Detection and correction of dropped write errors in a data storage system

A RAID system, comprising:

• an array of disk storage drives;
• an RAID array controller operable to receive first data from a host;
• a parity generator operable to generate a first checksum associated with first data;
• a data write module operable to write data, including the first data and the first checksum, to the array in a full stride write;
• a dropped write check module operable to periodically check the written data to determine whether a dropped write event has occurred;
• and
• a data state module operable to:
• classify a state of the first data as being in a first “new data, unconfirmed” state when the first data is written to the array;
• change the state of the first data from the first state to a second “single dropped write confirmed” state if a dropped write event is detected during a periodic check;
• change the state of the first data from the first state to a third “confirmed good” state if no dropped write event is detected during a periodic check;
• receive second data from the host;
• if the second data is to be written to the storage device in a full stride write, generate at least a second checksum and write the second data and the at least second checksum to the storage device in a full stride write;
• and
• if the second data is to be written to the storage device in a read modify write (RMW):
• generate at least one RMW checksum and write the second data and the at least one RMW checksum to the storage device in a read modify write;
• and
• change the state of the second data from the second state to a fourth “unquantifyable” state.