3Linux kernel development in the early 1990's was a pretty loose affair,
4with relatively small numbers of users and developers involved. With a
5user base in the millions and with some 2,000 developers involved over the
6course of one year, the kernel has since had to evolve a number of
7processes to keep development happening smoothly. A solid understanding of
8how the process works is required in order to be an effective part of it.
13The kernel developers use a loosely time-based release process, with a new
14major kernel release happening every two or three months. The recent
15release history looks like this:
17    2.6.26 July 13, 2008
18    2.6.25 April 16, 2008
19    2.6.24 January 24, 2008
20    2.6.23 October 9, 2007
21    2.6.22 July 8, 2007
22    2.6.21 April 25, 2007
23    2.6.20 February 4, 2007
25Every 2.6.x release is a major kernel release with new features, internal
26API changes, and more. A typical 2.6 release can contain over 10,000
27changesets with changes to several hundred thousand lines of code. 2.6 is
28thus the leading edge of Linux kernel development; the kernel uses a
29rolling development model which is continually integrating major changes.
31A relatively straightforward discipline is followed with regard to the
32merging of patches for each release. At the beginning of each development
33cycle, the "merge window" is said to be open. At that time, code which is
34deemed to be sufficiently stable (and which is accepted by the development
35community) is merged into the mainline kernel. The bulk of changes for a
36new development cycle (and all of the major changes) will be merged during
37this time, at a rate approaching 1,000 changes ("patches," or "changesets")
38per day.
40(As an aside, it is worth noting that the changes integrated during the
41merge window do not come out of thin air; they have been collected, tested,
42and staged ahead of time. How that process works will be described in
43detail later on).
45The merge window lasts for two weeks. At the end of this time, Linus
46Torvalds will declare that the window is closed and release the first of
47the "rc" kernels. For the kernel which is destined to be 2.6.26, for
48example, the release which happens at the end of the merge window will be
49called 2.6.26-rc1. The -rc1 release is the signal that the time to merge
50new features has passed, and that the time to stabilize the next kernel has
53Over the next six to ten weeks, only patches which fix problems should be
54submitted to the mainline. On occasion a more significant change will be
55allowed, but such occasions are rare; developers who try to merge new
56features outside of the merge window tend to get an unfriendly reception.
57As a general rule, if you miss the merge window for a given feature, the
58best thing to do is to wait for the next development cycle. (An occasional
59exception is made for drivers for previously-unsupported hardware; if they
60touch no in-tree code, they cannot cause regressions and should be safe to
61add at any time).
63As fixes make their way into the mainline, the patch rate will slow over
64time. Linus releases new -rc kernels about once a week; a normal series
65will get up to somewhere between -rc6 and -rc9 before the kernel is
66considered to be sufficiently stable and the final 2.6.x release is made.
67At that point the whole process starts over again.
69As an example, here is how the 2.6.25 development cycle went (all dates in
72    January 24 2.6.24 stable release
73    February 10 2.6.25-rc1, merge window closes
74    February 15 2.6.25-rc2
75    February 24 2.6.25-rc3
76    March 4 2.6.25-rc4
77    March 9 2.6.25-rc5
78    March 16 2.6.25-rc6
79    March 25 2.6.25-rc7
80    April 1 2.6.25-rc8
81    April 11 2.6.25-rc9
82    April 16 2.6.25 stable release
84How do the developers decide when to close the development cycle and create
85the stable release? The most significant metric used is the list of
86regressions from previous releases. No bugs are welcome, but those which
87break systems which worked in the past are considered to be especially
88serious. For this reason, patches which cause regressions are looked upon
89unfavorably and are quite likely to be reverted during the stabilization
92The developers' goal is to fix all known regressions before the stable
93release is made. In the real world, this kind of perfection is hard to
94achieve; there are just too many variables in a project of this size.
95There comes a point where delaying the final release just makes the problem
96worse; the pile of changes waiting for the next merge window will grow
97larger, creating even more regressions the next time around. So most 2.6.x
98kernels go out with a handful of known regressions though, hopefully, none
99of them are serious.
101Once a stable release is made, its ongoing maintenance is passed off to the
102"stable team," currently comprised of Greg Kroah-Hartman and Chris Wright.
103The stable team will release occasional updates to the stable release using
104the 2.6.x.y numbering scheme. To be considered for an update release, a
105patch must (1) fix a significant bug, and (2) already be merged into the
106mainline for the next development kernel. Continuing our 2.6.25 example,
107the history (as of this writing) is:
109    May 1
110    May 6
111    May 9
112    May 15
113    June 7
114    June 9
115    June 16
116    June 21
117    June 24
119Stable updates for a given kernel are made for approximately six months;
120after that, the maintenance of stable releases is solely the responsibility
121of the distributors which have shipped that particular kernel.
126Patches do not go directly from the developer's keyboard into the mainline
127kernel. There is, instead, a somewhat involved (if somewhat informal)
128process designed to ensure that each patch is reviewed for quality and that
129each patch implements a change which is desirable to have in the mainline.
130This process can happen quickly for minor fixes, or, in the case of large
131and controversial changes, go on for years. Much developer frustration
132comes from a lack of understanding of this process or from attempts to
133circumvent it.
135In the hopes of reducing that frustration, this document will describe how
136a patch gets into the kernel. What follows below is an introduction which
137describes the process in a somewhat idealized way. A much more detailed
138treatment will come in later sections.
140The stages that a patch goes through are, generally:
142 - Design. This is where the real requirements for the patch - and the way
143   those requirements will be met - are laid out. Design work is often
144   done without involving the community, but it is better to do this work
145   in the open if at all possible; it can save a lot of time redesigning
146   things later.
148 - Early review. Patches are posted to the relevant mailing list, and
149   developers on that list reply with any comments they may have. This
150   process should turn up any major problems with a patch if all goes
151   well.
153 - Wider review. When the patch is getting close to ready for mainline
154   inclusion, it will be accepted by a relevant subsystem maintainer -
155   though this acceptance is not a guarantee that the patch will make it
156   all the way to the mainline. The patch will show up in the maintainer's
157   subsystem tree and into the staging trees (described below). When the
158   process works, this step leads to more extensive review of the patch and
159   the discovery of any problems resulting from the integration of this
160   patch with work being done by others.
162 - Merging into the mainline. Eventually, a successful patch will be
163   merged into the mainline repository managed by Linus Torvalds. More
164   comments and/or problems may surface at this time; it is important that
165   the developer be responsive to these and fix any issues which arise.
167 - Stable release. The number of users potentially affected by the patch
168   is now large, so, once again, new problems may arise.
170 - Long-term maintenance. While it is certainly possible for a developer
171   to forget about code after merging it, that sort of behavior tends to
172   leave a poor impression in the development community. Merging code
173   eliminates some of the maintenance burden, in that others will fix
174   problems caused by API changes. But the original developer should
175   continue to take responsibility for the code if it is to remain useful
176   in the longer term.
178One of the largest mistakes made by kernel developers (or their employers)
179is to try to cut the process down to a single "merging into the mainline"
180step. This approach invariably leads to frustration for everybody
186There is exactly one person who can merge patches into the mainline kernel
187repository: Linus Torvalds. But, of the over 12,000 patches which went
188into the 2.6.25 kernel, only 250 (around 2%) were directly chosen by Linus
189himself. The kernel project has long since grown to a size where no single
190developer could possibly inspect and select every patch unassisted. The
191way the kernel developers have addressed this growth is through the use of
192a lieutenant system built around a chain of trust.
194The kernel code base is logically broken down into a set of subsystems:
195networking, specific architecture support, memory management, video
196devices, etc. Most subsystems have a designated maintainer, a developer
197who has overall responsibility for the code within that subsystem. These
198subsystem maintainers are the gatekeepers (in a loose way) for the portion
199of the kernel they manage; they are the ones who will (usually) accept a
200patch for inclusion into the mainline kernel.
202Subsystem maintainers each manage their own version of the kernel source
203tree, usually (but certainly not always) using the git source management
204tool. Tools like git (and related tools like quilt or mercurial) allow
205maintainers to track a list of patches, including authorship information
206and other metadata. At any given time, the maintainer can identify which
207patches in his or her repository are not found in the mainline.
209When the merge window opens, top-level maintainers will ask Linus to "pull"
210the patches they have selected for merging from their repositories. If
211Linus agrees, the stream of patches will flow up into his repository,
212becoming part of the mainline kernel. The amount of attention that Linus
213pays to specific patches received in a pull operation varies. It is clear
214that, sometimes, he looks quite closely. But, as a general rule, Linus
215trusts the subsystem maintainers to not send bad patches upstream.
217Subsystem maintainers, in turn, can pull patches from other maintainers.
218For example, the networking tree is built from patches which accumulated
219first in trees dedicated to network device drivers, wireless networking,
220etc. This chain of repositories can be arbitrarily long, though it rarely
221exceeds two or three links. Since each maintainer in the chain trusts
222those managing lower-level trees, this process is known as the "chain of
225Clearly, in a system like this, getting patches into the kernel depends on
226finding the right maintainer. Sending patches directly to Linus is not
227normally the right way to go.
232The chain of subsystem trees guides the flow of patches into the kernel,
233but it also raises an interesting question: what if somebody wants to look
234at all of the patches which are being prepared for the next merge window?
235Developers will be interested in what other changes are pending to see
236whether there are any conflicts to worry about; a patch which changes a
237core kernel function prototype, for example, will conflict with any other
238patches which use the older form of that function. Reviewers and testers
239want access to the changes in their integrated form before all of those
240changes land in the mainline kernel. One could pull changes from all of
241the interesting subsystem trees, but that would be a big and error-prone
244The answer comes in the form of staging trees, where subsystem trees are
245collected for testing and review. The older of these trees, maintained by
246Andrew Morton, is called "-mm" (for memory management, which is how it got
247started). The -mm tree integrates patches from a long list of subsystem
248trees; it also has some patches aimed at helping with debugging.
250Beyond that, -mm contains a significant collection of patches which have
251been selected by Andrew directly. These patches may have been posted on a
252mailing list, or they may apply to a part of the kernel for which there is
253no designated subsystem tree. As a result, -mm operates as a sort of
254subsystem tree of last resort; if there is no other obvious path for a
255patch into the mainline, it is likely to end up in -mm. Miscellaneous
256patches which accumulate in -mm will eventually either be forwarded on to
257an appropriate subsystem tree or be sent directly to Linus. In a typical
258development cycle, approximately 10% of the patches going into the mainline
259get there via -mm.
261The current -mm patch can always be found from the front page of
265Those who want to see the current state of -mm can get the "-mm of the
266moment" tree, found at:
270Use of the MMOTM tree is likely to be a frustrating experience, though;
271there is a definite chance that it will not even compile.
273The other staging tree, started more recently, is linux-next, maintained by
274Stephen Rothwell. The linux-next tree is, by design, a snapshot of what
275the mainline is expected to look like after the next merge window closes.
276Linux-next trees are announced on the linux-kernel and linux-next mailing
277lists when they are assembled; they can be downloaded from:
281Some information about linux-next has been gathered at:
285How the linux-next tree will fit into the development process is still
286changing. As of this writing, the first full development cycle involving
287linux-next (2.6.26) is coming to an end; thus far, it has proved to be a
288valuable resource for finding and fixing integration problems before the
289beginning of the merge window. See for
290more information on how linux-next has worked to set up the 2.6.27 merge
293Some developers have begun to suggest that linux-next should be used as the
294target for future development as well. The linux-next tree does tend to be
295far ahead of the mainline and is more representative of the tree into which
296any new work will be merged. The downside to this idea is that the
297volatility of linux-next tends to make it a difficult development target.
298See for more information on this topic, and
299stay tuned; much is still in flux where linux-next is involved.
3022.5: TOOLS
304As can be seen from the above text, the kernel development process depends
305heavily on the ability to herd collections of patches in various
306directions. The whole thing would not work anywhere near as well as it
307does without suitably powerful tools. Tutorials on how to use these tools
308are well beyond the scope of this document, but there is space for a few
311By far the dominant source code management system used by the kernel
312community is git. Git is one of a number of distributed version control
313systems being developed in the free software community. It is well tuned
314for kernel development, in that it performs quite well when dealing with
315large repositories and large numbers of patches. It also has a reputation
316for being difficult to learn and use, though it has gotten better over
317time. Some sort of familiarity with git is almost a requirement for kernel
318developers; even if they do not use it for their own work, they'll need git
319to keep up with what other developers (and the mainline) are doing.
321Git is now packaged by almost all Linux distributions. There is a home
322page at
326That page has pointers to documentation and tutorials. One should be
327aware, in particular, of the Kernel Hacker's Guide to git, which has
328information specific to kernel development:
332Among the kernel developers who do not use git, the most popular choice is
333almost certainly Mercurial:
337Mercurial shares many features with git, but it provides an interface which
338many find easier to use.
340The other tool worth knowing about is Quilt:
344Quilt is a patch management system, rather than a source code management
345system. It does not track history over time; it is, instead, oriented
346toward tracking a specific set of changes against an evolving code base.
347Some major subsystem maintainers use quilt to manage patches intended to go
348upstream. For the management of certain kinds of trees (-mm, for example),
349quilt is the best tool for the job.
354A great deal of Linux kernel development work is done by way of mailing
355lists. It is hard to be a fully-functioning member of the community
356without joining at least one list somewhere. But Linux mailing lists also
357represent a potential hazard to developers, who risk getting buried under a
358load of electronic mail, running afoul of the conventions used on the Linux
359lists, or both.
361Most kernel mailing lists are run on; the master list can
362be found at:
366There are lists hosted elsewhere, though; a number of them are at
369The core mailing list for kernel development is, of course, linux-kernel.
370This list is an intimidating place to be; volume can reach 500 messages per
371day, the amount of noise is high, the conversation can be severely
372technical, and participants are not always concerned with showing a high
373degree of politeness. But there is no other place where the kernel
374development community comes together as a whole; developers who avoid this
375list will miss important information.
377There are a few hints which can help with linux-kernel survival:
379- Have the list delivered to a separate folder, rather than your main
380  mailbox. One must be able to ignore the stream for sustained periods of
381  time.
383- Do not try to follow every conversation - nobody else does. It is
384  important to filter on both the topic of interest (though note that
385  long-running conversations can drift away from the original subject
386  without changing the email subject line) and the people who are
387  participating.
389- Do not feed the trolls. If somebody is trying to stir up an angry
390  response, ignore them.
392- When responding to linux-kernel email (or that on other lists) preserve
393  the Cc: header for all involved. In the absence of a strong reason (such
394  as an explicit request), you should never remove recipients. Always make
395  sure that the person you are responding to is in the Cc: list. This
396  convention also makes it unnecessary to explicitly ask to be copied on
397  replies to your postings.
399- Search the list archives (and the net as a whole) before asking
400  questions. Some developers can get impatient with people who clearly
401  have not done their homework.
403- Avoid top-posting (the practice of putting your answer above the quoted
404  text you are responding to). It makes your response harder to read and
405  makes a poor impression.
407- Ask on the correct mailing list. Linux-kernel may be the general meeting
408  point, but it is not the best place to find developers from all
409  subsystems.
411The last point - finding the correct mailing list - is a common place for
412beginning developers to go wrong. Somebody who asks a networking-related
413question on linux-kernel will almost certainly receive a polite suggestion
414to ask on the netdev list instead, as that is the list frequented by most
415networking developers. Other lists exist for the SCSI, video4linux, IDE,
416filesystem, etc. subsystems. The best place to look for mailing lists is
417in the MAINTAINERS file packaged with the kernel source.
422Questions about how to get started with the kernel development process are
423common - from both individuals and companies. Equally common are missteps
424which make the beginning of the relationship harder than it has to be.
426Companies often look to hire well-known developers to get a development
427group started. This can, in fact, be an effective technique. But it also
428tends to be expensive and does not do much to grow the pool of experienced
429kernel developers. It is possible to bring in-house developers up to speed
430on Linux kernel development, given the investment of a bit of time. Taking
431this time can endow an employer with a group of developers who understand
432the kernel and the company both, and who can help to train others as well.
433Over the medium term, this is often the more profitable approach.
435Individual developers are often, understandably, at a loss for a place to
436start. Beginning with a large project can be intimidating; one often wants
437to test the waters with something smaller first. This is the point where
438some developers jump into the creation of patches fixing spelling errors or
439minor coding style issues. Unfortunately, such patches create a level of
440noise which is distracting for the development community as a whole, so,
441increasingly, they are looked down upon. New developers wishing to
442introduce themselves to the community will not get the sort of reception
443they wish for by these means.
445Andrew Morton gives this advice for aspiring kernel developers
447    The #1 project for all kernel beginners should surely be "make sure
448    that the kernel runs perfectly at all times on all machines which
449    you can lay your hands on". Usually the way to do this is to work
450    with others on getting things fixed up (this can require
451    persistence!) but that's fine - it's a part of kernel development.
455In the absence of obvious problems to fix, developers are advised to look
456at the current lists of regressions and open bugs in general. There is
457never any shortage of issues in need of fixing; by addressing these issues,
458developers will gain experience with the process while, at the same time,
459building respect with the rest of the development community.

Archive Download this file