TODO 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312
  1. $Id$
  2. Legend:
  3. SPEC!! - Not specified
  4. SPEC - Spec not finalized
  5. N - nick claims
  6. R - arma claims
  7. P - phobos claims
  8. - Not done
  9. * Top priority
  10. . Partially done
  11. o Done
  12. D Deferred
  13. X Abandoned
  14. Non-Coding, Soon:
  15. N - Mark up spec; note unclear points about servers
  16. N - Clean up dir spec.
  17. N - Mention controller libs someplace.
  18. D FAQ entry: why gnutls is bad/not good for tor
  19. P - flesh out the rest of the section 6 of the faq
  20. P - gather pointers to livecd distros that include tor
  21. - put the logo on the website, in source form, so people can put it on
  22. stickers directly, etc.
  23. R . more pictures from ren. he wants to describe the tor handshake, i want to
  24. talk about hidden services.
  25. * clean up the places where our docs are redundant (or worse, obsolete in
  26. one file and correct elsewhere). agl has a start on a global
  27. list-of-tor-docs.
  28. P - update windows docs to clarify which versions of windows, and why a
  29. DOS window, how it's used, for the less technical users
  30. NR- write a spec appendix for 'being nice with tor'
  31. - tor-in-the-media page
  32. - Ask schanzle@cas.homelinux.org about a patch for rpm spec fixes against
  33. tor-0.1.0.7.rc
  34. - Remove need for HACKING file.
  35. for 0.1.1.x:
  36. N - if they're trying to be a tor server and they're running
  37. win 98 or win me, give them a message talking about The Bug.
  38. . Helper nodes
  39. . More testing and debugging
  40. R - If your helper nodes are unavailable, don't abandon them unless
  41. other nodes *are* reachable.
  42. N - Destroy and truncated cells should have reasons.
  43. - Specify
  44. - Implement
  45. o Add a panic-button config option to buy us time if we get sybiled:
  46. Limit nodes to those listed in fingerprints file.
  47. o Clients use Stable and Fast instead of uptime and bandwidth to
  48. pick which servers are stable/fast.
  49. N - Only use a routerdesc if you recognize its hash.
  50. - (Must defer till dirservers are upgraded to latest code, which
  51. actually generates these hashes.)
  52. - Of course, authdirservers must not do this.
  53. - Should directory mirrors do something else entirely?
  54. o If we have a routerdesc for Bob, and he says, "I'm 0.1.0.x", don't
  55. fetch a new one if it was published in the last 2 hours.
  56. - How does this interact with the 'recognized hash' rule?
  57. R - Christian Grothoff's attack of infinite-length circuit.
  58. the solution is to have a separate 'extend-data' cell type
  59. which is used for the first N data cells, and only
  60. extend-data cells can be extend requests.
  61. - Specify, including thought about
  62. - Implement
  63. R - When we connect to a Tor server, it sends back a signed cell listing
  64. the IP it believes it is using. Use this to block dvorak's attack.
  65. Also, this is a fine time to say what time you think it is.
  66. - Verify that a new cell type is okay with deployed codebase
  67. - Specify
  68. - Implement
  69. R - clients prefer to avoid exit nodes for non-exit path positions.
  70. - find 10 dirservers.
  71. - What are criteria to be a dirserver? Write a policy.
  72. Deferred from 0.1.1.x:
  73. N . Additional controller features
  74. o Find a way to make event info more extensible
  75. - change circuit status events to give more details, like purpose,
  76. whether they're internal, when they become dirty, when they become
  77. too dirty for further circuits, etc.
  78. R - What do we want here, exactly?
  79. N - Specify and implement it.
  80. - Change stream status events analogously.
  81. R - What do we want here, exactly?
  82. N - Specify and implement it.
  83. - Make other events "better".
  84. - Change stream status events analogously.
  85. R - What do we want here, exactly?
  86. N - Specify and implement it.
  87. - Make other events "better" analogously
  88. R - What do we want here, exactly?
  89. N - Specify and implement it.
  90. . Expose more information via getinfo:
  91. - import and export rendezvous descriptors
  92. - Review all static fields for additional candidates
  93. - Allow EXTENDCIRCUIT to unknown server.
  94. - We need some way to adjust server status, and to tell tor not to
  95. download directories/network-status, and a way to force a download.
  96. - It would be nice to request address lookups from the controller
  97. without using SOCKS.
  98. - Make everything work with hidden services
  99. X switch accountingmax to count total in+out, not either in or
  100. out. it's easy to move in this direction (not risky), but hard to
  101. back out if we decide we prefer it the way it already is. hm.
  102. - cpu fixes:
  103. - see if we should make use of truncate to retry
  104. o hardware accelerator support (configure engines.)
  105. o hardware accelerator support (use instead of aes.c when reasonable)
  106. - Benchmark this somehow to see whether using EVP_foo is slower in the
  107. non-engine case than AES_foo. If so, check for AES engine and fall
  108. back to AES_foo when it's not found.
  109. R - kill dns workers more slowly
  110. . Directory changes
  111. o recommended-versions for client / server ?
  112. . Some back-out mechanism for auto-approval
  113. o dirservers have blacklist of IPs and keys they hate
  114. - a way of rolling back approvals to before a timestamp
  115. - Consider minion-like fingerprint file/log combination.
  116. - Decentralization
  117. o Dirservers publish compressed network-status objects.
  118. o Support retrieving several-at-once
  119. o Everyone downloads network-status objects
  120. o Clients: from all directories, round-robin
  121. o Basic implementation: disable until 0.1.1.x is out.
  122. o On failure, mark trusted_dir_server as having failed
  123. o Retry, up to a point.
  124. X Launch retry immediately on failure.
  125. o Parse them
  126. o Cache them, reload on restart
  127. o Serve cached directories
  128. o Directories expose individual descriptors
  129. X By 'if-newer-than' (Does the spec require this??)
  130. o Support compression.
  131. o Alice acts on network-status objects
  132. o Alice downloads descriptors as needed.
  133. o Figure out what's needed
  134. o Store it
  135. o Implement store
  136. o Implement reload-from-store
  137. o Store downloaded descriptors
  138. o Download it
  139. o As-needed if we have 2 network-status objs.
  140. o Download "all" if we have less than 2 network-status objs.
  141. (This has vulnerabilities if we're not careful)
  142. o Call directory_has_arrived as needed; rename it.
  143. o Set has_fetched_directory properly.
  144. o Retry descriptors on failure
  145. o Give up after a while.
  146. - But try again after a long while (???)
  147. o Check software versions according to some sane plan.
  148. - Warn again after 24 hours.
  149. o Alice sets descriptor status from network-status
  150. o Implement
  151. o Use
  152. o Routerdesc download changes
  153. o Refactor combined-status to be its own type.
  154. o Change rule from "do not launch new connections when one exists" to
  155. "do not request any fingerprint that we're currently requesting."
  156. o Launch connections every minute, or whenever a download fails
  157. o Retry failed routerdescs after 0, 1, 5, 10 minutes.
  158. o Mirrors retry harder and more often. (0, 0, 1, 1, 2, 5, and 15)
  159. o Reset failure count every 60 minutes
  160. o Drop fallback to download-all. Also, always split download.
  161. o Use has_fetched_directory sanely, whatever that means.
  162. o Downgrade new directory events from notice to info
  163. o Call dirport_is_reachable from somewhere else.
  164. o Networkstatus should list who's an authority.
  165. o Add nickname element to dirserver line. Log this along with IP:Port.
  166. o Warn when using non-default directory servers.
  167. o When giving up on a non-finished dir request, log how many bytes
  168. dropped, to see whether it's worthwhile to use partial info.
  169. - config option to publish what ports you listen on, beyond
  170. ORPort/DirPort. It should support ranges and bit prefixes (?) too.
  171. - Parse this.
  172. - Relay this in networkstatus.
  173. X Make authorities rate-limit logging their complaints about given
  174. servers?
  175. o All versions of Tor should get cosmetic changes rate-limited.
  176. o Pick directories from networkstatus objects, not from routerlist.
  177. o But! We can't do this easily, since we want to know about platform,
  178. and networkstatus doesn't tell us Tor version. Can we solve this?
  179. Should we do it by adding flags to networkstatus or what?
  180. - packaging and ui stuff:
  181. . multiple sample torrc files
  182. - uninstallers
  183. . for os x
  184. . figure out how to make nt service stuff work?
  185. . Document it.
  186. . Add version number to directory.
  187. N - Vet all pending installer patches
  188. - Win32 installer plus privoxy, sockscap/freecap, etc.
  189. - Vet win32 systray helper code
  190. - document:
  191. - torcp needs more attention in the tor-doc-win32.
  192. - recommend gaim.
  193. - unrecommend IE because of ftp:// bug.
  194. - torrc.complete.in needs attention?
  195. o Dump "ports" from routerparse?
  196. o Let more config options (e.g. ORPort) change dynamically.
  197. o Add TTLs to DNS-related replies, and use them (when present) to adjust
  198. addressmap values.
  199. - Bind to random port when making outgoing connections to Tor servers,
  200. to reduce remote sniping attacks.
  201. - Have new people be in limbo and need to demonstrate usefulness
  202. before we approve them.
  203. - Clients should estimate their skew as median of skew from servers
  204. over last N seconds.
  205. - Security
  206. - Alices avoid duplicate class C nodes.
  207. - Analyze how bad the partitioning is or isn't.
  208. . Update the hidden service stuff for the new dir approach.
  209. - switch to an ascii format, maybe sexpr?
  210. - authdirservers publish blobs of them.
  211. - other authdirservers fetch these blobs.
  212. - hidserv people have the option of not uploading their blobs.
  213. - you can insert a blob via the controller.
  214. - and there's some amount of backwards compatibility.
  215. - teach clients, intro points, and hidservs about auth mechanisms.
  216. - come up with a few more auth mechanisms.
  217. . Come up with a coherent strategy for bandwidth buckets and TLS. (The
  218. logic for reading from TLS sockets is likely to overrun the bandwidth
  219. buckets under heavy load. (Really, the logic was never right in the
  220. first place.) Also, we should audit all users of get_pending_bytes().)
  221. - Make it harder to circumvent bandwidth caps: look at number of bytes
  222. sent across sockets, not number sent inside TLS stream.
  223. o Research memory use on Linux: what's happening?
  224. X Is it threading? (Maybe, maybe not)
  225. X Is it the buf_shrink bug? (Quite possibly)
  226. o Instrument the 0.1.1 code to figure out where our memory is going;
  227. apply the results. (all platforms?)
  228. - Make router_is_general_exit() a bit smarter once we're sure what it's for.
  229. - Directory "helper".
  230. - rewrite how libevent does select() on win32 so it's not so very slow.
  231. o enclaves (at least preliminary)
  232. - Write limiting; separate token bucket for write
  233. - Audit everything to make sure rend and intro points are just as likely to
  234. be us as not.
  235. - Do something to prevent spurious EXTEND cells from making middleman
  236. nodes connect all over. Rate-limit failed connections, perhaps?
  237. Future version:
  238. - Limit to 2 dir, 2 OR, N SOCKS connections per IP.
  239. - Handle full buffers without totally borking
  240. - Rate-limit OR and directory connections overall and per-IP and
  241. maybe per subnet.
  242. - Hold-open-until-flushed now works by accident; it should work by
  243. design.
  244. - DoS protection: TLS puzzles, public key ops, bandwidth exhaustion.
  245. - Specify?
  246. - tor-resolve script should use socks5 to get better error messages.
  247. - make min uptime a function of the available choices (say, choose 60th
  248. percentile, not 1 day.)
  249. - Track uptime as %-of-time-up, as well as time-since-last-down.
  250. - hidserv offerers shouldn't need to define a SocksPort
  251. * figure out what breaks for this, and do it.
  252. - auth mechanisms to let hidden service midpoint and responder filter
  253. connection requests.
  254. - Relax clique assumptions.
  255. - start handling server descriptors without a socksport?
  256. - tor should be able to have a pool of outgoing IP addresses
  257. that it is able to rotate through. (maybe)
  258. Blue-sky:
  259. - Patch privoxy and socks protocol to pass strings to the browser.
  260. - Standby/hotswap/redundant hidden services.
  261. - Robust decentralized storage for hidden service descriptors.
  262. - The "China problem"
  263. - Allow small cells and large cells on the same network?
  264. - Cell buffering and resending. This will allow us to handle broken
  265. circuits as long as the endpoints don't break, plus will allow
  266. connection (tls session key) rotation.
  267. - Implement Morphmix, so we can compare its behavior, complexity, etc.
  268. - Other transport. HTTP, udp, rdp, airhook, etc. May have to do our own
  269. link crypto, unless we can bully openssl into it.
  270. . Conn key rotation (we switch to a new one after a week, but
  271. old circuits don't get any benefit from this).
  272. - Need a relay teardown cell, separate from one-way ends.
  273. (Pending a user who needs this)
  274. - Handle half-open connections: right now we don't support all TCP
  275. streams, at least according to the protocol. But we handle all that
  276. we've seen in the wild.
  277. (Pending a user who needs this)