Page write to non-volatile data storage with failure recovery
Modifications to be made to a page of non-volatile data storage and a current transaction sequence number are store into a protected memory. The current transaction sequence number is added to a previous version of the page and the modifications are applied to the previous version of the page to create a new version of the page that is written to the non-volatile data storage. A failure of the write may cause the stored modifications to be re-applied to the page in response to a transaction sequence number in a retrieved version of the page not matching the saved current transaction sequence number. A write completion status stored in a protected memory may indicate which ones of multiple copies of the page have been overwritten with the new version. Alternatively, an error-detecting code may be generated and written to non-volatile data storage with the new version to the multiple copies.
1. A method of performing a page-write to a page of non-volatile data storage, comprising:
saving a set of modifications and a current transaction sequence number in a protected memory;
creating a new version of the page of non-volatile data storage by adding the current transaction sequence number to a previous version of the page and applying the set of modifications to the previous version of the page;
writing the new version of the page to non-volatile data storage; and
in response to a failure, retrieving a current version of the page from the non-volatile data storage, comparing a transaction sequence number contained in the current version of the page to a current transaction sequence number, and re-applying the set of modifications to the page of non-volatile data storage in response to the transaction sequence number contained in the current version of the page being less than the current transaction sequence number.
2. The method of claim 1 , wherein multiple copies of the page are maintained in the non-volatile data storage, each one of the multiple copies of the page containing a transaction sequence number, and further comprising:
in response to the new version of the page being written to the non-volatile data storage,
overwriting, within the non-volatile data storage, each of the multiple copies of the page with the new version of the page, and
after each one of the multiple copies of the page is overwritten with the new version of the page, updating a write completion status stored in protected memory to indicate which ones of the multiple copies of the page have been successfully overwritten with the new version of the page.
3. The method of claim 2 , further comprising:
in response to the failure, selecting one of the multiple copies of the page as the current version of the page based on the write completion status stored in the protected memory.
4. The method of claim 3 , wherein selecting one of the multiple copies of the page as the current version of the page based on the write completion status stored in the protected memory comprises:
in response to the write completion status stored in the protected memory indicating that no copies of the multiple copies of the page have been successfully overwritten with the new version of the page to which the set of modifications were previously applied, selecting one of the multiple copies of the page as the current version of the page that is the same as the previous version of the page.
5. The method of claim 3 , wherein selecting one of the multiple copies of the page as the current version of the page based on the write completion status stored in the protected memory comprises:
in response to the write completion status saved in the protected memory indicating that at least one of the multiple copies of the page has been successfully overwritten with the new version of the page to which the set of modifications were previously applied, selecting one of the multiple copies of the page as the current version of the page that is the same as the new version of the page.
6. The method of claim 1 , wherein multiple copies of the page are maintained in the non-volatile data storage, wherein each copy of the page in the non-volatile data storage has an error-detecting code stored therewith, and further comprising:
calculating an error-detecting code for the new version of the page;
adding the error-detecting code for the new version of the page to the new version of the page before writing the new version of the page to the non-volatile data storage; and
in response to the failure, selecting one of the multiple copies of the page as the current version of the page such that the error-detecting code for the current version of the page indicates that a contents of the current version of the page have not been corrupted.
7. The method of claim 1 , further comprising:
prior to saving the current transaction sequence number to the protected memory, generating the current transaction sequence number by incrementing a last previously used transaction sequence number.
8. A data storage system, comprising:
processing circuitry and memory coupled to the processing circuitry, the memory storing instructions, wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to:
save a set of modifications and a current transaction sequence number in a protected memory;
create a new version of the page of non-volatile data storage by adding the current transaction sequence number to a previous version of the page and applying the set of modifications to the previous version of the page;
write the new version of the page to non-volatile data storage; and
in response to a failure, retrieve a current version of the page from the non-volatile data storage, compare a transaction sequence number contained in the current version of the page to a current transaction sequence number, and re-apply the set of modifications to the page of non-volatile data storage in response the transaction sequence number contained in the current version of the page being less than the current transaction sequence number.
9. The data storage system of claim 8 , wherein multiple copies of the page are maintained in the non-volatile data storage, each one of the multiple copies of the page containing a transaction sequence number, and wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to:
in response to the new version of the page being written to the non-volatile data storage,
overwrite, within the non-volatile data storage, each of the multiple copies of the page with the new version of the page, and
after each one of the multiple copies of the page is overwritten with the new version of the page, update a write completion status stored in protected memory to indicate which ones of the multiple copies of the page have been successfully overwritten with the new version of the page.
10. The data storage system of claim 9 , wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to:
in response to the failure, select one of the multiple copies of the page as the current version of the page based on the write completion status stored in the protected memory.
11. The data storage system of claim 10 , wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to select one of the multiple copies of the page as the current version of the page based on the write completion status stored in the protected memory at least in part by causing the processing circuitry to:
in response to the write completion status stored in the protected memory indicating that no copies of the multiple copies of the page have been successfully overwritten with the new version of the page to which the set of modifications were previously applied, select one of the multiple copies of the page as the current version of the page that is the same as the previous version of the page.
12. The data storage system of claim 10 , wherein the instructions, when executed by the processing circuitry, cause the processing circuitry to select one of the multiple copies of the page as the current version of the page based on the write completion status stored in the protected memory at least in part by causing the processing circuitry to:
in response to the write completion status saved in the protected memory indicating that at least one of the multiple copies of the page has been successfully overwritten with the new version of the page to which the set of modifications were previously applied, select one of the multiple copies of the page as the current version of the page that is the same as the new version of the page.
13. The data storage system of claim 8 , wherein multiple copies of the page are maintained in the non-volatile data storage, wherein each copy of the page in the non-volatile data storage has an error-detecting code stored therewith, and wherein the instructions, when executed by the processing circuitry, further cause the processing circuitry to:
calculate an error-detecting code for the new version of the page;
add the error-detecting code for the new version of the page to the new version of the page before writing the new version of the page to the non-volatile data storage; and
in response to the failure, select one of the multiple copies of the page as the current version of the page such that the error-detecting code for the current version of the page indicates that a contents of the current version of the page have not been corrupted.
14. The data storage system of claim 8 , wherein the instructions, when executed by the processing circuitry, further cause the processing circuitry to:
prior to saving the current transaction sequence number to the protected memory, generate the current transaction sequence number by incrementing a last previously used transaction sequence number.
15. A computer program product, comprising:
a non-transitory, computer-readable medium having instructions stored thereon that, when executed by processing circuitry, cause the processing circuitry to perform a method comprising the steps of:
saving a set of modifications and a current transaction sequence number in a protected memory;
creating a new version of the page of non-volatile data storage by adding the current transaction sequence number to a previous version of the page and applying the set of modifications to the previous version of the page;
writing the new version of the page to non-volatile data storage; and
in response to a failure, retrieving a current version of the page from the non-volatile data storage, comparing a transaction sequence number contained in the current version of the page to a current transaction sequence number, and re-applying the set of modifications to the page of non-volatile data storage in response to the transaction sequence number contained in the current version of the page being less than the current transaction sequence number.