US 8,276,145 B2Granted

Protected mode scheduling of operations

AssigneeMicrosoft Corporation

InventorsEfstathios Papaefstathiou, Jinsong Yu, Stanislav A. Oks

Life of this patent

Estimating the term…

Loading patent…

Abstract

The present invention extends to methods, systems, and computer program products for protected mode scheduling of operations. Protected mode (e.g., user mode) scheduling can facilitate the development of programming frameworks that better reflect the requirements of the workloads through the use of workload-specific execution abstractions. In addition, the ability to define scheduling policies tuned to the characteristics of the hardware resources available and the workload requirements has the potential of better system scaling characteristics. Further, protected mode scheduling decentralizes the scheduling responsibility by moving significant portions of scheduling functionality from supervisor mode (e.g., kernel mode) to an application.

Claims18

Dependent claims

At a computer system, the computer system including a physical processor and an operating system having a protected mode and a supervisor mode, wherein protected mode limits program access to physical processor resources of programs running in protected mode, supervisor mode having unlimited access to physical processor resources, one or more functions gating access to supervisor mode to permit a program in protected mode to transition to supervisor mode to access physical processor resources, wherein programs executed within the operating system can transition from protected mode to supervisor mode and from supervisor mode to protected mode depending on program instructions, supervisor mode having a default supervisor-mode execution object format that consumes physical processor resources on a time-shared basis in accordance with a default supervisor-mode scheduling policy, a method for configuring the scheduling of processor resources to occur in protected mode to decentralize scheduling responsibility from supervisor mode, the method comprising:

an act of creating a protected-mode scheduling domain operating in the protected mode of the operating system, creating the protected-mode scheduling domain including:
creating a virtual processor, the virtual processor mapping at least a portion of the physical processor for use by execution objects of the protected-mode scheduling domain, the virtual processor including:
a supervisor-mode virtual processor, the supervisor-mode virtual processor configured to reserve at least a portion of the physical processor for use by execution objects of the protected-mode scheduling domain, the reservation being performed according to the default supervisor-mode scheduling policy;
and
a protected-mode virtual processor, the protected-mode virtual processor configured to process supervisor-mode notifications and dispatch execution objects for execution on the at least a portion of the physical processor during the reservation obtained by the supervisor-mode virtual processor such that multiple execution objects are scheduled by the protected-mode virtual processor to execute during the reservation;
loading at least one protected-mode scheduling policy that differs from the default supervisor-mode scheduling policy, the at least one protected-mode scheduling policy for scheduling the execution of the execution objects of the protected-mode scheduling domain on the virtual processor;
and
creating an execution object store configured to store state for execution objects of the protected-mode scheduling domain to facilitate switching context between different execution objects within the protected-mode scheduling domain.

5 more elements

At a computer system, the computer system including a physical processor and an operating system having a protected mode and a supervisor mode, wherein protected mode limits program access to physical processor resources of programs running in protected mode, supervisor mode having unlimited access to physical processor resources, one or more functions gating access to supervisor mode to permit a program in protected mode to transition to supervisor mode to access physical processor resources, wherein programs executed within the operating system can transition from protected mode to supervisor mode and from supervisor mode to protected mode depending on program instructions, supervisor mode having a default supervisor-mode execution object format that consumes physical processor resources on a time-shared basis in accordance with a default supervisor-mode scheduling policy, a method for scheduling the consumption of processor resources in protected mode to decentralize scheduling responsibility from supervisor mode, the method comprising:

an act of receiving an allocation of physical processor resources, the allocation being made according to a default supervisor-mode scheduling policy, the allocation comprising a duration of time on the physical processor resources;
an act of allocating, during the duration of time, the physical processor resources to a first execution object of a protected-mode scheduling domain within the protected-mode scheduling domain, physical processor resources allocated in accordance with a protected-mode scheduling policy that differs from the default supervisor-mode scheduling policy;
an act of executing the first execution object on the allocated physical processor resources to partially complete work indicated in the first execution object;
an act of determining, during the duration of time and in accordance with the protected-mode scheduling policy, that the second execution object is to be executed on the allocated physical resources during the duration of time within the protected-mode scheduling domain prior to fully completing the work indicated in the first execution object;
an act of switching context from the first execution object to the second execution object within protected mode and without transitioning to supervisor mode, including:
an act of persisting the state of first execution object in an execution object store, the execution object store resident within the protected-mode scheduling domain;
and
an act of loading the state of the second execution object from the execution object store;
an act of allocating the physical processor resources to the second execution object of a protected-mode scheduling domain within the protected-mode scheduling domain subsequent to switching context to the second execution object such that the first and second execution objects execute on the allocated physical processor resources during the duration of time;
and
an act of utilizing the allocated physical processor resources to partially complete work indicated in the second execution object during the duration of time.

8 more elements

A computer system, the computer system comprising:

one or more physical processors;
system memory;
one or more physical storage media having stored thereon computer executable instructions representing an operating system having a user mode and a kernel mode that, when executed at one of the processors, cause the computer system to perform the following:
create a user-mode scheduling domain operating in user mode so as to decentralize scheduling from kernel mode, creating the user-mode scheduling domain including:
creating a virtual processor, the virtual processor mapping at least a portion of the one or more physical processors for use by user-mode execution objects in the user-mode scheduling domain, the virtual processor including:
a kernel-mode virtual processor, the kernel-mode virtual processor configured to reserve at least a portion of the one or more physical processors for use by the user-mode execution objects, the reservation being performed according to the default supervisor-mode scheduling policy;
and
a user-mode virtual processor, the user-mode virtual processor configured to process kernel-mode notifications and dispatch execution objects for execution on the at least a portion of the physical processor during the reservation obtained by the kernel-mode virtual processor such that multiple execution objects are scheduled by the user-mode virtual processor to execute during the reservation;
loading a user-mode scheduling policy for scheduling the execution of the execution objects within the user-mode scheduling domain on the virtual processor;
and
creating a user-mode execution object store configured to store state for user-mode execution objects to facilitate switching context between different user-mode execution objects within the user-mode scheduling domain and without transitioning to kernel mode such that the kernel mode is relieved from having to store context information for user-mode execution objects;
and subsequent to creation of the user-mode scheduling domain:
receive an allocation of physical processor resources, the allocation being made according to a default kernel-mode scheduling policy, the allocation comprising a duration of time on the physical processor resources;
allocate the physical processor resources reserved by the kernel-mode virtual processor to a first user-mode execution object during the duration of time, the physical processor resources allocated in accordance with the user-mode scheduling policy;
utilize the allocated physical processor resources to partially complete work indicated in the first execution object;
determining, during the duration of time, and in accordance with the user-mode scheduling policy, that allocation of physical processor resources reserved by the kernel-mode virtual processor is to transition to a second user-mode execution object prior to fully completing the work indicated in the first user-mode execution object;
switch context from the first user-mode execution object to the second user-mode execution object within user mode and without transitioning to kernel mode, including:
an act of persisting the state of first user-mode execution object in the user-mode execution object store;
and
an act of loading the state of the second execution object from the user-mode execution object store;
allocate physical processor resources to the second user-mode execution object within the user-mode scheduling domain subsequent to switching context to the second execution object such that the first and second execution objects execute on the allocated physical processor resources during the duration of time;
and
utilize the allocated physical processor resources to partially complete work indicated in the second user-mode execution object.

20 more elements

Dependencies are inferred from the claim text.

Facts

Application: 12/163,726Filed 27 Jun 2008
Published: US 2009/0328058 A131 Dec 2009
Granted: 25 Sep 2012
Term ends: Not estimatedThe source records don’t support an estimate.

Classification4 codes

G06F 9/485
By program, e.g. task dispatcher, supervisor, operating system: task life-cycle, e.g. stopping, restarting, resuming execution
G06F 9/4881
By program, e.g. task dispatcher, supervisor, operating system: scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
G06F 9/545
Interprogram communication: where tasks reside in different layers, e.g. user- and kernel-space
G06F
Computing and data processing