Transparent dynamic code optimization

• US 9,483,295 B2
• Filed: 03/31/2014
• Issued: 11/01/2016
• Est. Priority Date: 03/31/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Gschwind, Michael K.
• Primary Examiner(s): Dao, Thuy
• Assistant Examiner(s): Nguyen, Mongbao
• Application Number:
  US14/231,658
• Publication Number:
  US 20150277870A1
• Time in Patent Office: 946 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 2009/45583 Memory management, e.g. access or allocation
• CPC Class Codes: G06F 9/45516 Runtime code conversion or optimisation
• CPC Class Codes: G06F 9/45525 Optimisation or modification within the same instruction set architecture, e.g. HP Dynamo
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects

Abstract

A transparent dynamic code optimization capability natively executes code and identifies hot regions within the code or identifies code that is to otherwise be optimized; notifies an optimizer that such hot region or code has been identified enabling the optimizer to create optimized code; efficiently passes control to the newly optimized code; and transfers control back from such optimized code.

Claims

1. A computer program product for facilitating processing in a computing environment, said computer program product comprising:

a non-transitory computer readable storage medium readable by a processing circuit and storing instructions for execution by the processing circuit for performing a method comprising;

executing, by a processor, an application;

determining that a defined event has occurred;

based on determining that the defined event has occurred, transferring control to a component of the computing environment to perform one or more modifications to provide modified code of the application, wherein the application comprises the modified code and unmodified code, the modified code being stored in separate memory regions different from the memory regions storing the unmodified code, wherein the unmodified code comprises original code corresponding to the modified code, and wherein the transferring control comprises using an event based branch to transfer to an address specified by the processor based on occurrence of the defined event, the address corresponding to an instruction to be performed by the component to perform the one or more modifications;

receiving control back at the processor from the component based on providing the modified code to continue executing the application;

continuing to execute the application, wherein the continuing comprises executing the modified code;

based on executing the modified code, gaining access to a target address register that specifies an address to which control is to be transferred; and

based on branching to the address, receiving control back at the unmodified code and continuing to execute the application by executing the unmodified code.

2. The computer program product of claim 1, wherein the continuing to execute the application comprises:

determining at a point in the application that modified code is to be executed;

based on determining that modified code is to be executed, fetching the modified code from memory using an address translation entry for instruction fetches, wherein a different address translation entry is used for data accesses; and

executing the modified code.

3. The computer program product of claim 1, wherein the target address register being a register separate from registers used for application state.

4. The computer program product of claim 1, wherein the method further comprises creating, by the component, the modified code, the creating comprising:

analyzing one or more statistics provided by a hardware performance monitor executing within the computing environment;

determining, based on the analyzing, that the application is to be modified; and

creating the modified code, based on the determining.

5. The computer program product of claim 4, wherein the analyzing comprises obtaining information from the hardware performance monitor to determine a hot region in the application and creating the modified code based on the obtained information.

6. The computer program product of claim 4, wherein the analyzing further comprises obtaining and analyzing information from a branch history rolling buffer related to one or more branches that have been taken in the application.

7. The computer program product of claim 4, wherein the method further comprises storing the modified code, wherein the modified code is stored in the separate memory regions, the separate memory regions being accessible via an instruction fetch, and wherein data accesses access the memory regions storing the unmodified code.

8. The computer program product of claim 1, wherein based on the transferring, execution of the application is stopped until execution is resumed based on completion of the modifications.

9. The computer program product of claim 1, wherein based on the transferring, execution of the application continues.

10. The computer program product of claim 1, wherein the one or more modifications are based at least in part on obtaining and analyzing information from a branch history rolling buffer related to one or more branches that have been taken or from a hot region in the application.

11. A computer system for facilitating processing in a computing environment, said computer program product comprising:

a memory; and

a processor in communications with the memory, wherein the computer system is configured to perform a method, said method comprising;

executing, by a processor, an application;

determining that a defined event has occurred;

based on determining that the defined event has occurred, transferring control to a component of the computing environment to perform one or more modifications to provide modified code of the application, wherein the application comprises the modified code and unmodified code, the modified code being stored in separate memory regions different from the memory regions storing the unmodified code, wherein the unmodified code comprises original code corresponding to the modified code, and wherein the transferring comprises using an event based branch to transfer to an address specified by the processor based on occurrence of the defined event, the address corresponding to an instruction to be performed by the component to perform the one or more modifications;

receiving control back at the processor from the component based on providing the modified code to continue executing the application;

continuing to execute the application, wherein the continuing comprises executing the modified code;

based on executing the modified code, gaining access to a target address register that specifies an address to which control is to be transferred; and

based on branching to the address, receiving control back at the unmodified code and continuing to execute the application by executing the unmodified code.

12. The computer system of claim 11, wherein the continuing to execute the application comprises:

determining at a point in the application that modified code is to be executed;

based on determining that modified code is to be executed, fetching the modified code from memory using an address translation entry for instruction fetches, wherein a different address translation entry is used for data accesses; and

executing the modified code.

13. The computer system of claim 11, wherein the target address register being a register separate from registers used for application state.

14. The computer system of claim 11, wherein the method further comprises creating, by the component, the modified code, the creating comprising:

analyzing one or more statistics provided by a hardware performance monitor executing within the computing environment;

determining, based on the analyzing, that the application is to be modified; and

creating the modified code, based on the determining.

15. The computer system of claim 14, wherein the analyzing comprises obtaining information from the hardware performance monitor to determine a hot region in the application and creating the modified code based on the obtained information.

16. The computer system of claim 14, wherein the analyzing further comprises obtaining and analyzing information from a branch history rolling buffer related to one or more branches that have been taken in the application.

17. The computer system of claim 11, wherein based on the transferring, execution of the application is stopped until execution is resumed based on completion of the modifications.

18. The computer system of claim 11, wherein based on the transferring, execution of the application continues.

19. The computer system of claim 11, wherein the one or more modifications are based at least in part on obtaining and analyzing information from a branch history rolling buffer related to one or more branches that have been taken or from a hot region in the application.