linux/Documentation/SubmittingPatches
<<
>>
Prefs
   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
  13Read Documentation/SubmitChecklist for a list of items to check
  14before submitting code.  If you are submitting a driver, also read
  15Documentation/SubmittingDrivers.
  16
  17Many of these steps describe the default behavior of the git version
  18control system; if you use git to prepare your patches, you'll find much
  19of the mechanical work done for you, though you'll still need to prepare
  20and document a sensible set of patches.
  21
  22--------------------------------------------
  23SECTION 1 - CREATING AND SENDING YOUR CHANGE
  24--------------------------------------------
  25
  26
  27
  281) "diff -up"
  29------------
  30
  31Use "diff -up" or "diff -uprN" to create patches.  git generates patches
  32in this form by default; if you're using git, you can skip this section
  33entirely.
  34
  35All changes to the Linux kernel occur in the form of patches, as
  36generated by diff(1).  When creating your patch, make sure to create it
  37in "unified diff" format, as supplied by the '-u' argument to diff(1).
  38Also, please use the '-p' argument which shows which C function each
  39change is in - that makes the resultant diff a lot easier to read.
  40Patches should be based in the root kernel source directory,
  41not in any lower subdirectory.
  42
  43To create a patch for a single file, it is often sufficient to do:
  44
  45        SRCTREE= linux-2.6
  46        MYFILE=  drivers/net/mydriver.c
  47
  48        cd $SRCTREE
  49        cp $MYFILE $MYFILE.orig
  50        vi $MYFILE      # make your change
  51        cd ..
  52        diff -up $SRCTREE/$MYFILE{.orig,} > /tmp/patch
  53
  54To create a patch for multiple files, you should unpack a "vanilla",
  55or unmodified kernel source tree, and generate a diff against your
  56own source tree.  For example:
  57
  58        MYSRC= /devel/linux-2.6
  59
  60        tar xvfz linux-2.6.12.tar.gz
  61        mv linux-2.6.12 linux-2.6.12-vanilla
  62        diff -uprN -X linux-2.6.12-vanilla/Documentation/dontdiff \
  63                linux-2.6.12-vanilla $MYSRC > /tmp/patch
  64
  65"dontdiff" is a list of files which are generated by the kernel during
  66the build process, and should be ignored in any diff(1)-generated
  67patch.  The "dontdiff" file is included in the kernel tree in
  682.6.12 and later.
  69
  70Make sure your patch does not include any extra files which do not
  71belong in a patch submission.  Make sure to review your patch -after-
  72generated it with diff(1), to ensure accuracy.
  73
  74If your changes produce a lot of deltas, you need to split them into
  75individual patches which modify things in logical stages; see section
  76#3.  This will facilitate easier reviewing by other kernel developers,
  77very important if you want your patch accepted.
  78
  79If you're using git, "git rebase -i" can help you with this process.  If
  80you're not using git, quilt <http://savannah.nongnu.org/projects/quilt>
  81is another popular alternative.
  82
  83
  84
  852) Describe your changes.
  86
  87Describe your problem.  Whether your patch is a one-line bug fix or
  885000 lines of a new feature, there must be an underlying problem that
  89motivated you to do this work.  Convince the reviewer that there is a
  90problem worth fixing and that it makes sense for them to read past the
  91first paragraph.
  92
  93Describe user-visible impact.  Straight up crashes and lockups are
  94pretty convincing, but not all bugs are that blatant.  Even if the
  95problem was spotted during code review, describe the impact you think
  96it can have on users.  Keep in mind that the majority of Linux
  97installations run kernels from secondary stable trees or
  98vendor/product-specific trees that cherry-pick only specific patches
  99from upstream, so include anything that could help route your change
 100downstream: provoking circumstances, excerpts from dmesg, crash
 101descriptions, performance regressions, latency spikes, lockups, etc.
 102
 103Quantify optimizations and trade-offs.  If you claim improvements in
 104performance, memory consumption, stack footprint, or binary size,
 105include numbers that back them up.  But also describe non-obvious
 106costs.  Optimizations usually aren't free but trade-offs between CPU,
 107memory, and readability; or, when it comes to heuristics, between
 108different workloads.  Describe the expected downsides of your
 109optimization so that the reviewer can weigh costs against benefits.
 110
 111Once the problem is established, describe what you are actually doing
 112about it in technical detail.  It's important to describe the change
 113in plain English for the reviewer to verify that the code is behaving
 114as you intend it to.
 115
 116The maintainer will thank you if you write your patch description in a
 117form which can be easily pulled into Linux's source code management
 118system, git, as a "commit log".  See #15, below.
 119
 120Solve only one problem per patch.  If your description starts to get
 121long, that's a sign that you probably need to split up your patch.
 122See #3, next.
 123
 124When you submit or resubmit a patch or patch series, include the
 125complete patch description and justification for it.  Don't just
 126say that this is version N of the patch (series).  Don't expect the
 127patch merger to refer back to earlier patch versions or referenced
 128URLs to find the patch description and put that into the patch.
 129I.e., the patch (series) and its description should be self-contained.
 130This benefits both the patch merger(s) and reviewers.  Some reviewers
 131probably didn't even receive earlier versions of the patch.
 132
 133Describe your changes in imperative mood, e.g. "make xyzzy do frotz"
 134instead of "[This patch] makes xyzzy do frotz" or "[I] changed xyzzy
 135to do frotz", as if you are giving orders to the codebase to change
 136its behaviour.
 137
 138If the patch fixes a logged bug entry, refer to that bug entry by
 139number and URL.  If the patch follows from a mailing list discussion,
 140give a URL to the mailing list archive; use the https://lkml.kernel.org/
 141redirector with a Message-Id, to ensure that the links cannot become
 142stale.
 143
 144However, try to make your explanation understandable without external
 145resources.  In addition to giving a URL to a mailing list archive or
 146bug, summarize the relevant points of the discussion that led to the
 147patch as submitted.
 148
 149If you want to refer to a specific commit, don't just refer to the
 150SHA-1 ID of the commit. Please also include the oneline summary of
 151the commit, to make it easier for reviewers to know what it is about.
 152Example:
 153
 154        Commit e21d2170f36602ae2708 ("video: remove unnecessary
 155        platform_set_drvdata()") removed the unnecessary
 156        platform_set_drvdata(), but left the variable "dev" unused,
 157        delete it.
 158
 159If your patch fixes a bug in a specific commit, e.g. you found an issue using
 160git-bisect, please use the 'Fixes:' tag with the first 12 characters of the
 161SHA-1 ID, and the one line summary.
 162Example:
 163
 164        Fixes: e21d2170f366 ("video: remove unnecessary platform_set_drvdata()")
 165
 166The following git-config settings can be used to add a pretty format for
 167outputting the above style in the git log or git show commands
 168
 169        [core]
 170                abbrev = 12
 171        [pretty]
 172                fixes = Fixes: %h (\"%s\")
 173
 1743) Separate your changes.
 175
 176Separate _logical changes_ into a single patch file.
 177
 178For example, if your changes include both bug fixes and performance
 179enhancements for a single driver, separate those changes into two
 180or more patches.  If your changes include an API update, and a new
 181driver which uses that new API, separate those into two patches.
 182
 183On the other hand, if you make a single change to numerous files,
 184group those changes into a single patch.  Thus a single logical change
 185is contained within a single patch.
 186
 187If one patch depends on another patch in order for a change to be
 188complete, that is OK.  Simply note "this patch depends on patch X"
 189in your patch description.
 190
 191If you cannot condense your patch set into a smaller set of patches,
 192then only post say 15 or so at a time and wait for review and integration.
 193
 194
 195
 1964) Style check your changes.
 197
 198Check your patch for basic style violations, details of which can be
 199found in Documentation/CodingStyle.  Failure to do so simply wastes
 200the reviewers time and will get your patch rejected, probably
 201without even being read.
 202
 203At a minimum you should check your patches with the patch style
 204checker prior to submission (scripts/checkpatch.pl).  You should
 205be able to justify all violations that remain in your patch.
 206
 207
 208
 2095) Select e-mail destination.
 210
 211Look through the MAINTAINERS file and the source code, and determine
 212if your change applies to a specific subsystem of the kernel, with
 213an assigned maintainer.  If so, e-mail that person.  The script
 214scripts/get_maintainer.pl can be very useful at this step.
 215
 216If no maintainer is listed, or the maintainer does not respond, send
 217your patch to the primary Linux kernel developer's mailing list,
 218linux-kernel@vger.kernel.org.  Most kernel developers monitor this
 219e-mail list, and can comment on your changes.
 220
 221
 222Do not send more than 15 patches at once to the vger mailing lists!!!
 223
 224
 225Linus Torvalds is the final arbiter of all changes accepted into the
 226Linux kernel.  His e-mail address is <torvalds@linux-foundation.org>. 
 227He gets a lot of e-mail, so typically you should do your best to -avoid-
 228sending him e-mail. 
 229
 230Patches which are bug fixes, are "obvious" changes, or similarly
 231require little discussion should be sent or CC'd to Linus.  Patches
 232which require discussion or do not have a clear advantage should
 233usually be sent first to linux-kernel.  Only after the patch is
 234discussed should the patch then be submitted to Linus.
 235
 236
 237
 2386) Select your CC (e-mail carbon copy) list.
 239
 240Unless you have a reason NOT to do so, CC linux-kernel@vger.kernel.org.
 241
 242Other kernel developers besides Linus need to be aware of your change,
 243so that they may comment on it and offer code review and suggestions.
 244linux-kernel is the primary Linux kernel developer mailing list.
 245Other mailing lists are available for specific subsystems, such as
 246USB, framebuffer devices, the VFS, the SCSI subsystem, etc.  See the
 247MAINTAINERS file for a mailing list that relates specifically to
 248your change.
 249
 250Majordomo lists of VGER.KERNEL.ORG at:
 251        <http://vger.kernel.org/vger-lists.html>
 252
 253If changes affect userland-kernel interfaces, please send
 254the MAN-PAGES maintainer (as listed in the MAINTAINERS file)
 255a man-pages patch, or at least a notification of the change,
 256so that some information makes its way into the manual pages.
 257
 258Even if the maintainer did not respond in step #5, make sure to ALWAYS
 259copy the maintainer when you change their code.
 260
 261For small patches you may want to CC the Trivial Patch Monkey
 262trivial@kernel.org which collects "trivial" patches. Have a look
 263into the MAINTAINERS file for its current manager.
 264Trivial patches must qualify for one of the following rules:
 265 Spelling fixes in documentation
 266 Spelling fixes which could break grep(1)
 267 Warning fixes (cluttering with useless warnings is bad)
 268 Compilation fixes (only if they are actually correct)
 269 Runtime fixes (only if they actually fix things)
 270 Removing use of deprecated functions/macros (eg. check_region)
 271 Contact detail and documentation fixes
 272 Non-portable code replaced by portable code (even in arch-specific,
 273 since people copy, as long as it's trivial)
 274 Any fix by the author/maintainer of the file (ie. patch monkey
 275 in re-transmission mode)
 276
 277
 278
 2797) No MIME, no links, no compression, no attachments.  Just plain text.
 280
 281Linus and other kernel developers need to be able to read and comment
 282on the changes you are submitting.  It is important for a kernel
 283developer to be able to "quote" your changes, using standard e-mail
 284tools, so that they may comment on specific portions of your code.
 285
 286For this reason, all patches should be submitting e-mail "inline".
 287WARNING:  Be wary of your editor's word-wrap corrupting your patch,
 288if you choose to cut-n-paste your patch.
 289
 290Do not attach the patch as a MIME attachment, compressed or not.
 291Many popular e-mail applications will not always transmit a MIME
 292attachment as plain text, making it impossible to comment on your
 293code.  A MIME attachment also takes Linus a bit more time to process,
 294decreasing the likelihood of your MIME-attached change being accepted.
 295
 296Exception:  If your mailer is mangling patches then someone may ask
 297you to re-send them using MIME.
 298
 299See Documentation/email-clients.txt for hints about configuring
 300your e-mail client so that it sends your patches untouched.
 301
 3028) E-mail size.
 303
 304When sending patches to Linus, always follow step #7.
 305
 306Large changes are not appropriate for mailing lists, and some
 307maintainers.  If your patch, uncompressed, exceeds 300 kB in size,
 308it is preferred that you store your patch on an Internet-accessible
 309server, and provide instead a URL (link) pointing to your patch.
 310
 311
 312
 3139) Name your kernel version.
 314
 315It is important to note, either in the subject line or in the patch
 316description, the kernel version to which this patch applies.
 317
 318If the patch does not apply cleanly to the latest kernel version,
 319Linus will not apply it.
 320
 321
 322
 32310) Don't get discouraged.  Re-submit.
 324
 325After you have submitted your change, be patient and wait.  If Linus
 326likes your change and applies it, it will appear in the next version
 327of the kernel that he releases.
 328
 329However, if your change doesn't appear in the next version of the
 330kernel, there could be any number of reasons.  It's YOUR job to
 331narrow down those reasons, correct what was wrong, and submit your
 332updated change.
 333
 334It is quite common for Linus to "drop" your patch without comment.
 335That's the nature of the system.  If he drops your patch, it could be
 336due to
 337* Your patch did not apply cleanly to the latest kernel version.
 338* Your patch was not sufficiently discussed on linux-kernel.
 339* A style issue (see section 2).
 340* An e-mail formatting issue (re-read this section).
 341* A technical problem with your change.
 342* He gets tons of e-mail, and yours got lost in the shuffle.
 343* You are being annoying.
 344
 345When in doubt, solicit comments on linux-kernel mailing list.
 346
 347
 348
 34911) Include PATCH in the subject
 350
 351Due to high e-mail traffic to Linus, and to linux-kernel, it is common
 352convention to prefix your subject line with [PATCH].  This lets Linus
 353and other kernel developers more easily distinguish patches from other
 354e-mail discussions.
 355
 356
 357
 35812) Sign your work
 359
 360To improve tracking of who did what, especially with patches that can
 361percolate to their final resting place in the kernel through several
 362layers of maintainers, we've introduced a "sign-off" procedure on
 363patches that are being emailed around.
 364
 365The sign-off is a simple line at the end of the explanation for the
 366patch, which certifies that you wrote it or otherwise have the right to
 367pass it on as an open-source patch.  The rules are pretty simple: if you
 368can certify the below:
 369
 370        Developer's Certificate of Origin 1.1
 371
 372        By making a contribution to this project, I certify that:
 373
 374        (a) The contribution was created in whole or in part by me and I
 375            have the right to submit it under the open source license
 376            indicated in the file; or
 377
 378        (b) The contribution is based upon previous work that, to the best
 379            of my knowledge, is covered under an appropriate open source
 380            license and I have the right under that license to submit that
 381            work with modifications, whether created in whole or in part
 382            by me, under the same open source license (unless I am
 383            permitted to submit under a different license), as indicated
 384            in the file; or
 385
 386        (c) The contribution was provided directly to me by some other
 387            person who certified (a), (b) or (c) and I have not modified
 388            it.
 389
 390        (d) I understand and agree that this project and the contribution
 391            are public and that a record of the contribution (including all
 392            personal information I submit with it, including my sign-off) is
 393            maintained indefinitely and may be redistributed consistent with
 394            this project or the open source license(s) involved.
 395
 396then you just add a line saying
 397
 398        Signed-off-by: Random J Developer <random@developer.example.org>
 399
 400using your real name (sorry, no pseudonyms or anonymous contributions.)
 401
 402Some people also put extra tags at the end.  They'll just be ignored for
 403now, but you can do this to mark internal company procedures or just
 404point out some special detail about the sign-off. 
 405
 406If you are a subsystem or branch maintainer, sometimes you need to slightly
 407modify patches you receive in order to merge them, because the code is not
 408exactly the same in your tree and the submitters'. If you stick strictly to
 409rule (c), you should ask the submitter to rediff, but this is a totally
 410counter-productive waste of time and energy. Rule (b) allows you to adjust
 411the code, but then it is very impolite to change one submitter's code and
 412make him endorse your bugs. To solve this problem, it is recommended that
 413you add a line between the last Signed-off-by header and yours, indicating
 414the nature of your changes. While there is nothing mandatory about this, it
 415seems like prepending the description with your mail and/or name, all
 416enclosed in square brackets, is noticeable enough to make it obvious that
 417you are responsible for last-minute changes. Example :
 418
 419        Signed-off-by: Random J Developer <random@developer.example.org>
 420        [lucky@maintainer.example.org: struct foo moved from foo.c to foo.h]
 421        Signed-off-by: Lucky K Maintainer <lucky@maintainer.example.org>
 422
 423This practice is particularly helpful if you maintain a stable branch and
 424want at the same time to credit the author, track changes, merge the fix,
 425and protect the submitter from complaints. Note that under no circumstances
 426can you change the author's identity (the From header), as it is the one
 427which appears in the changelog.
 428
 429Special note to back-porters: It seems to be a common and useful practice
 430to insert an indication of the origin of a patch at the top of the commit
 431message (just after the subject line) to facilitate tracking. For instance,
 432here's what we see in 2.6-stable :
 433
 434    Date:   Tue May 13 19:10:30 2008 +0000
 435
 436        SCSI: libiscsi regression in 2.6.25: fix nop timer handling
 437
 438        commit 4cf1043593db6a337f10e006c23c69e5fc93e722 upstream
 439
 440And here's what appears in 2.4 :
 441
 442    Date:   Tue May 13 22:12:27 2008 +0200
 443
 444        wireless, airo: waitbusy() won't delay
 445
 446        [backport of 2.6 commit b7acbdfbd1f277c1eb23f344f899cfa4cd0bf36a]
 447
 448Whatever the format, this information provides a valuable help to people
 449tracking your trees, and to people trying to trouble-shoot bugs in your
 450tree.
 451
 452
 45313) When to use Acked-by: and Cc:
 454
 455The Signed-off-by: tag indicates that the signer was involved in the
 456development of the patch, or that he/she was in the patch's delivery path.
 457
 458If a person was not directly involved in the preparation or handling of a
 459patch but wishes to signify and record their approval of it then they can
 460arrange to have an Acked-by: line added to the patch's changelog.
 461
 462Acked-by: is often used by the maintainer of the affected code when that
 463maintainer neither contributed to nor forwarded the patch.
 464
 465Acked-by: is not as formal as Signed-off-by:.  It is a record that the acker
 466has at least reviewed the patch and has indicated acceptance.  Hence patch
 467mergers will sometimes manually convert an acker's "yep, looks good to me"
 468into an Acked-by:.
 469
 470Acked-by: does not necessarily indicate acknowledgement of the entire patch.
 471For example, if a patch affects multiple subsystems and has an Acked-by: from
 472one subsystem maintainer then this usually indicates acknowledgement of just
 473the part which affects that maintainer's code.  Judgement should be used here.
 474When in doubt people should refer to the original discussion in the mailing
 475list archives.
 476
 477If a person has had the opportunity to comment on a patch, but has not
 478provided such comments, you may optionally add a "Cc:" tag to the patch.
 479This is the only tag which might be added without an explicit action by the
 480person it names.  This tag documents that potentially interested parties
 481have been included in the discussion
 482
 483
 48414) Using Reported-by:, Tested-by:, Reviewed-by:, Suggested-by: and Fixes:
 485
 486The Reported-by tag gives credit to people who find bugs and report them and it
 487hopefully inspires them to help us again in the future.  Please note that if
 488the bug was reported in private, then ask for permission first before using the
 489Reported-by tag.
 490
 491A Tested-by: tag indicates that the patch has been successfully tested (in
 492some environment) by the person named.  This tag informs maintainers that
 493some testing has been performed, provides a means to locate testers for
 494future patches, and ensures credit for the testers.
 495
 496Reviewed-by:, instead, indicates that the patch has been reviewed and found
 497acceptable according to the Reviewer's Statement:
 498
 499        Reviewer's statement of oversight
 500
 501        By offering my Reviewed-by: tag, I state that:
 502
 503         (a) I have carried out a technical review of this patch to
 504             evaluate its appropriateness and readiness for inclusion into
 505             the mainline kernel.
 506
 507         (b) Any problems, concerns, or questions relating to the patch
 508             have been communicated back to the submitter.  I am satisfied
 509             with the submitter's response to my comments.
 510
 511         (c) While there may be things that could be improved with this
 512             submission, I believe that it is, at this time, (1) a
 513             worthwhile modification to the kernel, and (2) free of known
 514             issues which would argue against its inclusion.
 515
 516         (d) While I have reviewed the patch and believe it to be sound, I
 517             do not (unless explicitly stated elsewhere) make any
 518             warranties or guarantees that it will achieve its stated
 519             purpose or function properly in any given situation.
 520
 521A Reviewed-by tag is a statement of opinion that the patch is an
 522appropriate modification of the kernel without any remaining serious
 523technical issues.  Any interested reviewer (who has done the work) can
 524offer a Reviewed-by tag for a patch.  This tag serves to give credit to
 525reviewers and to inform maintainers of the degree of review which has been
 526done on the patch.  Reviewed-by: tags, when supplied by reviewers known to
 527understand the subject area and to perform thorough reviews, will normally
 528increase the likelihood of your patch getting into the kernel.
 529
 530A Suggested-by: tag indicates that the patch idea is suggested by the person
 531named and ensures credit to the person for the idea. Please note that this
 532tag should not be added without the reporter's permission, especially if the
 533idea was not posted in a public forum. That said, if we diligently credit our
 534idea reporters, they will, hopefully, be inspired to help us again in the
 535future.
 536
 537A Fixes: tag indicates that the patch fixes an issue in a previous commit. It
 538is used to make it easy to determine where a bug originated, which can help
 539review a bug fix. This tag also assists the stable kernel team in determining
 540which stable kernel versions should receive your fix. This is the preferred
 541method for indicating a bug fixed by the patch. See #2 above for more details.
 542
 543
 54415) The canonical patch format
 545
 546The canonical patch subject line is:
 547
 548    Subject: [PATCH 001/123] subsystem: summary phrase
 549
 550The canonical patch message body contains the following:
 551
 552  - A "from" line specifying the patch author.
 553
 554  - An empty line.
 555
 556  - The body of the explanation, which will be copied to the
 557    permanent changelog to describe this patch.
 558
 559  - The "Signed-off-by:" lines, described above, which will
 560    also go in the changelog.
 561
 562  - A marker line containing simply "---".
 563
 564  - Any additional comments not suitable for the changelog.
 565
 566  - The actual patch (diff output).
 567
 568The Subject line format makes it very easy to sort the emails
 569alphabetically by subject line - pretty much any email reader will
 570support that - since because the sequence number is zero-padded,
 571the numerical and alphabetic sort is the same.
 572
 573The "subsystem" in the email's Subject should identify which
 574area or subsystem of the kernel is being patched.
 575
 576The "summary phrase" in the email's Subject should concisely
 577describe the patch which that email contains.  The "summary
 578phrase" should not be a filename.  Do not use the same "summary
 579phrase" for every patch in a whole patch series (where a "patch
 580series" is an ordered sequence of multiple, related patches).
 581
 582Bear in mind that the "summary phrase" of your email becomes a
 583globally-unique identifier for that patch.  It propagates all the way
 584into the git changelog.  The "summary phrase" may later be used in
 585developer discussions which refer to the patch.  People will want to
 586google for the "summary phrase" to read discussion regarding that
 587patch.  It will also be the only thing that people may quickly see
 588when, two or three months later, they are going through perhaps
 589thousands of patches using tools such as "gitk" or "git log
 590--oneline".
 591
 592For these reasons, the "summary" must be no more than 70-75
 593characters, and it must describe both what the patch changes, as well
 594as why the patch might be necessary.  It is challenging to be both
 595succinct and descriptive, but that is what a well-written summary
 596should do.
 597
 598The "summary phrase" may be prefixed by tags enclosed in square
 599brackets: "Subject: [PATCH tag] <summary phrase>".  The tags are not
 600considered part of the summary phrase, but describe how the patch
 601should be treated.  Common tags might include a version descriptor if
 602the multiple versions of the patch have been sent out in response to
 603comments (i.e., "v1, v2, v3"), or "RFC" to indicate a request for
 604comments.  If there are four patches in a patch series the individual
 605patches may be numbered like this: 1/4, 2/4, 3/4, 4/4.  This assures
 606that developers understand the order in which the patches should be
 607applied and that they have reviewed or applied all of the patches in
 608the patch series.
 609
 610A couple of example Subjects:
 611
 612    Subject: [patch 2/5] ext2: improve scalability of bitmap searching
 613    Subject: [PATCHv2 001/207] x86: fix eflags tracking
 614
 615The "from" line must be the very first line in the message body,
 616and has the form:
 617
 618        From: Original Author <author@example.com>
 619
 620The "from" line specifies who will be credited as the author of the
 621patch in the permanent changelog.  If the "from" line is missing,
 622then the "From:" line from the email header will be used to determine
 623the patch author in the changelog.
 624
 625The explanation body will be committed to the permanent source
 626changelog, so should make sense to a competent reader who has long
 627since forgotten the immediate details of the discussion that might
 628have led to this patch.  Including symptoms of the failure which the
 629patch addresses (kernel log messages, oops messages, etc.) is
 630especially useful for people who might be searching the commit logs
 631looking for the applicable patch.  If a patch fixes a compile failure,
 632it may not be necessary to include _all_ of the compile failures; just
 633enough that it is likely that someone searching for the patch can find
 634it.  As in the "summary phrase", it is important to be both succinct as
 635well as descriptive.
 636
 637The "---" marker line serves the essential purpose of marking for patch
 638handling tools where the changelog message ends.
 639
 640One good use for the additional comments after the "---" marker is for
 641a diffstat, to show what files have changed, and the number of
 642inserted and deleted lines per file.  A diffstat is especially useful
 643on bigger patches.  Other comments relevant only to the moment or the
 644maintainer, not suitable for the permanent changelog, should also go
 645here.  A good example of such comments might be "patch changelogs"
 646which describe what has changed between the v1 and v2 version of the
 647patch.
 648
 649If you are going to include a diffstat after the "---" marker, please
 650use diffstat options "-p 1 -w 70" so that filenames are listed from
 651the top of the kernel source tree and don't use too much horizontal
 652space (easily fit in 80 columns, maybe with some indentation).  (git
 653generates appropriate diffstats by default.)
 654
 655See more details on the proper patch format in the following
 656references.
 657
 658
 65916) Sending "git pull" requests  (from Linus emails)
 660
 661Please write the git repo address and branch name alone on the same line
 662so that I can't even by mistake pull from the wrong branch, and so
 663that a triple-click just selects the whole thing.
 664
 665So the proper format is something along the lines of:
 666
 667        "Please pull from
 668
 669                git://jdelvare.pck.nerim.net/jdelvare-2.6 i2c-for-linus
 670
 671         to get these changes:"
 672
 673so that I don't have to hunt-and-peck for the address and inevitably
 674get it wrong (actually, I've only gotten it wrong a few times, and
 675checking against the diffstat tells me when I get it wrong, but I'm
 676just a lot more comfortable when I don't have to "look for" the right
 677thing to pull, and double-check that I have the right branch-name).
 678
 679
 680Please use "git diff -M --stat --summary" to generate the diffstat:
 681the -M enables rename detection, and the summary enables a summary of
 682new/deleted or renamed files.
 683
 684With rename detection, the statistics are rather different [...]
 685because git will notice that a fair number of the changes are renames.
 686
 687-----------------------------------
 688SECTION 2 - HINTS, TIPS, AND TRICKS
 689-----------------------------------
 690
 691This section lists many of the common "rules" associated with code
 692submitted to the kernel.  There are always exceptions... but you must
 693have a really good reason for doing so.  You could probably call this
 694section Linus Computer Science 101.
 695
 696
 697
 6981) Read Documentation/CodingStyle
 699
 700Nuff said.  If your code deviates too much from this, it is likely
 701to be rejected without further review, and without comment.
 702
 703One significant exception is when moving code from one file to
 704another -- in this case you should not modify the moved code at all in
 705the same patch which moves it.  This clearly delineates the act of
 706moving the code and your changes.  This greatly aids review of the
 707actual differences and allows tools to better track the history of
 708the code itself.
 709
 710Check your patches with the patch style checker prior to submission
 711(scripts/checkpatch.pl).  The style checker should be viewed as
 712a guide not as the final word.  If your code looks better with
 713a violation then its probably best left alone.
 714
 715The checker reports at three levels:
 716 - ERROR: things that are very likely to be wrong
 717 - WARNING: things requiring careful review
 718 - CHECK: things requiring thought
 719
 720You should be able to justify all violations that remain in your
 721patch.
 722
 723
 724
 7252) #ifdefs are ugly
 726
 727Code cluttered with ifdefs is difficult to read and maintain.  Don't do
 728it.  Instead, put your ifdefs in a header, and conditionally define
 729'static inline' functions, or macros, which are used in the code.
 730Let the compiler optimize away the "no-op" case.
 731
 732Simple example, of poor code:
 733
 734        dev = alloc_etherdev (sizeof(struct funky_private));
 735        if (!dev)
 736                return -ENODEV;
 737        #ifdef CONFIG_NET_FUNKINESS
 738        init_funky_net(dev);
 739        #endif
 740
 741Cleaned-up example:
 742
 743(in header)
 744        #ifndef CONFIG_NET_FUNKINESS
 745        static inline void init_funky_net (struct net_device *d) {}
 746        #endif
 747
 748(in the code itself)
 749        dev = alloc_etherdev (sizeof(struct funky_private));
 750        if (!dev)
 751                return -ENODEV;
 752        init_funky_net(dev);
 753
 754
 755
 7563) 'static inline' is better than a macro
 757
 758Static inline functions are greatly preferred over macros.
 759They provide type safety, have no length limitations, no formatting
 760limitations, and under gcc they are as cheap as macros.
 761
 762Macros should only be used for cases where a static inline is clearly
 763suboptimal [there are a few, isolated cases of this in fast paths],
 764or where it is impossible to use a static inline function [such as
 765string-izing].
 766
 767'static inline' is preferred over 'static __inline__', 'extern inline',
 768and 'extern __inline__'.
 769
 770
 771
 7724) Don't over-design.
 773
 774Don't try to anticipate nebulous future cases which may or may not
 775be useful:  "Make it as simple as you can, and no simpler."
 776
 777
 778
 779----------------------
 780SECTION 3 - REFERENCES
 781----------------------
 782
 783Andrew Morton, "The perfect patch" (tpp).
 784  <http://www.ozlabs.org/~akpm/stuff/tpp.txt>
 785
 786Jeff Garzik, "Linux kernel patch submission format".
 787  <http://linux.yyz.us/patch-format.html>
 788
 789Greg Kroah-Hartman, "How to piss off a kernel subsystem maintainer".
 790  <http://www.kroah.com/log/linux/maintainer.html>
 791  <http://www.kroah.com/log/linux/maintainer-02.html>
 792  <http://www.kroah.com/log/linux/maintainer-03.html>
 793  <http://www.kroah.com/log/linux/maintainer-04.html>
 794  <http://www.kroah.com/log/linux/maintainer-05.html>
 795  <http://www.kroah.com/log/linux/maintainer-06.html>
 796
 797NO!!!! No more huge patch bombs to linux-kernel@vger.kernel.org people!
 798  <https://lkml.org/lkml/2005/7/11/336>
 799
 800Kernel Documentation/CodingStyle:
 801  <http://users.sosdg.org/~qiyong/lxr/source/Documentation/CodingStyle>
 802
 803Linus Torvalds's mail on the canonical patch format:
 804  <http://lkml.org/lkml/2005/4/7/183>
 805
 806Andi Kleen, "On submitting kernel patches"
 807  Some strategies to get difficult or controversial changes in.
 808  http://halobates.de/on-submitting-patches.pdf
 809
 810--
 811
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.