US Patent 11500581 Efficient TLP fragmentations in extended LBA environment

The present disclosure generally relates to efficient transfer layer packet (TLP) fragmentation in a data storage device. For an unaligned read from host flow, an amount of data sufficient to be aligned is transferred to the memory device from the host while the remainder of the data is stored in cache of the data storage device to be delivered to memory device at a later time. For an unaligned write to host flow, the unaligned data is written to cache and at a later time the cache will be flushed to the host device. In both cases, while the total data would be unaligned, a portion of the data is placed in cache so that the data not placed in cache is aligned. The data in cache is delivered at a later point in time.