1.. _process_howto:
2
3HOWTO do Linux kernel development
4=================================
5
6This is the be-all, end-all document on this topic.  It contains
7instructions on how to become a Linux kernel developer and how to learn
8to work with the Linux kernel development community.  It tries to not
9contain anything related to the technical aspects of kernel programming,
10but will help point you in the right direction for that.
11
12If anything in this document becomes out of date, please send in patches
13to the maintainer of this file, who is listed at the bottom of the
14document.
15
16
17Introduction
18------------
19
20So, you want to learn how to become a Linux kernel developer?  Or you
21have been told by your manager, "Go write a Linux driver for this
22device."  This document's goal is to teach you everything you need to
23know to achieve this by describing the process you need to go through,
24and hints on how to work with the community.  It will also try to
25explain some of the reasons why the community works like it does.
26
27The kernel is written mostly in C, with some architecture-dependent
28parts written in assembly. A good understanding of C is required for
29kernel development.  Assembly (any architecture) is not required unless
30you plan to do low-level development for that architecture.  Though they
31are not a good substitute for a solid C education and/or years of
32experience, the following books are good for, if anything, reference:
33
34 - "The C Programming Language" by Kernighan and Ritchie [Prentice Hall]
35 - "Practical C Programming" by Steve Oualline [O'Reilly]
36 - "C:  A Reference Manual" by Harbison and Steele [Prentice Hall]
37
38The kernel is written using GNU C and the GNU toolchain.  While it
39adheres to the ISO C89 standard, it uses a number of extensions that are
40not featured in the standard.  The kernel is a freestanding C
41environment, with no reliance on the standard C library, so some
42portions of the C standard are not supported.  Arbitrary long long
43divisions and floating point are not allowed.  It can sometimes be
44difficult to understand the assumptions the kernel has on the toolchain
45and the extensions that it uses, and unfortunately there is no
46definitive reference for them.  Please check the gcc info pages (`info
47gcc`) for some information on them.
48
49Please remember that you are trying to learn how to work with the
50existing development community.  It is a diverse group of people, with
51high standards for coding, style and procedure.  These standards have
52been created over time based on what they have found to work best for
53such a large and geographically dispersed team.  Try to learn as much as
54possible about these standards ahead of time, as they are well
55documented; do not expect people to adapt to you or your company's way
56of doing things.
57
58
59Legal Issues
60------------
61
62The Linux kernel source code is released under the GPL.  Please see the file
63COPYING in the main directory of the source tree. The Linux kernel licensing
64rules and how to use `SPDX <https://spdx.org/>`_ identifiers in source code are
65described in :ref:`Documentation/process/license-rules.rst <kernel_licensing>`.
66If you have further questions about the license, please contact a lawyer, and do
67not ask on the Linux kernel mailing list.  The people on the mailing lists are
68not lawyers, and you should not rely on their statements on legal matters.
69
70For common questions and answers about the GPL, please see:
71
72	https://www.gnu.org/licenses/gpl-faq.html
73
74
75Documentation
76-------------
77
78The Linux kernel source tree has a large range of documents that are
79invaluable for learning how to interact with the kernel community.  When
80new features are added to the kernel, it is recommended that new
81documentation files are also added which explain how to use the feature.
82When a kernel change causes the interface that the kernel exposes to
83userspace to change, it is recommended that you send the information or
84a patch to the manual pages explaining the change to the manual pages
85maintainer at mtk.manpages@gmail.com, and CC the list
86linux-api@vger.kernel.org.
87
88Here is a list of files that are in the kernel source tree that are
89required reading:
90
91  :ref:`Documentation/admin-guide/README.rst <readme>`
92    This file gives a short background on the Linux kernel and describes
93    what is necessary to do to configure and build the kernel.  People
94    who are new to the kernel should start here.
95
96  :ref:`Documentation/process/changes.rst <changes>`
97    This file gives a list of the minimum levels of various software
98    packages that are necessary to build and run the kernel
99    successfully.
100
101  :ref:`Documentation/process/coding-style.rst <codingstyle>`
102    This describes the Linux kernel coding style, and some of the
103    rationale behind it. All new code is expected to follow the
104    guidelines in this document. Most maintainers will only accept
105    patches if these rules are followed, and many people will only
106    review code if it is in the proper style.
107
108  :ref:`Documentation/process/submitting-patches.rst <submittingpatches>` and :ref:`Documentation/process/submitting-drivers.rst <submittingdrivers>`
109    These files describe in explicit detail how to successfully create
110    and send a patch, including (but not limited to):
111
112       - Email contents
113       - Email format
114       - Who to send it to
115
116    Following these rules will not guarantee success (as all patches are
117    subject to scrutiny for content and style), but not following them
118    will almost always prevent it.
119
120    Other excellent descriptions of how to create patches properly are:
121
122	"The Perfect Patch"
123		https://www.ozlabs.org/~akpm/stuff/tpp.txt
124
125	"Linux kernel patch submission format"
126		https://web.archive.org/web/20180829112450/http://linux.yyz.us/patch-format.html
127
128  :ref:`Documentation/process/stable-api-nonsense.rst <stable_api_nonsense>`
129    This file describes the rationale behind the conscious decision to
130    not have a stable API within the kernel, including things like:
131
132      - Subsystem shim-layers (for compatibility?)
133      - Driver portability between Operating Systems.
134      - Mitigating rapid change within the kernel source tree (or
135	preventing rapid change)
136
137    This document is crucial for understanding the Linux development
138    philosophy and is very important for people moving to Linux from
139    development on other Operating Systems.
140
141  :ref:`Documentation/admin-guide/security-bugs.rst <securitybugs>`
142    If you feel you have found a security problem in the Linux kernel,
143    please follow the steps in this document to help notify the kernel
144    developers, and help solve the issue.
145
146  :ref:`Documentation/process/management-style.rst <managementstyle>`
147    This document describes how Linux kernel maintainers operate and the
148    shared ethos behind their methodologies.  This is important reading
149    for anyone new to kernel development (or anyone simply curious about
150    it), as it resolves a lot of common misconceptions and confusion
151    about the unique behavior of kernel maintainers.
152
153  :ref:`Documentation/process/stable-kernel-rules.rst <stable_kernel_rules>`
154    This file describes the rules on how the stable kernel releases
155    happen, and what to do if you want to get a change into one of these
156    releases.
157
158  :ref:`Documentation/process/kernel-docs.rst <kernel_docs>`
159    A list of external documentation that pertains to kernel
160    development.  Please consult this list if you do not find what you
161    are looking for within the in-kernel documentation.
162
163  :ref:`Documentation/process/applying-patches.rst <applying_patches>`
164    A good introduction describing exactly what a patch is and how to
165    apply it to the different development branches of the kernel.
166
167The kernel also has a large number of documents that can be
168automatically generated from the source code itself or from
169ReStructuredText markups (ReST), like this one. This includes a
170full description of the in-kernel API, and rules on how to handle
171locking properly.
172
173All such documents can be generated as PDF or HTML by running::
174
175	make pdfdocs
176	make htmldocs
177
178respectively from the main kernel source directory.
179
180The documents that uses ReST markup will be generated at Documentation/output.
181They can also be generated on LaTeX and ePub formats with::
182
183	make latexdocs
184	make epubdocs
185
186Becoming A Kernel Developer
187---------------------------
188
189If you do not know anything about Linux kernel development, you should
190look at the Linux KernelNewbies project:
191
192	https://kernelnewbies.org
193
194It consists of a helpful mailing list where you can ask almost any type
195of basic kernel development question (make sure to search the archives
196first, before asking something that has already been answered in the
197past.)  It also has an IRC channel that you can use to ask questions in
198real-time, and a lot of helpful documentation that is useful for
199learning about Linux kernel development.
200
201The website has basic information about code organization, subsystems,
202and current projects (both in-tree and out-of-tree). It also describes
203some basic logistical information, like how to compile a kernel and
204apply a patch.
205
206If you do not know where you want to start, but you want to look for
207some task to start doing to join into the kernel development community,
208go to the Linux Kernel Janitor's project:
209
210	https://kernelnewbies.org/KernelJanitors
211
212It is a great place to start.  It describes a list of relatively simple
213problems that need to be cleaned up and fixed within the Linux kernel
214source tree.  Working with the developers in charge of this project, you
215will learn the basics of getting your patch into the Linux kernel tree,
216and possibly be pointed in the direction of what to go work on next, if
217you do not already have an idea.
218
219Before making any actual modifications to the Linux kernel code, it is
220imperative to understand how the code in question works.  For this
221purpose, nothing is better than reading through it directly (most tricky
222bits are commented well), perhaps even with the help of specialized
223tools.  One such tool that is particularly recommended is the Linux
224Cross-Reference project, which is able to present source code in a
225self-referential, indexed webpage format. An excellent up-to-date
226repository of the kernel code may be found at:
227
228	https://elixir.bootlin.com/
229
230
231The development process
232-----------------------
233
234Linux kernel development process currently consists of a few different
235main kernel "branches" and lots of different subsystem-specific kernel
236branches.  These different branches are:
237
238  - Linus's mainline tree
239  - Various stable trees with multiple major numbers
240  - Subsystem-specific trees
241  - linux-next integration testing tree
242
243Mainline tree
244~~~~~~~~~~~~~
245
246Mainline tree are maintained by Linus Torvalds, and can be found at
247https://kernel.org or in the repo.  Its development process is as follows:
248
249  - As soon as a new kernel is released a two weeks window is open,
250    during this period of time maintainers can submit big diffs to
251    Linus, usually the patches that have already been included in the
252    linux-next for a few weeks.  The preferred way to submit big changes
253    is using git (the kernel's source management tool, more information
254    can be found at https://git-scm.com/) but plain patches are also just
255    fine.
256  - After two weeks a -rc1 kernel is released and the focus is on making the
257    new kernel as rock solid as possible.  Most of the patches at this point
258    should fix a regression.  Bugs that have always existed are not
259    regressions, so only push these kinds of fixes if they are important.
260    Please note that a whole new driver (or filesystem) might be accepted
261    after -rc1 because there is no risk of causing regressions with such a
262    change as long as the change is self-contained and does not affect areas
263    outside of the code that is being added.  git can be used to send
264    patches to Linus after -rc1 is released, but the patches need to also be
265    sent to a public mailing list for review.
266  - A new -rc is released whenever Linus deems the current git tree to
267    be in a reasonably sane state adequate for testing.  The goal is to
268    release a new -rc kernel every week.
269  - Process continues until the kernel is considered "ready", the
270    process should last around 6 weeks.
271
272It is worth mentioning what Andrew Morton wrote on the linux-kernel
273mailing list about kernel releases:
274
275	*"Nobody knows when a kernel will be released, because it's
276	released according to perceived bug status, not according to a
277	preconceived timeline."*
278
279Various stable trees with multiple major numbers
280~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
281
282Kernels with 3-part versions are -stable kernels. They contain
283relatively small and critical fixes for security problems or significant
284regressions discovered in a given major mainline release, with the first
2852-part of version number are the same correspondingly.
286
287This is the recommended branch for users who want the most recent stable
288kernel and are not interested in helping test development/experimental
289versions.
290
291Stable trees are maintained by the "stable" team <stable@vger.kernel.org>, and
292are released as needs dictate.  The normal release period is approximately
293two weeks, but it can be longer if there are no pressing problems.  A
294security-related problem, instead, can cause a release to happen almost
295instantly.
296
297The file :ref:`Documentation/process/stable-kernel-rules.rst <stable_kernel_rules>`
298in the kernel tree documents what kinds of changes are acceptable for
299the -stable tree, and how the release process works.
300
301Subsystem-specific trees
302~~~~~~~~~~~~~~~~~~~~~~~~
303
304The maintainers of the various kernel subsystems --- and also many
305kernel subsystem developers --- expose their current state of
306development in source repositories.  That way, others can see what is
307happening in the different areas of the kernel.  In areas where
308development is rapid, a developer may be asked to base his submissions
309onto such a subsystem kernel tree so that conflicts between the
310submission and other already ongoing work are avoided.
311
312Most of these repositories are git trees, but there are also other SCMs
313in use, or patch queues being published as quilt series.  Addresses of
314these subsystem repositories are listed in the MAINTAINERS file.  Many
315of them can be browsed at https://git.kernel.org/.
316
317Before a proposed patch is committed to such a subsystem tree, it is
318subject to review which primarily happens on mailing lists (see the
319respective section below).  For several kernel subsystems, this review
320process is tracked with the tool patchwork.  Patchwork offers a web
321interface which shows patch postings, any comments on a patch or
322revisions to it, and maintainers can mark patches as under review,
323accepted, or rejected.  Most of these patchwork sites are listed at
324https://patchwork.kernel.org/.
325
326linux-next integration testing tree
327~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
328
329Before updates from subsystem trees are merged into the mainline tree,
330they need to be integration-tested.  For this purpose, a special
331testing repository exists into which virtually all subsystem trees are
332pulled on an almost daily basis:
333
334	https://git.kernel.org/?p=linux/kernel/git/next/linux-next.git
335
336This way, the linux-next gives a summary outlook onto what will be
337expected to go into the mainline kernel at the next merge period.
338Adventurous testers are very welcome to runtime-test the linux-next.
339
340
341Bug Reporting
342-------------
343
344https://bugzilla.kernel.org is where the Linux kernel developers track kernel
345bugs.  Users are encouraged to report all bugs that they find in this
346tool.  For details on how to use the kernel bugzilla, please see:
347
348	https://bugzilla.kernel.org/page.cgi?id=faq.html
349
350The file :ref:`admin-guide/reporting-bugs.rst <reportingbugs>`
351in the main kernel source directory has a good
352template for how to report a possible kernel bug, and details what kind
353of information is needed by the kernel developers to help track down the
354problem.
355
356
357Managing bug reports
358--------------------
359
360One of the best ways to put into practice your hacking skills is by fixing
361bugs reported by other people. Not only you will help to make the kernel
362more stable, you'll learn to fix real world problems and you will improve
363your skills, and other developers will be aware of your presence. Fixing
364bugs is one of the best ways to get merits among other developers, because
365not many people like wasting time fixing other people's bugs.
366
367To work in the already reported bug reports, go to https://bugzilla.kernel.org.
368
369
370Mailing lists
371-------------
372
373As some of the above documents describe, the majority of the core kernel
374developers participate on the Linux Kernel Mailing list.  Details on how
375to subscribe and unsubscribe from the list can be found at:
376
377	http://vger.kernel.org/vger-lists.html#linux-kernel
378
379There are archives of the mailing list on the web in many different
380places.  Use a search engine to find these archives.  For example:
381
382	http://dir.gmane.org/gmane.linux.kernel
383
384It is highly recommended that you search the archives about the topic
385you want to bring up, before you post it to the list. A lot of things
386already discussed in detail are only recorded at the mailing list
387archives.
388
389Most of the individual kernel subsystems also have their own separate
390mailing list where they do their development efforts.  See the
391MAINTAINERS file for a list of what these lists are for the different
392groups.
393
394Many of the lists are hosted on kernel.org. Information on them can be
395found at:
396
397	http://vger.kernel.org/vger-lists.html
398
399Please remember to follow good behavioral habits when using the lists.
400Though a bit cheesy, the following URL has some simple guidelines for
401interacting with the list (or any list):
402
403	http://www.albion.com/netiquette/
404
405If multiple people respond to your mail, the CC: list of recipients may
406get pretty large. Don't remove anybody from the CC: list without a good
407reason, or don't reply only to the list address. Get used to receiving the
408mail twice, one from the sender and the one from the list, and don't try
409to tune that by adding fancy mail-headers, people will not like it.
410
411Remember to keep the context and the attribution of your replies intact,
412keep the "John Kernelhacker wrote ...:" lines at the top of your reply, and
413add your statements between the individual quoted sections instead of
414writing at the top of the mail.
415
416If you add patches to your mail, make sure they are plain readable text
417as stated in :ref:`Documentation/process/submitting-patches.rst <submittingpatches>`.
418Kernel developers don't want to deal with
419attachments or compressed patches; they may want to comment on
420individual lines of your patch, which works only that way. Make sure you
421use a mail program that does not mangle spaces and tab characters. A
422good first test is to send the mail to yourself and try to apply your
423own patch by yourself. If that doesn't work, get your mail program fixed
424or change it until it works.
425
426Above all, please remember to show respect to other subscribers.
427
428
429Working with the community
430--------------------------
431
432The goal of the kernel community is to provide the best possible kernel
433there is.  When you submit a patch for acceptance, it will be reviewed
434on its technical merits and those alone.  So, what should you be
435expecting?
436
437  - criticism
438  - comments
439  - requests for change
440  - requests for justification
441  - silence
442
443Remember, this is part of getting your patch into the kernel.  You have
444to be able to take criticism and comments about your patches, evaluate
445them at a technical level and either rework your patches or provide
446clear and concise reasoning as to why those changes should not be made.
447If there are no responses to your posting, wait a few days and try
448again, sometimes things get lost in the huge volume.
449
450What should you not do?
451
452  - expect your patch to be accepted without question
453  - become defensive
454  - ignore comments
455  - resubmit the patch without making any of the requested changes
456
457In a community that is looking for the best technical solution possible,
458there will always be differing opinions on how beneficial a patch is.
459You have to be cooperative, and willing to adapt your idea to fit within
460the kernel.  Or at least be willing to prove your idea is worth it.
461Remember, being wrong is acceptable as long as you are willing to work
462toward a solution that is right.
463
464It is normal that the answers to your first patch might simply be a list
465of a dozen things you should correct.  This does **not** imply that your
466patch will not be accepted, and it is **not** meant against you
467personally.  Simply correct all issues raised against your patch and
468resend it.
469
470
471Differences between the kernel community and corporate structures
472-----------------------------------------------------------------
473
474The kernel community works differently than most traditional corporate
475development environments.  Here are a list of things that you can try to
476do to avoid problems:
477
478  Good things to say regarding your proposed changes:
479
480    - "This solves multiple problems."
481    - "This deletes 2000 lines of code."
482    - "Here is a patch that explains what I am trying to describe."
483    - "I tested it on 5 different architectures..."
484    - "Here is a series of small patches that..."
485    - "This increases performance on typical machines..."
486
487  Bad things you should avoid saying:
488
489    - "We did it this way in AIX/ptx/Solaris, so therefore it must be
490      good..."
491    - "I've being doing this for 20 years, so..."
492    - "This is required for my company to make money"
493    - "This is for our Enterprise product line."
494    - "Here is my 1000 page design document that describes my idea"
495    - "I've been working on this for 6 months..."
496    - "Here's a 5000 line patch that..."
497    - "I rewrote all of the current mess, and here it is..."
498    - "I have a deadline, and this patch needs to be applied now."
499
500Another way the kernel community is different than most traditional
501software engineering work environments is the faceless nature of
502interaction.  One benefit of using email and irc as the primary forms of
503communication is the lack of discrimination based on gender or race.
504The Linux kernel work environment is accepting of women and minorities
505because all you are is an email address.  The international aspect also
506helps to level the playing field because you can't guess gender based on
507a person's name. A man may be named Andrea and a woman may be named Pat.
508Most women who have worked in the Linux kernel and have expressed an
509opinion have had positive experiences.
510
511The language barrier can cause problems for some people who are not
512comfortable with English.  A good grasp of the language can be needed in
513order to get ideas across properly on mailing lists, so it is
514recommended that you check your emails to make sure they make sense in
515English before sending them.
516
517
518Break up your changes
519---------------------
520
521The Linux kernel community does not gladly accept large chunks of code
522dropped on it all at once.  The changes need to be properly introduced,
523discussed, and broken up into tiny, individual portions.  This is almost
524the exact opposite of what companies are used to doing.  Your proposal
525should also be introduced very early in the development process, so that
526you can receive feedback on what you are doing.  It also lets the
527community feel that you are working with them, and not simply using them
528as a dumping ground for your feature.  However, don't send 50 emails at
529one time to a mailing list, your patch series should be smaller than
530that almost all of the time.
531
532The reasons for breaking things up are the following:
533
5341) Small patches increase the likelihood that your patches will be
535   applied, since they don't take much time or effort to verify for
536   correctness.  A 5 line patch can be applied by a maintainer with
537   barely a second glance. However, a 500 line patch may take hours to
538   review for correctness (the time it takes is exponentially
539   proportional to the size of the patch, or something).
540
541   Small patches also make it very easy to debug when something goes
542   wrong.  It's much easier to back out patches one by one than it is
543   to dissect a very large patch after it's been applied (and broken
544   something).
545
5462) It's important not only to send small patches, but also to rewrite
547   and simplify (or simply re-order) patches before submitting them.
548
549Here is an analogy from kernel developer Al Viro:
550
551	*"Think of a teacher grading homework from a math student.  The
552	teacher does not want to see the student's trials and errors
553	before they came up with the solution. They want to see the
554	cleanest, most elegant answer.  A good student knows this, and
555	would never submit her intermediate work before the final
556	solution.*
557
558	*The same is true of kernel development. The maintainers and
559	reviewers do not want to see the thought process behind the
560	solution to the problem one is solving. They want to see a
561	simple and elegant solution."*
562
563It may be challenging to keep the balance between presenting an elegant
564solution and working together with the community and discussing your
565unfinished work. Therefore it is good to get early in the process to
566get feedback to improve your work, but also keep your changes in small
567chunks that they may get already accepted, even when your whole task is
568not ready for inclusion now.
569
570Also realize that it is not acceptable to send patches for inclusion
571that are unfinished and will be "fixed up later."
572
573
574Justify your change
575-------------------
576
577Along with breaking up your patches, it is very important for you to let
578the Linux community know why they should add this change.  New features
579must be justified as being needed and useful.
580
581
582Document your change
583--------------------
584
585When sending in your patches, pay special attention to what you say in
586the text in your email.  This information will become the ChangeLog
587information for the patch, and will be preserved for everyone to see for
588all time.  It should describe the patch completely, containing:
589
590  - why the change is necessary
591  - the overall design approach in the patch
592  - implementation details
593  - testing results
594
595For more details on what this should all look like, please see the
596ChangeLog section of the document:
597
598  "The Perfect Patch"
599      http://www.ozlabs.org/~akpm/stuff/tpp.txt
600
601
602All of these things are sometimes very hard to do. It can take years to
603perfect these practices (if at all). It's a continuous process of
604improvement that requires a lot of patience and determination. But
605don't give up, it's possible. Many have done it before, and each had to
606start exactly where you are now.
607
608
609
610
611----------
612
613Thanks to Paolo Ciarrocchi who allowed the "Development Process"
614(https://lwn.net/Articles/94386/) section
615to be based on text he had written, and to Randy Dunlap and Gerrit
616Huizenga for some of the list of things you should and should not say.
617Also thanks to Pat Mochel, Hanna Linder, Randy Dunlap, Kay Sievers,
618Vojtech Pavlik, Jan Kara, Josh Boyer, Kees Cook, Andrew Morton, Andi
619Kleen, Vadim Lobanov, Jesper Juhl, Adrian Bunk, Keri Harris, Frans Pop,
620David A. Wheeler, Junio Hamano, Michael Kerrisk, and Alex Shepard for
621their review, comments, and contributions.  Without their help, this
622document would not have been possible.
623
624
625
626Maintainer: Greg Kroah-Hartman <greg@kroah.com>
627