Dl!KPGTSR ---aeaIlocate DASD buffer and return to storage held by the dumay RECBLOKs. Dl!KSPL, DELSTART ---Por incoiplete allocation RECBLOK chains, deallocate by calling Dl!KPGTSD. DMKPGTSD ---neaIlocates a page at a time via SFBSTIRT and the IOBLOK until the
last page is reached. Dl!KFRFT -- Delete the SFBLOK, then go to Dl!KSPL and NEXTSFB. J2!KSR1, If the delete queue is not empty, process the next SFBLOK an
identical manner. Continue until all SFBLOK deletions are complete
then call Dl!KFRET. Dl!KFRET -- Delete the IOBLOK. DMKDSPCH the dispatcher. RECOVERY MANIGEMENT SUPPORT OPERATION Entry from CP initialization module to set up pointers to VM/370 error recording cylinders. Dl!KIOGF1 ---rhe-STIDP instruction stores processor version and model in CPUID of PSI. Dl!KIOG, ISSUEINS ---cbeck-attached channels. If standalone channel on the 165 or 168,
the address of the logout routines is stored in the DMKCCH module. 12MKIOG, Set up pointers for machine check and channel check record area and
extended logout areas. Dl!KIOG, Obtain storage for machine check record, extended logout area, and CPEXBLOK. The l!CBIREI is also initialized. Dl!KIOG, PISTDIVE ---netermine-the 901-full and 1001-full capacity of designated error
recording cylinders and store the amount in DMKIOEMX and DMKIOEBI respectively. DMKIOG, Check first record of the error recording cylinders for proper
format. If invalid, reformat. If valid but clear, store pointer value
in PSI as the first available slot for error record. If valid but
used, search for first unused slot and store its value in PSI. DMKIOGFR a 3031, 3032, or 3033 processor, read frames from the SRF (service record file) device, and write them to the beginning of the
error recording cylinders with unique record types. 12 MK IQ2, When error recording area is full, inform the operator, and continue. DMKIOEFL ---rurn-off the recording in progress switch and exit to caller. 1-220 IBM Vl!/370 System Logic and Problem Deteraination--Voluae 1
DMKMCHIN ----Entry via the machine check PSi upon detection of an unrecoverable
and nonfatal processor or storage error. Disable soft machine
recording store logout area on the machine check and channel check
recording cylinders. The system is enabled for hard machine checks
with a pointer to the termination routine. DMKBCH, ENHARD for virtual
user store status in VMBLOK. DMKMCH, BCHSYSIL for system damage
timing facility or uncorrectable retry, multibit storage error post
system operator message, flag system as terminated. Place wait state
code, if first hard error, record it. If the fault occurred in
problem state, terminate the active virtual machine. MMKMCH, For corrected ECC or processor retry, update soft error count and record the error and dispatch the virtual machine. M!!!£!!, For multibit storage error in problem mode, exercise storage location
to clear up or flag as unavailable (permanent error) • On an altered page condition, the virtual machine is reset,
otherwise, the error is recorded and the virtual machine is
redispatched. DBKMCH SPFTEST -- Storage-key failure. Exercise the 2K page key. If CP area and solid
error condition process as DMKMCH, BCHSYSIL, intermittent, restore
the key and go to the dispatcher. If key failure and in virtual machine area if permanent error, mark page as unavailable, terminate
the user. If intermittent condition refresh the key and dispatch the
virtual machine. On conditions that cause the termination or reset. The error is
recorded, and both the user and the operator receive status messages.
Per the termination flag, VMBLOK, the user is logged off and control
returns to the dispatcher or is reset via DBKCFPRR. DMKCFPRR ---Virtual storage is released, the virtual machine is flagged
dispatch able and placed in console f unc tion mode .. On a hard .achine check while handling a machine check. the machine check new PSi is loaded with a wait state PSi and the current PSi is
enabled for hard machine checks. Locate the system or the user's VMBLOK. Call DMKMCTPT if system is running in attached processor mode. Load disabled wait state for uniprocessor system. DMKMCTPT ---Co_plete processor termination for attached processor system. If the
error is on the attached processor and it is in problem state, signal
for automatic processor recovery and stop the attached processor. Make sure processing is on the main processor and set up the
appropriate wait state code. 121!KMCT, Issue a aessage to the operator and load a disabled wait state for
the attached processor system. DMKMCTPR ---Perform automatic processor recovery function. Allow system to
convert to uniprocessor mode by calling DMKCPUUP. CP Method of Operation and Program Organization 1-221
Previous Page Next Page