HACKING 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589
  1. Hacking Tor: An Incomplete Guide
  2. ================================
  3. Getting started
  4. ---------------
  5. For full information on how Tor is supposed to work, look at the files in
  6. https://gitweb.torproject.org/torspec.git/tree
  7. For an explanation of how to change Tor's design to work differently, look at
  8. https://gitweb.torproject.org/torspec.git/blob_plain/HEAD:/proposals/001-process.txt
  9. For the latest version of the code, get a copy of git, and
  10. git clone https://git.torproject.org/git/tor
  11. We talk about Tor on the tor-talk mailing list. Design proposals and
  12. discussion belong on the tor-dev mailing list. We hang around on
  13. irc.oftc.net, with general discussion happening on #tor and development
  14. happening on #tor-dev.
  15. How we use Git branches
  16. -----------------------
  17. Each main development series (like 0.2.1, 0.2.2, etc) has its main work
  18. applied to a single branch. At most one series can be the development series
  19. at a time; all other series are maintenance series that get bug-fixes only.
  20. The development series is built in a git branch called "master"; the
  21. maintenance series are built in branches called "maint-0.2.0", "maint-0.2.1",
  22. and so on. We regularly merge the active maint branches forward.
  23. For all series except the development series, we also have a "release" branch
  24. (as in "release-0.2.1"). The release series is based on the corresponding
  25. maintenance series, except that it deliberately lags the maint series for
  26. most of its patches, so that bugfix patches are not typically included in a
  27. maintenance release until they've been tested for a while in a development
  28. release. Occasionally, we'll merge an urgent bugfix into the release branch
  29. before it gets merged into maint, but that's rare.
  30. If you're working on a bugfix for a bug that occurs in a particular version,
  31. base your bugfix branch on the "maint" branch for the first supported series
  32. that has that bug. (As of June 2013, we're supporting 0.2.3 and later.) If
  33. you're working on a new feature, base it on the master branch.
  34. How we log changes
  35. ------------------
  36. When you do a commit that needs a ChangeLog entry, add a new file to
  37. the "changes" toplevel subdirectory. It should have the format of a
  38. one-entry changelog section from the current ChangeLog file, as in
  39. o Major bugfixes:
  40. - Fix a potential buffer overflow. Fixes bug 99999; bugfix on
  41. 0.3.1.4-beta.
  42. To write a changes file, first categorize the change. Some common categories
  43. are: Minor bugfixes, Major bugfixes, Minor features, Major features, Code
  44. simplifications and refactoring. Then say what the change does. If
  45. it's a bugfix, mention what bug it fixes and when the bug was
  46. introduced. To find out which Git tag the change was introduced in,
  47. you can use "git describe --contains <sha1 of commit>".
  48. If at all possible, try to create this file in the same commit where
  49. you are making the change. Please give it a distinctive name that no
  50. other branch will use for the lifetime of your change.
  51. When we go to make a release, we will concatenate all the entries
  52. in changes to make a draft changelog, and clear the directory. We'll
  53. then edit the draft changelog into a nice readable format.
  54. What needs a changes file?::
  55. A not-exhaustive list: Anything that might change user-visible
  56. behavior. Anything that changes internals, documentation, or the build
  57. system enough that somebody could notice. Big or interesting code
  58. rewrites. Anything about which somebody might plausibly wonder "when
  59. did that happen, and/or why did we do that" 6 months down the line.
  60. Why use changes files instead of Git commit messages?::
  61. Git commit messages are written for developers, not users, and they
  62. are nigh-impossible to revise after the fact.
  63. Why use changes files instead of entries in the ChangeLog?::
  64. Having every single commit touch the ChangeLog file tended to create
  65. zillions of merge conflicts.
  66. Useful tools
  67. ------------
  68. These aren't strictly necessary for hacking on Tor, but they can help track
  69. down bugs.
  70. Jenkins
  71. ~~~~~~~
  72. https://jenkins.torproject.org
  73. Dmalloc
  74. ~~~~~~~
  75. The dmalloc library will keep track of memory allocation, so you can find out
  76. if we're leaking memory, doing any double-frees, or so on.
  77. dmalloc -l ~/dmalloc.log
  78. (run the commands it tells you)
  79. ./configure --with-dmalloc
  80. Valgrind
  81. ~~~~~~~~
  82. valgrind --leak-check=yes --error-limit=no --show-reachable=yes src/or/tor
  83. (Note that if you get a zillion openssl warnings, you will also need to
  84. pass --undef-value-errors=no to valgrind, or rebuild your openssl
  85. with -DPURIFY.)
  86. Running lcov for unit test coverage
  87. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  88. Lcov is a utility that generates pretty HTML reports of test code coverage.
  89. To generate such a report:
  90. -----
  91. ./configure --enable-coverage
  92. make
  93. make coverage-html
  94. $BROWSER ./coverage_html/index.html
  95. -----
  96. This will run the tor unit test suite `./src/test/test` and generate the HTML
  97. coverage code report under the directory ./coverage_html/. To change the
  98. output directory, use `make coverage-html HTML_COVER_DIR=./funky_new_cov_dir`.
  99. Coverage diffs using lcov are not currently implemented, but are being
  100. investigated (as of July 2014).
  101. Running the unit tests
  102. ~~~~~~~~~~~~~~~~~~~~~~
  103. To quickly run all tests:
  104. -----
  105. make check
  106. -----
  107. To run unit tests only:
  108. -----
  109. make test
  110. -----
  111. To selectively run just some tests (the following can be combined
  112. arbitrarily):
  113. -----
  114. ./src/test/test <name_of_test> [<name of test 2>] ...
  115. ./src/test/test <prefix_of_name_of_test>.. [<prefix_of_name_of_test2>..] ...
  116. ./src/test/test :<name_of_excluded_test> [:<name_of_excluded_test2]...
  117. -----
  118. Running gcov for unit test coverage
  119. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  120. -----
  121. ./configure --enable-coverage
  122. make
  123. make check
  124. mkdir coverage-output
  125. ./scripts/test/coverage coverage-output
  126. -----
  127. (On OSX, you'll need to start with "--enable-coverage CC=clang".)
  128. Then, look at the .gcov files in coverage-output. '-' before a line means
  129. that the compiler generated no code for that line. '######' means that the
  130. line was never reached. Lines with numbers were called that number of times.
  131. If that doesn't work:
  132. * Try configuring Tor with --disable-gcc-hardening
  133. * You might need to run 'make clean' after you run './configure'.
  134. If you make changes to Tor and want to get another set of coverage results,
  135. you can run "make reset-gcov" to clear the intermediary gcov output.
  136. If you have two different "coverage-output" directories, and you want to see
  137. a meaningful diff between them, you can run:
  138. -----
  139. ./scripts/test/cov-diff coverage-output1 coverage-output2 | less
  140. -----
  141. In this diff, any lines that were visited at least once will have coverage
  142. "1". This lets you inspect what you (probably) really want to know: which
  143. untested lines were changed? Are there any new untested lines?
  144. Running integration tests
  145. ~~~~~~~~~~~~~~~~~~~~~~~~~
  146. We have the beginnings of a set of scripts to run integration tests using
  147. Chutney. To try them, set CHUTNEY_PATH to your chutney source directory, and
  148. run "make test-network".
  149. Profiling Tor with oprofile
  150. ~~~~~~~~~~~~~~~~~~~~~~~~~~~
  151. The oprofile tool runs (on Linux only!) to tell you what functions Tor is
  152. spending its CPU time in, so we can identify berformance pottlenecks.
  153. Here are some basic instructions
  154. - Build tor with debugging symbols (you probably already have, unless
  155. you messed with CFLAGS during the build process).
  156. - Build all the libraries you care about with debugging symbols
  157. (probably you only care about libssl, maybe zlib and Libevent).
  158. - Copy this tor to a new directory
  159. - Copy all the libraries it uses to that dir too (ldd ./tor will
  160. tell you)
  161. - Set LD_LIBRARY_PATH to include that dir. ldd ./tor should now
  162. show you it's using the libs in that dir
  163. - Run that tor
  164. - Reset oprofiles counters/start it
  165. * "opcontrol --reset; opcontrol --start", if Nick remembers right.
  166. - After a while, have it dump the stats on tor and all the libs
  167. in that dir you created.
  168. * "opcontrol --dump;"
  169. * "opreport -l that_dir/*"
  170. - Profit
  171. Coding conventions
  172. ------------------
  173. Patch checklist
  174. ~~~~~~~~~~~~~~~
  175. If possible, send your patch as one of these (in descending order of
  176. preference)
  177. - A git branch we can pull from
  178. - Patches generated by git format-patch
  179. - A unified diff
  180. Did you remember...
  181. - To build your code while configured with --enable-gcc-warnings?
  182. - To run "make check-spaces" on your code?
  183. - To run "make check-docs" to see whether all new options are on
  184. the manpage?
  185. - To write unit tests, as possible?
  186. - To base your code on the appropriate branch?
  187. - To include a file in the "changes" directory as appropriate?
  188. Whitespace and C conformance
  189. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  190. Invoke "make check-spaces" from time to time, so it can tell you about
  191. deviations from our C whitespace style. Generally, we use:
  192. - Unix-style line endings
  193. - K&R-style indentation
  194. - No space before newlines
  195. - A blank line at the end of each file
  196. - Never more than one blank line in a row
  197. - Always spaces, never tabs
  198. - No more than 79-columns per line.
  199. - Two spaces per indent.
  200. - A space between control keywords and their corresponding paren
  201. "if (x)", "while (x)", and "switch (x)", never "if(x)", "while(x)", or
  202. "switch(x)".
  203. - A space between anything and an open brace.
  204. - No space between a function name and an opening paren. "puts(x)", not
  205. "puts (x)".
  206. - Function declarations at the start of the line.
  207. We try hard to build without warnings everywhere. In particular, if you're
  208. using gcc, you should invoke the configure script with the option
  209. "--enable-gcc-warnings". This will give a bunch of extra warning flags to
  210. the compiler, and help us find divergences from our preferred C style.
  211. Getting emacs to edit Tor source properly
  212. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  213. Nick likes to put the following snippet in his .emacs file:
  214. -----
  215. (add-hook 'c-mode-hook
  216. (lambda ()
  217. (font-lock-mode 1)
  218. (set-variable 'show-trailing-whitespace t)
  219. (let ((fname (expand-file-name (buffer-file-name))))
  220. (cond
  221. ((string-match "^/home/nickm/src/libevent" fname)
  222. (set-variable 'indent-tabs-mode t)
  223. (set-variable 'c-basic-offset 4)
  224. (set-variable 'tab-width 4))
  225. ((string-match "^/home/nickm/src/tor" fname)
  226. (set-variable 'indent-tabs-mode nil)
  227. (set-variable 'c-basic-offset 2))
  228. ((string-match "^/home/nickm/src/openssl" fname)
  229. (set-variable 'indent-tabs-mode t)
  230. (set-variable 'c-basic-offset 8)
  231. (set-variable 'tab-width 8))
  232. ))))
  233. -----
  234. You'll note that it defaults to showing all trailing whitespace. The "cond"
  235. test detects whether the file is one of a few C free software projects that I
  236. often edit, and sets up the indentation level and tab preferences to match
  237. what they want.
  238. If you want to try this out, you'll need to change the filename regex
  239. patterns to match where you keep your Tor files.
  240. If you use emacs for editing Tor and nothing else, you could always just say:
  241. -----
  242. (add-hook 'c-mode-hook
  243. (lambda ()
  244. (font-lock-mode 1)
  245. (set-variable 'show-trailing-whitespace t)
  246. (set-variable 'indent-tabs-mode nil)
  247. (set-variable 'c-basic-offset 2)))
  248. -----
  249. There is probably a better way to do this. No, we are probably not going
  250. to clutter the files with emacs stuff.
  251. Functions to use
  252. ~~~~~~~~~~~~~~~~
  253. We have some wrapper functions like tor_malloc, tor_free, tor_strdup, and
  254. tor_gettimeofday; use them instead of their generic equivalents. (They
  255. always succeed or exit.)
  256. You can get a full list of the compatibility functions that Tor provides by
  257. looking through src/common/util.h and src/common/compat.h. You can see the
  258. available containers in src/common/containers.h. You should probably
  259. familiarize yourself with these modules before you write too much code, or
  260. else you'll wind up reinventing the wheel.
  261. Use 'INLINE' instead of 'inline', so that we work properly on Windows.
  262. Calling and naming conventions
  263. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  264. Whenever possible, functions should return -1 on error and 0 on success.
  265. For multi-word identifiers, use lowercase words combined with
  266. underscores. (e.g., "multi_word_identifier"). Use ALL_CAPS for macros and
  267. constants.
  268. Typenames should end with "_t".
  269. Function names should be prefixed with a module name or object name. (In
  270. general, code to manipulate an object should be a module with the same name
  271. as the object, so it's hard to tell which convention is used.)
  272. Functions that do things should have imperative-verb names
  273. (e.g. buffer_clear, buffer_resize); functions that return booleans should
  274. have predicate names (e.g. buffer_is_empty, buffer_needs_resizing).
  275. If you find that you have four or more possible return code values, it's
  276. probably time to create an enum. If you find that you are passing three or
  277. more flags to a function, it's probably time to create a flags argument that
  278. takes a bitfield.
  279. What To Optimize
  280. ~~~~~~~~~~~~~~~~
  281. Don't optimize anything if it's not in the critical path. Right now, the
  282. critical path seems to be AES, logging, and the network itself. Feel free to
  283. do your own profiling to determine otherwise.
  284. Log conventions
  285. ~~~~~~~~~~~~~~~
  286. https://trac.torproject.org/projects/tor/wiki/doc/TorFAQ#loglevel
  287. No error or warning messages should be expected during normal OR or OP
  288. operation.
  289. If a library function is currently called such that failure always means ERR,
  290. then the library function should log WARN and let the caller log ERR.
  291. Every message of severity INFO or higher should either (A) be intelligible
  292. to end-users who don't know the Tor source; or (B) somehow inform the
  293. end-users that they aren't expected to understand the message (perhaps
  294. with a string like "internal error"). Option (A) is to be preferred to
  295. option (B).
  296. Doxygen
  297. ~~~~~~~~
  298. We use the 'doxygen' utility to generate documentation from our
  299. source code. Here's how to use it:
  300. 1. Begin every file that should be documented with
  301. /**
  302. * \file filename.c
  303. * \brief Short description of the file.
  304. **/
  305. (Doxygen will recognize any comment beginning with /** as special.)
  306. 2. Before any function, structure, #define, or variable you want to
  307. document, add a comment of the form:
  308. /** Describe the function's actions in imperative sentences.
  309. *
  310. * Use blank lines for paragraph breaks
  311. * - and
  312. * - hyphens
  313. * - for
  314. * - lists.
  315. *
  316. * Write <b>argument_names</b> in boldface.
  317. *
  318. * \code
  319. * place_example_code();
  320. * between_code_and_endcode_commands();
  321. * \endcode
  322. */
  323. 3. Make sure to escape the characters "<", ">", "\", "%" and "#" as "\<",
  324. "\>", "\\", "\%", and "\#".
  325. 4. To document structure members, you can use two forms:
  326. struct foo {
  327. /** You can put the comment before an element; */
  328. int a;
  329. int b; /**< Or use the less-than symbol to put the comment
  330. * after the element. */
  331. };
  332. 5. To generate documentation from the Tor source code, type:
  333. $ doxygen -g
  334. To generate a file called 'Doxyfile'. Edit that file and run
  335. 'doxygen' to generate the API documentation.
  336. 6. See the Doxygen manual for more information; this summary just
  337. scratches the surface.
  338. Doxygen comment conventions
  339. ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  340. Say what functions do as a series of one or more imperative sentences, as
  341. though you were telling somebody how to be the function. In other words, DO
  342. NOT say:
  343. /** The strtol function parses a number.
  344. *
  345. * nptr -- the string to parse. It can include whitespace.
  346. * endptr -- a string pointer to hold the first thing that is not part
  347. * of the number, if present.
  348. * base -- the numeric base.
  349. * returns: the resulting number.
  350. */
  351. long strtol(const char *nptr, char **nptr, int base);
  352. Instead, please DO say:
  353. /** Parse a number in radix <b>base</b> from the string <b>nptr</b>,
  354. * and return the result. Skip all leading whitespace. If
  355. * <b>endptr</b> is not NULL, set *<b>endptr</b> to the first character
  356. * after the number parsed.
  357. **/
  358. long strtol(const char *nptr, char **nptr, int base);
  359. Doxygen comments are the contract in our abstraction-by-contract world: if
  360. the functions that call your function rely on it doing something, then your
  361. function should mention that it does that something in the documentation. If
  362. you rely on a function doing something beyond what is in its documentation,
  363. then you should watch out, or it might do something else later.
  364. Putting out a new release
  365. -------------------------
  366. Here are the steps Roger takes when putting out a new Tor release:
  367. 1) Use it for a while, as a client, as a relay, as a hidden service,
  368. and as a directory authority. See if it has any obvious bugs, and
  369. resolve those.
  370. 1.5) As applicable, merge the maint-X branch into the release-X branch.
  371. 2) Gather the changes/* files into a changelog entry, rewriting many
  372. of them and reordering to focus on what users and funders would find
  373. interesting and understandable.
  374. 2.1) Make sure that everything that wants a bug number has one.
  375. Make sure that everything which is a bugfix says what version
  376. it was a bugfix on.
  377. 2.2) Concatenate them.
  378. 2.3) Sort them by section. Within each section, sort by "version it's
  379. a bugfix on", else by numerical ticket order.
  380. 2.4) Clean them up:
  381. Standard idioms:
  382. "Fixes bug 9999; bugfix on 0.3.3.3-alpha."
  383. One space after a period.
  384. Make stuff very terse
  385. Make sure each section name ends with a colon
  386. Describe the user-visible problem right away
  387. Mention relevant config options by name. If they're rare or unusual,
  388. remind people what they're for
  389. Avoid starting lines with open-paren
  390. Present and imperative tense: not past.
  391. 'Relays', not 'servers' or 'nodes' or 'Tor relays'.
  392. "Stop FOOing", not "Fix a bug where we would FOO".
  393. Try not to let any given section be longer than about a page. Break up
  394. long sections into subsections by some sort of common subtopic. This
  395. guideline is especially important when organizing Release Notes for
  396. new stable releases.
  397. If a given changes stanza showed up in a different release (e.g.
  398. maint-0.2.1), be sure to make the stanzas identical (so people can
  399. distinguish if these are the same change).
  400. 2.5) Merge them in.
  401. 2.6) Clean everything one last time.
  402. 2.7) Run ./scripts/maint/format_changelog.py to make it prettier.
  403. 3) Compose a short release blurb to highlight the user-facing
  404. changes. Insert said release blurb into the ChangeLog stanza. If it's
  405. a stable release, add it to the ReleaseNotes file too. If we're adding
  406. to a release-0.2.x branch, manually commit the changelogs to the later
  407. git branches too.
  408. 4) Bump the version number in configure.ac and rebuild.
  409. 5) Make dist, put the tarball up somewhere, and tell #tor about it. Wait
  410. a while to see if anybody has problems building it. Try to get Sebastian
  411. or somebody to try building it on Windows.
  412. 6) Get at least two of weasel/arma/sebastian to put the new version number
  413. in their approved versions list.
  414. 7) Sign the tarball, then sign and push the git tag:
  415. gpg -ba <the_tarball>
  416. git tag -u <keyid> tor-0.2.x.y-status
  417. git push origin tag tor-0.2.x.y-status
  418. 8a) scp the tarball and its sig to the dist website, i.e.
  419. /srv/dist-master.torproject.org/htdocs/ on staticiforme. When you want
  420. it to go live, you run "static-update-component dist.torproject.org"
  421. on staticiforme.
  422. 8b)
  423. Edit "include/versions.wmi" and "Makefile" to note the new version.
  424. 9) Email the packagers (cc'ing tor-assistants) that a new tarball is up.
  425. 10) Add the version number to Trac. To do this, go to Trac, log in,
  426. select "Admin" near the top of the screen, then select "Versions" from
  427. the menu on the left. At the right, there will be an "Add version"
  428. box. By convention, we enter the version in the form "Tor:
  429. 0.2.2.23-alpha" (or whatever the version is), and we select the date as
  430. the date in the ChangeLog.
  431. 11) Forward-port the ChangeLog.
  432. 12) Wait up to a day or two (for a development release), or until most
  433. packages are up (for a stable release), and mail the release blurb and
  434. changelog to tor-talk or tor-announce.
  435. (We might be moving to faster announcements, but don't announce until
  436. the website is at least updated.)
  437. 13) If it's a stable release, bump the version number in the maint-x.y.z
  438. branch to "newversion-dev", and do a "merge -s ours" merge to avoid
  439. taking that change into master. Do a similar 'merge -s theirs'
  440. merge to get the change (and only that change) into release. (Some
  441. of the build scripts require that maint merge cleanly into release.)