Change to BadExit logic: Let authorities set an "I list bad exits" flag.  Consider an exit bad if it is listed as bad by more than half of *those* authorities.  This gives us a better migration path.


svn:r8756

 Add client support for a 'BadExit' flag, so authorities can say "Server X is a poor choise for your nytimes.com connections, as it seems to direct them to HoorayForMao.com or (more likely) WouldYouLikeToBuyTheseFineEncyclopedias.com"
 


svn:r8690

svn:r8630

svn:r8622

 Partial implementation of revised nickname syntax for controllers.  Implement ability to look up routers by "verbose" nicknames; add a per-v1-control-connection flag to turn the feature on in events.  Needs testing, spec, ability to actually turn on the flag, double-checking that we wont overflow any nickname buffers, and changelog.


svn:r8582

 Disprefer exit nodes for entry, middle positions (fixes bug 200).  Also, switch to using a uint64_t to hold "total bandwidth for all nodes" under consideration; crypt_rand_int would have died at 2GB/s network capacity.


svn:r8571

enable your dirport as a client, you'll crash.


svn:r8561

 Make distinct all non-bug messages at notice or higher that appear 3 or more times.


svn:r8541

 Reserve the nickname "Unnamed" for routers that can't pick a hostname; any
 router can call itself Unnamed; directory servers will never allocate Unnamed
 to any particular router; clients won't believe that any router is the
 canonical Unnamed.
 


svn:r8529

 Make "is a v1 authority", "is a v2 authority", and "is a hidden service authority" into separate flags so we can eventually migrate more trust away from moria.


svn:r8523

 Refactor entry guard status logic a lot; allow more factors [like not
 having a Guard flag or being listed in ExcludeNodes] to render a guard
 "unlisted" (now called "unusable"); track guard down status (now
 called "unreachable") separately from is_running.


svn:r8519

Add an EnforceDistinctSubnets option so that clients who know what they are doing (mainly people with private testing networks) can disable our same-/16 detection.

svn:r8504

svn:r8500

 Consider changes to has_old_dnsworkers as noncosmetic


svn:r8464

NEEDS REVIEW. Act on previous comment, and handle named servers differently: now, we allow multiple servers with the same name in the routerlist even if that name is reserved, but we check whether names are reserved when we try to look up routers by nickname. This is a minor security fix.  This makes router_add_to_routerlist O(1). This is a backport candidate.

svn:r8433

Stop searching routerlist for routers with the same identity as other routers (on router insert): we already have a map for that.  (We need to add an index field to routerinfo_t so we can figure out which point in the routerlist to replace.)  Also, add a comment to routerlist.c; arma, please advise?

svn:r8432

Switch routerlist.c to using memcmp on digests rather than crypto_pk_cmp_keys(); speed up find_whitespace a lot (8x for me) by using a switch statement.  This should speed parsing a lot of routers at once by a lot.

svn:r8430

 Add some client performance XXXXs; try to move some common case tests higher on their decision trees.


svn:r8410

we're reachable. Spread them over 4 circuits, so hopefully a few will
be fast. This exercises our bandwidth and bootstraps us quicker.


svn:r8399

 router_set_networkstatus() gets a list of status documents we asked for from
 connection_dir_client_reached_eof().  However, as a cache we (sometimes?) just
 ask for "all".  router_set_networkstatus() would freak out over that, meaning
 it would log a warning and drop the status document instead of caching it
 as it is supposed to.  Now we let router_set_networkstatus() know if the
 data comes from an all-request so it can do the right thing.


svn:r8398

 Minor documentation fix


svn:r8397

 Fix two corner cases in router_dir_info_changed().  This should have no observable effect.


svn:r8311

of cached-routers.new every time we read it.


svn:r8236

svn:r8235

/16 network when constructing a circuit.


svn:r7068

 Only do the expensive version of router_have_minimum_dir_info() when the dir info has changed.  Backport candidate, since oprofile suggests that this function and ones it calls account for 25-35% of oprofile samples.


svn:r7062

 Remove now-spurious size and data arguments from tor_mmap_file


svn:r6987

 Add an "mmap handle" type to encapsulate bookkeeping elements of mmap issues; add prelim win32 impl


svn:r6980

 Clean up wide lines from last patch.


svn:r6907

 Refactor connection_t into edge, or, dir, control, and base subtypes.  This might save some RAM on busy exit servers, but really matters most in terms of correctness.


svn:r6906

IP addresses, not IPs.


svn:r6764

First part of making mmap-based stuff work on win32: save descriptors as "binary" (no LF->CRLF tanslation) so that we can mmap them properly later.  Patch from Frediano Ziglio.


svn:r6748

svn:r6699

svn:r6698

svn:r6697

svn:r6696

Make sure that our calculated offsets for routers is correct; again. This time bug 309 may be gone gone gone.


svn:r6695

Likely fix for bug 309: when we calculate offsets after rebuilding the descriptor cache, do not reset the offset pointer half-way through.


svn:r6693

svn:r6678

svn:r6676