Document revision date: 30 March 2001 | |
Previous | Contents | Index |
To run SYSGEN to determine the minimum and default values of a quota, enter the following sequence of commands:
$ RUN SYS$SYSTEM:SYSGEN SYSGEN> SHOW/PQL |
Minimum values are named PQL_Mxxxxx, where xxxxx are the characters of the quota name that follow "PQL$_" in the quota name.
Default values are named PQL_Dxxxxx, where xxxxx are the characters of the quota name that follow "PQL$_" in the quota name.
PQL$_ASTLM
Asynchronous system trap (AST) limit. This quota restricts both the number of outstanding AST routines specified in system service calls that accept an AST address and the number of scheduled wakeup requests that can be issued.Minimum: PQL_MASTLM
Default: PQL_DASTLM
NondeductiblePQL$_BIOLM
Buffered I/O limit. This quota limits the number of outstanding system-buffered I/O operations. A buffered I/O operation is one that uses an intermediate buffer from the system pool rather than a buffer specified in a process's $QIO request.Minimum: PQL_MBIOLM
Default: PQL_DBIOLM
NondeductiblePQL$_BYTLM
Buffered I/O byte count quota. This quota limits the amount of system space that can be used to buffer I/O operations or to create temporary mailboxes.Minimum: PQL_MBYTLM
Default: PQL_DBYTLM
PooledPQL$_CPULM
CPU time limit, specified in units of 10 milliseconds. This quota limits the total amount of CPU time that a created process can use. When it has exhausted its CPU time limit quota, the created process is deleted and the status code SS$_EXCPUTIM is returned.If you do not specify this quota and the created process is a detached process, the detached process receives a default value of 0, that is, unlimited CPU time.
If you do not specify this quota and the created process is a subprocess, the subprocess receives half the CPU time limit quota of the creating process.
If you specify this quota as 0, the created process has unlimited CPU time, provided the creating process also has unlimited CPU time. If, however, the creating process does not have unlimited CPU time, the created process receives half the CPU time limit quota of the creating process.
The CPU time limit quota is a consumable quota; that is, the amount of CPU time used by the created process is not returned to the creating process when the created process is deleted.
Minimum: PQL_MCPULM
Default: PQL_DCPULM
DeductiblePQL$_DIOLM
Direct I/O quota. This quota limits the number of outstanding direct I/O operations. A direct I/O operation is one for which the system locks the pages containing the associated I/O buffer in memory for the duration of the I/O operation.Minimum: PQL_MDIOLM
Default: PQL_DDIOLM
NondeductiblePQL$_ENQLM
Lock request quota. This quota limits the number of lock requests that a process can queue.Minimum: PQL_MENQLM
Default: PQL_DENQLM
PooledPQL$_FILLM
Open file quota. This quota limits the number of files that a process can have open at one time.Minimum: PQL_MFILLM
Default: PQL_DFILLM
PooledPQL$_JTQUOTA
Job table quota. This quota limits the number of bytes of system paged pool used for the job logical name table. If the process being created is a subprocess, this item is ignored.A value of 0 represents an unlimited number of bytes.
Minimum: PQL_MJTQUOTA
Default: PQL_DJTQUOTA
NondeductiblePQL$_PGFLQUOTA
Paging file quota. This quota limits the number of pages (on VAX systems) or pagelets (adjusted up or down to represent CPU-specific pages on Alpha systems) that can be used to provide secondary storage in the paging file for the execution of a process.Minimum: PQL_MPGFLQUOTA
Default: PQL_DPGFLQUOTA
PooledPQL$_PRCLM
Subprocess quota. This quota limits the number of subprocesses a process can create.Minimum: PQL_MPRCLM
Default: PQL_DPRCLM
PooledPQL$_TQELM
Timer queue entry quota. This quota limits both the number of timer queue requests a process can have outstanding and the creation of temporary common event flag clusters.Minimum: PQL_MTQELM
Default: PQL_DTQELM
PooledPQL$_WSDEFAULT
Default working set size. This quota defines the number of pages (on VAX systems) or pagelets (adjusted up or down to represent CPU-specific pages on Alpha systems) in the default working set for any image the process executes. The working set size quota determines the maximum size you can specify for this quota.Minimum: PQL_MWSDEFAULT
Default: PQL_DWSDEFAULT
NondeductiblePQL$_WSEXTENT
Working set expansion quota. This quota limits the maximum size to which an image can expand its working set size with the Adjust Working Set Limit ($ADJWSL) system service.Minimum: PQL_MWSEXTENT
Default: PQL_DWSEXTENT
NondeductiblePQL$_WSQUOTA
Working set size quota. This quota limits the maximum size to which an image can lock pages in its working set with the Lock Pages in Memory ($LCKPAG) system service.Minimum: PQL_MWSQUOTA
Default: PQL_DWSQUOTA
Nondeductible
The values specified in the quota list are not necessarily the quotas that are actually assigned to the created process. The $CREPRC service performs the following steps to determine the quota values that are assigned when you create a process on the same node:
When you create a detached process on another OpenVMS Cluster node, the quotas assigned to the process are determined in the following way:
OpenVMS usage: | process_name |
type: | character-coded text string |
access: | read only |
mechanism: | by descriptor--fixed-length string descriptor |
If a subprocess is being created, the process name is implicitly qualified by the UIC group number of the creating process. If a detached process is being created, the process name is qualified by the group number specified in the uic argument.
OpenVMS usage: | longword_unsigned |
type: | longword (unsigned) |
access: | read only |
mechanism: | by value |
The OpenVMS VAX range is 0 to 31, where 31 is the highest priority and 0 is the lowest. Usual priorities are in the range 0 to 15, and real-time priorities are in the range 16 to 31.
The OpenVMS Alpha range is 0 to 63, with real-time priorities in the range 32 to 63.
If you want a created process to have a higher priority than its creating process, you must have ALTPRI privilege to raise the priority level. If the caller does not have this privilege, the specified base priority is compared with the caller's priority and the lower of the two values is used.
A process with ALTPRI privilege running on a VAX node can create a process with a priority greater than 31 on an Alpha node.
If the baspri argument is not specified, the priority defaults to 2 for VAX MACRO and VAX BLISS--32 and to 0 for all other languages.
OpenVMS usage: | uic |
type: | longword (unsigned) |
access: | read only |
mechanism: | by value |
If you do not specify the uic argument or specify it as 0 (the default), $CREPRC creates a process and assigns it the UIC of the creating process.
If you specify a nonzero value for the uic argument, $CREPRC creates a detached process. This value is interpreted as a 32-bit octal number, with two 16-bit fields:
bits 0--15---member number
bits 16--31---group number
You need IMPERSONATE or CMKRNL privilege to create a detached process with a UIC that is different from the UIC of the creating process.
If the image argument specifies the SYS$SYSTEM:LOGINOUT.EXE, the UIC of the created process will be the UIC of the caller of $CREPRC, and the UIC parameter is ignored.
OpenVMS usage: | word_unsigned |
type: | word (unsigned) |
access: | read only |
mechanism: | by value |
If you do not specify the mbxunt argument or specify it as 0 (the default), the operating system sends no termination message when it deletes the process.
The Get Device/Volume Information ($GETDVI) service can be used to obtain the unit number of the mailbox.
If you specify the mbxunt argument, the mailbox is used when the created process actually terminates. At that time, the $ASSIGN service is issued for the mailbox in the context of the terminating process and an accounting message is sent to the mailbox. If the mailbox no longer exists, cannot be assigned, or is full, the error is treated as if no mailbox had been specified.
If you specify this argument when you create a process on another node, an accounting message will be written to the mailbox when the process terminates. If the node is removed from the cluster before the created process terminates, an accounting message will be simulated. The simulated message will contain the created process's PID and name and a final status of SS$_NODELEAVE, but will lack execution statistics.
Note that two processes on different nodes cannot use the termination mailbox for general interprocess communication.
The accounting message is sent before process rundown is initiated but after the process name has been set to null. Thus, a significant interval of time can occur between the sending of the accounting message and the final deletion of the process.
To receive the accounting message, the caller must issue a read to the mailbox. When the I/O completes, the second longword of the I/O status block, if one is specified, contains the process identification of the deleted process.
The $ACCDEF macro defines symbolic names for offsets of fields within the accounting message. The offsets, their symbolic names, and the contents of each field are shown in the following table. Unless stated otherwise, the length of the field is 4 bytes.
Offset | Symbolic Name | Contents |
---|---|---|
0 | ACC$W_MSGTYP | MSG$_DELPROC (2 bytes) |
2 | Not used (2 bytes) | |
4 | ACC$L_FINALSTS | Exit status code |
8 | ACC$L_PID | External process identification |
12 | Not used (4 bytes) | |
16 | ACC$Q_TERMTIME | Current time in system format at process termination (8 bytes) |
24 | ACC$T_ACCOUNT | Account name for process, blank filled (8 bytes) |
32 | ACC$T_USERNAME | User name, blank filled (12 bytes) |
44 | ACC$L_CPUTIM | CPU time used by the process, in 10-millisecond units |
48 | ACC$L_PAGEFLTS | Number of page faults incurred by the process |
52 | ACC$L_PGFLPEAK | Peak paging file usage |
56 | ACC$L_WSPEAK | Peak working set size |
60 | ACC$L_BIOCNT | Count of buffered I/O operations performed by the process |
64 | ACC$L_DIOCNT | Count of direct I/O operations performed by the process |
68 | ACC$L_VOLUMES | Count of volumes mounted by the process |
72 | ACC$Q_LOGIN | Time, in system format, that process logged in (8 bytes) |
80 | ACC$L_OWNER | Process identification of owner |
The length of the termination message is equated to the constant ACC$K_TERMLEN.
OpenVMS usage: | mask_longword |
type: | longword (unsigned) |
access: | read only |
mechanism: | by value |
Each option (bit) has a symbolic name, which the $PRCDEF macro defines. You construct the stsflg argument by performing a logical OR operation using the symbolic names of each desired option. The following table describes the symbolic name of each option:
Symbolic Name | Description |
---|---|
PRC$M_BATCH | Create a batch process. IMPERSONATE privilege is required. |
PRC$M_IMPERSONATE | Create a detached process under another UIC. |
PRC$M_DISAWS | Disable system-initiated working set adjustment. |
PRC$M_HIBER | Force process to hibernate before it executes the image. |
PRC$M_HOME_RAD |
Assign process to specified home resource affinity domain (RAD).
Note: OpenVMS support for RADs is available only on the new AlphaServer GS series systems. For more information about using RADs, refer to the OpenVMS Alpha Partitioning and Galaxy Guide. |
PRC$M_IMGDMP | Enable image dump facility. If an image terminates due to an unhandled condition, the image dump facility writes the contents of the address space to a file in your current default directory. The file name is the same as the name of the terminated image. The file type is .DMP. |
PRC$M_INTER | Create an interactive process. This option is meaningful only if the image argument specifies SYS$SYSTEM:LOGINOUT.EXE. The purpose of this option is to provide you with information about the process. When you specify this option, it identifies the process as one that is in communication with another user (an interactive process). For example, if you use the DCL lexical function F$MODE to make an inquiry about a process that has specified the PRC$M_INTER option, F$MODE returns the value INTERACTIVE. |
PRC$M_NETWRK | Create a process that is a network connect object. IMPERSONATE privilege required. |
PRC$M_NOACNT | Do not perform accounting. ACNT privilege is required. |
PRC$M_NOPASSWORD | Do not display the Username: and Password: prompts if the process is interactive and detached and the image is SYS$SYSTEM:LOGINOUT.EXE. If you specify this option in your call to $CREPRC, the process created by the call is logged in under the user name associated with the creating process. If you do not specify this option for an interactive process, SYS$SYSTEM:LOGINOUT.EXE prompts you for the user name and password to be associated with the process. The prompts are displayed at the SYS$INPUT device. |
PRC$M_NOUAF |
Do not check authorization file if the process is detached and the
image is SYS$SYSTEM:LOGINOUT.EXE. You should not specify this option if
a subprocess is being created.
In previous versions of the operating system, the symbolic name of this option was PRC$M_LOGIN. The symbolic name has been changed to more accurately denote the effect of setting this bit. For compatibility with existing user programs, you can still specify this bit as PRC$M_LOGIN. |
PRC$M_PARSE_EXPANDED | Sets the PARSE_STYLE_PERM and the PARSE_STYLE_IMAGE properties for the new process to EXTENDED. |
PRC$M_PSWAPM | Inhibit process swapping. PSWAPM privilege is required. |
PRC$M_SSFEXCU | Enable system service failure exception mode. |
PRC$M_SSRWAIT | Disable resource wait mode. |
PRC$M_SUBSYSTEM | Inherit any protected subsystem identifiers. The default is that the new process does not inherit subsystem identifiers. |
PRC$M_TCB | Mark a process as part of the trusted computing base (TCB). As such, it is expected to perform its own auditing. IMPERSONATE privilege is required. |
Note that options PRC$M_BATCH, PRC$M_INTER, PRC$M_NOUAF, PRC$M_NETWRK, and PRC$M_NOPASSWORD are intended for use by Compaq software.
OpenVMS usage: | reserved |
type: | longword (unsigned) |
access: |
OpenVMS usage: | SCS_nodename |
type: | character-coded text string |
access: | read only |
mechanism: | by descriptor--fixed-length string descriptor |
OpenVMS usage: | longword_unsigned |
type: | longword (unsigned) |
access: | read only |
mechanism: | by value |
The home RAD is determined by the operating system, unless you explicitly request one. If bit PRC$M_HOME_RAD in the stsflg is set, home_rad is the RAD on which the process is to start. Note that you may set this bit to 0 on non-RAD systems.
Note: OpenVMS support for RADs is available only on the new AlphaServer GS series systems. For more information about using RADs, refer to the OpenVMS Alpha Partitioning and Galaxy Guide.
The Create Process service creates a subprocess or detached process on behalf of the calling process. A subprocess can be created only on the current OpenVMS Cluster node. A detached process can be created on the current OpenVMS Cluster node or on the node specified with the node argument.A detached process is a fully independent process. For example, the process that the system creates when you log in is a detached process. A subprocess, on the other hand, is related to its creating process in a treelike structure; it receives a portion of the creating process's resource quotas and must terminate before the creating process. Any subprocesses that still exist when their creator is being deleted are automatically deleted.
The presence of the uic argument, node argument, or the PRC$M_IMPERSONATE flag specifies that the created process is detached.
Creating a process is synchronous in that the process has actually been created and its PID determined before control returns to the program that requested the system service. Note, however, that the new process has not necessarily begun to execute at that point. Some error conditions are not detected until the created process executes. These conditions include an invalid or nonexistent image; invalid SYS$INPUT, SYS$OUTPUT, or SYS$ERROR logical name equivalence; inadequate quotas; or insufficient privilege to execute the requested image.
In creating a detached or subprocess, you can specify that the process run the image SYS$SYSTEM:LOGINOUT.EXE. During interactive logins, LOGINOUT performs the following functions:
- It validates user name and password.
- It reads the system authorization file record associated with that user and redefines the process environment based on information from the record.
- It maps a command language interpreter (CLI) into the process and passes control to it.
The CLI reads a command from SYS$INPUT, processes it, and reads another command. The presence of the CLI enables the process to execute multiple images. It also enables an image running in the process to use Run-Time Library procedures, such as LIB$SPAWN, LIB$DO_COMMAND, and LIB$SET_LOGICAL, that require a CLI.
Running in the context of a process you create through $CREPRC, LOGINOUT can perform some or all of the preceding steps, depending on whether the process is a subprocess or a detached process and on the values of PRC$M_NOPASSWORD and PRC$M_NOUAF in the stsflg argument.
Certain characteristics of a created process can be specified explicitly through $CREPRC system service arguments, while other characteristics are propagated implicitly from the $CREPRC caller. Implicit characteristics include the following:
- Current default directory
- Creator's equivalence name for SYS$DISK
- User and account names
- Command language interpreter (CLI) name and command table file name
Note, however, that after the process has been created, if it runs LOGINOUT and LOGINOUT redefines the process environment, those characteristics will be overridden by information from the system authorization file.
Several process characteristics are relevant to the creation of a process on another OpenVMS Cluster node, in particular, process quotas, default directory, SYS$DISK equivalence name, CLI name, and CLI command table name.
Quotas for a process created on another OpenVMS Cluster node are calculated as previously described in the section on the use of the quota list; namely, they are based on explicit values passed by the creator and system parameters on the other OpenVMS Cluster node. If the other node has its own authorization file with node-specific quotas, you might want to specify in the $CREPRC request that the process run LOGINOUT so it can redefine the process environment based on that node's quotas for the user.
Unless overridden by LOGINOUT, the new process will use its creator's default disk and directory. If the disk is not mounted clusterwide, the created process might need to redefine SYS$DISK with an equivalence name that specifies a disk accessible from that node.
Previous Next Contents Index
privacy and legal statement 4527PRO_019.HTML