1 2 How to Get Your Change Into the Linux Kernel 3 or 4 Care And Operation Of Your Linus Torvalds 5 6 7 8For a person or company who wishes to submit a change to the Linux 9kernel, the process can sometimes be daunting if you're not familiar 10with "the system." This text is a collection of suggestions which 11can greatly increase the chances of your change being accepted. 12 13If you are submitting a driver, also read Documentation/SubmittingDrivers. 14 15 16 17-------------------------------------------- 18SECTION 1 - CREATING AND SENDING YOUR CHANGE 19-------------------------------------------- 20 21 22 231) "diff -up" 24------------ 25 26Use "diff -up" or "diff -uprN" to create patches. 27 28All changes to the Linux kernel occur in the form of patches, as 29generated by diff(1). When creating your patch, make sure to create it 30in "unified diff" format, as supplied by the '-u' argument to diff(1). 31Also, please use the '-p' argument which shows which C function each 32change is in - that makes the resultant diff a lot easier to read. 33Patches should be based in the root kernel source directory, 34not in any lower subdirectory. 35 36To create a patch for a single file, it is often sufficient to do: 37 38 SRCTREE= linux-2.4 39 MYFILE= drivers/net/mydriver.c 40 41 cd $SRCTREE 42 cp $MYFILE $MYFILE.orig 43 vi $MYFILE # make your change 44 cd .. 45 diff -up $SRCTREE/$MYFILE{.orig,} > /tmp/patch 46 47To create a patch for multiple files, you should unpack a "vanilla", 48or unmodified kernel source tree, and generate a diff against your 49own source tree. For example: 50 51 MYSRC= /devel/linux-2.4 52 53 tar xvfz linux-2.4.0-test11.tar.gz 54 mv linux linux-vanilla 55 wget http://www.moses.uklinux.net/patches/dontdiff 56 diff -uprN -X dontdiff linux-vanilla $MYSRC > /tmp/patch 57 rm -f dontdiff 58 59"dontdiff" is a list of files which are generated by the kernel during 60the build process, and should be ignored in any diff(1)-generated 61patch. dontdiff is maintained by Tigran Aivazian <tigran@veritas.com> 62 63Make sure your patch does not include any extra files which do not 64belong in a patch submission. Make sure to review your patch -after- 65generated it with diff(1), to ensure accuracy. 66 67If your changes produce a lot of deltas, you may want to look into 68splitting them into individual patches which modify things in 69logical stages, this will facilitate easier reviewing by other 70kernel developers, very important if you want your patch accepted. 71There are a number of scripts which can aid in this; 72 73Quilt: 74http://savannah.nongnu.org/projects/quilt 75 76Randy Dunlap's patch scripts: 77http://developer.osdl.org/rddunlap/scripts/patching-scripts.tgz 78 79Andrew Morton's patch scripts: 80http://www.zip.com.au/~akpm/linux/patches/patch-scripts-0.16 81 822) Describe your changes. 83 84Describe the technical detail of the change(s) your patch includes. 85 86Be as specific as possible. The WORST descriptions possible include 87things like "update driver X", "bug fix for driver X", or "this patch 88includes updates for subsystem X. Please apply." 89 90If your description starts to get long, that's a sign that you probably 91need to split up your patch. See #3, next. 92 93 94 953) Separate your changes. 96 97Separate each logical change into its own patch. 98 99For example, if your changes include both bug fixes and performance 100enhancements for a single driver, separate those changes into two 101or more patches. If your changes include an API update, and a new 102driver which uses that new API, separate those into two patches. 103 104On the other hand, if you make a single change to numerous files, 105group those changes into a single patch. Thus a single logical change 106is contained within a single patch. 107 108If one patch depends on another patch in order for a change to be 109complete, that is OK. Simply note "this patch depends on patch X" 110in your patch description. 111 112 1134) Select e-mail destination. 114 115Look through the MAINTAINERS file and the source code, and determine 116if your change applies to a specific subsystem of the kernel, with 117an assigned maintainer. If so, e-mail that person. 118 119If no maintainer is listed, or the maintainer does not respond, send 120your patch to the primary Linux kernel developer's mailing list, 121linux-kernel@vger.kernel.org. Most kernel developers monitor this 122e-mail list, and can comment on your changes. 123 124Linus Torvalds is the final arbiter of all changes accepted into the 125Linux kernel. His e-mail address is <torvalds@osdl.org>. He gets 126a lot of e-mail, so typically you should do your best to -avoid- sending 127him e-mail. 128 129Patches which are bug fixes, are "obvious" changes, or similarly 130require little discussion should be sent or CC'd to Linus. Patches 131which require discussion or do not have a clear advantage should 132usually be sent first to linux-kernel. Only after the patch is 133discussed should the patch then be submitted to Linus. 134 135 136 1375) Select your CC (e-mail carbon copy) list. 138 139Unless you have a reason NOT to do so, CC linux-kernel@vger.kernel.org. 140 141Other kernel developers besides Linus need to be aware of your change, 142so that they may comment on it and offer code review and suggestions. 143linux-kernel is the primary Linux kernel developer mailing list. 144Other mailing lists are available for specific subsystems, such as 145USB, framebuffer devices, the VFS, the SCSI subsystem, etc. See the 146MAINTAINERS file for a mailing list that relates specifically to 147your change. 148 149Even if the maintainer did not respond in step #4, make sure to ALWAYS 150copy the maintainer when you change their code. 151 152For small patches you may want to CC the Trivial Patch Monkey 153trivial@rustcorp.com.au set up by Rusty Russell; which collects "trivial" 154patches. Trivial patches must qualify for one of the following rules: 155 Spelling fixes in documentation 156 Spelling fixes which could break grep(1). 157 Warning fixes (cluttering with useless warnings is bad) 158 Compilation fixes (only if they are actually correct) 159 Runtime fixes (only if they actually fix things) 160 Removing use of deprecated functions/macros (eg. check_region). 161 Contact detail and documentation fixes 162 Non-portable code replaced by portable code (even in arch-specific, 163 since people copy, as long as it's trivial) 164 Any fix by the author/maintainer of the file. (ie. patch monkey 165 in re-transmission mode) 166 167 168 1696) No MIME, no links, no compression, no attachments. Just plain text. 170 171Linus and other kernel developers need to be able to read and comment 172on the changes you are submitting. It is important for a kernel 173developer to be able to "quote" your changes, using standard e-mail 174tools, so that they may comment on specific portions of your code. 175 176For this reason, all patches should be submitting e-mail "inline". 177WARNING: Be wary of your editor's word-wrap corrupting your patch, 178if you choose to cut-n-paste your patch. 179 180Do not attach the patch as a MIME attachment, compressed or not. 181Many popular e-mail applications will not always transmit a MIME 182attachment as plain text, making it impossible to comment on your 183code. A MIME attachment also takes Linus a bit more time to process, 184decreasing the likelihood of your MIME-attached change being accepted. 185 186Exception: If your mailer is mangling patches then someone may ask 187you to re-send them using MIME. 188 189 190 1917) E-mail size. 192 193When sending patches to Linus, always follow step #6. 194 195Large changes are not appropriate for mailing lists, and some 196maintainers. If your patch, uncompressed, exceeds 40 kB in size, 197it is preferred that you store your patch on an Internet-accessible 198server, and provide instead a URL (link) pointing to your patch. 199 200 201 2028) Name your kernel version. 203 204It is important to note, either in the subject line or in the patch 205description, the kernel version to which this patch applies. 206 207If the patch does not apply cleanly to the latest kernel version, 208Linus will not apply it. 209 210 211 2129) Don't get discouraged. Re-submit. 213 214After you have submitted your change, be patient and wait. If Linus 215likes your change and applies it, it will appear in the next version 216of the kernel that he releases. 217 218However, if your change doesn't appear in the next version of the 219kernel, there could be any number of reasons. It's YOUR job to 220narrow down those reasons, correct what was wrong, and submit your 221updated change. 222 223It is quite common for Linus to "drop" your patch without comment. 224That's the nature of the system. If he drops your patch, it could be 225due to 226* Your patch did not apply cleanly to the latest kernel version 227* Your patch was not sufficiently discussed on linux-kernel. 228* A style issue (see section 2), 229* An e-mail formatting issue (re-read this section) 230* A technical problem with your change 231* He gets tons of e-mail, and yours got lost in the shuffle 232* You are being annoying (See Figure 1) 233 234When in doubt, solicit comments on linux-kernel mailing list. 235 236 237 23810) Include PATCH in the subject 239 240Due to high e-mail traffic to Linus, and to linux-kernel, it is common 241convention to prefix your subject line with [PATCH]. This lets Linus 242and other kernel developers more easily distinguish patches from other 243e-mail discussions. 244 245 246 24711) Sign your work 248 249To improve tracking of who did what, especially with patches that can 250percolate to their final resting place in the kernel through several 251layers of maintainers, we've introduced a "sign-off" procedure on 252patches that are being emailed around. 253 254The sign-off is a simple line at the end of the explanation for the 255patch, which certifies that you wrote it or otherwise have the right to 256pass it on as a open-source patch. The rules are pretty simple: if you 257can certify the below: 258 259 Developer's Certificate of Origin 1.1 260 261 By making a contribution to this project, I certify that: 262 263 (a) The contribution was created in whole or in part by me and I 264 have the right to submit it under the open source license 265 indicated in the file; or 266 267 (b) The contribution is based upon previous work that, to the best 268 of my knowledge, is covered under an appropriate open source 269 license and I have the right under that license to submit that 270 work with modifications, whether created in whole or in part 271 by me, under the same open source license (unless I am 272 permitted to submit under a different license), as indicated 273 in the file; or 274 275 (c) The contribution was provided directly to me by some other 276 person who certified (a), (b) or (c) and I have not modified 277 it. 278 279 (d) I understand and agree that this project and the contribution 280 are public and that a record of the contribution (including all 281 personal information I submit with it, including my sign-off) is 282 maintained indefinitely and may be redistributed consistent with 283 this project or the open source license(s) involved. 284 285then you just add a line saying 286 287 Signed-off-by: Random J Developer <random@developer.example.org> 288 289Some people also put extra tags at the end. They'll just be ignored for 290now, but you can do this to mark internal company procedures or just 291point out some special detail about the sign-off. 292 293 294----------------------------------- 295SECTION 2 - HINTS, TIPS, AND TRICKS 296----------------------------------- 297 298This section lists many of the common "rules" associated with code 299submitted to the kernel. There are always exceptions... but you must 300have a really good reason for doing so. You could probably call this 301section Linus Computer Science 101. 302 303 304 3051) Read Documentation/CodingStyle 306 307Nuff said. If your code deviates too much from this, it is likely 308to be rejected without further review, and without comment. 309 310 311 3122) #ifdefs are ugly 313 314Code cluttered with ifdefs is difficult to read and maintain. Don't do 315it. Instead, put your ifdefs in a header, and conditionally define 316'static inline' functions, or macros, which are used in the code. 317Let the compiler optimize away the "no-op" case. 318 319Simple example, of poor code: 320 321 dev = alloc_etherdev (sizeof(struct funky_private)); 322 if (!dev) 323 return -ENODEV; 324 #ifdef CONFIG_NET_FUNKINESS 325 init_funky_net(dev); 326 #endif 327 328Cleaned-up example: 329 330(in header) 331 #ifndef CONFIG_NET_FUNKINESS 332 static inline void init_funky_net (struct net_device *d) {} 333 #endif 334 335(in the code itself) 336 dev = alloc_etherdev (sizeof(struct funky_private)); 337 if (!dev) 338 return -ENODEV; 339 init_funky_net(dev); 340 341 342 3433) 'static inline' is better than a macro 344 345Static inline functions are greatly preferred over macros. 346They provide type safety, have no length limitations, no formatting 347limitations, and under gcc they are as cheap as macros. 348 349Macros should only be used for cases where a static inline is clearly 350suboptimal [there a few, isolated cases of this in fast paths], 351or where it is impossible to use a static inline function [such as 352string-izing]. 353 354'static inline' is preferred over 'static __inline__', 'extern inline', 355and 'extern __inline__'. 356 357 358 3594) Don't over-design. 360 361Don't try to anticipate nebulous future cases which may or may not 362be useful: "Make it as simple as you can, and no simpler" 363 364 365 366