Method and system for zero-copy disk IO using sector unaligned buffers

• US 9,047,018 B1
• Filed: 03/20/2012
• Issued: 06/02/2015
• Est. Priority Date: 03/20/2012
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Ananthabhotla, Anand
• Primary Examiner(s): SNYDER, STEVEN G
• Application Number:
  US13/424,589
• Time in Patent Office: 1,169 Days
• Field of Search: None
• US Class Current: 1/1
• 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 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0613 in relation to throughput
• CPC Class Codes: G06F 3/0634 by changing the state or mode of one or more devices
• CPC Class Codes: G06F 3/0638 Organizing or formatting or addressing of data
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• CPC Class Codes: G06F 3/0673 Single storage device

Abstract

Data is transferred between a requesting application and a storage device by passing the application IO buffer to the disk driver. Techniques are provided to avoid data copying to an intermediate sector aligned buffer and passing the intermediate buffer to the disk driver. The techniques allow the use of layered block device drivers such as a Logical Volume Manager (LVM), Distributed Replicated Block Device (DRBD), or both. A look-ahead can determine the IO constraints imposed by the layered block device drivers. Based on the constraints, an entire portion of the buffer may be added to an IO request, or the buffer may be split into a first portion and a second portion. The first portion may be added to a first IO request. The second portion may be added to a second IO request, different from the first IO request.

Claims

1. A system for transferring data between a requesting application and a storage device, the system comprising:

a processor-based input/output management system executed on a computer system and configured to;

receive from the requesting application a scatter-gather list having a plurality of scatter-gather elements referencing data buffers for a data transfer operation;

query at least one layered block device driver coupled between the requesting application and the storage device to determine an input/output constraint imposed by the layered block device driver, the constraint not including alignment of data with a boundary;

examine each scatter-gather element to determine whether there will be a violation of the input/output constraint;

generate at least one input/output request responsive to the examination; and

issue the input/output request to the storage device, wherein the input/output request comprises at least a portion of a data buffer referenced by a scatter-gather element.

2. The system of claim 1 wherein the processor-based input/output management system is configured to:

upon a determination that there will be a violation of the input/output constraint, partition the data buffer into first and second buffer portions to satisfy the input/output constraint;

add the first buffer portion to a first input/output request; and

add the second buffer portion to a second input/output request, different from the first input/output request, wherein the first input/output request is the at least one input/output request, and the first buffer portion is the at least a portion of a data buffer.

3. The system of claim 1 wherein the processor-based input/output management system is configured to:

upon a determination that there will not be a violation of the input/output constraint, add to the at least one input/output request an entire portion of the data buffer referenced by the scatter-gather element and additional data from a next buffer.

4. The system of claim 1 wherein the at least one layered block device driver comprises a Logical Volume Manager (LVM) of a Linux kernel.

5. The system of claim 1 wherein the at least one layered block device driver comprises a Distributed Replicated Block Device (DRBD) of a Linux kernel.

6. The system of claim 1 wherein the processor-based input/output management system is configured to:

split the data buffer referenced by the scatter-gather element into first and second buffer portions to satisfy the input/output constraint if there will be a violation of the input/output constraint;

add the first buffer portion to a first input/output request; and

add the second buffer portion to a second input/output request, different from the first input/output request, wherein the first input/output request is the at least one input/output request, and the first buffer portion is the at least a portion of a data buffer.

7. A method for transferring data between a requesting application and a storage device comprising:

receiving from the requesting application a scatter-gather list having a plurality of scatter-gather elements referencing data buffers for a data transfer operation;

querying at least one layered block device driver coupled between the requesting application and the storage device to determine an input/output constraint imposed by the layered block device driver, the constraint not including alignment of data with a boundary;

examining each scatter-gather element to determine whether there will be a violation of the input/output constraint;

generating at least one input/output request responsive to the examination; and

issuing the input/output request to the storage device, wherein the input/output request comprises at least a portion of a data buffer referenced by a scatter-gather element.

8. The method of claim 7 comprising:

upon a determination that there will be a violation of the input/output constraint, partitioning the data buffer into first and second buffer portions to satisfy the input/output constraint;

adding the first buffer portion to a first input/output request; and

adding the second buffer portion to a second input/output request, different from the first input/output request, wherein the first input/output request is the at least one input/output request, and the first buffer portion is the at least a portion of a data buffer.

9. The method of claim 7 comprising:

upon a determination that there will not be a violation of the input/output constraint, adding to the at least one input/output request an entire portion of the data buffer referenced by the scatter-gather element and at least a portion of another data buffer referenced by a next scatter-gather element.

10. The method of claim 7 wherein the at least one layered block device driver comprises a Logical Volume Manager (LVM) of a Linux kernel.

11. The method of claim 7 wherein the at least one layered block device driver comprises a Distributed Replicated Block Device (DRBD) of a Linux kernel.

12. The method of claim 7 wherein the generating at least one input/output request responsive to the examination comprises:

splitting the data buffer referenced by the scatter-gather element into first and second buffer portions to satisfy the input/output constraint if there will be a violation of the input/output constraint;

adding the first buffer portion to a first input/output request; and

adding the second buffer portion to a second input/output request, different from the first input/output request, wherein the first input/output request is the at least one input/output request, and the first buffer portion is the at least a portion of a data buffer.

13. The method of claim 7 wherein a size of the data buffer is not a multiple of 512 bytes and the at least one input/output request comprises a multiple of 512 bytes.

14. The method of claim 7 wherein the requesting application comprises an iSCSI server.

15. A computer program product, comprising a non-transitory computer-readable medium having a computer-readable program code embodied therein, the computer-readable program code adapted to be executed by one or more processors to implement a method comprising:

receiving from the requesting application a scatter-gather list having a plurality of scatter-gather elements referencing data buffers for a data transfer operation;

querying at least one layered block device driver coupled between the requesting application and the storage device to determine an input/output constraint imposed by the layered block device driver, the constraint not including alignment of data with a boundary;

examining each scatter-gather element to determine whether there will be a violation of the input/output constraint;

generating at least one input/output request responsive to the examination; and

issuing the input/output request to the storage device, wherein the input/output request comprises at least a portion of a data buffer referenced by a scatter-gather element.

16. The computer program product of claim 15 wherein the method comprises:

upon a determination that there will be a violation of the input/output constraint, partitioning the data buffer into first and second buffer portions to satisfy the input/output constraint;

adding the first buffer portion to a first input/output request; and

adding the second buffer portion to a second input/output request, different from the first input/output request, wherein the first input/output request is the at least one input/output request, and the first buffer portion is the at least a portion of a data buffer.

17. The computer program product of claim 15 wherein the method comprises:

upon a determination that there will not be a violation of the input/output constraint, adding to the at least one input/output request an entire portion of the data buffer referenced by the scatter-gather element and at least a portion of a data buffer referenced by a next scatter-gather element.

18. The computer program product of claim 15 wherein the at least one layered block device driver comprises a Logical Volume Manager (LVM) of a Linux kernel.

19. The computer program product of claim 15 wherein the at least one layered block device driver comprises a Distributed Replicated Block Device (DRBD) of a Linux kernel.

20. The computer program product of claim 15 wherein the generating at least one input/output request responsive to the examination comprises:

splitting the data buffer referenced by the scatter-gather element into first and second buffer portions to satisfy the input/output constraint if there will be a violation of the input/output constraint;

adding the first buffer portion to a first input/output request; and

adding the second buffer portion to a second input/output request, different from the first input/output request, wherein the first input/output request is the at least one input/output request, and the first buffer portion is the at least a portion of a data buffer.