So.   If the most current registers are in the TCB. The the PSW would be in 
STCBPPSW 

What I am doing is canceling the address space with CALLRTM. 
Before that a install a RESMGR exit

Which has the failing TCB and RB


So I know I would have to inspect the PSW to see where execution was taking 
place

Is there anyway from a RESMGR exit
To determine where execution was taking place


Thanks


Sent from my iPhone

On Nov 4, 2012, at 9:50 AM, Peter Relson <[email protected]> wrote:

> Most current when?
> 
> The "most current" registers for a running work unit are of course in the 
> registers, nowhere in control blocks. And status saved by the dispatcher 
> is in the TCB/STCB for an unlocked task (IHSA comes into play for a locked 
> task)
> 
> But I presume if you're looking at a dump from a cancel, you're referring 
> to the registers that were current at the point that the system allowed 
> the work unit to be abtermed.
> Abterming a work unit is accomplished by (in effect) getting the work unit 
> undispatched (which requires obtaining the local lock to issue STATUS 
> STOP) then setting it up to issue an SVC D upon redispatch. 
> 
> Once the work unit is undispatched (and its status saved in the TCB/STCB), 
> cancel results in RTM setting it up to abend, and making sure that in 
> doing so it does not lose information about the state before the 
> undispatch (since in order to abend, the PSW must be saved, and some 
> registers must be updated).  This is why the best answer is what Chris 
> Blaicher said: look in the RTM2WA. When there are multiple RBs, of course, 
> current registers for all the RBs cannot be in the TCB so registers and 
> PSW for the not-top RB are in RB/XSB.
> 
> By the way, PSAPSWSV has nothing to do with the current PSW.  PSAPSWSV (or 
> PSAPSWSV16 as of z/OS 1.13) might have the PSW(E) at the time the work 
> unit was last dispatched.
> And as of z/OS 1.13, XSBOPSW16 has the full 16-byte PSW. RBOPSW typically 
> has the scrunched 8-byte analog of that.
> 
> Peter Relson
> z/OS Core Technology Design
> 
> ----------------------------------------------------------------------
> For IBM-MAIN subscribe / signoff / archive access instructions,
> send email to [email protected] with the message: INFO IBM-MAIN

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to