And when you drive for hours, arrive to find you nowhere gone, you've just been
mouthing "brum, brum", rocking wheel, of course you have, the heap is rusted
through and off the road since you drove drunk through thirteen school yards,
laughing like Prescott.
Then welcome, ah, oo costrinzi welcome, in OProfile.
Please talk to the list before starting on something. We're not too scary.
You can find some documentation on how OProfile works in doc/internals.html
Here's a short list of some stuff you need to know to get started. Don't forget
to read doc/CodingStyle
Source organisation
-------------------
module/
The 2.4 module code. Sub-directories contain architecture-specific code.
daemon/
The daemon. liblegacy/ contains the daemon core for 2.4
utils/
Scripts for managing the daemon etc.
doc/
User and developer documentation
events/
Textual performance counter event descriptions.
libpp/
Classes for handling profiles
pp/
The post-profiling tools for showing results
libabi/
opimport and its ABI support library
libdb/
The sample file access library
libop/
C language oprofile-specific helper stuff
libopt++/
A simple C++ library for parsing command lines
libregex/
C++ demangling pattern matching for smart demangling feature.
libutil/
libutil++/
Generic helpers
gui/
The GUI for starting oprofile
m4/
Autoconf macros for ./configure stage
Tools
-----
You'll need autoconf 2.13+ and automake 2.5+ when using CVS. Don't forget to
autogen.sh first.
We still currently support gcc 2.91.66. Please bear this in mind.
Shell Scripts
-------------
Any shell scripts should aim to be as compatible as possible with different
shells and "bashisms" etc. should not be used. Busybox is often used instead
of bash on embedded devices for example.
Making patches, commit rights
-----------------------------
Patches should be in diff -u format, appliable by patch -p1 in the top-level
source directory. Patches should not include changes to generated files.
Even trivial patches must have a change log entry in the usual format (see
ChangeLog). Refer to bug numbers in the change log if relevant.
When you submit a patch, we ask that you include a "Signed-off-by"
line; for example:
Signed-off-by: Random J Developer <random@developer.example.org>
Including this line with your patch implies that you have read and comply with
the "Developer's Certificate of Origin 1.1" (DCO). This is the same process
the kernel community uses to try to ensure the originality of patches. The
DCO can be found in <kernel-source>/Documentation/SubmittingPatches, item 11,
"Sign your work". For convenience, a copy of the DCO is included below.
Developer's Certificate of Origin 1.1
By making a contribution to this project, I certify that:
(a) The contribution was created in whole or in part by me and I
have the right to submit it under the open source license
indicated in the file; or
(b) The contribution is based upon previous work that, to the best
of my knowledge, is covered under an appropriate open source
license and I have the right under that license to submit that
work with modifications, whether created in whole or in part
by me, under the same open source license (unless I am
permitted to submit under a different license), as indicated
in the file; or
(c) The contribution was provided directly to me by some other
person who certified (a), (b) or (c) and I have not modified
it.
(d) I understand and agree that this project and the contribution
are public and that a record of the contribution (including all
personal information I submit with it, including my sign-off) is
maintained indefinitely and may be redistributed consistent with
this project or the open source license(s) involved.
If you make a change visible to the user in some way, you should check the
website for any needed changes. Patches to oprofile-www CVS are preferred
but a notification of what needs changing is good enough. Any changes that
affect the docs (man-pages or oprofile.xml) must include documentation updates
as appropriate. Also see below.
You may after a while be given direct commit rights. You should be
subscribed to both the main list and the commits mailing list if you do.
Your cvs commit message only needs to briefly describe what your change
does - the change log should have the detailed description. Any
non-trivial change needs approval from either John, Phil or Maynard,
unless stated otherwise. The CVS tree will freeze occassionally for
release, in which case no commits are allowed at all without agreement
of John, Phil, or Maynard. CVS admin changes (-kb, .cvsignore etc.) do
not need a change log, and neither does changes to TODO. If you make a
change that affects the user (feature improvement, new feature, bug fix,
UI change), see the next section.
The oprofile website
--------------------
The oprofile website source is stored in the oprofile-www CVS module, excepting
the doc/ and srcdoc/ directories, which are updated by hand at release time.
The visible website (http://oprofile.sf.net/) must always describe the last
*released* version of OProfile, but the CVS contents should be up to date with
the CVS code. This means that if you make a user-visible change as described
in the last section, you should update the files in oprofile-www and commit.
You can do "cvs update" in home/groups/o/op/oprofile/htdocs/cvs on sourceforge
to get http://oprofile.sf.net/cvs/, so you can check your changes work (and
validate: see http://www.htmlhelp.com/tools/validator/).
Any user-visible change should have a short description in the file
release-notes/release-<nextversion> in the oprofile-www CVS module.
Do not document bug fixes that were not in the last released version.
CVS branches
------------
You may need at some point to do your work on a CVS branch, if it's
particularly invasive. CVS is a PITA in this respect unfortunately. It's
strongly recommended that you merge changes from the trunk to your branch at
regular intervals.
To create a branch, create a branch tag :
cvs rtag -b BRANCH_WHATEVER oprofile
And add a merge tag (in the trunk repository):
cvs rtag BRANCH_WHATEVER_MERGE oprofile
Now make your changes on the branch as you wish. When you want to merge some
fixes from the trunk in your branch, do something like this on a branch
checkout :
cvs update -j BRANCH_WHATEVER_MERGE -j HEAD
Fix up any conflicts and commit it the changes to the branch. Now move the
merge tag along for the next merge (in the trunk repository) :
cvs rtag -F BRANCH_WHATEVER_MERGE oprofile
When the time comes to merge the branch changes back into the trunk, I
recommend just doing a diff -Naur on the two trees, which will make sure CVS
hasn't done anything unusual. Don't forget to list your branch on the website
CVS page.