wireproto: add request IDs to frames
One of my primary goals with the new wire protocol is to make
operations faster and enable both client and server-side
operations to scale to multiple CPU cores.
One of the ways we make server interactions faster is by reducing
the number of round trips to that server.
With the existing wire protocol, the "batch" command facilitates
executing multiple commands from a single request payload. The way
it works is the requests for multiple commands are serialized. The
server executes those commands sequentially then serializes all
their results. As an optimization for reducing round trips, this
is very effective. The technical implementation, however, is pretty
bad and suffers from a number of deficiencies. For example, it
creates a new place where authorization to run a command must be
checked. (The lack of this checking in older Mercurial releases
was CVE-2018-1000132.)
The principles behind the "batch" command are sound. However, the
execution is not. Therefore, I want to ditch "batch" in the
new wire protocol and have protocol level support for issuing
multiple requests in a single round trip.
This commit introduces support in the frame-based wire protocol to
facilitate this. We do this by adding a "request ID" to each frame.
If a server sees frames associated with different "request IDs," it
handles them as separate requests. All of this happening possibly
as part of the same message from client to server (the same request
body in the case of HTTP).
We /could/ model the exchange the way pipelined HTTP requests do,
where the server processes requests in order they are issued and
received. But this artifically constrains scalability. A better
model is to allow multi-requests to be executed concurrently and
for responses to be sent and handled concurrently. So the
specification explicitly allows this. There is some work to be done
around specifying dependencies between multi-requests. We take
the easy road for now and punt on this problem, declaring that
if order is important, clients must not issue the request until
responses to dependent requests have been received.
This commit focuses on the boilerplate of implementing the request
ID. The server reactor still can't manage multiple, in-flight
request IDs. This will be addressed in a subsequent commit.
Because the wire semantics have changed, we bump the version of the
media type.
Differential Revision: https://phab.mercurial-scm.org/D2869
from __future__ import absolute_import, print_function
import unittest
from mercurial import (
util,
wireprotoframing as framing,
)
ffs = framing.makeframefromhumanstring
def makereactor(deferoutput=False):
return framing.serverreactor(deferoutput=deferoutput)
def sendframes(reactor, gen):
"""Send a generator of frame bytearray to a reactor.
Emits a generator of results from ``onframerecv()`` calls.
"""
for frame in gen:
rid, frametype, frameflags, framelength = framing.parseheader(frame)
payload = frame[framing.FRAME_HEADER_SIZE:]
assert len(payload) == framelength
yield reactor.onframerecv(rid, frametype, frameflags, payload)
def sendcommandframes(reactor, rid, cmd, args, datafh=None):
"""Generate frames to run a command and send them to a reactor."""
return sendframes(reactor,
framing.createcommandframes(rid, cmd, args, datafh))
class FrameTests(unittest.TestCase):
def testdataexactframesize(self):
data = util.bytesio(b'x' * framing.DEFAULT_MAX_FRAME_SIZE)
frames = list(framing.createcommandframes(1, b'command', {}, data))
self.assertEqual(frames, [
ffs(b'1 command-name have-data command'),
ffs(b'1 command-data continuation %s' % data.getvalue()),
ffs(b'1 command-data eos ')
])
def testdatamultipleframes(self):
data = util.bytesio(b'x' * (framing.DEFAULT_MAX_FRAME_SIZE + 1))
frames = list(framing.createcommandframes(1, b'command', {}, data))
self.assertEqual(frames, [
ffs(b'1 command-name have-data command'),
ffs(b'1 command-data continuation %s' % (
b'x' * framing.DEFAULT_MAX_FRAME_SIZE)),
ffs(b'1 command-data eos x'),
])
def testargsanddata(self):
data = util.bytesio(b'x' * 100)
frames = list(framing.createcommandframes(1, b'command', {
b'key1': b'key1value',
b'key2': b'key2value',
b'key3': b'key3value',
}, data))
self.assertEqual(frames, [
ffs(b'1 command-name have-args|have-data command'),
ffs(br'1 command-argument 0 \x04\x00\x09\x00key1key1value'),
ffs(br'1 command-argument 0 \x04\x00\x09\x00key2key2value'),
ffs(br'1 command-argument eoa \x04\x00\x09\x00key3key3value'),
ffs(b'1 command-data eos %s' % data.getvalue()),
])
class ServerReactorTests(unittest.TestCase):
def _sendsingleframe(self, reactor, s):
results = list(sendframes(reactor, [ffs(s)]))
self.assertEqual(len(results), 1)
return results[0]
def assertaction(self, res, expected):
self.assertIsInstance(res, tuple)
self.assertEqual(len(res), 2)
self.assertIsInstance(res[1], dict)
self.assertEqual(res[0], expected)
def assertframesequal(self, frames, framestrings):
expected = [ffs(s) for s in framestrings]
self.assertEqual(list(frames), expected)
def test1framecommand(self):
"""Receiving a command in a single frame yields request to run it."""
reactor = makereactor()
results = list(sendcommandframes(reactor, 1, b'mycommand', {}))
self.assertEqual(len(results), 1)
self.assertaction(results[0], 'runcommand')
self.assertEqual(results[0][1], {
'requestid': 1,
'command': b'mycommand',
'args': {},
'data': None,
})
result = reactor.oninputeof()
self.assertaction(result, 'noop')
def test1argument(self):
reactor = makereactor()
results = list(sendcommandframes(reactor, 41, b'mycommand',
{b'foo': b'bar'}))
self.assertEqual(len(results), 2)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'runcommand')
self.assertEqual(results[1][1], {
'requestid': 41,
'command': b'mycommand',
'args': {b'foo': b'bar'},
'data': None,
})
def testmultiarguments(self):
reactor = makereactor()
results = list(sendcommandframes(reactor, 1, b'mycommand',
{b'foo': b'bar', b'biz': b'baz'}))
self.assertEqual(len(results), 3)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'wantframe')
self.assertaction(results[2], 'runcommand')
self.assertEqual(results[2][1], {
'requestid': 1,
'command': b'mycommand',
'args': {b'foo': b'bar', b'biz': b'baz'},
'data': None,
})
def testsimplecommanddata(self):
reactor = makereactor()
results = list(sendcommandframes(reactor, 1, b'mycommand', {},
util.bytesio(b'data!')))
self.assertEqual(len(results), 2)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'runcommand')
self.assertEqual(results[1][1], {
'requestid': 1,
'command': b'mycommand',
'args': {},
'data': b'data!',
})
def testmultipledataframes(self):
frames = [
ffs(b'1 command-name have-data mycommand'),
ffs(b'1 command-data continuation data1'),
ffs(b'1 command-data continuation data2'),
ffs(b'1 command-data eos data3'),
]
reactor = makereactor()
results = list(sendframes(reactor, frames))
self.assertEqual(len(results), 4)
for i in range(3):
self.assertaction(results[i], 'wantframe')
self.assertaction(results[3], 'runcommand')
self.assertEqual(results[3][1], {
'requestid': 1,
'command': b'mycommand',
'args': {},
'data': b'data1data2data3',
})
def testargumentanddata(self):
frames = [
ffs(b'1 command-name have-args|have-data command'),
ffs(br'1 command-argument 0 \x03\x00\x03\x00keyval'),
ffs(br'1 command-argument eoa \x03\x00\x03\x00foobar'),
ffs(b'1 command-data continuation value1'),
ffs(b'1 command-data eos value2'),
]
reactor = makereactor()
results = list(sendframes(reactor, frames))
self.assertaction(results[-1], 'runcommand')
self.assertEqual(results[-1][1], {
'requestid': 1,
'command': b'command',
'args': {
b'key': b'val',
b'foo': b'bar',
},
'data': b'value1value2',
})
def testunexpectedcommandargument(self):
"""Command argument frame when not running a command is an error."""
result = self._sendsingleframe(makereactor(),
b'1 command-argument 0 ignored')
self.assertaction(result, 'error')
self.assertEqual(result[1], {
'message': b'expected command frame; got 2',
})
def testunexpectedcommanddata(self):
"""Command argument frame when not running a command is an error."""
result = self._sendsingleframe(makereactor(),
b'1 command-data 0 ignored')
self.assertaction(result, 'error')
self.assertEqual(result[1], {
'message': b'expected command frame; got 3',
})
def testmissingcommandframeflags(self):
"""Command name frame must have flags set."""
result = self._sendsingleframe(makereactor(),
b'1 command-name 0 command')
self.assertaction(result, 'error')
self.assertEqual(result[1], {
'message': b'missing frame flags on command frame',
})
def testmissingargumentframe(self):
frames = [
ffs(b'1 command-name have-args command'),
ffs(b'1 command-name 0 ignored'),
]
results = list(sendframes(makereactor(), frames))
self.assertEqual(len(results), 2)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'error')
self.assertEqual(results[1][1], {
'message': b'expected command argument frame; got 1',
})
def testincompleteargumentname(self):
"""Argument frame with incomplete name."""
frames = [
ffs(b'1 command-name have-args command1'),
ffs(br'1 command-argument eoa \x04\x00\xde\xadfoo'),
]
results = list(sendframes(makereactor(), frames))
self.assertEqual(len(results), 2)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'error')
self.assertEqual(results[1][1], {
'message': b'malformed argument frame: partial argument name',
})
def testincompleteargumentvalue(self):
"""Argument frame with incomplete value."""
frames = [
ffs(b'1 command-name have-args command'),
ffs(br'1 command-argument eoa \x03\x00\xaa\xaafoopartialvalue'),
]
results = list(sendframes(makereactor(), frames))
self.assertEqual(len(results), 2)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'error')
self.assertEqual(results[1][1], {
'message': b'malformed argument frame: partial argument value',
})
def testmissingcommanddataframe(self):
frames = [
ffs(b'1 command-name have-data command1'),
ffs(b'1 command-name eos command2'),
]
results = list(sendframes(makereactor(), frames))
self.assertEqual(len(results), 2)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'error')
self.assertEqual(results[1][1], {
'message': b'expected command data frame; got 1',
})
def testmissingcommanddataframeflags(self):
frames = [
ffs(b'1 command-name have-data command1'),
ffs(b'1 command-data 0 data'),
]
results = list(sendframes(makereactor(), frames))
self.assertEqual(len(results), 2)
self.assertaction(results[0], 'wantframe')
self.assertaction(results[1], 'error')
self.assertEqual(results[1][1], {
'message': b'command data frame without flags',
})
def testsimpleresponse(self):
"""Bytes response to command sends result frames."""
reactor = makereactor()
list(sendcommandframes(reactor, 1, b'mycommand', {}))
result = reactor.onbytesresponseready(1, b'response')
self.assertaction(result, 'sendframes')
self.assertframesequal(result[1]['framegen'], [
b'1 bytes-response eos response',
])
def testmultiframeresponse(self):
"""Bytes response spanning multiple frames is handled."""
first = b'x' * framing.DEFAULT_MAX_FRAME_SIZE
second = b'y' * 100
reactor = makereactor()
list(sendcommandframes(reactor, 1, b'mycommand', {}))
result = reactor.onbytesresponseready(1, first + second)
self.assertaction(result, 'sendframes')
self.assertframesequal(result[1]['framegen'], [
b'1 bytes-response continuation %s' % first,
b'1 bytes-response eos %s' % second,
])
def testapplicationerror(self):
reactor = makereactor()
list(sendcommandframes(reactor, 1, b'mycommand', {}))
result = reactor.onapplicationerror(1, b'some message')
self.assertaction(result, 'sendframes')
self.assertframesequal(result[1]['framegen'], [
b'1 error-response application some message',
])
def test1commanddeferresponse(self):
"""Responses when in deferred output mode are delayed until EOF."""
reactor = makereactor(deferoutput=True)
results = list(sendcommandframes(reactor, 1, b'mycommand', {}))
self.assertEqual(len(results), 1)
self.assertaction(results[0], 'runcommand')
result = reactor.onbytesresponseready(1, b'response')
self.assertaction(result, 'noop')
result = reactor.oninputeof()
self.assertaction(result, 'sendframes')
self.assertframesequal(result[1]['framegen'], [
b'1 bytes-response eos response',
])
def testmultiplecommanddeferresponse(self):
reactor = makereactor(deferoutput=True)
list(sendcommandframes(reactor, 1, b'command1', {}))
list(sendcommandframes(reactor, 3, b'command2', {}))
result = reactor.onbytesresponseready(1, b'response1')
self.assertaction(result, 'noop')
result = reactor.onbytesresponseready(3, b'response2')
self.assertaction(result, 'noop')
result = reactor.oninputeof()
self.assertaction(result, 'sendframes')
self.assertframesequal(result[1]['framegen'], [
b'1 bytes-response eos response1',
b'3 bytes-response eos response2'
])
def testrequestidtracking(self):
reactor = makereactor(deferoutput=True)
list(sendcommandframes(reactor, 1, b'command1', {}))
list(sendcommandframes(reactor, 3, b'command2', {}))
list(sendcommandframes(reactor, 5, b'command3', {}))
# Register results for commands out of order.
reactor.onbytesresponseready(3, b'response3')
reactor.onbytesresponseready(1, b'response1')
reactor.onbytesresponseready(5, b'response5')
result = reactor.oninputeof()
self.assertaction(result, 'sendframes')
self.assertframesequal(result[1]['framegen'], [
b'3 bytes-response eos response3',
b'1 bytes-response eos response1',
b'5 bytes-response eos response5',
])
if __name__ == '__main__':
import silenttestrunner
silenttestrunner.main(__name__)