@@ -485,7 +485,6 @@ dco_p2p_add_new_peer(struct context *c)
}
c->c2.tls_multi->dco_peer_id = multi->peer_id;
- c->c2.link_socket->dco_installed = true;
return 0;
}
@@ -605,17 +604,6 @@ dco_multi_add_new_peer(struct multi_context *m, struct multi_instance *mi)
c->c2.tls_multi->dco_peer_id = peer_id;
- if (c->mode == CM_CHILD_TCP)
- {
- multi_tcp_dereference_instance(m->mtcp, mi);
- if (close(sd))
- {
- msg(D_DCO|M_ERRNO, "error closing TCP socket after DCO handover");
- }
- c->c2.link_socket->dco_installed = true;
- c->c2.link_socket->sd = SOCKET_UNDEFINED;
- }
-
return 0;
}
@@ -127,15 +127,6 @@ void close_tun_dco(struct tuntap *tt, openvpn_net_ctx_t *ctx);
*/
int dco_do_read(dco_context_t *dco);
-/**
- * Write data to the DCO communication channel (control packet expected)
- *
- * @param dco the DCO context
- * @param peer_id the ID of the peer to send the data to
- * @param buf the buffer containing the data to send
- */
-int dco_do_write(dco_context_t *dco, int peer_id, struct buffer *buf);
-
/**
* Install a DCO in the main event loop
*/
@@ -301,13 +292,6 @@ dco_do_read(dco_context_t *dco)
return 0;
}
-static inline int
-dco_do_write(dco_context_t *dco, int peer_id, struct buffer *buf)
-{
- ASSERT(false);
- return 0;
-}
-
static inline void
dco_event_set(dco_context_t *dco, struct event_set *es, void *arg)
{
@@ -142,7 +142,6 @@ open_fd(dco_context_t *dco)
{
dco->open = true;
}
- dco->dco_packet_in = alloc_buf(PAGE_SIZE);
return dco->fd;
}
@@ -560,15 +559,6 @@ dco_do_read(dco_context_t *dco)
return 0;
}
-int
-dco_do_write(dco_context_t *dco, int peer_id, struct buffer *buf)
-{
- /* Control packets are passed through the socket, so this should never get
- * called. See should_use_dco_socket(). */
- ASSERT(0);
- return -EINVAL;
-}
-
bool
dco_available(int msglevel)
{
@@ -51,8 +51,6 @@ typedef struct dco_context {
char ifname[IFNAMSIZ];
- struct buffer dco_packet_in;
-
int dco_message_type;
int dco_message_peer_id;
int dco_del_peer_reason;
@@ -434,24 +434,6 @@ ovpn_dco_register(dco_context_t *dco)
{
msg(M_ERR, "%s: failed to join groups: %d", __func__, ret);
}
-
- /* Register for non-data packets that ovpn-dco may receive. They will be
- * forwarded to userspace
- */
- struct nl_msg *nl_msg = ovpn_dco_nlmsg_create(dco, OVPN_CMD_REGISTER_PACKET);
- if (!nl_msg)
- {
- msg(M_ERR, "%s: cannot allocate message to register for control packets",
- __func__);
- }
-
- ret = ovpn_nl_msg_send(dco, nl_msg, NULL, __func__);
- if (ret)
- {
- msg(M_ERR, "%s: failed to register for control packets: %d", __func__,
- ret);
- }
- nlmsg_free(nl_msg);
}
int
@@ -474,8 +456,6 @@ open_tun_dco(struct tuntap *tt, openvpn_net_ctx_t *ctx, const char *dev)
}
tt->actual_name = string_alloc(dev, NULL);
- uint8_t *dcobuf = malloc(65536);
- buf_set_write(&tt->dco.dco_packet_in, dcobuf, 65536);
tt->dco.dco_message_peer_id = -1;
ovpn_dco_register(&tt->dco);
@@ -490,7 +470,6 @@ close_tun_dco(struct tuntap *tt, openvpn_net_ctx_t *ctx)
net_iface_del(ctx, tt->actual_name);
ovpn_dco_uninit_netlink(&tt->dco);
- free(tt->dco.dco_packet_in.data);
}
int
@@ -821,51 +800,6 @@ ovpn_handle_msg(struct nl_msg *msg, void *arg)
break;
}
- case OVPN_CMD_PACKET:
- {
- if (!attrs[OVPN_ATTR_PACKET])
- {
- msg(D_DCO, "ovpn-dco: no packet in OVPN_CMD_PACKET message");
- return NL_SKIP;
- }
- struct nlattr *pkt_attrs[OVPN_PACKET_ATTR_MAX + 1];
-
- if (nla_parse_nested(pkt_attrs, OVPN_PACKET_ATTR_MAX,
- attrs[OVPN_ATTR_PACKET], NULL))
- {
- msg(D_DCO, "received bogus cmd packet data from ovpn-dco");
- return NL_SKIP;
- }
- if (!pkt_attrs[OVPN_PACKET_ATTR_PEER_ID])
- {
- msg(D_DCO, "ovpn-dco: Received OVPN_CMD_PACKET message without peer id");
- return NL_SKIP;
- }
- if (!pkt_attrs[OVPN_PACKET_ATTR_PACKET])
- {
- msg(D_DCO, "ovpn-dco: Received OVPN_CMD_PACKET message without packet");
- return NL_SKIP;
- }
-
- unsigned int peerid = nla_get_u32(pkt_attrs[OVPN_PACKET_ATTR_PEER_ID]);
-
- uint8_t *data = nla_data(pkt_attrs[OVPN_PACKET_ATTR_PACKET]);
- int len = nla_len(pkt_attrs[OVPN_PACKET_ATTR_PACKET]);
-
- msg(D_DCO_DEBUG, "ovpn-dco: received OVPN_PACKET_ATTR_PACKET, ifindex: %d peer-id: %d, len %d",
- ifindex, peerid, len);
- if (BLEN(&dco->dco_packet_in) > 0)
- {
- msg(D_DCO, "DCO packet buffer still full?!");
- return NL_SKIP;
- }
- buf_init(&dco->dco_packet_in, 0);
- buf_write(&dco->dco_packet_in, data, len);
- dco->dco_message_peer_id = peerid;
- dco->dco_message_type = OVPN_CMD_PACKET;
- break;
- }
-
default:
msg(D_DCO, "ovpn-dco: received unknown command: %d", gnlh->cmd);
dco->dco_message_type = 0;
@@ -884,41 +818,6 @@ dco_do_read(dco_context_t *dco)
return ovpn_nl_recvmsgs(dco, __func__);
}
-int
-dco_do_write(dco_context_t *dco, int peer_id, struct buffer *buf)
-{
- packet_size_type len = BLEN(buf);
- dmsg(D_STREAM_DEBUG, "DCO: WRITE %d offset=%d", (int)len, buf->offset);
-
- msg(D_DCO_DEBUG, "%s: peer-id %d, len=%d", __func__, peer_id, len);
-
- struct nl_msg *nl_msg = ovpn_dco_nlmsg_create(dco, OVPN_CMD_PACKET);
-
- if (!nl_msg)
- {
- return -ENOMEM;
- }
-
- struct nlattr *attr = nla_nest_start(nl_msg, OVPN_ATTR_PACKET);
- int ret = -EMSGSIZE;
- NLA_PUT_U32(nl_msg, OVPN_PACKET_ATTR_PEER_ID, peer_id);
- NLA_PUT(nl_msg, OVPN_PACKET_ATTR_PACKET, len, BSTR(buf));
- nla_nest_end(nl_msg, attr);
-
- ret = ovpn_nl_msg_send(dco, nl_msg, NULL, __func__);
- if (ret)
- {
- goto nla_put_failure;
- }
-
- /* return the length of the written data in case of success */
- ret = len;
-
-nla_put_failure:
- nlmsg_free(nl_msg);
- return ret;
-}
-
int
dco_get_peer_stats_multi(dco_context_t *dco, struct multi_context *m)
{
@@ -48,8 +48,6 @@ typedef struct
unsigned int ifindex;
- struct buffer dco_packet_in;
-
int dco_message_type;
int dco_message_peer_id;
int dco_del_peer_reason;
@@ -393,14 +393,6 @@ dco_do_read(dco_context_t *dco)
return 0;
}
-int
-dco_do_write(dco_context_t *dco, int peer_id, struct buffer *buf)
-{
- /* no-op on windows */
- ASSERT(0);
- return 0;
-}
-
int
dco_get_peer_stats_multi(dco_context_t *dco, struct multi_context *m)
{
@@ -1191,7 +1191,6 @@ static void
process_incoming_dco(struct context *c)
{
#if defined(ENABLE_DCO) && (defined(TARGET_LINUX) || defined(TARGET_FREEBSD))
- struct link_socket_info *lsi = get_link_socket_info(c);
dco_context_t *dco = &c->c1.tuntap->dco;
dco_do_read(dco);
@@ -1204,35 +1203,23 @@ process_incoming_dco(struct context *c)
msg(D_DCO_DEBUG, "%s: received message for mismatching peer-id %d, "
"expected %d", __func__, dco->dco_message_peer_id,
c->c2.tls_multi->dco_peer_id);
- /* ensure we also drop a message if there is one in the buffer */
- buf_init(&dco->dco_packet_in, 0);
return;
}
- if ((dco->dco_message_type == OVPN_CMD_DEL_PEER)
- && (dco->dco_del_peer_reason == OVPN_DEL_PEER_REASON_EXPIRED))
+ if (dco->dco_message_type != OVPN_CMD_DEL_PEER)
{
- msg(D_DCO_DEBUG, "%s: received peer expired notification of for peer-id "
- "%d", __func__, dco->dco_message_peer_id);
- trigger_ping_timeout_signal(c);
+ msg(D_DCO_DEBUG, "%s: received message of type %u - ignoring", __func__,
+ dco->dco_message_type);
return;
}
- if (dco->dco_message_type != OVPN_CMD_PACKET)
+ if (dco->dco_del_peer_reason == OVPN_DEL_PEER_REASON_EXPIRED)
{
- msg(D_DCO_DEBUG, "%s: received message of type %u - ignoring", __func__,
- dco->dco_message_type);
+ msg(D_DCO_DEBUG, "%s: received peer expired notification of for peer-id "
+ "%d", __func__, dco->dco_message_peer_id);
+ trigger_ping_timeout_signal(c);
return;
}
-
- struct buffer orig_buff = c->c2.buf;
- c->c2.buf = dco->dco_packet_in;
- c->c2.from = lsi->lsa->actual;
-
- process_incoming_link(c);
-
- c->c2.buf = orig_buff;
- buf_init(&dco->dco_packet_in, 0);
#endif /* if defined(ENABLE_DCO) && (defined(TARGET_LINUX) || defined(TARGET_FREEBSD)) */
}
@@ -1686,30 +1673,6 @@ process_ip_header(struct context *c, unsigned int flags, struct buffer *buf)
}
}
-/*
- * Linux DCO implementations pass the socket to the kernel and
- * disallow usage of it from userland for TCP, so (control) packets
- * sent and received by OpenVPN need to go through the DCO interface.
- *
- * Windows DCO needs control packets to be sent via the normal
- * standard Overlapped I/O.
- *
- * FreeBSD DCO allows control packets to pass through the socket in both
- * directions.
- *
- * Hide that complexity (...especially if more platforms show up
- * in the future...) in a small inline function.
- */
-static inline bool
-should_use_dco_socket(struct link_socket *ls)
-{
-#if defined(TARGET_LINUX)
- return ls->dco_installed && proto_is_tcp(ls->info.proto);
-#else
- return false;
-#endif
-}
-
/*
* Input: c->c2.to_link
*/
@@ -1783,17 +1746,7 @@ process_outgoing_link(struct context *c)
socks_preprocess_outgoing_link(c, &to_addr, &size_delta);
/* Send packet */
- if (should_use_dco_socket(c->c2.link_socket))
- {
- size = dco_do_write(&c->c1.tuntap->dco,
- c->c2.tls_multi->dco_peer_id,
- &c->c2.to_link);
- }
- else
- {
- size = link_socket_write(c->c2.link_socket, &c->c2.to_link,
- to_addr);
- }
+ size = link_socket_write(c->c2.link_socket, &c->c2.to_link, to_addr);
/* Undo effect of prepend */
link_socket_write_post_size_adjust(&size, size_delta, &c->c2.to_link);
@@ -3914,8 +3914,7 @@ do_close_link_socket(struct context *c)
/* in dco-win case, link socket is a tun handle which is
* closed in do_close_tun(). Set it to UNDEFINED so
* we won't use WinSock API to close it. */
- if (tuntap_is_dco_win(c->c1.tuntap) && c->c2.link_socket
- && c->c2.link_socket->dco_installed)
+ if (tuntap_is_dco_win(c->c1.tuntap) && c->c2.link_socket)
{
c->c2.link_socket->sd = SOCKET_UNDEFINED;
}
@@ -402,18 +402,6 @@ multi_tcp_wait_lite(struct multi_context *m, struct multi_instance *mi, const in
tv_clear(&c->c2.timeval); /* ZERO-TIMEOUT */
- if (mi && mi->context.c2.link_socket->dco_installed)
- {
- /* If we got a socket that has been handed over to the kernel
- * we must not call the normal socket function to figure out
- * if it is readable or writable */
- /* Assert that we only have the DCO exptected flags */
- ASSERT(action & (TA_SOCKET_READ | TA_SOCKET_WRITE));
-
- /* We are always ready! */
- return action;
- }
-
switch (action)
{
case TA_TUN_READ:
@@ -537,10 +525,7 @@ multi_tcp_dispatch(struct multi_context *m, struct multi_instance *mi, const int
case TA_INITIAL:
ASSERT(mi);
- if (!mi->context.c2.link_socket->dco_installed)
- {
- multi_tcp_set_global_rw_flags(m, mi);
- }
+ multi_tcp_set_global_rw_flags(m, mi);
multi_process_post(m, mi, mpp_flags);
break;
@@ -590,10 +575,7 @@ multi_tcp_post(struct multi_context *m, struct multi_instance *mi, const int act
}
else
{
- if (!c->c2.link_socket->dco_installed)
- {
- multi_tcp_set_global_rw_flags(m, mi);
- }
+ multi_tcp_set_global_rw_flags(m, mi);
}
break;
@@ -3202,37 +3202,6 @@ multi_signal_instance(struct multi_context *m, struct multi_instance *mi, const
#endif
#if defined(ENABLE_DCO) && (defined(TARGET_LINUX) || defined(TARGET_FREEBSD))
-static void
-process_incoming_dco_packet(struct multi_context *m, struct multi_instance *mi,
- dco_context_t *dco)
-{
- if (BLEN(&dco->dco_packet_in) < 1)
- {
- msg(D_DCO, "Received too short packet for peer %d",
- dco->dco_message_peer_id);
- goto done;
- }
-
- uint8_t *ptr = BPTR(&dco->dco_packet_in);
- uint8_t op = ptr[0] >> P_OPCODE_SHIFT;
- if ((op == P_DATA_V1) || (op == P_DATA_V2))
- {
- msg(D_DCO, "DCO: received data channel packet for peer %d",
- dco->dco_message_peer_id);
- goto done;
- }
-
- struct buffer orig_buf = mi->context.c2.buf;
- mi->context.c2.buf = dco->dco_packet_in;
-
- multi_process_incoming_link(m, mi, 0);
-
- mi->context.c2.buf = orig_buf;
-
-done:
- buf_init(&dco->dco_packet_in, 0);
-}
-
static void
process_incoming_del_peer(struct multi_context *m, struct multi_instance *mi,
dco_context_t *dco)
@@ -3299,11 +3268,7 @@ multi_process_incoming_dco(struct multi_context *m)
if ((peer_id < m->max_clients) && (m->instances[peer_id]))
{
mi = m->instances[peer_id];
- if (dco->dco_message_type == OVPN_CMD_PACKET)
- {
- process_incoming_dco_packet(m, mi, dco);
- }
- else if (dco->dco_message_type == OVPN_CMD_DEL_PEER)
+ if (dco->dco_message_type == OVPN_CMD_DEL_PEER)
{
process_incoming_del_peer(m, mi, dco);
}
@@ -3326,8 +3291,6 @@ multi_process_incoming_dco(struct multi_context *m)
msg(msglevel, "Received DCO message for unknown peer-id: %d, "
"type %d, del_peer_reason %d", peer_id, dco->dco_message_type,
dco->dco_del_peer_reason);
- /* Also clear the buffer if this was incoming packet for a dropped peer */
- buf_init(&dco->dco_packet_in, 0);
}
dco->dco_message_type = 0;
@@ -11,7 +11,7 @@
#ifndef _UAPI_LINUX_OVPN_DCO_H_
#define _UAPI_LINUX_OVPN_DCO_H_
-#define OVPN_NL_NAME "ovpn-dco"
+#define OVPN_NL_NAME "ovpn-dco-v2"
#define OVPN_NL_MULTICAST_GROUP_PEERS "peers"
@@ -45,19 +45,6 @@ enum ovpn_nl_commands {
OVPN_CMD_DEL_KEY,
- /**
- * @OVPN_CMD_REGISTER_PACKET: Register for specific packet types to be
- * forwarded to userspace
- */
- OVPN_CMD_REGISTER_PACKET,
-
- /**
- * @OVPN_CMD_PACKET: Send a packet from userspace to kernelspace. Also
- * used to send to userspace packets for which a process had registered
- * with OVPN_CMD_REGISTER_PACKET
- */
- OVPN_CMD_PACKET,
-
/**
* @OVPN_CMD_GET_PEER: Retrieve the status of a peer or all peers
*/
@@ -105,7 +92,6 @@ enum ovpn_netlink_attrs {
OVPN_ATTR_NEW_KEY,
OVPN_ATTR_SWAP_KEYS,
OVPN_ATTR_DEL_KEY,
- OVPN_ATTR_PACKET,
OVPN_ATTR_GET_PEER,
__OVPN_ATTR_AFTER_LAST,
@@ -2151,7 +2151,7 @@ create_socket_dco_win(struct context *c, struct link_socket *sock,
get_server_poll_remaining_time(sock->server_poll_timeout),
sig_info);
- sock->dco_installed = true;
+ sock->sockflags |= SF_DCO_WIN;
if (sig_info->signal_received)
{
@@ -3505,7 +3505,7 @@ link_socket_write_udp_posix_sendmsg(struct link_socket *sock,
static int
socket_get_last_error(const struct link_socket *sock)
{
- if (sock->dco_installed)
+ if (socket_is_dco_win(sock))
{
return GetLastError();
}
@@ -3546,7 +3546,7 @@ socket_recv_queue(struct link_socket *sock, int maxsize)
ASSERT(ResetEvent(sock->reads.overlapped.hEvent));
sock->reads.flags = 0;
- if (sock->dco_installed)
+ if (socket_is_dco_win(sock))
{
status = ReadFile((HANDLE)sock->sd, wsabuf[0].buf, wsabuf[0].len,
&sock->reads.size, &sock->reads.overlapped);
@@ -3651,7 +3651,7 @@ socket_send_queue(struct link_socket *sock, struct buffer *buf, const struct lin
ASSERT(ResetEvent(sock->writes.overlapped.hEvent));
sock->writes.flags = 0;
- if (sock->dco_installed)
+ if (socket_is_dco_win(sock))
{
status = WriteFile((HANDLE)sock->sd, wsabuf[0].buf, wsabuf[0].len,
&sock->writes.size, &sock->writes.overlapped);
@@ -168,7 +168,6 @@ struct link_socket
socket_descriptor_t sd;
socket_descriptor_t ctrl_sd; /* only used for UDP over Socks */
- bool dco_installed;
#ifdef _WIN32
struct overlapped_io reads;
@@ -207,6 +206,7 @@ struct link_socket
#define SF_PORT_SHARE (1<<2)
#define SF_HOST_RANDOMIZE (1<<3)
#define SF_GETADDRINFO_DGRAM (1<<4)
+#define SF_DCO_WIN (1<<5)
unsigned int sockflags;
int mark;
const char *bind_dev;
@@ -1021,6 +1021,17 @@ stream_buf_read_setup(struct link_socket *sock)
}
}
+/**
+ * Returns true if we are on Windows and this link is running on DCO-WIN.
+ * This helper is used to enable DCO-WIN specific logic that is not relevant
+ * to other platforms.
+ */
+static inline bool
+socket_is_dco_win(const struct link_socket *s)
+{
+ return s->sockflags & SF_DCO_WIN;
+}
+
/*
* Socket Read Routines
*/
@@ -1036,7 +1047,7 @@ link_socket_read_udp_win32(struct link_socket *sock,
struct link_socket_actual *from)
{
sockethandle_t sh = { .s = sock->sd };
- if (sock->dco_installed)
+ if (socket_is_dco_win(sock))
{
*from = sock->info.lsa->actual;
sh.is_handle = true;
@@ -1058,12 +1069,8 @@ link_socket_read(struct link_socket *sock,
struct buffer *buf,
struct link_socket_actual *from)
{
-#ifdef _WIN32
- if (proto_is_udp(sock->info.proto) || sock->dco_installed)
-#else
- if (proto_is_udp(sock->info.proto))
-#endif
- /* unified UDPv4 and UDPv6, for DCO the kernel
+ if (proto_is_udp(sock->info.proto) || socket_is_dco_win(sock))
+ /* unified UDPv4 and UDPv6, for DCO-WIN the kernel
* will strip the length header */
{
int res;
@@ -1105,7 +1112,7 @@ link_socket_write_win32(struct link_socket *sock,
{
int err = 0;
int status = 0;
- sockethandle_t sh = { .s = sock->sd, .is_handle = sock->dco_installed };
+ sockethandle_t sh = { .s = sock->sd, .is_handle = socket_is_dco_win(sock) };
if (overlapped_io_active(&sock->writes))
{
status = sockethandle_finalize(sh, &sock->writes, NULL, NULL);
@@ -1179,9 +1186,9 @@ link_socket_write(struct link_socket *sock,
struct buffer *buf,
struct link_socket_actual *to)
{
- if (proto_is_udp(sock->info.proto) || sock->dco_installed)
+ if (proto_is_udp(sock->info.proto) || socket_is_dco_win(sock))
{
- /* unified UDPv4 and UDPv6 and DCO (kernel adds size header) */
+ /* unified UDPv4, UDPv6 and DCO-WIN (driver adds length header) */
return link_socket_write_udp(sock, buf, to);
}
else if (proto_is_tcp(sock->info.proto)) /* unified TCPv4 and TCPv6 */
Using NetLink for control messages did not work out as it did lead to kernel side buffer congestion during heavy client activity. With this patch DCO will redirect control packets directly to the transport socket without altering them, so that userspace can happily process them as usual. Change-Id: Ia1297c3ae9a28b188ed21ad21ae96fff3d02ee4d [lev@openvpn.net: ensure win_dco flag is still exposed] Signed-off-by: Antonio Quartulli <a@unstable.cc> --- Changes from v1: * improved comments * improved commit message This patch was also reviewed and approved on gerrit at: https://gerrit.openvpn.net/c/openvpn/+/28 src/openvpn/dco.c | 12 ----- src/openvpn/dco.h | 16 ------ src/openvpn/dco_freebsd.c | 10 ---- src/openvpn/dco_freebsd.h | 2 - src/openvpn/dco_linux.c | 101 ----------------------------------- src/openvpn/dco_linux.h | 2 - src/openvpn/dco_win.c | 8 --- src/openvpn/forward.c | 63 +++------------------- src/openvpn/init.c | 3 +- src/openvpn/mtcp.c | 22 +------- src/openvpn/multi.c | 39 +------------- src/openvpn/ovpn_dco_linux.h | 16 +----- src/openvpn/socket.c | 8 +-- src/openvpn/socket.h | 29 ++++++---- 14 files changed, 35 insertions(+), 296 deletions(-)