Saving state data in parallel in a multi-processor system

• US 20070168760A1
• Filed: 11/29/2005
• Published: 07/19/2007
• Est. Priority Date: 11/29/2005
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Song, Cheng-Chung, Ko, Wenjeng
• Granted Patent: US 7,500,141 B2
• Time in Patent Office: Days
• US Class Current: 714/49
• CPC Class Codes: G06F 11/0715 in a system implementing multitasking multitasking per se G06F9/46
• CPC Class Codes: G06F 11/0724 in a multiprocessor or a multi-core unit multiprocessors per se G06F15/80
• CPC Class Codes: G06F 11/0766 Error or fault reporting or storing

Abstract

A method, system and program product save state data in a multi-processor system. A problem in the multi-processor system is detected and a statesave thread is spawned for each processor in the system. Each statesave thread directs a processor, in parallel with the other processors to attempt to identify a component in the system having a status of “incomplete”, indicating that state data of the component remains to be offloaded. When a component having a status of “incomplete” is identified, the processor executes statesave code to offload state data from the identified component. Upon completion of the state data offload from the identified component, the processor changes the status of the component to “complete”. The foregoing processes are repeated until no components are identified in the system having a status of “incomplete”.

Claims

1. A method for saving state data in a multi-processor system, the system including a plurality of software components having state data, the method comprising:

detecting a problem in a multi-processor system; and

for each processor in the system, spawning a statesave thread directing each processor in parallel to;

a) attempt to identify a component in the system having a status of “

incomplete”

, indicating that state data of the component remains to be offloaded;

b) when a component having a status of “

incomplete”

is identified, execute a statesave code to offload state data from the identified component;

c) upon completion of the state data offload from the identified component, change the status of the component to “

complete”

; and

d) repeat steps a) through c) until no components are identified in the system having a status of “

incomplete”

.

2. The method of claim 1, wherein the spawned statesave threads further direct the processors to:

prior to executing the statesave code, determine if the state data can be offloaded within a predetermined period of time and, if the state data cannot be offloaded within the predetermined period of time, split the state data into a plurality of portions, each portion capable of being offloaded within the predetermined period of time;

execute the statesave code to offload a first portion of the state data; and

leave the status of the component unchanged, whereby a second portion of the state data of the component is offloaded by any processor in the system.

3. The method of claim 1, wherein the spawned statesave threads further direct the processors to:

offload a first portion of the state data if, during execution of the statesave code, the state data cannot be offloaded within a predetermined period of time; and

leave the status of the component unchanged, whereby a second portion of the state data of the component is offloaded by any processor in the system.

4. The method of claim 1, further comprising calling a master statesave driver thread after detecting a problem in the system.

5. The method of claim 1, further comprising receiving the offloaded state data into a buffer.

6. The method of claim 5, further comprising destaging the offloaded state data from the buffer to a non-volatile storage device.

7. The method of claim 1, further comprising:

determining if all statesave threads are completed within a predetermined period of time; and

if all statesave threads are not completed within the predetermined period of time, transmitting a command to abort all incomplete statesave threads.

8. A multi-processor system, comprising:

a plurality of processors, including a first processor;

a plurality of software components, each logging state data comprising a history of states of the component;

a buffer;

a master thread configured to detect a problem in the system; and

a statesave thread spawned to each of the plurality of processors by the master thread upon detection of a problem and directing each processor, in parallel with each other processor, to;

a) attempt to identify a component in the system having a status of “

incomplete”

, indicating that state data of the component remains to be offloaded;

b) when a component having a status of “

incomplete”

is identified, execute a statesave code to offload state data from the identified component to the buffer;

c) upon completion of the state data offload from the identified component, change the status of the component to “

complete”

; and

d) repeat a) through c) until no components are identified in the system having a status of “

incomplete”

.

9. The system of claim 8, wherein the spawned statesave threads further direct the processors to:

prior to executing the statesave code, determine if the state data can be offloaded within a predetermined period of time and, if the state data cannot be offloaded within the predetermined period of time, split the state data into a plurality of portions, each portion capable of being offloaded within the predetermined period of time;

execute the statesave code to offload a first portion of the state data to the buffer; and

leave the status of the component unchanged, whereby a second portion of the state data of the component is offloaded by any processor in the system to the buffer.

10. The system of claim 8, wherein the spawned statesave threads further direct the processors to:

offload a first portion of the state data to the buffer if, during execution of the statesave code, the state data cannot be offloaded within a predetermined period of time; and

leave the status of the component unchanged, whereby a second portion of the state data of the component is offloaded by any processor in the system to the buffer.

11. The system of claim 8, further comprising a non-volatile storage device coupled to receive offloaded state data from the buffer.

12. A computer program product of a computer readable medium usable with a programmable computer, the computer program product having computer-readable code embodied therein for saving state data in a multi-processor system, the system including a plurality of components having state data, the computer-readable code comprising instructions for:

detecting a problem in a multi-processor system; and

for each processor in the system, spawning a statesave thread directing each processor in parallel to;

a) attempt to identify a component in the system having a status of “

incomplete”

, indicating that state data of the component remains to be offloaded;

b) when a component having a status of “

incomplete”

is identified, execute a statesave code to offload state data from the identified component;

c) upon completion of the state data offload from the identified component, change the status of the component to “

complete”

; and

d) repeat steps a) through c) until no components are identified in the system having a status of “

incomplete”

.

13. The computer program product of claim 12, wherein the spawned statesave threads further direct the processors to:

prior to executing the statesave code, determine if the state data can be offloaded within a predetermined period of time and, if the state data cannot be offloaded within the predetermined period of time, split the state data into a plurality of portions, each portion capable of being offloaded within the predetermined period of time;

execute the statesave code to offload a first portion of the state data; and

leave the status of the component unchanged, whereby a second portion of the state data of the component is offloaded by any processor in the system.

14. The computer program product of claim 12, wherein the spawned statesave threads further direct the processors to:

offload a first portion of the state data if, during execution of the statesave code, the state data cannot be offloaded within a predetermined period of time; and

leave the status of the component unchanged, whereby a second portion of the state data of the component is offloaded by any processor in the system.

15. The computer program product of claim 12, the computer-readable code further comprising instructions to call a master statesave driver thread after detecting a problem in the system.

16. The computer program product of claim 12, the computer-readable code further comprising instructions to receive the offloaded state data into a buffer.

17. The computer program product of claim 16, the computer-readable code further comprising instructions to destage the offloaded state data from the buffer to a non-volatile storage device.

18. The computer program product of claim 12, the computer-readable code further comprising instructions for:

determining if all statesave threads are completed within a predetermined period of time; and

if all statesave threads are not completed within the predetermined period of time, transmitting a command to abort all incomplete statesave threads.