• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
2<html lang="en">
3<head>
4  <meta http-equiv="content-type" content="text/html; charset=utf-8">
5  <title>Submitting patches</title>
6  <link rel="stylesheet" type="text/css" href="mesa.css">
7</head>
8<body>
9
10<div class="header">
11  <h1>The Mesa 3D Graphics Library</h1>
12</div>
13
14<iframe src="contents.html"></iframe>
15<div class="content">
16
17<h1>Submitting patches</h1>
18
19
20<ul>
21<li><a href="#guidelines">Basic guidelines</a>
22<li><a href="#formatting">Patch formatting</a>
23<li><a href="#testing">Testing Patches</a>
24<li><a href="#mailing">Mailing Patches</a>
25<li><a href="#reviewing">Reviewing Patches</a>
26<li><a href="#nominations">Nominating a commit for a stable branch</a>
27<li><a href="#criteria">Criteria for accepting patches to the stable branch</a>
28<li><a href="#backports">Sending backports for the stable branch</a>
29<li><a href="#gittips">Git tips</a>
30</ul>
31
32<h2 id="guidelines">Basic guidelines</h2>
33
34<ul>
35<li>Patches should not mix code changes with code formatting changes (except,
36perhaps, in very trivial cases.)
37<li>Code patches should follow Mesa
38<a href="codingstyle.html" target="_parent">coding conventions</a>.
39<li>Whenever possible, patches should only effect individual Mesa/Gallium
40components.
41<li>Patches should never introduce build breaks and should be bisectable (see
42<code>git bisect</code>.)
43<li>Patches should be properly <a href="#formatting">formatted</a>.
44<li>Patches should be sufficiently <a href="#testing">tested</a> before submitting.
45<li>Patches should be submitted to <a href="#mailing">mesa-dev</a>
46for <a href="#reviewing">review</a> using <code>git send-email</code>.
47
48</ul>
49
50<h2 id="formatting">Patch formatting</h2>
51
52<ul>
53<li>Lines should be limited to 75 characters or less so that git logs
54displayed in 80-column terminals avoid line wrapping.  Note that git
55log uses 4 spaces of indentation (4 + 75 &lt; 80).
56<li>The first line should be a short, concise summary of the change prefixed
57with a module name.  Examples:
58<pre>
59    mesa: Add support for querying GL_VERTEX_ATTRIB_ARRAY_LONG
60
61    gallium: add PIPE_CAP_DEVICE_RESET_STATUS_QUERY
62
63    i965: Fix missing type in local variable declaration.
64</pre>
65<li>Subsequent patch comments should describe the change in more detail,
66if needed.  For example:
67<pre>
68    i965: Remove end-of-thread SEND alignment code.
69
70    This was present in Eric's initial implementation of the compaction code
71    for Sandybridge (commit 077d01b6). There is no documentation saying this
72    is necessary, and removing it causes no regressions in piglit on any
73    platform.
74</pre>
75<li>A "Signed-off-by:" line is not required, but not discouraged either.
76<li>If a patch addresses a bugzilla issue, that should be noted in the
77patch comment.  For example:
78<pre>
79   Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89689
80</pre>
81<li>If a patch addresses a issue introduced with earlier commit, that should be
82noted in the patch comment.  For example:
83<pre>
84   Fixes: d7b3707c612 "util/disk_cache: use stat() to check if entry is a directory"
85</pre>
86<li>If there have been several revisions to a patch during the review
87process, they should be noted such as in this example:
88<pre>
89    st/mesa: add ARB_texture_stencil8 support (v4)
90
91    if we support stencil texturing, enable texture_stencil8
92    there is no requirement to support native S8 for this,
93    the texture can be converted to x24s8 fine.
94
95    v2: fold fixes from Marek in:
96       a) put S8 last in the list
97       b) fix renderable to always test for d/s renderable
98        fixup the texture case to use a stencil only format
99        for picking the format for the texture view.
100    v3: hit fallback for getteximage
101    v4: put s8 back in front, it shouldn't get picked now (Ilia)
102</pre>
103<li>If someone tested your patch, document it with a line like this:
104<pre>
105    Tested-by: Joe Hacker &lt;jhacker@foo.com&gt;
106</pre>
107<li>If the patch was reviewed (usually the case) or acked by someone,
108that should be documented with:
109<pre>
110    Reviewed-by: Joe Hacker &lt;jhacker@foo.com&gt;
111    Acked-by: Joe Hacker &lt;jhacker@foo.com&gt;
112</pre>
113<li>If sending later revision of a patch, add all the tags - ack, r-b,
114Cc: mesa-stable and/or other. This provides reviewers with quick feedback if the
115patch has already been reviewed.
116<li>In order for your patch to reach the prospective reviewer easier/faster,
117use the script scripts/get_reviewer.pl to get a list of individuals and include
118them in the CC list.
119<br>
120Please use common sense and do <strong>not</strong> blindly add everyone.
121<br>
122<pre>
123    $ scripts/get_reviewer.pl --help # to get the help screen
124    $ scripts/get_reviewer.pl -f src/egl/drivers/dri2/platform_android.c
125    Rob Herring <robh@kernel.org> (reviewer:ANDROID EGL SUPPORT,added_lines:188/700=27%,removed_lines:58/283=20%)
126    Tomasz Figa <tfiga@chromium.org> (reviewer:ANDROID EGL SUPPORT,authored:12/41=29%,added_lines:308/700=44%,removed_lines:115/283=41%)
127    Emil Velikov <emil.l.velikov@gmail.com> (authored:13/41=32%,removed_lines:76/283=27%)
128</pre>
129</ul>
130
131
132
133<h2 id="testing">Testing Patches</h2>
134
135<p>
136It should go without saying that patches must be tested.  In general,
137do whatever testing is prudent.
138</p>
139
140<p>
141You should always run the Mesa test suite before submitting patches.
142The test suite can be run using the 'make check' command. All tests
143must pass before patches will be accepted, this may mean you have
144to update the tests themselves.
145</p>
146
147<p>
148Whenever possible and applicable, test the patch with
149<a href="https://piglit.freedesktop.org">Piglit</a> and/or
150<a href="https://android.googlesource.com/platform/external/deqp/">dEQP</a>
151to check for regressions.
152</p>
153
154<p>
155As mentioned at the begining, patches should be bisectable.
156A good way to test this is to make use of the `git rebase` command,
157to run your tests on each commit. Assuming your branch is based off
158<code>origin/master</code>, you can run:
159<pre>
160$ git rebase --interactive --exec "make check" origin/master
161</pre>
162replacing <code>"make check"</code> with whatever other test you want to
163run.
164</p>
165
166
167<h2 id="mailing">Mailing Patches</h2>
168
169<p>
170Patches should be sent to the mesa-dev mailing list for review:
171<a href="https://lists.freedesktop.org/mailman/listinfo/mesa-dev">
172mesa-dev@lists.freedesktop.org</a>.
173When submitting a patch make sure to use
174<a href="https://git-scm.com/docs/git-send-email">git send-email</a>
175rather than attaching patches to emails. Sending patches as
176attachments prevents people from being able to provide in-line review
177comments.
178</p>
179
180<p>
181When submitting follow-up patches you can use --in-reply-to to make v2, v3,
182etc patches show up as replies to the originals. This usually works well
183when you're sending out updates to individual patches (as opposed to
184re-sending the whole series). Using --in-reply-to makes
185it harder for reviewers to accidentally review old patches.
186</p>
187
188<p>
189When submitting follow-up patches you should also login to
190<a href="https://patchwork.freedesktop.org">patchwork</a> and change the
191state of your old patches to Superseded.
192</p>
193
194<p>
195Some companies' mail server automatically append a legal disclaimer,
196usually containing something along the lines of "The information in this
197email is confidential" and "distribution is strictly prohibited".<br/>
198These legal notices prevent us from being able to accept your patch,
199rendering the whole process pointless. Please make sure these are
200disabled before sending your patches. (Note that you may need to contact
201your email administrator for this.)
202</p>
203
204<h2 id="reviewing">Reviewing Patches</h2>
205
206<p>
207When you've reviewed a patch on the mailing list, please be unambiguous
208about your review.  That is, state either
209</p>
210<pre>
211    Reviewed-by: Joe Hacker &lt;jhacker@foo.com&gt;
212</pre>
213or
214<pre>
215    Acked-by: Joe Hacker &lt;jhacker@foo.com&gt;
216</pre>
217<p>
218Rather than saying just "LGTM" or "Seems OK".
219</p>
220
221<p>
222If small changes are suggested, it's OK to say something like:
223</p>
224<pre>
225   With the above fixes, Reviewed-by: Joe Hacker &lt;jhacker@foo.com&gt;
226</pre>
227<p>
228which tells the patch author that the patch can be committed, as long
229as the issues are resolved first.
230</p>
231
232
233<h2 id="nominations">Nominating a commit for a stable branch</h2>
234
235<p>
236There are three ways to nominate a patch for inclusion in the stable branch and
237release.
238</p>
239<ul>
240<li> By adding the Cc: mesa-stable@ tag as described below.
241<li> Sending the commit ID (as seen in master branch) to the mesa-stable@ mailing list.
242<li> Forwarding the patch from the mesa-dev@ mailing list.
243</li>
244</ul>
245<p>
246Note: resending patch identical to one on mesa-dev@ or one that differs only
247by the extra mesa-stable@ tag is <strong>not</strong> recommended.
248</p>
249
250
251<h3 id="thetag">The stable tag</h3>
252
253<p>
254If you want a commit to be applied to a stable branch,
255you should add an appropriate note to the commit message.
256</p>
257
258<p>
259Here are some examples of such a note:
260</p>
261<ul>
262  <li>CC: &lt;mesa-stable@lists.freedesktop.org&gt;</li>
263</ul>
264
265Simply adding the CC to the mesa-stable list address is adequate to nominate
266the commit for all the active stable branches. If the commit is not applicable
267for said branch the stable-release manager will reply stating so.
268
269This "CC" syntax for patch nomination will cause patches to automatically be
270copied to the mesa-stable@ mailing list when you use "git send-email" to send
271patches to the mesa-dev@ mailing list. If you prefer using --suppress-cc that
272won't have any negative effect on the patch nomination.
273
274<p>
275Note: by removing the tag [as the commit is pushed] the patch is
276<strong>explicitly</strong> rejected from inclusion in the stable branch(es).
277<br>
278Thus, drop the line <strong>only</strong> if you want to cancel the nomination.
279</p>
280
281Alternatively, if one uses the "Fixes" tag as described in the "Patch formatting"
282section, it nominates a commit for all active stable branches that include the
283commit that is referred to.
284
285<h2 id="criteria">Criteria for accepting patches to the stable branch</h2>
286
287Mesa has a designated release manager for each stable branch, and the release
288manager is the only developer that should be pushing changes to these branches.
289Everyone else should nominate patches using the mechanism described above.
290
291The following rules define which patches are accepted and which are not. The
292stable-release manager is also given broad discretion in rejecting patches
293that have been nominated.
294
295<ul>
296  <li>Patch must conform with the <a href="#guidelines">Basic guidelines</a></li>
297
298  <li>Patch must have landed in master first. In case where the original
299  patch is too large and/or otherwise contradicts with the rules set within, a
300  backport is appropriate.</li>
301
302  <li>It must not introduce a regression - be that build or runtime wise.
303
304  Note:  If the regression is due to faulty piglit/dEQP/CTS/other test the
305  latter must be fixed first. A reference to the offending test(s) and
306  respective fix(es) should be provided in the nominated patch.</li>
307
308  <li>Patch cannot be larger than 100 lines.</li>
309
310  <li>Patches that move code around with no functional change should be
311  rejected.</li>
312
313  <li>Patch must be a bug fix and not a new feature.
314
315  Note: An exception to this rule, are hardware-enabling "features". For
316  example, <a href="#backports">backports</a> of new code to support a
317  newly-developed hardware product can be accepted if they can be reasonably
318  determined not to have effects on other hardware.</li>
319
320  <li>Patch must be reviewed, For example, the commit message has Reviewed-by,
321  Signed-off-by, or Tested-by tags from someone but the author.</li>
322
323  <li>Performance patches are considered only if they provide information
324  about the hardware, program in question and observed improvement. Use numbers
325  to represent your measurements.</li>
326</ul>
327
328If the patch complies with the rules it will be
329<a href="releasing.html#pickntest">cherry-picked</a>. Alternatively the release
330manager will reply to the patch in question stating why the patch has been
331rejected or would request a backport.
332
333A summary of all the picked/rejected patches will be presented in the
334<a href="releasing.html#prerelease">pre-release</a> announcement.
335
336The stable-release manager may at times need to force-push changes to the
337stable branches, for example, to drop a previously-picked patch that was later
338identified as causing a regression). These force-pushes may cause changes to
339be lost from the stable branch if developers push things directly. Consider
340yourself warned.
341
342<h2 id="backports">Sending backports for the stable branch</h2>
343By default merge conflicts are resolved by the stable-release manager. In which
344case he/she should provide a comment about the changes required, alongside the
345<code>Conflicts</code> section. Summary of which will be provided in the
346<a href="releasing.html#prerelease">pre-release</a> announcement.
347<br>
348Developers are interested in sending backports are recommended to use either a
349<code>[BACKPORT #branch]</code> subject prefix or provides similar information
350within the commit summary.
351
352<h2 id="gittips">Git tips</h2>
353
354<ul>
355<li><code>git rebase -i ...</code> is your friend. Don't be afraid to use it.
356<li>Apply a fixup to commit FOO.
357<pre>
358    git add ...
359    git commit --fixup=FOO
360    git rebase -i --autosquash ...
361</pre>
362<li>Test for build breakage between patches e.g last 8 commits.
363<pre>
364    git rebase -i --exec="make -j4" HEAD~8
365</pre>
366<li>Sets the default mailing address for your repo.
367<pre>
368    git config --local sendemail.to mesa-dev@lists.freedesktop.org
369</pre>
370<li> Add version to subject line of patch series in this case for the last 8
371commits before sending.
372<pre>
373    git send-email --subject-prefix="PATCH v4" HEAD~8
374    git send-email -v4 @~8 # shorter version, inherited from git format-patch
375</pre>
376<li> Configure git to use the get_reviewer.pl script interactively. Thus you
377can avoid adding the world to the CC list.
378<pre>
379    git config sendemail.cccmd "./scripts/get_reviewer.pl -i"
380</pre>
381</ul>
382
383
384</div>
385</body>
386</html>
387