ENLIGHTENMENT FOR LOW OVERHEAD HARDWARE ACCESS

• US 20100250230A1
• Filed: 03/30/2009
• Published: 09/30/2010
• Est. Priority Date: 03/30/2009
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Wohlgemuth, Jason S., Ganguly, Shuvabrata, Marshall, Allen
• Granted Patent: US 8,635,057 B2
• Time in Patent Office: Days
• US Class Current: 703/26
• CPC Class Codes: G06F 2009/45583 Memory management, e.g. access or allocation
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects

Abstract

A computing system in which a software component executing on a platform can reliably and efficiently obtain state information about a component supported by the platform through the use of a shared memory page. State information may be supplied by the platform, but any state translation information needed to map the state information as supplied to a format as used may be provided through the shared page. In a virtualized environment, the state translation information can be used to map the value of a virtual timer counter or other component from a value provided by a virtual processor to a normalized reference time that will yield the same result, regardless of whether the software component is migrated to or from another virtual processor. Use of a shared page avoids the inefficiency of an intercept into a virtualized environment or a system calls in native mode operation.

Claims

1. A method of operating a computing device, the method comprising:

operating at least one physical processor to;

implement a hypervisor, the hypervisor emulating an emulated processor having a time stamp counter;

execute computer-executable instructions implementing a computer operating system executing on the emulated processor that performs an operation of computing a normalized reference time,

wherein;

operating at least one physical processor to execute computer-executable instructions implementing a computer operating system comprises performing the operation, the performing comprising;

reading a timer value from the time stamp counter from the emulated processor;

obtaining through an interface with the hypervisor a scale factor and an offset; and

calculating the normalized reference time by applying the offset and the scale factor to the value obtained from the time stamp counter.

2. The method of claim 1, wherein the reading and the obtaining comprise reading and obtaining without causing an intercept into the hypervisor.

3. The method of claim 2, wherein the interface comprises a shared page, wherein the shared page is shared between the hypervisor and the operating system.

4. The method of claim 3, wherein applying the offset and the scale factor to the timer value obtained from the time stamp counter comprises adding the offset to the product of the timer value and the scale factor.

5. The method of claim 4, wherein:

the performing further comprises;

obtaining a sequence number from the shared page; and

prior to reading a timer value, determining if the sequence number is invalid; and

the reading, obtaining, and calculating are performed when the sequence number is not invalid.

6. The method of claim 5, wherein the performing further comprises:

when the sequence number is invalid, calculating the normalized reference time using an alternative method.

7. The method of claim 6, wherein calculating the normalized reference time using an alternative method comprises comprising the normalized reference time by obtaining a value of the Advanced Configuration and Power Interface Power Management timer and/or the High Precision Event Timer.

8. The method of claim 7, wherein:

the sequence number comprises a first sequence number obtained at a first time; and

wherein the performing further comprises;

subsequent to calculating the normalized reference time, obtaining a second sequence number from the shared page at a second time;

determining if the first sequence number is the same as the second sequence number;

when the first sequence number is the same as the second sequence number, returning the value of the calculated normalized reference time to a consumer of the operation; and

when the first sequence number is not the same as the second sequence number, then;

obtaining a third sequence number from the shared page at a third time;

prior to reading a timer value, determining if the third sequence number is invalid; and

when the third sequence number is not invalid, then performing the reading, the obtaining, and the calculating.

9. A method of operating an operating system in a guest partition, the method comprising:

operating at least one physical processor to;

implement a first emulated processor having a first time stamp counter;

implement a second emulated processor having a second time stamp counter;

execute computer-executable instructions implementing the operating system within the guest partition executing on the first emulated processor during a first period of time, the implementing comprising performing a timer operation based on a timer value obtained from the first time stamp counter;

execute computer-executable instructions implementing the operating system within the guest partition executing on the second emulated processor during a second period of time, the implementing comprising performing a timer operation based on a timer value obtained by computing a normalized reference time from a value obtained from the second time stamp counter, the computing the normalized reference time comprising;

obtaining a scale factor and an offset; and

calculating the normalized reference time by applying the offset and the scale factor to the value obtained from the second time stamp counter.

10. The method of claim 9, wherein obtaining a scale factor and an offset and obtaining a value from the second time stamp counter are performed without causing an intercept into the hypervisor.

11. The method of claim 10, further comprising migrating the guest partition from the first virtual processor to the second virtual processor between the first interval and the second interval.

12. The method of claim 11, wherein:

operating at least one physical processor to implement a second emulated processor comprises implementing a hypervisor; and

obtaining a scale factor and an offset comprises obtaining a scale factor and an offset through an interface with the hypervisor.

13. The method of claim 12, wherein the interface with the hypervisor comprises a shared page, wherein the shared page is shared between the operating system and the hypervisor.

14. The method of claim 13, wherein applying the offset and the scale factor to the value obtained from the time stamp counter comprises adding the offset to the product of the timer value and the scale factor.

15. The method of claim 14, wherein:

the computing the normalized reference time further comprises;

obtaining a sequence number from the shared page; and

prior to obtaining the value from the second time stamp counter,

determining if the sequence number is invalid; and

when the sequence number is not invalid, then;

obtaining the value from the second time stamp counter;

obtaining the scale factor and the offset; and

calculating the normalized reference time.

16. A method of supporting a normalized reference time in a guest partition executing on a hypervisor, the hypervisor executing on a computer, the method performed by the hypervisor when a guest partition is restored, the method comprising:

operating at least one physical processor on the computer to;

updating a maximum time stamp counter value;

store a value of a scale factor for the computer in a shared page, the shared page being shared between the hypervisor and the guest partition;

compute an offset based on the scale factor and the maximum time stamp counter value; and

store the offset in the shared page.

17. The method of claim 16, wherein computing an offset based on the scale factor and the maximum reference time comprises:

obtaining a saved reference time, wherein the saved reference time is a reference time for the guest partition prior when the guest partition was last saved;

subtracting the product of the scale factor and the maximum time stamp counter value from the saved reference time.

18. The method of claim 17, wherein the method further comprises:

incrementing a sequence number; and

storing the sequence number in the shared page.

19. The method of claim 18, wherein updating a maximum time stamp counter value comprises updating a maximum time stamp counter value across all virtual processors in the guest partition.

20. The method of claim 19, wherein the offset comprises a first offset, and wherein the method further comprises:

computing a second offset based on a value obtained by subtracting the maximum time stamp counter value from a time stamp counter value obtained from the at least one physical processor; and

setting an offset in all virtual processors for the guest partition to the value of the second offset.