Logical address direct memory access with multiple concurrent physical ports and internal switching

• US 8,271,700 B1
• Filed: 01/24/2011
• Issued: 09/18/2012
• Est. Priority Date: 11/23/2007
• Status: Active Grant

• Current Assignee: Microsemi Solutions (U.S.), Inc. (Microchip Technology Incorporated)
• Original Assignee: PMC-Sierra Incorporated (Microchip Technology Incorporated)
• Inventors: Annem, Babysaroja, Alexander, Praveen, Clinton, David J.
• Primary Examiner(s): Lee, Chun-Kuan
• Assistant Examiner(s): MARTINEZ, DAVID E
• Application Number:
  US13/012,296
• Time in Patent Office: 603 Days
• Field of Search: 710/22, 710/26
• US Class Current: 710/26
• 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 2213/0038 System on Chip

Abstract

A DMA engine is provided that is suitable for higher performance System On a Chip (SOC) devices that have multiple concurrent on-chip/off-chip memory spaces. The DMA engine operates either on logical addressing method or physical addressing method and provides random and sequential mapping function from logical address to physical address while supporting frequent context switching among a large number of logical address spaces. Embodiments of the present invention utilize per direction (source-destination) queuing and an internal switch to support non-blocking concurrent transfer of data on multiple directions. A caching technique can be incorporated to reduce the overhead of address translation.

Claims

1. A method for operating a direct memory access (DMA) engine for data transfer operations, comprising:

concurrently fetching at least two DMA descriptors, each of the at least two DMA descriptors corresponding to respective data transfer operations between devices;

processing each of the at least two DMA descriptors for

determining a source device and a destination device, read address information of data to be read from the source device and write address information of the destination device where the data is to be written, and

executing logical to physical address translation of at least one of the read address information and the write address information;

generating multiple read and write commands corresponding to each of the at least two DMA descriptors;

executing concurrent data transfer of data between source ports and destination ports of the DMA engine in response to the read and write commands; and

,

reporting in order DMA completion status for each of the at least two DMA descriptors.

2. The method of claim 1, further comprising pre-fetching a new DMA descriptor in response to completion of all requests related to the read and write command corresponding to a source port and a destination port.

3. The method of claim 1, further comprising pre-fetching a new DMA descriptor in response to acceptance of all requests related to the read and write command corresponding to a source port and a destination port.

4. The method of claim 1, wherein at least one of the at least two DMA descriptors comprises a field configured for DMA bypass mode operation to process control information for the at least one of the at least two DMA descriptors, and inhibiting the logical to physical address translation and the steps of generating and executing.

5. The method of claim 1, wherein at least one of the at least two DMA descriptors comprises a field configured for DMA bypass mode operation to process control information for the at least one of the at least two DMA descriptors and inhibiting the step of reporting.

6. The method of claim 1, wherein at least one of the at least two DMA descriptors comprises a field configured for DMA bypass mode operation to process control information for the at least one of the at least two DMA descriptors, and inhibiting the steps of generating and executing.

7. The method of claim 1, wherein a request for logical to physical address translation and fetching of physical address related to write data transfers are executed concurrently with read data transfers.

8. The method of claim 1, wherein processing comprises requesting logical to physical address translation of the write address information concurrently with storing the data read from the source device into a buffer.

9. The method of claim 1, wherein at least one of the source device and the destination device includes a fragmented physical buffer comprising arbitrarily aligned and arbitrarily sized memory fragments defined by at least one scatter/gather list (SGL), and logical to physical address translation comprises:

mapping each of the fragments to contiguous logical offset addresses in a contiguous logical address space equal to a total length of the fragments, and,

determining a physical address corresponding to a logical offset address of one of the fragments by;

accessing an SGL element associated with the fragment to determine a physical start address for the fragment; and

offsetting the physical start address by the logical offset address less the length of preceding fragments listed in the at least one SGL.

10. The method of claim 9, wherein the step of determining is executed for all the fragments.

11. The method of claim 9, wherein SGL fragment context information is fetched and stored concurrently with write data transfer and read data transfer.

12. The method of claim 1, further comprising monitoring internal control signals of the DMA engine.

13. The method of claim 12, wherein said internal control signals correspond to states of finite state machines, states of arbitration grants or read/write addresses.

14. A method for operating a direct memory access (DMA) engine for data transfer operations, comprising:

concurrently fetching at least two DMA descriptors, each of the at least two DMA descriptors corresponding to respective data transfer operations between devices;

processing each of the at least two DMA descriptors for determining a source device and a destination device, read address information of data to be read from the source device and write address information of the destination device where the data is to be written;

generating multiple read and write commands corresponding to each of the at least two DMA descriptors; and

,

executing concurrent data transfer of data between source ports and destination ports of the DMA engine in response to the read and write commands, including detecting an error code generated by sub-blocks of the DMA engine during data transfer operations, wherein detecting comprises decoding the error code to be a fatal error type or a non-fatal error type; and

the step of executing continues if the error code is decoded to be of the non-fatal error type.

15. The method of claim 14, wherein the step of executing is halted if the error code is decoded to be of the fatal error type.

16. The method of claim 14, further comprising monitoring internal control signals of the DMA engine.

17. The method of claim 16, wherein said internal control signals correspond to states of finite state machines, states of arbitration grants or read/write addresses.

18. The method of claim 14, further comprising pre-fetching a new DMA descriptor in response to completion of all requests related to the read and write command corresponding to a source port and a destination port.

19. The method of claim 14, further comprising pre-fetching a new DMA descriptor in response to acceptance of all requests related to the read and write command corresponding to a source port and a destination port.