HK1166390B

HK1166390B - Copyback optimization for memory system

Info

Publication number: HK1166390B
Application number: HK12106948.6A
Authority: HK
Inventors: Nir Jacob Wakrat; Mark Alan Helm
Original assignee: Apple Inc.
Priority date: 2008-04-30
Filing date: 2011-09-09
Publication date: 2013-11-29

Description

TECHNICAL FIELD

This subject matter is generally related to memory systems.

BACKGROUND

Flash memory is a type of electrically erasable programmable read-only memory (EEPROM). Because flash memories are non-volatile and relatively dense, they are used to store files and other persistent objects in handheld computers, mobile phones, digital cameras, portable music players, and many other devices in which other storage solutions (e.g., magnetic disks) are inappropriate. When data stored in flash memory is processed, a data page is read out of flash memory and stored in a register. The register contents is transferred to a controller that is external to the non-volatile memory for processing. The processed data page is placed back in the register so that the processed data page can be written back to flash memory.

US 2007/0048715 describes subtitle generation and retrieval combining document processing with voice processing.

US 2006/0050576 describes a NAND flash memory device and copyback program method.

EP 0691614 describes a device for determining an address of a modified line in a cache memory. A copyback routine to an external memory is executed.

SUMMARY

In a copyback or read operation for a non-volatile memory subsystem, data page change indicators are used to manage transfers of data pages between a register in non-volatile memory and a controller that is external to the non-volatile memory. According to the invention, which is defined in the appended independent claims 1 and 8, it is proposed a method and an apparatus for performing a copyback operation for a data page in a non-volatile flash memory.

DESCRIPTION OF DRAWINGS

FIG. 1A is a block diagram of an example memory system for implementing an optimized copyback process.

FIG. 1B is a block diagram of another example memory system for implementing an optimized copyback process.

FIG. 2 is a flow diagram of an example optimized copyback process 200 for computing and storing data change indicators used by the memory systems of FIGS. 1A and 1B.

FIGS. 3A-3C are flow diagrams of an example optimized copyback process implemented by the memory systems of FIGS. 1A and IB.

Fig. 4 is a flow diagram of an example optimized read operation implemented by the memory systems of Figs. 1A and 1B.

Fig. 5 illustrates example data structures used by the copyback process of Figs. 3A-3C.

Detailed description Examples of Memory Subsystems

Fig. 1A is a block diagram of an example memory system 100 for implementing an optimized copyback process. In some implementations, the system 100 can include a memory subsystem 102 coupled to a host device 124 through an external bus 122 (e.g., Open NAND Flash Interface (ONFI), ATA). The host device 124 can be any electronic device including but not limited to smart phones, media players, text messaging devices, portable computers, solid state drives, digital cameras, etc. The memory subsystem 102 can be any non-volatile memory subsystem (e.g., managed NAND).

The host device 124 can include a system-on-chip (SoC) 126 and volatile memory 128. The SoC 126 can include hardware and software for interacting with the memory subsystem 102, such as transmitting read and write requests made by applications running on the host device 124.

The memory subsystem 102 can include non-volatile memory 104 (also referred to as "raw memory") and an external controller 116. The memory 104 can include a memory cell array 106, an internal state machine 108, a memory register 110 and data change indicator (DO) 130. The controller 116 can include a processor 118, volatile memory 114 and error correction code (ECC) engine 120. Other configurations for the memory subsystem 102 are possible. For example, a cache register can be included in the data path between the memory cell array 106 and the memory register 110 to allow the internal state machine to read the next data page from the memory cell array 106 while transferring the current page to the controller 116 over internal bus 112.

FIG. IB is a block diagram of another example memory system 130 for implementing an optimized copyback process. The system 130 shows only the memory subsystem 102, the operation of which was described in reference to FIG. 1A.

Copyback Operations

Copyback can be a memory subsystem command to move data from one page to another page. Copyback can be used in wear leveling and other nonvolatile memory management operations. In a typical copyback operation, a data page is read from the memory cell array 106 and stored in the memory register 110 by the internal state machine. The external controller 116 reads or clocks the data page out of the memory register 110 so that the processor 118 can perform a desired operation on the data page (e.g., an ECC operation). The processed data page can be written back to the memory register 110 by the controller 116. The internal state machine 108 can write the contents of the memory register 110 into a new data page in the memory cell array 106. By avoiding the transfer of the entire contents of register 110 to the external controller 116, processing times and power consumption can be reduced.

Example Process For Writing New Data Pages

FIG. 2 is a flow diagram of an example optimized copyback process 200 for computing and storing data change indicators used by the memory systems of FIGS. 1A and IB. In some implementations, the process 200 can begin when new data page is transferred into a memory register of a memory subsystem (202). A DCI can be computed for the data page and stored in the memory subsystem (204). Some examples of data change indicators can include error detection codes (EDCs), including but not limited to: checksum, Hamming code, parity bit, cyclic redundancy check (CRC), polarity symbol reversal, Turbo code, etc. An EDC can also be part of an ECC, such as Reed-Solomon code, Reed-Muller code, Binary Golay code, and low-density parity-check codes. After the DCIs are computed and stored, the register contents can be written to non-volatile memory (206). DCIs can be used for the optimized copyback process described in reference to FIGS. 3A-3C.

Example Optimized Copyback Process

FIGS. 3A-3C are flow diagrams of an example optimized copyback process 300 implemented by the memory systems of FIGS. 1A and IB. In some implementations, the process 300 can begin by reading a data page from non-volatile memory (e.g., memory cell array 106) and storing the data page in a memory register (302) (e.g., register 110). A DCI (e.g., DCI130) can be verified for the data page in the memory register (304). In some implementations, the DCI can be previously computed and stored in the memory subsystem when the data page is first written to non-volatile memory (e.g., using process 200). In other implementations, the DCI can be computed "on the fly" as part of the read operation or as a separate operation. The memory subsystem can report to an external controller (e.g., external controller 116) that the read operation has completed (306).

If the DCI for the data page in the register indicates that the data page has changed (308), the data page can be transferred to the external controller (310), an ECC can be computed for the data page (312) and the ECC information for the data page can be corrected in the external controller (314). The data page and associated ECC can be transferred back to the memory register (316). A new DCI can be computed for the data page in the memory register (318).

If the DCI for the data page in the register indicates that the data page has not changed (308), an ECC can be computed for new metadata associated with the data page (320). The new metadata and associated ECC can be transferred to the memory register (322). A new DCI can be computed for the metadata in the memory register (324). The memory register contents can be written to a new data page in non-volatile memory (326).

Optimized Read Operation

FIG. 4 is a flow diagram of an example optimized read operation implemented by the memory systems of FIGS. 1A and IB. In some implementations, the process 400 can begin by reading a data page from non-volatile memory and storing the data page in a memory register (402). A pre-computed DCI associated with the data page can be clocked into the memory register from a storage location in the memory subsystem and verified (404). The DCI can also be computed "on the fly." The status of the DCI can be read to determine if the data page has changed (406).

[0020] If the data page has changed (408), an ECC for the new metadata associated with the data page can be computed and transferred, together with the new metadata to the memory register (412). A new DCI can be computed for the metadata in the memory register and the register contents can be written to a new data page in non-volatile memory (414).

[0021] If the data page has not changed (408), the memory register contents can be transferred to an external controller and an ECC can be computed by the external controller (410). An ECC for the new metadata associated with the data page can be computed and transferred by the external controller, together with the new metadata to the memory register (412). A new DCI can be computed for the metadata in the memory register and the register contents can be written to a new data page in non-volatile memory (414).

Example Data Structures

[0022] FIG. 5 illustrates example data structures used by the copyback process of FIGS. 3A-3C. In the example shown, a data page 500 in non-volatile memory includes metadata 502, metadata ECC 504, data 506 and data ECC 508. If the data has not changed then only the metadata 502 and the metadata ECC 504 are transferred from the register to the external controller 510 for processing. The processed metadata 502 and metadata ECC 504 are transferred back to the memory subsystem and stored in the memory register. The data page can the be written to a new data page in non-volatile memory (e.g., a memory cell array).

[0023] A number of implementations have been described. Nevertheless, it will be understood that various modifications may be made. For example, elements of one or more implementations may be combined, deleted, modified, or supplemented to form further implementations. As yet another example, the logic flows depicted in the figures do not require the particular order shown, or sequential order, to achieve desirable results. In addition, other steps may be provided, or steps may be eliminated, from the described flows, and other components may be added to, or removed from, the described systems. Accordingly, other implementations are within the scope of the following claims.

Claims

A method of performing a copyback operation for a data page in non-volatile Flash memory of a memory subsystem, the method comprising:
obtaining the data page from the non-volatile memory (106) of the memory subsystem;

storing the data page in a register (110) of the memory subsystem (102);

determining if the data page has changed based on a data change indicator (130) that indicates whether a change has occurred in the data page;

if the data page has changed, sending the data page to a controller (116) for processing; and

if the data page is unchanged:
sending metadata associated with the data page to the controller for processing, without sending the unchanged data page to the controller for processing;

receiving, from the controller, the metadata and an error correction code for the metadata;

storing the metadata and the error correction code in the register; and

writing the contents of the register to the non-volatile memory including:

writing the unchanged data page from the register to another data page in the non-volatile memory; and

writing the metadata and the error correction code, from the register, to the non-volatile memory.
The method of claim 1, where determining if the data page has changed further comprises:
reading the data change indicator (130) for the data page.
The method of claim 2, where the data change indicator is associated with an error detection code.
The method of claim 1, where determining if the data page has changed further comprises:
computing the data change indicator (130) for the data page; and

determining if the data page has changed based on the data change indicator.
The method of claim 4, where computing the data change indicator (130) is performed prior to or on-the-fly as part of the read operation for reading the data page from the non-volatile memory.
The method of claim 1, where if the data page has changed, the method further comprising:
determining, by the controller, an error correction code associated with the data page that has changed.
The method of claim 6, where obtaining the data page includes obtaining the data page from a register, the method further comprising:
sending the data page that has changed and the error correction code to the register.
A system for performing a copyback operation for a data page in non-volatile Flash memory, the system comprising:
an interface (112) adapted for coupling to a controller;

non-volatile memory (104);

a processor (118) coupled to the non-volatile memory and the interface, the processor operable to:
obtain the data page from the non-volatile memory;

store the data page in a register (110);

determine if the data page has changed based on a data change indicator (130) that indicates whether a change has occurred in the data page; and

if the data page has changed, send the data page through the interface to the controller for processing; and

if the data page has not changed:
send metadata associated with the data page through the interface to the controller for processing, without sending the unchanged data page to the controller for processing;

receive, from the controller through the interface, the metadata and an error correction code for the metadata;

store the metadata and the error correction code in the register; and write the contents of the register to the non-volatile memory including:
writing the unchanged data page from the register to another data page in the non-volatile memory; and

writing the metadata and the error correction code, from the register, to the non-volatile memory.
The system of claim 8, where the processor (118) is further operable to read the data change indicator (130) for the data page to determine if the data page has changed.
The system of claim 9, where the data change indicator (130) is associated with an error detection code.
The system of claim 8, where the processor (118) is further operable to compute the data change indicator (130) for the data page, and determine if the data page has changed based on the data change indicator.
The system of claim 11, where the processor (118) is further operable to compute the data change indicator (130) prior to or concurrently with obtaining the data page from the non-volatile memory (104).
The system of claim 8, where if the data page has changed, the controller (116) is further operable to determine an error correction code associated with the data page that has changed.
The system of claim 13, where:
the processor (118) further operable to obtain the data page from a register (110); and

the controller (116) is further operable to send the data page that has changed and the error correction code to the register.