Apparatus, system, and method for accessing memory

• US 9,208,071 B2
• Filed: 03/15/2013
• Issued: 12/08/2015
• Est. Priority Date: 12/13/2010
• Status: Active Grant

• Current Assignee: SanDisk Technologies LLC (Western Digital Corporation)
• Original Assignee: Sandisk Technologies Incorporated (Western Digital Corporation)
• Inventors: Talagala, Nisha, Flynn, David
• Primary Examiner(s): Li, Aimee
• Assistant Examiner(s): CHAN, TRACY C
• Application Number:
  US13/836,826
• Publication Number:
  US 20130227201A1
• Time in Patent Office: 998 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 12/0246 in block erasable memory, e.g. flash memory
• CPC Class Codes: G06F 12/0804 with main memory updating G06F12/0806 takes precedence
• CPC Class Codes: G06F 13/28 using burst mode transfer, e.g. direct memory access DMA, cycle steal G06F13/32 takes precedence
• CPC Class Codes: G06F 2212/202 Non-volatile memory
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/0656 Data buffering arrangements
• CPC Class Codes: G06F 3/0679 Non-volatile semiconductor memory device, e.g. flash memory, one time programmable memory [OTP]

Abstract

Apparatuses, systems, methods, and computer program products are disclosed for providing access to auto-commit memory. An auto-commit memory module is configured to cause a volatile memory buffer to commit data from the volatile memory buffer to a non-volatile memory medium in response to a trigger. A mapping module is configured to determine whether to associate a range of data with the volatile memory buffer. A bypass module is configured to service a request for the range of data directly from the volatile memory buffer in response to the mapping module determining to associate the range of data with the volatile memory buffer.

Claims

1. A method comprising:

receiving a write request for data, the write request comprising a namespace identifier;

identifying a relationship between the namespace identifier and one or more of a memory and a non-volatile memory medium;

servicing the write request using the memory in response to the identified relationship associating the namespace identifier with the memory; and

servicing the write request using the non-volatile medium in response to the identified relationship associating the namespace identifier with the non-volatile medium.

2. The method of claim 1, wherein a user space device driver services the write request using the memory directly by bypassing an operating system storage stack.

3. The method of claim 1, wherein identifying the relationship between the namespace identifier and one or more of the memory and the non-volatile memory medium comprises determining whether the write request comprises an auto-commit flag associating the namespace identifier with the memory.

4. The method of claim 1, wherein servicing the write request using the memory comprises mapping the data into virtual memory of a requesting client.

5. The method of claim 4, further comprising unmapping the data from virtual memory in response to access to the virtual memory exceeding a traffic threshold for the virtual memory.

6. The method of claim 1, further comprising,

arming the memory with metadata specifying a logical block address of a non-volatile medium to which the data of the memory is to be committed in response to a predefined trigger, wherein the namespace identifier is persistently mapped to the logical block address;

and

committing the data of the memory to the logical block address of the non-volatile medium in response to detecting the trigger.

7. The method of claim 1, further comprising,

tracking which portions of data of the memory have been updated;

and

committing the updated portions of data of the memory to a non-volatile medium separately from non-updated portions in response to detecting a predefined trigger.

8. The method of claim 1, wherein the namespace identifier is a member of a persistent namespace, the persistent namespace is configured to survive a restart event, and the persistent namespace configured to grant a client access to data of the namespace identifier subsequent to the restart event.

9. The method of claim 8, wherein the persistent namespace comprises a logical unit number (LUN) namespace for a storage device and the namespace identifier comprises a LUN address within the LUN namespace.

10. The method of claim 8, wherein the persistent namespace comprises a file system namespace and the namespace identifier comprises a file identifier of the file system namespace.

11. The method of claim 1, wherein identifying the relationship between the namespace identifier and the memory comprises one of identifying an existing relationship between the namespace identifier and the memory, and creating a relationship between the namespace identifier and the memory.

12. The method of claim 1, wherein the relationship associates the namespace identifier with the memory in response to one or more of detecting an existing relationship between the namespace identifier and the memory;

detecting an auto-commit flag for the write request; and

dynamically assigning the namespace identifier for association with the memory.

13. The method of claim 1, wherein the memory is within an isolation zone of a non-volatile device comprising a non-volatile medium and the isolation zone is configured to receive power from a secondary power source.

14. The method of claim 13, wherein a storage capacity of a plurality of memory buffers, including the memory, within the isolation zone is selected such that a power hold-up time provided by the secondary power source allows the plurality of memory buffers to commit data to a non-volatile medium during the power hold-up time in response to a restart event.

15. An apparatus comprising:

an auto-commit memory module configured to cause a volatile memory buffer to commit data from the volatile memory buffer to a non-volatile memory medium in response to the data filling at least a threshold amount of the volatile memory buffer;

a mapping module configured to determine whether to associate a range of addresses for data with the volatile memory buffer or the non-volatile memory medium; and

a bypass module configured to service a request for the range of addresses directly using the volatile memory buffer in response to the mapping module determining to associate the range of addresses with the volatile memory buffer and further configured to service the request for the range of addresses using the non-volatile memory medium in response to the mapping module determining to associate the range of addresses with the non-volatile memory medium, the request comprising a write request.

16. The apparatus of claim 15, further comprising a request module configured to receive the request for the range of addresses for data, the mapping module configured to determine to associate the range of addresses for data with the volatile memory buffer in response to an auto-commit flag of the request.

17. The apparatus of claim 15, further comprising a request module configured to intercept requests for the non-volatile memory medium, the mapping module configured to dynamically determine to associate the range of addresses for data with the volatile memory buffer in response to the request module intercepting the request for the range of addresses.

18. The apparatus of claim 15, wherein the bypass module is configured to service the request for the range of addresses for data directly from the volatile memory buffer by bypassing a kernel storage stack and servicing the request from user-space.

19. The apparatus of claim 15, wherein the request for the range of addresses for data comprises a write request and the bypass module is configured to service the write request by copying data of the write request into a virtual memory location of a requesting client in response to the mapping module determining to associate the range of addresses for data with the volatile memory buffer, the virtual memory location backed by the volatile memory buffer.

20. A system comprising:

a storage device comprising one or more auto-commit pages configured to preserve data of the auto-commit pages in a non-volatile memory medium in response to a restart event; and

a device driver for the storage device, the device driver configured to cause data of the storage device to be mapped, from kernel-space, into virtual memory and to service a write request, from user-space, the device driver using the one or more auto-commit pages to service the write request in response to determining an association of the write request with the one or more auto-commit pages and using the non-volatile memory medium to service the write request in response to determining an association of the write request with the non-volatile memory medium.

21. The system of claim 20, further comprising a host associated with the virtual memory, the host comprising a processor in communication with the storage device, the device driver executing on the processor.

22. A computer program product comprising a non-transitory computer readable storage medium storing computer usable program code executable to cause a computer to perform operations, the operations comprising:

intercepting, in user-space, a storage request for a memory device, the storage request comprising a file identifier and an offset for a write operation;

determining whether the offset and the file identifier are mapped to the volatile memory;

servicing the storage request in user-space directly using a volatile memory of the memory device in response to determining that the offset and the file identifier are mapped to the volatile memory;

determining whether the offset and the file identifier are mapped to a non-volatile memory of the memory device; and

servicing the storage request in user-space using the non-volatile memory of the memory device in response to determining that the offset and the file identifier are mapped to the non-volatile memory.

23. The computer program product of claim 22, wherein the operations further comprise enforcing file system access permissions for data of the offset and the file identifier using virtual memory access controls.

24. An apparatus comprising:

means for associating a logical identifier with one of a volatile memory and a non-volatile memory medium;

means for bypassing an operating system storage stack to service a storage request for data associated with the logical identifier using the volatile memory in response to the logical identifier being associated with the volatile memory and using the non-volatile memory medium in response to the logical identifier being associated with the non-volatile memory medium; and

means for preserving the data of the volatile memory in the non-volatile memory medium in response to a failure condition.

25. The apparatus of claim 24, further comprising means for providing access to the preserved data after the failure condition.