blob: 1d6560413cc593e001080759ce867c070956aeb7 [file] [log] [blame]
Greg Kroah-Hartmand36cc9d2005-11-18 09:31:11 -08001HOWTO do Linux kernel development
2---------------------------------
3
4This is the be-all, end-all document on this topic. It contains
5instructions on how to become a Linux kernel developer and how to learn
6to work with the Linux kernel development community. It tries to not
7contain anything related to the technical aspects of kernel programming,
8but will help point you in the right direction for that.
9
10If anything in this document becomes out of date, please send in patches
11to the maintainer of this file, who is listed at the bottom of the
12document.
13
14
15Introduction
16------------
17
18So, you want to learn how to become a Linux kernel developer? Or you
19have been told by your manager, "Go write a Linux driver for this
20device." This document's goal is to teach you everything you need to
21know to achieve this by describing the process you need to go through,
22and hints on how to work with the community. It will also try to
23explain some of the reasons why the community works like it does.
24
25The kernel is written mostly in C, with some architecture-dependent
26parts written in assembly. A good understanding of C is required for
27kernel development. Assembly (any architecture) is not required unless
28you plan to do low-level development for that architecture. Though they
29are not a good substitute for a solid C education and/or years of
30experience, the following books are good for, if anything, reference:
31 - "The C Programming Language" by Kernighan and Ritchie [Prentice Hall]
32 - "Practical C Programming" by Steve Oualline [O'Reilly]
33
34The kernel is written using GNU C and the GNU toolchain. While it
35adheres to the ISO C89 standard, it uses a number of extensions that are
36not featured in the standard. The kernel is a freestanding C
37environment, with no reliance on the standard C library, so some
38portions of the C standard are not supported. Arbitrary long long
39divisions and floating point are not allowed. It can sometimes be
40difficult to understand the assumptions the kernel has on the toolchain
41and the extensions that it uses, and unfortunately there is no
42definitive reference for them. Please check the gcc info pages (`info
43gcc`) for some information on them.
44
45Please remember that you are trying to learn how to work with the
46existing development community. It is a diverse group of people, with
47high standards for coding, style and procedure. These standards have
48been created over time based on what they have found to work best for
49such a large and geographically dispersed team. Try to learn as much as
50possible about these standards ahead of time, as they are well
51documented; do not expect people to adapt to you or your company's way
52of doing things.
53
54
55Legal Issues
56------------
57
58The Linux kernel source code is released under the GPL. Please see the
59file, COPYING, in the main directory of the source tree, for details on
60the license. If you have further questions about the license, please
61contact a lawyer, and do not ask on the Linux kernel mailing list. The
62people on the mailing lists are not lawyers, and you should not rely on
63their statements on legal matters.
64
65For common questions and answers about the GPL, please see:
66 http://www.gnu.org/licenses/gpl-faq.html
67
68
69Documentation
70------------
71
72The Linux kernel source tree has a large range of documents that are
73invaluable for learning how to interact with the kernel community. When
74new features are added to the kernel, it is recommended that new
75documentation files are also added which explain how to use the feature.
76When a kernel change causes the interface that the kernel exposes to
77userspace to change, it is recommended that you send the information or
78a patch to the manual pages explaining the change to the manual pages
79maintainer at mtk-manpages@gmx.net.
80
81Here is a list of files that are in the kernel source tree that are
82required reading:
83 README
84 This file gives a short background on the Linux kernel and describes
85 what is necessary to do to configure and build the kernel. People
86 who are new to the kernel should start here.
87
88 Documentation/Changes
89 This file gives a list of the minimum levels of various software
90 packages that are necessary to build and run the kernel
91 successfully.
92
93 Documentation/CodingStyle
94 This describes the Linux kernel coding style, and some of the
95 rationale behind it. All new code is expected to follow the
96 guidelines in this document. Most maintainers will only accept
97 patches if these rules are followed, and many people will only
98 review code if it is in the proper style.
99
100 Documentation/SubmittingPatches
101 Documentation/SubmittingDrivers
102 These files describe in explicit detail how to successfully create
103 and send a patch, including (but not limited to):
104 - Email contents
105 - Email format
106 - Who to send it to
107 Following these rules will not guarantee success (as all patches are
108 subject to scrutiny for content and style), but not following them
109 will almost always prevent it.
110
111 Other excellent descriptions of how to create patches properly are:
112 "The Perfect Patch"
113 http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt
114 "Linux kernel patch submission format"
115 http://linux.yyz.us/patch-format.html
116
117 Documentation/stable_api_nonsense.txt
118 This file describes the rationale behind the conscious decision to
119 not have a stable API within the kernel, including things like:
120 - Subsystem shim-layers (for compatibility?)
121 - Driver portability between Operating Systems.
122 - Mitigating rapid change within the kernel source tree (or
123 preventing rapid change)
124 This document is crucial for understanding the Linux development
125 philosophy and is very important for people moving to Linux from
126 development on other Operating Systems.
127
128 Documentation/SecurityBugs
129 If you feel you have found a security problem in the Linux kernel,
130 please follow the steps in this document to help notify the kernel
131 developers, and help solve the issue.
132
133 Documentation/ManagementStyle
134 This document describes how Linux kernel maintainers operate and the
135 shared ethos behind their methodologies. This is important reading
136 for anyone new to kernel development (or anyone simply curious about
137 it), as it resolves a lot of common misconceptions and confusion
138 about the unique behavior of kernel maintainers.
139
140 Documentation/stable_kernel_rules.txt
141 This file describes the rules on how the stable kernel releases
142 happen, and what to do if you want to get a change into one of these
143 releases.
144
145 Documentation/kernel-docs.txt
146 A list of external documentation that pertains to kernel
147 development. Please consult this list if you do not find what you
148 are looking for within the in-kernel documentation.
149
150 Documentation/applying-patches.txt
151 A good introduction describing exactly what a patch is and how to
152 apply it to the different development branches of the kernel.
153
154The kernel also has a large number of documents that can be
155automatically generated from the source code itself. This includes a
156full description of the in-kernel API, and rules on how to handle
157locking properly. The documents will be created in the
158Documentation/DocBook/ directory and can be generated as PDF,
159Postscript, HTML, and man pages by running:
160 make pdfdocs
161 make psdocs
162 make htmldocs
163 make mandocs
164respectively from the main kernel source directory.
165
166
167Becoming A Kernel Developer
168---------------------------
169
170If you do not know anything about Linux kernel development, you should
171look at the Linux KernelNewbies project:
172 http://kernelnewbies.org
173It consists of a helpful mailing list where you can ask almost any type
174of basic kernel development question (make sure to search the archives
175first, before asking something that has already been answered in the
176past.) It also has an IRC channel that you can use to ask questions in
177real-time, and a lot of helpful documentation that is useful for
178learning about Linux kernel development.
179
180The website has basic information about code organization, subsystems,
181and current projects (both in-tree and out-of-tree). It also describes
182some basic logistical information, like how to compile a kernel and
183apply a patch.
184
185If you do not know where you want to start, but you want to look for
186some task to start doing to join into the kernel development community,
187go to the Linux Kernel Janitor's project:
188 http://janitor.kernelnewbies.org/
189It is a great place to start. It describes a list of relatively simple
190problems that need to be cleaned up and fixed within the Linux kernel
191source tree. Working with the developers in charge of this project, you
192will learn the basics of getting your patch into the Linux kernel tree,
193and possibly be pointed in the direction of what to go work on next, if
194you do not already have an idea.
195
196If you already have a chunk of code that you want to put into the kernel
197tree, but need some help getting it in the proper form, the
198kernel-mentors project was created to help you out with this. It is a
199mailing list, and can be found at:
200 http://selenic.com/mailman/listinfo/kernel-mentors
201
202Before making any actual modifications to the Linux kernel code, it is
203imperative to understand how the code in question works. For this
204purpose, nothing is better than reading through it directly (most tricky
205bits are commented well), perhaps even with the help of specialized
206tools. One such tool that is particularly recommended is the Linux
207Cross-Reference project, which is able to present source code in a
208self-referential, indexed webpage format. An excellent up-to-date
209repository of the kernel code may be found at:
210 http://sosdg.org/~coywolf/lxr/
211
212
213The development process
214-----------------------
215
216Linux kernel development process currently consists of a few different
217main kernel "branches" and lots of different subsystem-specific kernel
218branches. These different branches are:
219 - main 2.6.x kernel tree
220 - 2.6.x.y -stable kernel tree
221 - 2.6.x -git kernel patches
222 - 2.6.x -mm kernel patches
223 - subsystem specific kernel trees and patches
224
2252.6.x kernel tree
226-----------------
2272.6.x kernels are maintained by Linus Torvalds, and can be found on
228kernel.org in the pub/linux/kernel/v2.6/ directory. Its development
229process is as follows:
230 - As soon as a new kernel is released a two weeks window is open,
231 during this period of time maintainers can submit big diffs to
232 Linus, usually the patches that have already been included in the
233 -mm kernel for a few weeks. The preferred way to submit big changes
234 is using git (the kernel's source management tool, more information
235 can be found at http://git.or.cz/) but plain patches are also just
236 fine.
237 - After two weeks a -rc1 kernel is released it is now possible to push
238 only patches that do not include new features that could affect the
239 stability of the whole kernel. Please note that a whole new driver
240 (or filesystem) might be accepted after -rc1 because there is no
241 risk of causing regressions with such a change as long as the change
242 is self-contained and does not affect areas outside of the code that
243 is being added. git can be used to send patches to Linus after -rc1
244 is released, but the patches need to also be sent to a public
245 mailing list for review.
246 - A new -rc is released whenever Linus deems the current git tree to
247 be in a reasonably sane state adequate for testing. The goal is to
248 release a new -rc kernel every week.
249 - Process continues until the kernel is considered "ready", the
250 process should last around 6 weeks.
251
252It is worth mentioning what Andrew Morton wrote on the linux-kernel
253mailing list about kernel releases:
254 "Nobody knows when a kernel will be released, because it's
255 released according to perceived bug status, not according to a
256 preconceived timeline."
257
2582.6.x.y -stable kernel tree
259---------------------------
260Kernels with 4 digit versions are -stable kernels. They contain
261relatively small and critical fixes for security problems or significant
262regressions discovered in a given 2.6.x kernel.
263
264This is the recommended branch for users who want the most recent stable
265kernel and are not interested in helping test development/experimental
266versions.
267
268If no 2.6.x.y kernel is available, then the highest numbered 2.6.x
269kernel is the current stable kernel.
270
2712.6.x.y are maintained by the "stable" team <stable@kernel.org>, and are
272released almost every other week.
273
274The file Documentation/stable_kernel_rules.txt in the kernel tree
275documents what kinds of changes are acceptable for the -stable tree, and
276how the release process works.
277
2782.6.x -git patches
279------------------
280These are daily snapshots of Linus' kernel tree which are managed in a
281git repository (hence the name.) These patches are usually released
282daily and represent the current state of Linus' tree. They are more
283experimental than -rc kernels since they are generated automatically
284without even a cursory glance to see if they are sane.
285
2862.6.x -mm kernel patches
287------------------------
288These are experimental kernel patches released by Andrew Morton. Andrew
289takes all of the different subsystem kernel trees and patches and mushes
290them together, along with a lot of patches that have been plucked from
291the linux-kernel mailing list. This tree serves as a proving ground for
292new features and patches. Once a patch has proved its worth in -mm for
293a while Andrew or the subsystem maintainer pushes it on to Linus for
294inclusion in mainline.
295
296It is heavily encouraged that all new patches get tested in the -mm tree
297before they are sent to Linus for inclusion in the main kernel tree.
298
299These kernels are not appropriate for use on systems that are supposed
300to be stable and they are more risky to run than any of the other
301branches.
302
303If you wish to help out with the kernel development process, please test
304and use these kernel releases and provide feedback to the linux-kernel
305mailing list if you have any problems, and if everything works properly.
306
307In addition to all the other experimental patches, these kernels usually
308also contain any changes in the mainline -git kernels available at the
309time of release.
310
311The -mm kernels are not released on a fixed schedule, but usually a few
312-mm kernels are released in between each -rc kernel (1 to 3 is common).
313
314Subsystem Specific kernel trees and patches
315-------------------------------------------
316A number of the different kernel subsystem developers expose their
317development trees so that others can see what is happening in the
318different areas of the kernel. These trees are pulled into the -mm
319kernel releases as described above.
320
321Here is a list of some of the different kernel trees available:
322 git trees:
323 - Kbuild development tree, Sam Ravnborg <sam@ravnborg.org>
324 kernel.org:/pub/scm/linux/kernel/git/sam/kbuild.git
325
326 - ACPI development tree, Len Brown <len.brown@intel.com>
327 kernel.org:/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6.git
328
329 - Block development tree, Jens Axboe <axboe@suse.de>
330 kernel.org:/pub/scm/linux/kernel/git/axboe/linux-2.6-block.git
331
332 - DRM development tree, Dave Airlie <airlied@linux.ie>
333 kernel.org:/pub/scm/linux/kernel/git/airlied/drm-2.6.git
334
335 - ia64 development tree, Tony Luck <tony.luck@intel.com>
336 kernel.org:/pub/scm/linux/kernel/git/aegl/linux-2.6.git
337
338 - ieee1394 development tree, Jody McIntyre <scjody@modernduck.com>
339 kernel.org:/pub/scm/linux/kernel/git/scjody/ieee1394.git
340
341 - infiniband, Roland Dreier <rolandd@cisco.com>
342 kernel.org:/pub/scm/linux/kernel/git/roland/infiniband.git
343
344 - libata, Jeff Garzik <jgarzik@pobox.com>
345 kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev.git
346
347 - network drivers, Jeff Garzik <jgarzik@pobox.com>
348 kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6.git
349
350 - pcmcia, Dominik Brodowski <linux@dominikbrodowski.net>
351 kernel.org:/pub/scm/linux/kernel/git/brodo/pcmcia-2.6.git
352
353 - SCSI, James Bottomley <James.Bottomley@SteelEye.com>
354 kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6.git
355
356 Other git kernel trees can be found listed at http://kernel.org/git
357
358 quilt trees:
359 - USB, PCI, Driver Core, and I2C, Greg Kroah-Hartman <gregkh@suse.de>
360 kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/
Andi Kleen027a51c2006-09-26 10:52:38 +0200361 - x86-64, partly i386, Andi Kleen <ak@suse.de>
362 ftp.firstfloor.org:/pub/ak/x86_64/quilt/
Greg Kroah-Hartmand36cc9d2005-11-18 09:31:11 -0800363
364Bug Reporting
365-------------
366
367bugzilla.kernel.org is where the Linux kernel developers track kernel
368bugs. Users are encouraged to report all bugs that they find in this
369tool. For details on how to use the kernel bugzilla, please see:
370 http://test.kernel.org/bugzilla/faq.html
371
372The file REPORTING-BUGS in the main kernel source directory has a good
373template for how to report a possible kernel bug, and details what kind
374of information is needed by the kernel developers to help track down the
375problem.
376
377
378Mailing lists
379-------------
380
381As some of the above documents describe, the majority of the core kernel
382developers participate on the Linux Kernel Mailing list. Details on how
383to subscribe and unsubscribe from the list can be found at:
384 http://vger.kernel.org/vger-lists.html#linux-kernel
385There are archives of the mailing list on the web in many different
386places. Use a search engine to find these archives. For example:
387 http://dir.gmane.org/gmane.linux.kernel
388It is highly recommended that you search the archives about the topic
389you want to bring up, before you post it to the list. A lot of things
390already discussed in detail are only recorded at the mailing list
391archives.
392
393Most of the individual kernel subsystems also have their own separate
394mailing list where they do their development efforts. See the
395MAINTAINERS file for a list of what these lists are for the different
396groups.
397
398Many of the lists are hosted on kernel.org. Information on them can be
399found at:
400 http://vger.kernel.org/vger-lists.html
401
402Please remember to follow good behavioral habits when using the lists.
403Though a bit cheesy, the following URL has some simple guidelines for
404interacting with the list (or any list):
405 http://www.albion.com/netiquette/
406
407If multiple people respond to your mail, the CC: list of recipients may
408get pretty large. Don't remove anybody from the CC: list without a good
409reason, or don't reply only to the list address. Get used to receiving the
410mail twice, one from the sender and the one from the list, and don't try
411to tune that by adding fancy mail-headers, people will not like it.
412
413Remember to keep the context and the attribution of your replies intact,
414keep the "John Kernelhacker wrote ...:" lines at the top of your reply, and
415add your statements between the individual quoted sections instead of
416writing at the top of the mail.
417
418If you add patches to your mail, make sure they are plain readable text
419as stated in Documentation/SubmittingPatches. Kernel developers don't
420want to deal with attachments or compressed patches; they may want
421to comment on individual lines of your patch, which works only that way.
422Make sure you use a mail program that does not mangle spaces and tab
423characters. A good first test is to send the mail to yourself and try
424to apply your own patch by yourself. If that doesn't work, get your
425mail program fixed or change it until it works.
426
427Above all, please remember to show respect to other subscribers.
428
429
430Working with the community
431--------------------------
432
433The goal of the kernel community is to provide the best possible kernel
434there is. When you submit a patch for acceptance, it will be reviewed
435on its technical merits and those alone. So, what should you be
436expecting?
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 - expect your patch to be accepted without question
452 - become defensive
453 - ignore comments
454 - resubmit the patch without making any of the requested changes
455
456In a community that is looking for the best technical solution possible,
457there will always be differing opinions on how beneficial a patch is.
458You have to be cooperative, and willing to adapt your idea to fit within
459the kernel. Or at least be willing to prove your idea is worth it.
460Remember, being wrong is acceptable as long as you are willing to work
461toward a solution that is right.
462
463It is normal that the answers to your first patch might simply be a list
464of a dozen things you should correct. This does _not_ imply that your
465patch will not be accepted, and it is _not_ meant against you
466personally. Simply correct all issues raised against your patch and
467resend it.
468
469
470Differences between the kernel community and corporate structures
471-----------------------------------------------------------------
472
473The kernel community works differently than most traditional corporate
474development environments. Here are a list of things that you can try to
475do to try to avoid problems:
476 Good things to say regarding your proposed changes:
477 - "This solves multiple problems."
478 - "This deletes 2000 lines of code."
479 - "Here is a patch that explains what I am trying to describe."
480 - "I tested it on 5 different architectures..."
481 - "Here is a series of small patches that..."
482 - "This increases performance on typical machines..."
483
484 Bad things you should avoid saying:
485 - "We did it this way in AIX/ptx/Solaris, so therefore it must be
486 good..."
487 - "I've being doing this for 20 years, so..."
488 - "This is required for my company to make money"
489 - "This is for our Enterprise product line."
490 - "Here is my 1000 page design document that describes my idea"
491 - "I've been working on this for 6 months..."
492 - "Here's a 5000 line patch that..."
493 - "I rewrote all of the current mess, and here it is..."
494 - "I have a deadline, and this patch needs to be applied now."
495
496Another way the kernel community is different than most traditional
497software engineering work environments is the faceless nature of
498interaction. One benefit of using email and irc as the primary forms of
499communication is the lack of discrimination based on gender or race.
500The Linux kernel work environment is accepting of women and minorities
501because all you are is an email address. The international aspect also
502helps to level the playing field because you can't guess gender based on
503a person's name. A man may be named Andrea and a woman may be named Pat.
504Most women who have worked in the Linux kernel and have expressed an
505opinion have had positive experiences.
506
507The language barrier can cause problems for some people who are not
508comfortable with English. A good grasp of the language can be needed in
509order to get ideas across properly on mailing lists, so it is
510recommended that you check your emails to make sure they make sense in
511English before sending them.
512
513
514Break up your changes
515---------------------
516
517The Linux kernel community does not gladly accept large chunks of code
518dropped on it all at once. The changes need to be properly introduced,
519discussed, and broken up into tiny, individual portions. This is almost
520the exact opposite of what companies are used to doing. Your proposal
521should also be introduced very early in the development process, so that
522you can receive feedback on what you are doing. It also lets the
523community feel that you are working with them, and not simply using them
524as a dumping ground for your feature. However, don't send 50 emails at
525one time to a mailing list, your patch series should be smaller than
526that almost all of the time.
527
528The reasons for breaking things up are the following:
529
5301) Small patches increase the likelihood that your patches will be
531 applied, since they don't take much time or effort to verify for
532 correctness. A 5 line patch can be applied by a maintainer with
533 barely a second glance. However, a 500 line patch may take hours to
534 review for correctness (the time it takes is exponentially
535 proportional to the size of the patch, or something).
536
537 Small patches also make it very easy to debug when something goes
538 wrong. It's much easier to back out patches one by one than it is
539 to dissect a very large patch after it's been applied (and broken
540 something).
541
5422) It's important not only to send small patches, but also to rewrite
543 and simplify (or simply re-order) patches before submitting them.
544
545Here is an analogy from kernel developer Al Viro:
546 "Think of a teacher grading homework from a math student. The
547 teacher does not want to see the student's trials and errors
548 before they came up with the solution. They want to see the
549 cleanest, most elegant answer. A good student knows this, and
550 would never submit her intermediate work before the final
551 solution."
552
553 The same is true of kernel development. The maintainers and
554 reviewers do not want to see the thought process behind the
555 solution to the problem one is solving. They want to see a
556 simple and elegant solution."
557
558It may be challenging to keep the balance between presenting an elegant
559solution and working together with the community and discussing your
560unfinished work. Therefore it is good to get early in the process to
561get feedback to improve your work, but also keep your changes in small
562chunks that they may get already accepted, even when your whole task is
563not ready for inclusion now.
564
565Also realize that it is not acceptable to send patches for inclusion
566that are unfinished and will be "fixed up later."
567
568
569Justify your change
570-------------------
571
572Along with breaking up your patches, it is very important for you to let
573the Linux community know why they should add this change. New features
574must be justified as being needed and useful.
575
576
577Document your change
578--------------------
579
580When sending in your patches, pay special attention to what you say in
581the text in your email. This information will become the ChangeLog
582information for the patch, and will be preserved for everyone to see for
583all time. It should describe the patch completely, containing:
584 - why the change is necessary
585 - the overall design approach in the patch
586 - implementation details
587 - testing results
588
589For more details on what this should all look like, please see the
590ChangeLog section of the document:
591 "The Perfect Patch"
592 http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt
593
594
595
596
597All of these things are sometimes very hard to do. It can take years to
598perfect these practices (if at all). It's a continuous process of
599improvement that requires a lot of patience and determination. But
600don't give up, it's possible. Many have done it before, and each had to
601start exactly where you are now.
602
603
604
605
606----------
Paolo Ciarrocchi5bd982e2006-04-25 22:47:51 +0200607Thanks to Paolo Ciarrocchi who allowed the "Development Process"
608(http://linux.tar.bz/articles/2.6-development_process) section
Greg Kroah-Hartmand36cc9d2005-11-18 09:31:11 -0800609to be based on text he had written, and to Randy Dunlap and Gerrit
610Huizenga for some of the list of things you should and should not say.
611Also thanks to Pat Mochel, Hanna Linder, Randy Dunlap, Kay Sievers,
612Vojtech Pavlik, Jan Kara, Josh Boyer, Kees Cook, Andrew Morton, Andi
613Kleen, Vadim Lobanov, Jesper Juhl, Adrian Bunk, Keri Harris, Frans Pop,
614David A. Wheeler, Junio Hamano, Michael Kerrisk, and Alex Shepard for
615their review, comments, and contributions. Without their help, this
616document would not have been possible.
617
618
619
620Maintainer: Greg Kroah-Hartman <greg@kroah.com>