Previous Topic: Using Exit Routines to Control Allocation RecoveryNext Topic: SWAP Processing


z/OS Dynamic Device Reconfiguration (SWAP) Processing

When a device fails, operators can enter the SWAP command to perform dynamic device reconfiguration (DDR) or the operating system can automatically initiate DDR in response to a detected device failure. DDR allows the operator to move or swap a de-mountable volume from one device to another with compatible attributes.

When the system detects a device failure or when an operator issues the SWAP command from a console, DDR processing under the MASTER address space issues a series of subsystem calls and unit allocation requests as the tape SWAP is processed. DDR uses the unit allocation calls to allocate, through UCB bit settings, potential TO devices for the SWAP and to determine if the potential TO devices can be used by SWAP processing. The SYSIEFSD/DDRTPUR EXCL enqueue is obtained during the allocation of the potential TO devices to serialize the SWAP device allocations with z/OS common allocations, other SWAPs that may be occurring, and HCD ACTIVATE.

While the MASTER address space is holding the SYSIEFSD/DDRTPUR EXCL during SWAP processing, no new allocation requests, no new SWAP requests, and no new HCD ACTIVATEs can start. Conversely, the MASTER address space is unable to obtain EXCLusive control of SYSIEFSD/DDRTPUR to process a new SWAP until all old allocations, SWAPs, and HCD ACTIVATEs have completed, therefore, current allocations, SWAPs, and ACTIVATEs can prevent SWAP processing from initiating.

In summary, SWAP is serialized with other device activity because SWAP must wait for all existing allocations, SWAPs, and ACTIVATEs to complete, and no new allocations, SWAPs, or ACTIVATEs can occur until SWAP completes.