af_xdp: workaround kernel race between poll() and sendmsg() - vpp

diff options

author	Benoît Ganne <bganne@cisco.com>	2021-04-29 18:24:24 +0200
committer	Damjan Marion <dmarion@me.com>	2021-05-21 19:50:14 +0000
commit	a42c41be4eed3e1ce2a42038b07ce1d3420891cd (patch)
tree	fc95c7c24cbef993cc2bef8742b3360123d70b66 /extras/vpp_config/scripts
parent	92a8d761c412590f5112239be4c511091b2b2d5a (diff)

af_xdp: workaround kernel race between poll() and sendmsg()

Prior to Linux 5.6 there is a race condition between poll() and sendmsg() in the kernel. This patch protects the syscalls with a lock to prevent it, unless the NO_SYSCALL_LOCK flag is set at create time. See https://lore.kernel.org/bpf/BYAPR11MB365382C5DB1E5FCC53242609C1549@BYAPR11MB3653.namprd11.prod.outlook.com/ Type: fix Change-Id: Ie7d4f5cb41f697b11a09b6046e54d190430d76df Signed-off-by: Benoît Ganne <bganne@cisco.com>

Diffstat (limited to 'extras/vpp_config/scripts')

0 files changed, 0 insertions, 0 deletions

/* * Copyright (c) 2015 Cisco and/or its affiliates. * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at: * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ /* Copyright (c) 2001, 2002, 2003, 2005 Eliot Dresselhaus Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ #include <stdio.h> #include <string.h> /* strchr */ #define __USE_GNU #include <sys/types.h> #include <sys/socket.h> #include <sys/un.h> #include <sys/stat.h> #include <netinet/in.h> #include <arpa/inet.h> #include <netdb.h> #include <unistd.h> #include <fcntl.h> #include <vppinfra/mem.h> #include <vppinfra/vec.h> #include <vppinfra/socket.h> #include <vppinfra/format.h> #include <vppinfra/error.h> void clib_socket_tx_add_formatted (clib_socket_t * s, char *fmt, ...) { va_list va; va_start (va, fmt); clib_socket_tx_add_va_formatted (s, fmt, &va); va_end (va); } /* Return and bind to an unused port. */ static word find_free_port (word sock) { word port; for (port = IPPORT_USERRESERVED; port < 1 << 16; port++) { struct sockaddr_in a; clib_memset (&a, 0, sizeof (a)); /* Warnings be gone */ a.sin_family = PF_INET; a.sin_addr.s_addr = INADDR_ANY; a.sin_port = htons (port); if (bind (sock, (struct sockaddr *) &a, sizeof (a)) >= 0) break; } return port < 1 << 16 ? port : -1; } /* Convert a config string to a struct sockaddr and length for use with bind or connect. */ static clib_error_t * socket_config (char *config, void *addr, socklen_t * addr_len, u32 ip4_default_address) { clib_error_t *error = 0; if (!config) config = ""; /* Anything that begins with a / is a local PF_LOCAL socket. */ if (config[0] == '/') { struct sockaddr_un *su = addr; su->sun_family = PF_LOCAL; clib_memcpy (&su->sun_path, config, clib_min (sizeof (su->sun_path), 1 + strlen (config))); *addr_len = sizeof (su[0]); } /* Hostname or hostname:port or port. */ else { char *host_name; int port = -1; struct sockaddr_in *sa = addr; host_name = 0; port = -1; if (config[0] != 0) { unformat_input_t i; unformat_init_string (&i, config, strlen (config)); if (unformat (&i, "%s:%d", &host_name, &port) || unformat (&i, "%s:0x%x", &host_name, &port)) ; else if (unformat (&i, "%s", &host_name)) ; else error = clib_error_return (0, "unknown input `%U'", format_unformat_error, &i); unformat_free (&i); if (error) goto done; } sa->sin_family = PF_INET; *addr_len = sizeof (sa[0]); if (port != -1) sa->sin_port = htons (port); else sa->sin_port = 0; if (host_name) { struct in_addr host_addr; /* Recognize localhost to avoid host lookup in most common cast. */ if (!strcmp (host_name, "localhost")) sa->sin_addr.s_addr = htonl (INADDR_LOOPBACK); else if (inet_aton (host_name, &host_addr)) sa->sin_addr = host_addr; else if (host_name && strlen (host_name) > 0) { struct hostent *host = gethostbyname (host_name); if (!host) error = clib_error_return (0, "unknown host `%s'", config); else clib_memcpy (&sa->sin_addr.s_addr, host->h_addr_list[0], host->h_length); } else sa->sin_addr.s_addr = htonl (ip4_default_address); vec_free (host_name); if (error) goto done; } } done: return error; } static clib_error_t * default_socket_write (clib_socket_t * s) { clib_error_t *err = 0; word written = 0; word fd = 0; word tx_len; fd = s->fd; /* Map standard input to standard output. Typically, fd is a socket for which read/write both work. */ if (fd == 0) fd = 1; tx_len = vec_len (s->tx_buffer); written = write (fd, s->tx_buffer, tx_len); /* Ignore certain errors. */ if (written < 0 && !unix_error_is_fatal (errno)) written = 0; /* A "real" error occurred. */ if (written < 0) { err = clib_error_return_unix (0, "write %wd bytes (fd %d, '%s')", tx_len, s->fd, s->config); vec_free (s->tx_buffer); goto done; } /* Reclaim the transmitted part of the tx buffer on successful writes. */ else if (written > 0) { if (written == tx_len) _vec_len (s->tx_buffer) = 0; else vec_delete (s->tx_buffer, written, 0); } /* If a non-fatal error occurred AND the buffer is full, then we must free it. */ else if (written == 0 && tx_len > 64 * 1024) { vec_free (s->tx_buffer); } done: return err; } static clib_error_t * default_socket_read (clib_socket_t * sock, int n_bytes) { word fd, n_read; u8 *buf; /* RX side of socket is down once end of file is reached. */ if (sock->flags & CLIB_SOCKET_F_RX_END_OF_FILE) return 0; fd = sock->fd; n_bytes = clib_max (n_bytes, 4096); vec_add2 (sock->rx_buffer, buf, n_bytes); if ((n_read = read (fd, buf, n_bytes)) < 0) { n_read = 0; /* Ignore certain errors. */ if (!unix_error_is_fatal (errno)) goto non_fatal; return clib_error_return_unix (0, "read %d bytes (fd %d, '%s')", n_bytes, sock->fd, sock->config); } /* Other side closed the socket. */ if (n_read == 0) sock->flags |= CLIB_SOCKET_F_RX_END_OF_FILE; non_fatal: _vec_len (sock->rx_buffer) += n_read - n_bytes; return 0; } static clib_error_t * default_socket_close (clib_socket_t * s) { if (close (s->fd) < 0) return clib_error_return_unix (0, "close (fd %d, %s)", s->fd, s->config); return 0; } static clib_error_t * default_socket_sendmsg (clib_socket_t * s, void *msg, int msglen, int fds[], int num_fds) { struct msghdr mh = { 0 }; struct iovec iov[1]; char ctl[CMSG_SPACE (sizeof (int) * num_fds)]; int rv; iov[0].iov_base = msg; iov[0].iov_len = msglen; mh.msg_iov = iov; mh.msg_iovlen = 1; if (num_fds > 0) { struct cmsghdr *cmsg; clib_memset (&ctl, 0, sizeof (ctl)); mh.msg_control = ctl; mh.msg_controllen = sizeof (ctl); cmsg = CMSG_FIRSTHDR (&mh); cmsg->cmsg_len = CMSG_LEN (sizeof (int) * num_fds); cmsg->cmsg_level = SOL_SOCKET; cmsg->cmsg_type = SCM_RIGHTS; memcpy (CMSG_DATA (cmsg), fds, sizeof (int) * num_fds); } rv = sendmsg (s->fd, &mh, 0); if (rv < 0) return clib_error_return_unix (0, "sendmsg"); return 0; } static clib_error_t * default_socket_recvmsg (clib_socket_t * s, void *msg, int msglen, int fds[], int num_fds) { #ifdef __linux__ char ctl[CMSG_SPACE (sizeof (int) * num_fds) + CMSG_SPACE (sizeof (struct ucred))]; struct ucred *cr = 0; #else char ctl[CMSG_SPACE (sizeof (int) * num_fds)]; #endif struct msghdr mh = { 0 }; struct iovec iov[1]; ssize_t size; struct cmsghdr *cmsg; iov[0].iov_base = msg; iov[0].iov_len = msglen; mh.msg_iov = iov; mh.msg_iovlen = 1; mh.msg_control = ctl; mh.msg_controllen = sizeof (ctl); clib_memset (ctl, 0, sizeof (ctl)); /* receive the incoming message */ size = recvmsg (s->fd, &mh, 0); if (size != msglen) { return (size == 0) ? clib_error_return (0, "disconnected") : clib_error_return_unix (0, "recvmsg: malformed message (fd %d, '%s')", s->fd, s->config); } cmsg = CMSG_FIRSTHDR (&mh); while (cmsg) { if (cmsg->cmsg_level == SOL_SOCKET) { #ifdef __linux__ if (cmsg->cmsg_type == SCM_CREDENTIALS) { cr = (struct ucred *) CMSG_DATA (cmsg); s->uid = cr->uid; s->gid = cr->gid; s->pid = cr->pid; } else #endif if (cmsg->cmsg_type == SCM_RIGHTS) { clib_memcpy_fast (fds, CMSG_DATA (cmsg), num_fds * sizeof (int)); } } cmsg = CMSG_NXTHDR (&mh, cmsg); } return 0; } static void socket_init_funcs (clib_socket_t * s) { if (!s->write_func) s->write_func = default_socket_write; if (!s->read_func) s->read_func = default_socket_read; if (!s->close_func) s->close_func = default_socket_close; if (!s->sendmsg_func) s->sendmsg_func = default_socket_sendmsg; if (!s->recvmsg_func) s->recvmsg_func = default_socket_recvmsg; } clib_error_t * clib_socket_init (clib_socket_t * s) { union { struct sockaddr sa; struct sockaddr_un su; } addr; socklen_t addr_len = 0; int socket_type; clib_error_t *error = 0; word port; error = socket_config (s->config, &addr.sa, &addr_len, (s->flags & CLIB_SOCKET_F_IS_SERVER ? INADDR_LOOPBACK : INADDR_ANY)); if (error) goto done; socket_init_funcs (s); socket_type = s->flags & CLIB_SOCKET_F_SEQPACKET ? SOCK_SEQPACKET : SOCK_STREAM; s->fd = socket (addr.sa.sa_family, socket_type, 0); if (s->fd < 0) { error = clib_error_return_unix (0, "socket (fd %d, '%s')", s->fd, s->config); goto done; } port = 0; if (addr.sa.sa_family == PF_INET) port = ((struct sockaddr_in *) &addr)->sin_port; if (s->flags & CLIB_SOCKET_F_IS_SERVER) { uword need_bind = 1; if (addr.sa.sa_family == PF_INET) { if (port == 0) { port = find_free_port (s->fd); if (port < 0) { error = clib_error_return (0, "no free port (fd %d, '%s')", s->fd, s->config); goto done; } need_bind = 0; } } if (addr.sa.sa_family == PF_LOCAL) unlink (((struct sockaddr_un *) &addr)->sun_path); /* Make address available for multiple users. */ { int v = 1; if (setsockopt (s->fd, SOL_SOCKET, SO_REUSEADDR, &v, sizeof (v)) < 0) clib_unix_warning ("setsockopt SO_REUSEADDR fails"); } #if __linux__ if (addr.sa.sa_family == PF_LOCAL && s->flags & CLIB_SOCKET_F_PASSCRED) { int x = 1; if (setsockopt (s->fd, SOL_SOCKET, SO_PASSCRED, &x, sizeof (x)) < 0) { error = clib_error_return_unix (0, "setsockopt (SO_PASSCRED, " "fd %d, '%s')", s->fd, s->config); goto done; } } #endif if (need_bind && bind (s->fd, &addr.sa, addr_len) < 0) { error = clib_error_return_unix (0, "bind (fd %d, '%s')", s->fd, s->config); goto done; } if (listen (s->fd, 5) < 0) { error = clib_error_return_unix (0, "listen (fd %d, '%s')", s->fd, s->config); goto done; } if (addr.sa.sa_family == PF_LOCAL && s->flags & CLIB_SOCKET_F_ALLOW_GROUP_WRITE) { struct stat st = { 0 }; if (stat (((struct sockaddr_un *) &addr)->sun_path, &st) < 0) { error = clib_error_return_unix (0, "stat (fd %d, '%s')", s->fd, s->config); goto done; } st.st_mode |= S_IWGRP; if (chmod (((struct sockaddr_un *) &addr)->sun_path, st.st_mode) < 0) { error = clib_error_return_unix (0, "chmod (fd %d, '%s', mode %o)", s->fd, s->config, st.st_mode); goto done; } } } else { if ((s->flags & CLIB_SOCKET_F_NON_BLOCKING_CONNECT) && fcntl (s->fd, F_SETFL, O_NONBLOCK) < 0) { error = clib_error_return_unix (0, "fcntl NONBLOCK (fd %d, '%s')", s->fd, s->config); goto done; } if (connect (s->fd, &addr.sa, addr_len) < 0 && !((s->flags & CLIB_SOCKET_F_NON_BLOCKING_CONNECT) && errno == EINPROGRESS)) { error = clib_error_return_unix (0, "connect (fd %d, '%s')", s->fd, s->config); goto done; } } return error; done: if (s->fd > 0) close (s->fd); return error; } clib_error_t * clib_socket_accept (clib_socket_t * server, clib_socket_t * client) { clib_error_t *err = 0; socklen_t len = 0; clib_memset (client, 0, sizeof (client[0])); /* Accept the new socket connection. */ client->fd = accept (server->fd, 0, 0); if (client->fd < 0) return clib_error_return_unix (0, "accept (fd %d, '%s')", server->fd, server->config); /* Set the new socket to be non-blocking. */ if (fcntl (client->fd, F_SETFL, O_NONBLOCK) < 0) { err = clib_error_return_unix (0, "fcntl O_NONBLOCK (fd %d)", client->fd); goto close_client; } /* Get peer info. */ len = sizeof (client->peer); if (getpeername (client->fd, (struct sockaddr *) &client->peer, &len) < 0) { err = clib_error_return_unix (0, "getpeername (fd %d)", client->fd); goto close_client; } client->flags = CLIB_SOCKET_F_IS_CLIENT; socket_init_funcs (client); return 0; close_client: close (client->fd); return err; } /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */


context:
space:
mode: