1.. _development_posting:
2
3Posting patches
4===============
5
6Sooner or later, the time comes when your work is ready to be presented to
7the community for review and, eventually, inclusion into the mainline
8kernel.  Unsurprisingly, the kernel development community has evolved a set
9of conventions and procedures which are used in the posting of patches;
10following them will make life much easier for everybody involved.  This
11document will attempt to cover these expectations in reasonable detail;
12more information can also be found in the files process/submitting-patches.rst,
13process/submitting-drivers.rst, and process/submit-checklist.rst in the kernel
14documentation directory.
15
16
17When to post
18------------
19
20There is a constant temptation to avoid posting patches before they are
21completely "ready."  For simple patches, that is not a problem.  If the
22work being done is complex, though, there is a lot to be gained by getting
23feedback from the community before the work is complete.  So you should
24consider posting in-progress work, or even making a git tree available so
25that interested developers can catch up with your work at any time.
26
27When posting code which is not yet considered ready for inclusion, it is a
28good idea to say so in the posting itself.  Also mention any major work
29which remains to be done and any known problems.  Fewer people will look at
30patches which are known to be half-baked, but those who do will come in
31with the idea that they can help you drive the work in the right direction.
32
33
34Before creating patches
35-----------------------
36
37There are a number of things which should be done before you consider
38sending patches to the development community.  These include:
39
40 - Test the code to the extent that you can.  Make use of the kernel's
41   debugging tools, ensure that the kernel will build with all reasonable
42   combinations of configuration options, use cross-compilers to build for
43   different architectures, etc.
44
45 - Make sure your code is compliant with the kernel coding style
46   guidelines.
47
48 - Does your change have performance implications?  If so, you should run
49   benchmarks showing what the impact (or benefit) of your change is; a
50   summary of the results should be included with the patch.
51
52 - Be sure that you have the right to post the code.  If this work was done
53   for an employer, the employer likely has a right to the work and must be
54   agreeable with its release under the GPL.
55
56As a general rule, putting in some extra thought before posting code almost
57always pays back the effort in short order.
58
59
60Patch preparation
61-----------------
62
63The preparation of patches for posting can be a surprising amount of work,
64but, once again, attempting to save time here is not generally advisable
65even in the short term.
66
67Patches must be prepared against a specific version of the kernel.  As a
68general rule, a patch should be based on the current mainline as found in
69Linus's git tree.  When basing on mainline, start with a well-known release
70point - a stable or -rc release - rather than branching off the mainline at
71an arbitrary spot.
72
73It may become necessary to make versions against -mm, linux-next, or a
74subsystem tree, though, to facilitate wider testing and review.  Depending
75on the area of your patch and what is going on elsewhere, basing a patch
76against these other trees can require a significant amount of work
77resolving conflicts and dealing with API changes.
78
79Only the most simple changes should be formatted as a single patch;
80everything else should be made as a logical series of changes.  Splitting
81up patches is a bit of an art; some developers spend a long time figuring
82out how to do it in the way that the community expects.  There are a few
83rules of thumb, however, which can help considerably:
84
85 - The patch series you post will almost certainly not be the series of
86   changes found in your working revision control system.  Instead, the
87   changes you have made need to be considered in their final form, then
88   split apart in ways which make sense.  The developers are interested in
89   discrete, self-contained changes, not the path you took to get to those
90   changes.
91
92 - Each logically independent change should be formatted as a separate
93   patch.  These changes can be small ("add a field to this structure") or
94   large (adding a significant new driver, for example), but they should be
95   conceptually small and amenable to a one-line description.  Each patch
96   should make a specific change which can be reviewed on its own and
97   verified to do what it says it does.
98
99 - As a way of restating the guideline above: do not mix different types of
100   changes in the same patch.  If a single patch fixes a critical security
101   bug, rearranges a few structures, and reformats the code, there is a
102   good chance that it will be passed over and the important fix will be
103   lost.
104
105 - Each patch should yield a kernel which builds and runs properly; if your
106   patch series is interrupted in the middle, the result should still be a
107   working kernel.  Partial application of a patch series is a common
108   scenario when the "git bisect" tool is used to find regressions; if the
109   result is a broken kernel, you will make life harder for developers and
110   users who are engaging in the noble work of tracking down problems.
111
112 - Do not overdo it, though.  One developer once posted a set of edits
113   to a single file as 500 separate patches - an act which did not make him
114   the most popular person on the kernel mailing list.  A single patch can
115   be reasonably large as long as it still contains a single *logical*
116   change.
117
118 - It can be tempting to add a whole new infrastructure with a series of
119   patches, but to leave that infrastructure unused until the final patch
120   in the series enables the whole thing.  This temptation should be
121   avoided if possible; if that series adds regressions, bisection will
122   finger the last patch as the one which caused the problem, even though
123   the real bug is elsewhere.  Whenever possible, a patch which adds new
124   code should make that code active immediately.
125
126Working to create the perfect patch series can be a frustrating process
127which takes quite a bit of time and thought after the "real work" has been
128done.  When done properly, though, it is time well spent.
129
130
131Patch formatting and changelogs
132-------------------------------
133
134So now you have a perfect series of patches for posting, but the work is
135not done quite yet.  Each patch needs to be formatted into a message which
136quickly and clearly communicates its purpose to the rest of the world.  To
137that end, each patch will be composed of the following:
138
139 - An optional "From" line naming the author of the patch.  This line is
140   only necessary if you are passing on somebody else's patch via email,
141   but it never hurts to add it when in doubt.
142
143 - A one-line description of what the patch does.  This message should be
144   enough for a reader who sees it with no other context to figure out the
145   scope of the patch; it is the line that will show up in the "short form"
146   changelogs.  This message is usually formatted with the relevant
147   subsystem name first, followed by the purpose of the patch.  For
148   example:
149
150   ::
151
152	gpio: fix build on CONFIG_GPIO_SYSFS=n
153
154 - A blank line followed by a detailed description of the contents of the
155   patch.  This description can be as long as is required; it should say
156   what the patch does and why it should be applied to the kernel.
157
158 - One or more tag lines, with, at a minimum, one Signed-off-by: line from
159   the author of the patch.  Tags will be described in more detail below.
160
161The items above, together, form the changelog for the patch.  Writing good
162changelogs is a crucial but often-neglected art; it's worth spending
163another moment discussing this issue.  When writing a changelog, you should
164bear in mind that a number of different people will be reading your words.
165These include subsystem maintainers and reviewers who need to decide
166whether the patch should be included, distributors and other maintainers
167trying to decide whether a patch should be backported to other kernels, bug
168hunters wondering whether the patch is responsible for a problem they are
169chasing, users who want to know how the kernel has changed, and more.  A
170good changelog conveys the needed information to all of these people in the
171most direct and concise way possible.
172
173To that end, the summary line should describe the effects of and motivation
174for the change as well as possible given the one-line constraint.  The
175detailed description can then amplify on those topics and provide any
176needed additional information.  If the patch fixes a bug, cite the commit
177which introduced the bug if possible (and please provide both the commit ID
178and the title when citing commits).  If a problem is associated with
179specific log or compiler output, include that output to help others
180searching for a solution to the same problem.  If the change is meant to
181support other changes coming in later patch, say so.  If internal APIs are
182changed, detail those changes and how other developers should respond.  In
183general, the more you can put yourself into the shoes of everybody who will
184be reading your changelog, the better that changelog (and the kernel as a
185whole) will be.
186
187Needless to say, the changelog should be the text used when committing the
188change to a revision control system.  It will be followed by:
189
190 - The patch itself, in the unified ("-u") patch format.  Using the "-p"
191   option to diff will associate function names with changes, making the
192   resulting patch easier for others to read.
193
194You should avoid including changes to irrelevant files (those generated by
195the build process, for example, or editor backup files) in the patch.  The
196file "dontdiff" in the Documentation directory can help in this regard;
197pass it to diff with the "-X" option.
198
199The tags mentioned above are used to describe how various developers have
200been associated with the development of this patch.  They are described in
201detail in the process/submitting-patches.rst document; what follows here is a
202brief summary.  Each of these lines has the format:
203
204::
205
206	tag: Full Name <email address>  optional-other-stuff
207
208The tags in common use are:
209
210 - Signed-off-by: this is a developer's certification that he or she has
211   the right to submit the patch for inclusion into the kernel.  It is an
212   agreement to the Developer's Certificate of Origin, the full text of
213   which can be found in Documentation/process/submitting-patches.rst.  Code
214   without a proper signoff cannot be merged into the mainline.
215
216 - Co-developed-by: states that the patch was also created by another developer
217   along with the original author.  This is useful at times when multiple
218   people work on a single patch.  Note, this person also needs to have a
219   Signed-off-by: line in the patch as well.
220
221 - Acked-by: indicates an agreement by another developer (often a
222   maintainer of the relevant code) that the patch is appropriate for
223   inclusion into the kernel.
224
225 - Tested-by: states that the named person has tested the patch and found
226   it to work.
227
228 - Reviewed-by: the named developer has reviewed the patch for correctness;
229   see the reviewer's statement in Documentation/process/submitting-patches.rst
230   for more detail.
231
232 - Reported-by: names a user who reported a problem which is fixed by this
233   patch; this tag is used to give credit to the (often underappreciated)
234   people who test our code and let us know when things do not work
235   correctly.
236
237 - Cc: the named person received a copy of the patch and had the
238   opportunity to comment on it.
239
240Be careful in the addition of tags to your patches: only Cc: is appropriate
241for addition without the explicit permission of the person named.
242
243
244Sending the patch
245-----------------
246
247Before you mail your patches, there are a couple of other things you should
248take care of:
249
250 - Are you sure that your mailer will not corrupt the patches?  Patches
251   which have had gratuitous white-space changes or line wrapping performed
252   by the mail client will not apply at the other end, and often will not
253   be examined in any detail.  If there is any doubt at all, mail the patch
254   to yourself and convince yourself that it shows up intact.
255
256   Documentation/process/email-clients.rst has some helpful hints on making
257   specific mail clients work for sending patches.
258
259 - Are you sure your patch is free of silly mistakes?  You should always
260   run patches through scripts/checkpatch.pl and address the complaints it
261   comes up with.  Please bear in mind that checkpatch.pl, while being the
262   embodiment of a fair amount of thought about what kernel patches should
263   look like, is not smarter than you.  If fixing a checkpatch.pl complaint
264   would make the code worse, don't do it.
265
266Patches should always be sent as plain text.  Please do not send them as
267attachments; that makes it much harder for reviewers to quote sections of
268the patch in their replies.  Instead, just put the patch directly into your
269message.
270
271When mailing patches, it is important to send copies to anybody who might
272be interested in it.  Unlike some other projects, the kernel encourages
273people to err on the side of sending too many copies; don't assume that the
274relevant people will see your posting on the mailing lists.  In particular,
275copies should go to:
276
277 - The maintainer(s) of the affected subsystem(s).  As described earlier,
278   the MAINTAINERS file is the first place to look for these people.
279
280 - Other developers who have been working in the same area - especially
281   those who might be working there now.  Using git to see who else has
282   modified the files you are working on can be helpful.
283
284 - If you are responding to a bug report or a feature request, copy the
285   original poster as well.
286
287 - Send a copy to the relevant mailing list, or, if nothing else applies,
288   the linux-kernel list.
289
290 - If you are fixing a bug, think about whether the fix should go into the
291   next stable update.  If so, stable@vger.kernel.org should get a copy of
292   the patch.  Also add a "Cc: stable@vger.kernel.org" to the tags within
293   the patch itself; that will cause the stable team to get a notification
294   when your fix goes into the mainline.
295
296When selecting recipients for a patch, it is good to have an idea of who
297you think will eventually accept the patch and get it merged.  While it
298is possible to send patches directly to Linus Torvalds and have him merge
299them, things are not normally done that way.  Linus is busy, and there are
300subsystem maintainers who watch over specific parts of the kernel.  Usually
301you will be wanting that maintainer to merge your patches.  If there is no
302obvious maintainer, Andrew Morton is often the patch target of last resort.
303
304Patches need good subject lines.  The canonical format for a patch line is
305something like:
306
307::
308
309	[PATCH nn/mm] subsys: one-line description of the patch
310
311where "nn" is the ordinal number of the patch, "mm" is the total number of
312patches in the series, and "subsys" is the name of the affected subsystem.
313Clearly, nn/mm can be omitted for a single, standalone patch.
314
315If you have a significant series of patches, it is customary to send an
316introductory description as part zero.  This convention is not universally
317followed though; if you use it, remember that information in the
318introduction does not make it into the kernel changelogs.  So please ensure
319that the patches, themselves, have complete changelog information.
320
321In general, the second and following parts of a multi-part patch should be
322sent as a reply to the first part so that they all thread together at the
323receiving end.  Tools like git and quilt have commands to mail out a set of
324patches with the proper threading.  If you have a long series, though, and
325are using git, please stay away from the --chain-reply-to option to avoid
326creating exceptionally deep nesting.
327