UBC Theses and Dissertations
A low overhead and consistent flash translation layer for embedded devices utilizing serial NOR flash Fazackerley, Scott Ronald
In today's world, embedded devices are playing an ever increasing and important role in daily life. Recent interest has focused on how data from embedded devices can be harnessed. Whether it is complex environmental parameters or the average temperature in our homes, devices store and process data. Due to the vast amount of data that is generated by devices, the ability to process data on device is beneficial as it reduces the amount of data that must be transferred off the device. Flash memory is the most commonly found storage medium on embedded devices but presents challenges for developers. Flash memory is managed through the use of a flash translation layer (FTL) to address the physical limitations of memory. No FTL is currently available for the smallest of devices due to resource limitations. This thesis examines the features that are currently used in FTLs and highlights their shortcomings for use with resource constrained devices. In response to these challenges, this work introduces an FTL architecture with a minimal RAM footprint that is robust and fault-tolerant for resource constrained embedded systems. Utilizing attributes of the Adesto serial NOR Dataflash memory, a unique flash translation layer for use with serial NOR flash has been developed. Key features focus on minimizing data transfer between host and flash while maintaining persistent address translation. In addition to a low overhead and robust flash translation layer, the FTL contains a deterministic garbage collection and wear levelling strategy. This work introduces the technique of masked overwriting for NOR flash, which demonstrates the use page overwriting for modification of specific data in place. This technique offers savings in terms of write times, page erases and complexity in managing data pages, resulting in a novel FTL strategy suitable for resource constrained devices.
Item Citations and Data
Attribution-NonCommercial-NoDerivatives 4.0 International