1 .. _development_process:
3 How the development process works
4 =================================
6 Linux kernel development in the early 1990's was a pretty loose affair,
7 with relatively small numbers of users and developers involved. With a
8 user base in the millions and with some 2,000 developers involved over the
9 course of one year, the kernel has since had to evolve a number of
10 processes to keep development happening smoothly. A solid understanding of
11 how the process works is required in order to be an effective part of it.
16 The kernel developers use a loosely time-based release process, with a new
17 major kernel release happening every two or three months. The recent
18 release history looks like this:
20 ====== =================
23 4.13 September 3, 2017
24 4.14 November 12, 2017
27 ====== =================
29 Every 4.x release is a major kernel release with new features, internal
30 API changes, and more. A typical 4.x release contain about 13,000
31 changesets with changes to several hundred thousand lines of code. 4.x is
32 thus the leading edge of Linux kernel development; the kernel uses a
33 rolling development model which is continually integrating major changes.
35 A relatively straightforward discipline is followed with regard to the
36 merging of patches for each release. At the beginning of each development
37 cycle, the "merge window" is said to be open. At that time, code which is
38 deemed to be sufficiently stable (and which is accepted by the development
39 community) is merged into the mainline kernel. The bulk of changes for a
40 new development cycle (and all of the major changes) will be merged during
41 this time, at a rate approaching 1,000 changes ("patches," or "changesets")
44 (As an aside, it is worth noting that the changes integrated during the
45 merge window do not come out of thin air; they have been collected, tested,
46 and staged ahead of time. How that process works will be described in
49 The merge window lasts for approximately two weeks. At the end of this
50 time, Linus Torvalds will declare that the window is closed and release the
51 first of the "rc" kernels. For the kernel which is destined to be 2.6.40,
52 for example, the release which happens at the end of the merge window will
53 be called 2.6.40-rc1. The -rc1 release is the signal that the time to
54 merge new features has passed, and that the time to stabilize the next
57 Over the next six to ten weeks, only patches which fix problems should be
58 submitted to the mainline. On occasion a more significant change will be
59 allowed, but such occasions are rare; developers who try to merge new
60 features outside of the merge window tend to get an unfriendly reception.
61 As a general rule, if you miss the merge window for a given feature, the
62 best thing to do is to wait for the next development cycle. (An occasional
63 exception is made for drivers for previously-unsupported hardware; if they
64 touch no in-tree code, they cannot cause regressions and should be safe to
67 As fixes make their way into the mainline, the patch rate will slow over
68 time. Linus releases new -rc kernels about once a week; a normal series
69 will get up to somewhere between -rc6 and -rc9 before the kernel is
70 considered to be sufficiently stable and the final 2.6.x release is made.
71 At that point the whole process starts over again.
73 As an example, here is how the 4.16 development cycle went (all dates in
76 ============== ===============================
77 January 28 4.15 stable release
78 February 11 4.16-rc1, merge window closes
85 April 1 4.17 stable release
86 ============== ===============================
88 How do the developers decide when to close the development cycle and create
89 the stable release? The most significant metric used is the list of
90 regressions from previous releases. No bugs are welcome, but those which
91 break systems which worked in the past are considered to be especially
92 serious. For this reason, patches which cause regressions are looked upon
93 unfavorably and are quite likely to be reverted during the stabilization
96 The developers' goal is to fix all known regressions before the stable
97 release is made. In the real world, this kind of perfection is hard to
98 achieve; there are just too many variables in a project of this size.
99 There comes a point where delaying the final release just makes the problem
100 worse; the pile of changes waiting for the next merge window will grow
101 larger, creating even more regressions the next time around. So most 4.x
102 kernels go out with a handful of known regressions though, hopefully, none
105 Once a stable release is made, its ongoing maintenance is passed off to the
106 "stable team," currently consisting of Greg Kroah-Hartman. The stable team
107 will release occasional updates to the stable release using the 4.x.y
108 numbering scheme. To be considered for an update release, a patch must (1)
109 fix a significant bug, and (2) already be merged into the mainline for the
110 next development kernel. Kernels will typically receive stable updates for
111 a little more than one development cycle past their initial release. So,
112 for example, the 4.13 kernel's history looked like:
114 ============== ===============================
115 September 3 4.13 stable release
123 ============== ===============================
125 4.13.16 was the final stable update of the 4.13 release.
127 Some kernels are designated "long term" kernels; they will receive support
128 for a longer period. As of this writing, the current long term kernels
129 and their maintainers are:
131 ====== ====================== ==============================
132 3.16 Ben Hutchings (very long-term stable kernel)
134 4.4 Greg Kroah-Hartman (very long-term stable kernel)
135 4.9 Greg Kroah-Hartman
136 4.14 Greg Kroah-Hartman
137 ====== ====================== ==============================
139 The selection of a kernel for long-term support is purely a matter of a
140 maintainer having the need and the time to maintain that release. There
141 are no known plans for long-term support for any specific upcoming
145 The lifecycle of a patch
146 ------------------------
148 Patches do not go directly from the developer's keyboard into the mainline
149 kernel. There is, instead, a somewhat involved (if somewhat informal)
150 process designed to ensure that each patch is reviewed for quality and that
151 each patch implements a change which is desirable to have in the mainline.
152 This process can happen quickly for minor fixes, or, in the case of large
153 and controversial changes, go on for years. Much developer frustration
154 comes from a lack of understanding of this process or from attempts to
157 In the hopes of reducing that frustration, this document will describe how
158 a patch gets into the kernel. What follows below is an introduction which
159 describes the process in a somewhat idealized way. A much more detailed
160 treatment will come in later sections.
162 The stages that a patch goes through are, generally:
164 - Design. This is where the real requirements for the patch - and the way
165 those requirements will be met - are laid out. Design work is often
166 done without involving the community, but it is better to do this work
167 in the open if at all possible; it can save a lot of time redesigning
170 - Early review. Patches are posted to the relevant mailing list, and
171 developers on that list reply with any comments they may have. This
172 process should turn up any major problems with a patch if all goes
175 - Wider review. When the patch is getting close to ready for mainline
176 inclusion, it should be accepted by a relevant subsystem maintainer -
177 though this acceptance is not a guarantee that the patch will make it
178 all the way to the mainline. The patch will show up in the maintainer's
179 subsystem tree and into the -next trees (described below). When the
180 process works, this step leads to more extensive review of the patch and
181 the discovery of any problems resulting from the integration of this
182 patch with work being done by others.
184 - Please note that most maintainers also have day jobs, so merging
185 your patch may not be their highest priority. If your patch is
186 getting feedback about changes that are needed, you should either
187 make those changes or justify why they should not be made. If your
188 patch has no review complaints but is not being merged by its
189 appropriate subsystem or driver maintainer, you should be persistent
190 in updating the patch to the current kernel so that it applies cleanly
191 and keep sending it for review and merging.
193 - Merging into the mainline. Eventually, a successful patch will be
194 merged into the mainline repository managed by Linus Torvalds. More
195 comments and/or problems may surface at this time; it is important that
196 the developer be responsive to these and fix any issues which arise.
198 - Stable release. The number of users potentially affected by the patch
199 is now large, so, once again, new problems may arise.
201 - Long-term maintenance. While it is certainly possible for a developer
202 to forget about code after merging it, that sort of behavior tends to
203 leave a poor impression in the development community. Merging code
204 eliminates some of the maintenance burden, in that others will fix
205 problems caused by API changes. But the original developer should
206 continue to take responsibility for the code if it is to remain useful
209 One of the largest mistakes made by kernel developers (or their employers)
210 is to try to cut the process down to a single "merging into the mainline"
211 step. This approach invariably leads to frustration for everybody
214 How patches get into the Kernel
215 -------------------------------
217 There is exactly one person who can merge patches into the mainline kernel
218 repository: Linus Torvalds. But, of the over 9,500 patches which went
219 into the 2.6.38 kernel, only 112 (around 1.3%) were directly chosen by Linus
220 himself. The kernel project has long since grown to a size where no single
221 developer could possibly inspect and select every patch unassisted. The
222 way the kernel developers have addressed this growth is through the use of
223 a lieutenant system built around a chain of trust.
225 The kernel code base is logically broken down into a set of subsystems:
226 networking, specific architecture support, memory management, video
227 devices, etc. Most subsystems have a designated maintainer, a developer
228 who has overall responsibility for the code within that subsystem. These
229 subsystem maintainers are the gatekeepers (in a loose way) for the portion
230 of the kernel they manage; they are the ones who will (usually) accept a
231 patch for inclusion into the mainline kernel.
233 Subsystem maintainers each manage their own version of the kernel source
234 tree, usually (but certainly not always) using the git source management
235 tool. Tools like git (and related tools like quilt or mercurial) allow
236 maintainers to track a list of patches, including authorship information
237 and other metadata. At any given time, the maintainer can identify which
238 patches in his or her repository are not found in the mainline.
240 When the merge window opens, top-level maintainers will ask Linus to "pull"
241 the patches they have selected for merging from their repositories. If
242 Linus agrees, the stream of patches will flow up into his repository,
243 becoming part of the mainline kernel. The amount of attention that Linus
244 pays to specific patches received in a pull operation varies. It is clear
245 that, sometimes, he looks quite closely. But, as a general rule, Linus
246 trusts the subsystem maintainers to not send bad patches upstream.
248 Subsystem maintainers, in turn, can pull patches from other maintainers.
249 For example, the networking tree is built from patches which accumulated
250 first in trees dedicated to network device drivers, wireless networking,
251 etc. This chain of repositories can be arbitrarily long, though it rarely
252 exceeds two or three links. Since each maintainer in the chain trusts
253 those managing lower-level trees, this process is known as the "chain of
256 Clearly, in a system like this, getting patches into the kernel depends on
257 finding the right maintainer. Sending patches directly to Linus is not
258 normally the right way to go.
264 The chain of subsystem trees guides the flow of patches into the kernel,
265 but it also raises an interesting question: what if somebody wants to look
266 at all of the patches which are being prepared for the next merge window?
267 Developers will be interested in what other changes are pending to see
268 whether there are any conflicts to worry about; a patch which changes a
269 core kernel function prototype, for example, will conflict with any other
270 patches which use the older form of that function. Reviewers and testers
271 want access to the changes in their integrated form before all of those
272 changes land in the mainline kernel. One could pull changes from all of
273 the interesting subsystem trees, but that would be a big and error-prone
276 The answer comes in the form of -next trees, where subsystem trees are
277 collected for testing and review. The older of these trees, maintained by
278 Andrew Morton, is called "-mm" (for memory management, which is how it got
279 started). The -mm tree integrates patches from a long list of subsystem
280 trees; it also has some patches aimed at helping with debugging.
282 Beyond that, -mm contains a significant collection of patches which have
283 been selected by Andrew directly. These patches may have been posted on a
284 mailing list, or they may apply to a part of the kernel for which there is
285 no designated subsystem tree. As a result, -mm operates as a sort of
286 subsystem tree of last resort; if there is no other obvious path for a
287 patch into the mainline, it is likely to end up in -mm. Miscellaneous
288 patches which accumulate in -mm will eventually either be forwarded on to
289 an appropriate subsystem tree or be sent directly to Linus. In a typical
290 development cycle, approximately 5-10% of the patches going into the
291 mainline get there via -mm.
293 The current -mm patch is available in the "mmotm" (-mm of the moment)
296 http://www.ozlabs.org/~akpm/mmotm/
298 Use of the MMOTM tree is likely to be a frustrating experience, though;
299 there is a definite chance that it will not even compile.
301 The primary tree for next-cycle patch merging is linux-next, maintained by
302 Stephen Rothwell. The linux-next tree is, by design, a snapshot of what
303 the mainline is expected to look like after the next merge window closes.
304 Linux-next trees are announced on the linux-kernel and linux-next mailing
305 lists when they are assembled; they can be downloaded from:
307 http://www.kernel.org/pub/linux/kernel/next/
309 Linux-next has become an integral part of the kernel development process;
310 all patches merged during a given merge window should really have found
311 their way into linux-next some time before the merge window opens.
317 The kernel source tree contains the drivers/staging/ directory, where
318 many sub-directories for drivers or filesystems that are on their way to
319 being added to the kernel tree live. They remain in drivers/staging while
320 they still need more work; once complete, they can be moved into the
321 kernel proper. This is a way to keep track of drivers that aren't
322 up to Linux kernel coding or quality standards, but people may want to use
323 them and track development.
325 Greg Kroah-Hartman currently maintains the staging tree. Drivers that
326 still need work are sent to him, with each driver having its own
327 subdirectory in drivers/staging/. Along with the driver source files, a
328 TODO file should be present in the directory as well. The TODO file lists
329 the pending work that the driver needs for acceptance into the kernel
330 proper, as well as a list of people that should be Cc'd for any patches to
331 the driver. Current rules require that drivers contributed to staging
332 must, at a minimum, compile properly.
334 Staging can be a relatively easy way to get new drivers into the mainline
335 where, with luck, they will come to the attention of other developers and
336 improve quickly. Entry into staging is not the end of the story, though;
337 code in staging which is not seeing regular progress will eventually be
338 removed. Distributors also tend to be relatively reluctant to enable
339 staging drivers. So staging is, at best, a stop on the way toward becoming
340 a proper mainline driver.
346 As can be seen from the above text, the kernel development process depends
347 heavily on the ability to herd collections of patches in various
348 directions. The whole thing would not work anywhere near as well as it
349 does without suitably powerful tools. Tutorials on how to use these tools
350 are well beyond the scope of this document, but there is space for a few
353 By far the dominant source code management system used by the kernel
354 community is git. Git is one of a number of distributed version control
355 systems being developed in the free software community. It is well tuned
356 for kernel development, in that it performs quite well when dealing with
357 large repositories and large numbers of patches. It also has a reputation
358 for being difficult to learn and use, though it has gotten better over
359 time. Some sort of familiarity with git is almost a requirement for kernel
360 developers; even if they do not use it for their own work, they'll need git
361 to keep up with what other developers (and the mainline) are doing.
363 Git is now packaged by almost all Linux distributions. There is a home
368 That page has pointers to documentation and tutorials.
370 Among the kernel developers who do not use git, the most popular choice is
371 almost certainly Mercurial:
373 http://www.selenic.com/mercurial/
375 Mercurial shares many features with git, but it provides an interface which
376 many find easier to use.
378 The other tool worth knowing about is Quilt:
380 http://savannah.nongnu.org/projects/quilt/
382 Quilt is a patch management system, rather than a source code management
383 system. It does not track history over time; it is, instead, oriented
384 toward tracking a specific set of changes against an evolving code base.
385 Some major subsystem maintainers use quilt to manage patches intended to go
386 upstream. For the management of certain kinds of trees (-mm, for example),
387 quilt is the best tool for the job.
393 A great deal of Linux kernel development work is done by way of mailing
394 lists. It is hard to be a fully-functioning member of the community
395 without joining at least one list somewhere. But Linux mailing lists also
396 represent a potential hazard to developers, who risk getting buried under a
397 load of electronic mail, running afoul of the conventions used on the Linux
400 Most kernel mailing lists are run on vger.kernel.org; the master list can
403 http://vger.kernel.org/vger-lists.html
405 There are lists hosted elsewhere, though; a number of them are at
408 The core mailing list for kernel development is, of course, linux-kernel.
409 This list is an intimidating place to be; volume can reach 500 messages per
410 day, the amount of noise is high, the conversation can be severely
411 technical, and participants are not always concerned with showing a high
412 degree of politeness. But there is no other place where the kernel
413 development community comes together as a whole; developers who avoid this
414 list will miss important information.
416 There are a few hints which can help with linux-kernel survival:
418 - Have the list delivered to a separate folder, rather than your main
419 mailbox. One must be able to ignore the stream for sustained periods of
422 - Do not try to follow every conversation - nobody else does. It is
423 important to filter on both the topic of interest (though note that
424 long-running conversations can drift away from the original subject
425 without changing the email subject line) and the people who are
428 - Do not feed the trolls. If somebody is trying to stir up an angry
429 response, ignore them.
431 - When responding to linux-kernel email (or that on other lists) preserve
432 the Cc: header for all involved. In the absence of a strong reason (such
433 as an explicit request), you should never remove recipients. Always make
434 sure that the person you are responding to is in the Cc: list. This
435 convention also makes it unnecessary to explicitly ask to be copied on
436 replies to your postings.
438 - Search the list archives (and the net as a whole) before asking
439 questions. Some developers can get impatient with people who clearly
440 have not done their homework.
442 - Avoid top-posting (the practice of putting your answer above the quoted
443 text you are responding to). It makes your response harder to read and
444 makes a poor impression.
446 - Ask on the correct mailing list. Linux-kernel may be the general meeting
447 point, but it is not the best place to find developers from all
450 The last point - finding the correct mailing list - is a common place for
451 beginning developers to go wrong. Somebody who asks a networking-related
452 question on linux-kernel will almost certainly receive a polite suggestion
453 to ask on the netdev list instead, as that is the list frequented by most
454 networking developers. Other lists exist for the SCSI, video4linux, IDE,
455 filesystem, etc. subsystems. The best place to look for mailing lists is
456 in the MAINTAINERS file packaged with the kernel source.
459 Getting started with Kernel development
460 ---------------------------------------
462 Questions about how to get started with the kernel development process are
463 common - from both individuals and companies. Equally common are missteps
464 which make the beginning of the relationship harder than it has to be.
466 Companies often look to hire well-known developers to get a development
467 group started. This can, in fact, be an effective technique. But it also
468 tends to be expensive and does not do much to grow the pool of experienced
469 kernel developers. It is possible to bring in-house developers up to speed
470 on Linux kernel development, given the investment of a bit of time. Taking
471 this time can endow an employer with a group of developers who understand
472 the kernel and the company both, and who can help to train others as well.
473 Over the medium term, this is often the more profitable approach.
475 Individual developers are often, understandably, at a loss for a place to
476 start. Beginning with a large project can be intimidating; one often wants
477 to test the waters with something smaller first. This is the point where
478 some developers jump into the creation of patches fixing spelling errors or
479 minor coding style issues. Unfortunately, such patches create a level of
480 noise which is distracting for the development community as a whole, so,
481 increasingly, they are looked down upon. New developers wishing to
482 introduce themselves to the community will not get the sort of reception
483 they wish for by these means.
485 Andrew Morton gives this advice for aspiring kernel developers
489 The #1 project for all kernel beginners should surely be "make sure
490 that the kernel runs perfectly at all times on all machines which
491 you can lay your hands on". Usually the way to do this is to work
492 with others on getting things fixed up (this can require
493 persistence!) but that's fine - it's a part of kernel development.
495 (http://lwn.net/Articles/283982/).
497 In the absence of obvious problems to fix, developers are advised to look
498 at the current lists of regressions and open bugs in general. There is
499 never any shortage of issues in need of fixing; by addressing these issues,
500 developers will gain experience with the process while, at the same time,
501 building respect with the rest of the development community.
git.samba.org / sfrench / cifs-2.6.git / blob