Containing machine check events in a virtual partition

1. A method of software execution executed by computer, comprising:

dividing a hard partition into first and second virtual partitions;

attempting to correct an error in a firmware layer of the first virtual partition;

determining, in the firmware layer of the first virtual partition, whether a processor in the first virtual partition holds a hard partition level lock;

rebooting both the first and second virtual partitions when the processor holds the hard partition level lock; and

rebooting the first virtual partition without disrupting hardware resources in the second virtual partition when the error is not correctable and the processor does not hold the hard partition level lock.

2. The method of claim 1 further comprising, if the error is correctable in the firmware layer, then returning to an interrupted context without disrupting processing in the second partition.

3. The method of claim 1 further comprising, performing clean-up operations of local error events in the first virtual partition without disrupting hardware resources in the second virtual partition.

4. The method of claim 1 further comprising, rebooting the first virtual partition without disrupting applications running in the second virtual partition.

5. The method of claim 1 further comprising, if the error is not correctable in the firmware layer, then transferring control to an operating system machine check abort (OS MCA) layer in the first virtual partition without disrupting operations of an operating system in the second virtual partition.

6. The method of claim 1 further comprising, if the error is an input/output (I/O) event, then performing a rendezvous of processors belonging to the first virtual partition.

7. The method of claim 1 further comprising, passing the error from a processor abstraction layer (PAL) to a system abstraction layer (SAL) if the error is not correctable in the PAL.

8. A computer readable storage medium having instructions for causing a computer to execute a method, comprising:

dividing a hard partition into plural virtual partitions;

receiving a machine check abort (MCA) event in a firmware layer of one of the plural virtual partitions;

determining, in the firmware layer of the one of the plural virtual partitions, whether a processor in the one of the plural virtual partitions holds a partition lock;

rebooting the plural virtual partitions in the hardware partition when the processor holds the partition lock; and

rebooting the one of the plural virtual partitions without affecting hardware resources in other virtual partitions of the plural virtual partitions when the error is not correctable and the processor does not hold the partition lock.

9. The computer readable storage medium of claims 8 further comprising, shutting down the one of the virtual partitions to perform system error analysis without shutting down the other of the plural virtual partitions.

10. The computer readable storage medium of claim 8 further comprising, isolating the MCA event to the one of the virtual partitions without disrupting processing events of the other of the plural virtual partitions running on the hard partition.

11. The computer readable storage medium of claim 8 further comprising, rebooting the one of the virtual partitions without affecting shared resources across the other of the plural virtual partitions.

12. The computer readable storage medium of claim 8 further comprising, if the MCA event us corrected in the firmware layer of the one of the virtual partitions, then returning to an interrupted context without disrupting the hardware resources in the other of the plural virtual partitions.

13. The computer readable storage medium of claim 8 further comprising, passing the MCA event between different firmware layers in the one of the virtual partitions without disrupting applications running in the other of the plural virtual partitions.

14. The computer readable storage medium of claim 8 further comprising, logging error information of platforms that are shared across the plural virtual partitions.

15. A computer system, comprising:

a hard partition that is divided into plural virtual partitions; and

plural firmware layers in one of the virtual partitions, the firmware layers containing a machine check event in the one of the virtual partitions, wherein a determination is made, in a firmware layer of the one of the virtual partitions, whether a processor in the one of the virtual partitions holds a partition lock; the virtual partitions in the hardware partition are rebooted when the processor holds the partition lock; and the one of the virtual partitions is rebooted without affecting hardware resources in other virtual partitions of the virtual partitions when the error is not correctable and the processor does not hold the partition lock.

16. The computer system of claim 15 further comprising, a processor abstraction layer (PAL) and a system abstraction layer (SAL) in the firmware layers, wherein the PAL passes the machine check event to the SAL without disrupting operations of applications in the other of the virtual partitions.

17. The computer system of claim 15 further comprising, a system abstraction layer (SAL) in th firmware layers that checks if a processor in the one of the virtual partitions hold a hard partition level lock.

18. The computer system of claim 15 further comprising, an operating system machine check abort (OS MCA) handler that reboots the one of the virtual partitions without rebooting the other of the virtual partitions.

19. The computer system of claim 15 , wherein the plural firmware layers correct the machine check event without disrupting operations in the other of the virtual partitions.

20. The computer system of claim 15 , wherein the plural firmware layers correct the machine check event and perform clean-up of local error events without affecting hardware resources belonging to the other of the virtual partitions.