Analysis And Optimization Of The UBIFS File System For Flash Memory

As an external medium frequently applied in the embedded system,Flash Memory has several advantages,such as low power consumption,high capacity,low cost,small size,shock resistance,non-volatility and so on.Depending on the physical structure,Flash Memory can be divided into two types:NAND Flash and NOR Flash.NAND Flash is more widely used in the daily consumer electronics,because of the progressive promotion in the performance-price ratio and the rapid growth in the capacity.Flash File System is one of the file systems designed for Flash Memory.Currently the mainstream Flash File System includes:JFFS/JFFS2,YAFFS/YAFFS2,UBIFS and so on.In these Flash File Systems,analysis literature for UBIFS is deficient,and source code analysis literature for UBIFS is even scarce.For these reasons,this article attempts to probe deeply into UBIFS source code and proposes two optimizing algorithms.UBIFS is a developing Flash File System,and it is the successor of JFFS2 which is widely used by the embedded system.UBIFS is designed to solve the bottleneck of NAND Flash Memory serving as Memory Technology devices.UBIFS was added into Linux kernel 2.6.27 in October,2008.Currently UBIFS is widely used in smart phones and so on.The performance of UBIFS in terms of mount time,wear-leveling,I/O,bad block management is much better than other Flash File Systems.Through comparative experiments of several commonly used Flash File Systems,we explain the reason why we choose UBIFS as our study object.Then we introduce the necessity of using wear-leveling algorithm,and describe several typical wear-leveling algorithms.Because source code analysis literature for UBIFS is scarce,we analysis the Linux source code and describe four modules associate with UBIFS which includes VFS,MTD,UBIFS and UBI in details.At the same time,we describe some important technologies used in UBIFS.For problems encountered in source code analysis literature,we propose two optimization algorithms which include threshold-queue wear-leveling algorithm,suffix name and access right TNC memory release algorithm.Using our own experimental procedures,we demonstrate the effectiveness of these two optimized algorithms.