[netstack] Implement MSG_ERRQUEUE flag for recvmsg(2).

Introduces the per-socket error queue and the necessary cmsg mechanisms.

PiperOrigin-RevId: 348028508

1 files changed, 43 insertions, 0 deletions

diff --git a/pkg/sentry/socket/netstack/netstack.go b/pkg/sentry/socket/netstack/netstack.go
index 23d5cab9c..a8ab6b385 100644
--- a/pkg/sentry/socket/netstack/netstack.go
+++ b/pkg/sentry/socket/netstack/netstack.go
@@ -2772,6 +2772,8 @@ func (s *socketOpsCommon) controlMessages() socket.ControlMessages {
 		IP: socket.IPControlMessages{
 			HasTimestamp:       s.readCM.HasTimestamp && s.sockOptTimestamp,
 			Timestamp:          s.readCM.Timestamp,
+			HasInq:             s.readCM.HasInq,
+			Inq:                s.readCM.Inq,
 			HasTOS:             s.readCM.HasTOS,
 			TOS:                s.readCM.TOS,
 			HasTClass:          s.readCM.HasTClass,
@@ -2779,6 +2781,7 @@ func (s *socketOpsCommon) controlMessages() socket.ControlMessages {
 			HasIPPacketInfo:    s.readCM.HasIPPacketInfo,
 			PacketInfo:         s.readCM.PacketInfo,
 			OriginalDstAddress: s.readCM.OriginalDstAddress,
+			SockErr:            s.readCM.SockErr,
 		},
 	}
 }
@@ -2795,9 +2798,49 @@ func (s *socketOpsCommon) updateTimestamp() {
 	}
 }
 
+// addrFamilyFromNetProto returns the address family identifier for the given
+// network protocol.
+func addrFamilyFromNetProto(net tcpip.NetworkProtocolNumber) int {
+	switch net {
+	case header.IPv4ProtocolNumber:
+		return linux.AF_INET
+	case header.IPv6ProtocolNumber:
+		return linux.AF_INET6
+	default:
+		panic(fmt.Sprintf("invalid net proto for addr family inference: %d", net))
+	}
+}
+
+// recvErr handles MSG_ERRQUEUE for recvmsg(2).
+// This is analogous to net/ipv4/ip_sockglue.c:ip_recv_error().
+func (s *socketOpsCommon) recvErr(t *kernel.Task, dst usermem.IOSequence) (int, int, linux.SockAddr, uint32, socket.ControlMessages, *syserr.Error) {
+	sockErr := s.Endpoint.SocketOptions().DequeueErr()
+	if sockErr == nil {
+		return 0, 0, nil, 0, socket.ControlMessages{}, syserr.ErrTryAgain
+	}
+
+	// The payload of the original packet that caused the error is passed as
+	// normal data via msg_iovec.  -- recvmsg(2)
+	msgFlags := linux.MSG_ERRQUEUE
+	if int(dst.NumBytes()) < len(sockErr.Payload) {
+		msgFlags |= linux.MSG_TRUNC
+	}
+	n, err := dst.CopyOut(t, sockErr.Payload)
+
+	// The original destination address of the datagram that caused the error is
+	// supplied via msg_name.  -- recvmsg(2)
+	dstAddr, dstAddrLen := socket.ConvertAddress(addrFamilyFromNetProto(sockErr.NetProto), sockErr.Dst)
+	cmgs := socket.ControlMessages{IP: socket.NewIPControlMessages(s.family, tcpip.ControlMessages{SockErr: sockErr})}
+	return n, msgFlags, dstAddr, dstAddrLen, cmgs, syserr.FromError(err)
+}
+
 // RecvMsg implements the linux syscall recvmsg(2) for sockets backed by
 // tcpip.Endpoint.
 func (s *socketOpsCommon) RecvMsg(t *kernel.Task, dst usermem.IOSequence, flags int, haveDeadline bool, deadline ktime.Time, senderRequested bool, controlDataLen uint64) (n int, msgFlags int, senderAddr linux.SockAddr, senderAddrLen uint32, controlMessages socket.ControlMessages, err *syserr.Error) {
+	if flags&linux.MSG_ERRQUEUE != 0 {
+		return s.recvErr(t, dst)
+	}
+
 	trunc := flags&linux.MSG_TRUNC != 0
 	peek := flags&linux.MSG_PEEK != 0
 	dontWait := flags&linux.MSG_DONTWAIT != 0