beyea@erc.msstate.edu>
fas@fibercom.com (Andy Smoak)
trinkle@cs.purdue.edu (Daniel Trinkle)
cameron@hobbes.intellistor.com
matt@wbst845e.xerox.com (Matt Goheen)
louis@andataco.com
poffen@sj.ate.slb.com (Russ Poffenberger)
margie%pgt1@Princeton.EDU (Margie Mott)
zeke@mpl.ucsd.edu (Rob Scott)
kbibb@qualcomm.com (Ken Bibb)
Tevian Dray <tevian@MATH.ORST.EDU>
Mike Raffety <miker@il.us.swissbank.com>

-----------------------------------------------------------------
| John A. Turner |
| Los Alamos National Laboratory, MS B226, Los Alamos, NM 87545 |
| Group: X-6 (Radiation Transport) |
| Location: TA-3, Bldg. 43, Rm. D263 |
| Phone: 505-665-1303 e-mail: turner@lanl.gov |
-----------------------------------------------------------------



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:20 CDT