TODO 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224
  1. Legend:
  2. SPEC!! - Not specified
  3. SPEC - Spec not finalized
  4. NICK - nick claims
  5. ARMA - arma claims
  6. - Not done
  7. * Top priority
  8. . Partially done
  9. o Done
  10. D Deferred
  11. X Abandoned
  12. Short-term:
  13. o routers have identity key, link key, onion key.
  14. . link key certs are
  15. - signed by identity key
  16. - not in descriptor
  17. o not in config
  18. - not on disk
  19. o identity and onion keys are in descriptor (and disk)
  20. o upon boot, if it doesn't find identity key, generate it and write it.
  21. o also write a file with the identity key fingerprint in it
  22. o router generates descriptor: flesh out router_get_my_descriptor()
  23. o Routers sign descriptors with identity key
  24. o routers put version number in descriptor
  25. o routers should maybe have `uname -a` in descriptor?
  26. o Give nicknames to routers
  27. o in config
  28. o in descriptors
  29. o router posts descriptor
  30. o when it boots
  31. D when it changes
  32. o change tls stuff so certs don't get written to disk, or read from disk
  33. o make directory.c 'thread'safe
  34. o dirserver parses descriptor
  35. o dirserver checks signature
  36. D client checks signature?
  37. o dirserver writes directory to file
  38. o reads that file upon boot
  39. o directory includes all routers, up and down
  40. o add "up" line to directory, listing nicknames
  41. ARMA . find an application that uses half-open connections: openssh
  42. NICK - instruments ORs to report stats
  43. - average cell fullness
  44. - average bandwidth used
  45. - others?
  46. . integrate rep_ok functions, see what breaks
  47. ARMA - configure log files. separate log file, separate severities.
  48. ARMA - what assumptions break if we fclose(0) when we daemonize?
  49. o make buffer struct elements opaque outside buffers.c
  50. ARMA - Go through log messages, reduce confusing error messages.
  51. ARMA - make the logs include more info (fd, etc)
  52. ARMA - add log convention to the HACKING file
  53. - make 'make install' do the right thing
  54. o change binary name to tor
  55. o change config files so you look at commandline, else look in
  56. /etc/torrc. no cascading.
  57. o have an absolute datadir with fixed names for files, and fixed-name
  58. keydir under that with fixed names
  59. ARMA - tor faq
  60. list all other systems, why we're different.
  61. NICK - Move (most of) the router/directory code out of main.c
  62. Mid-term:
  63. - What happens when a circuit's length is 1? What breaks?
  64. . streams / circuits
  65. o Implement streams
  66. o Rotate circuits after N minutes?
  67. X Circuits should expire when circuit->expire triggers
  68. NICK . Handle half-open connections
  69. o Figure out what causes connections to close, standardize
  70. when we mark a connection vs when we tear it down
  71. o Look at what ssl does to keep from mutating data streams
  72. ARMA - Reduce streamid footprint from 7 bytes to 2 bytes
  73. - Check for collisions in streamid (now possible with
  74. just 2 bytes), and back up & replace with padding if so
  75. - Use the 3 saved bytes to put pseudorandomness in each relay cell
  76. - Use the 4 reserved bytes in each cell header to keep 1/5
  77. of a sha1 of the relay payload (move into stream header)
  78. - Move length into the stream header too
  79. - Spec the stream_id stuff. Clarify that nobody on the backward
  80. stream should look at stream_id.
  81. ARMA . Exit policies
  82. o Spec how to write the exit policies
  83. - Path selection algorithms
  84. - Let user request certain nodes
  85. - And disallow certain nodes
  86. D Choose path by jurisdiction, etc?
  87. - Make relay end cells have failure status and payload attached
  88. - Streams that fail due to exit policy must reextend to new node
  89. - Add extend_wait state to edge connections, thumb through them
  90. when the AP get an extended cell.
  91. SPEC!! D Non-clique topologies
  92. D Implement our own memory management, at least for common structs
  93. . Put CPU workers in separate processes
  94. o Handle multiple cpu workers (one for each cpu, plus one)
  95. o Queue for pending tasks if all workers full
  96. o Support the 'process this onion' task
  97. D Merge dnsworkers and cpuworkers to some extent
  98. - Handle cpuworkers dying
  99. o Simple directory servers
  100. o Include key in source; sign directories
  101. o Signed directory backend
  102. o Document
  103. o Integrate
  104. o Add versions to code
  105. o Have directories list recommended-versions
  106. o Include (unused) line in directories
  107. o Check for presence of line.
  108. o Quit if running the wrong version
  109. o Command-line option to override quit
  110. o Add more information to directory server entries
  111. o Exit policies
  112. D Advanced directory servers
  113. D Automated reputation management
  114. SPEC!! D Figure out how to do threshold directory servers
  115. D jurisdiction info in dirserver entries? other info?
  116. . Scrubbing proxies
  117. - Find an smtp proxy?
  118. - Check the old smtp proxy code
  119. o Find an ftp proxy? wget --passive
  120. D Wait until there are packet redirectors for Linux
  121. . Get socks4a support into Mozilla
  122. . Get tor to act like a socks server
  123. o socks4, socks4a
  124. ARMA - socks5
  125. SPEC!! - Handle socks commands other than connect, eg, bind?
  126. . Develop rendezvous points
  127. o Design
  128. - Spec
  129. - Implement
  130. D Deploy and manage open source development site.
  131. . Documentation
  132. o Discussion of socks, tsocks, etc
  133. o On-the-network protocol
  134. o Onions
  135. o Cells
  136. . Better comments for functions!
  137. - Tests
  138. o Testing harness/infrastructure
  139. NICK . Unit tests
  140. D System tests (how?)
  141. - Performance tests, so we know when we've improved
  142. . webload infrastructure (Bruce)
  143. . httperf infrastructure (easy to set up)
  144. . oprofile (installed in RH >8.0)
  145. D Deploy a widespread network
  146. D Load balancing between router twins
  147. D Keep track of load over links/nodes, to
  148. know who's hosed
  149. NICK . Daemonize and package
  150. o Teach it to fork and background
  151. - Red Hat spec file
  152. - Debian spec file equivalent
  153. . Portability
  154. . Which .h files are we actually using?
  155. . Port to:
  156. o Linux
  157. o BSD
  158. . Solaris
  159. o Cygwin
  160. o Win32
  161. o OS X
  162. - deal with pollhup / reached_eof on all platforms
  163. o openssl randomness
  164. o inet_ntoa
  165. . stdint.h
  166. - Make a script to set up a local network on your machine
  167. X Move away from openssl
  168. o Abstract out crypto calls
  169. X Look at nss, others? Just include code?
  170. o Clearer bandwidth management
  171. o Do we want to remove bandwidth from OR handshakes?
  172. o What about OP handshakes?
  173. - More flexibility in node addressing
  174. D Support IPv6 rather than just 4
  175. - Handle multihomed servers (config variable to set IP)
  176. . Move from onions to ephemeral DH
  177. o incremental path building
  178. o transition circuit-level sendmes to hop-level sendmes
  179. o implement truncate, truncated
  180. o move from 192byte DH to 128byte DH, so it isn't so damn slow
  181. - exiting from not-last hop
  182. - OP logic to decide to extend/truncate a path
  183. - make sure exiting from the not-last hop works
  184. - logic to find last *open* hop, not last hop, in cpath
  185. - choose exit nodes by exit policies
  186. Older (done) todo stuff:
  187. o Use a stronger cipher
  188. o aes now, by including the code ourselves
  189. X On the fly compression of each stream
  190. o Clean up the event loop (optimize and sanitize)
  191. o Remove that awful concept of 'roles'
  192. o Terminology
  193. o Circuits, topics, cells stay named that
  194. o 'Connection' gets divided, or renamed, or something?
  195. o DNS farm
  196. o Distribute queries onto the farm, get answers
  197. o Preemptively grow a new worker before he's needed
  198. o Prune workers when too many are idle
  199. o DNS cache
  200. o Clear DNS cache over time
  201. D Honor DNS TTL info (how??)
  202. o Have strategy when all workers are busy
  203. o Keep track of which connections are in dns_wait
  204. o Need to cache positives/negatives on the tor side
  205. o Keep track of which queries have been asked
  206. o Better error handling when
  207. o An address doesn't resolve
  208. o We have max workers running
  209. o Consider taking the master out of the loop?
  210. X Implement reply onions
  211. o Total rate limiting
  212. o Look at OR handshake in more detail
  213. o Spec it
  214. o Merge OR and OP handshakes
  215. o rearrange connection_or so it doesn't suck so much to read
  216. D Periodic link key rotation. Spec?
  217. o wrap malloc with something that explodes when it fails
  218. o Clean up the number of places that get to look at prkey