zoukankan      html  css  js  c++  java
  • Tor路径选择说明

    Tor Path Specification

     

                                  Roger Dingledine

                                   Nick Mathewson

     

    Note: This is an attempt to specify Tor as currently implemented.  Future

    versions of Tor will implement improved algorithms.

     

    This document tries to cover how Tor chooses to build circuits and assign streams to circuits.  Other implementations MAY take other approaches, but implementors should be aware of the anonymity and load-balancing implications of their choices.

     

                        THIS SPEC ISN'T DONE YET.

     

          The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED",  "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119.

     

    1. General operation

     

       Tor begins building circuits as soon as it has enough directory information to do so (see section 5 of dir-spec.txt).  Some circuits are built preemptively because we expect to need them later (for user traffic), and some are built because of immediate need (for user traffic that no current circuit can handle, for testing the network or our reachability, and so on).

     

      [Newer versions of Tor (0.2.6.2-alpha and later): If the consensus contains Exits (the typical case), Tor will build both exit and internal circuits. When bootstrap completes, Tor will be ready to handle an application requesting an exit circuit to services like the World Wide Web.

     

       If the consensus does not contain Exits, Tor will only build internal circuits. In this case, earlier statuses will have included "internal" as indicated above. When bootstrap completes, Tor will be ready to handle an application requesting an internal circuit to hidden services at ".onion" addresses.

     

       If a future consensus contains Exits, exit circuits may become available.]

     

       When a client application creates a new stream (by opening a SOCKS connection or launching a resolve request), we attach it to an appropriate open circuit if one exists, or wait if an appropriate circuit is in-progress. We launch a new circuit only

       if no current circuit can handle the request.  We rotate circuits over time to avoid some profiling attacks.

     

       To build a circuit, we choose all the nodes we want to use, and then construct the circuit.  Sometimes, when we want a circuit that ends at a given hop, and we have an appropriate unused circuit, we "cannibalize" the existing circuit and extend it to the new terminus.

     

       These processes are described in more detail below.

     

       This document describes Tor's automatic path selection logic only; path selection can be overridden by a controller (with the EXTENDCIRCUIT and ATTACHSTREAM commands).  Paths constructed through these means may violate some constraints given below.

     

    1.1. Terminology

     

       A "path" is an ordered sequence of nodes, not yet built as a circuit.

     

       A "clean" circuit is one that has not yet been used for any traffic.

     

       A "fast" or "stable" or "valid" node is one that has the 'Fast' or 'Stable' or 'Valid' flag set respectively, based on our current directory information.  A "fast" or "stable" circuit is one consisting only of "fast" or "stable" nodes.

     

       In an "exit" circuit, the final node is chosen based on waiting stream requests if any, and in any case it avoids nodes with exit policy of "reject *:*". An "internal" circuit, on the other hand, is one where the final node is chosen just like a middle node (ignoring its exit policy).

     

       A "request" is a client-side stream or DNS resolve that needs to be served by a circuit.

     

       A "pending" circuit is one that we have started to build, but which has not yet completed.

     

       A circuit or path "supports" a request if it is okay to use the circuit/path to fulfill the request, according to the rules given below. A circuit or path "might support" a request if some aspect of the request is unknown (usually its target IP), but we believe the path probably supports the request according to the rules given below.

     

    1.1. A relay's bandwidth

     

       Old versions of Tor did not report bandwidths in network status documents, so clients had to learn them from the routers' advertised relay descriptors.

     

       For versions of Tor prior to 0.2.1.17-rc, everywhere below where we refer to a relay's "bandwidth", we mean its clipped advertised bandwidth, computed by taking the smaller of the 'rate' and 'observed' arguments to the "bandwidth" element in the relay's .  If a router's advertised bandwidth is greater than MAX_BELIEVABLE_BANDWIDTH (currently 10 MB/s), we clipped to that value.

     

       For more recent versions of Tor, we take the bandwidth value declared in the consensus, and fall back to the clipped advertised bandwidth only if the consensus does not have bandwidths listed.

     

    2. Building circuits

     

    2.1. When we build

     

    2.1.0. We don't build circuits until we have enough directory info

     

       There's a class of possible attacks where our directory servers only give us information about the relays that they would like us to use.  To prevent this attack, we don't build multi-hop circuits for real traffic (like those in 2.1.1, 2.1.2, 2.1.4 below) until we have enough directory information to be reasonably confident this attack isn't being done to us.

     

       Here, "enough" directory information is defined as:

     

          * Having a consensus that's been valid at some point in the last REASONABLY_LIVE_TIME interval (24 hourts).

     

          * Having enough descriptors that we could build at least some fraction F of all bandwidth-weighted paths, without taking ExitNodes/EntryNodes/etc into account.

     

            (F is set by the PathsNeededToBuildCircuits option, defaulting to the 'min_paths_for_circs_pct' consensus parameter, with a final default value of 60%.)

     

          * Having enough descriptors that we could build at least some fraction F of all bandwidth-weighted paths, _while_ taking ExitNodes/EntryNodes/etc into account.

     

            (F is as above.)

     

          * Having a descriptor for every one of the first NUM_GUARDS_TO_USE guards among our primary guards. (see guard-spec.txt)

     

     

    2.1.1. Clients build circuits preemptively

     

       When running as a client, Tor tries to maintain at least a certain number of clean circuits, so that new streams can be handled quickly.  To increase the likelihood of success, Tor tries to predict what circuits will be useful by choosing from among nodes that support the ports we have used in the recent past (by default one hour). Specifically, on startup Tor tries to maintain one clean fast exit circuit that allows connections to port 80, and at least two fast clean stable internal circuits in case we get a resolve request or hidden service request (at least three if we _run_ a hidden service).

     

       After that, Tor will adapt the circuits that it preemptively builds based on the requests it sees from the user: it tries to have two fast clean exit circuits available for every port seen within the past hour (each circuit can be adequate for many predicted ports -- it doesn't need two separate circuits for each port), and it tries to have the above internal circuits available if we've seen resolves or hidden service activity within the past hour. If there are 12 or more clean circuits open, it doesn't open more even if it has more predictions.

     

       Only stable circuits can "cover" a port that is listed in the LongLivedPorts config option. Similarly, hidden service requests to ports listed in LongLivedPorts make us create stable internal circuits.

     

       Note that if there are no requests from the user for an hour, Tor will predict no use and build no preemptive circuits.

     

       The Tor client SHOULD NOT store its list of predicted requests to a persistent medium.

     

    2.1.2. Clients build circuits on demand

     

       Additionally, when a client request exists that no circuit (built or pending) might support, we create a new circuit to support the request. For exit connections, we pick an exit node that will handle the most pending requests (choosing arbitrarily among ties), launch a circuit to end there, and repeat until every unattached request might be supported by a pending or built circuit. For internal circuits, we pick an arbitrary acceptable path, repeating as needed.

     

       In some cases we can reuse an already established circuit if it's clean; see Section 2.3 (cannibalizing circuits) for details.

     

    2.1.3. Relays build circuits for testing reachability and bandwidth

     

       Tor relays test reachability of their ORPort once they have successfully built a circuit (on startup and whenever their IP address changes). They build an ordinary fast internal circuit with themselves as the last hop. As soon as any testing circuit succeeds, the Tor relay decides it's reachable and is willing to publish a descriptor.

     

       We launch multiple testing circuits (one at a time), until we have NUM_PARALLEL_TESTING_CIRC (4) such circuits open. Then we do a "bandwidth test" by sending a certain number of relay drop cells down each circuit: BandwidthRate * 10 / CELL_NETWORK_SIZE total cells divided across the four circuits, but never more than CIRCWINDOW_START (1000) cells total. This exercises both outgoing and incoming bandwidth, and helps to jumpstart the observed bandwidth (see dir-spec.txt).

     

       Tor relays also test reachability of their DirPort once they have established a circuit, but they use an ordinary exit circuit for this purpose.

     

    2.1.4. Hidden-service circuits

     

       See section 4 below.

     

    2.1.5. Rate limiting of failed circuits

     

       If we fail to build a circuit N times in a X second period (see Section

       2.3 for how this works), we stop building circuits until the X seconds

       have elapsed.

       XXXX

     

    2.1.6. When to tear down circuits

     

       XXXX

     

     

    2.2. Path selection and constraints

     

       We choose the path for each new circuit before we build it.  We choose the exit node first, followed by the other nodes in the circuit.  All paths we generate obey the following constraints:

         - We do not choose the same router twice for the same path.

         - We do not choose any router in the same family as another in the same

           path. (Two routers are in the same family if each one lists the other

           in the "family" entries of its descriptor.)

         - We do not choose more than one router in a given /16 subnet

           (unless EnforceDistinctSubnets is 0).

         - We don't choose any non-running or non-valid router unless we have

           been configured to do so. By default, we are configured to allow

           non-valid routers in "middle" and "rendezvous" positions.

         - If we're using Guard nodes, the first node must be a Guard (see 5

           below)

         - XXXX Choosing the length

     

       For "fast" circuits, we only choose nodes with the Fast flag. For

       non-"fast" circuits, all nodes are eligible.

     

       For all circuits, we weight node selection according to router bandwidth.

     

       We also weight the bandwidth of Exit and Guard flagged nodes depending on

       the fraction of total bandwidth that they make up and depending upon the

       position they are being selected for.

     

       These weights are published in the consensus, and are computed as described

       in Section "Computing Bandwidth Weights" of dir-spec.txt. They are:

     

          Wgg - Weight for Guard-flagged nodes in the guard position

          Wgm - Weight for non-flagged nodes in the guard Position

          Wgd - Weight for Guard+Exit-flagged nodes in the guard Position

     

          Wmg - Weight for Guard-flagged nodes in the middle Position

          Wmm - Weight for non-flagged nodes in the middle Position

          Wme - Weight for Exit-flagged nodes in the middle Position

          Wmd - Weight for Guard+Exit flagged nodes in the middle Position

     

          Weg - Weight for Guard flagged nodes in the exit Position

          Wem - Weight for non-flagged nodes in the exit Position

          Wee - Weight for Exit-flagged nodes in the exit Position

          Wed - Weight for Guard+Exit-flagged nodes in the exit Position

     

          Wgb - Weight for BEGIN_DIR-supporting Guard-flagged nodes

          Wmb - Weight for BEGIN_DIR-supporting non-flagged nodes

          Web - Weight for BEGIN_DIR-supporting Exit-flagged nodes

          Wdb - Weight for BEGIN_DIR-supporting Guard+Exit-flagged nodes

     

          Wbg - Weight for Guard+Exit-flagged nodes for BEGIN_DIR requests

          Wbm - Weight for Guard+Exit-flagged nodes for BEGIN_DIR requests

          Wbe - Weight for Guard+Exit-flagged nodes for BEGIN_DIR requests

          Wbd - Weight for Guard+Exit-flagged nodes for BEGIN_DIR requests

     

       If any of those weights is malformed or not present in a consensus,

       clients proceed with the regular path selection algorithm setting

       the weights to the default value of 10000.

     

       Additionally, we may be building circuits with one or more requests in

       mind.  Each kind of request puts certain constraints on paths:

     

         - All service-side introduction circuits and all rendezvous paths

           should be Stable.

         - All connection requests for connections that we think will need to

           stay open a long time require Stable circuits.  Currently, Tor decides

           this by examining the request's target port, and comparing it to a

           list of "long-lived" ports. (Default: 21, 22, 706, 1863, 5050,

           5190, 5222, 5223, 6667, 6697, 8300.)

         - DNS resolves require an exit node whose exit policy is not equivalent

           to "reject *:*".

         - Reverse DNS resolves require a version of Tor with advertised eventdns

           support (available in Tor 0.1.2.1-alpha-dev and later).

         - All connection requests require an exit node whose exit policy

           supports their target address and port (if known), or which "might

           support it" (if the address isn't known).  See 2.2.1.

         - Rules for Fast? XXXXX

     

    2.2.1. Choosing an exit

     

       If we know what IP address we want to connect to or resolve, we can

       trivially tell whether a given router will support it by simulating

       its declared exit policy.

     

       Because we often connect to addresses of the form hostname:port, we do not

       always know the target IP address when we select an exit node.  In these

       cases, we need to pick an exit node that "might support" connections to a

       given address port with an unknown address.  An exit node "might support"

       such a connection if any clause that accepts any connections to that port

       precedes all clauses (if any) that reject all connections to that port.

     

       Unless requested to do so by the user, we never choose an exit node

       flagged as "BadExit" by more than half of the authorities who advertise

       themselves as listing bad exits.

     

    2.2.2. User configuration

     

       Users can alter the default behavior for path selection with configuration

       options.

     

       - If "ExitNodes" is provided, then every request requires an exit node on

         the ExitNodes list.  (If a request is supported by no nodes on that list,

         and StrictExitNodes is false, then Tor treats that request as if

         ExitNodes were not provided.)

     

       - "EntryNodes" and "StrictEntryNodes" behave analogously.

     

       - If a user tries to connect to or resolve a hostname of the form

         <target>.<servername>.exit, the request is rewritten to a request for

         <target>, and the request is only supported by the exit whose nickname

         or fingerprint is <servername>.

     

       - When set, "HSLayer2Nodes" and "HSLayer3Nodes" relax Tor's path

         restrictions to allow nodes in the same /16 and node family to reappear

         in the path. They also allow the guard node to be chosen as the RP, IP,

         and HSDIR, and as the hop before those positions.

     

    2.3. Cannibalizing circuits

     

       If we need a circuit and have a clean one already established, in

       some cases we can adapt the clean circuit for our new

       purpose. Specifically,

     

       For hidden service interactions, we can "cannibalize" a clean internal

       circuit if one is available, so we don't need to build those circuits

       from scratch on demand.

     

       We can also cannibalize clean circuits when the client asks to exit

       at a given node -- either via the ".exit" notation or because the

       destination is running at the same location as an exit node.

     

    2.4. Learning when to give up ("timeout") on circuit construction

     

       Since version 0.2.2.8-alpha, Tor attempts to learn when to give up on

       circuits based on network conditions.

     

    2.4.1 Distribution choice and parameter estimation

     

       Based on studies of build times, we found that the distribution of

       circuit build times appears to be a Frechet distribution. However,

       estimators and quantile functions of the Frechet distribution are

       difficult to work with and slow to converge. So instead, since we

       are only interested in the accuracy of the tail, we approximate

       the tail of the distribution with a Pareto curve.

     

       We calculate the parameters for a Pareto distribution fitting the data

       using the estimators in equation 4 from:

       http://portal.acm.org/citation.cfm?id=1647962.1648139

     

       This is:

     

          alpha_m = s/(ln(U(X)/Xm^n))

     

       where s is the total number of completed circuits we have seen, and

     

          U(X) = x_max^u * Prod_s{x_i}

     

       with x_i as our i-th completed circuit time, x_max as the longest

       completed circuit build time we have yet observed, u as the

       number of unobserved timeouts that have no exact value recorded,

       and n as u+s, the total number of circuits that either timeout or

       complete.

     

       Using log laws, we compute this as the sum of logs to avoid

       overflow and ln(1.0+epsilon) precision issues:

     

           alpha_m = s/(u*ln(x_max) + Sum_s{ln(x_i)} - n*ln(Xm))

     

       This estimator is closely related to the parameters present in:

       http://en.wikipedia.org/wiki/Pareto_distribution#Parameter_estimation

       except they are adjusted to handle the fact that our samples are

       right-censored at the timeout cutoff.

     

       Additionally, because this is not a true Pareto distribution, we alter

       how Xm is computed. The Xm parameter is computed as the midpoint of the most

       frequently occurring 50ms histogram bin, until the point where 1000

       circuits are recorded. After this point, the weighted average of the top

       'cbtnummodes' (default: 3) midpoint modes is used as Xm. All times below

       this value are counted as having the midpoint value of this weighted average

       bin.

     

       The timeout itself is calculated by using the Pareto Quantile function (the

       inverted CDF) to give us the value on the CDF such that 80% of the mass

       of the distribution is below the timeout value.

     

       Thus, we expect that the Tor client will accept the fastest 80% of

       the total number of paths on the network.

     

    2.4.2. How much data to record

     

       From our observations, the minimum number of circuit build times for a

       reasonable fit appears to be on the order of 100. However, to keep a

       good fit over the long term, we store 1000 most recent circuit build times

       in a circular array.

     

       The Tor client should build test circuits at a rate of one per

       minute up until 100 circuits are built. This allows a fresh Tor to have

       a CircuitBuildTimeout estimated within 1.5 hours after install,

       upgrade, or network change (see below).

     

       Timeouts are stored on disk in a histogram of 50ms bin width, the same

       width used to calculate the Xm value above. This histogram must be shuffled

       after being read from disk, to preserve a proper expiration of old values

       after restart.

     

    2.4.3. How to record timeouts

     

       Circuits that pass the timeout threshold should be allowed to continue

       building until a time corresponding to the point 'cbtclosequantile'

       (default 95) on the Pareto curve, or 60 seconds, whichever is greater.

     

       The actual completion times for these circuits should be recorded.

       Implementations should completely abandon a circuit and record a value

       as an 'unknown' timeout if the total build time exceeds this threshold.

     

       The reason for this is that right-censored pareto estimators begin to lose

       their accuracy if more than approximately 5% of the values are censored.

       Since we wish to set the cutoff at 20%, we must allow circuits to continue

       building past this cutoff point up to the 95th percentile.

     

    2.4.4. Detecting Changing Network Conditions

     

       We attempt to detect both network connectivity loss and drastic

       changes in the timeout characteristics.

     

       We assume that we've had network connectivity loss if a circuit

       times out and we've received no cells or TLS handshakes since that

       circuit began. We then temporarily stop counting timeouts until

       network activity resumes.

     

       To detect changing network conditions, we keep a history of

       the timeout or non-timeout status of the past 20 circuits that

       successfully completed at least one hop. If more than 90% of

       these circuits timeout, we discard all buildtimes history, reset

       the timeout to 60, and then begin recomputing the timeout.

     

       If the timeout was already 60 or higher, we double the timeout.

     

    2.4.5. Consensus parameters governing behavior

     

       Clients that implement circuit build timeout learning should obey the

       following consensus parameters that govern behavior, in order to allow

       us to handle bugs or other emergent behaviors due to client circuit

       construction. If these parameters are not present in the consensus,

       the listed default values should be used instead.

     

          cbtdisabled

            Default: 0

            Min: 0

            Max: 1

            Effect: If 1, all CircuitBuildTime learning code should be

                    disabled and history should be discarded. For use in

                    emergency situations only.

     

          cbtnummodes

            Default: 3

            Min: 1

            Max: 20

            Effect: This value governs how many modes to use in the weighted

            average calculation of Pareto parameter Xm. A value of 3 introduces

            some bias (2-5% of CDF) under ideal conditions, but allows for better

            performance in the event that a client chooses guard nodes of radically

            different performance characteristics.

     

          cbtrecentcount

            Default: 20

            Min: 3

            Max: 1000

            Effect: This is the number of circuit build times to keep track of

                    for the following option.

     

          cbtmaxtimeouts

            Default: 18

            Min: 3

            Max: 10000

            Effect: When this many timeouts happen in the last 'cbtrecentcount'

                    circuit attempts, the client should discard all of its

                    history and begin learning a fresh timeout value.

     

          cbtmincircs

            Default: 100

            Min: 1

            Max: 10000

            Effect: This is the minimum number of circuits to build before

                    computing a timeout.

     

          cbtquantile

            Default: 80

            Min: 10

            Max: 99

            Effect: This is the position on the quantile curve to use to set the

                    timeout value. It is a percent (10-99).

     

          cbtclosequantile

            Default: 95

            Min: Value of cbtquantile parameter

            Max: 99

            Effect: This is the position on the quantile curve to use to set the

                    timeout value to use to actually close circuits. It is a

                    percent (0-99).

     

          cbttestfreq

            Default: 60

            Min: 1

            Max: 2147483647 (INT32_MAX)

            Effect: Describes how often in seconds to build a test circuit to

                    gather timeout values. Only applies if less than 'cbtmincircs'

                    have been recorded.

     

          cbtmintimeout

            Default: 2000

            Min: 500

            Max: 2147483647 (INT32_MAX)

            Effect: This is the minimum allowed timeout value in milliseconds.

                    The minimum is to prevent rounding to 0 (we only check once

                    per second).

     

          cbtinitialtimeout

            Default: 60000

            Min: Value of cbtmintimeout

            Max: 2147483647 (INT32_MAX)

            Effect: This is the timeout value to use before computing a timeout,

                    in milliseconds.

     

          cbtlearntimeout

            Default: 180

            Min: 10

            Max: 60000

            Effect: This is how long idle circuits will be kept open while cbt is

                    learning a new timeout value.

     

          cbtmaxopencircs

            Default: 10

            Min: 0

            Max: 14

            Effect: This is the maximum number of circuits that can be open at

                    at the same time during the circuit build time learning phase.

     

    2.5. Handling failure

     

       If an attempt to extend a circuit fails (either because the first create

       failed or a subsequent extend failed) then the circuit is torn down and is

       no longer pending.  (XXXX really?)  Requests that might have been

       supported by the pending circuit thus become unsupported, and a new

       circuit needs to be constructed.

     

       If a stream "begin" attempt fails with an EXITPOLICY error, we

       decide that the exit node's exit policy is not correctly advertised,

       so we treat the exit node as if it were a non-exit until we retrieve

       a fresh descriptor for it.

     

       Excessive amounts of either type of failure can indicate an

       attack on anonymity. See section 7 for how excessive failure is handled.

     

    3. Attaching streams to circuits

     

       When a circuit that might support a request is built, Tor tries to attach

       the request's stream to the circuit and sends a BEGIN, BEGIN_DIR,

       or RESOLVE relay

       cell as appropriate.  If the request completes unsuccessfully, Tor

       considers the reason given in the CLOSE relay cell. [XXX yes, and?]

     

     

       After a request has remained unattached for SocksTimeout (2 minutes

       by default), Tor abandons the attempt and signals an error to the

       client as appropriate (e.g., by closing the SOCKS connection).

     

       XXX Timeouts and when Tor auto-retries.

        * What stream-end-reasons are appropriate for retrying.

     

       If no reply to BEGIN/RESOLVE, then the stream will timeout and fail.

     

    4. Hidden-service related circuits

     

      XXX Tracking expected hidden service use (client-side and hidserv-side)

     

    5. Guard nodes

     

      We use Guard nodes (also called "helper nodes" in the research

      literature) to prevent certain profiling attacks. For an overview of

      our Guard selection algorithm -- which has grown rather complex -- see

      guard-spec.txt.

     

    5.1. How consensus bandwidth weights factor into entry guard selection

     

      When weighting a list of routers for choosing an entry guard, the following

      consensus parameters (from the "bandwidth-weights" line) apply:

     

          Wgg - Weight for Guard-flagged nodes in the guard position

          Wgm - Weight for non-flagged nodes in the guard Position

          Wgd - Weight for Guard+Exit-flagged nodes in the guard Position

          Wgb - Weight for BEGIN_DIR-supporting Guard-flagged nodes

          Wmb - Weight for BEGIN_DIR-supporting non-flagged nodes

          Web - Weight for BEGIN_DIR-supporting Exit-flagged nodes

          Wdb - Weight for BEGIN_DIR-supporting Guard+Exit-flagged nodes

     

      Please see "bandwidth-weights" in §3.4.1 of dir-spec.txt for more in depth

      descriptions of these parameters.

     

      If a router has been marked as both an entry guard and an exit, then we

      prefer to use it more, with our preference for doing so (roughly) linearly

      increasing w.r.t. the router's non-guard bandwidth and bandwidth weight

      (calculated without taking the guard flag into account).  From proposal

      #236:

        |

        | Let Wpf denote the weight from the 'bandwidth-weights' line a

        | client would apply to N for position p if it had the guard

        | flag, Wpn the weight if it did not have the guard flag, and B the

        | measured bandwidth of N in the consensus.  Then instead of choosing

        | N for position p proportionally to Wpf*B or Wpn*B, clients should

        | choose N proportionally to F*Wpf*B + (1-F)*Wpn*B.

     

      where F is the weight as calculated using the above parameters.

     

    6. Server descriptor purposes

     

      There are currently three "purposes" supported for server descriptors:

      general, controller, and bridge. Most descriptors are of type general

      -- these are the ones listed in the consensus, and the ones fetched

      and used in normal cases.

     

      Controller-purpose descriptors are those delivered by the controller

      and labelled as such: they will be kept around (and expire like

      normal descriptors), and they can be used by the controller in its

      CIRCUITEXTEND commands. Otherwise they are ignored by Tor when it

      chooses paths.

     

      Bridge-purpose descriptors are for routers that are used as bridges. See

      doc/design-paper/blocking.pdf for more design explanation, or proposal

      125 for specific details. Currently bridge descriptors are used in place

      of normal entry guards, for Tor clients that have UseBridges enabled.

     

    7. Detecting route manipulation by Guard nodes (Path Bias)

     

      The Path Bias defense is designed to defend against a type of route

      capture where malicious Guard nodes deliberately fail or choke circuits

      that extend to non-colluding Exit nodes to maximize their network

      utilization in favor of carrying only compromised traffic.

     

      In the extreme, the attack allows an adversary that carries c/n

      of the network capacity to deanonymize c/n of the network

      connections, breaking the O((c/n)^2) property of Tor's original

      threat model. It also allows targeted attacks aimed at monitoring

      the activity of specific users, bridges, or Guard nodes.

     

      There are two points where path selection can be manipulated:

      during construction, and during usage. Circuit construction

      can be manipulated by inducing circuit failures during circuit

      extend steps, which causes the Tor client to transparently retry

      the circuit construction with a new path. Circuit usage can be

      manipulated by abusing the stream retry features of Tor (for

      example by withholding stream attempt responses from the client

      until the stream timeout has expired), at which point the tor client

      will also transparently retry the stream on a new path.

     

      The defense as deployed therefore makes two independent sets of

      measurements of successful path use: one during circuit construction,

      and one during circuit usage.

     

      The intended behavior is for clients to ultimately disable the use

      of Guards responsible for excessive circuit failure of either type

      (see section 7.4); however known issues with the Tor network currently

      restrict the defense to being informational only at this stage (see

      section 7.5).

     

    7.1. Measuring path construction success rates

     

      Clients maintain two counts for each of their guards: a count of the

      number of times a circuit was extended to at least two hops through that

      guard, and a count of the number of circuits that successfully complete

      through that guard. The ratio of these two numbers is used to determine

      a circuit success rate for that Guard.

     

      Circuit build timeouts are counted as construction failures if the

      circuit fails to complete before the 95% "right-censored" timeout

      interval, not the 80% timeout condition (see section 2.4).

     

      If a circuit closes prematurely after construction but before being

      requested to close by the client, this is counted as a failure.

     

    7.2. Measuring path usage success rates

     

      Clients maintain two usage counts for each of their guards: a count

      of the number of usage attempts, and a count of the number of

      successful usages.

     

      A usage attempt means any attempt to attach a stream to a circuit.

     

      Usage success status is temporarily recorded by state flags on circuits.

      Guard usage success counts are not incremented until circuit close. A

      circuit is marked as successfully used if we receive a properly

      recognized RELAY cell on that circuit that was expected for the current

      circuit purpose.

     

      If subsequent stream attachments fail or time out, the successfully used

      state of the circuit is cleared, causing it once again to be regarded

      as a usage attempt only.

     

      Upon close by the client, all circuits that are still marked as usage

      attempts are probed using a RELAY_BEGIN cell constructed with a

      destination of the form 0.a.b.c:25, where a.b.c is a 24 bit random

      nonce. If we get a RELAY_COMMAND_END in response matching our nonce,

      the circuit is counted as successfully used.

     

      If any unrecognized RELAY cells arrive after the probe has been sent,

      the circuit is counted as a usage failure.

     

      If the stream failure reason codes DESTROY, TORPROTOCOL, or INTERNAL

      are received in response to any stream attempt, such circuits are not

      probed and are declared usage failures.

     

      Prematurely closed circuits are not probed, and are counted as usage

      failures.

     

    7.3. Scaling success counts

     

      To provide a moving average of recent Guard activity while

      still preserving the ability to verify correctness, we periodically

      "scale" the success counts by multiplying them by a scale factor

      between 0 and 1.0.

     

      Scaling is performed when either usage or construction attempt counts

      exceed a parametrized value.

     

      To avoid error due to scaling during circuit construction and use,

      currently open circuits are subtracted from the usage counts before

      scaling, and added back after scaling.

     

    7.4. Parametrization

     

       The following consensus parameters tune various aspects of the

       defense.

     

         pb_mincircs

           Default: 150

           Min: 5

           Effect: This is the minimum number of circuits that must complete

                   at least 2 hops before we begin evaluating construction rates.

     

     

         pb_noticepct

           Default: 70

           Min: 0

           Max: 100

           Effect: If the circuit success rate falls below this percentage,

                   we emit a notice log message.

     

         pb_warnpct

           Default: 50

           Min: 0

           Max: 100

           Effect: If the circuit success rate falls below this percentage,

                   we emit a warn log message.

     

         pb_extremepct

           Default: 30

           Min: 0

           Max: 100

           Effect: If the circuit success rate falls below this percentage,

                   we emit a more alarmist warning log message. If

                   pb_dropguard is set to 1, we also disable the use of the

                   guard.

     

         pb_dropguards

           Default: 0

           Min: 0

           Max: 1

           Effect: If the circuit success rate falls below pb_extremepct,

                   when pb_dropguard is set to 1, we disable use of that

                   guard.

     

         pb_scalecircs

           Default: 300

           Min: 10

           Effect: After this many circuits have completed at least two hops,

                   Tor performs the scaling described in Section 7.3.

     

         pb_multfactor and pb_scalefactor

           Default: 1/2

           Min: 0.0

           Max: 1.0

           Effect: The double-precision result obtained from

                   pb_multfactor/pb_scalefactor is multiplied by our current

                   counts to scale them.

     

         pb_minuse

           Default: 20

           Min: 3

           Effect: This is the minimum number of circuits that we must attempt to

                   use before we begin evaluating construction rates.

     

         pb_noticeusepct

           Default: 80

           Min: 3

           Effect: If the circuit usage success rate falls below this percentage,

                   we emit a notice log message.

     

         pb_extremeusepct

           Default: 60

           Min: 3

           Effect: If the circuit usage success rate falls below this percentage,

                   we emit a warning log message. We also disable the use of the

                   guard if pb_dropguards is set.

     

         pb_scaleuse

           Default: 100

           Min: 10

           Effect: After we have attempted to use this many circuits,

                   Tor performs the scaling described in Section 7.3.

     

    7.5. Known barriers to enforcement

     

      Due to intermittent CPU overload at relays, the normal rate of

      successful circuit completion is highly variable. The Guard-dropping

      version of the defense is unlikely to be deployed until the ntor

      circuit handshake is enabled, or the nature of CPU overload induced

      failure is better understood.

     

     

     

    X. Old notes

     

    X.1. Do we actually do this?

     

    How to deal with network down.

      - While all helpers are down/unreachable and there are no established

        or on-the-way testing circuits, launch a testing circuit. (Do this

        periodically in the same way we try to establish normal circuits

        when things are working normally.)

        (Testing circuits are a special type of circuit, that streams won't

        attach to by accident.)

      - When a testing circuit succeeds, mark all helpers up and hold

        the testing circuit open.

      - If a connection to a helper succeeds, close all testing circuits.

        Else mark that helper down and try another.

      - If the last helper is marked down and we already have a testing

        circuit established, then add the first hop of that testing circuit

        to the end of our helper node list, close that testing circuit,

        and go back to square one. (Actually, rather than closing the

        testing circuit, can we get away with converting it to a normal

        circuit and beginning to use it immediately?)

     

      [Do we actually do any of the above?  If so, let's spec it.  If not, let's

      remove it. -NM]

     

    X.2. A thing we could do to deal with reachability.

     

    And as a bonus, it leads to an answer to Nick's attack ("If I pick

    my helper nodes all on 18.0.0.0:*, then I move, you'll know where I

    bootstrapped") -- the answer is to pick your original three helper nodes

    without regard for reachability. Then the above algorithm will add some

    more that are reachable for you, and if you move somewhere, it's more

    likely (though not certain) that some of the originals will become useful.

    Is that smart or just complex?

     

    X.3. Some stuff that worries me about entry guards. 2006 Jun, Nickm.

     

      It is unlikely for two users to have the same set of entry guards.

      Observing a user is sufficient to learn its entry guards.  So, as we move

      around, entry guards make us linkable.  If we want to change guards when

      our location (IP? subnet?) changes, we have two bad options.  We could

        - Drop the old guards.  But if we go back to our old location,

          we'll not use our old guards.  For a laptop that sometimes gets used

          from work and sometimes from home, this is pretty fatal.

        - Remember the old guards as associated with the old location, and use

          them again if we ever go back to the old location.  This would be

          nasty, since it would force us to record where we've been.

     

      [Do we do any of this now? If not, this should move into 099-misc or

      098-todo. -NM]

  • 相关阅读:
    安装VMWare tools 及安装后/mnt中有hgfs但没共享文件的解决办法
    linux挂载命令
    RHEL7/CentOS7 Network Service开机无法启动的解决方法
    linux消息队列编程实例
    Linux进程间通信——使用消息队列
    消息队列函数
    ipcs查看消息队列命令
    linux批量删除
    HTTP 请求消息头部实例:
    drf 序列化组件
  • 原文地址:https://www.cnblogs.com/masonmei/p/11616319.html
Copyright © 2011-2022 走看看