1.. _submittingpatches: 2 3Submitting patches: the essential guide to getting your code into the kernel 4============================================================================ 5 6For a person or company who wishes to submit a change to the Linux 7kernel, the process can sometimes be daunting if you're not familiar 8with "the system." This text is a collection of suggestions which 9can greatly increase the chances of your change being accepted. 10 11This document contains a large number of suggestions in a relatively terse 12format. For detailed information on how the kernel development process 13works, see Documentation/process/development-process.rst. Also, read 14Documentation/process/submit-checklist.rst 15for a list of items to check before submitting code. If you are submitting 16a driver, also read Documentation/process/submitting-drivers.rst; for device 17tree binding patches, read Documentation/process/submitting-patches.rst. 18 19This documentation assumes that you're using ``git`` to prepare your patches. 20If you're unfamiliar with ``git``, you would be well-advised to learn how to 21use it, it will make your life as a kernel developer and in general much 22easier. 23 24Some subsystems and maintainer trees have additional information about 25their workflow and expectations, see :ref:`Documentation/process/maintainer 26handbooks <maintainer_handbooks_main>`. 27 28Obtain a current source tree 29---------------------------- 30 31If you do not have a repository with the current kernel source handy, use 32``git`` to obtain one. You'll want to start with the mainline repository, 33which can be grabbed with:: 34 35 git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git 36 37Note, however, that you may not want to develop against the mainline tree 38directly. Most subsystem maintainers run their own trees and want to see 39patches prepared against those trees. See the **T:** entry for the subsystem 40in the MAINTAINERS file to find that tree, or simply ask the maintainer if 41the tree is not listed there. 42 43.. _describe_changes: 44 45Describe your changes 46--------------------- 47 48Describe your problem. Whether your patch is a one-line bug fix or 495000 lines of a new feature, there must be an underlying problem that 50motivated you to do this work. Convince the reviewer that there is a 51problem worth fixing and that it makes sense for them to read past the 52first paragraph. 53 54Describe user-visible impact. Straight up crashes and lockups are 55pretty convincing, but not all bugs are that blatant. Even if the 56problem was spotted during code review, describe the impact you think 57it can have on users. Keep in mind that the majority of Linux 58installations run kernels from secondary stable trees or 59vendor/product-specific trees that cherry-pick only specific patches 60from upstream, so include anything that could help route your change 61downstream: provoking circumstances, excerpts from dmesg, crash 62descriptions, performance regressions, latency spikes, lockups, etc. 63 64Quantify optimizations and trade-offs. If you claim improvements in 65performance, memory consumption, stack footprint, or binary size, 66include numbers that back them up. But also describe non-obvious 67costs. Optimizations usually aren't free but trade-offs between CPU, 68memory, and readability; or, when it comes to heuristics, between 69different workloads. Describe the expected downsides of your 70optimization so that the reviewer can weigh costs against benefits. 71 72Once the problem is established, describe what you are actually doing 73about it in technical detail. It's important to describe the change 74in plain English for the reviewer to verify that the code is behaving 75as you intend it to. 76 77The maintainer will thank you if you write your patch description in a 78form which can be easily pulled into Linux's source code management 79system, ``git``, as a "commit log". See :ref:`explicit_in_reply_to`. 80 81Solve only one problem per patch. If your description starts to get 82long, that's a sign that you probably need to split up your patch. 83See :ref:`split_changes`. 84 85When you submit or resubmit a patch or patch series, include the 86complete patch description and justification for it. Don't just 87say that this is version N of the patch (series). Don't expect the 88subsystem maintainer to refer back to earlier patch versions or referenced 89URLs to find the patch description and put that into the patch. 90I.e., the patch (series) and its description should be self-contained. 91This benefits both the maintainers and reviewers. Some reviewers 92probably didn't even receive earlier versions of the patch. 93 94Describe your changes in imperative mood, e.g. "make xyzzy do frotz" 95instead of "[This patch] makes xyzzy do frotz" or "[I] changed xyzzy 96to do frotz", as if you are giving orders to the codebase to change 97its behaviour. 98 99If you want to refer to a specific commit, don't just refer to the 100SHA-1 ID of the commit. Please also include the oneline summary of 101the commit, to make it easier for reviewers to know what it is about. 102Example:: 103 104 Commit e21d2170f36602ae2708 ("video: remove unnecessary 105 platform_set_drvdata()") removed the unnecessary 106 platform_set_drvdata(), but left the variable "dev" unused, 107 delete it. 108 109You should also be sure to use at least the first twelve characters of the 110SHA-1 ID. The kernel repository holds a *lot* of objects, making 111collisions with shorter IDs a real possibility. Bear in mind that, even if 112there is no collision with your six-character ID now, that condition may 113change five years from now. 114 115If related discussions or any other background information behind the change 116can be found on the web, add 'Link:' tags pointing to it. In case your patch 117fixes a bug, for example, add a tag with a URL referencing the report in the 118mailing list archives or a bug tracker; if the patch is a result of some 119earlier mailing list discussion or something documented on the web, point to 120it. 121 122When linking to mailing list archives, preferably use the lore.kernel.org 123message archiver service. To create the link URL, use the contents of the 124``Message-Id`` header of the message without the surrounding angle brackets. 125For example:: 126 127 Link: https://lore.kernel.org/r/30th.anniversary.repost@klaava.Helsinki.FI/ 128 129Please check the link to make sure that it is actually working and points 130to the relevant message. 131 132However, try to make your explanation understandable without external 133resources. In addition to giving a URL to a mailing list archive or bug, 134summarize the relevant points of the discussion that led to the 135patch as submitted. 136 137If your patch fixes a bug in a specific commit, e.g. you found an issue using 138``git bisect``, please use the 'Fixes:' tag with the first 12 characters of 139the SHA-1 ID, and the one line summary. Do not split the tag across multiple 140lines, tags are exempt from the "wrap at 75 columns" rule in order to simplify 141parsing scripts. For example:: 142 143 Fixes: 54a4f0239f2e ("KVM: MMU: make kvm_mmu_zap_page() return the number of pages it actually freed") 144 145The following ``git config`` settings can be used to add a pretty format for 146outputting the above style in the ``git log`` or ``git show`` commands:: 147 148 [core] 149 abbrev = 12 150 [pretty] 151 fixes = Fixes: %h (\"%s\") 152 153An example call:: 154 155 $ git log -1 --pretty=fixes 54a4f0239f2e 156 Fixes: 54a4f0239f2e ("KVM: MMU: make kvm_mmu_zap_page() return the number of pages it actually freed") 157 158.. _split_changes: 159 160Separate your changes 161--------------------- 162 163Separate each **logical change** into a separate patch. 164 165For example, if your changes include both bug fixes and performance 166enhancements for a single driver, separate those changes into two 167or more patches. If your changes include an API update, and a new 168driver which uses that new API, separate those into two patches. 169 170On the other hand, if you make a single change to numerous files, 171group those changes into a single patch. Thus a single logical change 172is contained within a single patch. 173 174The point to remember is that each patch should make an easily understood 175change that can be verified by reviewers. Each patch should be justifiable 176on its own merits. 177 178If one patch depends on another patch in order for a change to be 179complete, that is OK. Simply note **"this patch depends on patch X"** 180in your patch description. 181 182When dividing your change into a series of patches, take special care to 183ensure that the kernel builds and runs properly after each patch in the 184series. Developers using ``git bisect`` to track down a problem can end up 185splitting your patch series at any point; they will not thank you if you 186introduce bugs in the middle. 187 188If you cannot condense your patch set into a smaller set of patches, 189then only post say 15 or so at a time and wait for review and integration. 190 191 192 193Style-check your changes 194------------------------ 195 196Check your patch for basic style violations, details of which can be 197found in Documentation/process/coding-style.rst. 198Failure to do so simply wastes 199the reviewers time and will get your patch rejected, probably 200without even being read. 201 202One significant exception is when moving code from one file to 203another -- in this case you should not modify the moved code at all in 204the same patch which moves it. This clearly delineates the act of 205moving the code and your changes. This greatly aids review of the 206actual differences and allows tools to better track the history of 207the code itself. 208 209Check your patches with the patch style checker prior to submission 210(scripts/checkpatch.pl). Note, though, that the style checker should be 211viewed as a guide, not as a replacement for human judgment. If your code 212looks better with a violation then its probably best left alone. 213 214The checker reports at three levels: 215 - ERROR: things that are very likely to be wrong 216 - WARNING: things requiring careful review 217 - CHECK: things requiring thought 218 219You should be able to justify all violations that remain in your 220patch. 221 222 223Select the recipients for your patch 224------------------------------------ 225 226You should always copy the appropriate subsystem maintainer(s) on any patch 227to code that they maintain; look through the MAINTAINERS file and the 228source code revision history to see who those maintainers are. The 229script scripts/get_maintainer.pl can be very useful at this step. If you 230cannot find a maintainer for the subsystem you are working on, Andrew 231Morton (akpm@linux-foundation.org) serves as a maintainer of last resort. 232 233You should also normally choose at least one mailing list to receive a copy 234of your patch set. linux-kernel@vger.kernel.org should be used by default 235for all patches, but the volume on that list has caused a number of 236developers to tune it out. Look in the MAINTAINERS file for a 237subsystem-specific list; your patch will probably get more attention there. 238Please do not spam unrelated lists, though. 239 240Many kernel-related lists are hosted on vger.kernel.org; you can find a 241list of them at http://vger.kernel.org/vger-lists.html. There are 242kernel-related lists hosted elsewhere as well, though. 243 244Do not send more than 15 patches at once to the vger mailing lists!!! 245 246Linus Torvalds is the final arbiter of all changes accepted into the 247Linux kernel. His e-mail address is <torvalds@linux-foundation.org>. 248He gets a lot of e-mail, and, at this point, very few patches go through 249Linus directly, so typically you should do your best to -avoid- 250sending him e-mail. 251 252If you have a patch that fixes an exploitable security bug, send that patch 253to security@kernel.org. For severe bugs, a short embargo may be considered 254to allow distributors to get the patch out to users; in such cases, 255obviously, the patch should not be sent to any public lists. See also 256Documentation/admin-guide/security-bugs.rst. 257 258Patches that fix a severe bug in a released kernel should be directed 259toward the stable maintainers by putting a line like this:: 260 261 Cc: stable@vger.kernel.org 262 263into the sign-off area of your patch (note, NOT an email recipient). You 264should also read Documentation/process/stable-kernel-rules.rst 265in addition to this document. 266 267If changes affect userland-kernel interfaces, please send the MAN-PAGES 268maintainer (as listed in the MAINTAINERS file) a man-pages patch, or at 269least a notification of the change, so that some information makes its way 270into the manual pages. User-space API changes should also be copied to 271linux-api@vger.kernel.org. 272 273For small patches you may want to CC the Trivial Patch Monkey 274trivial@kernel.org which collects "trivial" patches. Have a look 275into the MAINTAINERS file for its current manager. 276 277Trivial patches must qualify for one of the following rules: 278 279- Spelling fixes in documentation 280- Spelling fixes for errors which could break :manpage:`grep(1)` 281- Warning fixes (cluttering with useless warnings is bad) 282- Compilation fixes (only if they are actually correct) 283- Runtime fixes (only if they actually fix things) 284- Removing use of deprecated functions/macros 285- Contact detail and documentation fixes 286- Non-portable code replaced by portable code (even in arch-specific, 287 since people copy, as long as it's trivial) 288- Any fix by the author/maintainer of the file (ie. patch monkey 289 in re-transmission mode) 290 291 292 293No MIME, no links, no compression, no attachments. Just plain text 294------------------------------------------------------------------- 295 296Linus and other kernel developers need to be able to read and comment 297on the changes you are submitting. It is important for a kernel 298developer to be able to "quote" your changes, using standard e-mail 299tools, so that they may comment on specific portions of your code. 300 301For this reason, all patches should be submitted by e-mail "inline". The 302easiest way to do this is with ``git send-email``, which is strongly 303recommended. An interactive tutorial for ``git send-email`` is available at 304https://git-send-email.io. 305 306If you choose not to use ``git send-email``: 307 308.. warning:: 309 310 Be wary of your editor's word-wrap corrupting your patch, 311 if you choose to cut-n-paste your patch. 312 313Do not attach the patch as a MIME attachment, compressed or not. 314Many popular e-mail applications will not always transmit a MIME 315attachment as plain text, making it impossible to comment on your 316code. A MIME attachment also takes Linus a bit more time to process, 317decreasing the likelihood of your MIME-attached change being accepted. 318 319Exception: If your mailer is mangling patches then someone may ask 320you to re-send them using MIME. 321 322See Documentation/process/email-clients.rst for hints about configuring 323your e-mail client so that it sends your patches untouched. 324 325Respond to review comments 326-------------------------- 327 328Your patch will almost certainly get comments from reviewers on ways in 329which the patch can be improved, in the form of a reply to your email. You must 330respond to those comments; ignoring reviewers is a good way to get ignored in 331return. You can simply reply to their emails to answer their comments. Review 332comments or questions that do not lead to a code change should almost certainly 333bring about a comment or changelog entry so that the next reviewer better 334understands what is going on. 335 336Be sure to tell the reviewers what changes you are making and to thank them 337for their time. Code review is a tiring and time-consuming process, and 338reviewers sometimes get grumpy. Even in that case, though, respond 339politely and address the problems they have pointed out. 340 341See Documentation/process/email-clients.rst for recommendations on email 342clients and mailing list etiquette. 343 344.. _resend_reminders: 345 346Don't get discouraged - or impatient 347------------------------------------ 348 349After you have submitted your change, be patient and wait. Reviewers are 350busy people and may not get to your patch right away. 351 352Once upon a time, patches used to disappear into the void without comment, 353but the development process works more smoothly than that now. You should 354receive comments within a week or so; if that does not happen, make sure 355that you have sent your patches to the right place. Wait for a minimum of 356one week before resubmitting or pinging reviewers - possibly longer during 357busy times like merge windows. 358 359It's also ok to resend the patch or the patch series after a couple of 360weeks with the word "RESEND" added to the subject line:: 361 362 [PATCH Vx RESEND] sub/sys: Condensed patch summary 363 364Don't add "RESEND" when you are submitting a modified version of your 365patch or patch series - "RESEND" only applies to resubmission of a 366patch or patch series which have not been modified in any way from the 367previous submission. 368 369 370Include PATCH in the subject 371----------------------------- 372 373Due to high e-mail traffic to Linus, and to linux-kernel, it is common 374convention to prefix your subject line with [PATCH]. This lets Linus 375and other kernel developers more easily distinguish patches from other 376e-mail discussions. 377 378``git send-email`` will do this for you automatically. 379 380 381Sign your work - the Developer's Certificate of Origin 382------------------------------------------------------ 383 384To improve tracking of who did what, especially with patches that can 385percolate to their final resting place in the kernel through several 386layers of maintainers, we've introduced a "sign-off" procedure on 387patches that are being emailed around. 388 389The sign-off is a simple line at the end of the explanation for the 390patch, which certifies that you wrote it or otherwise have the right to 391pass it on as an open-source patch. The rules are pretty simple: if you 392can certify the below: 393 394Developer's Certificate of Origin 1.1 395^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 396 397By making a contribution to this project, I certify that: 398 399 (a) The contribution was created in whole or in part by me and I 400 have the right to submit it under the open source license 401 indicated in the file; or 402 403 (b) The contribution is based upon previous work that, to the best 404 of my knowledge, is covered under an appropriate open source 405 license and I have the right under that license to submit that 406 work with modifications, whether created in whole or in part 407 by me, under the same open source license (unless I am 408 permitted to submit under a different license), as indicated 409 in the file; or 410 411 (c) The contribution was provided directly to me by some other 412 person who certified (a), (b) or (c) and I have not modified 413 it. 414 415 (d) I understand and agree that this project and the contribution 416 are public and that a record of the contribution (including all 417 personal information I submit with it, including my sign-off) is 418 maintained indefinitely and may be redistributed consistent with 419 this project or the open source license(s) involved. 420 421then you just add a line saying:: 422 423 Signed-off-by: Random J Developer <random@developer.example.org> 424 425using your real name (sorry, no pseudonyms or anonymous contributions.) 426This will be done for you automatically if you use ``git commit -s``. 427Reverts should also include "Signed-off-by". ``git revert -s`` does that 428for you. 429 430Some people also put extra tags at the end. They'll just be ignored for 431now, but you can do this to mark internal company procedures or just 432point out some special detail about the sign-off. 433 434Any further SoBs (Signed-off-by:'s) following the author's SoB are from 435people handling and transporting the patch, but were not involved in its 436development. SoB chains should reflect the **real** route a patch took 437as it was propagated to the maintainers and ultimately to Linus, with 438the first SoB entry signalling primary authorship of a single author. 439 440 441When to use Acked-by:, Cc:, and Co-developed-by: 442------------------------------------------------ 443 444The Signed-off-by: tag indicates that the signer was involved in the 445development of the patch, or that he/she was in the patch's delivery path. 446 447If a person was not directly involved in the preparation or handling of a 448patch but wishes to signify and record their approval of it then they can 449ask to have an Acked-by: line added to the patch's changelog. 450 451Acked-by: is often used by the maintainer of the affected code when that 452maintainer neither contributed to nor forwarded the patch. 453 454Acked-by: is not as formal as Signed-off-by:. It is a record that the acker 455has at least reviewed the patch and has indicated acceptance. Hence patch 456mergers will sometimes manually convert an acker's "yep, looks good to me" 457into an Acked-by: (but note that it is usually better to ask for an 458explicit ack). 459 460Acked-by: does not necessarily indicate acknowledgement of the entire patch. 461For example, if a patch affects multiple subsystems and has an Acked-by: from 462one subsystem maintainer then this usually indicates acknowledgement of just 463the part which affects that maintainer's code. Judgement should be used here. 464When in doubt people should refer to the original discussion in the mailing 465list archives. 466 467If a person has had the opportunity to comment on a patch, but has not 468provided such comments, you may optionally add a ``Cc:`` tag to the patch. 469This is the only tag which might be added without an explicit action by the 470person it names - but it should indicate that this person was copied on the 471patch. This tag documents that potentially interested parties 472have been included in the discussion. 473 474Co-developed-by: states that the patch was co-created by multiple developers; 475it is used to give attribution to co-authors (in addition to the author 476attributed by the From: tag) when several people work on a single patch. Since 477Co-developed-by: denotes authorship, every Co-developed-by: must be immediately 478followed by a Signed-off-by: of the associated co-author. Standard sign-off 479procedure applies, i.e. the ordering of Signed-off-by: tags should reflect the 480chronological history of the patch insofar as possible, regardless of whether 481the author is attributed via From: or Co-developed-by:. Notably, the last 482Signed-off-by: must always be that of the developer submitting the patch. 483 484Note, the From: tag is optional when the From: author is also the person (and 485email) listed in the From: line of the email header. 486 487Example of a patch submitted by the From: author:: 488 489 <changelog> 490 491 Co-developed-by: First Co-Author <first@coauthor.example.org> 492 Signed-off-by: First Co-Author <first@coauthor.example.org> 493 Co-developed-by: Second Co-Author <second@coauthor.example.org> 494 Signed-off-by: Second Co-Author <second@coauthor.example.org> 495 Signed-off-by: From Author <from@author.example.org> 496 497Example of a patch submitted by a Co-developed-by: author:: 498 499 From: From Author <from@author.example.org> 500 501 <changelog> 502 503 Co-developed-by: Random Co-Author <random@coauthor.example.org> 504 Signed-off-by: Random Co-Author <random@coauthor.example.org> 505 Signed-off-by: From Author <from@author.example.org> 506 Co-developed-by: Submitting Co-Author <sub@coauthor.example.org> 507 Signed-off-by: Submitting Co-Author <sub@coauthor.example.org> 508 509 510Using Reported-by:, Tested-by:, Reviewed-by:, Suggested-by: and Fixes: 511---------------------------------------------------------------------- 512 513The Reported-by tag gives credit to people who find bugs and report them and it 514hopefully inspires them to help us again in the future. Please note that if 515the bug was reported in private, then ask for permission first before using the 516Reported-by tag. 517 518A Tested-by: tag indicates that the patch has been successfully tested (in 519some environment) by the person named. This tag informs maintainers that 520some testing has been performed, provides a means to locate testers for 521future patches, and ensures credit for the testers. 522 523Reviewed-by:, instead, indicates that the patch has been reviewed and found 524acceptable according to the Reviewer's Statement: 525 526Reviewer's statement of oversight 527^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 528 529By offering my Reviewed-by: tag, I state that: 530 531 (a) I have carried out a technical review of this patch to 532 evaluate its appropriateness and readiness for inclusion into 533 the mainline kernel. 534 535 (b) Any problems, concerns, or questions relating to the patch 536 have been communicated back to the submitter. I am satisfied 537 with the submitter's response to my comments. 538 539 (c) While there may be things that could be improved with this 540 submission, I believe that it is, at this time, (1) a 541 worthwhile modification to the kernel, and (2) free of known 542 issues which would argue against its inclusion. 543 544 (d) While I have reviewed the patch and believe it to be sound, I 545 do not (unless explicitly stated elsewhere) make any 546 warranties or guarantees that it will achieve its stated 547 purpose or function properly in any given situation. 548 549A Reviewed-by tag is a statement of opinion that the patch is an 550appropriate modification of the kernel without any remaining serious 551technical issues. Any interested reviewer (who has done the work) can 552offer a Reviewed-by tag for a patch. This tag serves to give credit to 553reviewers and to inform maintainers of the degree of review which has been 554done on the patch. Reviewed-by: tags, when supplied by reviewers known to 555understand the subject area and to perform thorough reviews, will normally 556increase the likelihood of your patch getting into the kernel. 557 558Both Tested-by and Reviewed-by tags, once received on mailing list from tester 559or reviewer, should be added by author to the applicable patches when sending 560next versions. However if the patch has changed substantially in following 561version, these tags might not be applicable anymore and thus should be removed. 562Usually removal of someone's Tested-by or Reviewed-by tags should be mentioned 563in the patch changelog (after the '---' separator). 564 565A Suggested-by: tag indicates that the patch idea is suggested by the person 566named and ensures credit to the person for the idea. Please note that this 567tag should not be added without the reporter's permission, especially if the 568idea was not posted in a public forum. That said, if we diligently credit our 569idea reporters, they will, hopefully, be inspired to help us again in the 570future. 571 572A Fixes: tag indicates that the patch fixes an issue in a previous commit. It 573is used to make it easy to determine where a bug originated, which can help 574review a bug fix. This tag also assists the stable kernel team in determining 575which stable kernel versions should receive your fix. This is the preferred 576method for indicating a bug fixed by the patch. See :ref:`describe_changes` 577for more details. 578 579Note: Attaching a Fixes: tag does not subvert the stable kernel rules 580process nor the requirement to Cc: stable@vger.kernel.org on all stable 581patch candidates. For more information, please read 582Documentation/process/stable-kernel-rules.rst. 583 584.. _the_canonical_patch_format: 585 586The canonical patch format 587-------------------------- 588 589This section describes how the patch itself should be formatted. Note 590that, if you have your patches stored in a ``git`` repository, proper patch 591formatting can be had with ``git format-patch``. The tools cannot create 592the necessary text, though, so read the instructions below anyway. 593 594The canonical patch subject line is:: 595 596 Subject: [PATCH 001/123] subsystem: summary phrase 597 598The canonical patch message body contains the following: 599 600 - A ``from`` line specifying the patch author, followed by an empty 601 line (only needed if the person sending the patch is not the author). 602 603 - The body of the explanation, line wrapped at 75 columns, which will 604 be copied to the permanent changelog to describe this patch. 605 606 - An empty line. 607 608 - The ``Signed-off-by:`` lines, described above, which will 609 also go in the changelog. 610 611 - A marker line containing simply ``---``. 612 613 - Any additional comments not suitable for the changelog. 614 615 - The actual patch (``diff`` output). 616 617The Subject line format makes it very easy to sort the emails 618alphabetically by subject line - pretty much any email reader will 619support that - since because the sequence number is zero-padded, 620the numerical and alphabetic sort is the same. 621 622The ``subsystem`` in the email's Subject should identify which 623area or subsystem of the kernel is being patched. 624 625The ``summary phrase`` in the email's Subject should concisely 626describe the patch which that email contains. The ``summary 627phrase`` should not be a filename. Do not use the same ``summary 628phrase`` for every patch in a whole patch series (where a ``patch 629series`` is an ordered sequence of multiple, related patches). 630 631Bear in mind that the ``summary phrase`` of your email becomes a 632globally-unique identifier for that patch. It propagates all the way 633into the ``git`` changelog. The ``summary phrase`` may later be used in 634developer discussions which refer to the patch. People will want to 635google for the ``summary phrase`` to read discussion regarding that 636patch. It will also be the only thing that people may quickly see 637when, two or three months later, they are going through perhaps 638thousands of patches using tools such as ``gitk`` or ``git log 639--oneline``. 640 641For these reasons, the ``summary`` must be no more than 70-75 642characters, and it must describe both what the patch changes, as well 643as why the patch might be necessary. It is challenging to be both 644succinct and descriptive, but that is what a well-written summary 645should do. 646 647The ``summary phrase`` may be prefixed by tags enclosed in square 648brackets: "Subject: [PATCH <tag>...] <summary phrase>". The tags are 649not considered part of the summary phrase, but describe how the patch 650should be treated. Common tags might include a version descriptor if 651the multiple versions of the patch have been sent out in response to 652comments (i.e., "v1, v2, v3"), or "RFC" to indicate a request for 653comments. 654 655If there are four patches in a patch series the individual patches may 656be numbered like this: 1/4, 2/4, 3/4, 4/4. This assures that developers 657understand the order in which the patches should be applied and that 658they have reviewed or applied all of the patches in the patch series. 659 660Here are some good example Subjects:: 661 662 Subject: [PATCH 2/5] ext2: improve scalability of bitmap searching 663 Subject: [PATCH v2 01/27] x86: fix eflags tracking 664 Subject: [PATCH v2] sub/sys: Condensed patch summary 665 Subject: [PATCH v2 M/N] sub/sys: Condensed patch summary 666 667The ``from`` line must be the very first line in the message body, 668and has the form: 669 670 From: Patch Author <author@example.com> 671 672The ``from`` line specifies who will be credited as the author of the 673patch in the permanent changelog. If the ``from`` line is missing, 674then the ``From:`` line from the email header will be used to determine 675the patch author in the changelog. 676 677The explanation body will be committed to the permanent source 678changelog, so should make sense to a competent reader who has long since 679forgotten the immediate details of the discussion that might have led to 680this patch. Including symptoms of the failure which the patch addresses 681(kernel log messages, oops messages, etc.) are especially useful for 682people who might be searching the commit logs looking for the applicable 683patch. The text should be written in such detail so that when read 684weeks, months or even years later, it can give the reader the needed 685details to grasp the reasoning for **why** the patch was created. 686 687If a patch fixes a compile failure, it may not be necessary to include 688_all_ of the compile failures; just enough that it is likely that 689someone searching for the patch can find it. As in the ``summary 690phrase``, it is important to be both succinct as well as descriptive. 691 692The ``---`` marker line serves the essential purpose of marking for 693patch handling tools where the changelog message ends. 694 695One good use for the additional comments after the ``---`` marker is 696for a ``diffstat``, to show what files have changed, and the number of 697inserted and deleted lines per file. A ``diffstat`` is especially useful 698on bigger patches. If you are going to include a ``diffstat`` after the 699``---`` marker, please use ``diffstat`` options ``-p 1 -w 70`` so that 700filenames are listed from the top of the kernel source tree and don't 701use too much horizontal space (easily fit in 80 columns, maybe with some 702indentation). (``git`` generates appropriate diffstats by default.) 703 704Other comments relevant only to the moment or the maintainer, not 705suitable for the permanent changelog, should also go here. A good 706example of such comments might be ``patch changelogs`` which describe 707what has changed between the v1 and v2 version of the patch. 708 709Please put this information **after** the ``---`` line which separates 710the changelog from the rest of the patch. The version information is 711not part of the changelog which gets committed to the git tree. It is 712additional information for the reviewers. If it's placed above the 713commit tags, it needs manual interaction to remove it. If it is below 714the separator line, it gets automatically stripped off when applying the 715patch:: 716 717 <commit message> 718 ... 719 Signed-off-by: Author <author@mail> 720 --- 721 V2 -> V3: Removed redundant helper function 722 V1 -> V2: Cleaned up coding style and addressed review comments 723 724 path/to/file | 5+++-- 725 ... 726 727See more details on the proper patch format in the following 728references. 729 730.. _backtraces: 731 732Backtraces in commit mesages 733^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 734 735Backtraces help document the call chain leading to a problem. However, 736not all backtraces are helpful. For example, early boot call chains are 737unique and obvious. Copying the full dmesg output verbatim, however, 738adds distracting information like timestamps, module lists, register and 739stack dumps. 740 741Therefore, the most useful backtraces should distill the relevant 742information from the dump, which makes it easier to focus on the real 743issue. Here is an example of a well-trimmed backtrace:: 744 745 unchecked MSR access error: WRMSR to 0xd51 (tried to write 0x0000000000000064) 746 at rIP: 0xffffffffae059994 (native_write_msr+0x4/0x20) 747 Call Trace: 748 mba_wrmsr 749 update_domains 750 rdtgroup_mkdir 751 752.. _explicit_in_reply_to: 753 754Explicit In-Reply-To headers 755---------------------------- 756 757It can be helpful to manually add In-Reply-To: headers to a patch 758(e.g., when using ``git send-email``) to associate the patch with 759previous relevant discussion, e.g. to link a bug fix to the email with 760the bug report. However, for a multi-patch series, it is generally 761best to avoid using In-Reply-To: to link to older versions of the 762series. This way multiple versions of the patch don't become an 763unmanageable forest of references in email clients. If a link is 764helpful, you can use the https://lore.kernel.org/ redirector (e.g., in 765the cover email text) to link to an earlier version of the patch series. 766 767 768Providing base tree information 769------------------------------- 770 771When other developers receive your patches and start the review process, 772it is often useful for them to know where in the tree history they 773should place your work. This is particularly useful for automated CI 774processes that attempt to run a series of tests in order to establish 775the quality of your submission before the maintainer starts the review. 776 777If you are using ``git format-patch`` to generate your patches, you can 778automatically include the base tree information in your submission by 779using the ``--base`` flag. The easiest and most convenient way to use 780this option is with topical branches:: 781 782 $ git checkout -t -b my-topical-branch master 783 Branch 'my-topical-branch' set up to track local branch 'master'. 784 Switched to a new branch 'my-topical-branch' 785 786 [perform your edits and commits] 787 788 $ git format-patch --base=auto --cover-letter -o outgoing/ master 789 outgoing/0000-cover-letter.patch 790 outgoing/0001-First-Commit.patch 791 outgoing/... 792 793When you open ``outgoing/0000-cover-letter.patch`` for editing, you will 794notice that it will have the ``base-commit:`` trailer at the very 795bottom, which provides the reviewer and the CI tools enough information 796to properly perform ``git am`` without worrying about conflicts:: 797 798 $ git checkout -b patch-review [base-commit-id] 799 Switched to a new branch 'patch-review' 800 $ git am patches.mbox 801 Applying: First Commit 802 Applying: ... 803 804Please see ``man git-format-patch`` for more information about this 805option. 806 807.. note:: 808 809 The ``--base`` feature was introduced in git version 2.9.0. 810 811If you are not using git to format your patches, you can still include 812the same ``base-commit`` trailer to indicate the commit hash of the tree 813on which your work is based. You should add it either in the cover 814letter or in the first patch of the series and it should be placed 815either below the ``---`` line or at the very bottom of all other 816content, right before your email signature. 817 818 819References 820---------- 821 822Andrew Morton, "The perfect patch" (tpp). 823 <https://www.ozlabs.org/~akpm/stuff/tpp.txt> 824 825Jeff Garzik, "Linux kernel patch submission format". 826 <https://web.archive.org/web/20180829112450/http://linux.yyz.us/patch-format.html> 827 828Greg Kroah-Hartman, "How to piss off a kernel subsystem maintainer". 829 <http://www.kroah.com/log/linux/maintainer.html> 830 831 <http://www.kroah.com/log/linux/maintainer-02.html> 832 833 <http://www.kroah.com/log/linux/maintainer-03.html> 834 835 <http://www.kroah.com/log/linux/maintainer-04.html> 836 837 <http://www.kroah.com/log/linux/maintainer-05.html> 838 839 <http://www.kroah.com/log/linux/maintainer-06.html> 840 841NO!!!! No more huge patch bombs to linux-kernel@vger.kernel.org people! 842 <https://lore.kernel.org/r/20050711.125305.08322243.davem@davemloft.net> 843 844Kernel Documentation/process/coding-style.rst 845 846Linus Torvalds's mail on the canonical patch format: 847 <https://lore.kernel.org/r/Pine.LNX.4.58.0504071023190.28951@ppc970.osdl.org> 848 849Andi Kleen, "On submitting kernel patches" 850 Some strategies to get difficult or controversial changes in. 851 852 http://halobates.de/on-submitting-patches.pdf 853