Projects

Ext2 File System Driver for Windows

Ext2Fsd is an open source linux ext2/ext3 file system driver for Windows systems (NT/2K/XP/VISTA, X86/AMD64).

sreadahead

Sreadahead is a daemon that reads data sequential by use from disk. Sreadahead does this by retrieving the read order (in timestamp format) from an actual boot sequence and storing a list of data that was read during that boot sequence in a flat database. On the next boot this file can be used to ... [More] load all the data described in the database into memory and in order of use. Sreadahead requires a kernel patch but works on all filesystem types since release 1.0. [Less]

Partclone

Partclone is a project like the well-known backup utility "Partition Image" a.k.a partimage. Partclone provide utilities to back up used blocks and design for higher compatibility of the file system using supported library like e2fslibs.

EXTLINUX

EXTLINUX is a new syslinux derivative, which boots from a Linux ext2/ext3 filesystem. It works the same way as SYSLINUX, with a few slight modifications.

ext3grep

ext3grep is a utility that can help you recover cleanly deleted files on an ext3 partition.

oberfs

Welcome to the developer site for OberFS!Project:OberFS is a two to ten times faster file system, designed to replace the 15-year-old NTFS and ext2 file systems, by exploiting modern PC's faster processors and larger RAMs. License:OberFS is an open source project with a GPL license. ... [More] Requirements:Windows: A Windows XP or newer machine with an empty partition for OberFS. If you want to fairly race OberFS against another file system, you need two identical partitions (i.e. on the same drive). You also need the Dokan library. Linux and BSD: Nothing yet. It would be great if it were also buildable under Linux and BSD via the FUSE library. Status:Revision r12 released August 12th, 2009 - a proof of technology release OberFS launches Firefox in 2 seconds vs 5 seconds for NTFS! Source available at... http://code.google.com/p/oberfs/ ... launches Firefox in 2 seconds, whereas NTFS takes 5 seconds. The smart read-ahead is simulated by just a big read-ahead, but this should be a reasonable approximation of final performance. (Ask me why and I'll explain it.) To reproduce (without needing to reconfigure anything in OberFS): 1)Install Dokan library. 2)On a separate drive (if you want a fair race) add an NTFS partition (N:) and an empty partition (X:) 3)Zero out sector zero on X:. OberFS looks for this to be sure the partition is unused. OBVIOUSLY YOU NEED TO BE CAREFUL! 4)Download "Reformat OberFS.bat", "Mount OberFS.bat", "Unmount OberFS.bat", and "OberFS.exe" in OberFS_r12_bin.zip from http://groups.google.com/group/dokan/files/ 5)Execute "Reformat OberFS.bat" and "Mount OberFS.bat" 6)Launch Firefox. This pre-caches your profile, so that the race is fair. (I don't know how to set up 3 separate profiles on the 3 partitions yet.) 7)Drag the "Mozilla Firefox" folder into both N: and the new M: 8)Launch em. All suggestions, ideas, questions, help, etc. eagerly welcome. Revision r10 OberFS works (under Windows)! And IS much faster that NTFS in testing so far. Test Results from 28-May-09 NTFS OberFS File Create 13.28 sec 6.95 sec Random Access 18.14 sec 10.63 secThe test algorithms are by Hironobu Suzuki, run on 1000 files. Drive and OS caches were cleared at the start for each test by rebooting. Note OberFS was mounted on a much slower drive, so identical drive numbers should be much better! NTFS volume - WDC 160 GB - HD Tune 2.1 reports avg MB/sec of 45.5 OberFS volume - IBM 60 GB - HD Tune 2.1 reports avg MB/sec of 28.3For those who want to experiment or develop OberFS without formatting a hard drive, code is supplied to emulate the sector IO with either ram or a file. Major Features:The OberFS design has 3 main techniques for speeding up the file system. Keeps ALL metadata (all volume information other than the file data bytes) in RAM. This exploits the vastly larger memories in modern computers. This is roughly 100-200 bytes per inode/directory. (Planned) Logs are kept, and analyzed in slack time, and files are rearranged so that clusters of files typically read sequentially are arranged on sequential tracks. This exploits modern hard drive's sector skewing and minimal track-to-track seek times. (Planned) Files that are only read entirely (for example, executables), and are compressible (for example, not jpgs) are compressed using file type specific compression algorithms, and decompressed on the fly on read. This exploits the vastly faster processors in modern computers. (Planned) Finally, (I know, I know, that's 4, but all good file systems do this) a smart cache system will speed up many reads and writes. Near Term Goals:Entice developers. (So let me know what I should be doing to attract help.) What's With the Name:At conception, this project was submitted to Google's Project 10^100 as HOFS For Hard disk Optimized File System. When I discovered that HOFS was already an open source project, I asked my 9-year-old son for a name. He suggested Super Uber FS. I checked out UberFS, which was also taken. Since Uber sorta kinda means over, as does Ober, and since my last name begins with "Ober", I snagged OberFS as a project name. What the heck, it worked out pretty well for Linus. :-) [Less]

filesystem-debugger

A Linux kernel module to see how the virtual file system and particularly how ext3 works.

uberfs

Welcome to the developer site for UberFS!Want your computer, everyone's computer to boot in three seconds? Me too. When I do the math, it looks doable. You are welcome to check my math and comment on it! If you have any time to give, and have some skills, please jump in! With your help, UberFS ... [More] will substantially change the way computers behave. Sooo much faster as to be different. But right now, UberFS is at the drawing board pre-alpha not-a-line-of-code-to-show-for-itself stage. This means, of course, that you can jump in at this early stage to help make a few high quality design choices for this project. To learn more:Why - Why does the world need a new file system? How - For a technical overview, go here. Help - Let's get the ball rolling with some feedback, shall we? (The site for those that want to, you know, actually use UberFS will be hosted at http://uberfs.org once an installer exists.) This project was conceived in response to Google's Project 10 to the 100th challenge. It was submitted as HOFS, for "Hard disk Optimized File System", but that name was in use for another open source project, and, let's face it, was a little boring. [Less]

ext2-in-user-space

Workspace for implementing ext2/3 file system in user space

numa

nEW uNIFIED mEMORY aCCESS -numa recovers deleted files from harddisk of Ext3FS/ Ext2 file system.