CA1293811C - Domain related access lists - Google Patents

Domain related access lists

Info

Publication number
CA1293811C
CA1293811C CA000583573A CA583573A CA1293811C CA 1293811 C CA1293811 C CA 1293811C CA 000583573 A CA000583573 A CA 000583573A CA 583573 A CA583573 A CA 583573A CA 1293811 C CA1293811 C CA 1293811C
Authority
CA
Canada
Prior art keywords
program
address
access
access list
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CA000583573A
Other languages
French (fr)
Inventor
Carl E. Clark
Alan G. Ganek
Michael G. Mall
David R. Page
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
International Business Machines Corp
Original Assignee
International Business Machines Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by International Business Machines Corp filed Critical International Business Machines Corp
Application granted granted Critical
Publication of CA1293811C publication Critical patent/CA1293811C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/468Specific access rights for resources, e.g. using capability register
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/0223User address space allocation, e.g. contiguous or non contiguous base addressing
    • G06F12/0292User address space allocation, e.g. contiguous or non contiguous base addressing using tables or multilevel address translation means
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/14Protection against unauthorised use of memory or access to memory
    • G06F12/1458Protection against unauthorised use of memory or access to memory by checking the subject access rights
    • G06F12/1483Protection against unauthorised use of memory or access to memory by checking the subject access rights using an access-table, e.g. matrix or list

Abstract

DOMAIN RELATED ACCESS LISTS
Abstract Of The Disclosure A method and apparatus is provided to translate the contents of access registers into information for use in performing addressing functions for multiple virtual address spaces. The access registers represent the full addressing capability of the system but do not directly contain the addressing information. The system has a plurality of general purpose registers, a plurality of access registers associated with the general registers, an access list having access list entries which is addressed by the contents of the access register, memory storage for holding address space number second table entries (ASTE), where the contents of the access list entry locate the ASTE and where the ASTE contains the addressing information needed to locate a virtual address when combined with the contents of a general register, Access register translation (ART) consists of the process of determining addressing information by using the access list entry and the ASTE. The system has available to it at any one time a selection of one of two domains each represented by an access list. One domain is related to the dispatchable unit task to be performed and the other is related to the address space in which a particular program operates. The ART process selects the domain which the access register is using.

Description

3~

DOMAIN RELATED ACCESS LISTS

Field of the Invention The invention relates to a method and means for enabling a program or programs being executed in a data processing system to have concurrent access to multiple virtual address spaces. In this system, access registers corresponding to general purpose registers contain tokens to allow the system to identify address spaces. The tokens are used to specify an access list entry for obtaining a segment table designation in a translation process, The use of tokens allows system control of address spaces to be isolated from program control of acce:s registers. ~his invention is an PO987-007 ~ ~
~2938~:~

improvement over the access register system shown in U.S. patent 4,355,355.
-Description of The Prior Art Data processing systems using virtual addressing in multiple virtual address spaces, such as the IBM*
System/370 Systems using ~VS con~rolled programmingr are well known. The organization and hardware/architecturaI
aspects of the IBM System/370 are described in the "IBM
System/370-XA Principles of Operationl', form number SA22-7085 1. The described MVS system includes a central processing unit ~CPU) which contains the ~:
sequencing and processing facilities for instruction execution, interruption action, timing functions, initial program loading, and other machine related functions. Also included is a main storage, which is directly addressable and provides for high speed processing of data by the CPU. The main storage'may be either physically integrated with the CPU, or constructed in stand alone units.

U.S. patent number 4,096,573 to A. R. ~eller et al for "DLAT Synonym Control Means for Common Portions of All Address Spaces" and U.S. patent number 4,136,385 to P. M. Gannon et al for "Synon~m Control Means for Multiple Virtual Storage Systems" both assigned to the same assignee of the present invention, disclose MVS
systems in which the main storage may be allocated as address spaces for use by multiple users, each address space containing a portion defined as common among all of the users. The result is that a user may isolate programs or data from other users by placing them in a "private" portion of the user's assigned address space, or he may make them accessible to all other users hy : * ~egistered trade mark ~ PO987-007 ~. lZ5~381~
, ~ . .
`

placing the programs or data in "common". In such a system, data may be moved between two address spaces by having a program in the first address space move the data from its private area into common and then signal a program in the other address space to operate on, or further move, the~data The use of common as a communication area between address spaces increases the size of the common area and thus reduces the size of ~he private area available to all users. Signalling from one program to another can only be done by subsystems or - ~ `
the control program. Data is protected by storage protect keys. However, there are only 16 such keys, ; which are not enough to guarantee that the information ` is prote~ted from an inadvertent store by another ~il5 ~ subsystem or authorized program since the information is commonly addressable.

~, r~ ,` ,, U~S. patent number 4,355,355 assigned to the same-assignee as the present application, shows access registers (AR's~ associated with the general purpose ~20 ~ registers (GPRs) in a data processor. The AR's are each loaded with a unique STD ~segment table descriptor).
~ The STD comprises a segment table address in main ; storage and a segment table length field. There are a predetermined number of AR's associated respectively with the GP~'s in a processor to define a subset of up to the predetermined number of data address spaces up to a maximum of one address space for each GPR. The STD in an AR is selected for address translation when the associated GPR is selected as a storage operand base register, such as being the GPR selected by the B-field in an IBM System/370 instruction. Each AR may also specify that it does not use the STD in its associated AR to define its data address space, but instead uses the STD in the program address space AR. However, the .

PO987-007 ``~
:`;
~Z~338i~ -;

; STD content of an AR is not selected for an address translation if the associated GPR is selected for a purpose other than as a storage operand base register, such as if a GPR is selected as an index (X) register or as a data source or sink register (R) for an instruction. ~ sixteenth A~ may be provided to define and control the executing program address space, which may also contain data.

~ ~ U.S. patents 4,366,537, 4,430,705 and 4,500,952, assigned to the same assignee as the present ! ~ application, all deal with the dual address space (DAS) concept for which the present access register multiple address space ~MAS) concept i8 an improvement. These !;j~`
patents deal with systems in which one program in one ;~ Laddress space is permitted to obtain access to data in another address space or to call a program in another r ~ ~: address space without invoking a super~isor. Each o~ a pl~rality o~ address spaces assigned an Address Space Number (ASN) has an associated set of address ~ translation tables. A second address space can be designated by a program, and when authorized, the program can cause transfer of data in main memory from one physical location to another associated with the different address space. A program changeable space selection control bit controls use of two different sets of address translation tables associated with two ~ - ;
; different address spaces. Without in~oking a supexvisor, a particular program in an assigned-address space can call a program in another address space or ~30 obtain addressability to data in-another address space having an associated set of address translation tables.

U.S. patent 4,037,214 r assigned to the same assignee as the present application, shows a horizontal -~

r pO 9 8 7 - O 0 7 ~293~1 , . . .
.
~, --s addressing system in which multiple access keys in an access key register (AKR~ switch the address space of a storage access as a function of an instruction address, - - a sink operand address and a source operan~d address, respectively.

..
U.S. patent 4,521,846 issued to ~he same assignee as the present invention and entitled ~Mechani~m for Accessing Multiple Virtual Spaces n shows another mechanism for controlling access to plural virtual `~`10 ~ address spaces in a cross-memory implementation where "`'''J"~ data can be accessed in a non-privileged state.

U.S. patent 3,787,813 shows the concept of data processing devices using capability registers. The , ~ ~ patent shows a data processing device with a central -~ processing unit and a storage unit, the information in ' '~
i~"!"~ the storage unit being arranged in segments and the ;~
central processing unit having a plurality of capability reg~sters each arranged to store descriptor information , ~ indicative of the base and limit addresses of an 2~ information segment. One of the capability registers is arranged to hold information defining the base and limit ~-addresses of an information segment which contains a segment pointer table, particular to the program currently being executed b~ the central processin~ unit and a further one of the registers is arranged to hold information defining the base and limit addresses of an information segment which contains a master capability table having an entry for each information segment in the storage unit composed of information defining the ~30 base and limit addresses of a segment. The segment pointer table comprises a list of data words which are used as pointers to define different entries in the master segmont table.

, Poss7-007 lZ93~
; `
:-~ ' . - ...

U.S. patent 4,366,536 shows a digital computer system for selecting and linking multiple separately stored data proceissing procedures consisting o assembly level commands and for selecting a variable data area from a plurality of variable data areas. The system includes memories`for storing the data processing -~
procedures, the variable data areas and linking addresses; a program counter for accessing the memory containing ~he stored data processing procedures;
~lO registers or accessing the memories containing the data and the lin~ing addresses; and a hardware unit which is adapted to execute the assembly level commands contained in selected data processing procedures in accordance with assembly level commands in the data processing 15 ~ procedure being executed and previously selected addresses.

U.S. patent 4,268,903 shows a stack control ``~ register group for controlling a stack area. A data stack pointer register holds the start address o~ the data stack area which is formed in the stack facility and controlled by the user program directly.
.
U.S. patent 4,454,580 entitled "Program Call Method and Call Instruction Execution Apparatus", assigned to the same assignee as the present invention includes a method of passing execution from a program in one logical address space to another program in a new logical address space. The calling program controls selective allocation of segments to the called program but the called program controls the lengths of the ~30 segments being allocated. In this way, recursive calls to the ~ame program cannot a~fect the function or data of other programs or of the same program in a previous call. Also allocation of data segments can be postponed ~93~

until execution resulting in more flexible execution of pxograms written without knowledge of the details of other co-executing programs.

U.S. patent 4,297,743 entitled "Call and Stack Mechanism for Procedures Executing in Different Rings"
shows an architecture based on a hierarchy of rings where each ring represents a different level of privilege. Branches are allowed to rings having a lesser privilege and privilege levels are allowed to be different for read only status as opposed to read and write status. The patent shows a stack frame which has three areas: a work area for storing variables, a save area for saving the contents of registers and a communications area for passing parameters between procedures. Prior to a procedure call, the user must specify those registers to be saved and the user must load into the communications area the parameters to be passed to the called procedure. The system provides for a history of calls in a sequence of stack frames so that a return can be accomplished. Finally, U.S. patent 4,044,334 entitled "Database Instruction Unload" shows a system for retrieving a database pointer for locating database records in one of a plurality of segments of addressable space.

An article in the IBM* Technical Disclosure Bulletin, January 1982, Vol. 24, No. 8, pages 4401-4403 entitled "Method of Revoking a Capability Containing a Pointer-Type Identifier Without Accessing The Capability" deals with an Address Space Number (ASN) as a pointer-type identifier for the address space capability. This publication relates to the dual address space facility and the fact that an address space does not have to be entered to determine if the access *Reigstered Trade Mark P0987-007 '""' ~Z9;~

access is valid ~ince that information mav be determined using the ASN-second-table entry (ASTE) associated with address translation. In general, access to an object by means of the capability is permitted only when the ~nique codes in the capability and the object are equal.
The capability can be revoked simply by changing the unique code in the object without the need to locate and access the capability.
:
While the prior art previously discussed represents ~10 significant advancement in function for the computer user, there are significant areas in which improvement is desired. In particular, there is a need or a facility to move data between arbitrary address spaces.
In addition, there is a need for a facility to control an authorization index for programs in a space so that different authorization indexes can be used in the same space. Finally, there is a need to improve the ability to switch frequently between address spaces and to acqùire or relinquish the additional addressing ~20 capabilities on the switch of addressing environments.
Control and authority mechanisms are important considerations. The contents of an AR should be changeable by the user; however, the user should not have direct access to actual addressing information such as the STD. Thus, there is a need to combine ease of use of address space mechanisms with the proper authority and control mechanisms to prevent unwanted access to address spaces. As.will be described, the present invention provides improved performance in ~30 meeting these needs.

' PO987-007 ~` ~
~L2938 .

g Summary Of The Invention This invention is a data processing system which has multiple virtual address spaces under system control and in which the user's management of the address spaces is by mean-s of to~ens provided by the system for identifying the spaces. The tokens allow thé user to identify the address spacas to be accessed to the ~ystem but do not allow the user to directly control either real or virtual address spaces. Thus, the system provides proper authority and control over access to address spaces 90 that the user cannot directly work with a system managed resource. The user also has the option o selecting among operating modes as to which addressing system o~ several possi~le is invoked.
:, ~
In a computer system accordin~ to the present ~ t~`
invention, an access re~ister (AR) is provided to correspond to all of the general purpose registers that exist for users in ~he architecture. Tn AR mode, each access register contains an access list entry token ~20 (ALET) to signif~ an address space for which the general purpose register is to be allowed access. The ALET
points to an access list entry (ALE). Dynamic address translation (DAT) is to occur using the operand in the general purpose register for the address space signified ~ by the AhET. The process of obtaining the STD using access register translation (ART) is a two step process.
First, the ALET is used to identif~r an access list entry in an access list which contains an address space number second table entry address ~ASTE ADR). Then the ASTE
~30 ADR is used to access the ASTE which contains the STD
used in dynamic address translation (DAT~. Certain tests are performed usiny the components of the ALET, the access list and the ASTE before the STD is obtained from the ASTE.

"~

~0987-007 12~31311 The ART includes reliability, integrity and authority checks. When the ALET in the access register is first used to find the correct access list entry (ALE) in an access list, verification occurs to determine tha~ the entry is valid. Then ~he contents of the ALE are used to find the correct ASTE from which the STD is obtained. In addition, an ASTE sequence number (ASTESN) comparison is made between the contents of the ALE and the ~STE to verify the correctness of the ~10 reference to the ASTE.

These comparisons have to do with the use of a capability as defined in the computer arts. A
capability i~ an unforgèable object which allows its ~ possessor to perform operations on another object. In ~ the case of access registers, the capability deals with ; ~
objects which are address spaces. If an access register were allowed to directly contain the capability, that is ; the STD for the address space, the AR contents would have to be manipulated solely by privileged instructions in ordex to maintain the unforgeable characteristic.
The use of privileged instructions for address space control takes extra time and computer resource because it requires passing control between the problem state pro~ram and the privileged control program.

~ An access list contains the maxim~D potential addressability to address spaces at one time. The ART
process further includes the determination of the authority level of an A~E that may limit the maximum addressability by authority tests. All program code operates under some predetermined authority level and access to an ALE is controlled by authority tests.
There are three levels of tests. The first level is the public/private bit in the ALE. If it is public, any . ~ .
' user of the access list may use the entry. If it is private, the second level test is that an extended authorization index (EAX) in a control reg:ister must be the same as an authorization index (ALEAX) associated with the accessed address space in the ALE. The control register specifies the space accessing authority of the executing program. This is in effect the owner of a controlled space accessing the space. Finally, a third test is performed if the EAX does not equal the ALEAX. An authority table of the accessed ASTE is indexed by the EAX value from the control register. If the particular EAX finds an allowed access in the authority table then the access is permitted. This is the equivalent of the owner of a space allowing special usage.

An advantage of the present invention is that ARs contain ALETs which allows address space control without use of privileged instructions. Because the AR contains an ALET which specifies a space indirectly through an access list entry, an unprivileged instruction or subroutine can save access register contents, use the access register for another purpose and then restore the saved contents without using a control program service.
Thus, the access list controls the authority of the user. The reason the access list entry points to the ASTE which contains the STD, rather than the access list entry directly containing the STD is that control and authority over address spaces may be modified by manipulating the ASTE without having to find all the access list entries referencing the space. The address space second table (AST~ provides a definition of all program address spaces in the system so that the system has a central point of control for all address spaces.
Many users may have access to an address space and should have that space in their access list to do so.

PO987-007 ~
381~
.

More than one ALE in more than one access list can point to the same AST~. Thus, there are two levels of indirection between the AR and the STD which.allow ; separate control of the user's capability and of the ~ 5 authority to address spaces.
~ .
As a furthèr example of the control that is facilitated by the use o~ indirection between the AR and the STD, the management of real addresses in which the system places STD information is much simpler. Segment 0 table origins are identified by real addresses and are usually on.a page boundary. The segment table itself usually occupies full real frames of memory. Thus, it ~ is important that when an address space is swapped out ; of main storage or is terminated, the real frames that ~, map its segment table be reusable to map other segment . ~ . .- ., ~ . i tables or other virtual addresses. This re~uirement is ~;~ met by the use of the ASTE which is the only place the control proaram needs to update information before -~ reusing the real address in the STD.

The selected access list may be from one or more available domains. For example, the disclosed embodiment shows both a dispatchable unit domain (DUAL) and a primary address space domain (PSAL~. Although an access list is associated with either a dispatchable unit or a primary address space, the valid entries in the list are intended to be associated with the different programs that are executed to perform the work of the dispatchable unit. The DUAL is used in the sense that it belongs to the user and can be different for different users even when executing the same program.
The PSAL is used in the sense that it is common to a program executing in the associated primarv address space regardless of dispatchable unitr Two principal .

1;2931~1 ad~antages are gained by the primary address space domain. First, all users of the primary space domain have the capabilities of the domain without having to individually assign or acquire the capabilit~ to the dispatchable unit domain of every user. Second, a dispatchable unit`may acquire capabilities by switching primary address space domains without having to separately acquire those capabilities.

, , The dispatchable unit access list (DUAL) contains , the maY.imum potential addressability of all code running under the dispatchable unit regardless of what address space that code is in, while the PSAL con~ains the maximum space addressability of all code running in a particular address space regardless of what dispatchable ,, , unit it is running under. In either case, the ART
~15 process is subject to the EAX authority check. As an example, this allows a program to call a service in another address space allowing communication in a public address space while that service is forbidden access to the calling space itself for integrity or privacy reasons. Since a program call (PCJ can change the EAX, this is easy to accomplish. Even though the access list is usable in the service address space, the entry ~or the private space can be blocked from access while the dispatchable unit is in the second space.

In addition, the presen* in~tention includes the provision of an ART lookaside buffer (ALB) that reduces the number of references to storage for the ALD, ALE, ASTE and the authority table which otherwise must occur every time the associated GPR contains a storage operand reference. Because the number of storage references during ART can be quite high, the use of an ALB to provide the results of the,A~T function means that the .

~3811 ' access register functions can be provided with an efficient use of computer resources.

When the ALB is used, the CPU performs an ART
process in real storage only for the initial access using that AR entry. The information-from the ART
process is placed in the ALB, and subsequent ART
operations are performed using the information in the ALB, unless that information has become invalid, or been replaced by the results of other ART operations. The ALB also provides required authority checks before produclng an output. ~he presence of the ALB affects ART to the extent that a change to the contents of information used to perform ART in real storage does not necessarily have an immediate, if any, effect on whether an STD is obtained or on which ST~ is obtained from the ALB.

The ALB is logically a table, local to a CPU, consisting of some number of entry types and some number of entries of each type. The most complex im~lementation of an ALB provides the highest probability that one or more ALB entries will be usable in a particular instance o~ ART and, thus, that one or more references to real storage can be avoided in that instance. The most simple implementation of an ALB
provides the lowest probability o avoiding references to real storage.
.~ . .
A linkage stack facility permits programs operating at arbitraril~ different levels of authority to be linked directly without the intervention of the control ~30 program. The desree of authority of each program in a sequence of calling and called programs may be arbitrarily different, thus allowing a non hierarchical organization of programs to be established. Options `~

lZ93~1~

concerning authorities for the EAX changing option and the PSW key changing option as well as those for the PSW
key mask and the secondary address space provide means of associating different authorities with diff~rent programs or with the same called program. The authorit~
of each program IS prescribed in the entry tables and these tables are managed by the control program. By setting up the entry tables so that the same program can be called by means of different PC numbers, the program can be assigned different authorities depending on which PC number is used to call it. The entry tables also allow control over which PC numbers can be used by a program to call other programs.

Stacking PP~OGRAM CAL~ and PROGRAM RETURN linkage ~5 operations provided by the linkage stack facility can link programs residing in different address spaces and having different levels of authority. The execution state and the contents of the general registers and access registers are saved during the execution of a stacking PROGRAM CALL instruction and are partially restored during the execution of a PROGRP*l RETURN
instruction. A linkage stack provides an efficient means of saving and restoring both the execution state and the contents of registers during linkage operations.

In The Drawings Fig. 1 is a diagrammatic illustration of the use of an access register in addressing operands according to the present invention;
. .
Pig. 2 is a diagrammatic illustration of an access register translation of contents of an access register of Fig. l;

~2~

Pos-s7-oo7 ~ ~ ~

Fig. 3 depicts the organization and contents of the control registers for use with the MAS facility of the present invention;

Fig. 4, on the sheet of drawings showing Fig. l, depicts the contents of the PSW for use with the MAS
facility;

Fig. 5 depicts the format of an access-list-entry token for use with the MAS facility;

Fig. 6 depicts the format of an access-list entry for use with the MAS facility;

~ Fig. 7 depicts the format of a linkage-table entry ; according to the present invention;

Fig. 8 depicts the format of an entry-table entry for use with the MAS facility;

Fig. 9 is a diagrammatic illustration of a linkage ' stack for use with the MAS facility;
, , Fig. 10 depicts the format of an entry of the linkage stack of Fig. 9;

Fig. 11 depicts the format of an ASN-first-table entry;

Fig. 12 depicts the format of an ASN-second-table entry according to the present invention;

Fig. 13 depicts the format of an entry of an authority table fcr use with the w~5 facility;

l G

.. . .. . . ` ... `

~293~

Fig. 14 depicts the format of a dispatchable-unit-control table for use with the MAS
facilit~;

.
Fig. 15 is a diagrammatic illustratio:n of the logic-flow of a PC number translation of a PROGRA~1 CALL
operation;
, Fig. 16 is a diagrammatic illustration of the logic flow of a stacking operation of a stacking PROGRAM CALL
instruction;
, Fig. 17 is a diagrammatic illustration of the logic ~low o~ an ASN transla~ion o~ a PROGRAM CALL operation .
Fig. 18 is a diagrammatic illustration of the logic flow of an access register translation of Fig. 2; -: : Figs. l9A and l9B, when taken together, form a flow :
~15 chart of an access register t_anslation operation and ~: exceptions;

j Fig. 20 is an illustration of the use of the TEST
ACCESS REGISTER instruction for use with the MAS
facility;
::
Fig. 21 shows a first embodiment of an ALB entry according to the present invention;

: Fig. 22 shows a second embodiment of an ALB entry according to the present invention;
.
Fig. 23 shows a third embodiment of an ALB entry ~5 according to the present invention;

, PO987-~07 `~
~2~3~311 Fig. 24 shows a fourth embodiment of an ALB entry according to the present invention; and Fig. 25 shows a fifth embodiment of an ALB
~- according to the present invention in which the ALB
entries and au~hority are determined in separate tables~
.
Description of ~he Preerred Embodiment The multiple address spac~ ~MAS) facility of the present invention is an enhancement of the dual address , space facility and of the ascess register system. The ; 10 MAS facility is designed to run compatibly with, and in , addition to, the DAS facility and~ for the most part, to use the same tables and register arrangements as the DAS
facility, with certain changes and enhancements. The access register translation (A~T) system is an improvement which allows,full use of the access register system by the user while providing isolation and prot,ection of machine addressing functions from the user. The use of an ART lookaside buffer (ALB) enhances the performance of ART.

A service provider typically owns one or more address spaces containing data or programs, or both, which the service provider wants to make available to users. The service provider makes programs available to users by assigning them program call ~PC) numbers. This assigning operation includes establishing links for transferring program control, specifying the authorization characteristics needed by the service callers, and assigning the authorization characteristics of the service provider's programs. The transfer o~
program control may be from one address space to another or may remain in the same address .pace. In either ~;

~ Po987~007 1~3~1 case, it may change the authorization from one level to another to pro~ide greater, lesser or different authorization The service provider may run with an authorization level different ~han the caller's level, allowing the service provider routines to access data in address spaces which the caller cannot access. The usex and service provider can access all spaces on the access list which have not been designated as private address spaces. Additionally, the service provider can have access to selected addr~ss spaces which the user cannot access. Similarly, the service provider can be denied access to selected address spaces which the user can access.

The execution of a program instruction may be conveniently divided into two operations. The first operation is the fetching of the instruction to be executed: The second operation is the addressing of operands for the fetching and storing of data on which the instruction operates during its execution. In MAS
in the AR mode, the instruction is fetched from that address space established as its primary address space.
The establishment of the primary address space mav require a space switching operation.

Fig. 1 shows the use of an access register according to the present invention in addressing operands. The process of using the contents of an access register to obtain a S~D for use in a dynamic address .ranslation, is called an access-register-translation (ART) operation, which is generally designated at 10. An instruction 12 has an operation code, a B field which designates a general register 14 containing a base address, and a displacement D, which, when joined with the base address ' PO987-007 ~2931~i~

of general register 14 by an adder 15, ~orms a logical address of a storage op~rand. In the access register mode, the B field also designates an access register 16 which contains an ALET which, when ~ranslated at ART 10, provides the STD for the address space in which the data is stored. The STD from the ART lO may be joined with the logical address from the adder 15, and, when translated together in the dynamic ad~ress translation (DA~) opera~ion, designated generally at 18, provides the real address of the operand for use by the system.
` In addition to the B ~ield and displacement D shown in Fig. 1, an R field may be used for designating a general register containing a logical address of a storage operand.
, ' ' .
The use of an access register of the present ,~ ~ invention may be further illustrated by the following `~
move ~MVC) instruction:
,~, . .
:~ - MVC O(L,1),0(2) ` The second operand of this instruction, having length L, is to be moved to the first-operand location. The logical address of the second operand i5 in general register 2, and the logical address of the first-operand location is in general register 1. The address space containing the second operand is specified by the ALET
in access register 2, and the address space of the first-operand is specified by the ALET in access register 1. These two address spaces may be different address spaces, and each may be different from the current instruction address space.

~30 Fig. 2 provides an overview showing the translation of an ALET to a real address. Shown at 20 is an array ~2~38i~ .

of general registers numbered 0 through 15. An array 22 of access registers, also numbered 0 through 15, are arranged such that each access register is paixed with a respective one of the general registers of array 20, as previously described in connection with Fig. 1. An access-list entry number (ALEN) in the ALET selects an entry in one of the access lists 24 or 25. Access list 24 is the DUAL, and access list 25 is the PSAL. In the example of Fig. 2, the ALEN of access register 2 points to entry 3 of the PSAL 25. The origin of the DUAL is specified by a dispatchable-unit-access-list designation (DUALD) 26 which is found by decoding an entry in control register 2, as will be explained. The origin o~
the PSAL is s~ecified by a primary-space-access-list designation (PSALD) 27 which is found by decoding an entry in control register 5, as will be explained. The access-list designation used in the ART is known as the effective access-list designation ~ALD).
.
- Each entry in the access list includes an ASTE
address which points to an ASN second table entry (ASTE) 98 which may or may not be in an ASN second table (AST) 30. An ASTE may be created and perform its function for ; ART totally independently of actually being in an AST, although ASTEs used by PC are required to be in an AST.
Each ASTE is similar to that used in the DAS facility, ; ~ and includes an STD value to determine the real address by the DAT 18, as discussed in connection wi~h Fig. l.

There are two access lists available to a program at the same time each representing a different capability domain. One access list is called the dispatchable unit access list (DUAL) and the other primary space access list (PSAL). A bit in the ALET
determines whether the ALEN of the ALET is pointing to .'~ ,. ' ' ' .

:

' PO987-007 ~3~
.

an entry in the DUAL 24 or the PSAL 25. Each entry in the access lists 24 and 25 is available for use by programs.

The DUAL domain is intended to be permanently associated with the dispatchable unit ("task" or nprocess") on behalf of the program or programs executed by the dispatchable unit. There is a unique DUAL for every dispatchable unit i~ the system. The DUAL for a dispatchable unit does not change even though the ~10 dispatchable unit may execute programs in many different address spaces. The PSAh domain is associated with a primary address space. All programs which execute in a primary address space share the PSAL of the address space. This allows programs executing within a primary space to share access to a common set of address spaces.
The PSAL changes when the primary address space changes such as on a space switching PC operation. A user, in possession o~ a valid ~IET, m2y access an access list entry on either the DUAL 24 or the PSAL 25, and this entry specifies the desired address space. Other domains may also be implemented and controlled in a similar manner using the ALET and selected control - registers. For example, a system wide access list (SWALi domain may be created having the capabilities of all programs in the system. Various subsets of domains can be constructed as desired, such as a SASN domain access list (SSAL), to allow further exploitation of an existing mechanism.

Entries 0 and 1 of the DUAL are not used because the ALETS are reserved for accessing operands in the primary and the secondary address spaces, respectively, when in the AR addressing mode. The addressing mode of the CPU is designated by bits in the PSW, as will be ' ,.
' ' PO987-007 ~
~5338~

explainedO When the CPU is in the A~ addressing mode, an ALET of zero always refers to the primary address ' space and an ALET of one always refers to the secondary address space. See Fig. 1 in which box 28 identifies these special ALE~S and provide the correct STD for the PASN and the SAS~-to DAT when they occur. When the CPU
is in the home addressing mode, ~he home address space is the source of instructions ~o be executed and of data. The home address space is defined as that address space having the supervisor control informa~ion for the program being executed. By convention, the operating system assigns an ALEN of 2 for each home space for the purpose o~ data access and the STD for the home space is obtained by ART for such access. Since the STD values for the primary and the secondary address spaces are kept in control registers 1 and 7, respectively (see Fig. 3~ access list entries 0 and 1 are not used. As implemented, entries 0, 1 and 2 in the PSAL 25 are unused and are marked invalid.
; .
; 20 An ART lookaside buffer (ALB) 199 receives and saves inputs from the AR 22, the AL 25, and the AST 30 to hold the STD resulting from ART. The ALB also retains the access list designation, DUAL or PSAL. When `the same ALET is used again, ALB 199 provides the correct output directly to DAT 18 so that ART does not have to be repeated.

Figs. 3 and 4 show the control registers and the PSW word, respectively, for providing information for the control of a program and the state of the CPU during instruction execution.

Fig. 3 shows the contents of the control registers O through 15 for the MAS facility of the present invention. All of the contents of the control registers of Fig. 3 will not be discussed, as the majority of them have the identical functions of the control registers of the aforementioned DAS facility and are thus known.
Thus, primarily those changes necessary to provide the MAS facility will be discussed. A 1 in bit 15 of control register zero indicates the CPU is operating in the MAS mode and the control program supports ~S. The MAS facility includes new formats for the entry table entry, the ASN second table entry, the availability of a linkage stack, and the ability to enter the access register mode. Control register 1 contains the primary segment-table designation ~PSTD). Bits 1-19 specl~y the primary segment~table origin (PSTO) and bits 25-31 , designate the primarv segment-table length ~PSTL). Bits ~ 1-25 of control register 2 designate the ; dispatchable-unit-control-table origin ~DUCTO) used by the MAS facility to locate the DUALD, as will be discussed. Bits 1-25 of control register 5 designate the primary ASTE origin IPASTEO). AS will be discussed, . the entry in control register 5 points to the ASTE entry for finding the PSAL origin, and other information, in the ASTE for the primary address space.
~;
; Control register 7 includes the secondary segment-table designation (SSTD) in a format wherein bits 1-19 contain the secondary segment-table origin (SSTO) and bits 25-31 contain the secondary ~30 segment-table length (SSTL). Bits 0-15 of control register 8 contain an extended authorization index (EAX) for use by the MAS facility of the present invention.
As will be discussed, the EAX may be modified as specified by bit entries in the entry table entry llnder PO987-007 -~

lZ~3131~L

the control of the service provider such that authori%ation to access address spaces by a program may be changed.
' Control register 13 contains a home segment-table designation (HSTD) wherein bits 1-19 contain the home segment-table origin (HSTO), and bits 25-31 contain the home segment-table length (HSTL). Bits 1-28 of control regis~er 15 contain the address of a linkage-stack entry as defined in the last linkage-stack operation, ~o be discussed.
.
Fig. 4 shows the format of the program status word (PSW). Bit 5 of the PSW is a n~T mode bit (T) which defines if the DAT 18 of Figs. 1 and 2 is active. Bits 16 and 17 are combined to specify the addressing mode.
When the DAT is active, the combination of bits 16 and 17 define if the CPU is in the primary mode (003, the secondary mode (10), the access register mode l01), or the home mode (11). Bit 32 of the PSW is an addressing mode bit which defines the format of the instruction address in bits 33 through 63 of the PSW. The function and format of the remainder of the fields in the PSW are well understood and defined for IBM Systemt37o Operations.

Figs. 5 and 6 show the format of the ALET and the ~25 ascess list entry for defining the relationship of an access register and an address space.

Fig. 5 shows the format of the ALET discussed in connection with Fig. 2. In the ALET~ bit 7 is a primary-list bit which, when 1, indicates that the ALEN
refers to a PSAL. When the primary-list bit 7 is 0, the ALEN refers to the DUAL. Bits 16-31 contain the ALEN

PO987~007 ~29381~L

referred to in Fig. 2. When the ALE~J is multiplied by 16, the product is iequal to the number of bytes from the beginning of the effective access-list to the designated access-list entry. During ~he ART, an exception is recognized if ~he ~LEN designates an entry that is outside of the effective access-list or if the left most ; 7 bits of the ALET are not all zeroes. The access-list entry is outside of the effective access list if the ALEN points to an address past the end of the access-list as determined by the access-list length (ALL) of the effecti~e ALD. See Fig. 14 for the ALL.
The described format of the ALET does not apply when the ALET is 00000000 or 00000001 (hexadecimal notation), since these values have b~en ~ssigned a special meaning by the ART process.

An ALET can exist in an access register, in a general register or in s~orage, ~nd it is not protected from manipulation by a user's problem program. Through ~,; the use of instructions, any program can transfer the ~20 value of an ALET back and forth between access registers, general registers and storage. A called program can save the contents of the access registers in any storage area available to it, load and use the access registers for its own purposes, and then restore the original contents of the access registers be~ore returning to its caller. Bits 8-15 of the ALET con~ain an access-list-entry sequence number ~ALESN). Since the ALET is not protected from the problem program, and a user may inadvertently change its contents to any value, the ALESN is included in the ALET as a reliability mechanism that is checked during an ART.

Fig. 6 depicts the format of an access-list entry (ALE). ~it 0 of the ALE is an invalid bit which .
3~

indicates when the ALE is not valid. Bit 7 is a private bit which, when 0, specifies that any program is authorized to use this access-list entrv in an ART
operation, When bit 7 is 1, an access-list eY.tended authorization index (ALEAX) value in bits 16-31 of the ALE is used to determine if a calling program is authorized to use this access-list entry. The ALE
includes an ALESN value in bits 8-15, which is compared ~ to the ALESN value of the designating ALET, as discussed - lO in connection wi~h Fig. 5, to make a validity check.
Bits 65-89 of the ALE contains the corresponding ASTE
address of the associated address space. An ASTE
sequence number (ASTESN) iS located in bits 96-126 o~
the ALE for use as a validity check in connection with the ASTE entry, to be discussed.

It is intended that entries on the access-lists 24 and 25 be provided by the control program such that they may be protected from direct manipulation by any problem program. This protection may be obtained by means of key-controlled protection or by placing the access-lists in real storage not accessible by any problem program by means of the DAT. As determined by bit 0 in the entry, an ALE is either valid or invalid. A valid ALE
specifies an address space that can be used by a suitably authorized program to access that address space. An invalid ALE is available for allocation, or reallocation, as a valid entry. The control program provides services that allocates valid ALEs and that invalidates previously allocated ALEs.

Allocation of an ALE consists of the following steps, A problem proqram passes the identification of ; an address space to the control program, and it also passes an indicator specifying either the DUAL 24 or the ' 12~3381~l PSAL 25. This indication is th~ primary list bit 7 of the ALET. The control program then checks the authority of the problem program to access the address space, as will be explained. If the problem program is authorized, the control program selects an invàlid entry ; in the specified àccess list, changes it ~o a valid entry, includes the ASTE address and ASTESN thereby specifying the subject address space, and returns to the problem program the value of an ALET which designates the now allocated ALE. The problem program can then place the new ALET in an access register in order to access the address space. Later, through the use of the invalidation service of the control program, the ALE
that was allocated may be made invalid.
.
~ In this way, a particular ALE may be allocated, then invalidated, and then reallocated, this time specifying a different address space then was specified in the original allocation. To guard the user against erroneous use of an ALET that designates a conceptually wrong address space, the ALESN is stored in both the ALE~ and the ALE. Nhen the control program allocates an ALE, it places the same ALESN in both the ALE and the designated ALET that it returns to the problem program.
When the control program reallocates an ALE, it ch2nges the value of the ALESN in the reallocated A~E such that the value of the ALESN of previously designated ALETs no longer matches the ALESN in the new ALE.

Although the ASTESN portion of the ALE will be discussed further in connection with the ASrE and the associated figures, it is important to note here that comparison of the ASTESN value in the ALE with the value in the ASTE is the mechanism ~y which the ALE authority to designate the ASTE is confirmed. Thus, an ASTE can .

~3~

be reassigned and a different ~STESN assigned to control - its use without having to back tr~ck to all ALE entries which have referenced the ASTE. Through use of the ASTESN the control program does not have to retain every program or dispatchable unit which was able to use the - ASTE. Thus, the authority can be changed by changing the ASTESN and exceptions or interruptions generated when an attempt is ma~e to use the ASTE without the proper ASTESN. This allows the operating system to be made aware of attempts to access the ASTE with a capability granted in an ALE at a time before the ASTESN
was changed. Thus, an operating system has a mechanism to safely reuse an ASTE ~or a new and/or different space, or to revalidate the authority of the current accessors of an existing space to use it.
Figs. 7, 8 and 10 depict the formats of entries in the linkage table, the entry table and the linkage stack, respectively. These tables are used according to the present invention to establish linkage for ~20 transferring control between programs in either the same or different address spaces.

As previously described, a PC number identifies the particular PC routine that the system is to invoke and is constructed by a service provider. Each service provider that provides PC routines owns one or more entry tables for defining the service provider's routines. The entry tables are connected to linkage tables of those address spaces that require access to the PC routines. Each entrv in an entry table defines one PC routine, including its entry point, operating characteristics, and if the PC ins~ruction is a stacking PC. Fig. 7 depicts the format of a linkage table entry, wherein each entry includes an invalid bit at bit 0 7 an PO987-007 `

entry table origin (ETO), and an entry table léngth (~TL), which together define an entry-table designation.

Fig. 8 depicts the format of the entry of the entry table pointed to by the linkage-table entry of Fig. 7.
Bits 0-lS of the entry~table entry contain an authoriza~ion key mask ~AKM) which is used to verify whether the program issuing the PC instruction, when in the problem state, is authorized to call this entry point. Bits 16-31 contain an entry address-space numbex IEASN) which indicates whether a PC-ss or a PC-cp is to occur. When the EASN is all zeroes, a PC-cp is specified, When the EASN is not all zeroes, a PC-ss is specified, and ~he EASN identifies the address-space ~ number (ASN) which replaces the primary ASN ~PASN). Bit ;1~ 32 is an addressing mode bit that replacec the ~ addressing mode bit in the PSW as part of the PC
operation. The entry instruction address is the instruction address that replaces the instruction address in the PSW as part of the PC operation. Bit 63 ~20 is an entry problem state bit which replaces the problem ; state bit, bit 15 of the current PSW, as part of the PC
operation. Bits 64-95 are an entry parameter which is placed in general register 4 as part of the PC
operation. Bits 96-111 is an entry key mask (EKM~ which may be ORed into or repla~e the contents of control register 3, dependent upon the value of the M bit, as will be eY.plained. Bit 128 is a PC-type bit (T) which, when 1, specifies that the program call instruction is to perform a stacking operation. Bit 131 is a PSW-key ;30 control (K) which, when 1, specifies that the entry key (EK~ of bits 136-139 7 S to replace the PSW-key in the PSW as part of the stacking PC operation. When the K
bit is 0, the PSW key remains unchanged. Bit ~32 is a .

~ PSW-key-mask control (M~ which, when 1, specifies that ~29381~L

the EKM is to replace the PSW-key mask in control register 3 as part of the stacking PC operation. When this bit is 0, the EKM is ORed into the PSW-key-mas~ in control register 3 as part of the stacking PC operation.
Bit 133 is an extended-authorization-index control (E) which, when 1, specifies that the entry EAX of bits 144-159 is to replace the current EAX in control registèr 8 as part of the stacking PC operation. When the E bit is 0, the current EAX in control register 8 remains unchanged. Bit 134 is an address-space-control control (C) which, when 1, specifies that bit 17 of the current PSW is to be set to 1 as part of the stacking PC
operation. When this bit is 0, bit 17 o~ the current PSW is also set to 0. Because the CPU must be in either the primary-space mode or the access-register mode when a stacking PC instruction is issued, the result of this C bit is that the CPU is placed in the access-register mode if bit 134 is 1 or the primary-space mode if bit 134 is 0. Bit 135 is a secondary-ASN control (S) which, when 1, specifies that the EASN of bits 16-31 are to become the new secondary ASN, and a new secondary segment-table designation (SSTD) is to be set equal to the new primary segment-table designation (PSTD), as part of the stacking PC-ss operation. When this bit is ~25 0, the new secondary address-space number (SASN) and SSTD are set equal to the old primarv address-space number (PASN) and PSTD, respectively, of the calling .
program. When the EASN is not all zeroes, ~he AS~E
address of bits 161--185, with six zeroes appended on the ~30 right, forms the real ASTE address that results from applying the ASN translation of the EASM. It will thus be seen that the EASN and ASTE address entries in the entry-table entry ~oint to an entry in the AST 30 which contains the STD, as shown in connection with ~ig. 2.
It is unpredictable whether an ASN translation of the ~2~3~

EASN is performed to obtain an AST~ address, or whether the ASTE address of bits 161-185 is used to locate its designated ASTE. The CPU may do the lat~er to achieve ~, improved performance.

Fig. 9 shows a l1nkage stack 35 which may be formed by the control program for each dispatchable unit. The linkage s~ack is used to save the execution state and the contents of the general-registers and access-registers during a stacking operation. The ~la linkage stack is also used to restore a portion of the execution state and the general-register and access-register contents during a return operation, A
linkage stack resides in virtual storage, with the ; linkage stack for a dispatchable unit in the home ~15 address space for that dispatchable unit. As discussed in connection ~ith the control registers of Fig. 3, the home address space is designated by the HSTD in control register 13. This protects the linkage stack inf~rmation and allows recovery of the linkage stack information in the event of a failure in the users address space.

The linkage stack is intended to be protected from problem-state programs so that these programs cannot examine or modify the information saved in the linkage stack except by means of specific extract and modify instructions. The linkage stack 35 may consist of a number of linkage stack sections 36~ 37 and 38 which are chained together by forward pointers and backward pointers.

~30 There are three types of entries in the linkage stack: header entries 40 having a backward pointer, trailer entries 42 having a forward pointer, and state .

~Z93~

entries 43 (see linkage stack section 36). A header entry and a trailer entry are at the beginning and end, respectively, of a linkage-stack section, and are used to chain linkage-stack sections together. Header entries and trailer entries are formed by the control program, and a state entry is added to contain the execution state and register contents that are saved in the stacking operations. The linkage-stack-entry address in control register 15 points to either the current state entry 44 or, if the last state entry in a section has been unstacked, to the header entry for the current section. Fig. 10 depicts the contents of a linkage-stack state entry which, for a stacking PC
instruction, contains the contents o~ the general registers, the contents of the access registers, the PSW
key mask, the secondary address space number, the EAX
from control register 8, the primary address space n~mber, and the contents of the PSW, all at the beginning of the stacking instruction, and the PC number used. In the case of a BRAMCH AND 5TACK instruction (to be explzined), the addressing mode bit and the branch address are saved rather than the PC number.

Each type of linkage-stack entry has a length that is a multiple of eight bytes. A header entry and trailer entry each has a length of 16 bytes. A state entry has a length of 168 bytes (as shown by the numbers in Fig. 10). Each tvpe of entry has an eight-byte entry descriptor at its end ~shown at 46 of Fig. 10 for a linkage stack state entry).

Bit 0 of the entry descriptor is an unstack-suppression bit (U). When bit U is one in the entry descriptor of a header entry or a state entry, a stack-operation exception is recognized during the unstacking process in PROGRAM RETURN. Bit U is set to Po987-007 3~
.

zero in the entry descriptor of a state entry when the entry is formed during the stacking process.

Bits 1-7 of the entry descriptor are an entr~ type (ET) code that specifies the type of linkage stack entry containing the entry descriptor. The codes are:

0000001 Header en'ry 0000010 Trailer entry 0000100 Branch state entry - 0000101 Program call state entry Bits 8-15 of the entry descriptor are a section identification ~SI) provided by the control program. In the entry formed by a stacking process, the process sets the SI equal to the SI of the preceding linkage-stack entry. Bits 16-31 of the entry descriptor form the remaining free space (RFS) field which specifies the number of bytes between the end of this entry and the beginning of the trailer entry in the same linkage stack section. Bits 32-47 of the entry descriptor form the next entry size (NES) field which specifies the size, in ~20 bytes, of the next linkage stack entry, other than a trailer entry, in the same linkage stack section.

When a new state entry is to be formed in the linkage stack during the stacking process, the new entry is placed immediately after the entry descriptor of the ~25 current linkage stack entry, providing that there is enough remaining free space in the current linkage ~tack - section to contain the new entry. If there is not enough remaining free space in the current section, and if the trailer entry in the current section indicates ` 30 that another section follows the current section, the new entry is placed immediately after the entry .

.

~2~3~

descriptor of the header entry of that following section, provided that there is enough remaining free space in that section. If the ~railer entry indiciates that there is not a following section, an exception is recognized, and a program intèrruption occurs. The control program then allocates another section, chains it to the current section, and causes the stacking operation to be reexecuted. If there is a following section but there is not enough free space in it, an . .,~ .
`10 exception is recognized.
.,,,~:,' .
; When the stacking operation is successful in forming a new state entry 44, it updates the linkage-stack-entry address in control register 15 so that the address designates the leftmost byte of the ~15 ~ entry descriptor of the new entry, which thus becomes the new current linkage-stack entry. When a state entry is created during the stacking process, zeros are placed in the NES field in the created entry, and the length of the created state entry is placed in the NES field of ~20 the preceding entry. During a return operation, the contents of the general registers, access registers, and various contents of the control registers are restored from the linkage-stack-state entry 44, and the linkage-stack-entry address in control register 15 is ~5 changed to point to the previous linkage-stack entry.
;~ When the state entry is logically deleted during the unstacking process of a return operation, zeros are ; placed in the NES field in the preceding entry. It will thus be understood that the use of the linkage stack ~;30 allows the operating environment and authorization level of the calling program to be reinstated when program control is returned from the called program by a return instruction. Thus, the linkage operation is both :

, :

;~2~31!~

retraceable to the beginning point and enforceable against the user to that invalid changes may not occur.

An ASN number is assigned by the control program ~or each address space which contains programs. The ASN
- ~ may be ranslate* during a PC-ss operation as described in connection with the DAS facility. Xowe~er, since the ASTE address is found in the ~TE (see Fig. 8), access to the ASTE may be made directly through the ETE in a PC-ss operation without ASN translation. The control program ~10 associates a STD, an A~ , and a linkage table with each ASN by placing pointers in the ASTE associated with the address space. Data in these address spaces may also be accessed by having the control program construct an access list entry pointing to the ASTE. Certain address ~15 spaces may contain only data, no programs. These address spaces do not have ASNs. In the case of data only spaces, only the ASTE, STD, AT r and ALE are used.

Figs. ll and 12 show the format of entries in the AS~ first table and ASN second table, respecti~ely, and ~20 are very similar to those of the aforementioned DAS
facility. Each entry in the ASN tables of Figs. 11 and 12 represent an address space and are established by the control program to provide linkage and authorize addressability to the associated address space.

Fig. 12 shows the format of an ASTE. Bit O of the ASTE is an invalid bit for indicating the validity of the ASTE. The authority table origin (ATO) and the authority table length (ATL) indicate the authority table designation (ATD) of the associated authority ~able. Bits 96-127 contain the associated linkage-table ~ designation (LTD) and bits 1 8-160 contain the ; associated access-list designation. Bits 160-191 ~ .

~Z93l~

contain an ASTE sequence number (ASTESN) for the ASTE.
Since the ASTE may be reallocated as addressi spaces are created and deleted by the control program, each newly created ASTE has a new, unique ASTESN assigned to it.
When an ART operatio~ takes place, the ASTESN in the access-list is co`mpared with the ASTESN in the ASTE as a validity check so that the ASTE may safely be reused for a different address space or different authority.

Fig. 13 shows an authority table which is associated with each ASTE. As with the DAS facility, each au~hority table entry has a P bit and a S bit. The entries in the authority table are indexed such that there is one entry in the authority table for each of the values of EAX in use to access the associated address space. As will be discussed, the entry of the authority table which corresponds to the value of EAX in control register 8 may be used to determine if a program - is authorized to access the address space associated with the ASTE.
:
Fig. 14 shows the format of the dlspatchable unit control table (DUCT) whose address is located in control register 2, as previously discussied. The dispatchable-unit-access-list designation is stored in bytes 16-19 of the DUCT. The other bytes of the DUCT
are not used in the MAS facility, and will not be dis~ussed further.
;`:
The PROGRA~. CALL instruction has been enhanced to improve the function of the linkage facility. If the T
bit, bit 128, of the ETE (see Fig. 8) is one, a stacking PROGRAM CALL operation is ~erformed responsive to a PROGRAM CALL instruction. A stacking PROGR~1 CALL
; (stackin~ PC) is authorized to enter at a poin in an 12S~3~

entry table by the authorization key mask in the entry table entry. A stacking PC with space switching, among other operations, may place a new EAX (associated with the new program) in control register 8. The stacking PC
saves the contents of general registers 0-15, the contents of access registers 0-15, the complete PSW with an updat~d instruction address ~the return address), the primary and secondary ASNs, the PKM, the EAX, an indication $hat the entry was formed by PROG~AM CALL, ~10 the PC number used, and a two word modifiable area in the entry. The purpose of the modifiable area is to allow a program to "footprint" its progress so that appropriate recovery actions can be taken if a failure of the program occurs.

Two new instructions have been added to improve linkage function:
~ .
BRANCH AND STACK
PROGRAM RETURN

The BRANCH AND STACK instruction changes the instruction address in the PSW, and forms a state entry, called a branch state entry in the linkage stack of Fig.
9. The branch state entry is the same as a program call state entry except that it indicates that it was formed by BRANCH AND STACK and contains the branch address instead of the PC number. The BRANCH AND STACK
instruction can be used either in the calling program or at (or near) the entr~ point of the called program. The BRANCH AND STACK instruction at an entry point allows the linkage stack to be used witllout changing old ~30 calling programs.
',, .

. ..
.

~293~

The PROGRAM RETURN instruction is usecl to return from a program given control by means of either a stacking PROGRAM CALL or a ~RANCH AND STACK instruction.
PROGRAM RETU~N logically deletes the last linkage-s~ack state entry, which may be either a program call state entry or a branch`state entry. If the last state entry is a program call state entry, PROGRAM RETURN restoxes all of the state information that was saved in the entry, and the contents of general registers 2-14 and access registers 2-14. General and access registers 0, l and 15 are unchanged by PROGRAM RETURN. If the last state entry is a branch state entrv, PROGRAM RETURN
restores only the complete PSW ~subject to one exception noted) and the contents of general registers 2-14 and access registers 2-14. ~owever, the PER mask bit R, Fig. 4, is not restored by the PR operation. The combination of a stacking PROGRAM CALL and a PROGRAM
RETURN permits non hierarchical program linkage, that is, linkage from a program with some amount of authority to a program with less, more or completely different authority.

Figs. 15, 16 and 17 present the logic flow of the steps necessary to execute a stacking PC operation. Tt will be noted that the logic flow of Figs. 15, 16 and 17 ~25 can also be used to e~ecute a DAS PROGRAM CALL
; instruction. The textual information in the figures describe how various values may be mathematically manipulated to form addresses. Referring back to Fig.
3, if bit 15 of control register 0 (CR0.15) is equal to zero, ETE is 16 bytes and only a DAS PROGRAM CALL
operation can be performed. If CR0.15 is one, ETE is 32 bytes, and ETE bit 128 controls whether a DAS PC or a stacking PC is performed.
'' .

~3~

Fig. 15 is a logic flow diagram of the PC number translation operation of a program call. If CR0.15 = 1, the ASTE pointed at by ~he PASTEO entry in control register 5 (see Fig. 3) is fetched. This primary-ASTE
~5 includes an LTD at bits 96-127 ~see Fig. 12). If the PROGRAM CALL is a~DAS PROGRAM CALL (CR0.15 = 0) the LTD
is located in control register 5 as in a normal DAS
operation. The PROGRAM CALL instruction 50 includes an LX 51 and an EX 52, similar to that discussed in ~10 connection with the DAS facility. The LX Sl is joined with the linkage-table origin (LTO) 53 by an adder operation 53 to give the real address of a linkage table entry 55. The entry-table origin ~ETO) of the linkage table en~ry 55 is joined with the EX52 ~y an adder operation 56 to give the real address of an entry-table entry (ETE) 57 in the entry table.

Fig. 16 is a logic flow of the steps which are executed in addition to the those shown in Fig. 15 for performing a stacking PROGRAM CALL to current primary ~20 (PC-cp) and a stacking PROGRAM CALL with space switching ~, (PC-ss). As previously discussed, if the T bit 60 of the ETE 57 is equal to ~, a stacking operation is to be conducted. First, the value of the AR~ 6~ is ANDed at 63 with the PKM in control register 3 as it existed before the execution of the PROG~AM CAL~ instruction in the problem state, as shown at 64. If the result of the ANDing operation at 63 gives all zeroes, the PROGRA~
CALL instruction ic not authorized to enter at this point, and the PROGRAM CALL operation is terminated. Tf ~30 any one of the bits match in the ANDing operation of 63, the program is authorized to make the PROGRAM CALL at this entry, and the operation continues. If the PROGRAM
; CALL is authorized, the PSW at 65, the EAX at 66, the PKM 64, the SASN 68, and the PASN 69 as they all existed 1293~

before the PXOGRAM CALL are placed on the linkage stack.
Also placed on the linkage stack, but not shown, are the contents of the general registers, the contents of the access registers, and the PC number (see F:ig. 10). The addressing mode bit A and the entry instruction address are placed in the PSW at 70 and 71. The P--bit and C-bit of the ETE 57 are placed in the PSW at 72 and 73~ If the K-bit is equal to 1, the entry kev of the ~TE 57 is placed in the key of the PSW at 74. If the E-bit is equal to 1, the entry EAX is placed in control register 8 at 75. The entry parameter (EP) is placed in general register 4 at 76. If the M bit of the ETE 57 is equal to 1, the entry key mask ~EKM) replaces the PKM at 77 in the control register 3. I~, however, the M bit is equal to 0, the EKM is ORed into the PKM of control register 3 by the ORing operation 78. If a PC-cp operation is being executed or a stacking PC-ss is being conducted and the S-bit is equal to 0, the PASN at 69 replaces the SASN at 79 in contrbl register 3, and the PSTD at 80 in control register l replaces the SSTD 81 in control register 7. If a stacking PC-ss is being conducted and the S bit is equal to 1, the SASN in control register 3 is replaced by the new PASN and the SSTD in control register 7 is replaced by the new PSTD. After these operations, the ASN of ETE 56 is tested at 83. I~ the ASN is equal to 0, a PC-cp operation is being conducted and is complete. If, however, the ASN is not equal to 0, a PC-ss operation is being conducted and the ASTE is obtained for the destination space.

The PROGR~ CALL may change the PSW key 74 with the EX (X bit = 1) to give access to retch protected code of ~ the next instruction.
:' .

~2~38i~ -.

By changing the EAX in control register 8 (see 75~, each program executed to perform the work of the dispatchable unit can be differently authorized to use the ALEs in the DUAL and the PSAL. The EAX 75 in control register ~ can be set equal to the EEAX by a stacking PROGRAM CALL ~E bit = 1). The original EAX ~will then be restored from the linkage stack by a - PROGRAM RETURN. Thus, each program can be executed with an EAX that is specified in the ETE that is used to call the program. Alternately, the EAX can remain unchanged during a callin~ linkage (E bit = 0) ! allowing the called program to have the same authority as its caller.

By setting the PKM 77 in control register 3 equal to the E~ by a PROGRA~I CALL (M bit = 1), the called program has a PKM that is independent of the PKM of the calling program. This allows ~he callea program to have less authority, in terms of the PSW key values it can set, than the calling program. Alternately, the new PKM
77 may be set equal to the OR of the old PKM 64 and the i20 EKM (~1 bit = 0), if desired (see 78).

Setting the new SASN and new SSTD equal to the new PASN and new PSTD, respectively (S=1), prevents the called program from automatically having access, through the use of ALET 00000001 hex, to the caller's primary address space (access capability still may be obtainable by means of either an ALE or the DAS SET SECONDARY ASN
instruction). This is another way in which the authority of the called program can be less than that of the caller. Alternately, the new SASN at 79 and the new ~30 SSTD at 81 may be set equal to the old PASN at 69 and the old PSTD at 80, respectively (S bit - 0).

3~
.

Fig. 17 is a logic flow of the steps of an AS~
translation. As in the DAS facility, each address space containing programs is assigned an ASN, whose value is stored at 90 in the corresponding ETE 57. Also as in DAS, the ASN at 90 consists of two numbers, an AFX 91 and an ASX 92. Control register 14 includes an ASN-first-table origin (AFTO) 93 which, when joined with the AFX at 91 by an adder operation at 94 gi~es the real ; address of an AFTE 95 in the ASN first table. The AFTE
95 includes an ASN-second-table origin (ASTO~ 96 which, when joined with the ASX 92 by the adder operation at 97, ~orms the real address of the ASTE 98 in the ASN
second table 30, also discussed in connection with Fig.
2. Sinc~ the ASTE address 100 is located in the ETE 57 when CR0.15 is one, it may be used in place of the ASN
translation described. Control bit T, 101, located at bit 12 of control register 14 is an ASN translation bit.
- If bit 12 of control register 14 is zero, neither the ASTE address 100 nor the ASN 90 can be used. If bit 12 is one, either can be used. The AX 102 of ~STE 98 and the ASN 90 of ETE S7 are placed in control register 4 at 103 and 104, respectively, for PC-ss operations. The STD 105 of the ASTE 98 is placed in control register 1 i at 106. If CR0.15 = l, the ASTE address is placed in ~25 control register 5 at 107 as the PASTEO. If CR0.15 = 0, the LTD at 108 of ASTE 98 is placed in control register 5 at 107. It can thus be seen that the ASN translation of Fig. 17 provides for either DAS or MAS operations.

The PC-ss operation discussed in connection with Figs. 15, 16 and 17 may be used to transfer control to a new address space for instruction fetching operations, thereby establishing the new address space as the primary address spaceO TypicalIy, when the PC number, , PO~87-~07 3~;33.~.

the entry-tahle entry and the linkage-table entr~ are established by a service provider, an AKM is specified for setting the authoritv of programs calling that PC
number. If a calling program has the authority to enter the program defined bv the entry-table entxy, as ; determined bv the ANDing operation 63 of Fig. 16, the PC
operation may change the EAX stored in control register 8.

For example, the PC operation may also be used to call a system service to add a new ALE to one of the access lists 24 or 25, as discussed in connection with the access list entry of Fig. 6. The service program can establish a new acces~ list entry and provide a new ALET for use in access register mode operations by the calling user. When an access list entry is formed, the EA~ from the callers control register 8 is placed in the ALE as the ALEAX. Once the ALE is created, the service program returns the ALET for that ALE to the user program. The ALET may then be stored, or passed to other addres~ spaces, in any convenient manner for use in fetching or storing operands. The described authorization procedures prevent an unauthorized program from using an ALET.

Some access list entries may be designated by their owners at the time of creation either as private entries to provide address space access only by the owner or an authorizea user, or as public entries open to all users.
In the case of public entries ~P-bit, bit 7 is zero), the ALE is open and free to be used by any program. If the P-bit (bit 7J of the ALE is set to one, the ALE is to be used only by authorized programs. The control program provides facilities for adding entries to the AT

:' of the associated address space if more than one EAX is to be allowed to use the ALE.

Fig. 18 is a diagrammatic illustration of the access-register translation with program authorization ~5 checks. When an ALET is used in an access register i operation to fetch or store an operand, bits 0-6 of the ALET are examined at 115 to insure that the ALET is valid. If the P-bit 116 in the ALET is 0, the access list is a DUAL, and if the P-bit 116 is 1, the access list is a PSAL. If the access list is a ~UAL, the effective ALD is fetched from the DUCT whose address is stored in control register 2. If the access list is a PSAL, the effective ALD is fetched ~rom the primary ASTE
(PASTE) whose address is stored in control register 5.
The effective ALD includes an access list origin and an access list length (ALL). At 117, the ALEN is compared to the ALL to determine that the ALEN is not outside the bounds of the access list. If the ALEN passes this validity check, the effective access-list origin is ~20 joined with the ALEN by an adder operation at 119 to ; find the address of the ALE 120 in the access list 121.
The invalid bit, bit 0 of the ALE 120, is checked at 121 to see if it is 0, thereby determining if the ALE 120 is valid. If the ALE 120 is valid, the ALESN 122 of the ~25 ALET is compared to the ALESN 123 of the ALE 120 at 124.
If the ALESN 122 is equal to the ALESN 123, the ALET is still authorized to designate the ALE 120, and the ASTE
address 125 is used to fetch the ASTE 126. The validity of the ASTE 126 is confirmed by checking the invalid bit 127 at 128. If the ASTE 126 is valid, the ASTESN 130 is compared with the ASTESN 131 at 132 to insure that the ALE 120 is still authorized to designate the ASTE 126.
These checks complete the validity portlon o the ART.

~ .

.~.2~3~1~

The authority of the calling program to access the address space is now checked. The first check is made at 135 to determine if ~he P bit 136 is 0. I~ the P bit of 136 is 0, all programs are authorized to access the address space associated with the ALE, and no further checks are made. ~If the P bit 136 is 1, the ALEAX 137 is compared to the EAX 138 in control register 8 by the comparator 139. If the comparison at 139 is equal, then the program is specifically authorized to access ~he address space, and no further checks are made. If the comparison at 139 is not equal, then an ASN extended authorization check is made at 140. The ASN extended authorization check 140 is made by comparing the EAX in control register 8 with the authority table length (ATL) lS 141 to make sure that the EAX does not designate an entry outside of the bounds of the author~ty table. The EAX located in control register 8 is used as an index - into the authority table whose origin is ATO 142. If the S bit in the authority table is set equal to 1 for that EAX, then the program is authorized to have access into the address space. If the program is authorized to have access to the address space, as described, the STD
144 is provided ~or the DAT operation at 14S.

The private bit and the ALEAX field in the access list cntry provide high performance authorization - ~ mechanisms to grant or prohibi~ a program's access to an ; address space represented by the ALE. The private bit can be 0, thus allowin~ all programs which execute with the access list to access the address space represented by the ALE. The ALE private bit can be 1 and the user's EAX in control register 8 can be equal to the ALEAX
field. This allows programs with a particular EAX to access the address space represented by the ALE.
Finally, the ALE private bit can be one and the user's ~2~3 , control register 8 EAX can select an entry in the target space's authority table which has the S-bit equal to one. This allows multiple programs running with ; different EAXs to access the address space represented by the ALE.
~ f.
Fig. l9A and l9B, when taken together, form a flow chart of the access register translation steps and exceptions. When the ART logic is invoked, a check is ~ made at 150 to determine if access register 0 has been designated. If access register 0 has been designated, a check is made at 151 to determine if the ART was invoked by a TEST ACCESS operation (to be described). If access register 0 was not designated, or if this is a TEST
ACC~SS operation, the ALET in the access register is designated for use at 152. If access register 0 is designated and this is not a T,EST ACCESS operation, a ~ ; 00000000 hex is assigned to the ALET at 153. A check is 1~ made at 154 to determine if the ALET is equal to 00000000 hex. If yes, the STD for the primary address space is obtained from control register 1 at 155. At 156, a check is made to determine if the ALET has a value of 00000001 hex. If yes, the STD for the secondary address space is obtained from control register 7 at 157. A check is made at 158 to determine if bits 0-6 of the ALET are equal to 0. If bits 0-6 are not equal to 0, the assigned value of the ALET is not valid and an ALET specification exception is raised at 159 and the operation is suppressed.

A check is made at 160 to determine if the ALET bit ~30 7 is 1. If it is, the PASTEO entry in control register f~ 5 is decoded at 161 and the effecti~e ALD is fetched for the PSAL. If the ALET bit 7 is equal to 0, the~ DUCTO
entry in control register 2~is decoded at 162, and the effective ALD is ~etched for the DUAL. If the fetching ~2~38:~

address is not valid at 163, an addressing exception is raised at 164, and the operation is suppressed. If the address is valid at 163 r a check is made at 165 to determine if the ALEN of the ALET is outside the range of the effective ALL (bits 25-31 of the effective ALD).
If it is, an ALEN translation exception is raised at 166, and operation is nullified. If the answer at 165 is no, the ALE is located at 167 and a check is made to see if the ALE address is valid. If the ALE address is not valid, an addressing exception is raised at 168 and the operation is suppressed. If the address is valid at 167, the valid bit in the ALE is checked at 169 to see if the ALE is valid. If the ALE is not valid, an ALEN
translation exception is recognized at 170, and the operations is nullified. If the ALE is valid at 169, the ALESN of the ALET is compared to the ALESN of the ALE at 171. If the comparison at 171 is not equal, an ALE sequence exception is recognized at 172, and the ;~ operation is nullified. If there is an equal compare at 171, the ASTE is located at 173 using the ASTE address in the ALE. A check is made to determine if the ASTE
address is valid. If the address is not valid, an addressing exception is raised at 174 and the operation is suppressed. If the ASTE address is valid at 173, the validity bit of the ASTE is checked at 175 to determine if the ASTE is valid. If the ASTE is not valid, an ASTE
validity exception is raised at 176, and the operation is nullified. At 177, the ASTESN of the ALE is compared with the ASTESN of the ASTE. If there is not an equal ~30 comparison at 177, an ASTE sequence exception is raised at 178, and the operation is nullified.

The previous blocks 163-178 thus determine if the entries obtained are valid. At 179, the private bit of the ALE, bit 7, is checked to see if it is equal to 0.

POg87-007 12~31 31~L
;

Also at 179, the ALEAX entry in the ALE is compared to the EAX in control register 8. If either of the checks are equal, the STD for the operand is obtained from the ASTE of the address space, as shown at 180. When the private bit is 0, the program is authorized, and the authorization step of the access register translation is ; completed. When the private bit is 1 but the ALEAX is ~ equal to the EAX, the program is also authorized, and - the authorization step of the access register translation is completed.

If the program is not yet authorized at 179, then at 181, the validity of the ASTE is checked by determining if the ASTE bits 30, 31, and 60-63 are 0.
If not, an ASN translation specification exception is raised at 182 and the operation is suppressed. At 183, the value of the EAX bits 0-11 in control register 8 is compared against the length of the authority table to make sure that the EAX does not designate an entry outside of the bounds of the authority table. If the ~20 comparison at 183 is yes, an extended authorization exception is raised at 184 and the operation is suppressed. If the EAX designates an entry within the bounds of the authority table, the associated EAX entry is located in the authority table at 185. If the ~25 address of the authority table entry is not valid, an addressing exception is raised at 186 and the operation is suppressed. An extended authorization check is made at 187 by determining if the secondarv authorization bit tS-bit) of the authority table entry located at 185 is equal to 1. If the check at 187 is yes, the program is one of those authorized by the authority table associated with the address space, and the STD for the address space is obtained from the ASTE at 188. If the comparison at 187 is no, the program is not authorized 1~93~

and an extended authority exception is recognized at 189, and the operation is nullified.

The MAS facility includes a TEST ACCESS REGISTER
(TAR) instruction for performing the mentioned test access operation. TEST ACCESS has the following formatO

TAR A1, R2 The ALET specified as being in the access register of the first operand A1 is checked for ALET translation exceptions using the EAX in the general register specified by the second operand R2. As shown in Figs.
l9A and 19B, the TEST ACCESS REGISTER instruction, as d~termined in 151 of Fig. l9A, causes an ART operation to be performed. The TAR instruction returns the following results of the test in the PSW condition code ~15 (CC) see Fig. 4.

O = ALET specified is 0 and is valid for access.
1 = ALET specified is not 0 or 1, is in the DUAL
addressed by control reyister 2, and is valid for access with the specified EAX.
2 = ALET specified is not 0 or 1, is in the PSAL
addressed by control means of register 5, and is valid for access with the specified EAX.
3 = ALET specified either is 1 or is invalid for access with the specified EAX.

~25 The ability to test an ALET for authorization exceptions using an input EAX allows the program to determine if the ALET references the caller's PASN
(ALET=0), or if the ALET references the DUAL, or if the ALET references the caller's PSAL. This allows the , . .

3t~
' program to be independent of the internal format of the ALET.

~hen the TAR instruction is used and ART is pexformed, an A~B entry is created. Thus, when th~ ALET
in the AR is actually used, the ALB contains the entry, provided no exception occurred during ART.

igO 20 shows an example use of the TAR
instruction. A dispatchable unit task con~rol block TCBl, while executing at 200, has an EAX of 5. This EAX
allows the program to use specific entries on its dispatchable unit access list for TCBl. At 201, the first program makes a PC call to a second program which resides in address space ASN2, and the first program passes an A~ET which the second program in ASN2 must 1 15 use. At 202, the program in ASN2 is executed with an EAX = 8, which is different from its caller's EAX. If at 202, the program used the ALET provided by the ~; calling program, there could be a system integrity ~1 problem. The calling program may not have had the EAX
~j20 authority to reference the ALET, but the ASN2 program does. The ASN2 program must perform a validity check to determine if the caller had the authority to use the Z ALET that is passed. At 203, the program in ASN2 makes the validity check using the TAR instruction with the input ALET and the EAX = S of the caller. The caller's EAX is obtained from the linkage stac~ entry made on the program call to ASN2. If the TAR instruction gives a condition code which states that the caller was authorized to use the ALET, then the ASN2 program will continue to perform its function. If the caller was not authorized, then the ASN2 program will either A~END the . caller, or return to the caller with a return code which indicates that the call was not success~ul`. When PO~87-007 1~93~

control returns at 204 by means of a RETURM instruction, the callers EAX (EAX=5) is restored from the stack, and the ASNl program continues to execute with that EAX.

The ALET validity check function is needed quite frequently. In the example of Fig. 20, it is needed on every call to the program in ASN2. This function could be provided by an opexating system service routine, however, the performance overhead would be excessive.
If the TAR function is not provided, proyrams which must reference a caller's ALET and change the EAX, may need to use two PC instructions. The first PC would not change the EAX and the caller's parameters would be refexenced ~ith the callers EAX. Later, a second PC
~ould be executed to provide the new EAX for the called program to use. The called service may xequire a different EAX to do its function and this mechanism allows use of the correct EAX. The TAR function thus provides a more efficient performance. It will be understood that, although the example of Fig. 20 shows the TAR instruction used with an ALET on the DUAL, the ~` TAR instruction can be used with ALETs on both the DUAL
and PSAL.

Referring to Figs. 21-2S, the access register translation (ART) mechanism normally is implemented such ~25 that access list designations and information specified in access lists, ASN second tables, and authority tables are maintained in a special buffer, referred to as the ART lookaside buffer (ALB) previously shown at 199 in Fig. 2. Access list designations, access list entries, ASN second table entries, and authority table entries are collectively referred to as ART table entries. The CPU necessarily refers to an ART table entry in real storage only for the initial access to that entry. The PO987-0~7 3~

information in the entry may be placed in the ALB, and subsequent ART operations may be performed using the information in the ALB. The presence o~ the ALB affects the ART process to the extent that a modif:ication of an ALD, ALE, ASTE or ATE entry in real storage does not necessarily have an immediate, if any, effect on the translation.

The size and the structure of the AL~ depend on ~arious possible embodiments. For instance, the ALB may be implemented such as to contain at most 15 entries corresponding one to one with access registers 1-15, with each entry consisting of only a segment table designation (see Fig. 24); or it may contain arrays of values which are selected on the basis of an ALET, the ~15 cuxrent dispatchable unit control table origin or primary ASTE origin, and the current extended authorization index. In the first case, an ALB entry is cleared when the corresponding access register is reloaded, and the entire ALB is cleared upon a change to the contents of control register 2, 5 or 8. In the second case, information in the ALB persists despite changes of access register contents or control register contents.

Entries within the ALB are not explicitly ~25 addressable by the pro~ram.

Information is not necessarily retained in the ALB
under all conditions for which such retention is possible. Furthermore, information in the ALB may be cleared under conditions additional to those for which ~30 clearing is mandatory. All information in the ALB is necessarily cleared onlv by execution of PURGE ALB or SET PREFIX or by a CPU resét.

An ALB entry contains information fetched from an ART table entry in real storage and al50 the information used to select the ART table entry in real storage. An ; access list designation source origin (ALDSO) is used to select an ALD in real storage. The ALDSO is the dispatchable unit control table origin in control register 2 if the primary list bit in the ALET being translated is zero, or it is the primary ASN second table entry origin in control register 5 if the primary list bit in the ALET is one.

The access list origin part of an ALD, along with an ALET is used to select an ALE in real storage.

The ASTE address in an ALE is used to select an ASTE in real storage.

The authority table origin in an ASTE, along with the EAX in control register 8, is used to select an ATE
in real storage.

Referring to Fig. 21, in a first embodiment of an ALB, the ALB-ALD and the ALB ALE are combined into an ALB-ALD/ALE so that the access list origin need not be in the ALB entry. If the ALDSO and ALET for the ART
request match the content of the ALB entry, then the ALB
provides the following information: P bit, the ALEAX, the ASTE address and the ASTESN, all from the ALE.
Thus, this entry type allows the verification of authority to be relooked at and access to ASTE for the STD, with verification of the ASTESN entry, to be made at time of use. This provides efficient use of the ALB
because different ALETS may point to the same ASTE and ~381~

-54.1-STD. Thus, thi~ design of the ALB substitutes for use of the proper access list and determination of the ALE.
~owever, the ASTE and authority mechanisms are used as before.

Referring now to Fig. 2~, the function of the ASTE
may also be combined into a different embodiment of the ALB so that the STD is directly obtained from the ALB.
Thus, the ALDS~ and ALET, if a match exists in the ALB, provide the following information: P-bit, ALEAX, ;10 authority table origin (ATO), authority table length (ATL) and STD. Thus, the ASTE is not accessed for the STD, ATO and ATL because it has been retained in the ; ALB. However, if an ASTESN i5 changed in the ASTE, the ALB must be purged because the ALB contains the STD
`~ lS without reverification of the capability through the ASTESN.

i Referring to Fig. 23 a third embodiment of an ALB
combines into a single ALB entry the information and the attributes from the ALD, ALE, ASTE and ATE so that the ALB entry shown in the figure is all that is necessary.

~293~
, To further simplify, if the embodiment of Fig. 23 automatically clears the ALB of all entries whenever an ALD source origin is changed in control register 2 or 5, then the ALDSO field is not required in any AL~ entry.
As an additional simplification, the P, ALEAX, EAX and S
fields need not be implemented if the machine clears the ALB of entries whenever the EAX field is changed in control register 8. Such a simplified embodiment is shown in Fig. 4 where each A1ET simply fetches an STD, ` lO dependent on the necessary ALB purge operation~ to protect the STD's from improper use. Finally, if each entry corresponds one to one with one of access registers 1-15 and is cleared when the access register is r~loaded, the ALET field is not required.

Translations of ALET values of zero and one are not permitted to use the ALB . I f the actual implementation - has additional copies of the contents of control registers 1 and 7, the machine may have to perform some - type of special action in order to track changes to these control registers.

The formation of ALB entries and the effect of any manipulation of an ART table entry in real storage by the program depend on whether the ART table entry is attached to a particular CPU and on whether the entry is ~25 valid.
.
The attached state of an ART tab~e entrv denotes that the CPU to which the entry is attached can attempt to use the entry for access register translation. The ART table entry may be attached to more than one CPU at ~30 a time.

o~

~2938~l~

An access list entrv or ASN second ~able entry is valid when the invalid bit associated with the entry is zero. Access list designations and authority table entries have no invalid bit and are always valid. The primary space access list designation is valid regardless of the~value of the invalid bit in the primary ASTE.

An ART table entry may be placed in the ALB
whenever the entry is attached and valid. An access list designation is attached to a CPU when the designation is within the dispatchable unit control table specified by the dispatchable unit control table origin in control regis~er 2 or is within the primary ASTE specifi~d by the primary ASTE origin ~PASTEO) in control register 5. Control register 5 is considered to ~ontain the primary ASTE origin regardless of the value of the multiple address space control, bit 15 of control regi~ter 0.
: .
Referring now to Fig. 25, the preferred embodiment ~f an ALB is shown in which the ALB consists of several different tables which are accessed separately and sequentially during ALB usage and can thus provide more than one path to an STD. In the first step, a table or array referred to as an AL~-ALD/ALE table is accessed by an entry consisting of an ALDSO and an ALET which is compared (in the blocks labeled C) with all table entries and if a match is found the correct result is gated at blocks G to the ne~.t step. The ALB-ALD/ALE
table entries provide as resultant information the P-bit (private bit), the ALEAX, the ASTE address and the ~` ASTESN. In the second step, the ASTE address is used as the search tcrm in an ALB-ASTE table or array, again shown by the C compare blocks. If- a match is found, an ~,. . .

~2~38~

ALB-ASTE entry is gated, as shown by ~, consisting of an ASTESN, an ATO, and ATL and an STD. The ASTESN is in turn compared with the ASTESN provid,ed from the ALB-ALD/ALE entry and there must be a match or the ALB
process will not continue. If the P bit from the ALB-ALD/ALE entry~is one and the EAX in CR8 does not e~ual the ALEAX in the ALB-A~D/ALE entry, then, as a last step, authori.y is checked using an ALB-~TE table or array. The ATO and EAX are together used as a searcX
key. The EAX is determined from control register 8 and the ATO is used from the result found in the second step~ Again if a match is found, an entry consisting of , a secondary bit, S bit, from the ATE is gated as the result for testing authority.

An ASN second table entry is attached to a CPU when it is designated b~ the ASTE address in either an ALB~ALD/ALE array entry or an attached and valid ALE.

An authority table entry is attached to a CPU when it is within the authority table designated by either an ALB-ASTE array entry or an attached and valid ASTE.
' .
An ALB-ALD/AI,E entry may be used for ART onlv when all of the following conditions are met:

1. The ALET to be translated has a value larger than 1. (If the ALET is O or 1, the contents of CRl or CR7 are used.) .
2. The ALDSO field in the ALB-ALD/ALE matches the ALDSO being used.

; 3. The ALET field in the ALB-ALD/ALE matches the ALET to be translated.
' POg87-007 3~1 -~8-4. The ALB~ALD/ALE entry passes the ALE
authorization test; that is, one of the following conditions is true:

a. The private bit in the ALB-ALD/ALE entry is zero.
b. The ALEAX in the ALB-ALD/ALE entry equals the current EAX.
c. The current EAX selects a secondary bitls) that is one for the authority table designated by the ASTE that is addressed by the ALB-ALD/ALE.

An A~B-ASTE entry may be used ~or ART whenever the ASTE address and ASTESN in the ALB-ASTE entry match the ASTE address and ASTES~ in the ALE or ALB-ALD/ALE being used.
, In addition, two or more AL~-ALD/A~E entries may designate the same ALB-ASTE entry, thus providing more paths to the ALB-ASTE array and jus~ifying the use of separate array types in the ALB.

An ALB-ATE entry may be used for ART when both of the following conditions are met:
~ . .
1. The ATO in the ALB-ATE entry matches the AT0 in the ASTE or ALB-ASTE entry being used.
:
2~ The EAX in the AEB-ATE entry matches the current EAX.

When an attached but invalid ART table entry is made valid, or when an unattached but valid ART table entry is made attached, and no usable entry formed from the ART table entry is already in the ALB, the change ~30 takes effect no later than the end of the current instruction.

3~

The contents of the ALB need not be affected by a change of AR contents. The ALB can contain information pertaining to different AR contents or difEerent EAX
domains having different dispatchable units all at the same time. If a task is redispa~ched after being undispatched the ALB may still contain usable entries for ART.

When an attached and valid ~RT table entry is changed, and when, before the ALB is cleared of copies ~10 of that entry, an attempt is made to perform ART
requiring that entry, unpredictable results may occur, to the following extent. The use of the new value may begin between instruction~ or during the execution of an instruction, including the instruction that caused the ~15 change. Moreover, until the ALB is cleared of copies of the entry, the ALB may contain both the old and the new - values, and it is unpredictable whether the old or new value is selected for a particular ART operation.

When LOAD ACCESS MULTIPLE ox LOAD CONTROL changes the parameters associated with ART, the values of these parameters at the start of the operation are in effect for the duration of the operation.

; All entries are cleared from the ALB by the execution of PURGE ALB and SET PREFIX instructions and by a CPU reset. These instructions will have to be used ~ to prevent undesired conditions in the ALB.

.: .
~' .

7~ pOgg7-007 1 ~3~

The multiple address space ~MAS1 facility offers improvements in two major areas:

l. Data Accessing: Data in up to 16 different address spaces, including the instruction space, can be accessed concurrently by the program without changing any control parameters. This facility is provided by ~; means of 16 new registers named access registers. Stil~
more address spaces can be accessed by changing the contents of the access registers.

2. Program Linkage: The contents of an entry table entry are extended to allow increased status changing during a program call operation. A linkage stack is provided for saving status during program call and for restoring it by means of a new instruction named program return. There is also a new branch type linkage that uses the linkage stack.

MAS provides sixteen 32-bit access registers numbered 0-lS. Access registers are used to address storage operands in a new addressing mode called the access register mode. The access register mode results from new bit settings in the PSW.

In the access register mode, an instruction B or R
field that designates a general register containing a ~ storage operand address also designates the same numbered acccss register. The contents of the access register are used in a process called access register translation (ART) to obtain the segment table designation that will be used to translate, by means of ~ DAT, the storage operand address.

: '' ~' ~, Po987-007 ~Z~31313L
.

~ An address space specified by means of an access ;~ register is called an AR specified address space.

Access registers apply onlv to data addresses, not instruction addresses. Tn the access register mode, instructions are always fetched from the primary address space. (It is not possible to branch from one address space to another.) ;
The contents of the access register designated by ;~ the X field of a format RX instruction are ignored; only the access register designated by the B or R field is used in ART.

Through the use of access registers, data can be moved between any two address spaces and the complete instruction set can be used to operate on data in multiple different spaces all at the same time.
: . .
-The DAS instructions Move to Primary and Move to Secondary are not allowed to be executed in the access register mode. However, the DAS instruction Move with Key can be executed, so that the ability to have different access keys for the source and target data areas still is available.

The contents of an access register are called an access-list-entry token (ALET) because, in the general case, the~y designate an entry in a data area called an access list. ART uses the contents of the designated acce~s list entry to obtain the segment table designation that will be used by DAT.
.
The term "token" is used because an ALET does not directly convev any capability to access an address 3~1il space; it onl~ designates an access list entry, which represents the actual capability.

ALETs are manipulable as ordinary data. MA~
includes ins~ructions for transferring ALETs between access registers,~general registers, and storage.
Specifically, a called program can save the contents of the access registers in storage, load the access registers for its own purposes, and then restore the original contents so that the calling program will find them unchanged.

An ALET can be transferred to and from access register 0 e~en though access register 0 does not participate in the addrescing of a storage operand.

There are two special values of the ALET, 0 and 1, that specify the primary space and secondary space, respectively, without the use of an access list entry.
Thus, a program can have access to its own instruction address space without the need to form an access list entry that designates the space, and, after a space switching program call, the called pro~ram can similarly have access to the caller's space. A called program can be denied access to its callers space.
' ' E~tries in the access list are the addressing capabilities that are usable by means of access r 35 registers. The access list is intended to be protected from the problem state program to ensure the integrity of the addressing capabilities.
~ - .
The control program will provide a service that allocates an access list entry and returns an ALET
4~ designating the entry. The ALET can then be used by the 1~5a3~

requesting program to access the address space designated by the entry. The control program will also provide a service for deallocating an access list entry so the entry can be reused.
.~
An access list entry is marked invalid when it is not in the allocated state. An exception is recognized on an attempt to use an invalid access list entry.

There are actually two access lists available to a program at the same time. One is called the dispatchable unit access list and the other the primary space access list. The dispatchable unit access list is intended to be permanently associated with the dispatchable unit (the architectural term meaning "task"
or "process") on behalf of which the program is being executed. The primary space access list is a property of the primary address space in which the program is being executed. A bit in the ALET specifies which one of the dispatchable unit and primary space access lists is designated by the ALET.

A bit in the access list entry specifies whether the entry is public or private. No authorization is required for the use of a public access list entry. The use of a private access list entry must he authoriæed by an extended authorization index (EAX). The extended authorization index may be a property of either the dispatchable unit or the program, as will be described.
~; It is not a property of the primary space in which the program i5 executed.

Through the use of the extended authorization index, an entry in a dispatchable unit access list may he usable by some, but not all, of the programs that are .
'' 12~3~

executed to perform the work of the dispatchable unit.
Similarly, an entry on a primary space access list may be usable by some, but not all, of the programs that are executed in the corresponding primary space.

~5 The DAS autho~rization index has a bea:ring on the use of access registers since it authorizes the use of set secondary ASN in establishing a secondary space, and the secondary space can be accessed by means of an ALET
of 1. As has been said, the authorization index is a property of the primary space.

With MAS, program call is changed to test a new bit, named the PC type bit, in the entry table entry.
If this bit is zero, program call performs the DA5 operation described in DAS program linkage which is now ~15 called the basic operation. If the bit is one, program call performs a new operation called the stacking operation. The stacking operation makes some state changes differently than the basic operation, and it saves the old state in an entry it forms in a linkage stack. The linkage stack state entry is logically deleted, and the old state is restored, by a new instruction named program return.
, It is intended that there be a separate linkage stack for each dispatchable unit and that the linkage stack be protected rom direct manipulation hy the dispatchable unit. MAS includes instructions for extracting information from a state entry and for modifying one field in the entry.

~ M~S includes the branch and stack instruction, ;30 which may be used in place of branch and link. The only ; state information changed by branch and stack is the - .

~2~3~11 instruction address in the PSW. Branch ancl stack forms a state entry, called a branch state entry, that is the same as a program call state entry, except that it indicates that it was formed by branch and stack and ~5 contains the branch address instead of a PC number.
, The addressing mode bit and instruction address that are part of the complete PSW saved in a branch state entry can be either the current values in the PSW
or can be specified in a register as an operand of ~10 branch and stack. This register can be one that had link information placed in it by a branch and link, branch and save, branch and save and set mode, or branch and set mode instruction. Thus, branch and stack can be used either in a calling program or at (or near) the entry point of a called program, and in either case, a pxogram return at the end of the called program will return correctly to the calling program. The ability to use branch and stack at an entry point allows the linkage stack to be used without changing old calling programs.
:, The MAS instruction program return (PR) is used to return from a program given control b~ means of either stacking program call or branch and stack. Program return logically deletes the last linkage stack state entrv, which may be either a program call state entry or a branch state entry. If the last state entry is a program call state entry, program return restores all of the state information that was saved in the entry, except that it leaves the contents of general registers ~30 15, 0 and 1 and access registers 15, 0 and l unchanged.
If the last state entry is a branch state entry, program return restores only the complete PSW and the contents of general registers 2-14 and access registers 2-14.

, ~2~3~1~

However, program return always leaves the PER mask in the PSW unchanged in order not to counteract a PER
enablement or disablement that may have occurred while the called program was being executed.

A bit can be set to one in a linkage stack state entry to cause a program interruption if program return operates on the entry. The control program may set this bit to one to guard against an erroneous use of program return, for e~ample, when the last linkage instruction executed was a supervisor call instruction in which case the e~it service of the control program should be used before program return.

When a job step is started, which at least initially is a single dispatchable unit, it does so in ~,15 an address space that is unique to the job step. This address space is called the home address space of the job step. The system places the principal control blocks that represent the job step (for example, where ~ status is saved when the job step is undispatched) in the home address space o~ the job step. If the job step uses program call to giv~ control to another space and then an I/O or external interruption occurs, the control register contents must, without ~AS, be changed in order to gain access to the home address space so it can save the status of the step.

To improve the efficiency of accessing the home address space, ~S includes a home segment table designation and another address space mode, named the home space mode, which is conditioned by bit settings in the PSW. The new PSW that is loaded by the machine when an interruption occurs can specify the home space mode to provide immediate access to the home address space.

PO9~7-007 1;;~5~381 1 Access registers are 32-bit hardware re~isters available to the problem program. An access register ~,AR) may be used to associate an operand base register with an address space when storage is referenced. The basic function of ARs is to extend the 370-XA
instruction set to operate on instructions and storage operands in multiple spaces.

There are sixteen ARs, each one being directly associated with a GPR; i.e., AR0 with GPR0, ARl with GPRl, .. ..., AR15 wlth GPR15. ARs are only involved in the address~ng mechanism when the CPU is running in access rcgis~er mode as determined by program status word (PSW) bits 16 and 17.
, .
The general attributes of ARs are as follows:

1. The contents of access registers may be freely manipulated by a program in problem or supervisor state, whether in access register mode or not.
; ~
2~ Instructions in the architecture are provided to load and store AR contents from storage, transfer the contents from ARs to GPRs and vice versa, and to copy ; values from one access register to another.

3. The content of an access register is a token which can determine an address space via a hardware table lookup process: access register translation ART.
This token is called an Access List Entry Token (ALET).
' 4. The hardware associates the ALET value, in an AR, with an address space when storage references are made in access re~ister mode. The access register is implicitly determined in the hardware bv the base " ~

1~3~

register field of the instruction when used. The implicit designation of the AR allows the multiple address space access function to be extended to existing 370-XA instructions without modifying their machine code format.
5. The AR corresponding to the GPR specified in the index register field of an RX instruction does not participate in the selection of an address space.
6. All instructions and the target of an Execute instruction are always fetched from the primany address space ~7hen running in access register mode.
7. The same ALET value can be in more than one AR.
8. AR usage for addressing is done only in access register mode, when PSW bits 16 and 17 are 0 and 1 respectively.

The Access List ~A~) is an addressing capability table that is used with access registers ~Rs) and which is in the form of a dispatchable unit access list (DUAL) or a primary space access list (PSAL). The entries in the AL define the address spaces that can be addressed j via ARs for a given DU. When a storage reference instruction is executed in access register mode, the base register field of each operand is associated with ~25 an entry in the AL determined by the Access List Entry Token (ALET) contained in the corresponding access register.

An access list represents a list of addressing capabilities. These capabilities define address spaces , PO987~007 1293~1~

that can be accessed by the associated dispatchable unit. During addressing in access register mode, access . list entries provide the means for the hardware to : locate an alternate segment table origin to use for Dynamic Address Translation with respect to a storage operand of an instruction. An access list entry allows . -this by containing the real address of an Address Space Second Table Entry (ASTE) which in turn contains the addresses of the segment table and authority table associated with the address space.

.

3~1~

GLOSSARY

~ AKM Authorization Key Mask r AL Access List - An addressing capability table.

- AR Access Register - each access register is associated with a GPR.

ART Access Register Translation - A method of associating a STD - segment table designation with an access register.

AX Authorization Index .
~10 AL~ ART Lookaside Buffer - ART occurs each time an , .: .
~` AR is designated by a B field storage operand reference in a GPR, and the ALB reduces storage references during ART.

ALE Access List Entry ALEAX Access List Entry Authorization Index ':
ALEN Access List Entry Number - Bits 16-31 of the ALET are the access list entry number of the designated ALE.

A~L Access List Length - Stored in a control register as a predetermined number and can at most permit 1024 access list entries.

ALET Access List Entry Token - An ALET designates an entry in an access list.
,. .
ALESN Access List Entry Sequence Nu~ber - Bits 8-15 ~25 of the ALET and of the ALE.

~293~

ASN Address Space Number - Represents an address space.

ASTE ASN Second Table Entry - This is an expansion of the 370/XA AST~ shown in the prior art and includes~an I bit-and an STD.

ASTESN ASTE Sequence Number - The ASTESN in the ALE
is tested for equality with ASTESN in ASTE.

ATL Authority ~able Length.

DAS Dual Address Space DASD Direct Access storage device.
:
DAT Dynamic Address Translation - Uses an STD to convert virtual address to real storage addresses.
:
DUAL Dispatchable Unit Access List ,~
DUALD DUAL designation consisting of the real origin and length of the DUAL

DUCT Dispatchable Unit Control Table - contains DUALD and specified by CR2 EAX Extended Authorization Inde~

EK~I Entry Rey Mask ETE Entry Table Entry GPR General Purpose P.egister for containing operands and addresses ' ~LZ~?38~:~

-7~-LTD Linkage Table Designation MAS Multiple Address Space : P Bit Bit in AL~ that selects ketween DUAL and PSAL

PRIVATE-Bit Bit P in the ALE-that designates whether . 5 all users may have access or an authority mechanism is invoked.
, PASTE Primary ASN second table entry - contains PSAL
and LTD

PC-cp Program Call to Current Primary PC-ss Program Call with Space Switching PKM PSW Xey Mask . " .
PSAL ~rimary Space Access List i PSALD PSAL Designation consisting of the r~al origin : and length found in the primary ASTE

PSTD Primary Segment Table Designation PSW Program Status Word . ~
: SSTD Secondary Segment Table Designation STD Segment Table Designation , .
While the invention has been described with reference to the preferred embodimen's thereofj various : modifications and changes may be made by those skilled in the art without departing from the tr~e spirit and scope of the invention as defined by the claims hereof.
,

Claims (38)

1. A capability domain control mechanism in a computer system comprising:

a central processor unit;

a plurality of general purpose register (GPR) means connected to said central processor for holding at preselected times a first address component for virtual address translation;

a plurality of access register (AR) means connected to said central processor and having each of said access register means associated with one of said GPR means, said access registers containing means for identifying a second address component to be associated with said first address component in the GPR associated with said AR;

a first access list means containing a plurality of access list entries, each valid entry associated with an address to an address space, each access bit entry being capable of being indexed by the contents of an AR, said access list means being representative of a first domain of address spaces having predetermined capabilities;

a second access list means containing a plurality of access list entries, each valid entry associated with an address to an address space, each access list entry being capable of being indexed by the contents of an AR, said access list means being representative of a second domain of address spaces having predetermined capabilities;

indicating means for pointing to one of said access list means in response to the contents of a AR; and means for causing said second address component to be determined as a result of which access list means is pointed to by said indicating means to allow said system to switch capabilities by switching address domains.
2. The mechanism of Claim 1 wherein said first access list means is representative of a dispatchable unit access list domain; and wherein said second access list means is representative of a primary space access list domain;

wherein said system has more than one user each having different dispatchable unit access list domains unique to each said user and each user sharing said primary space access list domain.
3. The mechanism of Claim 2 wherein each of said access list means is uniquely identified by access list origin information;

wherein a plurality of control registers is associated with said control process and said access list origin information is stored in a specified control register associated with each access list means; and wherein the contents of each of said ARs contains information to identify the control register holding the access list origin information for the access list domain specified by said AR.
4. A method for controlling capability domains in a computer system having a plurality of general purpose registers (GPR) and a plurality of access registers associated with said general purpose registers comprising the steps of:

determining if said GPR holds a first address component;

determining from said access register associated with said GPR an access list pointing identifier where said access list represents a capability domain, if said GPR does hold a first address component;

determining from said access register associated with said GPR an access list entry identifier, if said GPR does hold a first address component;

selecting an access list representative of a capability domain from said access list pointing identifier;

selecting an access list entry from said selected access list, said access list entry containing means for identifying a second address component for determining real addresses from virtual addresses when used with said first component.
5. The method of Claim 4 wherein the selected access list domain is the dispatchable unit access list domain which has the capability of the task to be performed.
6. The method of Claim 4 wherein the selected access list is the primary space access list domain which has the capabilities of the primary address space.
7. The method of Claim 4 wherein the selected access list domain is the system wide access list (SWAL) domain which has the capabilities of all programs in the system.
8. The method of Claim 4 in which said computer system has control registers and in which said access list pointing identifier selects a control register having the address and size of the access list contained therein.
9. The method of Claim 4 wherein the selected access list domain is a secondary address space access list domain having the capabilities of a secondary address space number.
10. The method of Claim 4 wherein the capabilities of a user may be changed by changing from a first domain to a second domain.
11. The method of Claim 4 wherein the capabilities of all users of a domain are changed by changing the domain access list capabilities.
12. A capability domain control mechanism in a computer system comprising:

a plurality of general purpose register (GPR) means for containing a first component of a virtual address;

a plurality of access register (AR) means each associated with a particular one of said general register means, each of said AR means having the capability of holding an access list entry token (ALET);

a first access list means containing a plurality of access list entries, each valid entry associated with an address to an address space, each access list entry being capable of being indexed by an ALET from an AR, said access list means being representative of a first domain of address spaces having predetermined capabilities;

a second access list means containing a plurality of access list entries, each valid entry associated with an address to an address space, each access list entry being capable of being indexed by an ALET from an AR, said access list means being representative of a second domain of address spaces having predetermined capabilities;

first control register means for containing information to access said first access list means in response to the contents of an ALET from an AR;

second control register means for containing information to access said second access list means in response to the contents of an ALET from an AR;

selecting means for selecting said first control register means or said second control register means in response to the contents of an ALET from an AR to determine whether said first access list means or said second access list means is the selected access list for access list entries;

means for obtaining an access list entry from said selected access list corresponding to an ALET in an AR
means at the time the associated GPR means is activated in an addressing operation;

memory means for containing address space number second table entries (ASTE);

means for identifying an ASTE entry from said access list entry from said selected access list and for obtaining said ASTE, wherein said AST contains a second component for virtual address translation; and virtual address translation means for using said first and second components for virtual address translation to obtain real addresses.
13. The mechanism of Claim 12 wherein said first access list means is representative of a dispatchable unit access list domain;

wherein said second access list means is representative of a primary space access list domain;
and wherein said system has more than one user each having different dispatchable unit access list domains unique to each of said user and each user sharing said primary space access list domain.
14. The mechanism of Claim 12 wherein the capabilities of said first access list means and said second access list means are different and whereby the capabilities of the user may be changed by using an ALET
that changes the selected one of said access lists.
15. The mechanism of Claim 14 and further including:

access list capability changing means for changing the capabilities of an access list whereby the capabilities of all users of an access list may be changed by changing the capabilities of the access list.
16. The mechanism of Claim 12 and further including:

access list capability changing means for changing the capabilities of an access list whereby the capabilities of all users of an access list may be changed by changing the capabilities of the access list.
17. The mechanism of Claim 12 wherein one of said access lists is representative of a system wide access list (SWAL) domain having all capabilities of the system.
18. The mechanism of Claim 12 wherein one of said access lists is representative of a secondary address space access list (SSAL) domain having the capabilities of a secondary address space.
19. A computer system method for controlling the ability of an executing program to obtain access to data in a data domain, comprising:
recording program-address-space entries in a memory of the computer system in one or more system tables containing descriptors (STDs) for program address spaces, recording a plurality of addresses in relation to the STDs for locating associated access lists (PSALs) representing data domains having data required for the execution of program instruction in an associated one of the program address spaces, providing tokens for enabling an executing program to access data in an address space selected in a data domain defined by a PSAL of a current program address space having the executing program, each token having an AL-type indication field set to indicate which of plural types of access lists is to be accessed, calling a program for execution including loading an address into a program control register for defining a next program address space having a next program to execute, and concurrently loading a data domain control register with an address to a PSAL data domain associated with the next program address space, executing the next program in which a currently-executing instruction has an operand that selects one of a plurality of general purpose registers (GPRs) as a base GPR by using a base address component of a virtual address represented in the instruction operand, selecting an access register (AR) associated with the base GPR as a base AR containing one of the tokens previously loaded into the AR, selecting a control register associated with the token in the base AR, the selected control register containing an address defining a required data domain, the token locating a selected AL entry (ALE) in an access list addressed from the selected control register to obtain access to an address space having data required by the operand.
20. A computer system method for controlling the ability of an executing program to access a data domain as defined in claim 19 by using tokens in ARs, further comprising:
performing a program call instruction for calling a program in a next address space, processing the program call instruction to locate an entry in a system table containing a STD and an AL address, the AL address addressing a PSAL (primary space access list) assigned to the called program address space, accessing a PSAL control register in the processor, and loading the PSAL control register with an address for locating the entry in the system control table for the processor to obtain addressability to a PSAL data domain represented by address spaces located through the PSAL.
21. A computer system method for controlling the ability of an executing program to access a data domain as defined in claim 19 by using tokens in ARs, further comprising:
requesting a control program to provide a user with a token for a dispatchable user access list (DUAL) for providing the user with addressability to a user-indicated address space containing data to be accessed by the user, executing the control program to generate an AL
entry (ALE) in the DUAL in response to the request and to generate a token for locating the ALE, including the steps of:
finding an invalid ALE in the DUAL, locating information on the indicated address spaced in a system table, and writing the information in the ALE and setting the ALE to a valid state, generating a token containing at least a DUAL
indicator and an index to locate the ALE in the DUAL, and passing the token to the user for use by the user in any base AR; and accessing and loading a DUAL control register with an address for locating the DUAL to enable the processor to obtain addressability to a DUAL data domain of the user while the user is dispatched on a processor.
22. A computer system method for controlling the ability of an executing program to access different types of data domains as defined in claim 19 by using tokens in ARs, further comprising:
generating a plurality of ALs in the memory of the computer system that include a plurality of types of ALs, each type of AL representing a different type of data domain, loading a plurality of control registers in a processor with addresses that respectively provide addressability to ALs of the different types, and selecting a control register according to the AL-type field indicated in the token that restricts the token to be used only for accessing the type of AL
indicated by the field.
23. A computer system method for controlling the ability of an executing program to access different types of data domains as defined in claim 19, further comprising:
a dispatched program entity using the first type of AL (DUAL) for a DUAL type of data domain comprised of virtual address spaces which contain data accessible to a dispatched task for executing a job on a computer system that can involve a plurality of program address spaces, addressability to the DUAL domain being available to all programs executed for the dispatched task, and each currently-executing program in the dispatched task containing program instructions in a program address space having addressability to a second type of AL (PSAL) available only to the respective program address space, the PSAL providing a data domain exclusive to the currently-executing program address space.
24. A computer system method for obtaining domain control in a computer system as defined in claim 23, further comprising:
executing a series of program call instructions by a dispatched program entity for calling programs in different program address spaces, and sharing the addressability to data in the address spaces of the DUAL data domain among the program address spaces called by the dispatched program entity.
25. A computer system method for obtaining domain control in a computer system as defined in claim 24, further comprising:
executing dispatched program entities on different processors in the computer system which concurrently have addressability to different DUALs and PSALs while the processors are being controlled by dispatched program entities.
26. A computer system method for obtaining domain control in a computer system as defined in claim 25, each processor in a multiprocessing system further comprising:
assigning a first AL-type control register to contain addressability only to an AL associated with a currently-executing program address space for accessing an associated PSAL data domain, and also assigning a second AL-type control register to contain addressability to an AL providing accessibility to a DUAL data domain associated with all programs executable by a currently-dispatched program entity, the DUAL domain including all data address spaces associated with the currently-dispatched program entity.
27. A computer system method for obtaining domain control in a computer system as defined in claim 24, further comprising:
providing the addressability for the first AL-type control register by having an address therein for locating a PASTE (primary address space table entry) containing fields for defining a program address space in the system, and having a PSAL address in a predetermined field in the PASTE for locating the PSAL in MS, and obtaining the addressability for the second AL-type control register by having an address therein for locating a DUCT (dispatchable unit control table) containing entries for representing a dispatchable unit in the system, and having a DUAL address (DUCTO) at a predetermined location in the DUCT for locating the DUAL
in MS.
28. A computer system method for obtaining domain control in a computer system as defined in claim 24, further comprising:
terminating the ability of any program executing on a processor to access a data domain previously obtained by an AL address in an AL-type control register by reloading the content of the control register with another AL address for obtaining addressability to a different data domain.
29. A computer system method for quickly enabling and disabling a data domain associated with a virtual address space containing a set of executable program instructions (i.e. program address space), each data domain including a plurality of virtual address spaces providing independent ranges of virtual addresses for accessing data to be used by the operands of the instructions, the availability of data in a data domain only being allowed while an associated program address space is dispatched on a processor of the computer system, comprising:

storing a plurality of access lists (PSALS) in a memory of the processor, addresses (for locating PSALs in memory) being stored with descriptors of associated program address spaces, each PSAL providing a set of entries (ALES) representing respective accessing paths to data address spaces capable of containing data accessible to program instructions in the associated program address space, setting a valid indication for each ALE in a PSAL
when the ALE is in a valid state to indicate the ALE is usable for determining an associated address space for containing data, the ALE not being usable when the valid indication is set to an invalid state, loading an address from memory into a control register in a processor upon calling each next program address space to setup addressing to an associated PSAL
for providing an addressing path to a corresponding data domain for being accessed by instructions executed in the processor obtained from the next program address space, and the loaded address overlaying and destroying any address previously loaded in the control register in order to disable the ability of the instructions to access data from any other data domain than the currently-loaded domain, the loading operation providing single-address switching control for restricting the ability of instructions in any program address space in obtaining data required for execution of the instructions.
30. A computer system method as defined in claim 29 for quickly enabling and disabling a data domain on a processor from being accessible to a respectively associated program address space through a data access path (PSAL), further comprising:
providing in the PSAL the entries for paths to all address spaces capable of containing data accessible to a single program contained in the associated program address space.
31. A computer system method as defined in claim 29 having a multiplicity of central processing units (CPUs), each CPU having the ability to quickly enable and disable access to a data domain by loading a control register with addressability for a data access path (PSAL) to a data domain for an associated program address space, further comprising:
loading the same PSAL addressability in a control register in a plurality of the CPUs in the system, executing the same program or different programs in the same program address space concurrently in the plurality of CPUS, sharing the PSAL of the program address space by the programs concurrently executing on the plural CPUs by concurrently accessing entries in the PSAL for obtaining paths to address spaces containing data accessible to the different programs contained in the associated program address space, and the data in all of the address spaces accessible through the PSAL being shareable by all of the programs in the associated program address space executing on the plurality of CPUs.
32. A computer system method as defined in claim 29 for quickly enabling and disabling any data domain accessible to an associated program address space, further comprising:
providing an dispatchable unit access list (DUAL) associated with a dispatched programming job, the DUAL
containing a plurality of entries (ALES) providing addressability paths to address spaces capable of containing data usable by the programming job, the DUAL
defining address paths to all address spaces capable of containing data accessible to any program executed for the programming job, dispatching a program address space for initiating the execution of the dispatched programming job on a processor.

associating a first PSAL with an initiating program address space for the programming job to represent paths to all address spaces capable of containing data accessible to the initiating program address space, and calling a second program address space associated with a second PSAL to represent paths to all address spaces capable of containing data accessible to the second program address space being executed for the dispatched programming job, and the addressability to the first PSAL being replaced with addressability to the second PSAL to terminate addressability to the address spaces of the first PSAL.
33. A computer system method for controlling the ability of programs executed by any user dispatched on the system to access data, comprising:
recording in a system table a plurality of program-address-space entries in which each valid entry contains a descriptor (STD) for a program address space and an address for locating an associated access list (AL) defining a data domain for obtaining data needed for the execution of program instructions in the associated program address space, recording in a memory of the computer system a plurality of access lists (ALs) addressed by entries in the system table, providing tokens to each user prior to dispatching a user on the system for enabling a user dispatched task to access data in a user data domain defined by a dispatched access list (DUAL) associated with the dispatched task, each token having a DUAL indication field set to indicate that the user s DUAL is accessed by the token, executing in the computer system a plurality of user programs for the dispatched task, loading into a first control register of a processor for starting execution of any program in the dispatched task a designator of a program address space containing a program to execute on the processor, additionally loading into a PSAL (program space AL) control register in the processor in response to each program call instruction the address of a PSAL associated with the program address space designated in the first control register for connecting data to the processor required for execution of the program, entries in the PSAL containing addressing paths to virtual address spaces having data required by the program, and simultaneously disconnecting from the processor any accessibility to data obtainable through any other PSAL
previously connected for another program address space by the additional loading operation which overlays and destroys any previous address in the PSAL control register to restrict data accessing through the PSAL
assigned only to the current program address space.
34. A computer system method as defined in claim 33, further comprising:
permanently invalidating one or more ALEs at selected locations in a DUAL to allow tokens with predetermined values (that would select such DUAL
locations) to be used to directly select preassigned control registers that have predetermined addressability to respective data domains.
35. A computer system method for controlling the ability of an executing program to access data as defined in claim 34, comprising:
entering a user programming task into the computer system, the task indicating a sequence of one or more programs to be executed on a processor and indicating data files to be used by the executing programs, storing designators in the memory for program address spaces containing the programs indicated for the task, generating a task access list (DUAL) of addressing paths to address spaces which can contain data files of the task prior to executing any program for the task.

dispatching on a processor each program required for executing the task by loading into the first control register a designator of each program address space containing a program required by the task, and loading an address needed for locating the task access list (DUAL) in a DUAL control register in the processor, addressability to the DUAL remaining loaded in the DUAL control register while the task is dispatched to allow each program in a plurality of program address spaces sequentially designated in the first control register to have addressability to data in the DUAL
domain (any address space accessible through the DUAL) while also having addressability to the PSAL domain (any address space accessible through the PSAL) which is associated with one program address space, and reloading the PSAL control register with an address needed for locating another PSAL when the first control register is next loaded with another designator for another program address space during execution of the task, and the DUAL control register not being reloaded while the task is executing on the processor to continue addressability to the DUAL data domain during execution of the task, allowing a plurality of PSAL addresses to be loaded into the PSAL control register while a single DUAL
address is loaded in the DUAL control register.
36. A computer system method for controlling the ability of an executing program to access data as defined in claim 35, comprising:
writing an AL-type indicator in each token prior to use of the token by a task, the AL-type indicator indicating whether a PSAL or DUAL type of access list is to be used with each token, and providing the user with all tokens required for executing the task for enabling the programs of the task to select the ALEs in the DUAL
and in PSALs required for accessing address spaces having data needed by the operands of the instructions executed for the task.
37. Virtually addressed data domain control means for selecting one of plural address-translation tables from multiple data domains available to an executing program, the selected table to be used for translating a virtual address of an operand (of an instruction to be executed) to a real address for locating an operand in main storage (MS), the instruction being executed by a central processor (CP) in a data processing system, each address-translation table being located by a designator and being used for defining a virtual address space, the CP having general purpose registers (GPRs) and associated access registers (ARs), an instruction having one or more operands which may specify a virtual address by address components including a base GPR and optionally either one or both of a non-base GPR and a displacement field, CP
hardware/microcode means for selecting an address-translation table comprising:
plural control register means (CRs) for containing addresses to locate respective access lists (ALs), each AL defining a data domain of address spaces for containing data which can be made addressable to a program, means for indicating the validity of each ALE for representing an address space, means for selecting a token in a base AR associated with a base GPR selected by the base GPR address component in a current instruction operand, the token having fields for indicating the validity of the token and for selecting one of the CRs containing an address to locate a required AL and for selecting an entry (ALE) in the required AL to provide an address path to a designator that locates a respective address-translation table represented by the token, means for accessing the located address-translation table for performing an address translation of a virtual address to a real MS address for an instruction operand.
38. Virtually addressed data domain control means as defined in claim 37, further comprising:
the control register means addressing first and second access lists (ALs) including, a first access list of address spaces assigned to a dispatched user of programs executing on the data processing system, the first access list defining a data domain addressable by any program executed for the user through a one of the plural CRs, a second access list restricted for use only to program(s) in a program address space associated with the second access list, the second access list defining a data domain addressable only by the program(s) in the associated program address space through another of the plural CRs.
CA000583573A 1988-02-10 1988-11-18 Domain related access lists Expired - Fee Related CA1293811C (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US07/154,685 1988-02-10
US07/154,685 US4945480A (en) 1988-02-10 1988-02-10 Data domain switching on program address space switching and return

Publications (1)

Publication Number Publication Date
CA1293811C true CA1293811C (en) 1991-12-31

Family

ID=22552326

Family Applications (1)

Application Number Title Priority Date Filing Date
CA000583573A Expired - Fee Related CA1293811C (en) 1988-02-10 1988-11-18 Domain related access lists

Country Status (6)

Country Link
US (1) US4945480A (en)
EP (1) EP0331900B1 (en)
JP (1) JPH01207856A (en)
BR (1) BR8900567A (en)
CA (1) CA1293811C (en)
DE (1) DE68924720T2 (en)

Families Citing this family (39)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5210874A (en) * 1988-03-22 1993-05-11 Digital Equipment Corporation Cross-domain call system in a capability based digital data processing system
US5134696A (en) * 1988-07-28 1992-07-28 International Business Machines Corp. Virtual lookaside facility
JP2768503B2 (en) * 1989-07-25 1998-06-25 富士通株式会社 Virtual memory address space access control method
US5189733A (en) * 1989-08-22 1993-02-23 Borland International, Inc. Application program memory management system
JP2825550B2 (en) * 1989-09-21 1998-11-18 株式会社日立製作所 Multiple virtual space address control method and computer system
JPH0679296B2 (en) * 1989-09-22 1994-10-05 株式会社日立製作所 Multiple virtual address space access method and data processing device
US5522072A (en) * 1990-09-04 1996-05-28 At&T Corp. Arrangement for efficiently transferring program execution between subprograms
US5230069A (en) * 1990-10-02 1993-07-20 International Business Machines Corporation Apparatus and method for providing private and shared access to host address and data spaces by guest programs in a virtual machine computer system
EP0532744B1 (en) * 1991-03-07 1996-12-18 Digital Equipment Corporation Improved software debugging system and method especially adapted for code debugging within a multi-architecture environment
US5339422A (en) * 1991-03-07 1994-08-16 Digital Equipment Corporation System and method for jacketing cross-domain calls in a multi-code execution and debugging system within a multi-architecture environment
US5652869A (en) * 1991-03-07 1997-07-29 Digital Equipment Corporation System for executing and debugging multiple codes in a multi-architecture environment using jacketing means for jacketing the cross-domain calls
US5390312A (en) * 1991-09-24 1995-02-14 International Business Machines Corporation Access look-aside facility
GB2260004B (en) * 1991-09-30 1995-02-08 Apple Computer Memory management unit for a computer system
US5574936A (en) * 1992-01-02 1996-11-12 Amdahl Corporation Access control mechanism controlling access to and logical purging of access register translation lookaside buffer (ALB) in a computer system
US5361356A (en) * 1992-03-06 1994-11-01 International Business Machines Corporation Storage isolation with subspace-group facility
US5493661A (en) * 1992-03-06 1996-02-20 International Business Machines Corporation Method and system for providing a program call to a dispatchable unit's base space
US5479631A (en) * 1992-11-19 1995-12-26 International Business Machines Corporation System for designating real main storage addresses in instructions while dynamic address translation is on
US5584042A (en) * 1993-06-01 1996-12-10 International Business Machines Corporation Dynamic I/O data address relocation facility
JPH06348584A (en) * 1993-06-01 1994-12-22 Internatl Business Mach Corp <Ibm> Data processing system
US5611043A (en) * 1994-03-18 1997-03-11 Borland International, Inc. Debugger system and method for controlling child processes
US5729710A (en) * 1994-06-22 1998-03-17 International Business Machines Corporation Method and apparatus for management of mapped and unmapped regions of memory in a microkernel data processing system
US5590313A (en) * 1994-06-30 1996-12-31 International Business Machines Corporation Multiple protocol device interface subsystem and method
US5577231A (en) * 1994-12-06 1996-11-19 International Business Machines Corporation Storage access authorization controls in a computer system using dynamic translation of large addresses
US6681238B1 (en) * 1998-03-24 2004-01-20 International Business Machines Corporation Method and system for providing a hardware machine function in a protected virtual machine
US6832301B2 (en) * 2001-09-11 2004-12-14 International Business Machines Corporation Method for recovering memory
US6801993B2 (en) * 2001-09-28 2004-10-05 International Business Machines Corporation Table offset for shortening translation tables from their beginnings
US7533382B2 (en) * 2002-10-30 2009-05-12 Stmicroelectronics, Inc. Hyperprocessor
US6986017B2 (en) * 2003-04-24 2006-01-10 International Business Machines Corporation Buffer pre-registration
US7389411B2 (en) * 2003-08-29 2008-06-17 Sun Microsystems, Inc. Secure transfer of host identities
US7444396B2 (en) * 2003-08-29 2008-10-28 Sun Microsystems, Inc. Transferring system identities
US20090133022A1 (en) * 2007-11-15 2009-05-21 Karim Faraydon O Multiprocessing apparatus, system and method
US8307353B2 (en) * 2008-08-12 2012-11-06 Oracle America, Inc. Cross-domain inlining in a system virtual machine
JP2011139359A (en) * 2009-12-28 2011-07-14 Sharp Corp Display device, stopping method, program, and recording medium
US9037815B2 (en) * 2012-06-28 2015-05-19 International Business Machines Corporation Analyzing and reducing dump data
PL3204861T3 (en) 2015-12-15 2019-12-31 Lzlabs Gmbh Protection key management and prefixing in virtual address space legacy emulation system
US10891238B1 (en) 2019-06-28 2021-01-12 International Business Machines Corporation Dynamically joining and splitting dynamic address translation (DAT) tables based on operational context
US10970224B2 (en) 2019-06-28 2021-04-06 International Business Machines Corporation Operational context subspaces
US11074195B2 (en) 2019-06-28 2021-07-27 International Business Machines Corporation Access to dynamic address translation across multiple spaces for operational context subspaces
US11176056B2 (en) 2019-06-28 2021-11-16 International Business Machines Corporation Private space control within a common address space

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB1329721A (en) * 1970-05-26 1973-09-12 Plessey Co Ltd Data processing devices
FR2253425A5 (en) * 1973-11-30 1975-06-27 Honeywell Bull Soc Ind
US4044334A (en) * 1975-06-19 1977-08-23 Honeywell Information Systems, Inc. Database instruction unload
US4037214A (en) * 1976-04-30 1977-07-19 International Business Machines Corporation Key register controlled accessing system
JPS52149444A (en) * 1976-06-08 1977-12-12 Fujitsu Ltd Multiplex virtual space processing data processing system
US4136385A (en) * 1977-03-24 1979-01-23 International Business Machines Corporation Synonym control means for multiple virtual storage systems
US4096573A (en) * 1977-04-25 1978-06-20 International Business Machines Corporation DLAT Synonym control means for common portions of all address spaces
JPS5474651A (en) * 1977-11-28 1979-06-14 Toshiba Corp Stack control system
US4355355A (en) * 1980-03-19 1982-10-19 International Business Machines Corp. Address generating mechanism for multiple virtual spaces
US4366536A (en) * 1980-04-15 1982-12-28 National Semiconductor Corporation Modular digital computer system for storing and selecting data processing procedures and data
US4366537A (en) * 1980-05-23 1982-12-28 International Business Machines Corp. Authorization mechanism for transfer of program control or data between different address spaces having different storage protect keys
US4500952A (en) * 1980-05-23 1985-02-19 International Business Machines Corporation Mechanism for control of address translation by a program using a plurality of translation tables
US4430705A (en) * 1980-05-23 1984-02-07 International Business Machines Corp. Authorization mechanism for establishing addressability to information in another address space
US4521846A (en) * 1981-02-20 1985-06-04 International Business Machines Corporation Mechanism for accessing multiple virtual address spaces
US4455602A (en) * 1981-05-22 1984-06-19 Data General Corporation Digital data processing system having an I/O means using unique address providing and access priority control techniques
JPS57203279A (en) * 1981-06-10 1982-12-13 Hitachi Ltd Information processing device
US4454580A (en) * 1981-09-16 1984-06-12 International Business Machines Corporation Program call method and call instruction execution apparatus
JPS5853079A (en) * 1981-09-25 1983-03-29 Fujitsu Ltd Sto stack controlling system
US4731734A (en) * 1983-10-17 1988-03-15 Data General Corporation Digital computer system incorporating object-based addressing and access control and tables defining derivation of addresses of data from operands in instructions
JPS60225951A (en) * 1984-04-24 1985-11-11 Fujitsu Ltd Storage protecting system of expanded virtual storage space
EP0176939B1 (en) * 1984-09-28 1989-08-23 Siemens Aktiengesellschaft Data-processing systems with virtual storage addressing for a plurality of users

Also Published As

Publication number Publication date
EP0331900B1 (en) 1995-11-08
DE68924720T2 (en) 1996-06-13
US4945480A (en) 1990-07-31
JPH01207856A (en) 1989-08-21
BR8900567A (en) 1989-10-10
DE68924720D1 (en) 1995-12-14
JPH0571975B2 (en) 1993-10-08
EP0331900A3 (en) 1992-03-25
EP0331900A2 (en) 1989-09-13

Similar Documents

Publication Publication Date Title
CA1293811C (en) Domain related access lists
US5220669A (en) Linkage mechanism for program isolation
US4979098A (en) Multiple address space token designation, protection controls, designation translation and lookaside
US5023773A (en) Authorization for selective program access to data in multiple address spaces
US5809546A (en) Method for managing I/O buffers in shared storage by structuring buffer table having entries including storage keys for controlling accesses to the buffers
US9244856B2 (en) Dynamic address translation with translation table entry format control for identifying format of the translation table entry
US4521846A (en) Mechanism for accessing multiple virtual address spaces
US7213125B2 (en) Method for patching virtually aliased pages by a virtual-machine monitor
US8019964B2 (en) Dynamic address translation with DAT protection
US5845129A (en) Protection domains in a single address space
US8417916B2 (en) Perform frame management function instruction for setting storage keys and clearing blocks of main storage
US8117417B2 (en) Dynamic address translation with change record override
US5317717A (en) Apparatus and method for main memory unit protection using access and fault logic signals
EP0150521A2 (en) Data processing system with improved interrupt handling
JPS6022377B2 (en) Address control method
JPH0652050A (en) System for locking page of actual memory
US20090182974A1 (en) Dynamic address translation with access control
US5724551A (en) Method for managing I/O buffers in shared storage by structuring buffer table having entries include storage keys for controlling accesses to the buffers
US5787309A (en) Apparatus for protecting storage blocks from being accessed by unwanted I/O programs using I/O program keys and I/O storage keys having M number of bits
US5802397A (en) System for storage protection from unintended I/O access using I/O protection key by providing no control by I/O key entries over access by CP entity
EP0327839B1 (en) Information handling system
EP0297891B1 (en) Apparatus and method for main memory unit protection using access and fault logic signals
US5493661A (en) Method and system for providing a program call to a dispatchable unit&#39;s base space
CA1312142C (en) Linkage mechanism for program isolation

Legal Events

Date Code Title Description
MKLA Lapsed