platform_system_core/adb/test_adb.py

447 lines
18 KiB
Python
Raw Normal View History

#!/usr/bin/env python
#
# Copyright (C) 2015 The Android Open Source Project
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
"""Tests for the adb program itself.
This differs from things in test_device.py in that there is no API for these
things. Most of these tests involve specific error messages or the help text.
"""
from __future__ import print_function
import binascii
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
import contextlib
import os
import random
import select
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
import socket
import struct
import subprocess
import threading
import unittest
import adb
@contextlib.contextmanager
def fake_adbd(protocol=socket.AF_INET, port=0):
"""Creates a fake ADB daemon that just replies with a CNXN packet."""
serversock = socket.socket(protocol, socket.SOCK_STREAM)
serversock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
if protocol == socket.AF_INET:
serversock.bind(('127.0.0.1', port))
else:
serversock.bind(('::1', port))
serversock.listen(1)
# A pipe that is used to signal the thread that it should terminate.
readpipe, writepipe = os.pipe()
def _adb_packet(command, arg0, arg1, data):
bin_command = struct.unpack('I', command)[0]
buf = struct.pack('IIIIII', bin_command, arg0, arg1, len(data), 0,
bin_command ^ 0xffffffff)
buf += data
return buf
def _handle():
rlist = [readpipe, serversock]
cnxn_sent = {}
while True:
read_ready, _, _ = select.select(rlist, [], [])
for ready in read_ready:
if ready == readpipe:
# Closure pipe
os.close(ready)
serversock.shutdown(socket.SHUT_RDWR)
serversock.close()
return
elif ready == serversock:
# Server socket
conn, _ = ready.accept()
rlist.append(conn)
else:
# Client socket
data = ready.recv(1024)
if not data or data.startswith('OPEN'):
if ready in cnxn_sent:
del cnxn_sent[ready]
ready.shutdown(socket.SHUT_RDWR)
ready.close()
rlist.remove(ready)
continue
if ready in cnxn_sent:
continue
cnxn_sent[ready] = True
ready.sendall(_adb_packet('CNXN', 0x01000001, 1024 * 1024,
'device::ro.product.name=fakeadb'))
port = serversock.getsockname()[1]
server_thread = threading.Thread(target=_handle)
server_thread.start()
try:
yield port
finally:
os.close(writepipe)
server_thread.join()
@contextlib.contextmanager
def adb_connect(unittest, serial):
"""Context manager for an ADB connection.
This automatically disconnects when done with the connection.
"""
output = subprocess.check_output(['adb', 'connect', serial])
unittest.assertEqual(output.strip(), 'connected to {}'.format(serial))
try:
yield
finally:
# Perform best-effort disconnection. Discard the output.
subprocess.Popen(['adb', 'disconnect', serial],
stdout=subprocess.PIPE,
stderr=subprocess.PIPE).communicate()
@contextlib.contextmanager
def adb_server():
"""Context manager for an ADB server.
This creates an ADB server and returns the port it's listening on.
"""
port = 5038
# Kill any existing server on this non-default port.
subprocess.check_output(['adb', '-P', str(port), 'kill-server'],
stderr=subprocess.STDOUT)
read_pipe, write_pipe = os.pipe()
proc = subprocess.Popen(['adb', '-L', 'tcp:localhost:{}'.format(port),
'fork-server', 'server',
'--reply-fd', str(write_pipe)])
try:
os.close(write_pipe)
greeting = os.read(read_pipe, 1024)
assert greeting == 'OK\n', repr(greeting)
yield port
finally:
proc.terminate()
proc.wait()
class CommandlineTest(unittest.TestCase):
"""Tests for the ADB commandline."""
def test_help(self):
"""Make sure we get _something_ out of help."""
out = subprocess.check_output(
['adb', 'help'], stderr=subprocess.STDOUT)
self.assertGreater(len(out), 0)
def test_version(self):
"""Get a version number out of the output of adb."""
lines = subprocess.check_output(['adb', 'version']).splitlines()
version_line = lines[0]
self.assertRegexpMatches(
version_line, r'^Android Debug Bridge version \d+\.\d+\.\d+$')
if len(lines) == 2:
# Newer versions of ADB have a second line of output for the
# version that includes a specific revision (git SHA).
revision_line = lines[1]
self.assertRegexpMatches(
revision_line, r'^Revision [0-9a-f]{12}-android$')
def test_tcpip_error_messages(self):
"""Make sure 'adb tcpip' parsing is sane."""
proc = subprocess.Popen(['adb', 'tcpip'], stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
out, _ = proc.communicate()
self.assertEqual(1, proc.returncode)
self.assertIn('requires an argument', out)
proc = subprocess.Popen(['adb', 'tcpip', 'foo'], stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
out, _ = proc.communicate()
self.assertEqual(1, proc.returncode)
self.assertIn('invalid port', out)
class ServerTest(unittest.TestCase):
"""Tests for the ADB server."""
@staticmethod
def _read_pipe_and_set_event(pipe, event):
"""Reads a pipe until it is closed, then sets the event."""
pipe.read()
event.set()
def test_handle_inheritance(self):
"""Test that launch_server() does not inherit handles.
launch_server() should not let the adb server inherit
stdin/stdout/stderr handles, which can cause callers of adb.exe to hang.
This test also runs fine on unix even though the impetus is an issue
unique to Windows.
"""
# This test takes 5 seconds to run on Windows: if there is no adb server
# running on the the port used below, adb kill-server tries to make a
# TCP connection to a closed port and that takes 1 second on Windows;
# adb start-server does the same TCP connection which takes another
# second, and it waits 3 seconds after starting the server.
# Start adb client with redirected stdin/stdout/stderr to check if it
# passes those redirections to the adb server that it starts. To do
# this, run an instance of the adb server on a non-default port so we
# don't conflict with a pre-existing adb server that may already be
# setup with adb TCP/emulator connections. If there is a pre-existing
# adb server, this also tests whether multiple instances of the adb
# server conflict on adb.log.
port = 5038
# Kill any existing server on this non-default port.
subprocess.check_output(['adb', '-P', str(port), 'kill-server'],
stderr=subprocess.STDOUT)
try:
# Run the adb client and have it start the adb server.
proc = subprocess.Popen(['adb', '-P', str(port), 'start-server'],
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
# Start threads that set events when stdout/stderr are closed.
stdout_event = threading.Event()
stdout_thread = threading.Thread(
target=ServerTest._read_pipe_and_set_event,
args=(proc.stdout, stdout_event))
stdout_thread.daemon = True
stdout_thread.start()
stderr_event = threading.Event()
stderr_thread = threading.Thread(
target=ServerTest._read_pipe_and_set_event,
args=(proc.stderr, stderr_event))
stderr_thread.daemon = True
stderr_thread.start()
# Wait for the adb client to finish. Once that has occurred, if
# stdin/stderr/stdout are still open, it must be open in the adb
# server.
proc.wait()
# Try to write to stdin which we expect is closed. If it isn't
# closed, we should get an IOError. If we don't get an IOError,
# stdin must still be open in the adb server. The adb client is
# probably letting the adb server inherit stdin which would be
# wrong.
with self.assertRaises(IOError):
proc.stdin.write('x')
# Wait a few seconds for stdout/stderr to be closed (in the success
# case, this won't wait at all). If there is a timeout, that means
# stdout/stderr were not closed and and they must be open in the adb
# server, suggesting that the adb client is letting the adb server
# inherit stdout/stderr which would be wrong.
self.assertTrue(stdout_event.wait(5), "adb stdout not closed")
self.assertTrue(stderr_event.wait(5), "adb stderr not closed")
finally:
# If we started a server, kill it.
subprocess.check_output(['adb', '-P', str(port), 'kill-server'],
stderr=subprocess.STDOUT)
class EmulatorTest(unittest.TestCase):
"""Tests for the emulator connection."""
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
def _reset_socket_on_close(self, sock):
"""Use SO_LINGER to cause TCP RST segment to be sent on socket close."""
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
# The linger structure is two shorts on Windows, but two ints on Unix.
linger_format = 'hh' if os.name == 'nt' else 'ii'
l_onoff = 1
l_linger = 0
sock.setsockopt(socket.SOL_SOCKET, socket.SO_LINGER,
struct.pack(linger_format, l_onoff, l_linger))
# Verify that we set the linger structure properly by retrieving it.
linger = sock.getsockopt(socket.SOL_SOCKET, socket.SO_LINGER, 16)
self.assertEqual((l_onoff, l_linger),
struct.unpack_from(linger_format, linger))
def test_emu_kill(self):
"""Ensure that adb emu kill works.
Bug: https://code.google.com/p/android/issues/detail?id=21021
"""
with contextlib.closing(
socket.socket(socket.AF_INET, socket.SOCK_STREAM)) as listener:
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
# Use SO_REUSEADDR so subsequent runs of the test can grab the port
# even if it is in TIME_WAIT.
listener.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
listener.bind(('127.0.0.1', 0))
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
listener.listen(4)
port = listener.getsockname()[1]
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
# Now that listening has started, start adb emu kill, telling it to
# connect to our mock emulator.
proc = subprocess.Popen(
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
['adb', '-s', 'emulator-' + str(port), 'emu', 'kill'],
stderr=subprocess.STDOUT)
accepted_connection, addr = listener.accept()
with contextlib.closing(accepted_connection) as conn:
# If WSAECONNABORTED (10053) is raised by any socket calls,
# then adb probably isn't reading the data that we sent it.
conn.sendall('Android Console: type \'help\' for a list ' +
'of commands\r\n')
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
conn.sendall('OK\r\n')
with contextlib.closing(conn.makefile()) as connf:
line = connf.readline()
if line.startswith('auth'):
# Ignore the first auth line.
line = connf.readline()
self.assertEqual('kill\n', line)
self.assertEqual('quit\n', connf.readline())
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
conn.sendall('OK: killing emulator, bye bye\r\n')
# Use SO_LINGER to send TCP RST segment to test whether adb
# ignores WSAECONNRESET on Windows. This happens with the
# real emulator because it just calls exit() without closing
# the socket or calling shutdown(SD_SEND). At process
# termination, Windows sends a TCP RST segment for every
# open socket that shutdown(SD_SEND) wasn't used on.
self._reset_socket_on_close(conn)
# Wait for adb to finish, so we can check return code.
proc.communicate()
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
# If this fails, adb probably isn't ignoring WSAECONNRESET when
# reading the response from the adb emu kill command (on Windows).
self.assertEqual(0, proc.returncode)
def test_emulator_connect(self):
"""Ensure that the emulator can connect.
Bug: http://b/78991667
"""
with adb_server() as server_port:
with fake_adbd() as port:
serial = 'emulator-{}'.format(port - 1)
# Ensure that the emulator is not there.
try:
subprocess.check_output(['adb', '-P', str(server_port),
'-s', serial, 'get-state'],
stderr=subprocess.STDOUT)
self.fail('Device should not be available')
except subprocess.CalledProcessError as err:
self.assertEqual(
err.output.strip(),
'error: device \'{}\' not found'.format(serial))
# Let the ADB server know that the emulator has started.
with contextlib.closing(
socket.socket(socket.AF_INET, socket.SOCK_STREAM)) as sock:
sock.connect(('localhost', server_port))
command = 'host:emulator:{}'.format(port)
sock.sendall('%04x%s' % (len(command), command))
# Ensure the emulator is there.
subprocess.check_call(['adb', '-P', str(server_port),
'-s', serial, 'wait-for-device'])
output = subprocess.check_output(['adb', '-P', str(server_port),
'-s', serial, 'get-state'])
self.assertEqual(output.strip(), 'device')
class ConnectionTest(unittest.TestCase):
"""Tests for adb connect."""
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
def test_connect_ipv4_ipv6(self):
"""Ensure that `adb connect localhost:1234` will try both IPv4 and IPv6.
Bug: http://b/30313466
"""
for protocol in (socket.AF_INET, socket.AF_INET6):
try:
with fake_adbd(protocol=protocol) as port:
serial = 'localhost:{}'.format(port)
with adb_connect(self, serial):
pass
except socket.error:
print("IPv6 not available, skipping")
continue
def test_already_connected(self):
"""Ensure that an already-connected device stays connected."""
with fake_adbd() as port:
serial = 'localhost:{}'.format(port)
with adb_connect(self, serial):
# b/31250450: this always returns 0 but probably shouldn't.
output = subprocess.check_output(['adb', 'connect', serial])
self.assertEqual(
output.strip(), 'already connected to {}'.format(serial))
def test_reconnect(self):
"""Ensure that a disconnected device reconnects."""
with fake_adbd() as port:
serial = 'localhost:{}'.format(port)
with adb_connect(self, serial):
output = subprocess.check_output(['adb', '-s', serial,
'get-state'])
self.assertEqual(output.strip(), 'device')
# This will fail.
proc = subprocess.Popen(['adb', '-s', serial, 'shell', 'true'],
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT)
output, _ = proc.communicate()
self.assertEqual(output.strip(), 'error: closed')
subprocess.check_call(['adb', '-s', serial, 'wait-for-device'])
output = subprocess.check_output(['adb', '-s', serial,
'get-state'])
self.assertEqual(output.strip(), 'device')
# Once we explicitly kick a device, it won't attempt to
# reconnect.
output = subprocess.check_output(['adb', 'disconnect', serial])
self.assertEqual(
output.strip(), 'disconnected {}'.format(serial))
try:
subprocess.check_output(['adb', '-s', serial, 'get-state'],
stderr=subprocess.STDOUT)
self.fail('Device should not be available')
except subprocess.CalledProcessError as err:
self.assertEqual(
err.output.strip(),
'error: device \'{}\' not found'.format(serial))
adb: fix adb client running out of sockets on Windows Background ========== On Windows, if you run "adb shell exit" in a loop in two windows, eventually the adb client will be unable to connect to the adb server. I think connect() is returning WSAEADDRINUSE: "Only one usage of each socket address (protocol/network address/port) is normally permitted. (10048)". The Windows System Event Log may also show Event 4227, Tcpip. Netstat output is filled with: # for the adb server TCP 127.0.0.1:5037 127.0.0.1:65523 TIME_WAIT # for the adb client TCP 127.0.0.1:65523 127.0.0.1:5037 TIME_WAIT The error probably means that the client is running out of free address:port pairs. The first netstat line is unavoidable, but the second line exists because the adb client is not waiting for orderly/graceful shutdown of the socket, and that is apparently required on Windows to get rid of the second line. For more info, see https://github.com/CompareAndSwap/SocketCloseTest . This is exacerbated by the fact that "adb shell exit" makes 4 socket connections to the adb server: 1) host:version, 2) host:features, 3) host:version (again), 4) shell:exit. Also exacerbating is the fact that the adb protocol is length-prefixed so the client typically does not have to 'read() until zero' which effectively waits for orderly/graceful shutdown. The Fix ======= Introduce a function, ReadOrderlyShutdown(), that should be called in the adb client to wait for the server to close its socket, before closing the client socket. I reviewed all code where the adb client makes a connection to the adb server and added ReadOrderlyShutdown() when it made sense. I wasn't able to add it to the following: * interactive_shell: this doesn't matter because this is interactive and thus can't be run fast enough to use up ports. * adb sideload: I couldn't get enough test coverage and I don't think this is being called frequently enough to be a problem. * send_shell_command, backup, adb_connect_command, adb shell, adb exec-out, install_multiple_app, adb_send_emulator_command: These already wait for server socket shutdown since they already call recv() until zero. * restore, adb exec-in: protocol design can't have the server close first. * adb start-server: no fd is actually returned * create_local_service_socket, local_connect_arbitrary_ports, connect_device: probably called rarely enough not to be a problem. Also in this change =================== * Clarify comments in when adb_shutdown() is called before exit(). * add some missing adb_close() in adb sideload. * Fixup error handling and comments in adb_send_emulator_command(). * Make SyncConnection::SendQuit return a success boolean. * Add unittest for adb emu kill command. This gets code coverage over this very careful piece of code. Change-Id: Iad0b1336f5b74186af2cd35f7ea827d0fa77a17c Signed-off-by: Spencer Low <CompareAndSwap@gmail.com>
2015-10-15 02:32:44 +02:00
def main():
"""Main entrypoint."""
random.seed(0)
unittest.main(verbosity=3)
if __name__ == '__main__':
main()