summaryrefslogtreecommitdiffhomepage
path: root/pkg/tcpip
AgeCommit message (Collapse)Author
2020-11-19Merge release-20201109.0-84-ge5650d124 (automated)gVisor bot
2020-11-18[netstack] Move SO_KEEPALIVE and SO_ACCEPTCONN option to SocketOptions.Ayush Ranjan
PiperOrigin-RevId: 343217712
2020-11-19Merge release-20201109.0-83-g93750a600 (automated)gVisor bot
2020-11-18Remove unused methods from stack.RouteGhanan Gowripalan
PiperOrigin-RevId: 343211553
2020-11-18Merge release-20201109.0-79-gdf37babd5 (automated)gVisor bot
2020-11-18[netstack] Move SO_REUSEPORT and SO_REUSEADDR option to SocketOptions.Ayush Ranjan
This changes also introduces: - `SocketOptionsHandler` interface which can be implemented by endpoints to handle endpoint specific behavior on SetSockOpt. This is analogous to what Linux does. - `DefaultSocketOptionsHandler` which is a default implementation of the above. This is embedded in all endpoints so that we don't have to uselessly implement empty functions. Endpoints with specific behavior can override the embedded method by manually defining its own implementation. PiperOrigin-RevId: 343158301
2020-11-18Merge release-20201109.0-78-gc85bba038 (automated)gVisor bot
2020-11-18Automated rollback of changelist 342700744Nayana Bidari
PiperOrigin-RevId: 343152780
2020-11-18Merge release-20201109.0-77-g3e73c519a (automated)gVisor bot
2020-11-18[netstack] Move SO_NO_CHECK option to SocketOptions.Ayush Ranjan
PiperOrigin-RevId: 343146856
2020-11-18Merge release-20201109.0-75-g60b97bfda (automated)gVisor bot
2020-11-18Fix loopback subnet routing errorGhanan Gowripalan
Packets should be properly routed when sending packets to addresses in the loopback subnet which are not explicitly assigned to the loopback interface. Tests: - integration_test.TestLoopbackAcceptAllInSubnetUDP - integration_test.TestLoopbackAcceptAllInSubnetTCP PiperOrigin-RevId: 343135643
2020-11-18Merge release-20201109.0-71-gfc342fb43 (automated)gVisor bot
2020-11-18[netstack] Move SO_PASSCRED option to SocketOptions.Ayush Ranjan
This change also makes the following fixes: - Make SocketOptions use atomic operations instead of having to acquire/drop locks upon each get/set option. - Make documentation more consistent. - Remove tcpip.SocketOptions from socketOpsCommon because it already exists in transport.Endpoint. - Refactors get/set socket options tests to be easily extendable. PiperOrigin-RevId: 343103780
2020-11-18Merge release-20201109.0-69-g9d148627f (automated)gVisor bot
2020-11-18Introduce stack.WritePacketToRemote, remove LinkEndpoint.WriteRawPacketBruno Dal Bo
Redefine stack.WritePacket into stack.WritePacketToRemote which lets the NIC decide whether to append link headers. PiperOrigin-RevId: 343071742
2020-11-17Remove sniffer from gonet_test.Bhasker Hariharan
This was added by mistake in cl/342868552. PiperOrigin-RevId: 343021431
2020-11-18Merge release-20201109.0-67-g0e32d98f3 (automated)gVisor bot
2020-11-17Fix endpoint.Read() when endpoint is in StateError.Bhasker Hariharan
If the endpoint is in StateError but e.hardErrorLocked() returns nil then return ErrClosedForRecieve. This can happen if a concurrent write on the same endpoint was in progress when the endpoint transitioned to an error state. PiperOrigin-RevId: 343018257
2020-11-17Merge release-20201109.0-60-g7492ed6bd (automated)gVisor bot
2020-11-17Add per-sniffer instance log prefixBruno Dal Bo
A prefix associated with a sniffer instance can help debug situations where more than one NIC (i.e. more than one sniffer) exists. PiperOrigin-RevId: 342950027
2020-11-17Merge release-20201109.0-56-g05d2a26f7 (automated)gVisor bot
2020-11-17Fix possible deadlock in UDP.Write().Bhasker Hariharan
In UDP endpoint.Write() sendUDP is called with e.mu Rlocked. But if this happens to send a datagram over loopback which ends up generating an ICMP response of say ErrNoPortReachable, the handling of the response in HandleControlPacket also acquires e.mu using RLock. This is mostly fine unless there is a competing caller trying to acquire e.mu in exclusive mode using Lock(). This will deadlock as a caller waiting in Lock() disallows an new RLocks() to ensure it can actually acquire the Lock. This is documented here https://golang.org/pkg/sync/#RWMutex. This change releases the endpoint mutex before calling sendUDP to resolve the possibility of the deadlock. Reported-by: syzbot+537989797548c66e8ee3@syzkaller.appspotmail.com Reported-by: syzbot+eb0b73b4ab486f7673ba@syzkaller.appspotmail.com PiperOrigin-RevId: 342894148
2020-11-17Merge release-20201109.0-55-gfb9a649f3 (automated)gVisor bot
2020-11-17Fix SO_ERROR behavior for TCP in gVisor.Bhasker Hariharan
Fixes the behaviour of SO_ERROR for tcp sockets where in linux it returns sk->sk_err and if sk->sk_err is 0 then it returns sk->sk_soft_err. In gVisor TCP we endpoint.HardError is the equivalent of sk->sk_err and endpoint.LastError holds soft errors. This change brings this into alignment with Linux such that both hard/soft errors are cleared when retrieved using getsockopt(.. SO_ERROR) is called on a socket. Fixes #3812 PiperOrigin-RevId: 342868552
2020-11-16Merge release-20201109.0-51-gcc5cfce4c (automated)gVisor bot
2020-11-16Remove ARP address workaroundGhanan Gowripalan
- Make AddressableEndpoint optional for NetworkEndpoint. Not all NetworkEndpoints need to support addressing (e.g. ARP), so AddressableEndpoint should only be implemented for protocols that support addressing such as IPv4 and IPv6. With this change, tcpip.ErrNotSupported will be returned by the stack when attempting to modify addresses on a network endpoint that does not support addressing. Now that packets are fully handled at the network layer, and (with this change) addresses are optional for network endpoints, we no longer need the workaround for ARP where a fake ARP address was added to each NIC that performs ARP so that packets would be delivered to the ARP layer. PiperOrigin-RevId: 342722547
2020-11-16Merge release-20201109.0-49-g758e45618 (automated)gVisor bot
2020-11-16Clean up fragmentation.ProcessToshi Kikuchi
- Pass a PacketBuffer directly instead of releaseCB - No longer pass a VectorisedView, which is included in the PacketBuffer - Make it an error if data size is not equal to (last - first + 1) - Set the callback for the reassembly timeout on NewFragmentation PiperOrigin-RevId: 342702432
2020-11-16Merge release-20201109.0-48-ga73877ac9 (automated)gVisor bot
2020-11-16Automated rollback of changelist 342312166Bhasker Hariharan
PiperOrigin-RevId: 342700744
2020-11-16Merge release-20201109.0-47-g840a133c6 (automated)gVisor bot
2020-11-16Automated rollback of changelist 340274194Dean Deng
PiperOrigin-RevId: 342669574
2020-11-14Merge release-20201030.0-95-g0fee59c8c (automated)gVisor bot
2020-11-13Requested Comment/Message wording changesJulian Elischer
PiperOrigin-RevId: 342366891
2020-11-13Merge release-20201030.0-92-g839dd9700 (automated)gVisor bot
2020-11-13RACK: Detect DSACKNayana Bidari
Detect if the ACK is a duplicate and update in RACK. PiperOrigin-RevId: 342332569
2020-11-13Merge release-20201030.0-91-g6c0f53002 (automated)gVisor bot
2020-11-13Decrement TTL/Hop Limit when forwarding IP packetsGhanan Gowripalan
If the packet must no longer be forwarded because its TTL/Hop Limit reaches 0, send an ICMP Time Exceeded error to the source. Required as per relevant RFCs. See comments in code for RFC references. Fixes #1085 Tests: - ipv4_test.TestForwarding - ipv6.TestForwarding PiperOrigin-RevId: 342323610
2020-11-13Merge release-20201030.0-89-g7fff51e50 (automated)gVisor bot
2020-11-13Refactor loss recovery in TCP.Nayana Bidari
The current implementation of loss recovery algorithms SACK and NewReno are in the same file(snd.go). The functions have multiple checks to see which one is currently used by the endpoint. This CL will refactor and separate the implementation of existing recovery algorithms which will help us to implement new recovery algorithms(such as RACK) with less changes to the existing code. There is no change in the behavior. PiperOrigin-RevId: 342312166
2020-11-13Merge release-20201030.0-83-g5bb64ce1b (automated)gVisor bot
2020-11-12Refactor SOL_SOCKET optionsNayana Bidari
Store all the socket level options in a struct and call {Get/Set}SockOpt on this struct. This will avoid implementing socket level options on all endpoints. This CL contains implementing one socket level option for tcp and udp endpoints. PiperOrigin-RevId: 342203981
2020-11-13Merge release-20201030.0-80-g638d64c63 (automated)gVisor bot
2020-11-12Change AllocationSize to SizeWithPadding as requestedJulian Elischer
RELNOTES: n/a PiperOrigin-RevId: 342176296
2020-11-13Merge release-20201030.0-77-gd700ba22a (automated)gVisor bot
2020-11-12Pad with a loop rather than a copy from an allocation.Julian Elischer
Add a unit test for ipv4.Encode and a round trip test. PiperOrigin-RevId: 342169517
2020-11-13Merge release-20201030.0-74-g1a972411b (automated)gVisor bot
2020-11-12Move packet handling to NetworkEndpointGhanan Gowripalan
The NIC should not hold network-layer state or logic - network packet handling/forwarding should be performed at the network layer instead of the NIC. Fixes #4688 PiperOrigin-RevId: 342166985
2020-11-11Merge release-20201030.0-68-g9c4102896 (automated)gVisor bot
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-29 01:35:42 +0000