Memory configurations

Memory configurations offer the flexibility of keeping a specific memory configuration available on a physical processor no matter which logical processor it is.

The z/TPF system matches a memory configuration to a physical processor by using a best fit principle, as follows. The best fit principle must ensure two things:
  • There is enough physical storage for a memory configuration.
  • Areas can remain under the 2 GB bar (virtually) that must reside there.
After the z/TPF system is IPLed, the following process is used to select a memory configuration:
  1. CTIN selects the memory configuration that requires the most memory from the memory that is available.
  2. CTIN allocates storage for the memory configuration:
    • ECBs, IOBs, 4 KB frames, 1 MB frames, common blocks, SWBs, and XWBs
    • ECB associated areas, which include preallocated private area, preallocated ECB heap, preallocated application stack, trace buffers, and various control areas
    • Preallocated system heap
    • Tape buffer
  3. CTIN allocates storage for recoverable system heap. If possible, the storage that was used before the IPL is used during the current IPL, and the data in the area is not changed. The data is recovered. The storage area is not recovered if one of the following conditions occurs:
    • The z/TPF system is IPLed on another physical machine.
    • A hard IPL is done with storage cleared.
    • The memory configuration is changed, which causes the memory layout to be different.
    • A request is made to initialize recoverable system heap by entering the ZMRSH command with the SET and REINIT parameters specified.
  4. CTIN allocates all remaining storage as follows:
    • If the maximum size of VFA is specified as 0 (for the VFAMAX parameter for the CORREQ macro), all remaining storage above the 2 GB bar is allocated to VFA.
    • If the maximum size of VFA is a value other than 0, CTIN allocates less than or equal to the maximum VFA size that was specified.
    • The minimum amount of 31-bit system heap (as specified for the SHP parameter for the CORREQ macro) must be available. All remaining storage below the 2 GB bar is allocated to the 31-bit system heap.
If none of the memory configurations meet the minimum size of VFA, a predefined IBM memory configuration is used. This memory configuration permits the z/TPF system to be restarted. Once the z/TPF system is restarted, use the ZCTKA DEFINE command to define a memory configuration that meets the requirements of the current processor.
Note: You must IPL the z/TPF system for this change to take effect. If the z/TPF system detects the use of a predefined IBM configuration during restart, a message is sent to the operator and restart does not continue.

You can select a specific memory configuration to be used after the next IPL. This is called a preferred memory configuration and is designated, for example, when you have different logical processors performing different functions that might require different memory configurations. For this, you can select a preferred memory configuration and CTIN will use this memory configuration. If, however, CTIN is unsuccessful, a message is sent to the operator in restart to indicate that the preferred memory configuration was not used. Restart is suspended and the operator must respond to a message to indicate whether restart should continue or be canceled.

Restart will display the memory configuration that is in use. If the memory configuration is different from the memory configuration that was used for the previous IPL, a message is sent to the operator. Again, restart is suspended and the operator must respond to a message to indicate whether restart should continue or be canceled.