The Linux Kernel's VFS Layer
The Virtual File System (VFS) layer [1] providesa uniform interface for the kernel to deal with various I/O requestsand specifies a standard interface that each file system must support.Through this layer, one kernel can mount several different types offile systems (e.g. EXT2FS, ISO9660FS, NFS,...) into the same tree structure. We worked with version 2.2.12of the Linux kernel and confined our changes to the VFS layer. Bydoing all of our changes in the VFS layer we kept our predictiveprefetching totally independent of the underlying file system.
Arguably, the most important service the VFS layer provides is auniform I/O data cache. Linux maintains four caches of I/O data: page cache, i-node cache, buffer cache and directory cache. Figure 5 shows these caches andhow they interact with the kernel, each other and user level programs.The page cache combines virtual memory and file data. The i-node cache keeps recently accessed file i-nodes. The buffer cache interfaces with block devices, and caches recently usedmeta-data disk blocks. The Linux kernel reads file data through thebuffer cache, but keeps the data in the page cache for reuse on futurereads. The directory cache (d-cache) keeps in memory a tree thatrepresents a portion of the file system's directory structure. Thistree maps a file's path name to an i-node structure and speeds up filepath name look up. The basic element of the d-cache is a structurecalled the d-entry.
We implemented our methods of modeling file access patterns by addingone field to the d-entry structure. The various models wouldattach their modeling data structure to this pointer. For the lastsuccessor model this consisted of just a device and inode number. Forthe partitioned models this was a pointer to the partition that beganwith the file that the d-entry identified. After each fileaccess the model would update its predictions. The prefetch enginewas then called and would use these predictions to prefetch file data.