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 ====== =================
24 5.3 September 15, 2019
27 ====== =================
29 Every 5.x release is a major kernel release with new features, internal
30 API changes, and more. A typical release can contain about 13,000
31 changesets with changes to several hundred thousand lines of code. 5.x is
32 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 5.6,
52 for example, the release which happens at the end of the merge window will
53 be called 5.6-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 release is made.
71 At that point the whole process starts over again.
73 As an example, here is how the 5.4 development cycle went (all dates in
76 ============== ===============================
77 September 15 5.3 stable release
78 September 30 5.4-rc1, merge window closes
86 November 24 5.4 stable release
87 ============== ===============================
89 How do the developers decide when to close the development cycle and create
90 the stable release? The most significant metric used is the list of
91 regressions from previous releases. No bugs are welcome, but those which
92 break systems which worked in the past are considered to be especially
93 serious. For this reason, patches which cause regressions are looked upon
94 unfavorably and are quite likely to be reverted during the stabilization
97 The developers' goal is to fix all known regressions before the stable
98 release is made. In the real world, this kind of perfection is hard to
99 achieve; there are just too many variables in a project of this size.
100 There comes a point where delaying the final release just makes the problem
101 worse; the pile of changes waiting for the next merge window will grow
102 larger, creating even more regressions the next time around. So most 5.x
103 kernels go out with a handful of known regressions though, hopefully, none
106 Once a stable release is made, its ongoing maintenance is passed off to the
107 "stable team," currently Greg Kroah-Hartman. The stable team will release
108 occasional updates to the stable release using the 5.x.y numbering scheme.
109 To be considered for an update release, a patch must (1) fix a significant
110 bug, and (2) already be merged into the mainline for the next development
111 kernel. Kernels will typically receive stable updates for a little more
112 than one development cycle past their initial release. So, for example, the
113 5.2 kernel's history looked like this (all dates in 2019):
115 ============== ===============================
116 July 7 5.2 stable release
124 ============== ===============================
126 5.2.21 was the final stable update of the 5.2 release.
128 Some kernels are designated "long term" kernels; they will receive support
129 for a longer period. As of this writing, the current long term kernels
130 and their maintainers are:
132 ====== ================================ =======================
133 3.16 Ben Hutchings (very long-term kernel)
134 4.4 Greg Kroah-Hartman & Sasha Levin (very long-term kernel)
135 4.9 Greg Kroah-Hartman & Sasha Levin
136 4.14 Greg Kroah-Hartman & Sasha Levin
137 4.19 Greg Kroah-Hartman & Sasha Levin
138 5.4 Greg Kroah-Hartman & Sasha Levin
139 ====== ================================ =======================
141 The selection of a kernel for long-term support is purely a matter of a
142 maintainer having the need and the time to maintain that release. There
143 are no known plans for long-term support for any specific upcoming
147 The lifecycle of a patch
148 ------------------------
150 Patches do not go directly from the developer's keyboard into the mainline
151 kernel. There is, instead, a somewhat involved (if somewhat informal)
152 process designed to ensure that each patch is reviewed for quality and that
153 each patch implements a change which is desirable to have in the mainline.
154 This process can happen quickly for minor fixes, or, in the case of large
155 and controversial changes, go on for years. Much developer frustration
156 comes from a lack of understanding of this process or from attempts to
159 In the hopes of reducing that frustration, this document will describe how
160 a patch gets into the kernel. What follows below is an introduction which
161 describes the process in a somewhat idealized way. A much more detailed
162 treatment will come in later sections.
164 The stages that a patch goes through are, generally:
166 - Design. This is where the real requirements for the patch - and the way
167 those requirements will be met - are laid out. Design work is often
168 done without involving the community, but it is better to do this work
169 in the open if at all possible; it can save a lot of time redesigning
172 - Early review. Patches are posted to the relevant mailing list, and
173 developers on that list reply with any comments they may have. This
174 process should turn up any major problems with a patch if all goes
177 - Wider review. When the patch is getting close to ready for mainline
178 inclusion, it should be accepted by a relevant subsystem maintainer -
179 though this acceptance is not a guarantee that the patch will make it
180 all the way to the mainline. The patch will show up in the maintainer's
181 subsystem tree and into the -next trees (described below). When the
182 process works, this step leads to more extensive review of the patch and
183 the discovery of any problems resulting from the integration of this
184 patch with work being done by others.
186 - Please note that most maintainers also have day jobs, so merging
187 your patch may not be their highest priority. If your patch is
188 getting feedback about changes that are needed, you should either
189 make those changes or justify why they should not be made. If your
190 patch has no review complaints but is not being merged by its
191 appropriate subsystem or driver maintainer, you should be persistent
192 in updating the patch to the current kernel so that it applies cleanly
193 and keep sending it for review and merging.
195 - Merging into the mainline. Eventually, a successful patch will be
196 merged into the mainline repository managed by Linus Torvalds. More
197 comments and/or problems may surface at this time; it is important that
198 the developer be responsive to these and fix any issues which arise.
200 - Stable release. The number of users potentially affected by the patch
201 is now large, so, once again, new problems may arise.
203 - Long-term maintenance. While it is certainly possible for a developer
204 to forget about code after merging it, that sort of behavior tends to
205 leave a poor impression in the development community. Merging code
206 eliminates some of the maintenance burden, in that others will fix
207 problems caused by API changes. But the original developer should
208 continue to take responsibility for the code if it is to remain useful
211 One of the largest mistakes made by kernel developers (or their employers)
212 is to try to cut the process down to a single "merging into the mainline"
213 step. This approach invariably leads to frustration for everybody
216 How patches get into the Kernel
217 -------------------------------
219 There is exactly one person who can merge patches into the mainline kernel
220 repository: Linus Torvalds. But, for example, of the over 9,500 patches
221 which went into the 2.6.38 kernel, only 112 (around 1.3%) were directly
222 chosen by Linus himself. The kernel project has long since grown to a size
223 where no single developer could possibly inspect and select every patch
224 unassisted. The way the kernel developers have addressed this growth is
225 through the use of a lieutenant system built around a chain of trust.
227 The kernel code base is logically broken down into a set of subsystems:
228 networking, specific architecture support, memory management, video
229 devices, etc. Most subsystems have a designated maintainer, a developer
230 who has overall responsibility for the code within that subsystem. These
231 subsystem maintainers are the gatekeepers (in a loose way) for the portion
232 of the kernel they manage; they are the ones who will (usually) accept a
233 patch for inclusion into the mainline kernel.
235 Subsystem maintainers each manage their own version of the kernel source
236 tree, usually (but certainly not always) using the git source management
237 tool. Tools like git (and related tools like quilt or mercurial) allow
238 maintainers to track a list of patches, including authorship information
239 and other metadata. At any given time, the maintainer can identify which
240 patches in his or her repository are not found in the mainline.
242 When the merge window opens, top-level maintainers will ask Linus to "pull"
243 the patches they have selected for merging from their repositories. If
244 Linus agrees, the stream of patches will flow up into his repository,
245 becoming part of the mainline kernel. The amount of attention that Linus
246 pays to specific patches received in a pull operation varies. It is clear
247 that, sometimes, he looks quite closely. But, as a general rule, Linus
248 trusts the subsystem maintainers to not send bad patches upstream.
250 Subsystem maintainers, in turn, can pull patches from other maintainers.
251 For example, the networking tree is built from patches which accumulated
252 first in trees dedicated to network device drivers, wireless networking,
253 etc. This chain of repositories can be arbitrarily long, though it rarely
254 exceeds two or three links. Since each maintainer in the chain trusts
255 those managing lower-level trees, this process is known as the "chain of
258 Clearly, in a system like this, getting patches into the kernel depends on
259 finding the right maintainer. Sending patches directly to Linus is not
260 normally the right way to go.
266 The chain of subsystem trees guides the flow of patches into the kernel,
267 but it also raises an interesting question: what if somebody wants to look
268 at all of the patches which are being prepared for the next merge window?
269 Developers will be interested in what other changes are pending to see
270 whether there are any conflicts to worry about; a patch which changes a
271 core kernel function prototype, for example, will conflict with any other
272 patches which use the older form of that function. Reviewers and testers
273 want access to the changes in their integrated form before all of those
274 changes land in the mainline kernel. One could pull changes from all of
275 the interesting subsystem trees, but that would be a big and error-prone
278 The answer comes in the form of -next trees, where subsystem trees are
279 collected for testing and review. The older of these trees, maintained by
280 Andrew Morton, is called "-mm" (for memory management, which is how it got
281 started). The -mm tree integrates patches from a long list of subsystem
282 trees; it also has some patches aimed at helping with debugging.
284 Beyond that, -mm contains a significant collection of patches which have
285 been selected by Andrew directly. These patches may have been posted on a
286 mailing list, or they may apply to a part of the kernel for which there is
287 no designated subsystem tree. As a result, -mm operates as a sort of
288 subsystem tree of last resort; if there is no other obvious path for a
289 patch into the mainline, it is likely to end up in -mm. Miscellaneous
290 patches which accumulate in -mm will eventually either be forwarded on to
291 an appropriate subsystem tree or be sent directly to Linus. In a typical
292 development cycle, approximately 5-10% of the patches going into the
293 mainline get there via -mm.
295 The current -mm patch is available in the "mmotm" (-mm of the moment)
298 https://www.ozlabs.org/~akpm/mmotm/
300 Use of the MMOTM tree is likely to be a frustrating experience, though;
301 there is a definite chance that it will not even compile.
303 The primary tree for next-cycle patch merging is linux-next, maintained by
304 Stephen Rothwell. The linux-next tree is, by design, a snapshot of what
305 the mainline is expected to look like after the next merge window closes.
306 Linux-next trees are announced on the linux-kernel and linux-next mailing
307 lists when they are assembled; they can be downloaded from:
309 https://www.kernel.org/pub/linux/kernel/next/
311 Linux-next has become an integral part of the kernel development process;
312 all patches merged during a given merge window should really have found
313 their way into linux-next some time before the merge window opens.
319 The kernel source tree contains the drivers/staging/ directory, where
320 many sub-directories for drivers or filesystems that are on their way to
321 being added to the kernel tree live. They remain in drivers/staging while
322 they still need more work; once complete, they can be moved into the
323 kernel proper. This is a way to keep track of drivers that aren't
324 up to Linux kernel coding or quality standards, but people may want to use
325 them and track development.
327 Greg Kroah-Hartman currently maintains the staging tree. Drivers that
328 still need work are sent to him, with each driver having its own
329 subdirectory in drivers/staging/. Along with the driver source files, a
330 TODO file should be present in the directory as well. The TODO file lists
331 the pending work that the driver needs for acceptance into the kernel
332 proper, as well as a list of people that should be Cc'd for any patches to
333 the driver. Current rules require that drivers contributed to staging
334 must, at a minimum, compile properly.
336 Staging can be a relatively easy way to get new drivers into the mainline
337 where, with luck, they will come to the attention of other developers and
338 improve quickly. Entry into staging is not the end of the story, though;
339 code in staging which is not seeing regular progress will eventually be
340 removed. Distributors also tend to be relatively reluctant to enable
341 staging drivers. So staging is, at best, a stop on the way toward becoming
342 a proper mainline driver.
348 As can be seen from the above text, the kernel development process depends
349 heavily on the ability to herd collections of patches in various
350 directions. The whole thing would not work anywhere near as well as it
351 does without suitably powerful tools. Tutorials on how to use these tools
352 are well beyond the scope of this document, but there is space for a few
355 By far the dominant source code management system used by the kernel
356 community is git. Git is one of a number of distributed version control
357 systems being developed in the free software community. It is well tuned
358 for kernel development, in that it performs quite well when dealing with
359 large repositories and large numbers of patches. It also has a reputation
360 for being difficult to learn and use, though it has gotten better over
361 time. Some sort of familiarity with git is almost a requirement for kernel
362 developers; even if they do not use it for their own work, they'll need git
363 to keep up with what other developers (and the mainline) are doing.
365 Git is now packaged by almost all Linux distributions. There is a home
370 That page has pointers to documentation and tutorials.
372 Among the kernel developers who do not use git, the most popular choice is
373 almost certainly Mercurial:
375 https://www.selenic.com/mercurial/
377 Mercurial shares many features with git, but it provides an interface which
378 many find easier to use.
380 The other tool worth knowing about is Quilt:
382 https://savannah.nongnu.org/projects/quilt/
384 Quilt is a patch management system, rather than a source code management
385 system. It does not track history over time; it is, instead, oriented
386 toward tracking a specific set of changes against an evolving code base.
387 Some major subsystem maintainers use quilt to manage patches intended to go
388 upstream. For the management of certain kinds of trees (-mm, for example),
389 quilt is the best tool for the job.
395 A great deal of Linux kernel development work is done by way of mailing
396 lists. It is hard to be a fully-functioning member of the community
397 without joining at least one list somewhere. But Linux mailing lists also
398 represent a potential hazard to developers, who risk getting buried under a
399 load of electronic mail, running afoul of the conventions used on the Linux
402 Most kernel mailing lists are run on vger.kernel.org; the master list can
405 http://vger.kernel.org/vger-lists.html
407 There are lists hosted elsewhere, though; a number of them are at
408 redhat.com/mailman/listinfo.
410 The core mailing list for kernel development is, of course, linux-kernel.
411 This list is an intimidating place to be; volume can reach 500 messages per
412 day, the amount of noise is high, the conversation can be severely
413 technical, and participants are not always concerned with showing a high
414 degree of politeness. But there is no other place where the kernel
415 development community comes together as a whole; developers who avoid this
416 list will miss important information.
418 There are a few hints which can help with linux-kernel survival:
420 - Have the list delivered to a separate folder, rather than your main
421 mailbox. One must be able to ignore the stream for sustained periods of
424 - Do not try to follow every conversation - nobody else does. It is
425 important to filter on both the topic of interest (though note that
426 long-running conversations can drift away from the original subject
427 without changing the email subject line) and the people who are
430 - Do not feed the trolls. If somebody is trying to stir up an angry
431 response, ignore them.
433 - When responding to linux-kernel email (or that on other lists) preserve
434 the Cc: header for all involved. In the absence of a strong reason (such
435 as an explicit request), you should never remove recipients. Always make
436 sure that the person you are responding to is in the Cc: list. This
437 convention also makes it unnecessary to explicitly ask to be copied on
438 replies to your postings.
440 - Search the list archives (and the net as a whole) before asking
441 questions. Some developers can get impatient with people who clearly
442 have not done their homework.
444 - Avoid top-posting (the practice of putting your answer above the quoted
445 text you are responding to). It makes your response harder to read and
446 makes a poor impression.
448 - Ask on the correct mailing list. Linux-kernel may be the general meeting
449 point, but it is not the best place to find developers from all
452 The last point - finding the correct mailing list - is a common place for
453 beginning developers to go wrong. Somebody who asks a networking-related
454 question on linux-kernel will almost certainly receive a polite suggestion
455 to ask on the netdev list instead, as that is the list frequented by most
456 networking developers. Other lists exist for the SCSI, video4linux, IDE,
457 filesystem, etc. subsystems. The best place to look for mailing lists is
458 in the MAINTAINERS file packaged with the kernel source.
461 Getting started with Kernel development
462 ---------------------------------------
464 Questions about how to get started with the kernel development process are
465 common - from both individuals and companies. Equally common are missteps
466 which make the beginning of the relationship harder than it has to be.
468 Companies often look to hire well-known developers to get a development
469 group started. This can, in fact, be an effective technique. But it also
470 tends to be expensive and does not do much to grow the pool of experienced
471 kernel developers. It is possible to bring in-house developers up to speed
472 on Linux kernel development, given the investment of a bit of time. Taking
473 this time can endow an employer with a group of developers who understand
474 the kernel and the company both, and who can help to train others as well.
475 Over the medium term, this is often the more profitable approach.
477 Individual developers are often, understandably, at a loss for a place to
478 start. Beginning with a large project can be intimidating; one often wants
479 to test the waters with something smaller first. This is the point where
480 some developers jump into the creation of patches fixing spelling errors or
481 minor coding style issues. Unfortunately, such patches create a level of
482 noise which is distracting for the development community as a whole, so,
483 increasingly, they are looked down upon. New developers wishing to
484 introduce themselves to the community will not get the sort of reception
485 they wish for by these means.
487 Andrew Morton gives this advice for aspiring kernel developers
491 The #1 project for all kernel beginners should surely be "make sure
492 that the kernel runs perfectly at all times on all machines which
493 you can lay your hands on". Usually the way to do this is to work
494 with others on getting things fixed up (this can require
495 persistence!) but that's fine - it's a part of kernel development.
497 (https://lwn.net/Articles/283982/).
499 In the absence of obvious problems to fix, developers are advised to look
500 at the current lists of regressions and open bugs in general. There is
501 never any shortage of issues in need of fixing; by addressing these issues,
502 developers will gain experience with the process while, at the same time,
503 building respect with the rest of the development community.
projects / platform / kernel / linux-rpi.git / blob