The reason the "streams problem" occurs is due to the complicated
interaction between Tor's congestion control and libevent. At some point
during the experiment, the circuit window is exhausted, which blocks all
edge streams. When a circuit level sendme is received at Exit, it
resumes edge reading by looping over linked list of edge streams, and
calling connection_start_reading() to inform libevent to resume reading.
When the streams are activated again, Tor gets the chance to service the
first three streams activated before the circuit window is exhausted
again, which causes all streams to be blocked again. As an experiment,
we reversed the order in which the streams are activated, and indeed the
first three streams, rather than the last three, got service, while the
others starved.
Our solution is to change the order in which streams are activated. We
choose a random edge connection from the linked list, and then we
activate streams starting from that chosen stream. When we reach the end
of the list, then we continue from the head of the list until our chosen
stream (treating the linked list as a circular linked list). It would
probably be better to actually remember which streams have received
service recently, but this way is simple and effective.
A node_t is an abstraction over routerstatus_t, routerinfo_t, and
microdesc_t. It should try to present a consistent interface to all
of them. There should be a node_t for a server whenever there is
* A routerinfo_t for it in the routerlist
* A routerstatus_t in the current_consensus.
(note that a microdesc_t alone isn't enough to make a node_t exist,
since microdescriptors aren't usable on their own.)
There are three ways to get a node_t right now: looking it up by ID,
looking it up by nickname, and iterating over the whole list of
microdescriptors.
All (or nearly all) functions that are supposed to return "a router"
-- especially those used in building connections and circuits --
should return a node_t, not a routerinfo_t or a routerstatus_t.
A node_t should hold all the *mutable* flags about a node. This
patch moves the is_foo flags from routerinfo_t into node_t. The
flags in routerstatus_t remain, but they get set from the consensus
and should not change.
Some other highlights of this patch are:
* Looking up routerinfo and routerstatus by nickname is now
unified and based on the "look up a node by nickname" function.
This tries to look only at the values from current consensus,
and not get confused by the routerinfo_t->is_named flag, which
could get set for other weird reasons. This changes the
behavior of how authorities (when acting as clients) deal with
nodes that have been listed by nickname.
* I tried not to artificially increase the size of the diff here
by moving functions around. As a result, some functions that
now operate on nodes are now in the wrong file -- they should
get moved to nodelist.c once this refactoring settles down.
This moving should happen as part of a patch that moves
functions AND NOTHING ELSE.
* Some old code is now left around inside #if 0/1 blocks, and
should get removed once I've verified that I don't want it
sitting around to see how we used to do things.
There are still some unimplemented functions: these are flagged
with "UNIMPLEMENTED_NODELIST()." I'll work on filling in the
implementation here, piece by piece.
I wish this patch could have been smaller, but there did not seem to
be any piece of it that was independent from the rest. Moving flags
forces many functions that once returned routerinfo_t * to return
node_t *, which forces their friends to change, and so on.
When the CellStatistics option is off, we don't store cell insertion
times. Doing so would also not be very smart, because there seem to
still be some performance issues with this type of statistics. Nothing
harmful happens when we don't have insertion times, so we don't need to
alarm the user.
Previously[*], the function would start with the first stream on the
circuit, and let it package as many cells as it wanted before
proceeding to the next stream in turn. If a circuit had many live
streams that all wanted to package data, the oldest would get
preference, and the newest would get ignored.
Now, we figure out how many cells we're willing to send per stream,
and try to allocate them fairly.
Roger diagnosed this in the comments for bug 1298.
[*] This bug has existed since before the first-ever public release
of Tor. It was added by r152 of Tor on 26 Jan 2003, which was
the first commit to implement streams (then called "topics").
This is not the oldest bug to be fixed in 0.2.2.x: that honor
goes to the windowing bug in r54, which got fixed in e50b7768 by
Roger with diagnosis by Karsten. This is, however, the most
long-lived bug to be fixed in 0.2.2.x: the r54 bug was fixed
2580 days after it was introduced, whereas I am writing this
commit message 2787 days after r152.
I'm going to use this to implement more fairness in
circuit_resume_edge_reading_helper in an attempt to fix bug 1298.
(Updated with fixes from arma and Sebastian)
We frequently add cells to stream-blocked queues for valid reasons
that don't mean we need to block streams. The most obvious reason
is if the cell arrives over a circuit rather than from an edge: we
don't block circuits, no matter how full queues get. The next most
obvious reason is that we allow CONNECTED cells from a newly created
stream to get delivered just fine.
This patch changes the behavior so that we only iterate over the
streams on a circuit when the cell in question came from a stream,
and we only block the stream that generated the cell, so that other
streams can still get their CONNECTEDs in.
When this happens, run through the streams on the circuit and make
sure they're all blocked. If some aren't, that's a bug: block them
all and log it! If they all are, where did the cell come from? Log
it!
(I suspect that this actually happens pretty frequently, so I'm making
these log messages appear at INFO.)
Do not start reading on exit streams when we get a SENDME unless we
have space in the appropriate circuit's cell queue.
Draft fix for bug 1653.
(commit message by nickm)
At best, this patch helps us avoid sending queued relayed cells that
would get ignored during the time between when a destroy cell is
sent and when the circuit is finally freed. At worst, it lets us
release some memory a little earlier than it would otherwise.
Fix for bug #1184. Bugfix on 0.2.0.1-alpha.
Everything that accepted the 'Circ' name handled it wrong, so even now
that we fixed the handling of the parameter, we wouldn't be able to
set it without making all the 0.2.2.7..0.2.2.10 relays act wonky.
This patch makes Tors accept the 'Circuit' name instead, so we can
turn on circuit priorities without confusing the versions that treated
the 'Circ' name as occasion to act weird.
When you mean (a=b(c,d)) >= 0, you had better not say (a=b(c,d)>=0).
We did the latter, and so whenever CircPriorityHalflife was in the
consensus, it was treated as having a value of 1 msec (that is,
boolean true).
The rule is now: take the value from the CircuitPriorityHalflife
config option if it is set. If it zero, disable the cell_ewma
algorithm. If it is set, use it to calculate the scaling factor.
If it is not set, look for a CircPriorityHalflifeMsec parameter in the
consensus networkstatus. If *that* is zero, then disable the cell_ewma
algorithm; if it is set, use it to calculate the scaling factor.
If it is not set at all, disable the algorithm.
There are two big changes here:
- We store active circuits in a priority queue for each or_conn,
rather than doing a linear search over all the active circuits
before we send each cell.
- Rather than multiplying every circuit's cell-ewma by a decay
factor every time we send a cell (thus normalizing the value of a
current cell to 1.0 and a past cell to alpha^t), we instead
only scale down the cell-ewma every tick (ten seconds atm),
normalizing so that a cell sent at the start of the tick has
value 1.0).
Each circuit is ranked in terms of how many cells from it have been
relayed recently, using a time-weighted average.
This patch has been tested this on a private Tor network on PlanetLab,
and gotten improvements of 12-35% in time it takes to fetch a small
web page while there's a simultaneous large data transfer going on
simultaneously.
[Commit msg by nickm based on mail from Ian Goldberg.]
Some *_free functions threw asserts when passed NULL. Now all of them
accept NULL as input and perform no action when called that way.
This gains us consistence for our free functions, and allows some
code simplifications where an explicit null check is no longer necessary.
- Refactor geoip.c by moving duplicate code into rotate_request_period().
- Don't leak memory when cleaning up cell queues.
- Make sure that exit_(streams|bytes_(read|written)) are initialized in all
places accessing these arrays.
- Read only the last block from *stats files and ensure that its timestamp
is not more than 25 hours in the past and not more than 1 hour in the
future.
- Stop truncating the last character when reading *stats files.
The only thing that's left now is to avoid reading whole *stats files into
memory.
Send circuit or stream sendme cells when our window has decreased
by 100 cells, not when it has decreased by 101 cells. Bug uncovered
by Karsten when testing the "reduce circuit window" performance
patch. Bugfix on the 54th commit on Tor -- from July 2002,
before the release of Tor 0.0.0. This is the new winner of the
oldest-bug prize.
The problem is that clients and hidden services are receiving
relay_early cells, and they tear down the circuit.
Hack #1 is for rendezvous points to rewrite relay_early cells to
relay cells. That way there are never any incoming relay_early cells.
Hack #2 is for clients and hidden services to never send a relay_early
cell on an established rendezvous circuit. That works around rendezvous
points that haven't upgraded yet.
Hack #3 is for clients and hidden services to not tear down the circuit
when they receive an inbound relay_early cell. We already refuse extend
cells at clients.
When determining how long directory requests take or how long cells spend
in queues, we were comparing timestamps on microsecond detail only to
convert results to second or millisecond detail later on. But on 32-bit
architectures this means that 2^31 microseconds only cover time
differences of up to 36 minutes. Instead, compare timestamps on
millisecond detail.
Changes to directory request statistics:
- Rename GEOIP statistics to DIRREQ statistics, because they now include
more than only GeoIP-based statistics, whereas other statistics are
GeoIP-dependent, too.
- Rename output file from geoip-stats to dirreq-stats.
- Add new config option DirReqStatistics that is required to measure
directory request statistics.
- Clean up ChangeLog.
Also ensure that entry guards statistics have access to a local GeoIP
database.
Fix an edge case where a malicious exit relay could convince a
controller that the client's DNS question resolves to an internal IP
address. Bug found and fixed by "optimist"; bugfix on 0.1.2.8-beta.
Fix an edge case where a malicious exit relay could convince a
controller that the client's DNS question resolves to an internal IP
address. Bug found and fixed by "optimist"; bugfix on 0.1.2.8-beta.
The subversion $Id$ fields made every commit force a rebuild of
whatever file got committed. They were not actually useful for
telling the version of Tor files in the wild.
svn:r17867
"connecting" and it receives an "end" relay cell, the exit relay
would silently ignore the end cell and not close the stream. If
the client never closes the circuit, then the exit relay never
closes the TCP connection. Bug introduced in Tor 0.1.2.1-alpha;
reported by "wood".
svn:r17625
The "ClientDNSRejectInternalAddresses" config option wasn't being
consistently obeyed: if an exit relay refuses a stream because its
exit policy doesn't allow it, we would remember what IP address
the relay said the destination address resolves to, even if it's
an internal IP address. Bugfix on 0.2.0.7-alpha; patch by rovv.
svn:r17135
Initial conversion of uint32_t addr to tor_addr_t addr in connection_t and related types. Most of the Tor wire formats using these new types are in, but the code to generate and use it is not. This is a big patch. Let me know what it breaks for you.
svn:r16435
Part of fix for bug 617: allow connection_ap_handshake_attach_circuit() to mark connections, to avoid double-mark warnings. Note that this is an incomplete refactoring.
svn:r14066
Re-tune mempool parametes based on testing on peacetime: use smaller chuncks, free them a little more aggressively, and try very hard to concentrate allocations on fuller chunks. Also, lots of new documentation.
svn:r13484
Add a couple of (currently disabled) strategies for trying to avoid using too much ram in memory pools: prefer putting new cells in almost-full chunks, and be willing to free the last empty chunk if we have not needed it for a while. Also add better output to mp_pool_log_status to track how many mallocs a given memory pool strategy is saving us, so we can tune the mempool parameters.
svn:r13428
Be more thorough about memory poisoning and clearing. Add an in-place version of aes_crypt in order to remove a memcpy from relay_crypt_one_payload.
svn:r13414
Tor can warn and/or refuse connections to ports commonly used with
vulnerable-plaintext protocols.
We still need to figure out some good defaults for them.
svn:r13198
Use reference-counting to avoid allocating a zillion little addr_policy_t objects. (This is an old patch that had been sitting on my hard drive for a while.)
svn:r13017
Add a new ClientDNSRejectInternalAddresses option (default: on) to refuse to believe that any address can map to or from an internal address. This blocks some kinds of potential browser-based attacks, especially on hosts using DNSPort. Also clarify behavior in some comments. Backport candiate?
svn:r11287
Fix warnings from -Wunsafe-loop-optimizations, which incidentally turned up a logic bug in connection_or_flush_from_first_active_circuit that would overcount the number of cells flushed.
svn:r10199
Fix a bug in displaying memory pool usage. Also dump cell allocation, and track padded_cell_ts as they are allocated and freed, to make sure we are not leaking cells.
svn:r9992
Add code to shrink the cell memory pool by discarding empty chunks that have been empty for the last 60 seconds. Also, instead of having test.c duplicate declarations for exposed functions, put them inside #ifdef foo_PRIVATE blocks in the headers. This prevents bugs where test.c gets out of sync.
svn:r9944
Yet another attempted Bug 411 fix: Under some circumstances, a circuit can have cells without being active. The likeliest is that it has been unlinked from all connections in preparation for closing. Therefore, stop enforcing this non-invariant.
svn:r9936
Split type of "packed cell" from "parsed cell"; pack cells before queueing them on circuits. This will help us avoid dumb errors when we confuse the two types.
svn:r9935
A surprisingly simple patch to stop reading on edge connections when their circuits get too full, and start again when they empty out. This lets us remove the logic to block begin_dir conns when the corresponding or conns get full: it was already broken by cell queues anyway.
svn:r9905
Initial version of circuit-based cell queues. Instead of hammering or_conns with piles of cells, queue cells on their corresponding circuits, and append them to the or_conn as needed. This seems to work so far, but needs a bit more work. This will break the memory-use-limitation patch for begin_dir conns: the solution will be a fun but fiddly.
svn:r9904
Make all LD_BUG log messsages get prefixed with "Bug: ". Remove manually-generated "Bug: "s from log-messages. (Apparently, we remembered to add them about 40% of the time.)
svn:r9733
Make remap stream events have a souce; make them generated every time we get a successful connected or resolved cell. Also change reported stream target address to IP consistently when we get the IP from an exit node.
svn:r9624
Report stream end events where a resolve succeeded or where we got a socks protocol error correctly, rather than calling both of them "INTERNAL". Turn ALREADY_SOCKS_REPLIED into a flag rather than a reason. This will help debug 367 part 2 a little.
svn:r9511
Add a REMAP state to stream events so that controllers can learn exactly when the target address for a stream has changed. May help Vidalia resolve confusions related to bug 375.
svn:r9484
New socks command CONNECT_DIR. New config option TunnelDirConns that
builds a circ ending at the directory server and delivers a BEGIN_DIR
cell if it's running 0.1.2.2-alpha or later. We still need to make
one-hop circs when appropriate, while making other conns avoid them.
svn:r9098
Have connection_about_to_close use an end_reason field in edge_connection_t to tell what reason to tell the controller for closing the stream. Set end_reason in connection_edge_end, connection_mark_unattached_ap, and everwhere we set edge_has_sent_end. Add a changelog entry.
svn:r8779
Weasel correctly notes that we should not discard the return value from connection_exit_begin_con. Right now, the return value is always discardable, so this does not actually cause a bug, but it might later. So fix it.
svn:r8774
Fix longstanding bug in connection_exit_begin_conn(): Since connection_edge_end() exits when the connection is unattached, we were never sending RELAY_END cells back for failed RELAY_BEGIN attempts. Fix this. This might make clients that were otherwise timing out either fail faster or retry faster, which is good news for us.
svn:r8770
[Needs review.] Add a BEGIN_DIR relay cell type for an easier
in-protocol way to connect to directory servers through Tor.
Previously, clients could only connect to director servers over Tor
from exit nodes, but couldn't get directory information anonymously
from a non-exit cache without getting a directory server involved.
This needs testing, and needs client-side code to actually exercise it.
svn:r8527
Patch from Tup to add support for transparent AP connections: this basically bundles the functionality of trans-proxy-tor into the tor mainline. Now hosts with compliant pf/netfilter implementations can redirect TCP connections straight to Tor without diverting through SOCKS.
svn:r7007
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
circuit_t into origin_circuit_t and or_circuit_t. I fixed some
segaults; there may be more. We still need to move more rendezvous
stuff into subtypes.
This is a trial run for splitting up connection_t; if the approach is
insane, please say so soon so we can do something smarter.
Also, this discards the old HALF_OPEN code, which nobody seems to
want.
svn:r6817
we screwed up the formatting in wild and unpredictable ways.
fix it before it becomes convention to format logs in wild and
unpredictable ways.
still need to do src/common/ someday.
svn:r5551
don't recognize. now we just drop it. perhaps this will make us
more forward-compatible? or perhaps it will bite us? one day we
will find out.
svn:r5405
its advertised exit policy is different from its real one) as soon
as it refused any requests.
After fixing that bug, another bug appeared: we would try the same
server again and again, since once we learned an IP address for a
hostname, we still kept think of it as the hostname. now pass it to
the remapper before trying to reattach.
svn:r4962
- Add a new extend_info_t datatype to hold information needed to
extend a circuit (addr,port,keyid,onion_key). Use it in cpath and
build_state. Make appropriate functions take or return it instead of
routerinfo_t or keyid.
- #if 0 needless check in circuit_get_by_edge_conn; if nobody triggers this
error in 0.1.0.10, nobody will trigger it.
- Implement new hidden service descriptor format, which contains "extend
info" for introduction points, along with protocol version list.
- Parse new format.
- Generate new format
- Cache old and new formats alongside each other.
- Directories serve "old" format if asked in old way, "newest available"
format if asked in new way.
- Use new format to find introduction points if possible; otherwise fall
back. Keep nickname lists and extendinfo lists in sync.
- Tests for new format.
- Implement new "v2" INTRODUCE cell format.
- Accept new format
- Use new format if we have a versioned service descriptor that says the
server accepts the new format.
- Add documentation for functions and data types.
svn:r4506
