Advanced management of a non-volatile memory

• US 9,372,792 B1
• Filed: 10/10/2013
• Issued: 06/21/2016
• Est. Priority Date: 05/12/2011
• Status: Active Grant

• Current Assignee: Avago Technologies International Sales Pte Limited (Broadcom, Inc.)
• Original Assignee: Avago Technologies General IP PTE Limited (Broadcom, Inc.)
• Inventors: Weingarten, Hanan, Katz, Michael, Bar, Ilan
• Primary Examiner(s): Bragdon, Reginald
• Assistant Examiner(s): Namazi, Mehdi
• Application Number:
  US14/050,913
• Time in Patent Office: 985 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 12/0246 in block erasable memory, e.g. flash memory
• CPC Class Codes: G06F 2212/1016 Performance improvement
• CPC Class Codes: G06F 2212/7203 Temporary buffering, e.g. using volatile buffer or dedicated buffer blocks
• CPC Class Codes: G06F 2212/7204 Capacity control, e.g. partitioning, end-of-life degradation

Abstract

A method of managing a non-volatile memory module, the method may include: receiving data sectors during a set of at least one write transactions; selecting, out of the currently buffered portions, to-be-merged memory space portions and to-be-cleaned memory space portions; merging, for each to-be-merged memory space portion and before the buffer becomes full, data sectors that belong to the to-be-merged memory space portion into a sequential portion of the non-volatile memory module, wherein the sequential portion differs from the buffer; and copying, for each to-be-cleaned memory space and before the buffer becomes full, data sectors that belong to the to-be-cleaned memory space portion into a buffer block of the buffer.

Claims

1. A method of managing a non-volatile memory module, the method comprising:

receiving data sectors during a set of at least one write transactions;

wherein each data sector belongs to a memory space portion out of multiple memory space portions;

writing the data sectors into a buffer of the non-volatile memory module;

wherein the buffer comprises multiple buffer blocks;

wherein the multiple memory space portions comprise currently buffered memory space portions that are associated with currently buffered data sectors;

selecting, out of the currently buffered portions, to-be-merged memory space portions and to-be-cleaned memory space portions;

wherein a target ratio between (a) a number of to-be-cleaned memory space portions and (b) a sum of the number of to-be-cleaned memory space portions and a number of to-be-merged memory space portions ranges between an upper limit and a lower limit, wherein the upper limit differs from the lower limit and exceeds one half;

merging, for each to-be-merged memory space portion and before the buffer becomes full, data sectors that belong to the to-be-merged memory space portion into a sequential portion of the non-volatile memory module, wherein the sequential portion differs from the buffer;

copying, for each to-be-cleaned memory space and before the buffer becomes full, data sectors that belong to the to-be-cleaned memory space portion into a buffer block of the buffer.

2. The method according to claim 1 wherein the target ratio exceeds 0.75.

3. The method according to claim 1 wherein the target ratio exceeds 0.8.

4. The method according to claim 1 comprising selecting the to-be-merged memory space portions and the to-be-cleaned memory portions based upon a number of buffered data sectors for each of the currently buffered portions.

5. The method according to claim 1 comprising selecting the to-be-merged memory space portions as memory space portions that have higher numbers of buffered data sectors in relation to numbers of buffered data sectors of to-be-cleaned memory space portions.

6. The method according to claim 1 comprising selecting the to-be-merged memory space portions as memory space portions that have highest numbers of buffered data sectors out of the currently buffered portions.

7. The method according to claim 1 comprising selecting the to-be-cleaned memory space portions as memory space portions that have lowest numbers of buffered data sectors out of the currently buffered portions.

8. The method according to claim 1 comprising selecting the to-be-merged memory space portions and the to-be-cleaned memory portions based upon timings of buffering data sectors that belong to the currently buffered portions.

9. The method according to claim 1 comprising selecting the to-be-merged memory space portions as memory space portions that have older buffered data sectors in relation to data sectors of to-be-cleaned memory space portions.

10. The method according to claim 1 comprising selecting the to-be-merged memory space portions and the to-be-cleaned memory portions based upon (a) a number of buffered data sectors for each of the currently buffered portions, and (b) timings of buffering data sectors that belong to the currently buffered portions.

11. The method according to claim 1 comprising:

receiving data sectors during multiple sets of at least one write transactions; and

repeating the writing, the selecting, the merging and the copying for each set of at least one write transactions to provide multiple actual ratios between numbers of to-be-cleaned memory space portions and to-be-merged memory space portions;

wherein the target ratio between each number of to-be-cleaned memory space portions and each number of to-be-merged memory space portions is a statistical function of the multiple actual ratios.

12. The method according to claim 11 wherein the statistical function is an average.

13. The method according to claim 1 wherein the merging comprises executing multiple merging fragments and wherein the copying comprises executing multiple cleaning fragments.

14. The method according to claim 13 wherein the multiple merging fragments and multiple cleaning fragments are executed in an interleaved manner.

15. The method according to claim 14 wherein the multiple merging fragments and multiple cleaning fragments are executed in a non-interleaved manner.

16. The method according to claim 14 comprising determining an inter-fragment threshold that represents a number of data segments to be written to the buffer between an execution of two consecutive fragments of any type of fragment out of merging fragments and cleaning fragment.

17. The method according to claim 16 comprising determining the inter-fragment threshold such as to prevent an overflow of the buffer due to (a) a completion of merging and the copying and (b) a reception of new data sectors during the completion of merging and the copying.

18. The method according to claim 16 wherein the determining of the inter-fragment threshold is responsive to a size of a residual buffer space that is allocated for receiving data sectors during an execution of the merging.

19. The method according to claim 16 wherein the determining of the inter-fragment threshold is responsive to at least some parameters out of:

a number (Np) of data sectors required to fill the buffer block;

a size (g) of residual buffer space that is allocated for receiving data sectors during an execution of the merging;

a number (Pc) of data sectors copied during an entire cleaning;

a number (Nu) of currently not-empty buffer blocks;

numbers (L(1)−

L(Nu)) of primary memory space portions per each buffer block;

wherein a primary memory space portion of a buffer block is a memory space portion of which an earliest written data sector that is still buffered was written to the buffer block;

a number (nl) of data sectors per memory space portion;

the target ratio (alpha);

a number (f) of fragments required for completing the merging and the cleaning;

a number (NF) of free buffer blocks; and

a number (Q) of uncompleted fragments requires for a completion of the merging and copying that are executed when determining of the inter-fragment threshold.

20. The method according to claim 19 wherein the determining of the inter-fragment threshold comprises calculating a maximal value (t) out of intermediate values calculated for different values of an index j that ranges between 1 and Nu of a ratio between a sum of primary space portions per buffer block from L(1) to a k'"'"'th primary space portion L(k) for index k values ranging between and j plus g, and (b) a value of index j plus NF minus 1.

21. The method according to claim 20 wherein the determining of the inter-fragment threshold comprises:

calculating a first intermediate value that equals Np−

Pc;

calculating r wherein r is an absolute value of (f*Pc/Np);

calculating a second intermediate value that equals f*t*(1-alpha)+Q+r; and

calculating the inter-fragment threshold by dividing the first intermediate value by the second intermediate value.

22. The method according to claim 20 comprising changing a relationship between the number of to-be-merged memory space portions and to-be-cleaned memory space portions if the first intermediate value is negative.

23. The method according to claim 1 comprising selecting memory space portions that are associated with data sectors stored in an oldest buffer block of the buffer;

wherein the oldest buffer block stores an earliest written data sector out of all data sectors currently buffered in the buffer.

24. The method according to claim 1 comprising checking when the merging and the copying to the buffer block are completed, whether the buffer block is full;

and delaying further merging and copying until selecting new to-be-merged memory space portions and new to-be-cleaned memory space portions, if it is determined that the buffer block is not full.

25. A non-transitory computer readable medium that stores instructions that once executed by a memory controller cause the memory controller to:

receive data sectors during a set of at least one write transactions;

wherein each data sector belongs to a memory space portion out of multiple memory space portions;

write the data sectors into a buffer of the non-volatile memory module;

wherein the buffer comprises multiple buffer blocks;

wherein the multiple memory space portions comprise currently buffered memory space portions that are associated with currently buffered data sectors;

select, out of the currently buffered portions, to-be-merged memory space portions and to-be-cleaned memory space portions;

wherein a target ratio between (a) a number of to-be-cleaned memory space portions and (b) a sum of the number of to-be-cleaned memory space portions and a number of to-be-merged memory space portions ranges between an upper limit and a lower limit, wherein the upper limit differs from the lower limit and exceeds one half;

merge, for each to-be-merged memory space portion and before the buffer becomes full, data sectors that belong to the to-be-merged memory space portion into a sequential portion of the non-volatile memory module, wherein the sequential portion differs from the buffer; and

copy, for each to-be-cleaned memory space and before the buffer becomes full, data sectors that belong to the to-be-cleaned memory space portion into a buffer block of the buffer.

26. A system, comprising an interface and a memory controller;

wherein interface is arranged to receive data sectors during a set of at least one write transactions;

wherein each data sector belongs to a memory space portion out of multiple memory space portions;

wherein the memory controller is arranged to;

write the data sectors into a buffer of a non-volatile memory module;

wherein the buffer comprises multiple buffer blocks;

wherein the multiple memory space portions comprise currently buffered memory space portions that are associated with currently buffered data sectors;

select, out of the currently buffered portions, to-be-merged memory space portions and to-be-cleaned memory space portions;

wherein a target ratio between (a) a number of to-be-cleaned memory space portions and (b) a sum of the number of to-be-cleaned memory space portions and a number of to-be-merged memory space portions ranges between an upper limit and a lower limit, wherein the upper limit differs from the lower limit and exceeds one half;

merge, for each to-be-merged memory space portion and before the buffer becomes full, data sectors that belong to the to-be-merged memory space portion into a sequential portion of the non-volatile memory module, wherein the sequential portion differs from the buffer;

copy, for each to-be-cleaned memory space and before the buffer becomes full, data sectors that belong to the to-be-cleaned memory space portion into a buffer block of the buffer.