开云体育

-----Original Message-----
From: [email protected] <[email protected]> On Behalf Of Fish Fish
Sent: Saturday, March 16, 2024 7:36 PM
To: [email protected]
Subject: Re: [h390-vm] Tape Error?

Mark A. Stevens wrote:

I have been attaching and detaching tapes from VM/370 for quite a
while and have seen this notice every time.
Nothing seems to have damaged the tape files when read-only or when
writing. Is this a problem/error, or just normal?

[...]

09:20:12 HHC00201I 0:0581 Tape file tapes/Original_EDGAR_tape.aws,
type aws: tape closed
09:20:12 HHC01315I 0:0581 CHAN: ccw 0F000000 20000001
09:20:12 HHC01312I 0:0581 CHAN: stat 2E00, count 0001
09:20:12 HHC01313I 0:0581 CHAN: sense 40220000 00C00300 00000000
00800100 010000FF FF000000 00000000 00000000
09:20:12 HHC01314I 0:0581 CHAN: sense INTREQ EOC WRI

(separated for readability/emphasis)

09:20:12 /08:20:12 DMKTAP516I TAPE 581 BAD SENSE CMD= 0F
09:20:12 HHC00009I RRR...RING...GGG!^G
09:20:12 / SNS=

4022000000C003000000000000800100010000FFFF000000

09:20:12 / CSW= 00F3D978 2E000001 USERID= SYSTEM

This is "normal" -- for Hercules. :)

It is a long standing "known problem" that causes absolutely no harm (other
than to cause certain guests such as VM/370 for example to issue a slightly scary
error/warning message to the user).

The problem is that we don't present CUE, CE, DE and/or UC as separate
interrupts, as Ivan explains in our "Unresolved Issues" README:

-->
"Unresolved Issues" -->
390/hyperion/blob/master/readme/README.ISSUES.md:


"Tape Device Handler"

"CE+DE on tape rewind-unload"

"Technically, they should be presented separately to the
guest O/S: CE upon completion the actual rewind-unload
CCW and then DE once the tape actually finishes unloading,
but doing so is somewhat problematic especially for SCSI
tapes. (Might be possible to do for AWS/HET tapes but
I haven't looked into it yet)."

"Addendum (Ivan 2004/10/02)"

"Final DE on Rewind Unload must be accompanied by UC
(Unit Check) (all device types) and possibly CUE
(Control Unit End) for 3420 (3803 Control Unit Type,
to indicate an end of Control Unit Busy condition).
It is possible the CUE should only occur if a BUSY
was presented at some point, but I am not 100% sure
about that (and that's not supported either, as we
don't support shared Control Units). However, VM/370
for example insists that CUE be presented."

"On the same line, VM/370 DDR is failing on Rewind unload.
This could be part of the above problem."


BOTTOM LINE: with some guests (most notably VM/370) you will *always* an
"error" reported whenever a tape drive unloads a tape, since, as explained in the
README above, a UC (Unit Check) is *required* to be presented in such a
situation, and it is nothing to worry about. It's completely normal in the Hercules
environment.

Now whether or not the actual SENSE data that we're presenting to the guest in
that situation is accurate or not is an entirely different matter altogether. But I
believe the sense data is very likely correct/accurate as well. I suspect that
VM/370 is probably just issuing that particular message of its in the described
situation. I doubt that it's actually complaining about the sense data, but of
course I do not know that for certain. I'm far too lazy to pour through VM's
source code to see under what conditions that particular messages of its is being
issued.

Hope that helps!

--
"Fish" (David B. Trout)
Software Development Laboratories

mail: fish@...