webrtc_m130/p2p/base/tcpport_unittest.cc

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

234 lines
9.4 KiB
C++
Raw Normal View History

Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
/*
* Copyright 2016 The WebRTC Project Authors. All rights reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
#include <list>
#include <memory>
#include "p2p/base/basicpacketsocketfactory.h"
#include "p2p/base/tcpport.h"
#include "rtc_base/gunit.h"
#include "rtc_base/thread.h"
#include "rtc_base/virtualsocketserver.h"
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
using rtc::SocketAddress;
using cricket::Connection;
using cricket::Port;
using cricket::TCPPort;
using cricket::ICE_UFRAG_LENGTH;
using cricket::ICE_PWD_LENGTH;
static int kTimeout = 1000;
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
static const SocketAddress kLocalAddr("11.11.11.11", 0);
static const SocketAddress kLocalIPv6Addr("2401:fa00:4:1000:be30:5bff:fee5:c3",
0);
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
static const SocketAddress kAlternateLocalAddr("1.2.3.4", 0);
static const SocketAddress kRemoteAddr("22.22.22.22", 0);
static const SocketAddress kRemoteIPv6Addr("2401:fa00:4:1000:be30:5bff:fee5:c4",
0);
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
class ConnectionObserver : public sigslot::has_slots<> {
public:
explicit ConnectionObserver(Connection* conn) {
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
conn->SignalDestroyed.connect(this, &ConnectionObserver::OnDestroyed);
}
bool connection_destroyed() { return connection_destroyed_; }
private:
void OnDestroyed(Connection*) { connection_destroyed_ = true; }
bool connection_destroyed_ = false;
};
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
class TCPPortTest : public testing::Test, public sigslot::has_slots<> {
public:
TCPPortTest()
: ss_(new rtc::VirtualSocketServer()),
main_(ss_.get()),
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
socket_factory_(rtc::Thread::Current()),
username_(rtc::CreateRandomString(ICE_UFRAG_LENGTH)),
password_(rtc::CreateRandomString(ICE_PWD_LENGTH)) {}
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
rtc::Network* MakeNetwork(const SocketAddress& addr) {
networks_.emplace_back("unittest", "unittest", addr.ipaddr(), 32);
networks_.back().AddIP(addr.ipaddr());
return &networks_.back();
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
}
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
std::unique_ptr<TCPPort> CreateTCPPort(const SocketAddress& addr) {
return std::unique_ptr<TCPPort>(
TCPPort::Create(&main_, &socket_factory_, MakeNetwork(addr), 0, 0,
username_, password_, true));
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
}
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
std::unique_ptr<TCPPort> CreateTCPPort(rtc::Network* network) {
return std::unique_ptr<TCPPort>(TCPPort::Create(
&main_, &socket_factory_, network, 0, 0, username_, password_, true));
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
}
protected:
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
// When a "create port" helper method is called with an IP, we create a
// Network with that IP and add it to this list. Using a list instead of a
// vector so that when it grows, pointers aren't invalidated.
std::list<rtc::Network> networks_;
std::unique_ptr<rtc::VirtualSocketServer> ss_;
rtc::AutoSocketServerThread main_;
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
rtc::BasicPacketSocketFactory socket_factory_;
std::string username_;
std::string password_;
};
TEST_F(TCPPortTest, TestTCPPortWithLocalhostAddress) {
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
SocketAddress local_address("127.0.0.1", 0);
// After calling this, when TCPPort attempts to get a socket bound to
// kLocalAddr, it will end up using localhost instead.
ss_->SetAlternativeLocalAddress(kLocalAddr.ipaddr(), local_address.ipaddr());
auto local_port = CreateTCPPort(kLocalAddr);
auto remote_port = CreateTCPPort(kRemoteAddr);
local_port->PrepareAddress();
remote_port->PrepareAddress();
Connection* conn = local_port->CreateConnection(remote_port->Candidates()[0],
Port::ORIGIN_MESSAGE);
EXPECT_TRUE_WAIT(conn->connected(), kTimeout);
// Verify that the socket actually used localhost, otherwise this test isn't
// doing what it meant to.
ASSERT_EQ(local_address.ipaddr(),
local_port->Candidates()[0].address().ipaddr());
}
// If the address the socket ends up bound to does not match any address of the
// TCPPort's Network, then the socket should be discarded and no candidates
// should be signaled. In the context of ICE, where one TCPPort is created for
// each Network, when this happens it's likely that the unexpected address is
// associated with some other Network, which another TCPPort is already
// covering.
TEST_F(TCPPortTest, TCPPortDiscardedIfBoundAddressDoesNotMatchNetwork) {
// Sockets bound to kLocalAddr will actually end up with kAlternateLocalAddr.
ss_->SetAlternativeLocalAddress(kLocalAddr.ipaddr(),
kAlternateLocalAddr.ipaddr());
// Create ports (local_port is the one whose IP will end up reassigned).
auto local_port = CreateTCPPort(kLocalAddr);
auto remote_port = CreateTCPPort(kRemoteAddr);
local_port->PrepareAddress();
remote_port->PrepareAddress();
// Tell port to create a connection; it should be destroyed when it's
// realized that it's using an unexpected address.
Connection* conn = local_port->CreateConnection(remote_port->Candidates()[0],
Port::ORIGIN_MESSAGE);
ConnectionObserver observer(conn);
EXPECT_TRUE_WAIT(observer.connection_destroyed(), kTimeout);
}
// A caveat for the above logic: if the socket ends up bound to one of the IPs
// associated with the Network, just not the "best" one, this is ok.
TEST_F(TCPPortTest, TCPPortNotDiscardedIfNotBoundToBestIP) {
// Sockets bound to kLocalAddr will actually end up with kAlternateLocalAddr.
ss_->SetAlternativeLocalAddress(kLocalAddr.ipaddr(),
kAlternateLocalAddr.ipaddr());
// Set up a network with kLocalAddr1 as the "best" IP, and kAlternateLocalAddr
// as an alternate.
rtc::Network* network = MakeNetwork(kLocalAddr);
network->AddIP(kAlternateLocalAddr.ipaddr());
ASSERT_EQ(kLocalAddr.ipaddr(), network->GetBestIP());
// Create ports (using our special 2-IP Network for local_port).
auto local_port = CreateTCPPort(network);
auto remote_port = CreateTCPPort(kRemoteAddr);
local_port->PrepareAddress();
remote_port->PrepareAddress();
// Expect connection to succeed.
Connection* conn = local_port->CreateConnection(remote_port->Candidates()[0],
Port::ORIGIN_MESSAGE);
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
EXPECT_TRUE_WAIT(conn->connected(), kTimeout);
Make Port (and subclasses) fully "Network"-based, instead of IP-based. For ICE, we want sockets that are bound to specific network interfaces, rather than to specific IP addresses. So, a while ago, we added a "Network" class that gets passed into the Port constructor, in addition to the IP address as before. But we never finished the job of removing the IP address field, such that a Port only guarantees something about the network interface it's associated with, and not the specific IP address it ends up with. This CL does that, and as a consequence, if a port ends up bound to an IP address other than the "best" one (returned by Network::GetBestIP), this *won't* be treated as an error. This is relevant to Android, where even though we pass an IP address into "Bind" as a way of identifying the network, the socket actually gets bound using "android_setsocknetwork", which doesn't provide any guarantees about the IP address. So, if a network interface has multiple IPv6 addresses (for instance), we may not correctly predict the one the OS will choose, and that's ok. This CL also moves "SetAlternateLocalAddress" from VirtualSocket to VirtualSocketServer, which makes for much more readable test code. The next step, if there is one, is to pass along the Network class all the way to SocketServer::Bind. Then the socket server could do smart things with the network information. We could even stick a platform- specific network handle in the Network object, such that the socket server could use it for the binding, or for "sendmsg", for example. See bug 7026 for more context about the sendmsg idea. BUG=webrtc:7715 Review-Url: https://codereview.webrtc.org/2989303002 Cr-Commit-Position: refs/heads/master@{#19251}
2017-08-04 15:01:57 -07:00
// Verify that the socket actually used the alternate address, otherwise this
// test isn't doing what it meant to.
ASSERT_EQ(kAlternateLocalAddr.ipaddr(),
local_port->Candidates()[0].address().ipaddr());
Reland of Allow the localhost IP address even if it does not match the tcp port address (patchset #1 id:1 of https://codereview.webrtc.org/1979463003/ ) Reason for revert: Relanding this change since the revert didn't make a difference. Original issue's description: > Revert of Allow the localhost IP address even if it does not match the tcp port address (patchset #4 id:120001 of https://codereview.webrtc.org/1914803002/ ) > > Reason for revert: > Speculatively reverting due to failures on the memcheck bot (and possibly other bots): > > https://build.chromium.org/p/client.webrtc/builders/Linux%20Memcheck/builds/5910/steps/video_engine_tests/logs/EndToEndTest.SendsAndReceivesH264 > > Original issue's description: > > This fixes an issue similar to > > https://bugs.chromium.org/p/webrtc/issues/detail?id=3927 > > where the localhost IP does not match the turn port address. > > The issue here is in the TCP port. > > > > BUG= > > R=pthatcher@webrtc.org > > > > Committed: https://crrev.com/6705012904e6cbbefce6fbce0a3f615b7aeafd8f > > Cr-Commit-Position: refs/heads/master@{#12707} > > TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org > # Skipping CQ checks because original CL landed less than 1 days ago. > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG= > > Committed: https://crrev.com/1cbf0a73eb4b475e8beb878ea3a4d650191f0c08 > Cr-Commit-Position: refs/heads/master@{#12728} TBR=pthatcher@webrtc.org,deadbeef@webrtc.org,honghaiz@webrtc.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.webrtc.org/1979073002 Cr-Commit-Position: refs/heads/master@{#12746}
2016-05-14 03:19:31 -07:00
}
// Regression test for crbug.com/webrtc/8972, caused by buggy comparison
// between rtc::IPAddress and rtc::InterfaceAddress.
TEST_F(TCPPortTest, TCPPortNotDiscardedIfBoundToTemporaryIP) {
networks_.emplace_back("unittest", "unittest", kLocalIPv6Addr.ipaddr(), 32);
networks_.back().AddIP(rtc::InterfaceAddress(
kLocalIPv6Addr.ipaddr(), rtc::IPV6_ADDRESS_FLAG_TEMPORARY));
auto local_port = CreateTCPPort(&networks_.back());
auto remote_port = CreateTCPPort(kRemoteIPv6Addr);
local_port->PrepareAddress();
remote_port->PrepareAddress();
// Connection should succeed if the port isn't discarded.
Connection* conn = local_port->CreateConnection(remote_port->Candidates()[0],
Port::ORIGIN_MESSAGE);
ASSERT_NE(nullptr, conn);
EXPECT_TRUE_WAIT(conn->connected(), kTimeout);
}
class SentPacketCounter : public sigslot::has_slots<> {
public:
explicit SentPacketCounter(TCPPort* p) {
p->SignalSentPacket.connect(this, &SentPacketCounter::OnSentPacket);
}
int sent_packets() const { return sent_packets_; }
private:
void OnSentPacket(const rtc::SentPacket&) { ++sent_packets_; }
int sent_packets_ = 0;
};
// Test that SignalSentPacket is fired when a packet is successfully sent, for
// both TCP client and server sockets.
TEST_F(TCPPortTest, SignalSentPacket) {
std::unique_ptr<TCPPort> client(CreateTCPPort(kLocalAddr));
std::unique_ptr<TCPPort> server(CreateTCPPort(kRemoteAddr));
client->SetIceRole(cricket::ICEROLE_CONTROLLING);
server->SetIceRole(cricket::ICEROLE_CONTROLLED);
client->PrepareAddress();
server->PrepareAddress();
Connection* client_conn =
client->CreateConnection(server->Candidates()[0], Port::ORIGIN_MESSAGE);
ASSERT_NE(nullptr, client_conn);
ASSERT_TRUE_WAIT(client_conn->connected(), kTimeout);
// Need to get the port of the actual outgoing socket, not the server socket..
cricket::Candidate client_candidate = client->Candidates()[0];
client_candidate.set_address(static_cast<cricket::TCPConnection*>(client_conn)
->socket()
->GetLocalAddress());
Connection* server_conn =
server->CreateConnection(client_candidate, Port::ORIGIN_THIS_PORT);
ASSERT_NE(nullptr, server_conn);
ASSERT_TRUE_WAIT(server_conn->connected(), kTimeout);
client_conn->Ping(rtc::TimeMillis());
server_conn->Ping(rtc::TimeMillis());
ASSERT_TRUE_WAIT(client_conn->writable(), kTimeout);
ASSERT_TRUE_WAIT(server_conn->writable(), kTimeout);
SentPacketCounter client_counter(client.get());
SentPacketCounter server_counter(server.get());
static const char kData[] = "hello";
for (int i = 0; i < 10; ++i) {
client_conn->Send(&kData, sizeof(kData), rtc::PacketOptions());
server_conn->Send(&kData, sizeof(kData), rtc::PacketOptions());
}
EXPECT_EQ_WAIT(10, client_counter.sent_packets(), kTimeout);
EXPECT_EQ_WAIT(10, server_counter.sent_packets(), kTimeout);
}