Mercurial > hg
comparison mercurial/sshpeer.py @ 35976:48a3a9283f09
sshpeer: initial definition and implementation of new SSH protocol
The existing SSH protocol has several design flaws. Future commits
will elaborate on these flaws as new features are introduced
to combat these flaws. For now, hopefully you can take me for my
word that a ground up rewrite of the SSH protocol is needed.
This commit lays the foundation for a new SSH protocol by defining
a mechanism to upgrade the SSH transport channel away from the
default (version 1) protocol to something modern (which we'll call
"version 2" for now).
This upgrade process is detailed in the internals documentation
for the wire protocol. The gist of it is the client sends a
request line preceding the "hello" command/line which basically
says "I'm requesting an upgrade: here's what I support." If the
server recognizes that line, it processes the upgrade request and
the transport channel is switched to use the new version of the
protocol. If not, it sends an empty response, which is how all
Mercurial SSH servers from the beginning of time reacted to unknown
commands. The upgrade request is effectively ignored and the client
continues to use the existing version of the protocol as if nothing
happened.
The new version of the SSH protocol is completely identical to
version 1 aside from the upgrade dance and the bytes that follow.
The immediate bytes that follow the protocol switch are defined to
be a length framed "capabilities: " line containing the remote's
advertised capabilities. In reality, this looks very similar to
what the "hello" response would look like. But it will evolve
quickly.
The methodology by which the protocol will evolve is important.
I'm not going to introduce the new protocol all at once. That would
likely lead to endless bike shedding and forward progress would
stall. Instead, I intend to tricle out new features and diversions
from the existing protocol in small, incremental changes.
To support the gradual evolution of the protocol, the on-the-wire
advertised protocol name contains an "exp" to denote "experimental"
and a 4 digit field to capture the sub-version of the protocol.
Whenever we make a BC change to the wire protocol, we can increment
this version and lock out all older clients because it will appear
as a completely different protocol version. This means we can incur
as many breaking changes as we want. We don't have to commit to
supporting any one feature or idea for a long period of time. We
can even evolve the handshake mechanism, because that is defined
as being an implementation detail of the negotiated protocol version!
Hopefully this lowers the barrier to accepting changes to the
protocol and for experimenting with "radical" ideas during its
development.
In core, sshpeer received most of the attention. We haven't even
implemented the server bits for the new protocol in core yet.
Instead, we add very primitive support to our test server, mainly
just to exercise the added code paths in sshpeer.
Differential Revision: https://phab.mercurial-scm.org/D2061
# no-check-commit because of required foo_bar naming
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Tue, 06 Feb 2018 11:08:36 -0800 |
parents | 556218e08e25 |
children | 625038cb4b1d |
comparison
equal
deleted
inserted
replaced
35975:40d94ea51402 | 35976:48a3a9283f09 |
---|---|
6 # GNU General Public License version 2 or any later version. | 6 # GNU General Public License version 2 or any later version. |
7 | 7 |
8 from __future__ import absolute_import | 8 from __future__ import absolute_import |
9 | 9 |
10 import re | 10 import re |
11 import uuid | |
11 | 12 |
12 from .i18n import _ | 13 from .i18n import _ |
13 from . import ( | 14 from . import ( |
14 error, | 15 error, |
15 pycompat, | 16 pycompat, |
16 util, | 17 util, |
17 wireproto, | 18 wireproto, |
19 wireprotoserver, | |
18 ) | 20 ) |
19 | 21 |
20 def _serverquote(s): | 22 def _serverquote(s): |
21 """quote a string for the remote shell ... which we assume is sh""" | 23 """quote a string for the remote shell ... which we assume is sh""" |
22 if not s: | 24 if not s: |
160 def badresponse(): | 162 def badresponse(): |
161 msg = _('no suitable response from remote hg') | 163 msg = _('no suitable response from remote hg') |
162 hint = ui.config('ui', 'ssherrorhint') | 164 hint = ui.config('ui', 'ssherrorhint') |
163 raise error.RepoError(msg, hint=hint) | 165 raise error.RepoError(msg, hint=hint) |
164 | 166 |
165 # The handshake consists of sending 2 wire protocol commands: | 167 # The handshake consists of sending wire protocol commands in reverse |
166 # ``hello`` and ``between``. | 168 # order of protocol implementation and then sniffing for a response |
167 # | 169 # to one of them. |
168 # The ``hello`` command (which was introduced in Mercurial 0.9.1) | 170 # |
169 # instructs the server to advertise its capabilities. | 171 # Those commands (from oldest to newest) are: |
170 # | 172 # |
171 # The ``between`` command (which has existed in all Mercurial servers | 173 # ``between`` |
172 # for as long as SSH support has existed), asks for the set of revisions | 174 # Asks for the set of revisions between a pair of revisions. Command |
173 # between a pair of revisions. | 175 # present in all Mercurial server implementations. |
176 # | |
177 # ``hello`` | |
178 # Instructs the server to advertise its capabilities. Introduced in | |
179 # Mercurial 0.9.1. | |
180 # | |
181 # ``upgrade`` | |
182 # Requests upgrade from default transport protocol version 1 to | |
183 # a newer version. Introduced in Mercurial 4.6 as an experimental | |
184 # feature. | |
174 # | 185 # |
175 # The ``between`` command is issued with a request for the null | 186 # The ``between`` command is issued with a request for the null |
176 # range. If the remote is a Mercurial server, this request will | 187 # range. If the remote is a Mercurial server, this request will |
177 # generate a specific response: ``1\n\n``. This represents the | 188 # generate a specific response: ``1\n\n``. This represents the |
178 # wire protocol encoded value for ``\n``. We look for ``1\n\n`` | 189 # wire protocol encoded value for ``\n``. We look for ``1\n\n`` |
184 # value. If the server doesn't support ``hello`` (which should be | 195 # value. If the server doesn't support ``hello`` (which should be |
185 # rare), that line will be ``0\n``. Otherwise, the value will contain | 196 # rare), that line will be ``0\n``. Otherwise, the value will contain |
186 # RFC 822 like lines. Of these, the ``capabilities:`` line contains | 197 # RFC 822 like lines. Of these, the ``capabilities:`` line contains |
187 # the capabilities of the server. | 198 # the capabilities of the server. |
188 # | 199 # |
200 # The ``upgrade`` command isn't really a command in the traditional | |
201 # sense of version 1 of the transport because it isn't using the | |
202 # proper mechanism for formatting insteads: instead, it just encodes | |
203 # arguments on the line, delimited by spaces. | |
204 # | |
205 # The ``upgrade`` line looks like ``upgrade <token> <capabilities>``. | |
206 # If the server doesn't support protocol upgrades, it will reply to | |
207 # this line with ``0\n``. Otherwise, it emits an | |
208 # ``upgraded <token> <protocol>`` line to both stdout and stderr. | |
209 # Content immediately following this line describes additional | |
210 # protocol and server state. | |
211 # | |
189 # In addition to the responses to our command requests, the server | 212 # In addition to the responses to our command requests, the server |
190 # may emit "banner" output on stdout. SSH servers are allowed to | 213 # may emit "banner" output on stdout. SSH servers are allowed to |
191 # print messages to stdout on login. Issuing commands on connection | 214 # print messages to stdout on login. Issuing commands on connection |
192 # allows us to flush this banner output from the server by scanning | 215 # allows us to flush this banner output from the server by scanning |
193 # for output to our well-known ``between`` command. Of course, if | 216 # for output to our well-known ``between`` command. Of course, if |
194 # the banner contains ``1\n\n``, this will throw off our detection. | 217 # the banner contains ``1\n\n``, this will throw off our detection. |
195 | 218 |
196 requestlog = ui.configbool('devel', 'debug.peer-request') | 219 requestlog = ui.configbool('devel', 'debug.peer-request') |
220 | |
221 # Generate a random token to help identify responses to version 2 | |
222 # upgrade request. | |
223 token = bytes(uuid.uuid4()) | |
224 upgradecaps = [ | |
225 ('proto', wireprotoserver.SSHV2), | |
226 ] | |
227 upgradecaps = util.urlreq.urlencode(upgradecaps) | |
197 | 228 |
198 try: | 229 try: |
199 pairsarg = '%s-%s' % ('0' * 40, '0' * 40) | 230 pairsarg = '%s-%s' % ('0' * 40, '0' * 40) |
200 handshake = [ | 231 handshake = [ |
201 'hello\n', | 232 'hello\n', |
202 'between\n', | 233 'between\n', |
203 'pairs %d\n' % len(pairsarg), | 234 'pairs %d\n' % len(pairsarg), |
204 pairsarg, | 235 pairsarg, |
205 ] | 236 ] |
206 | 237 |
238 # Request upgrade to version 2 if configured. | |
239 if ui.configbool('experimental', 'sshpeer.advertise-v2'): | |
240 ui.debug('sending upgrade request: %s %s\n' % (token, upgradecaps)) | |
241 handshake.insert(0, 'upgrade %s %s\n' % (token, upgradecaps)) | |
242 | |
207 if requestlog: | 243 if requestlog: |
208 ui.debug('devel-peer-request: hello\n') | 244 ui.debug('devel-peer-request: hello\n') |
209 ui.debug('sending hello command\n') | 245 ui.debug('sending hello command\n') |
210 if requestlog: | 246 if requestlog: |
211 ui.debug('devel-peer-request: between\n') | 247 ui.debug('devel-peer-request: between\n') |
215 stdin.write(''.join(handshake)) | 251 stdin.write(''.join(handshake)) |
216 stdin.flush() | 252 stdin.flush() |
217 except IOError: | 253 except IOError: |
218 badresponse() | 254 badresponse() |
219 | 255 |
256 # Assume version 1 of wire protocol by default. | |
257 protoname = wireprotoserver.SSHV1 | |
258 reupgraded = re.compile(b'^upgraded %s (.*)$' % re.escape(token)) | |
259 | |
220 lines = ['', 'dummy'] | 260 lines = ['', 'dummy'] |
221 max_noise = 500 | 261 max_noise = 500 |
222 while lines[-1] and max_noise: | 262 while lines[-1] and max_noise: |
223 try: | 263 try: |
224 l = stdout.readline() | 264 l = stdout.readline() |
225 _forwardoutput(ui, stderr) | 265 _forwardoutput(ui, stderr) |
266 | |
267 # Look for reply to protocol upgrade request. It has a token | |
268 # in it, so there should be no false positives. | |
269 m = reupgraded.match(l) | |
270 if m: | |
271 protoname = m.group(1) | |
272 ui.debug('protocol upgraded to %s\n' % protoname) | |
273 # If an upgrade was handled, the ``hello`` and ``between`` | |
274 # requests are ignored. The next output belongs to the | |
275 # protocol, so stop scanning lines. | |
276 break | |
277 | |
278 # Otherwise it could be a banner, ``0\n`` response if server | |
279 # doesn't support upgrade. | |
280 | |
226 if lines[-1] == '1\n' and l == '\n': | 281 if lines[-1] == '1\n' and l == '\n': |
227 break | 282 break |
228 if l: | 283 if l: |
229 ui.debug('remote: ', l) | 284 ui.debug('remote: ', l) |
230 lines.append(l) | 285 lines.append(l) |
233 badresponse() | 288 badresponse() |
234 else: | 289 else: |
235 badresponse() | 290 badresponse() |
236 | 291 |
237 caps = set() | 292 caps = set() |
238 for l in reversed(lines): | 293 |
239 # Look for response to ``hello`` command. Scan from the back so | 294 # For version 1, we should see a ``capabilities`` line in response to the |
240 # we don't misinterpret banner output as the command reply. | 295 # ``hello`` command. |
241 if l.startswith('capabilities:'): | 296 if protoname == wireprotoserver.SSHV1: |
242 caps.update(l[:-1].split(':')[1].split()) | 297 for l in reversed(lines): |
243 break | 298 # Look for response to ``hello`` command. Scan from the back so |
244 | 299 # we don't misinterpret banner output as the command reply. |
245 # Error if we couldn't find a response to ``hello``. This could | 300 if l.startswith('capabilities:'): |
246 # mean: | 301 caps.update(l[:-1].split(':')[1].split()) |
302 break | |
303 elif protoname == wireprotoserver.SSHV2: | |
304 # We see a line with number of bytes to follow and then a value | |
305 # looking like ``capabilities: *``. | |
306 line = stdout.readline() | |
307 try: | |
308 valuelen = int(line) | |
309 except ValueError: | |
310 badresponse() | |
311 | |
312 capsline = stdout.read(valuelen) | |
313 if not capsline.startswith('capabilities: '): | |
314 badresponse() | |
315 | |
316 caps.update(capsline.split(':')[1].split()) | |
317 # Trailing newline. | |
318 stdout.read(1) | |
319 | |
320 # Error if we couldn't find capabilities, this means: | |
247 # | 321 # |
248 # 1. Remote isn't a Mercurial server | 322 # 1. Remote isn't a Mercurial server |
249 # 2. Remote is a <0.9.1 Mercurial server | 323 # 2. Remote is a <0.9.1 Mercurial server |
250 # 3. Remote is a future Mercurial server that dropped ``hello`` | 324 # 3. Remote is a future Mercurial server that dropped ``hello`` |
251 # support. | 325 # and other attempted handshake mechanisms. |
252 if not caps: | 326 if not caps: |
253 badresponse() | 327 badresponse() |
254 | 328 |
255 return caps | 329 return caps |
256 | 330 |