<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<
meta http-equiv="content-type" content="text/html; charset=utf-8">
<
link rel="stylesheet" type="text/css" href="mesa.css">
<
h1>The Mesa 3D Graphics Library<
/h1>
<
h1>Development Notes<
/h1>
<
h2>Adding Extensions<
/h2>
To add a new GL extension to Mesa you have to do at least the following.
If glext.h doesn't define the extension, edit include/GL/gl.h and add
code like this:
<pre>
#ifndef GL_EXT_the_extension_name
#define GL_EXT_the_extension_name 1
/* declare the new enum tokens */
/* prototype the new functions */
/* TYPEDEFS for the new functions */
#endif
</pre>
</li>
<li>
In the src/mapi/glapi/gen/ directory, add the new extension functions and
enums to the gl_API.xml file.
Then, a bunch of source files must be regenerated by executing the
corresponding Python scripts.
</li>
<li>
Add a new entry to the <code>gl_extensions</code> struct in mtypes.h
</li>
<li>
Update the <code>extensions.c</code> file.
</li>
<li>
From this point, the best way to proceed is to find another extension,
similar to the new one, that's already implemented in Mesa and use it
as an example.
If the new extension adds new GL state, the functions in get.c, enable.c
and attrib.c will most likely require new code.
The dispatch tests check_table.cpp and dispatch_sanity.cpp
should be updated with details about the new extensions functions. These
tests are run using 'make check'
Mesa's code style has changed over the years. Here's the latest.
Comment your code! It's extremely important that open-source code be
well documented. Also, strive to write clean, easily understandable code.
</p>
<p>
3-space indentation
</p>
<p>
If you use tabs, set them to 8 columns
</p>
<p>
Line width: the preferred width to fill comments and code in Mesa is 78
columns. Exceptions are sometimes made for clarity (e.g. tabular data is
sometimes filled to a much larger width so that extraneous carriage returns
don't obscure the table).
Brace example:
if (condition) {
foo;
}
else {
bar;
}
switch (condition) {
case 0:
foo();
break;
case 1: {
...
break;
}
default:
...
break;
}
Here's the GNU indent command which will best approximate my preferred style:
(Note that it won't format switch statements in the preferred way)
indent -br -i3 -npcs --no-tabs infile.c -o outfile.c
Local variable name example: localVarName (no underscores)
Constants and macros are ALL_UPPERCASE, with _ between words
Global variables are not allowed.
Function name examples:
glFooBar() - a public GL entry point (in glapi_dispatch.c)
_mesa_FooBar() - the internal immediate mode function
save_FooBar() - retained mode (display list) function in dlist.c
foo_bar() - a static (private) function
_mesa_foo_bar() - an internal non-static Mesa function
Places that are not directly visible to the GL API should prefer the use
<
tt>false<
/tt> over <
tt>GLboolean<
/tt>, <
tt>GL_TRUE<
/tt>, and
<
tt>GL_FALSE<
/tt>. In C
code, this may mean that
<
tt>#include <stdbool.h><
/tt> needs to be added. The
<
tt>try_emit_<
/tt>* methods in
src/mesa
/program
/ir_to_mesa.cpp and
src/mesa/state_tracker/st_glsl_to_tgsi.cpp can serve as examples.
<
h2>Submitting patches<
/h2>
You should always run the Mesa Testsuite before submitting patches.
The Testsuite can be run using the 'make check' command. All tests
must pass before patches will be accepted, this may mean you have
to update the tests themselves.
Patches should be sent to the Mesa mailing list for review.
When submitting a patch make sure to use git send-email rather than attaching
patches to emails. Sending patches as attachments prevents people from being
able to provide in-line review comments.
When submitting follow-up patches you can use --in-reply-to to make v2, v3,
etc patches show up as replies to the originals. This usually works well
when you're sending out updates to individual patches (as opposed to
re-sending the whole series). Using --in-reply-to makes
it harder for reviewers to accidentally review old patches.
</p>
<h2>Marking a commit as a candidate for a stable branch</h2>
<p>
If you want a commit to be applied to a stable branch,
you should add an appropriate note to the commit message.
</p>
<p>
Here are some examples of such a note:
</p>
<ul>
<li>CC: <mesa-stable@lists.freedesktop.org></li>
<li>CC: "9.2 10.0" <mesa-stable@lists.freedesktop.org></li>
<li>CC: "10.0" <mesa-stable@lists.freedesktop.org></li>
</ul>
Simply adding the CC to the mesa-stable list address is adequate to nominate
the commit for the most-recently-created stable branch. It is only necessary
to specify a specific branch name, (such as "9.2 10.0" or "10.0" in the
examples above), if you want to nominate the commit for an older stable
branch. And, as in these examples, you can nominate the commit for the older
branch in addition to the more recent branch, or nominate the commit
exclusively for the older branch.
This "CC" syntax for patch nomination will cause patches to automatically be
copied to the mesa-stable@ mailing list when you use "git send-email" to send
patches to the mesa-dev@ mailing list. Also, if you realize that a commit
should be nominated for the stable branch after it has already been committed,
you can send a note directly to the mesa-stable@lists.freedesktop.org where
the Mesa stable-branch maintainers will receive it. Be sure to mention the
commit ID of the commit of interest (as it appears in the mesa master branch).
The latest set of patches that have been nominated, accepted, or rejected for
the upcoming stable release can always be seen on the
<a href="http://cworth.org/~cworth/mesa-stable-queue/">Mesa Stable Queue</a>
page.
<h2>Criteria for accepting patches to the stable branch</h2>
Mesa has a designated release manager for each stable branch, and the release
manager is the only developer that should be pushing changes to these
branches. Everyone else should simply nominate patches using the mechanism
described above.
The stable-release manager will work with the list of nominated patches, and
for each patch that meets the crtieria below will cherry-pick the patch with:
<code>git cherry-pick -x <commit></code>. The <code>-x</code> option is
important so that the picked patch references the comit ID of the original
patch.
The stable-release manager may at times need to force-push changes to the
stable branches, for example, to drop a previously-picked patch that was later
identified as causing a regression). These force-pushes may cause changes to
be lost from the stable branch if developers push things directly. Consider
yourself warned.
The stable-release manager is also given broad discretion in rejecting patches
that have been nominated for the stable branch. The most basic rule is that
the stable branch is for bug fixes only, (no new features, no
regressions). Here is a non-exhaustive list of some reasons that a patch may
be rejected:
<ul>
<li>Patch introduces a regression. Any reported build breakage or other
regression caused by a particular patch, (game no longer work, piglit test
changes from PASS to FAIL), is justification for rejecting a patch.</li>
<li>Patch is too large, (say, larger than 100 lines)</li>
<li>Patch is not a fix. For example, a commit that moves code around with no
functional change should be rejected.</li>
<li>Patch fix is not clearly described. For example, a commit message
of only a single line, no description of the bug, no mention of bugzilla,
etc.</li>
<li>Patch has not obviously been reviewed, For example, the commit message
has no Reviewed-by, Signed-off-by, nor Tested-by tags from anyone but the
author.</li>
<li>Patch has not already been merged to the master branch. As a rule, bug
fixes should never be applied first to a stable branch. Patches should land
first on the master branch and then be cherry-picked to a stable
branch. (This is to avoid future releases causing regressions if the patch
is not also applied to master.) The only things that might look like
exceptions would be backports of patches from master that happen to look
significantly different.</li>
<li>Patch depends on too many other patches. Ideally, all stable-branch
patches should be self-contained. It sometimes occurs that a single, logical
bug-fix occurs as two separate patches on master, (such as an original
patch, then a subsequent fix-up to that patch). In such a case, these two
patches should be squashed into a single, self-contained patch for the
stable branch. (Of course, if the squashing makes the patch too large, then
that could be a reason to reject the patch.)</li>
<li>Patch includes new feature development, not bug fixes. New OpenGL
features, extensions, etc. should be applied to Mesa master and included in
the next major release. Stable releases are intended only for bug fixes.
Note: As an exception to this rule, the stable-release manager may accept
hardware-enabling "features". For example, backports of new code to support
a newly-developed hardware product can be accepted if they can be reasonably
determined to not have effects on other hardware.</li>
<li>Patch is a performance optimization. As a rule, performance patches are
not candidates for the stable branch. The only exception might be a case
where an application's performance was recently severely impacted so as to
become unusable. The fix for this performance regression could then be
considered for a stable branch. The optimization must also be
non-controversial and the patches still need to meet the other criteria of
being simple and self-contained<
/li>
<
li>Patch introduces a new failure mode
(such as an assert
). While the new
assert might technically be correct, for example to make Mesa more
conformant, this is not the kind of "bug fix" we want in a stable
release. The potential problem here is that an OpenGL program that was
previously working, (even if technically non-compliant with the
specification), could stop working after this patch. So that would be a
regression that is unaacceptable
for the stable branch.<
/li>
<
h2>Making a New Mesa Release<
/h2>
These are the instructions for making a new Mesa release.
<
h3>Get latest source files<
/h3>
Use git to get the latest Mesa files from the git repository, from whatever
branch is relevant. This document uses the convention X.Y.Z for the release
being created, which should be created from a branch named X.Y.
<
h3>Perform basic testing<
/h3>
The release manager should, at the very least, test the code by compiling it,
installing it, and running the latest piglit to ensure that no piglit tests
have regressed since the previous release.
The release manager should do this testing with at least one hardware driver,
(say, whatever is contained in the local development machine), as well as on
both Gallium and non-Gallium software drivers. The software testing can be
performed by running piglit with the following environment-variable set:
LIBGL_ALWAYS_SOFTWARE=1
And Gallium vs. non-Gallium software drivers can be obtained by using the
following configure flags on separate builds:
--with-dri-drivers=swrast
--with-gallium-drivers=swrast
Note: If both options are given in one build, both swrast_dri.so drivers will
be compiled, but only one will be installed. The following command can be used
to ensure the correct driver is being tested:
LIBGL_ALWAYS_SOFTWARE=1 glxinfo | grep "renderer string"
If any regressions are found in this testing with piglit, stop here, and do
not perform a release until regressions are fixed.
<
h3>Update
version in file VERSION<
/h3>
Increment the version contained in the file VERSION at Mesa's top-level, then
commit this change.
</p>
<h3>Create release notes for the new release</h3>
<p>
Create a new file docs/relnotes/X.Y.Z.html, (follow the style of the previous
release notes). Note that the sha256sums section of the release notes should
be empty at this point.
</p>
<p>
Two scripts are available to help generate portions of the release notes:
<pre>
./bin/bugzilla_mesa.sh
./bin/shortlog_mesa.sh
</pre>
<p>
The first script identifies commits that reference bugzilla bugs and obtains
the descriptions of those bugs from bugzilla. The second script generates a
log of all commits. In both cases, HTML-formatted lists are printed to stdout
to be included in the release notes.
</p>
<p>
Commit these changes
</p>
<h3>Make the release archives, signatures, and the release tag</h3>
<p>
From inside the Mesa directory:
<pre>
./autogen.sh
make -j1 tarballs
</pre>
<p>
After the tarballs are created, the sha256 checksums for the files will
be computed and printed. These will be used in a step below.
</p>
<p>
It's important at this point to also verify that the constructed tar file
actually builds:
tar xjf MesaLib-X.Y.Z.tar.bz2
cd Mesa-X.Y.Z
./configure --enable-gallium-llvm
make -j6
make install
Some touch testing should also be performed at this point, (run glxgears or
more involved OpenGL programs against the installed Mesa).
Create detached GPG signatures for each of the archive files created above:
gpg --sign --detach MesaLib-X.Y.Z.tar.gz
gpg --sign --detach MesaLib-X.Y.Z.tar.bz2
gpg --sign --detach MesaLib-X.Y.Z.zip
Tag the commit used for the build:
git tag -s mesa-X.Y.X -m "Mesa X.Y.Z release"
Note: It would be nice to investigate and fix the issue that causes the
tarballs target to fail with multiple build process, such as with "-j4". It
would also be nice to incorporate all of the above commands into a single
makefile target. And instead of a custom "tarballs" target, we should
incorporate things into the standard "make dist" and "make distcheck" targets.
<
h3>Add the sha256sums to the release notes<
/h3>
Edit docs/relnotes/X.Y.Z.html to add the sha256sums printed as part of "make
tarballs" in the previous step. Commit this change.
<
h3>Push all commits and the tag creates above<
/h3>
This is the first step that cannot easily be undone. The release is going
forward from this point:
git push origin X.Y --tags
<
h3>Install the release files and signatures on the distribution server<
/h3>
The following commands can be used to copy the release archive files and
signatures to the freedesktop.org server:
scp MesaLib-X.Y.Z* people.freedesktop.org:
ssh people.freedesktop.org
cd /srv/ftp.freedesktop.org/pub/mesa
mkdir X.Y.Z
cd X.Y.Z
mv ~/MesaLib-X.Y.Z* .
<
h3>Back on mesa master, andd the new release notes into the tree<
/h3>
Something like the following steps will do the trick:
cp docs/relnotes/X.Y.Z.html /tmp
git checkout master
cp /tmp/X.Y.Z.html docs/relnotes
git add docs/relnotes/X.Y.Z.html
Also, edit docs/relnotes.html to add a link to the new release notes, and edit
docs/index.html to add a news entry. Then commit and push:
git commit -a -m "docs: Import X.Y.Z release notes, add news item."
git push origin
<
h3>Update the mesa3d.org website<
/h3>
NOTE: The recent release managers have not been performing this step
themselves, but leaving this to Brian Paul, (who has access to the
sourceforge.net hosting for mesa3d.org). Brian is more than willing to grant
the permission necessary to future release managers to do this step on their
own.
Update the web site by copying the docs/ directory's files to
/home/users/b/br/brianp/mesa-www/htdocs/ with:
<br>
<code>
sftp USERNAME,mesa3d@web.sourceforge.net
</code>
</p>
<h3>Announce the release</h3>
<p>
Make an announcement on the mailing lists:
<em>mesa-dev@lists.freedesktop.org</em>,
and
<em>mesa-announce@lists.freedesktop.org</em>
Follow the template of previously-sent release announcements. The following
command can be used to generate the log of changes to be included in the
release announcement:
<pre>
git shortlog mesa-X.Y.Z-1..mesa-X.Y.Z
</pre>
</p>
</div>
</body>
</html>