SubmittingPatches 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509
  1. How to Get Your Change Into the Linux Kernel
  2. or
  3. Care And Operation Of Your Linus Torvalds
  4. For a person or company who wishes to submit a change to the Linux
  5. kernel, the process can sometimes be daunting if you're not familiar
  6. with "the system." This text is a collection of suggestions which
  7. can greatly increase the chances of your change being accepted.
  8. Read Documentation/SubmitChecklist for a list of items to check
  9. before submitting code. If you are submitting a driver, also read
  10. Documentation/SubmittingDrivers.
  11. --------------------------------------------
  12. SECTION 1 - CREATING AND SENDING YOUR CHANGE
  13. --------------------------------------------
  14. 1) "diff -up"
  15. ------------
  16. Use "diff -up" or "diff -uprN" to create patches.
  17. All changes to the Linux kernel occur in the form of patches, as
  18. generated by diff(1). When creating your patch, make sure to create it
  19. in "unified diff" format, as supplied by the '-u' argument to diff(1).
  20. Also, please use the '-p' argument which shows which C function each
  21. change is in - that makes the resultant diff a lot easier to read.
  22. Patches should be based in the root kernel source directory,
  23. not in any lower subdirectory.
  24. To create a patch for a single file, it is often sufficient to do:
  25. SRCTREE= linux-2.6
  26. MYFILE= drivers/net/mydriver.c
  27. cd $SRCTREE
  28. cp $MYFILE $MYFILE.orig
  29. vi $MYFILE # make your change
  30. cd ..
  31. diff -up $SRCTREE/$MYFILE{.orig,} > /tmp/patch
  32. To create a patch for multiple files, you should unpack a "vanilla",
  33. or unmodified kernel source tree, and generate a diff against your
  34. own source tree. For example:
  35. MYSRC= /devel/linux-2.6
  36. tar xvfz linux-2.6.12.tar.gz
  37. mv linux-2.6.12 linux-2.6.12-vanilla
  38. diff -uprN -X linux-2.6.12-vanilla/Documentation/dontdiff \
  39. linux-2.6.12-vanilla $MYSRC > /tmp/patch
  40. "dontdiff" is a list of files which are generated by the kernel during
  41. the build process, and should be ignored in any diff(1)-generated
  42. patch. The "dontdiff" file is included in the kernel tree in
  43. 2.6.12 and later. For earlier kernel versions, you can get it
  44. from <http://www.xenotime.net/linux/doc/dontdiff>.
  45. Make sure your patch does not include any extra files which do not
  46. belong in a patch submission. Make sure to review your patch -after-
  47. generated it with diff(1), to ensure accuracy.
  48. If your changes produce a lot of deltas, you may want to look into
  49. splitting them into individual patches which modify things in
  50. logical stages. This will facilitate easier reviewing by other
  51. kernel developers, very important if you want your patch accepted.
  52. There are a number of scripts which can aid in this:
  53. Quilt:
  54. http://savannah.nongnu.org/projects/quilt
  55. Andrew Morton's patch scripts:
  56. http://www.zip.com.au/~akpm/linux/patches/
  57. Instead of these scripts, quilt is the recommended patch management
  58. tool (see above).
  59. 2) Describe your changes.
  60. Describe the technical detail of the change(s) your patch includes.
  61. Be as specific as possible. The WORST descriptions possible include
  62. things like "update driver X", "bug fix for driver X", or "this patch
  63. includes updates for subsystem X. Please apply."
  64. If your description starts to get long, that's a sign that you probably
  65. need to split up your patch. See #3, next.
  66. 3) Separate your changes.
  67. Separate _logical changes_ into a single patch file.
  68. For example, if your changes include both bug fixes and performance
  69. enhancements for a single driver, separate those changes into two
  70. or more patches. If your changes include an API update, and a new
  71. driver which uses that new API, separate those into two patches.
  72. On the other hand, if you make a single change to numerous files,
  73. group those changes into a single patch. Thus a single logical change
  74. is contained within a single patch.
  75. If one patch depends on another patch in order for a change to be
  76. complete, that is OK. Simply note "this patch depends on patch X"
  77. in your patch description.
  78. If you cannot condense your patch set into a smaller set of patches,
  79. then only post say 15 or so at a time and wait for review and integration.
  80. 4) Select e-mail destination.
  81. Look through the MAINTAINERS file and the source code, and determine
  82. if your change applies to a specific subsystem of the kernel, with
  83. an assigned maintainer. If so, e-mail that person.
  84. If no maintainer is listed, or the maintainer does not respond, send
  85. your patch to the primary Linux kernel developer's mailing list,
  86. linux-kernel@vger.kernel.org. Most kernel developers monitor this
  87. e-mail list, and can comment on your changes.
  88. Do not send more than 15 patches at once to the vger mailing lists!!!
  89. Linus Torvalds is the final arbiter of all changes accepted into the
  90. Linux kernel. His e-mail address is <torvalds@linux-foundation.org>.
  91. He gets a lot of e-mail, so typically you should do your best to -avoid-
  92. sending him e-mail.
  93. Patches which are bug fixes, are "obvious" changes, or similarly
  94. require little discussion should be sent or CC'd to Linus. Patches
  95. which require discussion or do not have a clear advantage should
  96. usually be sent first to linux-kernel. Only after the patch is
  97. discussed should the patch then be submitted to Linus.
  98. 5) Select your CC (e-mail carbon copy) list.
  99. Unless you have a reason NOT to do so, CC linux-kernel@vger.kernel.org.
  100. Other kernel developers besides Linus need to be aware of your change,
  101. so that they may comment on it and offer code review and suggestions.
  102. linux-kernel is the primary Linux kernel developer mailing list.
  103. Other mailing lists are available for specific subsystems, such as
  104. USB, framebuffer devices, the VFS, the SCSI subsystem, etc. See the
  105. MAINTAINERS file for a mailing list that relates specifically to
  106. your change.
  107. Majordomo lists of VGER.KERNEL.ORG at:
  108. <http://vger.kernel.org/vger-lists.html>
  109. If changes affect userland-kernel interfaces, please send
  110. the MAN-PAGES maintainer (as listed in the MAINTAINERS file)
  111. a man-pages patch, or at least a notification of the change,
  112. so that some information makes its way into the manual pages.
  113. Even if the maintainer did not respond in step #4, make sure to ALWAYS
  114. copy the maintainer when you change their code.
  115. For small patches you may want to CC the Trivial Patch Monkey
  116. trivial@kernel.org managed by Adrian Bunk; which collects "trivial"
  117. patches. Trivial patches must qualify for one of the following rules:
  118. Spelling fixes in documentation
  119. Spelling fixes which could break grep(1)
  120. Warning fixes (cluttering with useless warnings is bad)
  121. Compilation fixes (only if they are actually correct)
  122. Runtime fixes (only if they actually fix things)
  123. Removing use of deprecated functions/macros (eg. check_region)
  124. Contact detail and documentation fixes
  125. Non-portable code replaced by portable code (even in arch-specific,
  126. since people copy, as long as it's trivial)
  127. Any fix by the author/maintainer of the file (ie. patch monkey
  128. in re-transmission mode)
  129. URL: <http://www.kernel.org/pub/linux/kernel/people/bunk/trivial/>
  130. 6) No MIME, no links, no compression, no attachments. Just plain text.
  131. Linus and other kernel developers need to be able to read and comment
  132. on the changes you are submitting. It is important for a kernel
  133. developer to be able to "quote" your changes, using standard e-mail
  134. tools, so that they may comment on specific portions of your code.
  135. For this reason, all patches should be submitting e-mail "inline".
  136. WARNING: Be wary of your editor's word-wrap corrupting your patch,
  137. if you choose to cut-n-paste your patch.
  138. Do not attach the patch as a MIME attachment, compressed or not.
  139. Many popular e-mail applications will not always transmit a MIME
  140. attachment as plain text, making it impossible to comment on your
  141. code. A MIME attachment also takes Linus a bit more time to process,
  142. decreasing the likelihood of your MIME-attached change being accepted.
  143. Exception: If your mailer is mangling patches then someone may ask
  144. you to re-send them using MIME.
  145. WARNING: Some mailers like Mozilla send your messages with
  146. ---- message header ----
  147. Content-Type: text/plain; charset=us-ascii; format=flowed
  148. ---- message header ----
  149. The problem is that "format=flowed" makes some of the mailers
  150. on receiving side to replace TABs with spaces and do similar
  151. changes. Thus the patches from you can look corrupted.
  152. To fix this just make your mozilla defaults/pref/mailnews.js file to look like:
  153. pref("mailnews.send_plaintext_flowed", false); // RFC 2646=======
  154. pref("mailnews.display.disable_format_flowed_support", true);
  155. 7) E-mail size.
  156. When sending patches to Linus, always follow step #6.
  157. Large changes are not appropriate for mailing lists, and some
  158. maintainers. If your patch, uncompressed, exceeds 40 kB in size,
  159. it is preferred that you store your patch on an Internet-accessible
  160. server, and provide instead a URL (link) pointing to your patch.
  161. 8) Name your kernel version.
  162. It is important to note, either in the subject line or in the patch
  163. description, the kernel version to which this patch applies.
  164. If the patch does not apply cleanly to the latest kernel version,
  165. Linus will not apply it.
  166. 9) Don't get discouraged. Re-submit.
  167. After you have submitted your change, be patient and wait. If Linus
  168. likes your change and applies it, it will appear in the next version
  169. of the kernel that he releases.
  170. However, if your change doesn't appear in the next version of the
  171. kernel, there could be any number of reasons. It's YOUR job to
  172. narrow down those reasons, correct what was wrong, and submit your
  173. updated change.
  174. It is quite common for Linus to "drop" your patch without comment.
  175. That's the nature of the system. If he drops your patch, it could be
  176. due to
  177. * Your patch did not apply cleanly to the latest kernel version.
  178. * Your patch was not sufficiently discussed on linux-kernel.
  179. * A style issue (see section 2).
  180. * An e-mail formatting issue (re-read this section).
  181. * A technical problem with your change.
  182. * He gets tons of e-mail, and yours got lost in the shuffle.
  183. * You are being annoying.
  184. When in doubt, solicit comments on linux-kernel mailing list.
  185. 10) Include PATCH in the subject
  186. Due to high e-mail traffic to Linus, and to linux-kernel, it is common
  187. convention to prefix your subject line with [PATCH]. This lets Linus
  188. and other kernel developers more easily distinguish patches from other
  189. e-mail discussions.
  190. 11) Sign your work
  191. To improve tracking of who did what, especially with patches that can
  192. percolate to their final resting place in the kernel through several
  193. layers of maintainers, we've introduced a "sign-off" procedure on
  194. patches that are being emailed around.
  195. The sign-off is a simple line at the end of the explanation for the
  196. patch, which certifies that you wrote it or otherwise have the right to
  197. pass it on as a open-source patch. The rules are pretty simple: if you
  198. can certify the below:
  199. Developer's Certificate of Origin 1.1
  200. By making a contribution to this project, I certify that:
  201. (a) The contribution was created in whole or in part by me and I
  202. have the right to submit it under the open source license
  203. indicated in the file; or
  204. (b) The contribution is based upon previous work that, to the best
  205. of my knowledge, is covered under an appropriate open source
  206. license and I have the right under that license to submit that
  207. work with modifications, whether created in whole or in part
  208. by me, under the same open source license (unless I am
  209. permitted to submit under a different license), as indicated
  210. in the file; or
  211. (c) The contribution was provided directly to me by some other
  212. person who certified (a), (b) or (c) and I have not modified
  213. it.
  214. (d) I understand and agree that this project and the contribution
  215. are public and that a record of the contribution (including all
  216. personal information I submit with it, including my sign-off) is
  217. maintained indefinitely and may be redistributed consistent with
  218. this project or the open source license(s) involved.
  219. then you just add a line saying
  220. Signed-off-by: Random J Developer <random@developer.example.org>
  221. using your real name (sorry, no pseudonyms or anonymous contributions.)
  222. Some people also put extra tags at the end. They'll just be ignored for
  223. now, but you can do this to mark internal company procedures or just
  224. point out some special detail about the sign-off.
  225. 12) The canonical patch format
  226. The canonical patch subject line is:
  227. Subject: [PATCH 001/123] subsystem: summary phrase
  228. The canonical patch message body contains the following:
  229. - A "from" line specifying the patch author.
  230. - An empty line.
  231. - The body of the explanation, which will be copied to the
  232. permanent changelog to describe this patch.
  233. - The "Signed-off-by:" lines, described above, which will
  234. also go in the changelog.
  235. - A marker line containing simply "---".
  236. - Any additional comments not suitable for the changelog.
  237. - The actual patch (diff output).
  238. The Subject line format makes it very easy to sort the emails
  239. alphabetically by subject line - pretty much any email reader will
  240. support that - since because the sequence number is zero-padded,
  241. the numerical and alphabetic sort is the same.
  242. The "subsystem" in the email's Subject should identify which
  243. area or subsystem of the kernel is being patched.
  244. The "summary phrase" in the email's Subject should concisely
  245. describe the patch which that email contains. The "summary
  246. phrase" should not be a filename. Do not use the same "summary
  247. phrase" for every patch in a whole patch series.
  248. Bear in mind that the "summary phrase" of your email becomes
  249. a globally-unique identifier for that patch. It propagates
  250. all the way into the git changelog. The "summary phrase" may
  251. later be used in developer discussions which refer to the patch.
  252. People will want to google for the "summary phrase" to read
  253. discussion regarding that patch.
  254. A couple of example Subjects:
  255. Subject: [patch 2/5] ext2: improve scalability of bitmap searching
  256. Subject: [PATCHv2 001/207] x86: fix eflags tracking
  257. The "from" line must be the very first line in the message body,
  258. and has the form:
  259. From: Original Author <author@example.com>
  260. The "from" line specifies who will be credited as the author of the
  261. patch in the permanent changelog. If the "from" line is missing,
  262. then the "From:" line from the email header will be used to determine
  263. the patch author in the changelog.
  264. The explanation body will be committed to the permanent source
  265. changelog, so should make sense to a competent reader who has long
  266. since forgotten the immediate details of the discussion that might
  267. have led to this patch.
  268. The "---" marker line serves the essential purpose of marking for patch
  269. handling tools where the changelog message ends.
  270. One good use for the additional comments after the "---" marker is for
  271. a diffstat, to show what files have changed, and the number of inserted
  272. and deleted lines per file. A diffstat is especially useful on bigger
  273. patches. Other comments relevant only to the moment or the maintainer,
  274. not suitable for the permanent changelog, should also go here.
  275. Use diffstat options "-p 1 -w 70" so that filenames are listed from the
  276. top of the kernel source tree and don't use too much horizontal space
  277. (easily fit in 80 columns, maybe with some indentation).
  278. See more details on the proper patch format in the following
  279. references.
  280. -----------------------------------
  281. SECTION 2 - HINTS, TIPS, AND TRICKS
  282. -----------------------------------
  283. This section lists many of the common "rules" associated with code
  284. submitted to the kernel. There are always exceptions... but you must
  285. have a really good reason for doing so. You could probably call this
  286. section Linus Computer Science 101.
  287. 1) Read Documentation/CodingStyle
  288. Nuff said. If your code deviates too much from this, it is likely
  289. to be rejected without further review, and without comment.
  290. 2) #ifdefs are ugly
  291. Code cluttered with ifdefs is difficult to read and maintain. Don't do
  292. it. Instead, put your ifdefs in a header, and conditionally define
  293. 'static inline' functions, or macros, which are used in the code.
  294. Let the compiler optimize away the "no-op" case.
  295. Simple example, of poor code:
  296. dev = alloc_etherdev (sizeof(struct funky_private));
  297. if (!dev)
  298. return -ENODEV;
  299. #ifdef CONFIG_NET_FUNKINESS
  300. init_funky_net(dev);
  301. #endif
  302. Cleaned-up example:
  303. (in header)
  304. #ifndef CONFIG_NET_FUNKINESS
  305. static inline void init_funky_net (struct net_device *d) {}
  306. #endif
  307. (in the code itself)
  308. dev = alloc_etherdev (sizeof(struct funky_private));
  309. if (!dev)
  310. return -ENODEV;
  311. init_funky_net(dev);
  312. 3) 'static inline' is better than a macro
  313. Static inline functions are greatly preferred over macros.
  314. They provide type safety, have no length limitations, no formatting
  315. limitations, and under gcc they are as cheap as macros.
  316. Macros should only be used for cases where a static inline is clearly
  317. suboptimal [there a few, isolated cases of this in fast paths],
  318. or where it is impossible to use a static inline function [such as
  319. string-izing].
  320. 'static inline' is preferred over 'static __inline__', 'extern inline',
  321. and 'extern __inline__'.
  322. 4) Don't over-design.
  323. Don't try to anticipate nebulous future cases which may or may not
  324. be useful: "Make it as simple as you can, and no simpler."
  325. ----------------------
  326. SECTION 3 - REFERENCES
  327. ----------------------
  328. Andrew Morton, "The perfect patch" (tpp).
  329. <http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt>
  330. Jeff Garzik, "Linux kernel patch submission format".
  331. <http://linux.yyz.us/patch-format.html>
  332. Greg Kroah-Hartman, "How to piss off a kernel subsystem maintainer".
  333. <http://www.kroah.com/log/2005/03/31/>
  334. <http://www.kroah.com/log/2005/07/08/>
  335. <http://www.kroah.com/log/2005/10/19/>
  336. <http://www.kroah.com/log/2006/01/11/>
  337. NO!!!! No more huge patch bombs to linux-kernel@vger.kernel.org people!
  338. <http://marc.theaimsgroup.com/?l=linux-kernel&m=112112749912944&w=2>
  339. Kernel Documentation/CodingStyle:
  340. <http://sosdg.org/~coywolf/lxr/source/Documentation/CodingStyle>
  341. Linus Torvalds's mail on the canonical patch format:
  342. <http://lkml.org/lkml/2005/4/7/183>
  343. --