4 SPEC - Spec not finalized
16 X <nickm> "Let's try to find a way to make it run and make the version
17 match, but if not, let's just make it run."
18 X <arma> "should we detect if we have a --with-ssl-dir and try the -R
19 by default, if it works?"
21 Items for 0.1.2.x, real soon now:
22 ? - Bug: combination of things:
23 When we've been idle a long time, we stop fetching server
24 descriptors. When we then get a socks request, we build circuits
25 immediately using whatever descriptors we have, rather than waiting
26 until we've fetched correct ones.
28 N - Test guard unreachable logic; make sure that we actually attempt to
29 connect to guards that we think are unreachable from time to time.
30 Make sure that we don't freak out when the network is down.
32 R - Reconstruct ChangeLog; put rolled-up info in ReleaseNotes or something.
35 N - enumerate events of important things that occur in tor, so vidalia can
37 o Backend implementation
38 - Actually list all the events (notice and warn log messages are a good
39 place to look.) Divide messages into categories, perhaps.
40 - Specify general event system
41 - Specify actual events.
42 - and implement the rest
44 . Have (and document) a BEGIN_DIR relay cell that means "Connect to your
47 R - turn the received socks addr:port into a digest for setting .exit
48 - be able to connect without having a server descriptor, to bootstrap.
49 R - handle connect-dir streams that don't have a chosen_exit_name set.
50 o include ORPort in DirServers lines so we can know where to connect.
51 list the orport as 0 if it can't handle begin_dir.
52 - List orports of actual dirservers..
54 - Servers are easy to setup and run: being a relay is about as easy as
56 . Reduce resource load
57 o A way to alert controller when router flags change.
58 o Specify: SETEVENTS NS
60 R - Hunt for places that change networkstatus info that I might have
63 . option to dl directory info via tor:
64 TunnelDirConns and PreferTunneledDirConns
65 R - actually cause the directory.c functions to know about or_port
66 and use it when we're supposed to.
67 N - for tunneled edge conns, stop reading to the bridge connection
68 when the or_conn we're writing to has a full outbuf.
72 - Make evdns use windows strerror equivalents.
73 - Make sure patches get into libevent.
74 - Verify that it works well on windows
75 - Debug and re-enable server-side reverse DNS caching
77 - Critical but minor bugs, backport candidates.
78 - support dir 503s better
79 o clients don't log as loudly when they receive them
80 o they don't count toward the 3-strikes rule
81 D But eventually, we give up after getting a lot of 503s.
82 D Delay when we get a lot of 503s, rather than punting onto the
83 servers that have given us 503s?
84 o Add a 'BadDirectory' flag to statuses.
85 o authorities should *never* 503 a cache, and should never 503
86 network status requests.
87 D They can 503 client descriptor requests when they feel like it.
88 How can they distinguish? Not implemented for now, maybe
90 - update dir-spec with what we decided for each of these
92 o Have a mode that doesn't write to disk much, so we can run Tor on
93 flash memory (e.g. Linksys routers or USB keys).
94 o Add AvoidDiskWrites config option.
95 o only write state file when it's "changed"
96 o crank up the numbers if avoiddiskwrites is on.
97 D some things may not want to get written at all.
98 o stop writing fingerprint every restart
101 NR. Write path-spec.txt
104 - Profile client and server; fix slow spots
105 - Address XXX012 items
108 - Tell people about OSX Uninstaller
109 - Quietly document NT Service options
110 - Switch canonical win32 compiler to mingw.
111 NR D Get some kind of "meta signing key" to be used solely to sign
112 releases/to certify releases when signed by the right people/
113 to certify sign the right people's keys? Also use this to cert the SSL
115 - If we haven't replaced privoxy, lock down its configuration in all
116 packages, as documented in tor-doc-unix.html
119 - More prominently, we should have a recommended apps list.
121 - unrecommend IE because of ftp:// bug.
122 N - torrc.complete.in needs attention?
123 N - we should add a preamble to tor-design saying it's out of date.
124 N - Document transport and natdport
125 N - Look into generating torrc.{complete|sample}.in, tor.1.in,
126 the HTML manual, and the online config documentation from a single
129 - Improvements to bandwidth counting
130 R - look into "uncounting" bytes spent on local connections, so
131 we can bandwidthrate but still have fast downloads.
132 R - "bandwidth classes", for incoming vs initiated-here conns,
133 and to give dir conns lower priority.
134 . Write limiting; separate token bucket for write
135 o preemptively give a 503 to some v1 dir requests
136 o preemptively give a 503 to some v2 dir requests
137 o Write function to estimate bytes needed for N descriptors
139 D per-conn write buckets
140 D separate config options for read vs write limiting
141 (It's hard to support read > write, since we need better
142 congestion control to avoid overfull buffers there. So,
143 defer the whole thing.)
145 - Forward compatibility fixes
146 - Caches should start trying to cache consensus docs?
148 N - Implement, if we think it's smart.
149 - Start uploading short and long descriptors; authorities should support
150 URLs to retrieve long descriptors, and should discard short descriptors
151 for now. Later, once tools use the "long descriptor" URLs, authorities
152 will serve the short descriptors every time they're asked for
155 N - Implement, if we think it's smart.
157 Topics to think about during 0.1.2.x development:
158 * Figure out incentives.
159 - (How can we make this tolerant of a bad v0?)
160 * Figure out non-clique.
162 - Figure out partial network knowledge.
163 - Figure out hidden services.
164 - Design next-version protocol for directories
165 - Design next-version protocol for connections
167 Deferred from 0.1.2.x:
168 P - Figure out why dll's compiled in mingw don't work right in WinXP.
169 P - Figure out why openssl 0.9.8d "make test" fails at sha256t test.
171 - RAM use in directory authorities.
172 - Memory use improvements:
173 - Look into pulling serverdescs off buffers as they arrive.
174 - Save and mmap v1 directories, and networkstatus docs; store them
175 zipped, not uncompressed.
176 - Switch cached_router_t to use mmap.
177 - What to do about reference counts on windows? (On Unix, this is
178 easy: unlink works fine. (Right?) On Windows, I have doubts. Do we
179 need to keep multiple files?)
180 - What do we do about the fact that people can't read zlib-
181 compressed files manually?
183 - If the client's clock is too far in the past, it will drop (or
184 just not try to get) descriptors, so it'll never build circuits.
185 - Tolerate clock skew on bridge relays.
187 - A more efficient dir protocol.
188 - Authorities should fetch the network-statuses amongst each
189 other, consensus them, and advertise a communal network-status.
190 This is not so much for safety/complexity as it is to reduce
191 bandwidth requirements for Alice.
192 - How does this interact with our goal of being able to choose
193 your own dir authorities? I guess we're now assuming that all
194 dir authorities know all the other authorities in their "group"?
195 - Should we also look into a "delta since last network-status
196 checkpoint" scheme, to reduce overhead further?
197 - Extend the "r" line in network-status to give a set of buckets (say,
198 comma-separated) for that router.
199 - Buckets are deterministic based on IP address.
200 - Then clients can choose a bucket (or set of buckets) to
203 - Improvements to versioning.
204 - When we connect to a Tor server, it sends back a cell listing
205 the IP it believes it is using. Use this to block dvorak's attack.
206 Also, this is a fine time to say what time you think it is.
207 o Verify that a new cell type is okay with deployed codebase
208 . Specify HELLO cells
209 . Figure out v0 compatibility.
212 - Eventdns improvements
213 - Have a way to query for AAAA and A records simultaneously.
214 - Improve request API: At the very least, add the ability to construct
215 a more-or-less arbitrary request and get a response.
216 - (Can we suppress cnames? Should we?)
218 - Now that we're avoiding exits when picking non-exit positions,
219 we need to consider how to pick nodes for internal circuits. If
220 we avoid exits for all positions, we skew the load balancing. If
221 we accept exits for all positions, we leak whether it's an internal
222 circuit at every step. If we accept exits only at the last hop, we
223 reintroduce Lasse's attacks from the Oakland paper.
225 - We should ship with a list of stable dir mirrors -- they're not
226 trusted like the authorities, but they'll provide more robustness
227 and diversity for bootstrapping clients.
229 - Simplify authority operation
230 - Follow weasel's proposal, crossed with mixminion dir config format
232 - A way to adjust router flags from the controller.
233 (How do we prevent the authority from clobbering them soon after?)
234 - a way to pick entry guards based wholly on extend_info equivalent;
235 a way to export extend_info equivalent.
237 - Better estimates in the directory of whether servers have good uptime
238 (high expected time to failure) or good guard qualities (high
240 - AKA Track uptime as %-of-time-up, as well as time-since-last-down
242 - Have a "Faster" status flag that means it. Fast2, Fast4, Fast8?
246 - Failed rend desc fetches sometimes don't get retried. True/false?
248 - Windows server usability
249 - Solve the ENOBUFS problem.
250 - make tor's use of openssl operate on buffers rather than sockets,
251 so we can make use of libevent's buffer paradigm once it has one.
252 - make tor's use of libevent tolerate either the socket or the
253 buffer paradigm; includes unifying the functions in connect.c.
254 - We need a getrlimit equivalent on Windows so we can reserve some
255 file descriptors for saving files, etc. Otherwise we'll trigger
256 asserts when we're out of file descriptors and crash.
257 M - rewrite how libevent does select() on win32 so it's not so very slow.
260 - Add an option (related to AvoidDiskWrites) to disable directory caching.
262 Minor items for 0.1.2.x as time permits:
263 - when reporting clock skew (both to logs and to controller), if it's
264 taken 126 seconds to read from the directory, our clock skew estimate
265 is 126 seconds wrong. use conn->timestamp_create or _lastwritten
266 for a closer estimate?
267 - a way to generate the website diagrams from source, so we can
268 translate them as utf-8 text rather than with gimp.
269 R - add d64 and fp64 along-side d and fp so people can paste status
270 entries into a url. since + is a valid base64 char, only allow one
271 at a time. spec and then do.
272 D don't do dns hijacking tests if we're reject *:* exit policy?
273 (deferred until 0.1.1.x is less common)
274 - When we export something from foo.c file for testing purposes only,
275 make a foo_test.h file for test.c to include.
276 - The Debian package now uses --verify-config when (re)starting,
277 to distinguish configuration errors from other errors. Perhaps
278 the RPM and other startup scripts should too?
279 - add a "default.action" file to the tor/vidalia bundle so we can fix the
280 https thing in the default configuration:
281 http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#PrivoxyWeirdSSLPort
282 . Flesh out options_description array in src/or/config.c
283 - Don't let 'newnym' be triggered more often than every n seconds.
284 X If we try to publish as a nickname that's already claimed, should
285 we append a number (or increment the number) and try again? This
286 way people who read their logs can fix it as before, but people
287 who don't read their logs will still offer Tor servers.
288 - Fall back to unnamed; warn user; send controller event. ("When we
289 notice a 'Rejected: There is already a named server with this nickname'
290 message... or maybe instead when we see in the networkstatuses that
291 somebody else is Named with the name we want: warn the user, send a
292 STATUS_SERVER message, and fall back to unnamed.")
293 ! - Tor should bind its ports before dropping privs, so users don't
294 have to do the ipchains dance.
295 - Rate limit exit connections to a given destination -- this helps
296 us play nice with websites when Tor users want to crawl them; it
297 also introduces DoS opportunities.
298 - Streamline how we pick entry nodes: Make choose_random_entry() have
299 less magic and less control logic.
300 - Christian Grothoff's attack of infinite-length circuit.
301 the solution is to have a separate 'extend-data' cell type
302 which is used for the first N data cells, and only
303 extend-data cells can be extend requests.
304 - Specify, including thought about anonymity implications.
305 - Display the reasons in 'destroy' and 'truncated' cells under some
307 - We need a way for the authorities to declare that nodes are
308 in a family. Also, it kinda sucks that family declarations use O(N^2)
309 space in the descriptors.
310 - If the server is spewing complaints about raising your ulimit -n,
311 we should add a note about this to the server descriptor so other
312 people can notice too.
314 - see if we should make use of truncate to retry
315 X kill dns workers more slowly
317 . Some back-out mechanism for auto-approval
318 - a way of rolling back approvals to before a timestamp
319 - Consider minion-like fingerprint file/log combination.
320 - packaging and ui stuff:
321 . multiple sample torrc files
322 . figure out how to make nt service stuff work?
324 - Vet all pending installer patches
325 - Win32 installer plus privoxy, sockscap/freecap, etc.
326 - Vet win32 systray helper code
329 - a NEWSTATUS event similar to NEWDESC.
330 - change circuit status events to give more details, like purpose,
331 whether they're internal, when they become dirty, when they become
332 too dirty for further circuits, etc.
333 - What do we want here, exactly?
334 - Specify and implement it.
335 - Change stream status events analogously.
336 - What do we want here, exactly?
337 - Specify and implement it.
338 - Make other events "better".
339 - Change stream status events analogously.
340 - What do we want here, exactly?
341 - Specify and implement it.
342 - Make other events "better" analogously
343 - What do we want here, exactly?
344 - Specify and implement it.
345 . Expose more information via getinfo:
346 - import and export rendezvous descriptors
347 - Review all static fields for additional candidates
348 - Allow EXTENDCIRCUIT to unknown server.
349 - We need some way to adjust server status, and to tell tor not to
350 download directories/network-status, and a way to force a download.
351 - Make everything work with hidden services
352 - Directory system improvements
353 - config option to publish what ports you listen on, beyond
354 ORPort/DirPort. It should support ranges and bit prefixes (?) too.
356 - Relay this in networkstatus.
359 - Need a way to request address lookups (and allocate a stream ID for
360 them) without having a corresponding client socket.
361 - Once this is done, it would be nice to have a way to request address
362 lookups from the controller without using SOCKS.
363 - Refactor exit/entry side of DNS resolve: we don't need a connection_t;
364 we can have an edge_connection_t and (say) dns_request_t both extend an
365 edge_stream_t, and have p_streams and n_streams both be linked lists
369 - Configuration format really wants sections.
370 - Good RBL substitute.
371 - Authorities should try using exits for http to connect to some URLS
372 (specified in a configuration file, so as not to make the List Of Things
373 Not To Censor completely obvious) and ask them for results. Exits that
374 don't give good answers should have the BadExit flag set.
375 - Our current approach to block attempts to use Tor as a single-hop proxy
376 is pretty lame; we should get a better one.
377 . Update the hidden service stuff for the new dir approach.
378 - switch to an ascii format, maybe sexpr?
379 - authdirservers publish blobs of them.
380 - other authdirservers fetch these blobs.
381 - hidserv people have the option of not uploading their blobs.
382 - you can insert a blob via the controller.
383 - and there's some amount of backwards compatibility.
384 - teach clients, intro points, and hidservs about auth mechanisms.
385 - come up with a few more auth mechanisms.
386 - auth mechanisms to let hidden service midpoint and responder filter
388 - Bind to random port when making outgoing connections to Tor servers,
389 to reduce remote sniping attacks.
390 - Have new people be in limbo and need to demonstrate usefulness
391 before we approve them.
392 - Clients should estimate their skew as median of skew from servers
394 - Make router_is_general_exit() a bit smarter once we're sure what it's for.
395 - Audit everything to make sure rend and intro points are just as likely to
397 - Do something to prevent spurious EXTEND cells from making middleman
398 nodes connect all over. Rate-limit failed connections, perhaps?
399 - Automatically determine what ports are reachable and start using
400 those, if circuits aren't working and it's a pattern we recognize
401 ("port 443 worked once and port 9001 keeps not working").
402 - Limit to 2 dir, 2 OR, N SOCKS connections per IP.
403 - Handle full buffers without totally borking
404 - Rate-limit OR and directory connections overall and per-IP and
406 - Hold-open-until-flushed now works by accident; it should work by
408 - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
410 - tor-resolve script should use socks5 to get better error messages.
411 - hidserv offerers shouldn't need to define a SocksPort
412 * figure out what breaks for this, and do it.
413 - tor should be able to have a pool of outgoing IP addresses
414 that it is able to rotate through. (maybe)
415 - Specify; implement.
416 - let each hidden service (or other thing) specify its own
418 - Stop using tor_socketpair to make connection bridges: do an
419 implementation that uses buffers only.
422 - Patch privoxy and socks protocol to pass strings to the browser.
423 - Standby/hotswap/redundant hidden services.
424 - Robust decentralized storage for hidden service descriptors.
425 - The "China problem"
426 - Allow small cells and large cells on the same network?
427 - Cell buffering and resending. This will allow us to handle broken
428 circuits as long as the endpoints don't break, plus will allow
429 connection (tls session key) rotation.
430 - Implement Morphmix, so we can compare its behavior, complexity, etc.
431 - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
432 link crypto, unless we can bully openssl into it.
433 - Need a relay teardown cell, separate from one-way ends.
434 (Pending a user who needs this)
435 - Handle half-open connections: right now we don't support all TCP
436 streams, at least according to the protocol. But we handle all that
437 we've seen in the wild.
438 (Pending a user who needs this)
441 - Mark up spec; note unclear points about servers
442 - Mention controller libs someplace.
443 . more pictures from ren. he wants to describe the tor handshake
444 NR- write a spec appendix for 'being nice with tor'
445 - tor-in-the-media page
446 - Remove need for HACKING file.
447 - Figure out licenses for website material.
448 - Specify the keys and key rotation schedules and stuff
451 - and remove home and make the "Tor" picture be the link to home.
452 - put the logo on the website, in source form, so people can put it on
453 stickers directly, etc.
454 R - make a page with the hidden service diagrams.
456 - ask Jan to be the translation coordinator? add to volunteer page.
458 - add a page for localizing all tor's components.