summaryrefslogtreecommitdiffhomepage
path: root/pkg/tcpip/stack
AgeCommit message (Collapse)Author
2020-10-31net/tcpip: connect to unset loopback address has to return EADDRNOTAVAILAndrei Vagin
In the docker container, the ipv6 loopback address is not set, and connect("::1") has to return ENEADDRNOTAVAIL in this case. Without this fix, it returns EHOSTUNREACH. PiperOrigin-RevId: 340002915
2020-10-30Automated rollback of changelist 339750876Dean Deng
PiperOrigin-RevId: 339945377
2020-10-29Automated rollback of changelist 339675182Dean Deng
PiperOrigin-RevId: 339750876
2020-10-29Make RedirectTarget thread safeKevin Krakauer
Fixes #4613. PiperOrigin-RevId: 339746784
2020-10-29Keep magic constants out of netstackKevin Krakauer
PiperOrigin-RevId: 339721152
2020-10-29Delay goroutine creation during TCP handshake for accept/connect.Dean Deng
Refactor TCP handshake code so that when connect is initiated, the initial SYN is sent before creating a goroutine to handle the rest of the handshake (which blocks). Similarly, the initial SYN-ACK is sent inline when SYN is received during accept. Some additional cleanup is done as well. Eventually we would like to complete connections in the dispatcher without requiring a wakeup to complete the handshake. This refactor makes that easier. Updates #231 PiperOrigin-RevId: 339675182
2020-10-28Avoid time.Now in NUDTamir Duberstein
Use the stack clock instead. Change NeighborEntry.UpdatedAt to UpdatedAtNanos. PiperOrigin-RevId: 339520566
2020-10-27Add support for Timestamp and RecordRoute IP optionsJulian Elischer
IPv4 options extend the size of the IP header and have a basic known format. The framework can process that format without needing to know about every possible option. We can add more code to handle additional option types as we need them. Bad options or mangled option entries can result in ICMP Parameter Problem packets. The first types we support are the Timestamp option and the Record Route option, included in this change. The options are processed at several points in the packet flow within the Network stack, with slightly different requirements. The framework includes a mechanism to control this at each point. Support has been added for such points which are only present in upcoming CLs such as during packet forwarding and fragmentation. With this change, 'ping -R' and 'ping -T' work against gVisor and Fuchsia. $ ping -R 192.168.1.2 PING 192.168.1.2 (192.168.1.2) 56(124) bytes of data. 64 bytes from 192.168.1.2: icmp_seq=1 ttl=64 time=0.990 ms NOP RR: 192.168.1.1 192.168.1.2 192.168.1.1 $ ping -T tsprespec 192.168.1.2 192.168.1.1 192.168.1.2 PING 192.168.1.2 (192.168.1.2) 56(124) bytes of data. 64 bytes from 192.168.1.2: icmp_seq=1 ttl=64 time=1.20 ms TS: 192.168.1.2 71486821 absolute 192.168.1.1 746 Unit tests included for generic options, Timestamp options and Record Route options. PiperOrigin-RevId: 339379076
2020-10-27Pass NeighborEntry in NUD callbacksTamir Duberstein
...instead of passing its fields piecemeal. PiperOrigin-RevId: 339345899
2020-10-27Add basic address deletion to netlinkIan Lewis
Updates #3921 PiperOrigin-RevId: 339195417
2020-10-23Wait before transitioning NUD entries from Probe to FailedSam Balana
Wait an additional RetransmitTimer duration after the last probe before transitioning to Failed. The previous implementation transitions immediately to Failed after sending the last probe, which is erroneous behavior. PiperOrigin-RevId: 338723794
2020-10-23Do not hold NIC local address in neighbor entriesGhanan Gowripalan
Previously, the NIC local address used when completing link resolution was held in the neighbor entry. A neighbor is not identified by any NIC local address so remove it. PiperOrigin-RevId: 338699695
2020-10-22Pass NetworkInterface to LinkAddressRequestGhanan Gowripalan
Previously a link endpoint was passed to stack.LinkAddressResolver.LinkAddressRequest. With this change, implementations that want a route for the link address request may find one through the stack. Other implementations that want to send a packet without a route may continue to do so using the network interface directly. Test: - arp_test.TestLinkAddressRequest - ipv6.TestLinkAddressRequest PiperOrigin-RevId: 338577474
2020-10-20Fix locking in AddressableEndpointStateGhanan Gowripalan
PiperOrigin-RevId: 338156438
2020-10-20Fix nogo tests.Ting-Yu Wang
//pkg/tcpip/stack:stack_x_test_nogo //pkg/tcpip/transport/raw:raw_nogo PiperOrigin-RevId: 338153265
2020-10-16Don't include link header when forwarding packetsGhanan Gowripalan
Before this change, if a link header was included in an incoming packet that is forwarded, the packet that gets sent out will take the original packet and add a link header to it while keeping the old link header. This would make the sent packet look like: OUTGOING LINK HDR | INCOMING LINK HDR | NETWORK HDR | ... Obviously this is incorrect as we should drop the incoming link header and only include the outgoing link header. This change fixes this bug. Test: integration_test.TestForwarding PiperOrigin-RevId: 337571447
2020-10-15Process NAs without target link-layer addressesSam Balana
RFC 4861 section 4.4 comments the Target link-layer address option is sometimes optional in a Neighbor Advertisement packet: "When responding to a unicast Neighbor Solicitation this option SHOULD be included." Tests: pkg/tcpip/stack:stack_test - TestEntryStaleToReachableWhenSolicitedConfirmationWithoutAddress - TestEntryDelayToReachableWhenSolicitedConfirmationWithoutAddress - TestEntryProbeToReachableWhenSolicitedConfirmationWithoutAddress pkg/tcpip/network/ipv6:ipv6_test - TestCallsToNeighborCache PiperOrigin-RevId: 337396493
2020-10-14Find route before sending NA responseGhanan Gowripalan
This change also brings back the stack.Route.ResolveWith method so that we can immediately resolve a route when sending an NA in response to a a NS with a source link layer address option. Test: ipv6_test.TestNeighorSolicitationResponse PiperOrigin-RevId: 337185461
2020-10-09Automated rollback of changelist 336304024Ghanan Gowripalan
PiperOrigin-RevId: 336339194
2020-10-09Automated rollback of changelist 336185457Bhasker Hariharan
PiperOrigin-RevId: 336304024
2020-10-08Only block resolution when NUD is incompleteGhanan Gowripalan
When a completed entry exists for a neighbor, there is no need to block while reachability is (re)confirmed. The stack should continue to use the neighbor's link address while NUD is performed. Test: stack_test.TestNeighborCacheReplace PiperOrigin-RevId: 336199043
2020-10-08Do not resolve routes immediatelyGhanan Gowripalan
When a response needs to be sent to an incoming packet, the stack should consult its neighbour table to determine the remote address's link address. When an entry does not exist in the stack's neighbor table, the stack should queue the packet while link resolution completes. See comments. PiperOrigin-RevId: 336185457
2020-10-05Remove AssignableAddressEndpoint.NetworkEndpointGhanan Gowripalan
We can get the network endpoint directly from the NIC. This is a preparatory CL for when a Route needs to hold a dedicated NIC as its output interface. This is because when forwarding is enabled, packets may be sent from a NIC different from the NIC a route's local address is associated with. PiperOrigin-RevId: 335484500
2020-09-30ip6tables: redirect supportKevin Krakauer
Adds support for the IPv6-compatible redirect target. Redirection is a limited form of DNAT, where the destination is always the localhost. Updates #3549. PiperOrigin-RevId: 334698344
2020-09-30Count IP OutgoingPacketErrors in the NetworkEndpoint methodsArthur Sfez
Before this change, OutgoingPacketErrors was incremented in the stack.Route methods. This was going to be a problem once IPv4/IPv6 WritePackets support fragmentation because Route.WritePackets might now know how many packets are left after an error occurs. Test: - pkg/tcpip/network/ipv4:ipv4_test - pkg/tcpip/network/ipv6:ipv6_test PiperOrigin-RevId: 334687983
2020-09-30Use the ICMP error response facilityJulian Elischer
Add code in IPv6 to send ICMP packets while processing extension headers. Add some accounting in processing IPV6 Extension headers which allows us to report meaningful information back in ICMP parameter problem packets. IPv4 also needs to send a message when an unsupported protocol is requested. Add some tests to generate both ipv4 and ipv6 packets with various errors and check the responses. Add some new checkers and cleanup some inconsistencies in the messages in that file. Add new error types for the ICMPv4/6 generators. Fix a bug in the ICMPv4 generator that stopped it from generating "Unknown protocol" messages. Updates #2211 PiperOrigin-RevId: 334661716
2020-09-30Make all Target.Action implementation pointer receiversKevin Krakauer
PiperOrigin-RevId: 334652998
2020-09-29iptables: remove unused min/max NAT range fieldsKevin Krakauer
PiperOrigin-RevId: 334531794
2020-09-29Return permanent addresses when NIC is downGhanan Gowripalan
Test: stack_test.TestGetMainNICAddressWhenNICDisabled PiperOrigin-RevId: 334513286
2020-09-29Don't allow broadcast/multicast source addressGhanan Gowripalan
As per relevant IP RFCS (see code comments), broadcast (for IPv4) and multicast addresses are not allowed. Currently checks for these are done at the transport layer, but since it is explicitly forbidden at the IP layers, check for them there. This change also removes the UDP.InvalidSourceAddress stat since there is no longer a need for it. Test: ip_test.TestSourceAddressValidation PiperOrigin-RevId: 334490971
2020-09-29iptables: refactor to make targets extendableKevin Krakauer
Like matchers, targets should use a module-like register/lookup system. This replaces the brittle switch statements we had before. The only behavior change is supporing IPT_GET_REVISION_TARGET. This makes it much easier to add IPv6 redirect in the next change. Updates #3549. PiperOrigin-RevId: 334469418
2020-09-29Trim Network/Transport Endpoint/ProtocolGhanan Gowripalan
* Remove Capabilities and NICID methods from NetworkEndpoint. * Remove linkEP and stack parameters from NetworkProtocol.NewEndpoint. The LinkEndpoint can be fetched from the NetworkInterface. The stack is passed to the NetworkProtocol when it is created so the NetworkEndpoint can get it from its protocol. * Remove stack parameter from TransportProtocol.NewEndpoint. Like the NetworkProtocol/Endpoint, the stack is passed to the TransportProtocol when it is created. PiperOrigin-RevId: 334332721
2020-09-29Move IP state from NIC to NetworkEndpoint/ProtocolGhanan Gowripalan
* Add network address to network endpoints. Hold network-specific state in the NetworkEndpoint instead of the stack. This results in the stack no longer needing to "know" about the network endpoints and special case certain work for various endpoints (e.g. IPv6 DAD). * Provide NetworkEndpoints with an NetworkInterface interface. Instead of just passing the NIC ID of a NIC, pass an interface so the network endpoint may query other information about the NIC such as whether or not it is a loopback device. * Move NDP code and state to the IPv6 package. NDP is IPv6 specific so there is no need for it to live in the stack. * Control forwarding through NetworkProtocols instead of Stack Forwarding should be controlled on a per-network protocol basis so forwarding configurations are now controlled through network protocols. * Remove stack.referencedNetworkEndpoint. Now that addresses are exposed via AddressEndpoint and only one NetworkEndpoint is created per interface, there is no need for a referenced NetworkEndpoint. * Assume network teardown methods are infallible. Fixes #3871, #3916 PiperOrigin-RevId: 334319433
2020-09-28Support creating protocol instances with Stack refGhanan Gowripalan
Network or transport protocols may want to reach the stack. Support this by letting the stack create the protocol instances so it can pass a reference to itself at protocol creation time. Note, protocols do not yet use the stack in this CL but later CLs will make use of the stack from protocols. PiperOrigin-RevId: 334260210
2020-09-26Remove generic ICMP errorsGhanan Gowripalan
Generic ICMP errors were required because the transport dispatcher was given the responsibility of sending ICMP errors in response to transport packet delivery failures. Instead, the transport dispatcher should let network layer know it failed to deliver a packet (and why) and let the network layer make the decision as to what error to send (if any). Fixes #4068 PiperOrigin-RevId: 333962333
2020-09-24Change segment/pending queue to use receive buffer limits.Bhasker Hariharan
segment_queue today has its own standalone limit of MaxUnprocessedSegments but this can be a problem in UnlockUser() we do not release the lock till there are segments to be processed. What can happen is as handleSegments dequeues packets more keep getting queued and we will never release the lock. This can keep happening even if the receive buffer is full because nothing can read() till we release the lock. Further having a separate limit for pending segments makes it harder to track memory usage etc. Unifying the limits makes it easier to reason about memory in use and makes the overall buffer behaviour more consistent. PiperOrigin-RevId: 333508122
2020-09-23Remove unused field from neighborEntryGhanan Gowripalan
PiperOrigin-RevId: 333405169
2020-09-23Extract ICMP error sender from UDPJulian Elischer
Store transport protocol number on packet buffers for use in ICMP error generation. Updates #2211. PiperOrigin-RevId: 333252762
2020-09-22Move stack.fakeClock into a separate packageToshi Kikuchi
PiperOrigin-RevId: 333138701
2020-09-20Merge pull request #3651 from ianlewis:ip-forwardinggVisor bot
PiperOrigin-RevId: 332760843
2020-09-18Count packets dropped by iptables in IPStatsKevin Krakauer
PiperOrigin-RevId: 332486383
2020-09-17ip6tables: filter table supportKevin Krakauer
`ip6tables -t filter` is now usable. NAT support will come in a future CL. #3549 PiperOrigin-RevId: 332381801
2020-09-16Bind loopback subnets' lifetime to perm addressGhanan Gowripalan
The lifetime of addreses in a loopback interface's associated subnets should be bound to their respective permanent addresses. This change also fixes a race when the stack attempts to get an IPv4 rereferencedNetworkEndpoint for an address in an associated subnet on a loopback interface. Before this change, the stack would only check if an IPv4 address is contained in an associated subnet while holding a read lock but wouldn't do this same check after releasing the read lock for a write lock to create a temporary address. This may cause the stack to bind the lifetime of the address to a new (temporary) endpoint instead of the associated subnet's permanent address. Test: integration_test.TestLoopbackSubnetLifetimeBoundToAddr PiperOrigin-RevId: 332094719
2020-09-16Receive broadcast packets on interested endpointsGhanan Gowripalan
When a broadcast packet is received by the stack, the packet should be delivered to each endpoint that may be interested in the packet. This includes all any address and specified broadcast address listeners. Test: integration_test.TestReuseAddrAndBroadcast PiperOrigin-RevId: 332060652
2020-09-15Don't conclude broadcast from route destinationGhanan Gowripalan
The routing table (in its current) form should not be used to make decisions about whether a remote address is a broadcast address or not (for IPv4). Note, a destination subnet does not always map to a network. E.g. RouterA may have a route to 192.168.0.0/22 through RouterB, but RouterB may be configured with 4x /24 subnets on 4 different interfaces. See https://github.com/google/gvisor/issues/3938. PiperOrigin-RevId: 331819868
2020-09-08Increase resolution timeout for TestCacheResolutionSam Balana
Fixes pkg/tcpip/stack:stack_test flake experienced while running TestCacheResolution with gotsan. This occurs when the test-runner takes longer than the resolution timeout to call linkAddrCache.get. In this test we don't care about the resolution timeout, so set it to the maximum and rely on test-runner timeouts to avoid deadlocks. PiperOrigin-RevId: 330566250
2020-09-08Improve type safety for transport protocol optionsGhanan Gowripalan
The existing implementation for TransportProtocol.{Set}Option take arguments of an empty interface type which all types (implicitly) implement; any type may be passed to the functions. This change introduces marker interfaces for transport protocol options that may be set or queried which transport protocol option types implement to ensure that invalid types are caught at compile time. Different interfaces are used to allow the compiler to enforce read-only or set-only socket options. RELNOTES: n/a PiperOrigin-RevId: 330559811
2020-09-03Use fine-grained mutex for stack.cleanupEndpoints.Bhasker Hariharan
stack.cleanupEndpoints is protected by the stack.mu but that can cause contention as the stack mutex is already acquired in a lot of hot paths during new endpoint creation /cleanup etc. Moving this to a fine grained mutex should reduce contention on the stack.mu. PiperOrigin-RevId: 330026151
2020-09-03Use atomic.Value for Stack.tcpProbeFunc.Jamie Liu
b/166980357#comment56 shows: - 837 goroutines blocked in: gvisor/pkg/sync/sync.(*RWMutex).Lock gvisor/pkg/tcpip/stack/stack.(*Stack).StartTransportEndpointCleanup gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).cleanupLocked gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).completeWorkerLocked gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).protocolMainLoop.func1 gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).protocolMainLoop - 695 goroutines blocked in: gvisor/pkg/sync/sync.(*RWMutex).Lock gvisor/pkg/tcpip/stack/stack.(*Stack).CompleteTransportEndpointCleanup gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).cleanupLocked gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).completeWorkerLocked gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).protocolMainLoop.func1 gvisor/pkg/tcpip/transport/tcp/tcp.(*endpoint).protocolMainLoop - 3882 goroutines blocked in: gvisor/pkg/sync/sync.(*RWMutex).Lock gvisor/pkg/tcpip/stack/stack.(*Stack).GetTCPProbe gvisor/pkg/tcpip/transport/tcp/tcp.newEndpoint gvisor/pkg/tcpip/transport/tcp/tcp.(*protocol).NewEndpoint gvisor/pkg/tcpip/stack/stack.(*Stack).NewEndpoint All of these are contending on Stack.mu. Stack.StartTransportEndpointCleanup() and Stack.CompleteTransportEndpointCleanup() insert/delete TransportEndpoints in a map (Stack.cleanupEndpoints), and the former also does endpoint unregistration while holding Stack.mu, so it's not immediately clear how feasible it is to replace the map with a mutex-less implementation or how much doing so would help. However, Stack.GetTCPProbe() just reads a function object (Stack.tcpProbeFunc) that is almost always nil (as far as I can tell, Stack.AddTCPProbe() is only called in tests), and it's called for every new TCP endpoint. So converting it to an atomic.Value should significantly reduce contention on Stack.mu, improving TCP endpoint creation latency and allowing TCP endpoint cleanup to proceed. PiperOrigin-RevId: 330004140
2020-09-02Fix Accept to not return error for sockets in accept queue.Bhasker Hariharan
Accept on gVisor will return an error if a socket in the accept queue was closed before Accept() was called. Linux will return the new fd even if the returned socket is already closed by the peer say due to a RST being sent by the peer. This seems to be intentional in linux more details on the github issue. Fixes #3780 PiperOrigin-RevId: 329828404
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-16 19:54:15 +0000