Cooperative mechanism for efficient application memory allocation

• US 7,516,291 B2
• Filed: 11/21/2005
• Issued: 04/07/2009
• Est. Priority Date: 11/21/2005
• Status: Active Grant

• Current Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Original Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Inventors: van de Ven, Adriaan D M, van Riel, Henri Han
• Primary Examiner(s): Kim; Matt
• Assistant Examiner(s): Krofcheck; Michael C
• Application Number:
  US11/164,393
• Publication Number:
  US 20070118712A1
• Time in Patent Office: 1,233 Days
• Field of Search: 711/170
• US Class Current: 711/170
• CPC Class Codes: G06F 12/023 Free address space management
• CPC Class Codes: G06F 12/08 in hierarchically structured memory systems, e.g. virtual memory systems
• CPC Class Codes: G06F 12/12 Replacement control
• CPC Class Codes: G06F 12/14 Protection against unauthorised use of memory or access to memory

Abstract

System, method and computer program product for allocating physical memory to processes. The method includes enabling a kernel to free memory in a physical memory space corresponding to arbitrarily sized memory allocations released by processes or applications in a virtual memory space. After freeing the memory, the system determines whether freed physical memory in the physical memory space spans one or more fixed size memory units (e.g., page frames). The method further includes designating a status of the one or more page frames as available for reuse; the freed page frames marked as available for reuse being available for backing a new process without requiring the kernel to delete data included in the freed memory released by the process. The kernel may organize pages marked as available for reuse in one or more local “pools” that is organized according to a variety of schemes which provide system efficiencies in that the kernel can eliminate the need for deleting of old data in those page frames without compromising data security.

Claims

1. A method for managing memory freed by processes executing in a virtual memory space in a computer system implementing means for mapping virtual memory addresses into physical memory locations accessible by an operating system (O/S), said method comprising:

receiving, by said O/S, an indication of freed memory in said virtual memory space that spans one or more fixed size memory units in physical memory;

adding said fixed size memory units to a free memory pool, said fixed size memory units in a free memory pool being directly allocatable to a process requiring backing physical memory store without the 0/S deleting data in the freed fixed size memory units; and

maintaining information about said process and one or more free memory pools having fixed size memory units, said information used to directly allocate said freed fixed size memory units to a process, wherein said fixed size memory unit comprises a physical memory page frame, and prior to said O/S adding one or more fixed size memory units into a free memory pool, unmapping a physical memory page frame from virtual memory space.

2. The method as claimed in claim 1, wherein said fixed size memory unit comprises a physical memory page frame, and prior to said O/S adding one or more fixed size memory units into a free memory pool, rendering said physical memory page frame as available for reuse by a process while leaving said physical memory page frame mapped to said process'"'"'s virtual memory space.

3. The method as claimed in claim 2, wherein said O/S receives a page fault exception, and in response, said O/S:

determining whether a free memory pool includes a physical memory page available for reuse, and if said free memory pool includes a physical memory page frame available for reuse;

removing said page from said free memory pool;

determining whether the freed memory page is to be used by a process that freed it, and allocating said page if said page is to be used by said freeing process.

4. The method as claimed in claim 3, wherein if it is determined that a memory page frame is not available for reuse in said free memory pool:

determining if a memory page frame is available in a system wide free memory pool, and if a page frame is available in a system wide free memory pool, deleting data included in said memory page frame; and

allocating memory page frames from a system wide free memory pool for said process.

5. The method as claimed in claim 4, wherein if a memory page frame is not available in a system wide free memory pool, invoking pageout mechanism for increasing availability of system-wide free memory pages.

6. The method as claimed in claim 1, wherein said O/S receives a page fault exception, and in response, said O/S:

determining if a pool having freed memory units will not result in a data security breach if data included in said freed memory units corresponding to said pool is accessed by said process; and

if no data security breach results, allocating one or more freed memory page frames of that pool for said process without the O/S expending overhead of deleting data included in said one or more freed memory units.

7. The method as claimed in claim 6, wherein if no free page is available in a suitable free page pool:

determining if a memory page frame is available in a system wide free memory pool, and if a page frame is available in a system wide free memory pool, deleting data included in said memory page frame; and

allocating memory page frames from a system wide free memory pool for said process.

8. The method as claimed in claim 7, wherein if a memory page frame is not available in a system wide free memory pool, invoking pageout mechanism for increasing availability of system-wide free memory pages.

9. The method as claimed in claim 1, wherein said O/S tracks fixed size memory units that are available system-wide, said method including:

determining whether an amount of fixed size memory units tracked system-wide is below a threshold amount.

10. The method as claimed in claim 9, further comprising:

the step of releasing said freed fixed size memory units into a system wide free memory pool when a process terminates, wherein said freed fixed size memory units of said system-wide free pool are available system-wide.

11. The method as claimed in claim 10, wherein said process having freed fixed size memory units marked available for reuse is associated with an executing parent process, wherein when said process terminates, said method further comprising:

comparing computer system performance for executing a step of releasing said freed fixed size memory units marked available for reuse by a terminating process to its parent process against a computer system performance for executing a step of releasing said freed fixed size memory units marked available for reuse into a global free pool of fixed size memory units when said process terminates; and

choosing a method of releasing said freed fixed size memory units marked available for reuse according to said performance comparison.

12. The method as claimed in claim 11, wherein said comparing comprises:

measuring the computer system performance for releasing said freed fixed size memory units marked available for reuse at one of;

a computer system start-up time, or at a computer system run time when said freed fixed size memory units marked available for reuse are to be allocated.

13. The method as claimed in claim 9, wherein if said amount of fixed size memory units tracked system-wide is below a threshold amount said O/S:

determining if a suitable free page pool to reclaim freed fixed size memory units from; and

, reclaiming said freed fixed size memory units from that free memory pool.

14. The method as claimed in claim 13, wherein said freed memory units designated as available for reuse from a pool are reclaimed based on criteria including at least one of:

a size of said free memory pool;

an allocation rate from the pool;

a rate at which memory is freed into the pool;

a net allocation rate of the pool; and

a length of time memory has been assigned to the pool.

15. The method as claimed in claim 13, wherein if it is determined that no page frames are available in said free memory pools, or no suitable free page pool exists to reclaim freed memory pages from, then invoking pageout mechanism for increasing availability of system-wide free memory pages.

16. The method as claimed in claim 9, wherein a process having freed fixed size memory units marked available for reuse is associated with an executing parent process, wherein upon determination of said process terminating, said method further comprising the step of:

determining that said process does not present a security conflict with said parent process; and

, upon determining no security conflict exists, releasing said freed fixed size memory units marked available for reuse to said parent process.

17. A method for managing memory freed by processes executing in a virtual memory space in a computer system implementing means for mapping virtual memory addresses into physical memory locations accessible by an operating system (O/S), said method comprising:

receiving, by said O/S, an indication of freed memory in said virtual memory space that spans one or more fixed size memory units in physical memory;

adding said fixed size memory units to a free memory pool, said fixed size memory units in a free memory pool being directly allocatable to a process requiring backing physical memory store without the O/S deleting data in the freed fixed size memory units; and

maintaining information about said process and one or more free memory pools having fixed size memory units, said information used to directly allocate said freed fixed size memory units to a process, wherein said fixed size memory unit comprises a physical memory page frame, and prior to said O/S adding one or more fixed size memory units into a free memory pool, designating said physical memory page frame as read-only by a process while leaving said physical memory page frame mapped to said process'"'"'s virtual memory space.

18. The method as claimed in claim 17, wherein said O/S receives a page fault exception, and in response, said O/S:

determining whether a memory page frame needs to be allocated, and, if a memory page frame does not need to be allocated, then;

determining whether a free memory pool includes a physical memory page frame designated as read-only by said process, and if said free memory pool includes a physical memory page frame designated as read-only, removing said page from said free memory pool and designating said memory page frame as available for reuse by re-enabling a process'"'"'s write access to its associated memory page frames.

19. The method as claimed in claim 18, wherein if it is determined that a memory page frame needs to be allocated, said O/S:

determining if a memory page frame is available for reuse in said free memory pool; and

, allocating said freed memory page frame of that pool for said process.

20. The method as claimed in claim 19, wherein if it is determined that a memory page frame is not available for reuse in said free memory pool:

determining if a memory page frame is available in a system wide free memory pool, and if a page frame is available in a system wide free memory pool, deleting data included in said memory page frame; and

allocating memory page frames from a system wide free memory pool for said process.

21. The method as claimed in claim 20, wherein if a memory page frame is not available in a system wide free memory pool, invoking pageout mechanism for increasing availability of system-wide free memory pages.

22. The method as claimed in claim 18, wherein if it is determined that a memory page frame needs to be allocated, said O/S determining whether said process allocates memory elsewhere in its virtual memory space, said O/S further:

unmapping said freed page frame associated with that page frame'"'"'s old virtual memory address in the process'"'"'s virtual memory space, and marking said freed page frame as no longer available for reuse, and, mapping said freed page frame associated with said process to a new virtual address.