summaryrefslogtreecommitdiffstats
path: root/daemon/network.h
diff options
context:
space:
mode:
authorDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-07 15:26:00 +0000
committerDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-07 15:26:00 +0000
commit830407e88f9d40d954356c3754f2647f91d5c06a (patch)
treed6a0ece6feea91f3c656166dbaa884ef8a29740e /daemon/network.h
parentInitial commit. (diff)
downloadknot-resolver-830407e88f9d40d954356c3754f2647f91d5c06a.tar.xz
knot-resolver-830407e88f9d40d954356c3754f2647f91d5c06a.zip
Adding upstream version 5.6.0.upstream/5.6.0upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to '')
-rw-r--r--daemon/network.h162
1 files changed, 162 insertions, 0 deletions
diff --git a/daemon/network.h b/daemon/network.h
new file mode 100644
index 0000000..e21651f
--- /dev/null
+++ b/daemon/network.h
@@ -0,0 +1,162 @@
+/* Copyright (C) CZ.NIC, z.s.p.o. <knot-resolver@labs.nic.cz>
+ * SPDX-License-Identifier: GPL-3.0-or-later
+ */
+
+#pragma once
+
+#include "daemon/tls.h"
+
+#include "lib/generic/array.h"
+#include "lib/generic/trie.h"
+
+#include <uv.h>
+#include <stdbool.h>
+
+#include <sys/socket.h>
+#ifndef AF_XDP
+#define AF_XDP 44
+#endif
+
+struct engine;
+struct session;
+
+/** Ways to listen on a socket (which may exist already). */
+typedef struct {
+ int sock_type; /**< SOCK_DGRAM or SOCK_STREAM */
+ bool tls; /**< only used together with .kind == NULL and SOCK_STREAM */
+ bool http; /**< DoH2, implies .tls (in current implementation) */
+ bool xdp; /**< XDP is special (not a normal socket, in particular) */
+ bool freebind; /**< used for binding to non-local address */
+ const char *kind; /**< tag for other types: "control" or module-handled kinds */
+} endpoint_flags_t;
+
+struct endpoint_key;
+
+static inline bool endpoint_flags_eq(endpoint_flags_t f1, endpoint_flags_t f2)
+{
+ if (f1.sock_type != f2.sock_type)
+ return false;
+ if (f1.kind && f2.kind)
+ return strcasecmp(f1.kind, f2.kind);
+ else
+ return f1.tls == f2.tls && f1.kind == f2.kind;
+}
+
+/** Wrapper for a single socket to listen on.
+ * There are two types: normal have handle, special have flags.kind (and never both).
+ *
+ * LATER: .family might be unexpected for IPv4-in-IPv6 addresses.
+ * ATM AF_UNIX is only supported with flags.kind != NULL
+ */
+struct endpoint {
+ /** uv_{udp,tcp,poll}_t (poll for XDP);
+ * NULL in case of endpoints that are to be handled by modules. */
+ uv_handle_t *handle;
+ int fd; /**< POSIX file-descriptor; always used. */
+ int family; /**< AF_INET or AF_INET6 or AF_UNIX or AF_XDP */
+ uint16_t port; /**< TCP/UDP port. Meaningless with AF_UNIX. */
+ int16_t nic_queue; /**< -1 or queue number of the interface for AF_XDP use. */
+ bool engaged; /**< to some module or internally */
+ endpoint_flags_t flags;
+};
+
+/** @cond internal Array of endpoints */
+typedef array_t(struct endpoint) endpoint_array_t;
+/* @endcond */
+
+struct net_tcp_param {
+ uint64_t in_idle_timeout;
+ uint64_t tls_handshake_timeout;
+};
+
+/** Information about an address that is allowed to use PROXYv2. */
+struct net_proxy_data {
+ union kr_in_addr addr;
+ uint8_t netmask; /**< Number of bits to be matched */
+};
+
+struct network {
+ uv_loop_t *loop;
+
+ /** Map: address string -> endpoint_array_t.
+ * \note even same address-port-flags tuples may appear. */
+ trie_t *endpoints;
+
+ /** Registry of callbacks for special endpoint kinds (for opening/closing).
+ * Map: kind (lowercased) -> lua function ID converted to void *
+ * The ID is the usual: raw int index in the LUA_REGISTRYINDEX table. */
+ trie_t *endpoint_kinds;
+ /** See network_engage_endpoints() */
+ bool missing_kind_is_error : 1;
+
+ /** True: All IPv4 addresses are allowed to use the PROXYv2 protocol */
+ bool proxy_all4 : 1;
+ /** True: All IPv6 addresses are allowed to use the PROXYv2 protocol */
+ bool proxy_all6 : 1;
+
+ /** IPv4 addresses and networks allowed to use the PROXYv2 protocol */
+ trie_t *proxy_addrs4;
+ /** IPv6 addresses and networks allowed to use the PROXYv2 protocol */
+ trie_t *proxy_addrs6;
+
+ struct tls_credentials *tls_credentials;
+ tls_client_params_t *tls_client_params; /**< Use tls_client_params_*() functions. */
+ struct tls_session_ticket_ctx *tls_session_ticket_ctx;
+ struct net_tcp_param tcp;
+ int tcp_backlog;
+};
+
+void network_init(struct network *net, uv_loop_t *loop, int tcp_backlog);
+void network_deinit(struct network *net);
+
+/** Start listening on addr#port with flags.
+ * \note if we did listen on that combination already,
+ * nothing is done and kr_error(EADDRINUSE) is returned.
+ * \note there's no short-hand to listen both on UDP and TCP.
+ * \note ownership of flags.* is taken on success. TODO: non-success?
+ * \param nic_queue == -1 for auto-selection or non-XDP.
+ * \note In XDP mode, addr may be also interface name, so kr_error(ENODEV)
+ * is returned if some nonsense is passed
+ */
+int network_listen(struct network *net, const char *addr, uint16_t port,
+ int16_t nic_queue, endpoint_flags_t flags);
+
+/** Allow the specified address to send the PROXYv2 header.
+ * \note the address may be specified with a netmask
+ */
+int network_proxy_allow(struct network *net, const char* addr);
+
+/** Reset all addresses allowed to send the PROXYv2 header. No addresses will
+ * be allowed to send PROXYv2 headers from the point of calling this function
+ * until re-allowed via network_proxy_allow again.
+ */
+void network_proxy_reset(struct network *net);
+
+/** Start listening on an open file-descriptor.
+ * \note flags.sock_type isn't meaningful here.
+ * \note ownership of flags.* is taken on success. TODO: non-success?
+ */
+int network_listen_fd(struct network *net, int fd, endpoint_flags_t flags);
+
+/** Stop listening on all endpoints with matching addr#port.
+ * port < 0 serves as a wild-card.
+ * \return kr_error(ENOENT) if nothing matched. */
+int network_close(struct network *net, const char *addr, int port);
+
+/** Close all endpoints immediately (no waiting for UV loop). */
+void network_close_force(struct network *net);
+
+/** Enforce that all endpoints are registered from now on.
+ * This only does anything with struct endpoint::flags.kind != NULL. */
+int network_engage_endpoints(struct network *net);
+
+/** Returns a string representation of the specified endpoint key.
+ *
+ * The result points into key or is on static storage like for kr_straddr() */
+const char *network_endpoint_key_str(const struct endpoint_key *key);
+
+int network_set_tls_cert(struct network *net, const char *cert);
+int network_set_tls_key(struct network *net, const char *key);
+void network_new_hostname(struct network *net, struct engine *engine);
+int network_set_bpf(struct network *net, int bpf_fd);
+void network_clear_bpf(struct network *net);