Automated transition to a recovery kernel via firmware-assisted-dump flows providing automated operating system diagnosis and repair

• US 8,132,057 B2
• Filed: 08/07/2009
• Issued: 03/06/2012
• Est. Priority Date: 08/07/2009
• Status: Expired due to Fees

• Current Assignee: GlobalFoundries, Inc.
• Original Assignee: International Business Machines Corporation
• Inventors: Jann, Joefon, Lindeman, James Arthur, Burugula, Ramanjaneva Sarma, Wu, Ching-Farn Eric, El Maghroaui, Kaoutar
• Primary Examiner(s): Le, Dieu-Minh
• Application Number:
  US12/537,486
• Publication Number:
  US 20110035618A1
• Time in Patent Office: 942 Days
• Field of Search: 714/38.1, 714/38.11, 714/38.12, 714/38.13, 714/39, 714/2, 714/25, 717/124, 717/126, 717/128
• US Class Current: 714/38.1
• CPC Class Codes: G06F 11/0712 in a virtual computing platform, e.g. logically partitioned systems
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40

Abstract

A method (and structure) of operating an operating system (OS) on a computer. When a failure of the OS is detected, the computer automatically performs a diagnosis of the OS failure. The computer also attempts to automatically repair/recover the failed OS, based on the diagnosis, without requiring a reboot.

Claims

1. A method of operating an operating system (OS) instance on a computer, said method comprising:

detecting a failure of said OS instance; and

automatically performing a diagnosis of said OS failure, using a processing code on said computer that is created from the originally-booted OS instance itself, before an occurrence of failure, and is located in a hidden memory region associated with said OS instance.

2. The method of claim 1, further comprising using said processing on said computer to attempt to automatically repair/recover said failed OS, based on said diagnosis, without requiring a reboot.

3. The method of claim 2, wherein said automatic diagnosis/repair/recovery of said failed OS instance comprises:

creating a diagnosis-and-recovery kernel using parts from said OS instance as originally booted, on said computer;

hiding the memory region containing said diagnostic-and-repair kernel from its associated running OS instance; and

registering a memory location of said diagnosis-and-recovery kernel with the firmware on said computer that controls said OS instance,

wherein, upon said failure of said OS instance, said firmware automatically transfers control to said diagnosis-and-recovery kernel image.

4. The method of claim 3, further comprising:

providing a user command that enables said diagnosis-and-recovery kernel by creating an image of said diagnosis-and-recovery kernel in a protected region of said OS instance; and

allocating protected memory region/space for storing information during a failure of said OS instance.

5. The method of claim 4, wherein said protected region of said OS instance comprises a contiguous real-address memory region of said OS instance that is hidden from said OS instance during a normal operation of said OS instance.

6. The method of claim 5, wherein said failure of said OS instance is at least one of:

detected by said firmware; and

reported to said firmware for action.

7. The method of claim 6, wherein, upon detecting a failure of said OS instance or receiving a notification of said failure, said firmware automatically saves a state of said OS instance into said protected region.

8. The method of claim 7, further comprising, upon saving said OS instance'"'"'s state into said protected region, said firmware invokes said diagnosis-and-recovery kernel to execute said diagnosis of said OS instance failure.

9. The method of claim 8, further comprising:

automatically attempting to repair said failed OS instance;

if said failed OS instance can be repaired, said firmware reloads the repaired OS instance and passes control back to said repaired OS instance for resuming operation without rebooting said OS instance; and

if said failed OS instance cannot be repaired, said firmware transmits a diagnosis report of said failed OS instance.

10. The method of claim 1, wherein said OS comprises an OS instance, as supported by a firmware with a hypervisor.

11. The method of claim 10, wherein said OS instance is implemented on a SMP (Symmetric Multi-Processor) server with LPARs (Logical PARtitions), including a hypervisor firmware that oversees the LPARs, said method automatically performing diagnosis in the same LPAR instance as the LPAR instance where a failure in the OS has been detected.

12. The method of claim 11, wherein said computer comprises an SMP server having a plurality of LPARs (Logical PARtitions) each running an OS instance with a diagnosis-and-recovery kernel in an ALDR (Automated LPAR Diagnosis and Recovery) image.

13. The method of claim 12, said ALDR image comprising:

an ALDR kernel;

a plurality of device drivers;

a plurality of commands; and

a table containing addresses of key items from the OS instance before it crashes.

14. A tangible computer-readable storage medium tangibly embodying a program of machine-readable instructions executable by a digital processing apparatus to perform the method of claim 1.

15. An apparatus, comprising:

at least one processor executing at least one operating system (OS) instance,

wherein, when a failure of said OS instance is detected, said processor automatically performs a diagnosis of said OS failure, using a section of processing code created from the originally-booted OS instance itself, before an occurrence of failure, and located in a hidden region in a same LPAR (Logical PARtition) as the failed OS instance.

16. The apparatus of claim 15, wherein said processor also automatically attempts to repair/recover said failed OS instance, based on said diagnosis, without a reboot.

17. The apparatus of claim 16, wherein said failed OS comprises an OS instance, and said automatic diagnosis/repair/recovery of said failed OS instance comprises:

creating a diagnosis-and-recovery kernel using parts from said OS instance as originally booted on said apparatus;

hiding a memory region containing said diagnosis-and-repair kernel from its associated running OS instance; and

registering a memory location of said diagnosis-and-recovery kernel with a firmware on said apparatus that controls said OS instance,

wherein, upon said failure of said OS instance, said firmware automatically transfers control to said diagnosis-and-recovery kernel image.

18. The apparatus of claim 17, wherein a user command is provided that enables said diagnosis-and-recovery kernel by creating an image of said diagnosis-and-recovery kernel in a protected region of said OS instance and to allocate memory space for storing information during a failure of said OS instance.

19. The apparatus of claim 17, wherein said diagnosis-and-recovery kernel automatically attempts to repair said failed OS instance, and

if said failed OS instance can be repaired, said firmware reloads the repaired OS instance and passes control back to said repaired OS instance for resuming operation without rebooting said OS instance, and

if said failed OS instance cannot be repaired, said firmware transmits a diagnosis report of said failed OS instance.

20. The apparatus of claim 15, wherein said OS comprises an OS instance implemented on a SMP (Symmetric Multi-Processor) server with LPARs (Logical PARtitions), including a hypervisor firmware that oversees the LPAR instances, said apparatus automatically performing diagnosis executed in the same LPAR instance as the LPAR of the OS instance where a failure has been detected.

21. A mechanism to attempt to automatically recover a failed operating system (OS) instance on a computer or LPAR, said mechanism comprising:

a detector that provides an indication that an OS instance has failed;

a diagnosis-and-recovery kernel embedded in a protected region of said OS instance; and

memory space allocated for automatically storing information during a failure of said OS instance,

wherein, upon said OS instance having been detected as failed, said diagnosis-and-recovery kernel is caused to automatically perform a diagnosis of said OS failure, using said automatically-stored information, and attempts to automatically repair/recover said failed OS, based on said diagnosis, without rebooting.

22. A computer, comprising a SMP (Symmetric Multi-Processor) server with LPAR (Logical PARtitioning) capability, said computer comprising:

a plurality of processors, each said processor capable of implementing at least one LPAR having a operating system (OS) instance;

a hypervisor firmware overseeing instances of OSs on the LPARs; and

the automatic recovery mechanism of claim 21.