docs/submittingpatches.html

   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="#gittips">Git tips</a>
  29 </ul>
  30
  31 <h2 id="guidelines">Basic guidelines</h2>
  32
  33 <ul>
  34 <li>Patches should not mix code changes with code formatting changes (except,
  35 perhaps, in very trivial cases.)
  36 <li>Code patches should follow Mesa
  37 <a href="codingstyle.html" target="_parent">coding conventions</a>.
  38 <li>Whenever possible, patches should only effect individual Mesa/Gallium
  39 components.
  40 <li>Patches should never introduce build breaks and should be bisectable (see
  41 <code>git bisect</code>.)
  42 <li>Patches should be properly <a href="#formatting">formatted</a>.
  43 <li>Patches should be sufficiently <a href="#testing">tested</a> before submitting.
  44 <li>Patches should be submitted to <a href="#mailing">mesa-dev</a>
  45 for <a href="#reviewing">review</a> using <code>git send-email</code>.
  46
  47 </ul>
  48
  49 <h2 id="formatting">Patch formatting</h2>
  50
  51 <ul>
  52 <li>Lines should be limited to 75 characters or less so that git logs
  53 displayed in 80-column terminals avoid line wrapping.  Note that git
  54 log uses 4 spaces of indentation (4 + 75 &lt; 80).
  55 <li>The first line should be a short, concise summary of the change prefixed
  56 with a module name.  Examples:
  57 <pre>
  58     mesa: Add support for querying GL_VERTEX_ATTRIB_ARRAY_LONG
  59
  60     gallium: add PIPE_CAP_DEVICE_RESET_STATUS_QUERY
  61
  62     i965: Fix missing type in local variable declaration.
  63 </pre>
  64 <li>Subsequent patch comments should describe the change in more detail,
  65 if needed.  For example:
  66 <pre>
  67     i965: Remove end-of-thread SEND alignment code.
  68
  69     This was present in Eric's initial implementation of the compaction code
  70     for Sandybridge (commit 077d01b6). There is no documentation saying this
  71     is necessary, and removing it causes no regressions in piglit on any
  72     platform.
  73 </pre>
  74 <li>A "Signed-off-by:" line is not required, but not discouraged either.
  75 <li>If a patch address a bugzilla issue, that should be noted in the
  76 patch comment.  For example:
  77 <pre>
  78    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89689
  79 </pre>
  80 <li>If there have been several revisions to a patch during the review
  81 process, they should be noted such as in this example:
  82 <pre>
  83     st/mesa: add ARB_texture_stencil8 support (v4)
  84
  85     if we support stencil texturing, enable texture_stencil8
  86     there is no requirement to support native S8 for this,
  87     the texture can be converted to x24s8 fine.
  88
  89     v2: fold fixes from Marek in:
  90        a) put S8 last in the list
  91        b) fix renderable to always test for d/s renderable
  92         fixup the texture case to use a stencil only format
  93         for picking the format for the texture view.
  94     v3: hit fallback for getteximage
  95     v4: put s8 back in front, it shouldn't get picked now (Ilia)
  96 </pre>
  97 <li>If someone tested your patch, document it with a line like this:
  98 <pre>
  99     Tested-by: Joe Hacker &lt;jhacker@foo.com&gt;
 100 </pre>
 101 <li>If the patch was reviewed (usually the case) or acked by someone,
 102 that should be documented with:
 103 <pre>
 104     Reviewed-by: Joe Hacker &lt;jhacker@foo.com&gt;
 105     Acked-by: Joe Hacker &lt;jhacker@foo.com&gt;
 106 </pre>
 107 <li>If sending later revision of a patch, add all the tags - ack, r-b,
 108 Cc: mesa-stable and/or other. This provides reviewers with quick feedback if the
 109 patch has already been reviewed.
 110 <li>In order for your patch to reach the prospective reviewer easier/faster,
 111 use the script scripts/get_reviewer.pl to get a list of individuals and include
 112 them in the CC list.
 113 <br>
 114 Please use common sense and do <strong>not</strong> blindly add everyone.
 115 <br>
 116 <pre>
 117     $ scripts/get_reviewer.pl --help # to get the the help screen
 118     $ scripts/get_reviewer.pl -f src/egl/drivers/dri2/platform_android.c
 119     Rob Herring <robh@kernel.org> (reviewer:ANDROID EGL SUPPORT,added_lines:188/700=27%,removed_lines:58/283=20%)
 120     Tomasz Figa <tfiga@chromium.org> (reviewer:ANDROID EGL SUPPORT,authored:12/41=29%,added_lines:308/700=44%,removed_lines:115/283=41%)
 121     Emil Velikov <emil.l.velikov@gmail.com> (authored:13/41=32%,removed_lines:76/283=27%)
 122 </pre>
 123 </ul>
 124
 125
 126
 127 <h2 id="testing">Testing Patches</h2>
 128
 129 <p>
 130 It should go without saying that patches must be tested.  In general,
 131 do whatever testing is prudent.
 132 </p>
 133
 134 <p>
 135 You should always run the Mesa test suite before submitting patches.
 136 The test suite can be run using the 'make check' command. All tests
 137 must pass before patches will be accepted, this may mean you have
 138 to update the tests themselves.
 139 </p>
 140
 141 <p>
 142 Whenever possible and applicable, test the patch with
 143 <a href="http://piglit.freedesktop.org">Piglit</a> and/or
 144 <a href="https://android.googlesource.com/platform/external/deqp/">dEQP</a>
 145 to check for regressions.
 146 </p>
 147
 148
 149 <h2 id="mailing">Mailing Patches</h2>
 150
 151 <p>
 152 Patches should be sent to the mesa-dev mailing list for review:
 153 <a href="https://lists.freedesktop.org/mailman/listinfo/mesa-dev">
 154 mesa-dev@lists.freedesktop.org</a>.
 155 When submitting a patch make sure to use
 156 <a href="https://git-scm.com/docs/git-send-email">git send-email</a>
 157 rather than attaching patches to emails. Sending patches as
 158 attachments prevents people from being able to provide in-line review
 159 comments.
 160 </p>
 161
 162 <p>
 163 When submitting follow-up patches you can use --in-reply-to to make v2, v3,
 164 etc patches show up as replies to the originals. This usually works well
 165 when you're sending out updates to individual patches (as opposed to
 166 re-sending the whole series). Using --in-reply-to makes
 167 it harder for reviewers to accidentally review old patches.
 168 </p>
 169
 170 <p>
 171 When submitting follow-up patches you should also login to
 172 <a href="https://patchwork.freedesktop.org">patchwork</a> and change the
 173 state of your old patches to Superseded.
 174 </p>
 175
 176 <h2 id="reviewing">Reviewing Patches</h2>
 177
 178 <p>
 179 When you've reviewed a patch on the mailing list, please be unambiguous
 180 about your review.  That is, state either
 181 </p>
 182 <pre>
 183     Reviewed-by: Joe Hacker &lt;jhacker@foo.com&gt;
 184 </pre>
 185 or
 186 <pre>
 187     Acked-by: Joe Hacker &lt;jhacker@foo.com&gt;
 188 </pre>
 189 <p>
 190 Rather than saying just "LGTM" or "Seems OK".
 191 </p>
 192
 193 <p>
 194 If small changes are suggested, it's OK to say something like:
 195 </p>
 196 <pre>
 197    With the above fixes, Reviewed-by: Joe Hacker &lt;jhacker@foo.com&gt;
 198 </pre>
 199 <p>
 200 which tells the patch author that the patch can be committed, as long
 201 as the issues are resolved first.
 202 </p>
 203
 204
 205 <h2 id="nominations">Nominating a commit for a stable branch</h2>
 206
 207 <p>
 208 There are three ways to nominate patch for inclusion of the stable branch and
 209 release.
 210 </p>
 211 <ul>
 212 <li> By adding the Cc: mesa-stable@ tag as described below.
 213 <li> Sending the commit ID (as seen in master branch) to the mesa-stable@ mailing list.
 214 <li> Forwarding the patch from the mesa-dev@ mailing list.
 215 </li>
 216 </ul>
 217 <p>
 218 Note: resending patch identical to one on mesa-dev@ or one that differs only
 219 by the extra mesa-stable@ tag is <strong>not</strong> recommended.
 220 </p>
 221
 222
 223 <h3 id="thetag">The stable tag</h3>
 224
 225 <p>
 226 If you want a commit to be applied to a stable branch,
 227 you should add an appropriate note to the commit message.
 228 </p>
 229
 230 <p>
 231 Here are some examples of such a note:
 232 </p>
 233 <ul>
 234   <li>CC: &lt;mesa-stable@lists.freedesktop.org&gt;</li>
 235   <li>CC: "9.2 10.0" &lt;mesa-stable@lists.freedesktop.org&gt;</li>
 236   <li>CC: "10.0" &lt;mesa-stable@lists.freedesktop.org&gt;</li>
 237 </ul>
 238
 239 Simply adding the CC to the mesa-stable list address is adequate to nominate
 240 the commit for the most-recently-created stable branch. It is only necessary
 241 to specify a specific branch name, (such as "9.2 10.0" or "10.0" in the
 242 examples above), if you want to nominate the commit for an older stable
 243 branch. And, as in these examples, you can nominate the commit for the older
 244 branch in addition to the more recent branch, or nominate the commit
 245 exclusively for the older branch.
 246
 247 This "CC" syntax for patch nomination will cause patches to automatically be
 248 copied to the mesa-stable@ mailing list when you use "git send-email" to send
 249 patches to the mesa-dev@ mailing list. If you prefer using --suppress-cc that
 250 won't have any effect negative effect on the patch nomination.
 251
 252 <p>
 253 Note: by removing the tag [as the commit is pushed] the patch is
 254 <strong>explicitly</strong> rejected from inclusion in the stable branch(es).
 255 <br>
 256 Thus, drop the line <strong>only</strong> if you want to cancel the nomination.
 257 </p>
 258
 259 <h2 id="criteria">Criteria for accepting patches to the stable branch</h2>
 260
 261 Mesa has a designated release manager for each stable branch, and the release
 262 manager is the only developer that should be pushing changes to these
 263 branches. Everyone else should simply nominate patches using the mechanism
 264 described above.
 265
 266 The stable-release manager will work with the list of nominated patches, and
 267 for each patch that meets the criteria below will cherry-pick the patch with:
 268 <code>git cherry-pick -x &lt;commit&gt;</code>. The <code>-x</code> option is
 269 important so that the picked patch references the commit ID of the original
 270 patch.
 271
 272 The stable-release manager may at times need to force-push changes to the
 273 stable branches, for example, to drop a previously-picked patch that was later
 274 identified as causing a regression). These force-pushes may cause changes to
 275 be lost from the stable branch if developers push things directly. Consider
 276 yourself warned.
 277
 278 The stable-release manager is also given broad discretion in rejecting patches
 279 that have been nominated for the stable branch. The most basic rule is that
 280 the stable branch is for bug fixes only, (no new features, no
 281 regressions). Here is a non-exhaustive list of some reasons that a patch may
 282 be rejected:
 283
 284 <ul>
 285   <li>Patch introduces a regression. Any reported build breakage or other
 286   regression caused by a particular patch, (game no longer work, piglit test
 287   changes from PASS to FAIL), is justification for rejecting a patch.</li>
 288
 289   <li>Patch is too large, (say, larger than 100 lines)</li>
 290
 291   <li>Patch is not a fix. For example, a commit that moves code around with no
 292   functional change should be rejected.</li>
 293
 294   <li>Patch fix is not clearly described. For example, a commit message
 295   of only a single line, no description of the bug, no mention of bugzilla,
 296   etc.</li>
 297
 298   <li>Patch has not obviously been reviewed, For example, the commit message
 299   has no Reviewed-by, Signed-off-by, nor Tested-by tags from anyone but the
 300   author.</li>
 301
 302   <li>Patch has not already been merged to the master branch. As a rule, bug
 303   fixes should never be applied first to a stable branch. Patches should land
 304   first on the master branch and then be cherry-picked to a stable
 305   branch. (This is to avoid future releases causing regressions if the patch
 306   is not also applied to master.) The only things that might look like
 307   exceptions would be backports of patches from master that happen to look
 308   significantly different.</li>
 309
 310   <li>Patch depends on too many other patches. Ideally, all stable-branch
 311   patches should be self-contained. It sometimes occurs that a single, logical
 312   bug-fix occurs as two separate patches on master, (such as an original
 313   patch, then a subsequent fix-up to that patch). In such a case, these two
 314   patches should be squashed into a single, self-contained patch for the
 315   stable branch. (Of course, if the squashing makes the patch too large, then
 316   that could be a reason to reject the patch.)</li>
 317
 318   <li>Patch includes new feature development, not bug fixes. New OpenGL
 319   features, extensions, etc. should be applied to Mesa master and included in
 320   the next major release. Stable releases are intended only for bug fixes.
 321
 322   Note: As an exception to this rule, the stable-release manager may accept
 323   hardware-enabling "features". For example, backports of new code to support
 324   a newly-developed hardware product can be accepted if they can be reasonably
 325   determined to not have effects on other hardware.</li>
 326
 327   <li>Patch is a performance optimization. As a rule, performance patches are
 328   not candidates for the stable branch. The only exception might be a case
 329   where an application's performance was recently severely impacted so as to
 330   become unusable. The fix for this performance regression could then be
 331   considered for a stable branch. The optimization must also be
 332   non-controversial and the patches still need to meet the other criteria of
 333   being simple and self-contained</li>
 334
 335   <li>Patch introduces a new failure mode (such as an assert). While the new
 336   assert might technically be correct, for example to make Mesa more
 337   conformant, this is not the kind of "bug fix" we want in a stable
 338   release. The potential problem here is that an OpenGL program that was
 339   previously working, (even if technically non-compliant with the
 340   specification), could stop working after this patch. So that would be a
 341   regression that is unacceptable for the stable branch.</li>
 342 </ul>
 343
 344 <h2 id="gittips">Git tips</h2>
 345
 346 <ul>
 347 <li><code>git rebase -i ...</code> is your friend. Don't be afraid to use it.
 348 <li>Apply a fixup to commit FOO.
 349 <pre>
 350     git add ...
 351     git commit --fixup=FOO
 352     git rebase -i --autosquash ...
 353 </pre>
 354 <li>Test for build breakage between patches e.g last 8 commits.
 355 <pre>
 356     git rebase -i --exec="make -j4" HEAD~8
 357 </pre>
 358 <li>Sets the default mailing address for your repo.
 359 <pre>
 360     git config --local sendemail.to mesa-dev@lists.freedesktop.org
 361 </pre>
 362 <li> Add version to subject line of patch series in this case for the last 8
 363 commits before sending.
 364 <pre>
 365     git send-email --subject-prefix="PATCH v4" HEAD~8
 366     git send-email -v4 @~8 # shorter version, inherited from git format-patch
 367 </pre>
 368 <li> Configure git to use the get_reviewer.pl script interactively. Thus you
 369 can avoid adding the world to the CC list.
 370 <pre>
 371     git config sendemail.cccmd "./scripts/get_reviewer.pl -i"
 372 </pre>
 373 </ul>
 374
 375
 376 </div>
 377 </body>
 378 </html>