Method of operating a data processing system having a dynamic software update facility

• US 5,359,730 A
• Filed: 12/04/1992
• Issued: 10/25/1994
• Est. Priority Date: 12/04/1992
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Marron, Assaf
• Primary Examiner(s): Heckler, Thomas M.
• Application Number:
  US07/985,762
• Time in Patent Office: 690 Days
• Field of Search: 364/DIG. 1 MS File, 395/650, 395/700
• US Class Current: 717/169
• CPC Class Codes: G06F 8/656 while running

Abstract

A dynamic software update facility (DSUF) is installed in a data processing system for the purpose of non-disruptively replacing old operating system programs or modules with new updated versions thereof while providing continuous availability and operation of the system. The new versions are loaded into the system along with change instructions providing information controlling the update. Task or process control blocks contain markers indicating the corresponding tasks are safe or unsafe to run the new programs. The markers are set initially to unsafe. A change descriptor table is stored and contains control information derived from the change instructions. When the DSUF is activated, an interrupt handler is installed and traps are stored in the old programs at entry points and safety points therein. Entry point traps are tripped when a task or process enters the old program and interrupts are generated that are handled by the interrupt handler to route tasks which are unsafe to the old program and tasks which are safe to a new program. When all tasks are safe, the new programs replace the old programs. When safety point traps are tripped, a task or process may change its state from unsafe to safe when predetermined conditions are met.

Claims

1. The method of dynamically updating an old operating system program (hereinafter "old program") stored in a main memory of a data processing system (DPS) while said DPS is running and executing at least one task that accesses said old program from time to time in a multitasking mode, said method comprising:

(A) storing in said memory a new program that is an updated version of said old program;

(B) establishing an executable safety point in said DPS which produces a machine observable safety point condition;

(C) storing in said memory a selectively settable first marker for indicating whether said one task is safe or unsafe for executing said new program, said first marker being initially set to indicate said one task is unsafe;

(D) executing said one task and entering said old program through an entry point therein;

(E) in response to entering said entry point, examining said first marker and in response to noting said one task is unsafe, passing control to said old program for execution thereof;

(F) executing said safety point to produce said safety point condition;

(G) in response to observing said safety point condition, setting said first marker to safe;

and (H) after said marker has been set to safe, executing said new program each time said first task enters said old program.

2. The method in accordance with claim 1 comprising:

(I) installing an entry point trap in said old program at said entry point before performing step (D);

step (D) comprises tripping said entry point trap;

and step (E) is performed in response to tripping said entry point trap.

3. The method in accordance with claim 2 wherein said entry point trap is effective when tripped to generate a first interrupt, and said method further comprises

(J) installing an interrupt handler in said DPS for intercepting said first interrupt and branching to step (E).

4. The method in accordance with claim 3 wherein:

said safety point comprises a safety point trap which when executed in step (F) generates an interrupt of the same type as said first interrupt generated by said entry point trap;

and step (J) further comprises intercepting said second interrupt and branching to step (G).

5. The method in accordance with claim 4 wherein

said step (E) is done by first deciding that said first interrupt is generated by said entry point trap, before examining said marker and passing control;

and step (G) is done by first deciding said second interrupt is generated by said safety point trap, before setting said marker to safe.

6. The method in accordance with claim 3 wherein said entry point trap comprises an invalid op code effective when executed to generate a program check interrupt, and said interrupt handler comprises a program check first level interrupt handler having an intercept for said program check interrupt.

7. The method in accordance with claim 1 wherein a second task is being executed which also attempts to access said old program, and said method further comprises:

storing in said memory a second marker which is selectively settable to alternatively indicate the status of said second task is safe or unsafe for executing said new program, said second marker being initially set to indicate said second task is unsafe;

and thereafter performing for said second task steps similar to steps (D) through (H) whereby both said second task thereafter causes said new program to be executed.

8. The method in accordance with claim 7 wherein said first task and said second task form a coordinated group wherein said first task is suspended when said first marker is set to safe, until said second marker is set to safe whereupon said first task is unsuspended.

9. The method in accordance with claim 1 wherein said one task comprises executable code and a control block stored in said memory for storing information controlling operation of said one task, said first marker being located in said control block.

10. The method in accordance with claim 1 wherein said safety point comprises a safety point trap which is executed in step (F).

11. The method of dynamically updating an old operating system program (hereinafter "old program") stored in a main memory of a data processing system (DPS) while said DPS is running and executing at least one task that accesses said old program from time to time, said method comprising:

(A) installing in said DPS a selectively activated dynamic software update facility (DSUF) having a trap processing routine;

(B) storing in said memory prior to activation of said DSUF a new program and a status marker for said one task, said new program being an updated version of said old program, each status marker being selectively settable to indicate that said one task is safe or unsafe to execute said new program, said status marker being initially set to unsafe;

(C) activating said DSUF by

(C1) storing an entry point trap at each entry point into said old program, each entry point trap generating an entry point interrupt when such each trap is tripped,

(C2) establishing a safety point in said DPS which safety point includes safety point code effective when executed to signify said one task is eligible to be marked safe and execute said new program,

and (C3) installing in said DPS an interrupt handler for intercepting said entry point interrupt and executing said trap processing routine;

(D), after activation of said DSUF, running said DPS to execute said one task and, in response to said one task entering said old program, performing steps comprising

(D1) tripping said entry point trap to generate said entry point interrupt,

and (D2) executing said trap processing routine in response to said entry point interrupt from step (D1) to examine said status marker and in response to noting said status is unsafe, branching to said old program for execution thereof;

(E) further operating said system until said safety point code is executed and, in response thereto, setting said status marker safe;

and (F) thereafter executing said new program each time said one task enters said old program and trips one of said entry point traps.

12. The method in accordance with claim 11 wherein:

said safety point code is a trap operative when tripped to generate a safety point interrupt;

and step (E) comprises executing said trap to generate said safety point interrupt and branching to said trap processing routine, said trap processing routine being operative to set said status marker safe.

13. The method of claim 12 wherein said safety point trap and said entry point trap generate the same type of interrupt, and said trap processing routine performs the steps comprising:

examining the source of each interrupt and deciding whether such interrupt is a safety point interrupt or an entry point interrupt, and branching to a routing routine when said interrupt is an entry point interrupt and to a safety point processing routine when said interrupt is a safety point interrupt;

and executing said routing routine to branch to said old program when said task is unsafe and to said new program when said task is safe.

14. The method in accordance with claim 13 comprising:

executing said safety point processing routine to decide if said task meets predetermined conditions for such task to become safe, and to set said status marker to safe when said task meets said predetermined conditions.

15. The method in accordance with claim 14 wherein a safety point is selected from a group of conditions comprising:

when a task is started, upon a task entering a particular module, upon a task exiting a particular module, when a task makes a particular system call, if a task executes an instruction at a given offset at a given module, when a task is swapped out, when a task is observed as being in a problem state, when a task is observed as being in a wait state awaiting completion of some other task, when a task is awaiting new work to be assigned to it, when a task is running under a given job name, and when a task is running other than under a given job name.

16. The method in accordance with claim 11 wherein said system includes at least one additional task which accesses said old program, said DPS operating in a multitasking mode, and said method further comprises;

storing in said memory a second status marker for said additional task, said second status marker being selectively settable to indicate that said additional task is safe or unsafe to execute said new program, said second status marker being initially set or unsafe;

establishing a second safety point in said DPS which includes second safety point code effective when executed to signify said additional task is safe to execute said new program;

and thereafter performing in steps (D) through (F) in such a manner as to performs steps for said additional task similar to those performed for said one task whereby said additional task executes said new program each time it enters said old programs and trips an entry point trap therein.

17. The method in accordance with claim 16 wherein said one task and said additional task are coordinated, and said method comprises:

suspending execution of said one task after it becomes safe, until said second safety point is executed and said additional task becomes safe;

setting said second status marker to safe;

and then unsuspending said one task whereby each task can then execute said new program.

18. The method in,accordance with claim 17 wherein said DPS is a multiprocessing system comprising a plurality of processors for executing said tasks, said memory and programs stored therein being shared by tasks executing in said processors.

19. The method in accordance with claim 16 wherein said memory further stores a data structure which is accessed by one or more of said tasks, and said method further comprises:

storing in said memory a data structure change effector program for changing data in said data structure;

and, in response to said one or more tasks being marked safe, executing said data structure change effector program to change the data in said data structure.