Richard,

 

CP is probably expecting to see an unsolicited x'80' status from the
tape device, indicating that it is ready. But since the device isn't
really a tape it may never provide an attention when it's ready to read
or write, which results in the devices showing I/R from CP's
perspective.

 

The vendor would have to better emulate a tape device, providing the
attention when it's ready and also no-op any of the several other tape
commands like rewind unload etc. it doesn't support. The device could
simply provide CE/DE status to any of the basic tape commands it can't
perform to no-op them, making CP happy. Otherwise you may have to define
address ranges of unsupported device types.

 

 (my posts have had long delays - this one was sent 12:17pm CT on 9/25)

 

Regards,

 

Mike

 

________________________________

From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Schuh, Richard
Sent: Thursday, September 25, 2008 11:21 AM
To: [email protected]
Subject: Re: Error Flags

 

No, READY does not help, it only works on virtual devices that have a
pending interrupt. The only things I know of that do affect it, and I
have been going down the list, are SET RDEVICE cuu CLEAR, which requires
that the devices be varied offline, and an IPL of VM. From looking at
the CP code, about the only way to clear this other than the two
previously mentioned would be for there to be a clean, unsolicited,
device end from the device. 

 

The devices really are not virtual tape devices, they are special
processors that are written to and read from. I presume that tape was
chosen as the medium to masquerade as because it is the simplest device
that has the ability to do both. You probably cannot consider a printer,
even though it is possible to read the print buffer on at least some of
the models available, so you are left with tape and dasd., and tape is
simpler to emulate. The problem is that the vendor chose to not handle
things quite right when CP issues a RUN command at DETACH time. Their
previous model never told CP that the devices were NOT READY, so they do
not have this effect. 

 

Since the device never pays attention to the I/R condition and neither
does TPF, the equipment is usable under VM. However, it should behave
better. Unwanted conditions, ones that cannot be handled by the
equipment, should never purposefully be reflected to CP by the devices.

 

I could call the devices unsupported class tape devices. If I could
count on their staying at the same addresses, that might be OK.
Unfortunately, I just found out this week that 16 of them had been
readdressed a month ago, so I cannot count on it. I do not like playing
catch up with the device configuration when it comes to strange devices.
It gives me too many chances for error, too many chances to screw up my
SYSTEM CONFIG. And besides, the vendor had it right and chose to mess it
up. 

 

 

Regards, 
Richard Schuh 

 

 

         

        
________________________________


        From: The IBM z/VM Operating System
[mailto:[EMAIL PROTECTED] On Behalf Of Mike Rydberg
        Sent: Wednesday, September 24, 2008 2:58 PM
        To: [email protected]
        Subject: Re: Error Flags

        Richard,

         

        Does the CP READY vdev command have any effect on these virtual
tape devices?

         

        Mike

         

        
________________________________


        From: The IBM z/VM Operating System
[mailto:[EMAIL PROTECTED] On Behalf Of Schuh, Richard
        Sent: Wednesday, September 24, 2008 4:45 PM
        To: [email protected]
        Subject: Error Flags

         

        Why does the RDEVINTR flag in RDEVRFLG persist, as evidenced in
the messages below, even though the device has been attached to a user
and is actively being used? Examination of the RDEV confirms that the
bit is on. 

        q 69e 69f                               
        A tape  069E intervention required.     
        TAPE 069E ATTACHED TO JOEUSER 069E R/W 
        A tape  069F intervention required.     
        TAPE 069F ATTACHED TO JOEUSER 069F R/W 

         

        Note: These are not real tape drives, so it is not possible to
actually load a tape. Also, the TAPSENSE command that comes with the
VSSI VTAPE product indicates that, at least in the virtual machine to
which they are attached, they are ready with tapes at load point and
they appear to have autoloaders which contain tapes.

        Related question, Is there any way to clear the bit other than
varying the devices off and entering a  SET RDEVICE 62E-62F CLEAR
command? Storing into memory does not count. 

        Regards,
        Richard Schuh 

         

Reply via email to