Notice just a little "churn" on this subject. Thought I'd be able to provide some "reference" opinions even if I cannot point to publicly available documents. So many items so little time but bottom line at the top:
HP7980SX on A-Class 0/0/2/0 path should work as configured assuming good cables and terminators & the SCSI address not conflicting with other devices & no other bad device on the bus.
#1 visibility on Mapper or Mapper2 doesn't equate with a working device.
It only equates to the ability to pass a few bytes to the SCSI controller for the device -- your mechanism can be fried and the controller may still respond to the Identify Command.
#2 Mapper2 <> MPE.
Mapper2 uses a set of PDC drivers which I was told for A/N-Class were pared down HP-UX drivers; some device support therefore is PDC and ISL dependent as well as MPE release and patch dependent but Mapper/2 can typically see just about anything since IDENTIFY is a mandatory command.
#3 A-Class and Dual port HBAs
A-Class only allows one to use FUCTION 0 of the dual function card so A5150A works like A5149A as long as you use only the "A" or "0" labeled part. Same holds for A5159A/B->A4800A and A6829A->A6828A
#4 While 0/0/2/0 on A-Class is served by the same chip (symbois/Logitech ? xxx896) as the A5150A HBA it is really an embedded "Core I/O" or motherboard function. These Core I/O functions are exempted from the dual-port HBA rule -- otherwise the internal disk at 0/0/2/1 wouldn't work (worried feeling -- anyone check that recently?) ;-). So even if we fell prey to the dual port issue we ought to be able to use 0/0/2/0 but not 0/0/2/1.
#5 0/0/2/0 on A-Class is EXTRA special -- despite the xxx896 chipset supporting Ultra2 LVD the external port is an 50 pin HD connector so ONLY Narrow SCSI devices (7980, DDS up to DDS4, CD-ROM) may be attached. Hooking up a "Wide" device will cause a world of hurt since the software driver is not path self-aware (a design feature of all MPE drivers).
The problem would be:
Pci_scsi_dam starts (narrow) & asks the device: "can you talk wide?"
Device replies (narrow) "Yes I can talk wide"
Pci_scsi_dam sets wide mode.
Subsequently every other data byte is dropped as there are no wires there!
#6 HP7980S/SX is supported on A/N and was tested in the lab (by me).
Physical: you're connecting a 50 pin "Low Density" SE device to either 50 pin HD (or 68pin VHD) connectors -- use trusted cables only (HP Sourced best), known good terminator. 99% of all SCSI Problems come from there, AND yes Mapper2 can sometimes see devices across a flaky cable as it repeats itself SLOWLY to give a device every chance to make itself known. As per #2 above physical connection doesn't equate to a device which will configure in MPE or allow a STORE/VSTORE cycle.
Device Firmware: Latest/Last -- no idea what it is. . .
MPE Software: I actually did a pci_scsi_dam code change when I found configuration didn't always work for HP7980S/SX as the device firmware returned a "funny" status. Since there was no firmware fix possible (put to bed around 1990?) I put a kludge in place of along the lines of: "If we get a funny status here, this might be a HP7980S/SX, try again just in case". I don't recall if this was base 7.0 or a patch to the pci_scsi_dam -- latest patches always a good idea.
Whew, I think that covers all the questions and a bit more. No actual source code was referenced in the writing of this document. . .
Jim
-----Original Message-----
From: HP-3000 Systems Discussion [mailto:[log in to unmask]] On Behalf Of Craig Lalley
Sent: Thursday, April 30, 2009 3:03 PM
To: [log in to unmask]
Subject: Re: [HP3000-L] Mpe: HP7980S Configuration on A400 - mapper2 output
--- On Thu, 4/30/09, donna hofmeister <[log in to unmask]> wrote:
> PATH: 0/0/1/0.4.0 LDEV: 7
> ID: HPC1553A TYPE: TAPE
>
> PATH: 0/0/2/0.5.0 LDEV: 9
> ID: HP7980S TYPE: TAPE
>
> (how DO I know these things? ;-) - d
>
******************************************************
donna,
you do know that is a different path, not the other side.
the other side of 0/0/2/0 is 0/0/2/1. :-)
that's ok, if you were always right, we would have to worship you, and that is Jame's job.
-Craig
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|