2015-05-30 11:05:37 +00:00
|
|
|
|
.\" Man page generated from reStructuredText.
|
|
|
|
|
.
|
2020-02-19 06:45:29 +00:00
|
|
|
|
.TH "SYNCTHING-DEVICE-IDS" "7" "Feb 13, 2020" "v1" "Syncthing"
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.SH NAME
|
|
|
|
|
syncthing-device-ids \- Understanding Device IDs
|
|
|
|
|
.
|
|
|
|
|
.nr rst2man-indent-level 0
|
|
|
|
|
.
|
|
|
|
|
.de1 rstReportMargin
|
|
|
|
|
\\$1 \\n[an-margin]
|
|
|
|
|
level \\n[rst2man-indent-level]
|
|
|
|
|
level margin: \\n[rst2man-indent\\n[rst2man-indent-level]]
|
|
|
|
|
-
|
|
|
|
|
\\n[rst2man-indent0]
|
|
|
|
|
\\n[rst2man-indent1]
|
|
|
|
|
\\n[rst2man-indent2]
|
|
|
|
|
..
|
|
|
|
|
.de1 INDENT
|
|
|
|
|
.\" .rstReportMargin pre:
|
|
|
|
|
. RS \\$1
|
|
|
|
|
. nr rst2man-indent\\n[rst2man-indent-level] \\n[an-margin]
|
|
|
|
|
. nr rst2man-indent-level +1
|
|
|
|
|
.\" .rstReportMargin post:
|
|
|
|
|
..
|
|
|
|
|
.de UNINDENT
|
|
|
|
|
. RE
|
|
|
|
|
.\" indent \\n[an-margin]
|
|
|
|
|
.\" old: \\n[rst2man-indent\\n[rst2man-indent-level]]
|
|
|
|
|
.nr rst2man-indent-level -1
|
|
|
|
|
.\" new: \\n[rst2man-indent\\n[rst2man-indent-level]]
|
|
|
|
|
.in \\n[rst2man-indent\\n[rst2man-indent-level]]u
|
|
|
|
|
..
|
|
|
|
|
.sp
|
|
|
|
|
Every device is identified by a device ID. The device ID is used for address
|
2019-03-27 06:45:25 +00:00
|
|
|
|
resolution, authentication and authorization. The term “device ID” could
|
|
|
|
|
interchangeably have been “key ID” since the device ID is a direct property of
|
2015-05-30 11:05:37 +00:00
|
|
|
|
the public key in use.
|
|
|
|
|
.SH KEYS
|
|
|
|
|
.sp
|
|
|
|
|
To understand device IDs we need to look at the underlying mechanisms. At first
|
2015-09-27 20:31:19 +00:00
|
|
|
|
startup, Syncthing will create a public/private keypair.
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.sp
|
2018-02-28 06:45:16 +00:00
|
|
|
|
Currently this is a 384 bit ECDSA key (3072 bit RSA prior to v0.12.5,
|
|
|
|
|
which is what is used as an example in this article). The keys are saved in
|
|
|
|
|
the form of the private key (\fBkey.pem\fP) and a self signed certificate
|
2019-03-27 06:45:25 +00:00
|
|
|
|
(\fBcert.pem\fP). The self signing part doesn’t actually add any security or
|
2018-02-28 06:45:16 +00:00
|
|
|
|
functionality as far as Syncthing is concerned but it enables the use of the
|
|
|
|
|
keys in a standard TLS exchange.
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.sp
|
|
|
|
|
The typical certificate will look something like this, inspected with
|
|
|
|
|
\fBopenssl x509\fP:
|
|
|
|
|
.INDENT 0.0
|
|
|
|
|
.INDENT 3.5
|
|
|
|
|
.sp
|
|
|
|
|
.nf
|
|
|
|
|
.ft C
|
|
|
|
|
Certificate:
|
|
|
|
|
Data:
|
|
|
|
|
Version: 3 (0x2)
|
|
|
|
|
Serial Number: 0 (0x0)
|
|
|
|
|
Signature Algorithm: sha1WithRSAEncryption
|
|
|
|
|
Issuer: CN=syncthing
|
|
|
|
|
Validity
|
|
|
|
|
Not Before: Mar 30 21:10:52 2014 GMT
|
|
|
|
|
Not After : Dec 31 23:59:59 2049 GMT
|
|
|
|
|
Subject: CN=syncthing
|
|
|
|
|
Subject Public Key Info:
|
|
|
|
|
Public Key Algorithm: rsaEncryption
|
|
|
|
|
RSA Public Key: (3072 bit)
|
|
|
|
|
Modulus (3072 bit):
|
|
|
|
|
00:da:83:8a:c0:95:af:0a:42:af:43:74:65:29:f2:
|
|
|
|
|
30:e3:b9:12:d2:6b:70:93:da:0b:7b:8a:1e:e5:79:
|
|
|
|
|
...
|
|
|
|
|
99:09:4c:a9:7b:ba:4a:6a:8b:3b:e6:e7:c7:2c:00:
|
|
|
|
|
90:aa:bc:ad:94:e7:80:95:d2:1b
|
|
|
|
|
Exponent: 65537 (0x10001)
|
|
|
|
|
X509v3 extensions:
|
|
|
|
|
X509v3 Key Usage: critical
|
|
|
|
|
Digital Signature, Key Encipherment
|
|
|
|
|
X509v3 Extended Key Usage:
|
|
|
|
|
TLS Web Server Authentication, TLS Web Client Authentication
|
|
|
|
|
X509v3 Basic Constraints: critical
|
|
|
|
|
CA:FALSE
|
|
|
|
|
Signature Algorithm: sha1WithRSAEncryption
|
|
|
|
|
68:72:43:8b:83:61:09:68:f0:ef:f0:43:b7:30:a6:73:1e:a8:
|
|
|
|
|
d9:24:6c:2d:b4:bc:c9:e8:3e:0b:1e:3c:cc:7a:b2:c8:f1:1d:
|
|
|
|
|
...
|
|
|
|
|
88:7e:e2:61:aa:4c:02:e3:64:b0:da:70:3a:cd:1c:3d:86:db:
|
|
|
|
|
df:54:b9:4e:be:1b
|
|
|
|
|
.ft P
|
|
|
|
|
.fi
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.sp
|
|
|
|
|
We can see here that the certificate is little more than a container for the
|
|
|
|
|
public key; the serial number is zero and the Issuer and Subject are both
|
2019-03-27 06:45:25 +00:00
|
|
|
|
“syncthing” where a qualified name might otherwise be expected.
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.sp
|
|
|
|
|
An advanced user could replace the \fBkey.pem\fP and \fBcert.pem\fP files with a
|
|
|
|
|
keypair generated directly by the \fBopenssl\fP utility or other mechanism.
|
|
|
|
|
.SH DEVICE IDS
|
|
|
|
|
.sp
|
|
|
|
|
To form a device ID the SHA\-256 hash of the certificate data in DER form is
|
|
|
|
|
calculated. This means the hash covers all information under the
|
|
|
|
|
\fBCertificate:\fP section above.
|
|
|
|
|
.sp
|
2015-09-27 20:31:19 +00:00
|
|
|
|
The hashing results in a 256 bit hash which we encode using base32. Base32
|
|
|
|
|
encodes five bits per character so we need 256 / 5 = 51.2 characters to encode
|
2015-05-30 11:05:37 +00:00
|
|
|
|
the device ID. This becomes 52 characters in practice, but 52 characters of
|
2015-09-27 20:31:19 +00:00
|
|
|
|
base32 would decode to 260 bits which is not a whole number of bytes. The
|
2015-05-30 11:05:37 +00:00
|
|
|
|
base32 encoding adds padding to 280 bits (the next multiple of both 5 and 8
|
2015-06-07 10:57:26 +00:00
|
|
|
|
bits) so the resulting ID looks something like:
|
|
|
|
|
.INDENT 0.0
|
|
|
|
|
.INDENT 3.5
|
|
|
|
|
.sp
|
|
|
|
|
.nf
|
|
|
|
|
.ft C
|
|
|
|
|
MFZWI3DBONSGYYLTMRWGC43ENRQXGZDMMFZWI3DBONSGYYLTMRWA====
|
|
|
|
|
.ft P
|
|
|
|
|
.fi
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.UNINDENT
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.sp
|
2015-09-27 20:31:19 +00:00
|
|
|
|
The padding (\fB====\fP) is stripped away, the device ID split into four
|
2015-05-30 11:05:37 +00:00
|
|
|
|
groups, and \fI\%check
|
2016-12-26 13:23:55 +00:00
|
|
|
|
digits\fP <\fBhttps://forum.syncthing.net/t/v0-9-0-new-node-id-format/478\fP>
|
2015-05-30 11:05:37 +00:00
|
|
|
|
are added for each group. For presentation purposes the device ID is
|
|
|
|
|
grouped with dashes, resulting in the final value:
|
2015-06-07 10:57:26 +00:00
|
|
|
|
.INDENT 0.0
|
|
|
|
|
.INDENT 3.5
|
|
|
|
|
.sp
|
|
|
|
|
.nf
|
|
|
|
|
.ft C
|
2015-06-14 11:52:00 +00:00
|
|
|
|
MFZWI3D\-BONSGYC\-YLTMRWG\-C43ENR5\-QXGZDMM\-FZWI3DP\-BONSGYY\-LTMRWAD
|
2015-06-07 10:57:26 +00:00
|
|
|
|
.ft P
|
|
|
|
|
.fi
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.UNINDENT
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.SS Connection Establishment
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
Now we know what device IDs are, here’s how they are used in Syncthing. When
|
2015-09-27 20:31:19 +00:00
|
|
|
|
you add a device ID to the configuration, Syncthing will attempt to
|
2015-05-30 11:05:37 +00:00
|
|
|
|
connect to that device. The first thing we need to do is figure out the IP and
|
2015-09-27 20:31:19 +00:00
|
|
|
|
port to connect to. There are three possibilities here:
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.INDENT 0.0
|
|
|
|
|
.IP \(bu 2
|
|
|
|
|
The IP and port can be set statically in the configuration. The IP
|
2015-09-27 20:31:19 +00:00
|
|
|
|
can equally well be a host name, so if you have a static IP or a
|
2015-05-30 11:05:37 +00:00
|
|
|
|
dynamic DNS setup this might be a good option.
|
|
|
|
|
.IP \(bu 2
|
|
|
|
|
Using local discovery, if enabled. Every Syncthing instance on a LAN
|
|
|
|
|
periodically broadcasts information about itself (device ID, address,
|
2019-03-27 06:45:25 +00:00
|
|
|
|
port number). If we’ve seen one of these broadcasts for a given
|
|
|
|
|
device ID that’s where we try to connect.
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.IP \(bu 2
|
|
|
|
|
Using global discovery, if enabled. Every Syncthing instance
|
|
|
|
|
announces itself to the global discovery service (device ID and
|
|
|
|
|
external port number \- the internal address is not announced to the
|
2019-03-27 06:45:25 +00:00
|
|
|
|
global server). If we don’t have a static address and haven’t seen
|
2015-05-30 11:05:37 +00:00
|
|
|
|
any local announcements the global discovery server will be queried
|
|
|
|
|
for an address.
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.sp
|
2015-09-27 20:31:19 +00:00
|
|
|
|
Once we have an address and port a TCP connection is established and a TLS
|
2015-05-30 11:05:37 +00:00
|
|
|
|
handshake performed. As part of the handshake both devices present their
|
|
|
|
|
certificates. Once the handshake has completed and the peer certificate is
|
2015-09-27 20:31:19 +00:00
|
|
|
|
known, the following steps are performed:
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.INDENT 0.0
|
|
|
|
|
.IP 1. 3
|
2015-09-27 20:31:19 +00:00
|
|
|
|
Calculate the remote device ID by processing the received certificate as above.
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.IP 2. 3
|
|
|
|
|
Weed out a few possible misconfigurations \- i.e. if the device ID is
|
|
|
|
|
that of the local device or of a device we already have an active
|
|
|
|
|
connection to. Drop the connection in these cases.
|
|
|
|
|
.IP 3. 3
|
|
|
|
|
Verify the remote device ID against the configuration. If it is not a
|
|
|
|
|
device ID we are expecting to talk to, drop the connection.
|
|
|
|
|
.IP 4. 3
|
|
|
|
|
Verify the certificate \fBCommonName\fP against the configuration. By
|
|
|
|
|
default, we expect it to be \fBsyncthing\fP, but when using custom
|
|
|
|
|
certificates this can be changed.
|
|
|
|
|
.IP 5. 3
|
|
|
|
|
If everything checks out so far, accept the connection.
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.SH AN ASIDE ABOUT COLLISIONS
|
|
|
|
|
.sp
|
|
|
|
|
The SHA\-256 hash is cryptographically collision resistant. This means
|
|
|
|
|
that there is no way that we know of to create two different messages
|
|
|
|
|
with the same hash.
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
You can argue that of course there are collisions \- there’s an infinite
|
2015-09-27 20:31:19 +00:00
|
|
|
|
amount of inputs and a finite amount of outputs \- so by definition there
|
2015-05-30 11:05:37 +00:00
|
|
|
|
are infinitely many messages that result in the same hash.
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
I’m going to quote \fI\%stack
|
2017-02-05 12:54:49 +00:00
|
|
|
|
overflow\fP <\fBhttps://stackoverflow.com/questions/4014090/is-it-safe-to-ignore-the-possibility-of-sha-collisions-in-practice\fP>
|
2015-05-30 11:05:37 +00:00
|
|
|
|
here:
|
|
|
|
|
.INDENT 0.0
|
|
|
|
|
.INDENT 3.5
|
|
|
|
|
The usual answer goes thus: what is the probability that a rogue
|
|
|
|
|
asteroid crashes on Earth within the next second, obliterating
|
|
|
|
|
civilization\-as\-we\- know\-it, and killing off a few billion people ?
|
|
|
|
|
It can be argued that any unlucky event with a probability lower
|
|
|
|
|
than that is not actually very important.
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
If we have a “perfect” hash function with output size n, and we have
|
2015-05-30 11:05:37 +00:00
|
|
|
|
p messages to hash (individual message length is not important),
|
|
|
|
|
then probability of collision is about p2/2n+1 (this is an
|
2019-03-27 06:45:25 +00:00
|
|
|
|
approximation which is valid for “small” p, i.e. substantially
|
2015-05-30 11:05:37 +00:00
|
|
|
|
smaller than 2n/2). For instance, with SHA\-256 (n=256) and one
|
|
|
|
|
billion messages (p=10^9) then the probability is about 4.3*10^\-60.
|
|
|
|
|
.sp
|
|
|
|
|
A mass\-murderer space rock happens about once every 30 million years
|
|
|
|
|
on average. This leads to a probability of such an event occurring
|
2019-03-27 06:45:25 +00:00
|
|
|
|
in the next second to about 10^\-15. That’s 45 orders of magnitude
|
2015-05-30 11:05:37 +00:00
|
|
|
|
more probable than the SHA\-256 collision. Briefly stated, if you
|
|
|
|
|
find SHA\-256 collisions scary then your priorities are wrong.
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
It’s also worth noting that the property of SHA\-256 that we are using is not
|
2015-09-27 20:31:19 +00:00
|
|
|
|
simply collision resistance but resistance to a preimage attack, i.e. even if
|
2019-03-27 06:45:25 +00:00
|
|
|
|
you can find two messages that result in a hash collision that doesn’t help you
|
2015-05-30 11:05:37 +00:00
|
|
|
|
attack Syncthing (or TLS in general). You need to create a message that hashes
|
2019-03-27 06:45:25 +00:00
|
|
|
|
to exactly the hash that my certificate already has or you won’t get in.
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
Note also that it’s not good enough to find a random blob of bits that happen to
|
2015-09-27 20:31:19 +00:00
|
|
|
|
have the same hash as my certificate. You need to create a valid DER\-encoded,
|
2015-05-30 11:05:37 +00:00
|
|
|
|
signed certificate that has the same hash as mine. The difficulty of this is
|
|
|
|
|
staggeringly far beyond the already staggering difficulty of finding a SHA\-256
|
|
|
|
|
collision.
|
|
|
|
|
.SH PROBLEMS AND VULNERABILITIES
|
|
|
|
|
.sp
|
|
|
|
|
As far as I know, these are the issues or potential issues with the
|
|
|
|
|
above mechanism.
|
|
|
|
|
.SS Discovery Spoofing
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
Currently, the local discovery mechanism isn’t protected by crypto. This
|
2018-02-07 06:45:17 +00:00
|
|
|
|
means that any device can in theory announce itself for any device ID and
|
|
|
|
|
potentially receive connections for that device from the local network.
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.SS Long Device IDs are Painful
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
It’s a mouthful to read over the phone, annoying to type into an SMS or even
|
2015-05-30 11:05:37 +00:00
|
|
|
|
into a computer. And it needs to be done twice, once for each side.
|
|
|
|
|
.sp
|
2019-03-27 06:45:25 +00:00
|
|
|
|
This isn’t a vulnerability as such, but a user experience problem. There are
|
2015-05-30 11:05:37 +00:00
|
|
|
|
various possible solutions:
|
|
|
|
|
.INDENT 0.0
|
|
|
|
|
.IP \(bu 2
|
2019-03-27 06:45:25 +00:00
|
|
|
|
Use shorter device IDs with verification based on the full ID (“You
|
2015-05-30 11:05:37 +00:00
|
|
|
|
entered MFZWI3; I found and connected to a device with the ID
|
|
|
|
|
MFZWI3\-DBONSG\-YYLTMR\-WGC43E\-NRQXGZ\-DMMFZW\-I3DBON\-SGYYLT\-MRWA, please
|
2019-03-27 06:45:25 +00:00
|
|
|
|
confirm that this is correct”).
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.IP \(bu 2
|
|
|
|
|
Use shorter device IDs with an out of band authentication, a la
|
|
|
|
|
Bluetooth pairing. You enter a one time PIN into Syncthing and give
|
|
|
|
|
that PIN plus a short device ID to another user. On initial connect,
|
|
|
|
|
both sides verify that the other knows the correct PIN before
|
|
|
|
|
accepting the connection.
|
|
|
|
|
.UNINDENT
|
|
|
|
|
.SH AUTHOR
|
|
|
|
|
The Syncthing Authors
|
|
|
|
|
.SH COPYRIGHT
|
2019-02-13 06:45:23 +00:00
|
|
|
|
2014-2019, The Syncthing Authors
|
2015-05-30 11:05:37 +00:00
|
|
|
|
.\" Generated by docutils manpage writer.
|
|
|
|
|
.
|