Reallocation of pools of fixed size buffers based on metrics collected for maximum number of concurrent requests for each distinct memory size

• US 6,070,202 A
• Filed: 05/11/1998
• Issued: 05/30/2000
• Est. Priority Date: 05/11/1998
• Status: Expired due to Fees

• Current Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Original Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Inventors: Abbate, Alain D., Monahan-Mitchell, Timothy A., Minkoff, Ronald M., Sandvos, Jerry Lee Albert
• Primary Examiner(s): Lee, Thomas C.
• Assistant Examiner(s): WANG, ALBERT C
• Application Number:
  US09/075,364
• Time in Patent Office: 750 Days
• Field of Search: 707/202, 707/206, 709/1, 709/104, 710/57, 710/56, 710/52, 711/147, 711/148, 711/149, 711/150, 711/151, 711/152, 711/153, 711/170, 711/171, 711/172, 711/173, 713/1
• US Class Current: 710/56
• CPC Class Codes: G06F 9/5016 the resource being the memory

Abstract

A computing device (100) has memory (180) subdivided into a first configuration of pools of fixed size memory buffers (188). In response to requests for memory specifying a particular memory size, a fixed size memory of at least the particular size is allocated from the memory area (510, 520). Metrics are collected for each memory request, the metrics including distinct memory sizes based on the particular size requested, and the peak number of concurrent requests for each distinct memory size (530). The memory area is reconfigured as needed into a second configuration of pools of fixed size memory buffers using the metrics collected (540, 550).

Claims

1. In a computing device having a memory area subdivided into a first configuration of pools of fixed size memory buffers, a method comprising the steps of:

receiving a plurality of memory requests for memory from the memory area, each memory request specifying a particular memory size;

allocating a fixed size memory buffer from the memory area in response to each memory request, the fixed size memory buffer having a size of at least the particular size;

collecting metrics for each memory request, the metrics including distinct memory sizes based on the particular size requested, and maximum number of concurrent requests for each distinct memory size, wherein the collection of metrics on maximum number of concurrent requests is independent of the number of fixed size memory buffers in any of the pools of fixed size memory buffers; and

reconfiguring the memory area into a second configuration of pools of fixed size memory buffers using the metrics collected.

2. The method of claim 1, wherein:

the step of collecting metrics comprises the step of identifying particular pools of fixed size memory buffers that have been exhausted; and

the step of reconfiguring the memory area comprises the step of allocating an increase in number of memory buffers for the particular pools of fixed size memory buffers identified.

3. The method of claim 1, wherein the step of reconfiguring the memory area comprises the step of the automatically configuring the memory area using the metrics collected during normal operation of the computing device.

4. The method of claim 1, wherein the step of reconfiguring the memory area comprises the step of automatically configuring the memory area using the metrics collected as part of a power up sequence.

5. The method of claim 1, wherein the step of reconfiguring the memory area comprises the step of automatically configuring the memory area using the metrics collected as part of a power down sequence.

6. A method for memory management in a computing device, the method comprising the steps of:

organizing memory in the computing device into at least first, second, and third memory areas, the first memory area being subdivided into pools of fixed size buffers, the second memory area having configuration information for the first memory area, and the third memory area having a metrics storage portion;

collecting metrics on memory usage of the first memory area, the metrics including distinct memory size requests and peak number of concurrent requests for each of the distinct memory size requests, wherein the collection of metrics on peak number of concurrent requests is independent of the number of fixed size memory buffers in any of the pools of fixed size memory buffers;

storing the metrics in the third memory area;

reconfiguring the first memory area using the metrics stored in the third memory area, including the steps of;

generating new configuration for the first memory area; and

storing the new configuration in the second memory area.

7. The method of claim 6, wherein the step of reconfiguring the first memory area comprises the step of adjusting buffer size for existing buffer pools based on the distinct memory size requests and peak number of concurrent requests for each of the distinct memory size requests.

8. The method of claim 6, wherein:

the step of collecting metrics comprises the step of identifying particular pools of fixed size buffers for which an allocated number of memory buffers have been exhausted; and

the step of reconfiguring the first memory area comprises the step of increasing the allocated number of memory buffers for the particular pools of fixed size buffers identified.

9. The method of claim 6, wherein the step of reconfiguring the first memory area comprises the step of the automatically configuring the first memory area during normal operation of the computing device.

10. The method of claim 6, wherein the step of reconfiguring the first memory area comprises the step of the automatically configuring the first memory area, as part of a power up sequence.

11. The method of claim 6, wherein the step of reconfiguring the first memory area comprises the step of automatically configuring the first memory area, as part of a power down sequence.

12. The method of claim 6, wherein the step of reconfiguring the first memory area comprises the steps of:

determining that a particular buffer pool has an insufficient number of buffers based on a peak number of concurrent requests stored as a metric, and based on an allocated number of buffers for the particular buffer pool;

determine an actual deficiency for the particular buffer pool by comparing the peak number of concurrent requests stored with the allocated number of buffers for the particular buffer pool;

identifying another buffer pool having excess capacity based on the peak number of concurrent requests and the allocated number of buffers for that other buffer pool; and

reallocating at least a portion of the excess capacity from the other buffer pool to the particular buffer pool.

13. The method of claim 6, wherein the metrics of peak number of concurrent requests is independent of the number of fixed size buffers in any of the pools of fixed size buffers.

14. The method of claim 6, wherein the step of collecting includes the step of tracking at least one peak number of concurrent requests for at least one distinct memory size that does not match the size of a fixed size buffer in any of the pools of fixed size buffers.

15. A computing device, comprising:

a memory subdivided into first, second, and third areas, a first area being subdivided into pools of fixed size memory buffers, and a second area having configuration information for the first area, and a third area having metrics on usage of the first area; and

a real-time operating system that is responsive to requests for memory from the first area, to collect and store the metrics from the requests for memory in the third area, wherein the metrics include distinct memory sizes associated with requests for memory from the first area, and peak number of concurrent requests corresponding to each of the distinct memory sizes, which peak number of concurrent requests is independent of the number of fixed size memory buffers in any of the pools of fixed size memory buffers.

16. The computing device of claim 15, further comprising a memory block in the third area that stores data indicating which pools have had their memory buffers exhausted.

17. The method of claim 15, wherein the peak number of concurrent requests for at least one metric exceeds the number of buffers allocated for any of the pools of fixed size memory buffers.

18. The method of claim 15, wherein at least some of the distinct memory sizes are distinct from the size of buffers in any of the pools of fixed size memory buffers.