Error recovery from submission queue fetching errors
An apparatus includes a non-volatile memory media and a storage controller. The storage controller is configured to fetch a storage command from a submission queue of a host. The submission queue has a submission queue identifier (SQID). The storage controller then determines a submission queue fetch error in response to receiving a Transport Layer Packet (TLP) error as a result of fetching the storage command. Next, the storage controller is configured to determine a command identifier (CID) for the storage command associated with the submission queue fetch error. The storage controller then sends a completion message to the host. The completion message uniquely identifies the storage command of the submission queue associated with the submission queue fetch error using the SQID and CID.
1. An apparatus, comprising:
a non-volatile memory media; and
a storage controller configured to:
fetch a storage command from a submission queue in a host, the submission queue having a submission queue identifier (SQID);
determine a submission queue fetch error in response to receiving a Transport Layer Packet (TLP) error as a result of fetching the storage command;
determine a command identifier (CID) for the storage command associated with the submission queue fetch error based on a command identifier allocation model, wherein the command identifier allocation model is provided to the storage controller by the host prior to the storage controller fetching the storage command; and
send a completion message to the host, the completion message comprising the SQID and the CID to uniquely identify the storage command of the submission queue associated with the submission queue fetch error, wherein the completion message indicates to the host that the storage command is to be either:
retired; or
resubmitted to the submission queue for the controller to fetch.
2. The apparatus of claim 1 , wherein storage controller is further configured to derive the command identifier allocation model based on a storage command history.
3. The apparatus of claim 1 , wherein storage controller is further configured to predict the command identifier allocation model.
4. A system, comprising:
a host having a submission queue;
a Peripheral Component Interconnect Express (PCIe) bus;
a storage device coupled to the host by way of the PCIe bus, the storage device comprising:
non-volatile memory media; and
a storage controller configured to communicate with the host over the PCIe bus using a Non-Volatile Memory express (NVMe) protocol, the storage controller comprising:
a PCIe medium access control (MAC) and physical layer (PHY) interface;
a command fetcher configured to fetch a storage command from the submission queue in the host;
a queue manager configured to manage a submission queue identifier (SQID) and to determine a submission queue fetch error in response to receiving a Transport Layer Packet (TLP) error as a result of fetching the storage command;
a command identifier tracker configured to determine a command identifier (CID) for the storage command associated with the submission queue fetch error, the command identifier based on a command identifier allocation model, wherein the command identifier allocation model is provided to the storage controller by the host prior to the command fetcher fetching the storage command; and
wherein the queue manager is configured to post a completion message to the host, the completion message comprising the SQID and the CID and configured to uniquely identify the storage command of the submission queue associated with the submission queue fetch error, wherein the completion message indicates to the host that the storage command is to be either:
retired; or
resubmitted to the submission queue for the controller to fetch.
5. The system of claim 4 , wherein the host defines the command identifier allocation model when the host sets up a submission queue and a completion queue with the storage device.
6. The system of claim 4 , wherein the command identifier tracker defines the command identifier allocation model as a default command identifier allocation model in response to the host failing to define the command identifier allocation model.
7. The system of claim 4 , wherein the command identifier tracker is configured to derive the command identifier allocation model based on a storage command history.
8. The system of claim 4 , wherein the Transport Layer Packet (TLP) error comprises an uncorrectable packet error.
9. The system of claim 4 , wherein the queue manager is configured to retry fetching the storage command in response to the submission queue fetch error.
10. The system of claim 4 , wherein the command identifier tracker is configured to monitor a plurality of storage commands and to validate the command identifier allocation model by comparing command identifiers of the plurality of storage commands with corresponding command identifier based on the command identifier allocation model.
11. The system of claim 4 , wherein the submission queue fetch error comprises a PCIe Express Transport Layer Protocol fetch error.
12. A method, comprising:
attempting to fetch a storage command from a submission queue in a host, the submission queue having a submission queue identifier (SQID);
receiving a storage command from the host, the storage command configured to set up the submission queue and a completion queue as a submission completion queue pair for storage operations between the host and a storage device and wherein the storage command comprises a command identifier allocation model for use in allocating command identifiers for storage commands for the submission completion queue pair;
detecting a submission queue fetch error as a result of attempting to fetch the storage command;
determining a command identifier (CID) for the storage command associated with the submission queue fetch error based on the command identifier model; and
posting a completion message to the host, the completion message comprising the SQID and the CID to uniquely identify the storage command of the submission queue associated with the submission queue fetch error, wherein the completion message indicates to the host that the storage command is to be either:
retired; or
resubmitted to the submission queue for to be re-fetched.
13. The method of claim 12 , wherein determining the command identifier based on the command identifier model comprises deriving the command identifier based on a storage command history.
14. The method of claim 12 , wherein determining the command identifier based on the command identifier model comprises forecasting the command identifier based on an industry convention.
15. The method of claim 12 , further comprising:
associating a default command identifier allocation model with the submission completion queue pair, in response to the host failing to designate the command identifier allocation model.
16. The method of claim 12 , wherein determining the command identifier for the storage command comprises determining the storage command based on the command identifier allocation model provided by the host.
17. The method of claim 12 , wherein the completion message comprises a completion identifier and the method further comprises generating the completion identifier using the CID and the SQID.