Commit Graph

310 Commits

Author SHA1 Message Date
George Kadianakis
6009c89165 Set guard state on bridge descriptor fetches.
We used to not set the guard state in launch_direct_bridge_descriptor_fetch().
So when a bridge descriptor fetch failed, the guard subsystem would never
learn about the fail (and hence the guard's reachability state would not
be updated).
2017-05-22 15:56:32 +03:00
Roger Dingledine
a18b41cc77 fix typo in comment 2017-05-19 22:14:56 -04:00
Nick Mathewson
5d8061cd01 Merge remote-tracking branch 'public/bug21415_testfix_030' into maint-0.3.0 2017-04-03 09:27:40 -04:00
Nick Mathewson
4d5b46aad9 Merge remote-tracking branch 'origin/maint-0.3.0' 2017-03-27 15:44:31 +02:00
George Kadianakis
ef4c10fb42 Fix max sampled size logic when in bridge mode.
When calculating max sampled size, Tor would only count the number of
bridges in torrc, without considering that our state file might already
have sampled bridges in it. This caused problems when people swap
bridges, since the following error would trigger:

         [warn] Not expanding the guard sample any further; just hit the
                maximum sample threshold of 1
2017-03-27 15:39:26 +02:00
Nick Mathewson
7505f452c8 Run the copyright update script. 2017-03-15 16:13:17 -04:00
George Kadianakis
41324b5ae1 Revert "Restore correct behavior of 0.3.0.4-rc with bridges+ipv6-min"
This reverts commit 5298ab5917.
2017-03-09 09:19:12 -05:00
Nick Mathewson
5298ab5917 Restore correct behavior of 0.3.0.4-rc with bridges+ipv6-min
In that chutney test, the bridge client is configured to connect to
the same bridge at 127.0.0.1:5003 _and_ at [::1]:5003, with no
change in transports.

That meant, I think, that the descriptor is only assigned to the
first bridge when it arrives, and never the second.
2017-03-01 15:02:16 -05:00
George Kadianakis
18a98206ed Improve descriptor checks in the new guard algorithm.
- Make sure we check at least two guards for descriptor before making
  circuits. We typically use the first primary guard for circuits, but
  it can also happen that we use the second primary guard (e.g. if we
  pick our first primary guard as an exit), so we should make sure we
  have descriptors for both of them.

- Remove BUG() from the guard_has_descriptor() check since we now know
  that this can happen in rare but legitimate situations as well, and we
  should just move to the next guard in that case.
2017-03-01 08:46:53 -05:00
Nick Mathewson
1582adabbb Change approach to preventing duplicate guards.
Previously I'd made a bad assumption in the implementation of
prop271 in 0.3.0.1-alpha: I'd assumed that there couldn't be two
guards with the same identity.  That's true for non-bridges, but in
the bridge case, we allow two bridges to have the same ID if they
have different addr:port combinations -- in order to have the same
bridge ID running multiple PTs.

Fortunately, this assumption wasn't deeply ingrained: we stop
enforcing the "one guard per ID" rule in the bridge case, and
instead enforce "one guard per <id,addr,port>".

We also needed to tweak our implementation of
get_bridge_info_for_guard, since it made the same incorrect
assumption.

Fixes bug 21027; bugfix on 0.3.0.1-alpha.
2017-02-28 08:16:33 -05:00
Nick Mathewson
f5995692da Replace entry_guard_get_by_id_digest_for_guard_selection impl.
We already implemented this whole function somewhere else; no need
to have the same code twice.
2017-02-14 10:28:54 -05:00
Nick Mathewson
2670844b2b whoops, removed a semicolon :( 2017-02-09 10:59:48 -05:00
Nick Mathewson
f594bdb3ad One more prop271 XXX. 2017-02-09 10:52:47 -05:00
Nick Mathewson
14c2a1f403 Update some more XXXXprop271 comments to refer to actual tickets or to be up-to-date 2017-02-09 10:48:28 -05:00
Nick Mathewson
3919f4f529 Remove an XXXprop271 comment: turns out we didn't need a tristate 2017-02-09 10:30:20 -05:00
Nick Mathewson
d15273e9f5 Change "prop271" in XXXXs about guard Ed identity to refer to #20872. 2017-02-09 10:29:02 -05:00
Nick Mathewson
2d2ab29ce8 Merge remote-tracking branch 'asn/bug21052' 2017-02-01 15:53:16 -05:00
Nick Mathewson
78011bb7ba Merge branch 'bug21242' 2017-02-01 09:09:58 -05:00
Nick Mathewson
2e93bffa1d Merge remote-tracking branch 'public/bug21129' 2017-02-01 09:01:44 -05:00
Nick Mathewson
35d8270942 When marking guard state instances on a channel, don't mark NULL
It's okay for guard_state to be null: we might have a fallback
circuit, or we might not be using guards.

Fixes bug 211228; bugfix on 0.3.0.1-alpha
2017-01-31 14:44:14 -05:00
Nick Mathewson
0f0d4356b2 Don't try to use confirmed_idx in remove_guard_from_...lists()
Since we can call this function more than once before we update all
the confirmed_idx fields, we can't rely on all the relays having an
accurate confirmed_idx.

Fixes bug 21129; bugfix on 0.3.0.1-alpha
2017-01-31 14:34:32 -05:00
Nick Mathewson
746d959100 Don't build circuits till primary guards have descriptors
In addition to not wanting to build circuits until we can see most
of the paths in the network, and in addition to not wanting to build
circuits until we have a consensus ... we shouldn't build circuits
till all of our (in-use) primary guards have descriptors that we can
use for them.

This is another bug 21242 fix.
2017-01-31 12:31:43 -05:00
Nick Mathewson
02da24f8e5 Don't (usually) return any guards that are missing descriptors.
Actually, it's _fine_ to use a descriptorless guard for fetching
directory info -- we just shouldn't use it when building circuits.
Fortunately, we already have a "usage" flag that we can use here.

Partial fix for bug 21242.
2017-01-31 12:30:33 -05:00
Nick Mathewson
26957a127a entry_guard_pick_for_circuit(): TRAFFIC guards must have descriptors
This relates to the 21242 fix -- entry_guard_pick_for_circuit()
should never yield nodes without descriptors when the node is going
to be used for traffic, since we won't be able to extend through
them.
2017-01-31 11:47:09 -05:00
Nick Mathewson
cccd3f1dae entrynodes: Remove "split these functions" XXXXs
They now have a ticket: #21349.
2017-01-30 10:49:40 -05:00
Nick Mathewson
ead934e61e Remove prop271 "spec deviation" comments -- the spec has been updated
In some cases, replace those comments with better ones.
2017-01-30 10:30:09 -05:00
Nick Mathewson
818b44cc7c Repair the (deprecated, ugly) DROPGUARDS controller function.
This actually is much easier to write now that guard_selection_t is
first-class.
2017-01-24 09:18:56 -05:00
George Kadianakis
b047d97b28 Remove some more remnants of legacy guard selection. 2017-01-24 13:35:57 +02:00
Nick Mathewson
d5d7c3e638 Remove argument from guards_choose_dirguard 2017-01-18 15:58:19 -05:00
Nick Mathewson
3efe8bb8ac Remove some now-spurious blocks and indentation. 2017-01-18 15:45:02 -05:00
Nick Mathewson
6d03e36fd0 Remove GS_TYPE_LEGACY 2017-01-18 15:37:01 -05:00
Nick Mathewson
a31a5581ee Remove UseDeprecatedGuardAlgorithm. 2017-01-18 15:33:26 -05:00
Nick Mathewson
472b277207 Remove the (no longer compiled) code for legacy guard selection.
Part of 20830.
2017-01-18 15:27:10 -05:00
George Kadianakis
def7115fe4 prop271: Move new funcs to top, to avoid compiler warnings. 2017-01-17 14:35:38 +02:00
George Kadianakis
2938fd3b85 prop271: When we exhaust all guards, mark all of them for retry.
In the past, when we exhausted all guards in our sampled set, we just
waited there till we mark a guard for retry again (usually takes 10 mins
for a primary guard, 1 hour for a non-primary guard). This patch marks
all guards as maybe-reachable when we exhaust all guards (this can
happen when network is down for some time).
2017-01-17 14:35:38 +02:00
Nick Mathewson
b317813485 Make GETINFO entry-guards work again with prop271
This is not a great solution, but it's as close to
backward-compatible as possible.  A better GETINFO API should expose
more information.
2017-01-02 10:10:03 -05:00
Nick Mathewson
762b799545 Rename 'remove' -> 'rmv' to avoid shadowing a libc global 2016-12-16 14:04:57 -05:00
Nick Mathewson
23c09b6bc2 Resolve a division-by-zero complaint from coverity. CID 1397272 2016-12-16 12:21:02 -05:00
Nick Mathewson
990a863d7c Merge branch 'ticket20831_v2' 2016-12-16 11:40:19 -05:00
Nick Mathewson
506bd6d47c Make NumDirectoryGuards work with the new guard algorithm.
Now that we support NumEntryGuards, NumDirectoryGuards is pretty
easy to put back in.
2016-12-16 11:34:31 -05:00
Nick Mathewson
d9200d853d Make NumEntryGuards work as expected again.
Further, add a "guard-n-primary-guards-to-use" parameter, defaulting
to 1, for NumEntryGuards to override.
2016-12-16 11:34:31 -05:00
Nick Mathewson
deb8bcadce 271: Algorithm tweak to allow multiple entry guards.
Previously, we had NumEntryGuards kind of hardwired to 1.  Now we
have the code (but not the configuarability) to choose randomly from
among the first N primary guards that would work, where N defaults
to 1.

Part of 20831 support for making NumEntryGuards work again.
2016-12-16 11:34:31 -05:00
Nick Mathewson
385602e982 Respect GuardLifetime in prop271 code.
It overrides both the GUARD_LIFETIME and the
GUARD_CONFIRMED_MIN_LIFETIME options.
2016-12-16 11:34:31 -05:00
Nick Mathewson
988b0afbd6 Merge branch 'ticket20826_v2' 2016-12-16 11:29:02 -05:00
Nick Mathewson
e044b4f8ce Support restrictive ENTRYNODES configurations correctly.
Since we already had a separate function for getting the universe of
possible guards, all we had to do was tweak it to handle very the
GS_TYPE_RESTRICTED case.
2016-12-16 11:28:27 -05:00
Nick Mathewson
4ec9751c14 guard->nickname is never NULL. 2016-12-16 11:25:59 -05:00
Nick Mathewson
2cee38f76a Merge branch 'prop271_030_v1_squashed' 2016-12-16 11:20:59 -05:00
Nick Mathewson
20292ec497 Per suggestion, increase the retry frequency for primary guards. 2016-12-16 11:06:22 -05:00
Nick Mathewson
fc7751a989 Rewrite state transition logic in entry_guards_note_success()
asn found while testing that this function can be reached with
GUARD_STATE_COMPLETE circuits; I believe this happens when
cannibalization occurs.

The added complexity of handling one more state made it reasonable
to turn the main logic here into a switch statement.
2016-12-16 11:06:22 -05:00
Nick Mathewson
2e2f3a4d99 Add a separate, non-fractional, limit to the sampled guard set size.
Letting the maximum sample size grow proportionally to the number of
guards defeats its purpose to a certain extent.  Noted by asn during
code review.

Fixes bug 20920; bug not in any released (or merged) version of Tor.
2016-12-16 11:06:22 -05:00
Nick Mathewson
e50d85b90c Clean check for live consensus when updating the guard sample.
The valid_until check was redundant.
2016-12-16 11:06:22 -05:00
Nick Mathewson
b7088e5b5a Move a TODO comment into doxygen comments. 2016-12-16 11:06:22 -05:00
George Kadianakis
50783d0123 Easy code fixes.
- Correctly maintain the previous guard selection in choose_guard_selection().
- Print bridge identifier instead of nothing in entry_guard_describe()._
2016-12-16 11:06:22 -05:00
George Kadianakis
7ab2678074 Trivial documentation improvements. 2016-12-16 11:06:22 -05:00
Nick Mathewson
6867950432 Wrap all of the legacy guard code, and its users, in #ifdefs
This will make it easier to see what we remove down the line.
2016-12-16 11:06:22 -05:00
Nick Mathewson
2b4bfe62ee Fix a signed/unsigned warning on 32-bit 2016-12-16 11:06:21 -05:00
Nick Mathewson
52e196bab5 Don't make $hexid nicknames persistent.
(That's asking for trouble, and also totally completely redundant.)
2016-12-16 11:06:21 -05:00
Nick Mathewson
79d3e94f8b prop271: Tests for the highlevel or_state_t encode/decode functions 2016-12-16 11:06:20 -05:00
Nick Mathewson
217590ad05 Extract guard_selection_infer_type into its own function. 2016-12-16 11:06:20 -05:00
Nick Mathewson
2c8c58ab2f Another tweak for guard restrictions: don't let complete circs block
If a complete circuit C2 doesn't obey the restrictions of C1, then
C2 cannot block C1.

The patch here is a little big-ish, since we can no longer look
through all the complete circuits and all the waiting circuits on a
single pass: we have to find the best waiting circuit first.
2016-12-16 11:06:20 -05:00
Nick Mathewson
87f9b42179 Implement support for per-circuit guard restrictions.
This is an important thing I hadn't considered when writing prop271:
sometimes you have to restrict what guard you use for a particular
circuit.  Most frequently, that would be because you plan to use a
certain node as your exit, and so you can't choose that for your
guard.

This change means that the upgrade-waiting-circuits algorithm needs
a slight tweak too: circuit A cannot block circuit B from upgrading
if circuit B needs to follow a restriction that circuit A does not
follow.
2016-12-16 11:06:20 -05:00
Nick Mathewson
17c3faa2e3 guards_choose_dirguard(): replace one XXXX with another.
I had been asking myself, "hey, doesn't the new code need to look at
this "info" parameter? The old code did!"  But it turns out that the
old code hasn't, since 05f7336624.

So instead of "support this!" the comment now says "we can remove
this!"
2016-12-16 11:06:20 -05:00
Nick Mathewson
9d065ecc3d Fix a magic number in get_max_sample_size 2016-12-16 11:06:20 -05:00
Nick Mathewson
1e9cd5d2bb Note a couple of XXX-prop271s as spec deviations. 2016-12-16 11:06:19 -05:00
Nick Mathewson
f4e64c04f4 Remove some resolved "XXXX prop271" comments. 2016-12-16 11:06:19 -05:00
Nick Mathewson
80fa404625 Fix for small test networks: don't refuse to have any sampled guards.
Don't restrict the sample size if the network size is less than 20
guards.  Maybe we'll think of a better rule later on?
2016-12-16 11:06:19 -05:00
Nick Mathewson
eac8b3f758 Remove a few unused arguments. 2016-12-16 11:06:19 -05:00
Nick Mathewson
84bfa895d7 Change return value of entry_guard_succeeded to an enum.
George pointed out that (-1,0,1) for (never usable, maybe usable
later, usable right now) was a pretty rotten convention that made
the code harder to read.
2016-12-16 11:06:19 -05:00
Nick Mathewson
46619ec914 Note some large functions that could be split.
George Kadianakis pointed these out.
2016-12-16 11:06:19 -05:00
Nick Mathewson
3bcbbea350 More progress on bridge implementation with prop271 guards
Here we handle most (all?) of the remaining tasks, and fix some
bugs, in the prop271 bridge implementation.

  * We record bridge identities as we learn them.
  * We only call deprecated functions from bridges.c when the
    deprecated guard algorithm is in use.
  * We update any_bridge_descriptors_known() and
    num_bridges_usable() to work correctly with the new backend
    code. (Previously, they called into the guard selection logic.
  * We update bridge directory fetches to work with the new
    guard code.
  * We remove some erroneous assertions where we assumed that we'd
    never load a guard that wasn't for the current selection.

Also, we fix a couple of typos.
2016-12-16 11:06:18 -05:00
Nick Mathewson
82fa71610d Implement bridge backends for sampling, filtering guards.
Still missing is functionality for picking bridges when we don't
know a descriptor for them yet, and functionality for learning a
bridge ID.

Everything else remains (basically) the same. Neat!
2016-12-16 11:06:18 -05:00
Nick Mathewson
53f248f6c9 Add some needed accessors/inspectors for bridge/guard convergence 2016-12-16 11:06:18 -05:00
Nick Mathewson
1d52ac4d3f Lay down some infrastructure for bridges in the New Guard Order.
This includes:
  * making bridge_info_t exposed but opaque
  * allowing guards where we don't know an identity
  * making it possible to learn the identity of a guard
  * creating a guard that lacks a node_t
  * remembering a guard's address and port.
  * Looking up a guard by address and port.
  * Only enforcing the rule that we need a live consensus to update
    the "listed" status for guards when we are not using bridges.
2016-12-16 11:06:18 -05:00
Nick Mathewson
89f5f149df Remove guard_selection argument from status-reporting functions
This prevents us from mixing up multiple guard_selections
2016-12-16 11:06:18 -05:00
Nick Mathewson
6dcbc24a4e Add a backpointer from entry_guard_t to guard_selection_t
This is safe, because no entry_guard_t ever outlives its
guard_selection_t.

I want this because now that multiple guard selections can be active
during one tor session, we should make sure that any information we
register about guards is with respect to the selection that they came
from.
2016-12-16 11:06:18 -05:00
Nick Mathewson
404e9e5611 Have multiple guard contexts we can switch between.
Currently, this code doesn't actually have the contexts behave
differently, (except for the legacy context), but it does switch
back and forth between them nicely.
2016-12-16 11:06:18 -05:00
Nick Mathewson
c6d218c44b Unit tests for entry_guard_{pick_for_circuit,succeeded,failed} 2016-12-16 11:06:17 -05:00
Nick Mathewson
9493711077 Mark confirmed guards primary as appropriate.
If a guard becomes primary as a result of confirming it, consider
the circuit through that guard as a primary circuit.

Also, note open questions on behavior when confirming nonprimary guards
2016-12-16 11:06:17 -05:00
Nick Mathewson
d2af9826fd Turn #defines for prop271 into networkstatus params
Some of these will get torrc options to override them too; this
is just the mechanical conversion.

Also, add documentation for a couple of undocumented (but now used)
parameters.
2016-12-16 11:06:17 -05:00
Nick Mathewson
2ea5aa7182 Expire circuits that have been WAITING_FOR_BETTER_GUARD too long
(This is required by 3.9 in prop271, but is better done as a
separate function IMO)
2016-12-16 11:06:16 -05:00
Nick Mathewson
e56bc1e5de Move the 'dirty' flag for the guards to a global again
It makes more sense to have a single dirty flag, since we always
regenerate the whole state file when we save it.
2016-12-16 11:06:16 -05:00
Nick Mathewson
bce0f79252 Mark some more BUG lines as unreachable. 2016-12-16 11:06:16 -05:00
Nick Mathewson
a7bc73935b Test get_guard_selection_by_name 2016-12-16 11:06:15 -05:00
Nick Mathewson
526b0e2ce2 Avoid division-by-zero in pathbias_check_*_success_count 2016-12-16 11:06:15 -05:00
Nick Mathewson
ac67819396 Make sure primary-guards are up-to-date when we inspect them.
(Plus some magic to prevent and detect recursive invocation of
entry_guards_update_primary(), since that can cause some pretty
tricky misbehavior.)
2016-12-16 11:06:15 -05:00
Nick Mathewson
897626953b Rebuild the guard lists as appropriate on torrc change.
(Also, prepare to tie guard changes into the mark-all-old-circuits
logic.)
2016-12-16 11:06:15 -05:00
Nick Mathewson
6788418f28 Propagate Ed25519 identities downwards into more functions.
Actually set ed25519 identities on channels when we set a channel's
identity.
2016-12-08 16:47:58 -05:00
Nick Mathewson
a20c8a81d7 Migrate main data loop for set_bad_connections to use channel structures
This was the last user of our or_connections-by-ID map.  It also had
a tendency to be O(N) in cases that only had to be O(1).
2016-12-08 16:47:57 -05:00
Nick Mathewson
d98b9b6d65 Fix pathbias interactions with entry guards
entry_guard_get_by_id_digest() was always returning NULL, which was
causing "adventure" and "fun"
2016-11-30 14:44:43 -05:00
Nick Mathewson
783fa2f586 Make pathbias fields persistent for new guards 2016-11-30 14:44:43 -05:00
Nick Mathewson
858c8f5593 Make new prop271 entry guards persistent
To do this, it makes sense to treat legacy guards as a separate
guard_selection_t *, and handle them separately.  This also means we
add support here for having multiple guard selections.

Note that we don't persist pathbias information yet; that will take
some refactoring.
2016-11-30 14:44:43 -05:00
Nick Mathewson
dbbaa51518 Use the new guard notification/selection APIs throughout Tor
This patch doesn't cover every case; omitted cases are marked with
"XXXX prop271", as usual.  It leaves both the old interface and the
new interface for guard status notification, since they don't
actually work in the same way: the new API wants to be told when a
circuit has failed or succeeded, whereas the old API wants to know
when a channel has failed or succeeded.

I ran into some trouble with directory guard stuff, since when we
pick the directory guard, we don't actually have a circuit to
associate it with.  I solved that by allowing guard states to be
associated with directory connections, not just circuits.
2016-11-30 14:42:53 -05:00
Nick Mathewson
8e43398986 Function to cancel a guard state.
We'll want to use this if we allocate a guard state then decide,
"whoops, we don't want to use this."
2016-11-30 14:42:53 -05:00
Nick Mathewson
4689096ed1 No need to say success/failure when recording failure; remove returnval
(We can fail at succeeding, but there's no plausible way to fail at failing)
2016-11-30 14:42:53 -05:00
Nick Mathewson
af1918d289 New entry_guard_chan_failed function
To be called when an entire channel has failed: tell any/all
circuits pending for the guard of that channel that they have
failed.
2016-11-30 14:42:53 -05:00
Nick Mathewson
1fd0a547bb New function to tell the guard module "We're on the net!"
(Call it whenever we read a cell.)
2016-11-30 14:42:53 -05:00
Nick Mathewson
8dc6048c02 Add an (as yet) unused UseDeprecatedGuardAlgorithm_ option.
I expect we'll be ripping this out somewhere in 0.3.0, but let's
keep it around for a little while in case it turns out to be the
only way to avert disaster?
2016-11-30 14:42:53 -05:00
Nick Mathewson
36e9fbd752 Backend for upgrading 'waiting' circuits to 'complete'
When a nonprimary guard's circuit is complete, we don't call it
actually usable until we are pretty sure that every better guard
is indeed not going to give us a working circuit.
2016-11-30 14:42:52 -05:00
Nick Mathewson
dd6bdab3f6 Write the easy parts of the public entryguard interface.
Here we add a little bit of state to origin circuits, and set up
the necessary functions for the circuit code to call in order to
find guards, use guards, and decide when circuits can be used.

There's also an incomplete function for the hard part of the
circuit-maintenance code, where we figure out whether any waiting
guards are ready to become usable.

(This patch finally uses the handle.c code to make safe handles to
entry_guard_t objects, so that we are allowed to free an
entry_guard_t without checking whether any origin_circuit_t is
holding a reference to it.)
2016-11-30 14:42:52 -05:00
Nick Mathewson
7bf946965b Implement most of the prop271 data structure backends.
This code handles:
  * Maintaining the sampled set, the filtered set, and the
    usable_filtered set.
  * Maintaining the confirmed and primary guard lists.
  * Picking guards for circuits, and updating guard state when
    circuit state changes.

Additionally, I've done code structure movement: even more constants
and structures from entrynodes.c have become ENTRYNODES_PRIVATE
fields of entrynodes.h.

I've also included a bunch of documentation and a bunch of unit
tests.  Coverage on the new code is pretty high.

I've noted important things to resolve before this branch is done
with the /XXXX.*prop271/ regex.
2016-11-30 14:42:52 -05:00