There are many ways for people to start out in Linux kernel development. One good place to start is the KernelJanitors project, where you can become familiar with the Linux kernel source tree and development style by making small cleanups and bug fixes all over the tree, together with the other kernel janitors.
For Kernel Hackers
Please add suitable projects here to help Computer Science students do something useful in the time they need to spend on projects anyway. Suitable projects:
- Are relatively self contained, so the code could be merged into the kernel after the student is done with the project.
- Have clearly defined functionality, so the student has goals and can determine whether (s)he achieved them.
- Can be of various sizes. Students need projects anywhere from 6 weeks part-time to 6 months full-time effort.
- If possible, are useful to the Linux kernel. Maybe something from your own TODO list that you did not get around to?
Small Linux features
A good next step is to implement small, relatively self contained, features that the Linux kernel needs but have not been implemented yet. When you "take" such a feature, please add your name in the "Developer" column and the date you decided to take the project in the "Date taken" column.
Project | Summary | Contact | Difficulty (1-10) | Developer | Date taken | Merged in |
64 bit versions of set/getrlimit, see bug 5042 | Andrew Morton - akpm (at) linux-foundation.org | 6 | Narendra Prasad (narendramind (at) gmail.com) | 26-Jan-2008 | 31-Mar-2008 | |
/dev/random statistics | interface that shows the amount of entropy generated via the interfaces per second or so and also the amount consumed | 1 | Marat Stanichenko (maratrus (at) mail.ru) | August 2008 | ||
power saving tweaks | 3 | various | ||||
swap out virtually contiguous pages together | Rik van Riel - riel (at) redhat.com | 5 | Rajesh.A.R raajeshar(at)gmail(dot)com Michael Pellon pellon (at) bcm (dot) tmc (dot) edu | Sept 15th 2009 December 17th 2009 | ||
mac80211 suspend/resume support | Johannes Berg - johannes (at) sipsolutions.net | 5 | Pankaj ( pankaj.linux@gmail.com ) / BobCopeland | 16-Jun-2008 | 19-Jan-2009 (in wireless-testing) | |
add reporting of socket information to pfiles SystemTap script | Eugene Teo - eteo (at) redhat.com | 3 | Luís Henriques - lhenriques (at) netvisao.pt | 17 Oct 2007 | 19 Jan 2008, see here | |
Increase speed for a build with no updates | Sam Ravnborg - sam (at) ravnborg.org | 5 | Kumar Vishal | 17 Sep 2008 | ASAP | |
Increase speed for a build which updates a single file | Sam Ravnborg - sam (at) ravnborg.org | ? | Kumar Vishal | 17 Sep 2008 | ASAP | |
Update menuconfig to a modern ncurses look & feel | Sam Ravnborg - sam (at) ravnborg.org | 5 | Nir Tzachar | 1 Sep 2008 | ||
Allow removal of select from Kconfig files | Jörn Engel - joern (at) logfs.org | 4 | Narendra Prasad (narendramind (at) gmail.com) | 16 Sep 2008 | ||
Add read balancing to RAID 1 | Rik van Riel - riel (at) redhat.com | 4 | Konstantin Sharlaimov - konstantin.sharlaimov (at) gmail.com ,prajith - emailprajith@gmail.com | 17 Oct 2007 | ||
Make all system calls asynchronous - or at least the ones that can stall. [As a MS dissertation idea, to work on] -- Sourish | b.phani@yahoo.co.in ,Sourish Banerjee - sourish (dot) banerjee (dot) 05 (at) gmail (dot) com | 5 | Phanisharma, Sourish Banerjee | 1st Apr 2008 15th, Jan 2009 Onwards | ||
Implement network interface default labels for Smack | Casey Schaufler - casey (at) schaufler-ca.com | 5 | Narendra Prasad (narendramind (at) gmail.com) | 23 Apr 2008 | ||
Implement IPv6 networking for Smack | Casey Schaufler - casey (at) schaufler-ca.com | 8 | Narendra Prasad (narendramind (at) gmail.com) | 23 Apr 2008 | ||
Place module text in to large pages instead of the small pages that vmalloc() uses | Dave Hansen - dave@sr71.net | 4 | Daniel Webster <dannywebster@gmail.com> "Joshi, Kedar" <kedar.joshi@intel.com> shashank karkare <karkareshashank@gmail.com> | 2014-06 |
Larger projects
Alternatively, there are some larger projects. Again, when you "take" such a project, please add your name in the "Developer" column and the date you decided to take the project in the "Date taken" column.
Project | Summary | Contact | Difficulty (1-10) | Developer | Date taken | Merged in |
Test mac80211 without real hardware | Johannes Berg - johannes (at) sipsolutions.net | depends | ? | various commits, mostly by Jouni | ||
Block rootkits using virtualization (Xen) | Rik van Riel - riel (at) redhat.com | 7 | prabir - android_online (at) yahoo.com | 16 Oct 2007 | ||
Block rootkits using virtualization (KVM) | Rik van Riel - riel (at) redhat.com | 7 | ||||
Better resource control | kalium (at) gmx.de | 7 | Sri Ram, kishore.kernel.dev at gmail.com | 02-05-10 | ||
Self-optimizing read-write block devices | kalium (at) gmx.de | 7 | Micah Gruber - micah.gruber (at) gmail.com | 19 Oct 2007 | ||
Proof that caching/buffering/transactions/syncs are handled correctly. | kalium (at) gmx.de | 10+ | ||||
Add impersonation feature to the kernel | Alex (dot) Slesarev (at) gmail.com | 7 | Onkar Mahajan - onkar[dot]n[dot]mahajan[at]gmail[dot]com | 21 Oct 2008 | ||
VMware image mounter | Being able to map different types of virtual disk formats to appear as blockdevice or as real ide/scsi device. For example, being able to mount VMware .vmdk virtual disks or use them as blockdevice. replace ancient/unstable vmware-mount.pl. See http://communities.vmware.com/message/749768 . news on 12/26/07: Apparently, it looks vmware is adressing this. Latest beta for their hosted products has a fuse based replacement for vmware-mount.pl/vmware-loop which is able to map a virtual disk to a flat file. mapping that file to a blockdevice is easy then (losetup or device-mapper) . closed source, though. | devzero (at) web.de | ? | |||
Real mode emulation for virtualization | Real mode support: VT support for real mode is terrible, so we need to do it in software. This means extending the x86 emulator (x86_emulate.c) to handle more instructions, and changing the execution loop to call the emulator for real mode | dor.laor (at) qumranet.com | 9 | Mohammed Gamal - m.gamal005[at]gmail[dot]com | ||
Guest shared memory | Add a qemu interface for sharing memory between guests. Using a pci device to expose the shared memory is probably a good starting point. | dor.laor (at) qumranet.com | 7 | Onkar Mahajan - onkar[dot]n[dot]mahajan[at]gmail[dot]com | 21 Oct 2008 | |
A device mapper target to use flash memory as a disk accelerator | riel (at) redhat.com | 6 | Himanshu Barthwal- himanshu_barthu@yahoomail.com Wilson Bright - rombright@gmail.com | 23 October 2008 15 Dec 2008 | ||
ext4 currently limits extents to 8MB | willy (at) infradead.org | 8 |
Projects that could use more help
Another possibility is to get involved with an existing kernel project that needs some help. Since these projects need multiple people, there is no need to claim these.
Project | Summary |
Compressed Caching is a new level in the virtual memory hierarchy, where pages are stored in some compressed format, decreasing the number of page faults that are serviced by slow hard disks. | |
The Linux1394 project has a seemingly ever-growing to-do list of bugs and other items, ranging from small cleanups to implementation of full drivers (e.g. IP over FireWire). If you have FireWire hardware, working on these drivers may be a good entry into kernel hacking because most FireWire specs are open, some even gratis. See the Wiki at linux1394.org for the to-do list and links to specifications. | |
This project addresses some shortcomings (fairness, performance, privacy) of the TCP/IP Stack in mesh networks with a new layer3/4 protocol. Some things are running and need more testing. | |
OpenSSI is a comprehensive clustering solution offering a full, highly available single-system image (SSI) environment for Linux. Development involves porting to newer kernels, regression testing, and fixing bugs in the clustering file system, process management, networking, IPC, shared memory, and init as well as userspace system utilities in a distributed computing environment. | |
There are more than 1000 open bugreports in kernel bugzilla. The bugs to fix are of difficulty level 1-10 and there`s a lot to learn there, because nearly every part of the kernel gets touched. Please help identifying and fixing Linux kernel bugs. | |
The Linux kernel evolves at a breath-taking speed. But while adding new features and fixing others, developers often break things and introduce regression. Some call that ....progress. | |
Tux3 is a write-anywhere, atomic commit, btree-based versioning filesystem. |
Please help finding bugs and regression in the kernel. Be part of the QA-team. - "Don't feel bad about being a pest, because we need more pests to keep all of us kernel developers in line" ( 