SpadFS

"SpadFS" is a newly developed filesystem project, designed and developed as a PhD thesis. It aims to incorporate advanced filesystem features such as crash recovery and faster directories, while keeping code complexity in check. This is achieved by using crash counts instead of journaling, as journaling is often very complicated and bug-prone. Additionally, hash is used to organize directories, rather than btrees.

The "SpadFS" filesystem boasts several key features, including a new method for maintaining consistency across system crashes, which involves crash counts. It also supports 48-bit sector numbers, which can accommodate device sizes up to 144PB. The variable block size ranges from 512 bytes to the machine page size, taking into account the Linux page cache design. Smaller block sizes can significantly increase CPU usage.

"SpadFS" uses a specialized method for organizing large directories, similar to Fagin's extendible hashing, without using btrees. Files are embedded directly into the directory structure unless a hardlink is created. This allows for quicker ls -la commands, without the need to seek to inodes. Free space is described in lists of extents, rather than bitmaps, which is commonly seen in most filesystems. If the filesystem becomes too fragmented, a list of free extents is converted into a bitmap.

The latest release of "SpadFS" addresses some key issues, including crashing problems related to "spadfsck," and the creation of bad directory entries while creating files with names longer than 255 characters. Overall, "SpadFS" is a highly advanced and efficient filesystem, with many unique features to offer to users.