Wasn't there a time way back in the dark ages when CP would check for 
RESERVE/RELEASE hardware during its IPL roll call and unconditionally mark the 
(real) device as sharable if present? 

Regards,
Richard Schuh


> -----Original Message-----
> From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED]
> Behalf Of Alan Altmark
> Sent: Wednesday, September 06, 2006 12:24 PM
> To: [email protected]
> Subject: Re: VM directory entry for shared DASD
> 
> 
> On Wednesday, 09/06/2006 at 01:10 AST, "Wakser, David" 
> <[EMAIL PROTECTED]> wrote:
> > I believe it is probably overhead that should be avoided? 
> But even if 
> that is 
> > not the case, why buy bicycle tires if you have no bicycle?
> 
> For clarity:
> - The "V" enables CP simulation of RESERVE/RELEASE for guests on the 
> *same* VM system.  Without it, CP causes a Command Reject on 
> the RESERVE 
> or RELEASE (resulting the VSE message)
> - IF a minidisk link mode has the "V" on it
>    AND it is full-pack (not just starting on cyl 0)
>    AND it is defined as SHARED
>   THEN
>    CP will propagate a RESERVE/RELEASE to the real DASD volume.
> 
> The "V" is optional to allow continued simulation of older dasd 
> configuration that don't have the then-optional "two-channel switch" 
> feature.  You *do* have 2835 and 3880 control units, right?  
> :-)  Back 
> then on the real dasd you would get a Unit Check with Command 
> Reject.  You 
> get the same thing on a minidisk without the "V".
> 
> [20 years pass]
> 
> So, it is now the 21st century and Walter Cronkite no longer 
> hosts the CBS 
> Evening News.  Should we change CP such that "V" is the "unchangable 
> default"(SM)?  Is there any reason RESERVE/RELEASE should 
> remain disabled 
> in an ECKD world?  (VM no longer supports CKD devices, where 
> RESERVE would 
> be optional.)  If the disk is a full-pack mini, should CP 
> treat the volume 
> as if it were defined as SHARED?
> 
> The floor is open for discussion.
> 
> Alan Altmark
> z/VM Development
> IBM Endicott
> 

Reply via email to