Merging mapping metadata to promote reference counting efficiency

1. A method of managing metadata in a data storage system, the method comprising:

maintaining (i) a parent BPS (block pointer set) for mapping a base range of a storage object and (ii) a child BPS for mapping a logical copy of the base range, each BPS including multiple block pointers that map data of the respective range, each block pointer having an attribute that indicates whether that block pointer is a source or a copy, the block pointers in each child BPS initially designated as copies when that child BPS is created;

maintaining reference counts on lower-level mapping structures pointed to by the block pointers of the parent BPS and the child BPS, the reference count for each lower-level mapping structure counting a number of source block pointers pointing to that mapping structure and being unaffected by any copy block pointers that also point to that lower-level mapping structure; and

subsequent to the parent BPS becoming inaccessible from a namespace of the data storage system, performing a metadata-merge operation between the parent BPS and the child BPS, the metadata-merge operation (i) updating each copy block pointer at a respective pointer location in the child BPS to a source, in response to a block pointer at the corresponding location in the parent BPS being a source, and (ii) invalidating, in the parent BPS, the block pointer corresponding to each updated block pointer in the child BPS, the metadata-merge operation thereby transferring reference-counted source attributes from the parent BPS to the child BPS without updating reference counts of lower-level mapping structures pointed to by the updated block pointers.

2. The method of claim 1 , wherein the parent BPS is part of a parent BPS tree that includes the parent BPS and multiple lower-level BPS's that provide lower-level mapping structures of the parent BPS, and wherein the block pointers in the parent BPS point to the lower-level BPS's in the parent BPS tree.

3. The method of claim 2 , wherein the child BPS is a sole child BPS of the parent BPS, such that the parent BPS has no other child BPS's.

4. The method of claim 3 , further comprising performing a cleanup operation, the cleanup operation including:

identifying a set of unique source block pointers in the parent BPS, which are not invalidated by the metadata-merge operation; and

for each of the set of unique source block pointers, decrementing the reference count of a respective lower-level BPS pointed to by that unique source block pointer.

5. The method of claim 4 , wherein performing the cleanup operation further includes, in response to the decremented reference count of a lower-level BPS indicating zero references, freeing that lower-level BPS such that it is no longer used to map the base range.

6. The method of claim 4 , wherein the child BPS has a parent field configured to store a quasi-physical address of the parent BPS, and wherein performing the cleanup operation further includes:

freeing the parent BPS such that it is no longer used to map the base range; and

deleting or otherwise invalidating the quasi-physical address of the parent BPS from the parent field of the child BPS.

7. The method of claim 1 ,

wherein the child BPS is a first child BPS created along with a second child BPS as part of a copy-copy-delete operation that created the first and second child BPS's and orphaned the parent BPS, such that the parent BPS became inaccessible from a namespace of the data storage system, and

wherein the metadata-merge operation is performed in response to a parent-child relationship being severed between the parent BPS and the second child BPS.

8. The method of claim 7 , the parent-child relationship is severed between the parent BPS and the second child BPS in response to one of (i) the second child BPS being deleted or (ii) the second child BPS being entirely overwritten.

9. The method of claim 1 , further comprising:

providing the parent BPS in a first block of a quasi-physical address space of the data storage system; and

providing the child BPS in a second block of the quasi-physical address space,

wherein each of the first block and the second block includes:

(i) multiple block pointer fields, each block pointer field configured to store an address in the quasi-physical address space of a pointed-to structure and an attribute field configured to identify the block pointer as a source or a copy;

(ii) a reference count field configured to indicate a number of source block pointers that point to the respective block;

(iii) a parent field configured to identify a quasi-physical address of another BPS from which the BPS in the respective block was copied; and

(iv) at least one child field configured to identify a quasi-physical address of another BPS that was created as a copy of the BPS in the respective block.

10. A data storage system, comprising control circuitry that includes a set of processing units coupled to memory, the control circuitry constructed and arranged to:

maintain (i) a parent BPS (block pointer set) for mapping a base range of a storage object and (ii) a child BPS for mapping a logical copy of the base range, each BPS including multiple block pointers that map data of the respective range, each block pointer having an attribute that indicates whether that block pointer is a source or a copy, the block pointers in each child BPS initially designated as copies when that child BPS is created;

maintain reference counts on lower-level mapping structures pointed to by the block pointers of the parent BPS and the child BPS, the reference count for each lower-level mapping structure counting a number of source block pointers pointing to that mapping structure and being unaffected by any copy block pointers that also point to that lower-level mapping structure; and

subsequent to the parent BPS becoming inaccessible from a namespace of the data storage system, perform a metadata-merge operation between the parent BPS and the child BPS, the metadata-merge operation (i) updating each copy block pointer at a respective pointer location in the child BPS to a source, in response to a block pointer at the corresponding location in the parent BPS being a source, and (ii) invalidating, in the parent BPS, the block pointer corresponding to each updated block pointer in the child BPS, the metadata-merge operation thereby transferring reference-counted source attributes from the parent BPS to the child BPS without updating reference counts of lower-level mapping structures pointed to by the updated block pointers.

11. The data storage system of claim 10 , wherein the control circuitry constructed and arranged to:

provide the parent BPS in a first block of a quasi-physical address space of the data storage system; and

provide the child BPS in a second block of the quasi-physical address space,

wherein each of the first block and the second block includes:

(i) multiple block pointer fields, each block pointer field configured to store an address in the quasi-physical address space of a pointed-to structure and an attribute field configured to identify the block pointer as a source or a copy;

(ii) a reference count field configured to indicate a number of source block pointers that point to the respective block;

(iii) a parent field configured to identify a quasi-physical address of another BPS from which the BPS in the respective block was copied; and

(iv) at least one child field configured to identify a quasi-physical address of another BPS that was created as a copy of the BPS in the respective block.

12. A computer program product including a set of non-transitory, computer-readable media having instructions which, when executed by control circuitry of a data storage system, cause the control circuitry to perform a method of managing metadata, the method comprising:

maintaining (i) a parent BPS (block pointer set) for mapping a base range of a storage object and (ii) a child BPS for mapping a logical copy of the base range, each BPS including multiple block pointers that map data of the respective range, each block pointer having an attribute that indicates whether that block pointer is a source or a copy, the block pointers in each child BPS initially designated as copies when that child BPS is created;

maintaining reference counts on lower-level mapping structures pointed to by the block pointers of the parent BPS and the child BPS, the reference count for each lower-level mapping structure counting a number of source block pointers pointing to that mapping structure and being unaffected by any copy block pointers that also point to that lower-level mapping structure; and

subsequent to the parent BPS becoming inaccessible from a namespace of the data storage system, performing a metadata-merge operation between the parent BPS and the child BPS, the metadata-merge operation (i) updating each copy block pointer at a respective pointer location in the child BPS to a source, in response to a block pointer at the corresponding location in the parent BPS being a source, and (ii) invalidating, in the parent BPS, the block pointer corresponding to each updated block pointer in the child BPS, the metadata-merge operation thereby transferring reference-counted source attributes from the parent BPS to the child BPS without updating reference counts of lower-level mapping structures pointed to by the updated block pointers.

13. The computer program product of claim 12 , wherein the parent BPS is part of a parent BPS tree that includes the parent BPS and multiple lower-level BPS's that provide lower-level mapping structures of the parent BPS, and wherein the block pointers in the parent BPS point to the lower-level BPS's in the parent BPS tree.

14. The computer program product of claim 13 , wherein the child BPS is a sole child BPS of the parent BPS, such that the parent BPS has no other child BPS's.

15. The computer program product of claim 14 , wherein the method further comprises performing a cleanup operation, the cleanup operation including:

identifying a set of unique source block pointers in the parent BPS, which are not invalidated by the metadata-merge operation; and

for each of the set of unique source block pointers, decrementing the reference count of a respective lower-level BPS pointed to by that unique source block pointer.

16. The computer program product of claim 15 , wherein performing the cleanup operation further includes, in response to the decremented reference count of a lower-level BPS indicating zero references, freeing that lower-level BPS such that it is no longer used to map the base range.

17. The computer program product of claim 15 , wherein the child BPS has a parent field configured to store a quasi-physical address of the parent BPS, and wherein performing the cleanup operation further includes:

freeing the parent BPS such that it is no longer used to map the base range; and

deleting or otherwise invalidating the quasi-physical address of the parent BPS from the parent field of the child BPS.

18. The computer program product of claim 12 ,

wherein the child BPS is a first child BPS created along with a second child BPS as part of a copy-copy-delete operation that created the first and second child BPS's and orphaned the parent BPS, such that the parent BPS became inaccessible from a namespace of the data storage system, and

wherein the metadata-merge operation is performed in response to a parent-child relationship being severed between the parent BPS and the second child BPS.

19. The computer program product of claim 18 , the parent-child relationship is severed between the parent BPS and the second child BPS in response to one of (i) the second child BPS being deleted or (ii) the second child BPS being entirely overwritten.

20. The computer program product of claim 12 , wherein the method further comprises:

providing the parent BPS in a first block of a quasi-physical address space of the data storage system; and

providing the child BPS in a second block of the quasi-physical address space,

wherein each of the first block and the second block includes:

(i) multiple block pointer fields, each block pointer field configured to store an address in the quasi-physical address space of a pointed-to structure and an attribute field configured to identify the block pointer as a source or a copy;

(ii) a reference count field configured to indicate a number of source block pointers that point to the respective block;

(iii) a parent field configured to identify a quasi-physical address of another BPS from which the BPS in the respective block was copied; and

(iv) at least one child field configured to identify a quasi-physical address of another BPS that was created as a copy of the BPS in the respective block.