Method and computer readable medium for providing checkpointing to windows application groups

• US 8,752,049 B1
• Filed: 12/15/2008
• Issued: 06/10/2014
• Est. Priority Date: 12/15/2008
• Status: Expired due to Fees

• Current Assignee: Open Invention Network LLC
• Original Assignee: Open Invention Network LLC
• Inventors: Backensto, Keith Richard, Havemose, Allan
• Primary Examiner(s): Puente, Emerson
• Assistant Examiner(s): MILLS, PAUL V
• Application Number:
  US12/334,671
• Time in Patent Office: 2,003 Days
• Field of Search: None
• US Class Current: 718/100
• CPC Class Codes: G06F 11/1466 to make the backup process non-disruptive
• CPC Class Codes: G06F 11/1469 Backup restoration techniques
• CPC Class Codes: G06F 11/1479 Generic software techniques for error detection or fault masking
• CPC Class Codes: G06F 11/203 using migration
• CPC Class Codes: G06F 2201/82 Solving problems relating to consistency ensuring consistency in mirrored systems G06F11/2064
• CPC Class Codes: G06F 9/485 Task life-cycle, e.g. stopping, restarting, resuming execution G06F9/4881 takes precedence
• CPC Class Codes: G06F 9/547 Remote procedure calls [RPC]; Web services

Abstract

A computer readable medium and method providing checkpointing to Windows application groups, the computer readable medium having computer-executable instructions for execution by a processing system. The computer-executable instructions may be for launching an application and creating one or more application threads, receiving a checkpoint signal by an application thread, and entering a checkpoint Asynchronous Procedure Call (APC) handler at IRQL APC_LEVEL, the APC handler disposed in a kernel module, acquiring an ETHREAD block and user-mode context for the application thread, and examining an execution state of the one or more application threads at a time of the checkpoint signal.

Claims

1. A non-transitory computer readable medium having computer-executable instructions which, when executed by a processor, cause a processing system to perform a method of checkpointing an application group, the method comprising:

loading a checkpoint kernel module and registering a coordinator process with the checkpoint kernel module;

launching a plurality of independent applications, each including a plurality of threads, as an application group via the coordinator, wherein launching an application via the coordinator causes the launched application to load a user-space checkpoint library;

wherein an application loading the checkpoint library comprises installing a plurality of function interceptors including at least wait function interceptors which together are configured to intercept system calls made by the threads of the loading application during execution, and, in response to intercepting a system call from a calling thread and determining the coordinator has initiated a group checkpoint, are configured to cause the calling thread to block and wait in an alertable state;

initiating a group checkpoint of the application group via the coordinator including causing each of the independent applications to perform an application checkpoint, wherein performing an application checkpoint includes sending a kernel-mode checkpoint signal to each thread of the application; and

responsive to receiving the kernel-mode checkpoint signal, the receiving thread performs steps comprising;

entering a checkpoint signal handler in the kernel module;

determining the receiving thread was active in kernel-space at the time the kernel-mode signal was received utilizing kernel attributes directly accessible from the checkpoint signal handler; and

inserting a user-mode checkpoint signal at the front of a signal queue of the receiving thread and returning from the checkpoint signal handler in response to the determining that the receiving thread was active in kernel-space, wherein a user-mode signal is only processed by a signaled thread when the signaled thread waits in the alertable state, and wherein the user-mode checkpoint signal is configured to cause the thread to enter a user-mode signal handler in the checkpoint library when processed.

2. The non-transitory computer readable medium of claim 1, wherein performing the application checkpoint further comprises directing the receiving thread to wait at a first checkpoint barrier in the kernel module in response to determining that the receiving thread was not active in kernel-space at the time the kernel-mode signal was received.

3. The non-transitory computer readable medium of claim 2, wherein the user-mode signal handler in the checkpoint library is configured to direct the thread processing the user-mode signal to wait at the first checkpoint barrier.

4. The non-transitory computer readable medium of claim 3, wherein performing the application checkpoint further comprises, after all threads of the application are waiting at the first checkpoint barrier, releasing all of threads from the first checkpoint barrier which each proceed to save their respective thread state.

5. The non-transitory computer readable medium of claim 4, wherein a thread saving its thread state comprises saving the thread state to a heap of an application checkpoint thread.

6. The non-transitory computer readable medium of claim 4, wherein the threads of the application performing the application checkpoint wait at a final checkpoint barrier after saving their respective thread states.

7. The non-transitory computer readable medium of claim 6, wherein the coordinator process releases the threads waiting at the final checkpoint barrier in response to determining that all threads of each of the independent applications are waiting at the final checkpoint barrier.

8. The non-transitory computer readable medium of claim 1, wherein the kernel-mode checkpoint signal comprises a kernel Asynchronous Procedure Call (APC) and the user-mode checkpoint signal comprises a user APC.

9. The non-transitory computer readable medium of claim 1, wherein performing the application checkpoint further comprises setting a user-mode instruction pointer of the receiving thread to an invalid address prior to returning from the checkpoint signal handler in response to the determining that the receiving thread was active in kernel-space at the time the kernel-mode checkpoint signal was received.

10. The non-transitory computer readable medium of claim 9, wherein the user-mode instruction pointer of the receiving thread being set to the invalid address will generate an exception in response to the receiving thread returning to user-space after returning from the checkpoint handler, wherein the exception is caught by an exception interceptor of the checkpoint library configured to direct the receiving thread to wait at a first checkpoint barrier.

11. A method of checkpointing an application group comprising:

loading a checkpoint kernel module and registering a coordinator process with the checkpoint kernel module;

launching a plurality of independent applications, each including a plurality of threads, as an application group via the coordinator, wherein launching an application via the coordinator causes the launched application to load a user-space checkpoint library;

wherein an application loading the checkpoint library comprises installing a plurality of function interceptors including at least wait function interceptors which together are configured to intercept system calls made by the threads of the loading application during execution, and, in response to intercepting a system call from a calling thread and determining the coordinator has initiated a group checkpoint, are configured to cause the calling thread to block and wait in an alertable state;

initiating a group checkpoint of the application group via the coordinator including causing each of the independent applications to perform an application checkpoint, wherein performing an application checkpoint includes sending a kernel-mode checkpoint signal to each thread of the application; and

responsive to receiving the kernel-mode checkpoint signal, the receiving thread performs steps comprising;

entering a checkpoint signal handler in the kernel module;

determining the receiving thread was active in kernel-space at the time the kernel-mode signal was received utilizing kernel attributes directly accessible from the checkpoint signal handler; and

inserting a user-mode checkpoint signal at the front of a signal queue of the receiving thread and returning from the checkpoint signal handler in response to the determining that the receiving thread was active in kernel-space, wherein a user-mode signal is only processed by a signaled thread when the signaled thread waits in the alertable state, and wherein the user-mode checkpoint signal is configured to cause the thread to enter a user-mode signal handler in the checkpoint library when processed.

12. The method of claim 11, wherein performing the application checkpoint further comprises directing the receiving thread to wait at a first checkpoint barrier in the kernel module in response to determining that the receiving thread was not active in kernel-space at the time the kernel-mode signal was received.

13. The method of claim 12, wherein the user-mode signal handler in the checkpoint library is configured to direct the thread processing the user-mode signal to wait at the first checkpoint barrier.

14. The method of claim 13, wherein performing the application checkpoint further comprises, after all threads of the application are waiting at the first checkpoint barrier, releasing all of threads from the first checkpoint barrier which each proceed to save their respective thread state.

15. The method of claim 14, wherein a thread saving its thread state comprises saving the thread state to a heap of an application checkpoint thread.

16. The method of claim 14, wherein the threads of the application performing the application checkpoint wait at a final checkpoint barrier after saving their respective thread states.

17. The method of claim 16, wherein the coordinator process releases the threads waiting at the final checkpoint barrier in response to determining that all threads of each of the independent applications are waiting at the final checkpoint barrier.

18. The method of claim 11, wherein the kernel-mode checkpoint signal comprises a kernel Asynchronous Procedure Call (APC) and the user-mode checkpoint signal comprises a user APC.

19. The method of claim 11, wherein performing the application checkpoint further comprises setting a user-mode instruction pointer of the receiving thread to an invalid address prior to returning from the checkpoint signal handler in response to the determining that the receiving thread was active in kernel-space at the time the kernel-mode checkpoint signal was received.

20. The method of claim 19, wherein the user-mode instruction pointer of the receiving thread being set to the invalid address will generate an exception in response to the receiving thread returning to user-space after returning from the checkpoint handler, wherein the exception is caught by an exception interceptor of the checkpoint library configured to direct the receiving thread to wait at a first checkpoint barrier.