Subject: scsi cable problems?
To: None <netbsd-help@netbsd.org>
From: Nick <nmanisca@vt.edu>
List: netbsd-help
Date: 07/03/2000 20:56:26
I have a very strange scsi bus on an AlphaStation 200.

Under heavy disk load the system will have a problem.  Thy symptom is
that disk reads/writes appear to hang.  After a few seconds on the console
appears a few messages:

sd1(ahc0:1:0): timed out in datain phase, SCSISIGI == 0x44
sd1(ahc0:1:0): asserted ATN - device reset in message buffer
sd1(ahc0:1:0): timed out in datain phase, SCSISIGI == 0x54
ahc0: Issued Channel A Bus Reset #1. 5 SCBs aborted
ahc0: target 1 using 16Bit transfers
ahc0: target 0 using 16Bit transfers
ahc0:A:0: refuses synchronous negotiation. Using asynchronous transfers
ahc0: target 1 synchronous at 20.0MHz, offset = 0x8
ahc0: target 2 using 16Bit transfers
ahc0: target 3 using 16Bit transfers
ahc0: target 0 synchronous at 20.0MHz, offset = 0x8
ahc0: target 0 synchronous at 20.0MHz, offset = 0x8
ahc0: target 0 synchronous at 20.0MHz, offset = 0x8
ahc0: target 0 synchronous at 20.0MHz, offset = 0x8
ahc0: target 0 synchronous at 20.0MHz, offset = 0x8
ahc0: target 0 synchronous at 20.0MHz, offset = 0x8

The last few lines repeat forever (or until I turn the machine off).

Here is an example of what my scsi bus looks like

HD0++HD1++CD0++SA0--HD2--HD3--SA1--ET0

All cables are 68 pin ribbons.  When a device is narrow, a 68->50
converter is used.  There are two cables.  One is denoted by ++
the other is denoted by --


HD0 = hard disk with termination enabled (fast/wide)
HD1 = hard disk (fast/wide)
CD0 = cdrom drive (fast/narrow)
SA0 = adaptech 2940uw with termination set to auto
HD2 = hard disk (ultra/wide)
HD3 = hard disk (ultra/wide)
SD1 = on-board ncr810 not sure what termination is set to (fast/narrow)
ET0 = an external terminator that it attached to the cable

Does this look like a problem wtih my scsi bus?  I am sure that I am violating
some scsi rules, but I was wondering if these messages were a dead give away.

The reason for the two adapters is that the firmware will only boot one of
them (the SA1).  The kernel does not have drivers in it for the SA1 so there
are no duplicate devices (two sd1's etc...)


One more note:
If I leave the bus as it is now and do an fsck on HD3 I get the situation
I described above (messages and all).  But, if I unplug the ++ cable from
the SA0 then the fsck on HD3 works just fine.  I think I am just making the
bus too long?  It is a pretty long bus, but it I am sure that I am very close
to having it work...


Thanks in advance.


Nick Maniscalco
nm@vt.edu