OpenIoT & ELC 2016

The NAND framework has been introduced a long time ago (in Linux 2.4.6) and has organically evolved since then to support new NAND chips and NAND controllers of growing complexity.

While, from a functional point of view, the NAND framework is able to support new NAND controllers and new NAND chips with minimal changes, the way it has been designed prevents code factorization and complexifies the addition of new features like multi-plane and sequential operations or DDR NAND support. Another issue is the lack of separation between the NAND controller and NAND chip aspects, which misleads many developers.

After giving an overview of the current architecture of the MTD and NAND subsystems and highlighting its shortcomings, this talk will detail some of the reworks we have done or are planning to do.