/* $NetBSD: clientloop.c,v 1.26.2.1 2023/12/25 12:31:03 martin Exp $ */ /* $OpenBSD: clientloop.c,v 1.402 2023/11/24 00:31:30 dtucker Exp $ */ /* * Author: Tatu Ylonen * Copyright (c) 1995 Tatu Ylonen , Espoo, Finland * All rights reserved * The main loop for the interactive session (client side). * * As far as I am concerned, the code I have written for this software * can be used freely for any purpose. Any derived versions of this * software must be clearly marked as such, and if the derived work is * incompatible with the protocol description in the RFC file, it must be * called by a name other than "ssh" or "Secure Shell". * * * Copyright (c) 1999 Theo de Raadt. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * * SSH2 support added by Markus Friedl. * Copyright (c) 1999, 2000, 2001 Markus Friedl. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "includes.h" __RCSID("$NetBSD: clientloop.c,v 1.26.2.1 2023/12/25 12:31:03 martin Exp $"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "xmalloc.h" #include "ssh.h" #include "ssh2.h" #include "packet.h" #include "sshbuf.h" #include "compat.h" #include "channels.h" #include "dispatch.h" #include "sshkey.h" #include "cipher.h" #include "kex.h" #include "myproposal.h" #include "log.h" #include "misc.h" #include "readconf.h" #include "clientloop.h" #include "sshconnect.h" #include "authfd.h" #include "atomicio.h" #include "sshpty.h" #include "match.h" #include "msg.h" #include "getpeereid.h" #include "ssherr.h" #include "hostfile.h" /* Permitted RSA signature algorithms for UpdateHostkeys proofs */ #define HOSTKEY_PROOF_RSA_ALGS "rsa-sha2-512,rsa-sha2-256" /* Uncertainty (in percent) of keystroke timing intervals */ #define SSH_KEYSTROKE_TIMING_FUZZ 10 /* import options */ extern Options options; /* Control socket */ extern int muxserver_sock; /* XXX use mux_client_cleanup() instead */ /* * Name of the host we are connecting to. This is the name given on the * command line, or the Hostname specified for the user-supplied name in a * configuration file. */ extern char *host; /* * If this field is not NULL, the ForwardAgent socket is this path and different * instead of SSH_AUTH_SOCK. */ extern char *forward_agent_sock_path; /* * Flag to indicate that we have received a window change signal which has * not yet been processed. This will cause a message indicating the new * window size to be sent to the server a little later. This is volatile * because this is updated in a signal handler. */ static volatile sig_atomic_t received_window_change_signal = 0; static volatile sig_atomic_t received_signal = 0; /* Time when backgrounded control master using ControlPersist should exit */ static time_t control_persist_exit_time = 0; /* Common data for the client loop code. */ volatile sig_atomic_t quit_pending; /* Set non-zero to quit the loop. */ static int last_was_cr; /* Last character was a newline. */ static int exit_status; /* Used to store the command exit status. */ static struct sshbuf *stderr_buffer; /* Used for final exit message. */ static int connection_in; /* Connection to server (input). */ static int connection_out; /* Connection to server (output). */ static int need_rekeying; /* Set to non-zero if rekeying is requested. */ static int session_closed; /* In SSH2: login session closed. */ static time_t x11_refuse_time; /* If >0, refuse x11 opens after this time. */ static time_t server_alive_time; /* Time to do server_alive_check */ static int hostkeys_update_complete; static int session_setup_complete; static void client_init_dispatch(struct ssh *ssh); int session_ident = -1; /* Track escape per proto2 channel */ struct escape_filter_ctx { int escape_pending; int escape_char; }; /* Context for channel confirmation replies */ struct channel_reply_ctx { const char *request_type; int id; enum confirm_action action; }; /* Global request success/failure callbacks */ /* XXX move to struct ssh? */ struct global_confirm { TAILQ_ENTRY(global_confirm) entry; global_confirm_cb *cb; void *ctx; int ref_count; }; TAILQ_HEAD(global_confirms, global_confirm); static struct global_confirms global_confirms = TAILQ_HEAD_INITIALIZER(global_confirms); void ssh_process_session2_setup(int, int, int, struct sshbuf *); static void quit_message(const char *fmt, ...) __attribute__((__format__ (printf, 1, 2))); static void quit_message(const char *fmt, ...) { char *msg; va_list args; int r; va_start(args, fmt); xvasprintf(&msg, fmt, args); va_end(args); if ((r = sshbuf_putf(stderr_buffer, "%s\r\n", msg)) != 0) fatal_fr(r, "sshbuf_putf"); free(msg); quit_pending = 1; } /* * Signal handler for the window change signal (SIGWINCH). This just sets a * flag indicating that the window has changed. */ static void window_change_handler(int sig) { received_window_change_signal = 1; } /* * Signal handler for signals that cause the program to terminate. These * signals must be trapped to restore terminal modes. */ static void signal_handler(int sig) { received_signal = sig; quit_pending = 1; } /* * Sets control_persist_exit_time to the absolute time when the * backgrounded control master should exit due to expiry of the * ControlPersist timeout. Sets it to 0 if we are not a backgrounded * control master process, or if there is no ControlPersist timeout. */ static void set_control_persist_exit_time(struct ssh *ssh) { if (muxserver_sock == -1 || !options.control_persist || options.control_persist_timeout == 0) { /* not using a ControlPersist timeout */ control_persist_exit_time = 0; } else if (channel_still_open(ssh)) { /* some client connections are still open */ if (control_persist_exit_time > 0) debug2_f("cancel scheduled exit"); control_persist_exit_time = 0; } else if (control_persist_exit_time <= 0) { /* a client connection has recently closed */ control_persist_exit_time = monotime() + (time_t)options.control_persist_timeout; debug2_f("schedule exit in %d seconds", options.control_persist_timeout); } /* else we are already counting down to the timeout */ } #define SSH_X11_VALID_DISPLAY_CHARS ":/.-_" static int client_x11_display_valid(const char *display) { size_t i, dlen; if (display == NULL) return 0; dlen = strlen(display); for (i = 0; i < dlen; i++) { if (!isalnum((u_char)display[i]) && strchr(SSH_X11_VALID_DISPLAY_CHARS, display[i]) == NULL) { debug("Invalid character '%c' in DISPLAY", display[i]); return 0; } } return 1; } #define SSH_X11_PROTO "MIT-MAGIC-COOKIE-1" #define X11_TIMEOUT_SLACK 60 int client_x11_get_proto(struct ssh *ssh, const char *display, const char *xauth_path, u_int trusted, u_int timeout, char **_proto, char **_data) { char *cmd, line[512], xdisplay[512]; char xauthfile[PATH_MAX], xauthdir[PATH_MAX]; static char proto[512], data[512]; FILE *f; int got_data = 0, generated = 0, do_unlink = 0, r; struct stat st; u_int now, x11_timeout_real; *_proto = proto; *_data = data; proto[0] = data[0] = xauthfile[0] = xauthdir[0] = '\0'; if (!client_x11_display_valid(display)) { if (display != NULL) logit("DISPLAY \"%s\" invalid; disabling X11 forwarding", display); return -1; } if (xauth_path != NULL && stat(xauth_path, &st) == -1) { debug("No xauth program."); xauth_path = NULL; } if (xauth_path != NULL) { /* * Handle FamilyLocal case where $DISPLAY does * not match an authorization entry. For this we * just try "xauth list unix:displaynum.screennum". * XXX: "localhost" match to determine FamilyLocal * is not perfect. */ if (strncmp(display, "localhost:", 10) == 0) { if ((r = snprintf(xdisplay, sizeof(xdisplay), "unix:%s", display + 10)) < 0 || (size_t)r >= sizeof(xdisplay)) { error_f("display name too long"); return -1; } display = xdisplay; } if (trusted == 0) { /* * Generate an untrusted X11 auth cookie. * * The authentication cookie should briefly outlive * ssh's willingness to forward X11 connections to * avoid nasty fail-open behaviour in the X server. */ mktemp_proto(xauthdir, sizeof(xauthdir)); if (mkdtemp(xauthdir) == NULL) { error_f("mkdtemp: %s", strerror(errno)); return -1; } do_unlink = 1; if ((r = snprintf(xauthfile, sizeof(xauthfile), "%s/xauthfile", xauthdir)) < 0 || (size_t)r >= sizeof(xauthfile)) { error_f("xauthfile path too long"); rmdir(xauthdir); return -1; } if (timeout == 0) { /* auth doesn't time out */ xasprintf(&cmd, "%s -f %s generate %s %s " "untrusted 2>%s", xauth_path, xauthfile, display, SSH_X11_PROTO, _PATH_DEVNULL); } else { /* Add some slack to requested expiry */ if (timeout < UINT_MAX - X11_TIMEOUT_SLACK) x11_timeout_real = timeout + X11_TIMEOUT_SLACK; else { /* Don't overflow on long timeouts */ x11_timeout_real = UINT_MAX; } xasprintf(&cmd, "%s -f %s generate %s %s " "untrusted timeout %u 2>%s", xauth_path, xauthfile, display, SSH_X11_PROTO, x11_timeout_real, _PATH_DEVNULL); } debug2_f("xauth command: %s", cmd); if (timeout != 0 && x11_refuse_time == 0) { now = monotime() + 1; if (SSH_TIME_T_MAX - timeout < now) x11_refuse_time = SSH_TIME_T_MAX; else x11_refuse_time = now + timeout; channel_set_x11_refuse_time(ssh, x11_refuse_time); } if (system(cmd) == 0) generated = 1; free(cmd); } /* * When in untrusted mode, we read the cookie only if it was * successfully generated as an untrusted one in the step * above. */ if (trusted || generated) { xasprintf(&cmd, "%s %s%s list %s 2>" _PATH_DEVNULL, xauth_path, generated ? "-f " : "" , generated ? xauthfile : "", display); debug2("x11_get_proto: %s", cmd); f = popen(cmd, "r"); if (f && fgets(line, sizeof(line), f) && sscanf(line, "%*s %511s %511s", proto, data) == 2) got_data = 1; if (f) pclose(f); free(cmd); } } if (do_unlink) { unlink(xauthfile); rmdir(xauthdir); } /* Don't fall back to fake X11 data for untrusted forwarding */ if (!trusted && !got_data) { error("Warning: untrusted X11 forwarding setup failed: " "xauth key data not generated"); return -1; } /* * If we didn't get authentication data, just make up some * data. The forwarding code will check the validity of the * response anyway, and substitute this data. The X11 * server, however, will ignore this fake data and use * whatever authentication mechanisms it was using otherwise * for the local connection. */ if (!got_data) { u_int8_t rnd[16]; u_int i; logit("Warning: No xauth data; " "using fake authentication data for X11 forwarding."); strlcpy(proto, SSH_X11_PROTO, sizeof proto); arc4random_buf(rnd, sizeof(rnd)); for (i = 0; i < sizeof(rnd); i++) { snprintf(data + 2 * i, sizeof data - 2 * i, "%02x", rnd[i]); } } return 0; } /* * Checks if the client window has changed, and sends a packet about it to * the server if so. The actual change is detected elsewhere (by a software * interrupt on Unix); this just checks the flag and sends a message if * appropriate. */ static void client_check_window_change(struct ssh *ssh) { if (!received_window_change_signal) return; received_window_change_signal = 0; debug2_f("changed"); channel_send_window_changes(ssh); } static int client_global_request_reply(int type, u_int32_t seq, struct ssh *ssh) { struct global_confirm *gc; if ((gc = TAILQ_FIRST(&global_confirms)) == NULL) return 0; if (gc->cb != NULL) gc->cb(ssh, type, seq, gc->ctx); if (--gc->ref_count <= 0) { TAILQ_REMOVE(&global_confirms, gc, entry); freezero(gc, sizeof(*gc)); } ssh_packet_set_alive_timeouts(ssh, 0); return 0; } static void schedule_server_alive_check(void) { if (options.server_alive_interval > 0) server_alive_time = monotime() + options.server_alive_interval; } static void server_alive_check(struct ssh *ssh) { int r; if (ssh_packet_inc_alive_timeouts(ssh) > options.server_alive_count_max) { logit("Timeout, server %s not responding.", host); cleanup_exit(255); } if ((r = sshpkt_start(ssh, SSH2_MSG_GLOBAL_REQUEST)) != 0 || (r = sshpkt_put_cstring(ssh, "keepalive@openssh.com")) != 0 || (r = sshpkt_put_u8(ssh, 1)) != 0 || /* boolean: want reply */ (r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send packet"); /* Insert an empty placeholder to maintain ordering */ client_register_global_confirm(NULL, NULL); schedule_server_alive_check(); } /* Try to send a dummy keystroke */ static int send_chaff(struct ssh *ssh) { int r; if ((ssh->kex->flags & KEX_HAS_PING) == 0) return 0; /* XXX probabilistically send chaff? */ /* * a SSH2_MSG_CHANNEL_DATA payload is 9 bytes: * 4 bytes channel ID + 4 bytes string length + 1 byte string data * simulate that here. */ if ((r = sshpkt_start(ssh, SSH2_MSG_PING)) != 0 || (r = sshpkt_put_cstring(ssh, "PING!")) != 0 || (r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send packet"); return 1; } /* Sets the next interval to send a keystroke or chaff packet */ static void set_next_interval(const struct timespec *now, struct timespec *next_interval, u_int interval_ms, int starting) { struct timespec tmp; long long interval_ns, fuzz_ns; static long long rate_fuzz; interval_ns = interval_ms * (1000LL * 1000); fuzz_ns = (interval_ns * SSH_KEYSTROKE_TIMING_FUZZ) / 100; /* Center fuzz around requested interval */ if (fuzz_ns > INT_MAX) fuzz_ns = INT_MAX; if (fuzz_ns > interval_ns) { /* Shouldn't happen */ fatal_f("internal error: fuzz %u%% %lldns > interval %lldns", SSH_KEYSTROKE_TIMING_FUZZ, fuzz_ns, interval_ns); } /* * Randomise the keystroke/chaff intervals in two ways: * 1. Each interval has some random jitter applied to make the * interval-to-interval time unpredictable. * 2. The overall interval rate is also randomly perturbed for each * chaffing session to make the average rate unpredictable. */ if (starting) rate_fuzz = arc4random_uniform(fuzz_ns); interval_ns -= fuzz_ns; interval_ns += arc4random_uniform(fuzz_ns) + rate_fuzz; tmp.tv_sec = interval_ns / (1000 * 1000 * 1000); tmp.tv_nsec = interval_ns % (1000 * 1000 * 1000); timespecadd(now, &tmp, next_interval); } /* * Performs keystroke timing obfuscation. Returns non-zero if the * output fd should be polled. */ static int obfuscate_keystroke_timing(struct ssh *ssh, struct timespec *timeout, int channel_did_enqueue) { static int active; static struct timespec next_interval, chaff_until; struct timespec now, tmp; int just_started = 0, had_keystroke = 0; static unsigned long long nchaff; const char *stop_reason = NULL; long long n; monotime_ts(&now); if (options.obscure_keystroke_timing_interval <= 0) return 1; /* disabled in config */ if (!channel_tty_open(ssh) || quit_pending) { /* Stop if no channels left of we're waiting for one to close */ stop_reason = "no active channels"; } else if (ssh_packet_is_rekeying(ssh)) { /* Stop if we're rekeying */ stop_reason = "rekeying started"; } else if (!ssh_packet_interactive_data_to_write(ssh) && ssh_packet_have_data_to_write(ssh)) { /* Stop if the output buffer has more than a few keystrokes */ stop_reason = "output buffer filling"; } else if (active && channel_did_enqueue && ssh_packet_have_data_to_write(ssh)) { /* Still in active mode and have a keystroke queued. */ had_keystroke = 1; } else if (active) { if (timespeccmp(&now, &chaff_until, >=)) { /* Stop if there have been no keystrokes for a while */ stop_reason = "chaff time expired"; } else if (timespeccmp(&now, &next_interval, >=)) { /* Otherwise if we were due to send, then send chaff */ if (send_chaff(ssh)) nchaff++; } } if (stop_reason != NULL) { if (active) { debug3_f("stopping: %s (%llu chaff packets sent)", stop_reason, nchaff); active = 0; } return 1; } /* * If we're in interactive mode, and only have a small amount * of outbound data, then we assume that the user is typing * interactively. In this case, start quantising outbound packets to * fixed time intervals to hide inter-keystroke timing. */ if (!active && ssh_packet_interactive_data_to_write(ssh) && channel_did_enqueue && ssh_packet_have_data_to_write(ssh)) { debug3_f("starting: interval ~%dms", options.obscure_keystroke_timing_interval); just_started = had_keystroke = active = 1; nchaff = 0; set_next_interval(&now, &next_interval, options.obscure_keystroke_timing_interval, 1); } /* Don't hold off if obfuscation inactive */ if (!active) return 1; if (had_keystroke) { /* * Arrange to send chaff packets for a random interval after * the last keystroke was sent. */ ms_to_timespec(&tmp, SSH_KEYSTROKE_CHAFF_MIN_MS + arc4random_uniform(SSH_KEYSTROKE_CHAFF_RNG_MS)); timespecadd(&now, &tmp, &chaff_until); } ptimeout_deadline_monotime_tsp(timeout, &next_interval); if (just_started) return 1; /* Don't arm output fd for poll until the timing interval has elapsed */ if (timespeccmp(&now, &next_interval, <)) return 0; /* Calculate number of intervals missed since the last check */ n = (now.tv_sec - next_interval.tv_sec) * 1000LL * 1000 * 1000; n += now.tv_nsec - next_interval.tv_nsec; n /= options.obscure_keystroke_timing_interval * 1000LL * 1000; n = (n < 0) ? 1 : n + 1; /* Advance to the next interval */ set_next_interval(&now, &next_interval, options.obscure_keystroke_timing_interval * n, 0); return 1; } /* * Waits until the client can do something (some data becomes available on * one of the file descriptors). */ static void client_wait_until_can_do_something(struct ssh *ssh, struct pollfd **pfdp, u_int *npfd_allocp, u_int *npfd_activep, int channel_did_enqueue, sigset_t *sigsetp, int *conn_in_readyp, int *conn_out_readyp) { struct timespec timeout; int ret, oready; u_int p; *conn_in_readyp = *conn_out_readyp = 0; /* Prepare channel poll. First two pollfd entries are reserved */ ptimeout_init(&timeout); channel_prepare_poll(ssh, pfdp, npfd_allocp, npfd_activep, 2, &timeout); if (*npfd_activep < 2) fatal_f("bad npfd %u", *npfd_activep); /* shouldn't happen */ /* channel_prepare_poll could have closed the last channel */ if (session_closed && !channel_still_open(ssh) && !ssh_packet_have_data_to_write(ssh)) { /* clear events since we did not call poll() */ for (p = 0; p < *npfd_activep; p++) (*pfdp)[p].revents = 0; return; } oready = obfuscate_keystroke_timing(ssh, &timeout, channel_did_enqueue); /* Monitor server connection on reserved pollfd entries */ (*pfdp)[0].fd = connection_in; (*pfdp)[0].events = POLLIN; (*pfdp)[1].fd = connection_out; (*pfdp)[1].events = (oready && ssh_packet_have_data_to_write(ssh)) ? POLLOUT : 0; /* * Wait for something to happen. This will suspend the process until * some polled descriptor can be read, written, or has some other * event pending, or a timeout expires. */ set_control_persist_exit_time(ssh); if (control_persist_exit_time > 0) ptimeout_deadline_monotime(&timeout, control_persist_exit_time); if (options.server_alive_interval > 0) ptimeout_deadline_monotime(&timeout, server_alive_time); if (options.rekey_interval > 0 && !ssh_packet_is_rekeying(ssh)) { ptimeout_deadline_sec(&timeout, ssh_packet_get_rekey_timeout(ssh)); } ret = ppoll(*pfdp, *npfd_activep, ptimeout_get_tsp(&timeout), sigsetp); if (ret == -1) { /* * We have to clear the events because we return. * We have to return, because the mainloop checks for the flags * set by the signal handlers. */ for (p = 0; p < *npfd_activep; p++) (*pfdp)[p].revents = 0; if (errno == EINTR) return; /* Note: we might still have data in the buffers. */ quit_message("poll: %s", strerror(errno)); return; } *conn_in_readyp = (*pfdp)[0].revents != 0; *conn_out_readyp = (*pfdp)[1].revents != 0; if (options.server_alive_interval > 0 && !*conn_in_readyp && monotime() >= server_alive_time) { /* * ServerAlive check is needed. We can't rely on the poll * timing out since traffic on the client side such as port * forwards can keep waking it up. */ server_alive_check(ssh); } } static void client_suspend_self(struct sshbuf *bin, struct sshbuf *bout, struct sshbuf *berr) { /* Flush stdout and stderr buffers. */ if (sshbuf_len(bout) > 0) atomicio(vwrite, fileno(stdout), sshbuf_mutable_ptr(bout), sshbuf_len(bout)); if (sshbuf_len(berr) > 0) atomicio(vwrite, fileno(stderr), sshbuf_mutable_ptr(berr), sshbuf_len(berr)); leave_raw_mode(options.request_tty == REQUEST_TTY_FORCE); sshbuf_reset(bin); sshbuf_reset(bout); sshbuf_reset(berr); /* Send the suspend signal to the program itself. */ kill(getpid(), SIGTSTP); /* Reset window sizes in case they have changed */ received_window_change_signal = 1; enter_raw_mode(options.request_tty == REQUEST_TTY_FORCE); } static void client_process_net_input(struct ssh *ssh) { int r; /* * Read input from the server, and add any such data to the buffer of * the packet subsystem. */ schedule_server_alive_check(); if ((r = ssh_packet_process_read(ssh, connection_in)) == 0) return; /* success */ if (r == SSH_ERR_SYSTEM_ERROR) { if (errno == EAGAIN || errno == EINTR) return; if (errno == EPIPE) { quit_message("Connection to %s closed by remote host.", host); return; } } quit_message("Read from remote host %s: %s", host, ssh_err(r)); } static void client_status_confirm(struct ssh *ssh, int type, Channel *c, void *ctx) { struct channel_reply_ctx *cr = (struct channel_reply_ctx *)ctx; char errmsg[256]; int r, tochan; /* * If a TTY was explicitly requested, then a failure to allocate * one is fatal. */ if (cr->action == CONFIRM_TTY && (options.request_tty == REQUEST_TTY_FORCE || options.request_tty == REQUEST_TTY_YES)) cr->action = CONFIRM_CLOSE; /* XXX suppress on mux _client_ quietmode */ tochan = options.log_level >= SYSLOG_LEVEL_ERROR && c->ctl_chan != -1 && c->extended_usage == CHAN_EXTENDED_WRITE; if (type == SSH2_MSG_CHANNEL_SUCCESS) { debug2("%s request accepted on channel %d", cr->request_type, c->self); } else if (type == SSH2_MSG_CHANNEL_FAILURE) { if (tochan) { snprintf(errmsg, sizeof(errmsg), "%s request failed\r\n", cr->request_type); } else { snprintf(errmsg, sizeof(errmsg), "%s request failed on channel %d", cr->request_type, c->self); } /* If error occurred on primary session channel, then exit */ if (cr->action == CONFIRM_CLOSE && c->self == session_ident) fatal("%s", errmsg); /* * If error occurred on mux client, append to * their stderr. */ if (tochan) { debug3_f("channel %d: mux request: %s", c->self, cr->request_type); if ((r = sshbuf_put(c->extended, errmsg, strlen(errmsg))) != 0) fatal_fr(r, "sshbuf_put"); } else error("%s", errmsg); if (cr->action == CONFIRM_TTY) { /* * If a TTY allocation error occurred, then arrange * for the correct TTY to leave raw mode. */ if (c->self == session_ident) leave_raw_mode(0); else mux_tty_alloc_failed(ssh, c); } else if (cr->action == CONFIRM_CLOSE) { chan_read_failed(ssh, c); chan_write_failed(ssh, c); } } free(cr); } static void client_abandon_status_confirm(struct ssh *ssh, Channel *c, void *ctx) { free(ctx); } void client_expect_confirm(struct ssh *ssh, int id, const char *request, enum confirm_action action) { struct channel_reply_ctx *cr = xcalloc(1, sizeof(*cr)); cr->request_type = request; cr->action = action; channel_register_status_confirm(ssh, id, client_status_confirm, client_abandon_status_confirm, cr); } void client_register_global_confirm(global_confirm_cb *cb, void *ctx) { struct global_confirm *gc, *last_gc; /* Coalesce identical callbacks */ last_gc = TAILQ_LAST(&global_confirms, global_confirms); if (last_gc && last_gc->cb == cb && last_gc->ctx == ctx) { if (++last_gc->ref_count >= INT_MAX) fatal_f("last_gc->ref_count = %d", last_gc->ref_count); return; } gc = xcalloc(1, sizeof(*gc)); gc->cb = cb; gc->ctx = ctx; gc->ref_count = 1; TAILQ_INSERT_TAIL(&global_confirms, gc, entry); } /* * Returns non-zero if the client is able to handle a hostkeys-00@openssh.com * hostkey update request. */ static int can_update_hostkeys(void) { if (hostkeys_update_complete) return 0; if (options.update_hostkeys == SSH_UPDATE_HOSTKEYS_ASK && options.batch_mode) return 0; /* won't ask in batchmode, so don't even try */ if (!options.update_hostkeys || options.num_user_hostfiles <= 0) return 0; return 1; } static void client_repledge(void) { debug3_f("enter"); /* Might be able to tighten pledge now that session is established */ if (options.control_master || options.control_path != NULL || options.forward_x11 || options.fork_after_authentication || can_update_hostkeys() || (session_ident != -1 && !session_setup_complete)) { /* Can't tighten */ return; } #ifdef __OpenBSD__ /* * LocalCommand and UpdateHostkeys have finished, so can get rid of * filesystem. * * XXX protocol allows a server can to change hostkeys during the * connection at rekey time that could trigger a hostkeys update * but AFAIK no implementations support this. Could improve by * forcing known_hosts to be read-only or via unveil(2). */ if (options.num_local_forwards != 0 || options.num_remote_forwards != 0 || options.num_permitted_remote_opens != 0 || options.enable_escape_commandline != 0) { /* rfwd needs inet */ debug("pledge: network"); if (pledge("stdio unix inet dns proc tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } else if (options.forward_agent != 0) { /* agent forwarding needs to open $SSH_AUTH_SOCK at will */ debug("pledge: agent"); if (pledge("stdio unix proc tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } else { debug("pledge: fork"); if (pledge("stdio proc tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } #endif /* XXX further things to do: * * - might be able to get rid of proc if we kill ~^Z * - ssh -N (no session) * - stdio forwarding * - sessions without tty */ } static void process_cmdline(struct ssh *ssh) { void (*handler)(int); char *s, *cmd; int ok, delete = 0, local = 0, remote = 0, dynamic = 0; struct Forward fwd; memset(&fwd, 0, sizeof(fwd)); leave_raw_mode(options.request_tty == REQUEST_TTY_FORCE); handler = ssh_signal(SIGINT, SIG_IGN); cmd = s = read_passphrase("\r\nssh> ", RP_ECHO); if (s == NULL) goto out; while (isspace((u_char)*s)) s++; if (*s == '-') s++; /* Skip cmdline '-', if any */ if (*s == '\0') goto out; if (*s == 'h' || *s == 'H' || *s == '?') { logit("Commands:"); logit(" -L[bind_address:]port:host:hostport " "Request local forward"); logit(" -R[bind_address:]port:host:hostport " "Request remote forward"); logit(" -D[bind_address:]port " "Request dynamic forward"); logit(" -KL[bind_address:]port " "Cancel local forward"); logit(" -KR[bind_address:]port " "Cancel remote forward"); logit(" -KD[bind_address:]port " "Cancel dynamic forward"); if (!options.permit_local_command) goto out; logit(" !args " "Execute local command"); goto out; } if (*s == '!' && options.permit_local_command) { s++; ssh_local_cmd(s); goto out; } if (*s == 'K') { delete = 1; s++; } if (*s == 'L') local = 1; else if (*s == 'R') remote = 1; else if (*s == 'D') dynamic = 1; else { logit("Invalid command."); goto out; } while (isspace((u_char)*++s)) ; /* XXX update list of forwards in options */ if (delete) { /* We pass 1 for dynamicfwd to restrict to 1 or 2 fields. */ if (!parse_forward(&fwd, s, 1, 0)) { logit("Bad forwarding close specification."); goto out; } if (remote) ok = channel_request_rforward_cancel(ssh, &fwd) == 0; else if (dynamic) ok = channel_cancel_lport_listener(ssh, &fwd, 0, &options.fwd_opts) > 0; else ok = channel_cancel_lport_listener(ssh, &fwd, CHANNEL_CANCEL_PORT_STATIC, &options.fwd_opts) > 0; if (!ok) { logit("Unknown port forwarding."); goto out; } logit("Canceled forwarding."); } else { /* -R specs can be both dynamic or not, so check both. */ if (remote) { if (!parse_forward(&fwd, s, 0, remote) && !parse_forward(&fwd, s, 1, remote)) { logit("Bad remote forwarding specification."); goto out; } } else if (!parse_forward(&fwd, s, dynamic, remote)) { logit("Bad local forwarding specification."); goto out; } if (local || dynamic) { if (!channel_setup_local_fwd_listener(ssh, &fwd, &options.fwd_opts)) { logit("Port forwarding failed."); goto out; } } else { if (channel_request_remote_forwarding(ssh, &fwd) < 0) { logit("Port forwarding failed."); goto out; } } logit("Forwarding port."); } out: ssh_signal(SIGINT, handler); enter_raw_mode(options.request_tty == REQUEST_TTY_FORCE); free(cmd); free(fwd.listen_host); free(fwd.listen_path); free(fwd.connect_host); free(fwd.connect_path); } /* reasons to suppress output of an escape command in help output */ #define SUPPRESS_NEVER 0 /* never suppress, always show */ #define SUPPRESS_MUXCLIENT 1 /* don't show in mux client sessions */ #define SUPPRESS_MUXMASTER 2 /* don't show in mux master sessions */ #define SUPPRESS_SYSLOG 4 /* don't show when logging to syslog */ #define SUPPRESS_NOCMDLINE 8 /* don't show when cmdline disabled*/ struct escape_help_text { const char *cmd; const char *text; unsigned int flags; }; static struct escape_help_text esc_txt[] = { {".", "terminate session", SUPPRESS_MUXMASTER}, {".", "terminate connection (and any multiplexed sessions)", SUPPRESS_MUXCLIENT}, {"B", "send a BREAK to the remote system", SUPPRESS_NEVER}, {"C", "open a command line", SUPPRESS_MUXCLIENT|SUPPRESS_NOCMDLINE}, {"R", "request rekey", SUPPRESS_NEVER}, {"V/v", "decrease/increase verbosity (LogLevel)", SUPPRESS_MUXCLIENT}, {"^Z", "suspend ssh", SUPPRESS_MUXCLIENT}, {"#", "list forwarded connections", SUPPRESS_NEVER}, {"&", "background ssh (when waiting for connections to terminate)", SUPPRESS_MUXCLIENT}, {"?", "this message", SUPPRESS_NEVER}, }; static void print_escape_help(struct sshbuf *b, int escape_char, int mux_client, int using_stderr) { unsigned int i, suppress_flags; int r; if ((r = sshbuf_putf(b, "%c?\r\nSupported escape sequences:\r\n", escape_char)) != 0) fatal_fr(r, "sshbuf_putf"); suppress_flags = (mux_client ? SUPPRESS_MUXCLIENT : 0) | (mux_client ? 0 : SUPPRESS_MUXMASTER) | (using_stderr ? 0 : SUPPRESS_SYSLOG) | (options.enable_escape_commandline == 0 ? SUPPRESS_NOCMDLINE : 0); for (i = 0; i < sizeof(esc_txt)/sizeof(esc_txt[0]); i++) { if (esc_txt[i].flags & suppress_flags) continue; if ((r = sshbuf_putf(b, " %c%-3s - %s\r\n", escape_char, esc_txt[i].cmd, esc_txt[i].text)) != 0) fatal_fr(r, "sshbuf_putf"); } if ((r = sshbuf_putf(b, " %c%c - send the escape character by typing it twice\r\n" "(Note that escapes are only recognized immediately after " "newline.)\r\n", escape_char, escape_char)) != 0) fatal_fr(r, "sshbuf_putf"); } /* * Process the characters one by one. */ static int process_escapes(struct ssh *ssh, Channel *c, struct sshbuf *bin, struct sshbuf *bout, struct sshbuf *berr, const char *buf, int len) { pid_t pid; int r, bytes = 0; u_int i; u_char ch; char *s; struct escape_filter_ctx *efc; if (c == NULL || c->filter_ctx == NULL || len <= 0) return 0; efc = (struct escape_filter_ctx *)c->filter_ctx; for (i = 0; i < (u_int)len; i++) { /* Get one character at a time. */ ch = buf[i]; if (efc->escape_pending) { /* We have previously seen an escape character. */ /* Clear the flag now. */ efc->escape_pending = 0; /* Process the escaped character. */ switch (ch) { case '.': /* Terminate the connection. */ if ((r = sshbuf_putf(berr, "%c.\r\n", efc->escape_char)) != 0) fatal_fr(r, "sshbuf_putf"); if (c && c->ctl_chan != -1) { channel_force_close(ssh, c, 1); return 0; } else quit_pending = 1; return -1; case 'Z' - 64: /* XXX support this for mux clients */ if (c && c->ctl_chan != -1) { char b[16]; noescape: if (ch == 'Z' - 64) snprintf(b, sizeof b, "^Z"); else snprintf(b, sizeof b, "%c", ch); if ((r = sshbuf_putf(berr, "%c%s escape not available to " "multiplexed sessions\r\n", efc->escape_char, b)) != 0) fatal_fr(r, "sshbuf_putf"); continue; } /* Suspend the program. Inform the user */ if ((r = sshbuf_putf(berr, "%c^Z [suspend ssh]\r\n", efc->escape_char)) != 0) fatal_fr(r, "sshbuf_putf"); /* Restore terminal modes and suspend. */ client_suspend_self(bin, bout, berr); /* We have been continued. */ continue; case 'B': if ((r = sshbuf_putf(berr, "%cB\r\n", efc->escape_char)) != 0) fatal_fr(r, "sshbuf_putf"); channel_request_start(ssh, c->self, "break", 0); if ((r = sshpkt_put_u32(ssh, 1000)) != 0 || (r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send packet"); continue; case 'R': if (ssh->compat & SSH_BUG_NOREKEY) logit("Server does not " "support re-keying"); else need_rekeying = 1; continue; case 'V': /* FALLTHROUGH */ case 'v': if (c && c->ctl_chan != -1) goto noescape; if (!log_is_on_stderr()) { if ((r = sshbuf_putf(berr, "%c%c [Logging to syslog]\r\n", efc->escape_char, ch)) != 0) fatal_fr(r, "sshbuf_putf"); continue; } if (ch == 'V' && options.log_level > SYSLOG_LEVEL_QUIET) log_change_level(--options.log_level); if (ch == 'v' && options.log_level < SYSLOG_LEVEL_DEBUG3) log_change_level(++options.log_level); if ((r = sshbuf_putf(berr, "%c%c [LogLevel %s]\r\n", efc->escape_char, ch, log_level_name(options.log_level))) != 0) fatal_fr(r, "sshbuf_putf"); continue; case '&': if (c->ctl_chan != -1) goto noescape; /* * Detach the program (continue to serve * connections, but put in background and no * more new connections). */ /* Restore tty modes. */ leave_raw_mode( options.request_tty == REQUEST_TTY_FORCE); /* Stop listening for new connections. */ channel_stop_listening(ssh); if ((r = sshbuf_putf(berr, "%c& " "[backgrounded]\n", efc->escape_char)) != 0) fatal_fr(r, "sshbuf_putf"); /* Fork into background. */ pid = fork(); if (pid == -1) { error("fork: %.100s", strerror(errno)); continue; } if (pid != 0) { /* This is the parent. */ /* The parent just exits. */ exit(0); } /* The child continues serving connections. */ /* fake EOF on stdin */ if ((r = sshbuf_put_u8(bin, 4)) != 0) fatal_fr(r, "sshbuf_put_u8"); return -1; case '?': print_escape_help(berr, efc->escape_char, (c && c->ctl_chan != -1), log_is_on_stderr()); continue; case '#': if ((r = sshbuf_putf(berr, "%c#\r\n", efc->escape_char)) != 0) fatal_fr(r, "sshbuf_putf"); s = channel_open_message(ssh); if ((r = sshbuf_put(berr, s, strlen(s))) != 0) fatal_fr(r, "sshbuf_put"); free(s); continue; case 'C': if (c && c->ctl_chan != -1) goto noescape; if (options.enable_escape_commandline == 0) { if ((r = sshbuf_putf(berr, "commandline disabled\r\n")) != 0) fatal_fr(r, "sshbuf_putf"); continue; } process_cmdline(ssh); continue; default: if (ch != efc->escape_char) { if ((r = sshbuf_put_u8(bin, efc->escape_char)) != 0) fatal_fr(r, "sshbuf_put_u8"); bytes++; } /* Escaped characters fall through here */ break; } } else { /* * The previous character was not an escape char. * Check if this is an escape. */ if (last_was_cr && ch == efc->escape_char) { /* * It is. Set the flag and continue to * next character. */ efc->escape_pending = 1; continue; } } /* * Normal character. Record whether it was a newline, * and append it to the buffer. */ last_was_cr = (ch == '\r' || ch == '\n'); if ((r = sshbuf_put_u8(bin, ch)) != 0) fatal_fr(r, "sshbuf_put_u8"); bytes++; } return bytes; } /* * Get packets from the connection input buffer, and process them as long as * there are packets available. * * Any unknown packets received during the actual * session cause the session to terminate. This is * intended to make debugging easier since no * confirmations are sent. Any compatible protocol * extensions must be negotiated during the * preparatory phase. */ static void client_process_buffered_input_packets(struct ssh *ssh) { ssh_dispatch_run_fatal(ssh, DISPATCH_NONBLOCK, &quit_pending); } /* scan buf[] for '~' before sending data to the peer */ /* Helper: allocate a new escape_filter_ctx and fill in its escape char */ void * client_new_escape_filter_ctx(int escape_char) { struct escape_filter_ctx *ret; ret = xcalloc(1, sizeof(*ret)); ret->escape_pending = 0; ret->escape_char = escape_char; return (void *)ret; } /* Free the escape filter context on channel free */ void client_filter_cleanup(struct ssh *ssh, int cid, void *ctx) { free(ctx); } int client_simple_escape_filter(struct ssh *ssh, Channel *c, const char *buf, int len) { if (c->extended_usage != CHAN_EXTENDED_WRITE) return 0; return process_escapes(ssh, c, c->input, c->output, c->extended, buf, len); } static void client_channel_closed(struct ssh *ssh, int id, int force, void *arg) { channel_cancel_cleanup(ssh, id); session_closed = 1; leave_raw_mode(options.request_tty == REQUEST_TTY_FORCE); } /* * Implements the interactive session with the server. This is called after * the user has been authenticated, and a command has been started on the * remote host. If escape_char != SSH_ESCAPECHAR_NONE, it is the character * used as an escape character for terminating or suspending the session. */ int client_loop(struct ssh *ssh, int have_pty, int escape_char_arg, int ssh2_chan_id) { struct pollfd *pfd = NULL; u_int npfd_alloc = 0, npfd_active = 0; double start_time, total_time; int channel_did_enqueue = 0, r, len; u_int64_t ibytes, obytes; int conn_in_ready, conn_out_ready; sigset_t bsigset, osigset; debug("Entering interactive session."); session_ident = ssh2_chan_id; #ifdef __OpenBSD__ if (options.control_master && !option_clear_or_none(options.control_path)) { debug("pledge: id"); if (pledge("stdio rpath wpath cpath unix inet dns recvfd sendfd proc exec id tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } else if (options.forward_x11 || options.permit_local_command) { debug("pledge: exec"); if (pledge("stdio rpath wpath cpath unix inet dns proc exec tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } else if (options.update_hostkeys) { debug("pledge: filesystem"); if (pledge("stdio rpath wpath cpath unix inet dns proc tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } else if (!option_clear_or_none(options.proxy_command) || options.fork_after_authentication) { debug("pledge: proc"); if (pledge("stdio cpath unix inet dns proc tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } else { debug("pledge: network"); if (pledge("stdio unix inet dns proc tty", NULL) == -1) fatal_f("pledge(): %s", strerror(errno)); } #endif /* might be able to tighten now */ client_repledge(); start_time = monotime_double(); /* Initialize variables. */ last_was_cr = 1; exit_status = -1; connection_in = ssh_packet_get_connection_in(ssh); connection_out = ssh_packet_get_connection_out(ssh); quit_pending = 0; /* Initialize buffer. */ if ((stderr_buffer = sshbuf_new()) == NULL) fatal_f("sshbuf_new failed"); client_init_dispatch(ssh); /* * Set signal handlers, (e.g. to restore non-blocking mode) * but don't overwrite SIG_IGN, matches behaviour from rsh(1) */ if (ssh_signal(SIGHUP, SIG_IGN) != SIG_IGN) ssh_signal(SIGHUP, signal_handler); if (ssh_signal(SIGINT, SIG_IGN) != SIG_IGN) ssh_signal(SIGINT, signal_handler); if (ssh_signal(SIGQUIT, SIG_IGN) != SIG_IGN) ssh_signal(SIGQUIT, signal_handler); if (ssh_signal(SIGTERM, SIG_IGN) != SIG_IGN) ssh_signal(SIGTERM, signal_handler); ssh_signal(SIGWINCH, window_change_handler); if (have_pty) enter_raw_mode(options.request_tty == REQUEST_TTY_FORCE); if (session_ident != -1) { if (escape_char_arg != SSH_ESCAPECHAR_NONE) { channel_register_filter(ssh, session_ident, client_simple_escape_filter, NULL, client_filter_cleanup, client_new_escape_filter_ctx( escape_char_arg)); } channel_register_cleanup(ssh, session_ident, client_channel_closed, 0); } schedule_server_alive_check(); if (sigemptyset(&bsigset) == -1 || sigaddset(&bsigset, SIGHUP) == -1 || sigaddset(&bsigset, SIGINT) == -1 || sigaddset(&bsigset, SIGQUIT) == -1 || sigaddset(&bsigset, SIGTERM) == -1) error_f("bsigset setup: %s", strerror(errno)); /* Main loop of the client for the interactive session mode. */ while (!quit_pending) { channel_did_enqueue = 0; /* Process buffered packets sent by the server. */ client_process_buffered_input_packets(ssh); if (session_closed && !channel_still_open(ssh)) break; if (ssh_packet_is_rekeying(ssh)) { debug("rekeying in progress"); } else if (need_rekeying) { /* manual rekey request */ debug("need rekeying"); if ((r = kex_start_rekex(ssh)) != 0) fatal_fr(r, "kex_start_rekex"); need_rekeying = 0; } else { /* * Make packets from buffered channel data, and * enqueue them for sending to the server. */ if (ssh_packet_not_very_much_data_to_write(ssh)) channel_did_enqueue = channel_output_poll(ssh); /* * Check if the window size has changed, and buffer a * message about it to the server if so. */ client_check_window_change(ssh); } /* * Wait until we have something to do (something becomes * available on one of the descriptors). */ if (sigprocmask(SIG_BLOCK, &bsigset, &osigset) == -1) error_f("bsigset sigprocmask: %s", strerror(errno)); if (quit_pending) break; client_wait_until_can_do_something(ssh, &pfd, &npfd_alloc, &npfd_active, channel_did_enqueue, &osigset, &conn_in_ready, &conn_out_ready); if (sigprocmask(SIG_UNBLOCK, &bsigset, &osigset) == -1) error_f("osigset sigprocmask: %s", strerror(errno)); if (quit_pending) break; /* Do channel operations. */ channel_after_poll(ssh, pfd, npfd_active); /* Buffer input from the connection. */ if (conn_in_ready) client_process_net_input(ssh); if (quit_pending) break; /* A timeout may have triggered rekeying */ if ((r = ssh_packet_check_rekey(ssh)) != 0) fatal_fr(r, "cannot start rekeying"); /* * Send as much buffered packet data as possible to the * sender. */ if (conn_out_ready) { if ((r = ssh_packet_write_poll(ssh)) < 0) { sshpkt_fatal(ssh, r, "%s: ssh_packet_write_poll", __func__); } } /* * If we are a backgrounded control master, and the * timeout has expired without any active client * connections, then quit. */ if (control_persist_exit_time > 0) { if (monotime() >= control_persist_exit_time) { debug("ControlPersist timeout expired"); break; } } } free(pfd); /* Terminate the session. */ /* Stop watching for window change. */ ssh_signal(SIGWINCH, SIG_DFL); if ((r = sshpkt_start(ssh, SSH2_MSG_DISCONNECT)) != 0 || (r = sshpkt_put_u32(ssh, SSH2_DISCONNECT_BY_APPLICATION)) != 0 || (r = sshpkt_put_cstring(ssh, "disconnected by user")) != 0 || (r = sshpkt_put_cstring(ssh, "")) != 0 || /* language tag */ (r = sshpkt_send(ssh)) != 0 || (r = ssh_packet_write_wait(ssh)) < 0) fatal_fr(r, "send disconnect"); channel_free_all(ssh); if (have_pty) leave_raw_mode(options.request_tty == REQUEST_TTY_FORCE); /* * If there was no shell or command requested, there will be no remote * exit status to be returned. In that case, clear error code if the * connection was deliberately terminated at this end. */ if (options.session_type == SESSION_TYPE_NONE && received_signal == SIGTERM) { received_signal = 0; exit_status = 0; } if (received_signal) { verbose("Killed by signal %d.", (int) received_signal); cleanup_exit(255); } /* * In interactive mode (with pseudo tty) display a message indicating * that the connection has been closed. */ if (have_pty && options.log_level >= SYSLOG_LEVEL_INFO) quit_message("Connection to %s closed.", host); /* Output any buffered data for stderr. */ if (sshbuf_len(stderr_buffer) > 0) { len = atomicio(vwrite, fileno(stderr), sshbuf_mutable_ptr(stderr_buffer), sshbuf_len(stderr_buffer)); if (len < 0 || (u_int)len != sshbuf_len(stderr_buffer)) error("Write failed flushing stderr buffer."); else if ((r = sshbuf_consume(stderr_buffer, len)) != 0) fatal_fr(r, "sshbuf_consume"); } /* Clear and free any buffers. */ sshbuf_free(stderr_buffer); /* Report bytes transferred, and transfer rates. */ total_time = monotime_double() - start_time; ssh_packet_get_bytes(ssh, &ibytes, &obytes); verbose("Transferred: sent %llu, received %llu bytes, in %.1f seconds", (unsigned long long)obytes, (unsigned long long)ibytes, total_time); if (total_time > 0) verbose("Bytes per second: sent %.1f, received %.1f", obytes / total_time, ibytes / total_time); /* Return the exit status of the program. */ debug("Exit status %d", exit_status); return exit_status; } /*********/ static Channel * client_request_forwarded_tcpip(struct ssh *ssh, const char *request_type, int rchan, u_int rwindow, u_int rmaxpack) { Channel *c = NULL; struct sshbuf *b = NULL; char *listen_address, *originator_address; u_int listen_port, originator_port; int r; /* Get rest of the packet */ if ((r = sshpkt_get_cstring(ssh, &listen_address, NULL)) != 0 || (r = sshpkt_get_u32(ssh, &listen_port)) != 0 || (r = sshpkt_get_cstring(ssh, &originator_address, NULL)) != 0 || (r = sshpkt_get_u32(ssh, &originator_port)) != 0 || (r = sshpkt_get_end(ssh)) != 0) fatal_fr(r, "parse packet"); debug_f("listen %s port %d, originator %s port %d", listen_address, listen_port, originator_address, originator_port); if (listen_port > 0xffff) error_f("invalid listen port"); else if (originator_port > 0xffff) error_f("invalid originator port"); else { c = channel_connect_by_listen_address(ssh, listen_address, listen_port, "forwarded-tcpip", originator_address); } if (c != NULL && c->type == SSH_CHANNEL_MUX_CLIENT) { if ((b = sshbuf_new()) == NULL) { error_f("alloc reply"); goto out; } /* reconstruct and send to muxclient */ if ((r = sshbuf_put_u8(b, 0)) != 0 || /* padlen */ (r = sshbuf_put_u8(b, SSH2_MSG_CHANNEL_OPEN)) != 0 || (r = sshbuf_put_cstring(b, request_type)) != 0 || (r = sshbuf_put_u32(b, rchan)) != 0 || (r = sshbuf_put_u32(b, rwindow)) != 0 || (r = sshbuf_put_u32(b, rmaxpack)) != 0 || (r = sshbuf_put_cstring(b, listen_address)) != 0 || (r = sshbuf_put_u32(b, listen_port)) != 0 || (r = sshbuf_put_cstring(b, originator_address)) != 0 || (r = sshbuf_put_u32(b, originator_port)) != 0 || (r = sshbuf_put_stringb(c->output, b)) != 0) { error_fr(r, "compose for muxclient"); goto out; } } out: sshbuf_free(b); free(originator_address); free(listen_address); return c; } static Channel * client_request_forwarded_streamlocal(struct ssh *ssh, const char *request_type, int rchan) { Channel *c = NULL; char *listen_path; int r; /* Get the remote path. */ if ((r = sshpkt_get_cstring(ssh, &listen_path, NULL)) != 0 || (r = sshpkt_get_string(ssh, NULL, NULL)) != 0 || /* reserved */ (r = sshpkt_get_end(ssh)) != 0) fatal_fr(r, "parse packet"); debug_f("request: %s", listen_path); c = channel_connect_by_listen_path(ssh, listen_path, "forwarded-streamlocal@openssh.com", "forwarded-streamlocal"); free(listen_path); return c; } static Channel * client_request_x11(struct ssh *ssh, const char *request_type, int rchan) { Channel *c = NULL; char *originator; u_int originator_port; int r, sock; if (!options.forward_x11) { error("Warning: ssh server tried X11 forwarding."); error("Warning: this is probably a break-in attempt by a " "malicious server."); return NULL; } if (x11_refuse_time != 0 && monotime() >= x11_refuse_time) { verbose("Rejected X11 connection after ForwardX11Timeout " "expired"); return NULL; } if ((r = sshpkt_get_cstring(ssh, &originator, NULL)) != 0 || (r = sshpkt_get_u32(ssh, &originator_port)) != 0 || (r = sshpkt_get_end(ssh)) != 0) fatal_fr(r, "parse packet"); /* XXX check permission */ /* XXX range check originator port? */ debug("client_request_x11: request from %s %u", originator, originator_port); free(originator); sock = x11_connect_display(ssh); if (sock < 0) return NULL; /* again is this really necessary for X11? */ if (options.hpn_disabled) c = channel_new(ssh, "x11-connection", SSH_CHANNEL_X11_OPEN, sock, sock, -1, CHAN_TCP_WINDOW_DEFAULT, CHAN_X11_PACKET_DEFAULT, 0, "x11", 1); else c = channel_new(ssh, "x11-connection", SSH_CHANNEL_X11_OPEN, sock, sock, -1, options.hpn_buffer_size, CHAN_X11_PACKET_DEFAULT, 0, "x11", 1); c->force_drain = 1; return c; } static Channel * client_request_agent(struct ssh *ssh, const char *request_type, int rchan) { Channel *c = NULL; int r, sock; if (!options.forward_agent) { error("Warning: ssh server tried agent forwarding."); error("Warning: this is probably a break-in attempt by a " "malicious server."); return NULL; } if (forward_agent_sock_path == NULL) { r = ssh_get_authentication_socket(&sock); } else { r = ssh_get_authentication_socket_path(forward_agent_sock_path, &sock); } if (r != 0) { if (r != SSH_ERR_AGENT_NOT_PRESENT) debug_fr(r, "ssh_get_authentication_socket"); return NULL; } if ((r = ssh_agent_bind_hostkey(sock, ssh->kex->initial_hostkey, ssh->kex->session_id, ssh->kex->initial_sig, 1)) == 0) debug_f("bound agent to hostkey"); else debug2_fr(r, "ssh_agent_bind_hostkey"); if (options.hpn_disabled) c = channel_new(ssh, "agent-connection", SSH_CHANNEL_OPEN, sock, sock, -1, CHAN_X11_WINDOW_DEFAULT, CHAN_TCP_PACKET_DEFAULT, 0, "authentication agent connection", 1); else c = channel_new(ssh, "agent connection", SSH_CHANNEL_OPEN, sock, sock, -1, options.hpn_buffer_size, options.hpn_buffer_size, 0, "authentication agent connection", 1); c->force_drain = 1; return c; } char * client_request_tun_fwd(struct ssh *ssh, int tun_mode, int local_tun, int remote_tun, channel_open_fn *cb, void *cbctx) { Channel *c; int r, fd; char *ifname = NULL; if (tun_mode == SSH_TUNMODE_NO) return 0; debug("Requesting tun unit %d in mode %d", local_tun, tun_mode); /* Open local tunnel device */ if ((fd = tun_open(local_tun, tun_mode, &ifname)) == -1) { error("Tunnel device open failed."); return NULL; } debug("Tunnel forwarding using interface %s", ifname); if(options.hpn_disabled) c = channel_new(ssh, "tun-connection", SSH_CHANNEL_OPENING, fd, fd, -1, CHAN_TCP_WINDOW_DEFAULT, CHAN_TCP_PACKET_DEFAULT, 0, "tun", 1); else c = channel_new(ssh, "tun-connection", SSH_CHANNEL_OPENING, fd, fd, -1, options.hpn_buffer_size, CHAN_TCP_PACKET_DEFAULT, 0, "tun", 1); c->datagram = 1; if (cb != NULL) channel_register_open_confirm(ssh, c->self, cb, cbctx); if ((r = sshpkt_start(ssh, SSH2_MSG_CHANNEL_OPEN)) != 0 || (r = sshpkt_put_cstring(ssh, "tun@openssh.com")) != 0 || (r = sshpkt_put_u32(ssh, c->self)) != 0 || (r = sshpkt_put_u32(ssh, c->local_window_max)) != 0 || (r = sshpkt_put_u32(ssh, c->local_maxpacket)) != 0 || (r = sshpkt_put_u32(ssh, tun_mode)) != 0 || (r = sshpkt_put_u32(ssh, remote_tun)) != 0 || (r = sshpkt_send(ssh)) != 0) sshpkt_fatal(ssh, r, "%s: send reply", __func__); return ifname; } /* XXXX move to generic input handler */ static int client_input_channel_open(int type, u_int32_t seq, struct ssh *ssh) { Channel *c = NULL; char *ctype = NULL; int r; u_int rchan; size_t len; u_int rmaxpack, rwindow; if ((r = sshpkt_get_cstring(ssh, &ctype, &len)) != 0 || (r = sshpkt_get_u32(ssh, &rchan)) != 0 || (r = sshpkt_get_u32(ssh, &rwindow)) != 0 || (r = sshpkt_get_u32(ssh, &rmaxpack)) != 0) goto out; debug("client_input_channel_open: ctype %s rchan %d win %d max %d", ctype, rchan, rwindow, rmaxpack); if (strcmp(ctype, "forwarded-tcpip") == 0) { c = client_request_forwarded_tcpip(ssh, ctype, rchan, rwindow, rmaxpack); } else if (strcmp(ctype, "forwarded-streamlocal@openssh.com") == 0) { c = client_request_forwarded_streamlocal(ssh, ctype, rchan); } else if (strcmp(ctype, "x11") == 0) { c = client_request_x11(ssh, ctype, rchan); } else if (strcmp(ctype, "auth-agent@openssh.com") == 0) { c = client_request_agent(ssh, ctype, rchan); } if (c != NULL && c->type == SSH_CHANNEL_MUX_CLIENT) { debug3("proxied to downstream: %s", ctype); } else if (c != NULL) { debug("confirm %s", ctype); c->remote_id = rchan; c->have_remote_id = 1; c->remote_window = rwindow; c->remote_maxpacket = rmaxpack; if (c->type != SSH_CHANNEL_CONNECTING) { if ((r = sshpkt_start(ssh, SSH2_MSG_CHANNEL_OPEN_CONFIRMATION)) != 0 || (r = sshpkt_put_u32(ssh, c->remote_id)) != 0 || (r = sshpkt_put_u32(ssh, c->self)) != 0 || (r = sshpkt_put_u32(ssh, c->local_window)) != 0 || (r = sshpkt_put_u32(ssh, c->local_maxpacket)) != 0 || (r = sshpkt_send(ssh)) != 0) sshpkt_fatal(ssh, r, "%s: send reply", __func__); } } else { debug("failure %s", ctype); if ((r = sshpkt_start(ssh, SSH2_MSG_CHANNEL_OPEN_FAILURE)) != 0 || (r = sshpkt_put_u32(ssh, rchan)) != 0 || (r = sshpkt_put_u32(ssh, SSH2_OPEN_ADMINISTRATIVELY_PROHIBITED)) != 0 || (r = sshpkt_put_cstring(ssh, "open failed")) != 0 || (r = sshpkt_put_cstring(ssh, "")) != 0 || (r = sshpkt_send(ssh)) != 0) sshpkt_fatal(ssh, r, "%s: send failure", __func__); } r = 0; out: free(ctype); return r; } static int client_input_channel_req(int type, u_int32_t seq, struct ssh *ssh) { Channel *c = NULL; char *rtype = NULL; u_char reply; u_int id, exitval; int r, success = 0; if ((r = sshpkt_get_u32(ssh, &id)) != 0) return r; if (id <= INT_MAX) c = channel_lookup(ssh, id); if (channel_proxy_upstream(c, type, seq, ssh)) return 0; if ((r = sshpkt_get_cstring(ssh, &rtype, NULL)) != 0 || (r = sshpkt_get_u8(ssh, &reply)) != 0) goto out; debug("client_input_channel_req: channel %u rtype %s reply %d", id, rtype, reply); if (c == NULL) { error("client_input_channel_req: channel %d: " "unknown channel", id); } else if (strcmp(rtype, "eow@openssh.com") == 0) { if ((r = sshpkt_get_end(ssh)) != 0) goto out; chan_rcvd_eow(ssh, c); } else if (strcmp(rtype, "exit-status") == 0) { if ((r = sshpkt_get_u32(ssh, &exitval)) != 0) goto out; if (c->ctl_chan != -1) { mux_exit_message(ssh, c, exitval); success = 1; } else if ((int)id == session_ident) { /* Record exit value of local session */ success = 1; exit_status = exitval; } else { /* Probably for a mux channel that has already closed */ debug_f("no sink for exit-status on channel %d", id); } if ((r = sshpkt_get_end(ssh)) != 0) goto out; } if (reply && c != NULL && !(c->flags & CHAN_CLOSE_SENT)) { if (!c->have_remote_id) fatal_f("channel %d: no remote_id", c->self); if ((r = sshpkt_start(ssh, success ? SSH2_MSG_CHANNEL_SUCCESS : SSH2_MSG_CHANNEL_FAILURE)) != 0 || (r = sshpkt_put_u32(ssh, c->remote_id)) != 0 || (r = sshpkt_send(ssh)) != 0) sshpkt_fatal(ssh, r, "%s: send failure", __func__); } r = 0; out: free(rtype); return r; } struct hostkeys_update_ctx { /* The hostname and (optionally) IP address string for the server */ char *host_str, *ip_str; /* * Keys received from the server and a flag for each indicating * whether they already exist in known_hosts. * keys_match is filled in by hostkeys_find() and later (for new * keys) by client_global_hostkeys_prove_confirm(). */ struct sshkey **keys; u_int *keys_match; /* mask of HKF_MATCH_* from hostfile.h */ int *keys_verified; /* flag for new keys verified by server */ size_t nkeys, nnew, nincomplete; /* total, new keys, incomplete match */ /* * Keys that are in known_hosts, but were not present in the update * from the server (i.e. scheduled to be deleted). * Filled in by hostkeys_find(). */ struct sshkey **old_keys; size_t nold; /* Various special cases. */ int complex_hostspec; /* wildcard or manual pattern-list host name */ int ca_available; /* saw CA key for this host */ int old_key_seen; /* saw old key with other name/addr */ int other_name_seen; /* saw key with other name/addr */ }; static void hostkeys_update_ctx_free(struct hostkeys_update_ctx *ctx) { size_t i; if (ctx == NULL) return; for (i = 0; i < ctx->nkeys; i++) sshkey_free(ctx->keys[i]); free(ctx->keys); free(ctx->keys_match); free(ctx->keys_verified); for (i = 0; i < ctx->nold; i++) sshkey_free(ctx->old_keys[i]); free(ctx->old_keys); free(ctx->host_str); free(ctx->ip_str); free(ctx); } /* * Returns non-zero if a known_hosts hostname list is not of a form that * can be handled by UpdateHostkeys. These include wildcard hostnames and * hostnames lists that do not follow the form host[,ip]. */ static int hostspec_is_complex(const char *hosts) { char *cp; /* wildcard */ if (strchr(hosts, '*') != NULL || strchr(hosts, '?') != NULL) return 1; /* single host/ip = ok */ if ((cp = strchr(hosts, ',')) == NULL) return 0; /* more than two entries on the line */ if (strchr(cp + 1, ',') != NULL) return 1; /* XXX maybe parse cp+1 and ensure it is an IP? */ return 0; } /* callback to search for ctx->keys in known_hosts */ static int hostkeys_find(struct hostkey_foreach_line *l, void *_ctx) { struct hostkeys_update_ctx *ctx = (struct hostkeys_update_ctx *)_ctx; size_t i; struct sshkey **tmp; if (l->key == NULL) return 0; if (l->status != HKF_STATUS_MATCHED) { /* Record if one of the keys appears on a non-matching line */ for (i = 0; i < ctx->nkeys; i++) { if (sshkey_equal(l->key, ctx->keys[i])) { ctx->other_name_seen = 1; debug3_f("found %s key under different " "name/addr at %s:%ld", sshkey_ssh_name(ctx->keys[i]), l->path, l->linenum); return 0; } } return 0; } /* Don't proceed if revocation or CA markers are present */ /* XXX relax this */ if (l->marker != MRK_NONE) { debug3_f("hostkeys file %s:%ld has CA/revocation marker", l->path, l->linenum); ctx->complex_hostspec = 1; return 0; } /* If CheckHostIP is enabled, then check for mismatched hostname/addr */ if (ctx->ip_str != NULL && strchr(l->hosts, ',') != NULL) { if ((l->match & HKF_MATCH_HOST) == 0) { /* Record if address matched a different hostname. */ ctx->other_name_seen = 1; debug3_f("found address %s against different hostname " "at %s:%ld", ctx->ip_str, l->path, l->linenum); return 0; } else if ((l->match & HKF_MATCH_IP) == 0) { /* Record if hostname matched a different address. */ ctx->other_name_seen = 1; debug3_f("found hostname %s against different address " "at %s:%ld", ctx->host_str, l->path, l->linenum); } } /* * UpdateHostkeys is skipped for wildcard host names and hostnames * that contain more than two entries (ssh never writes these). */ if (hostspec_is_complex(l->hosts)) { debug3_f("hostkeys file %s:%ld complex host specification", l->path, l->linenum); ctx->complex_hostspec = 1; return 0; } /* Mark off keys we've already seen for this host */ for (i = 0; i < ctx->nkeys; i++) { if (!sshkey_equal(l->key, ctx->keys[i])) continue; debug3_f("found %s key at %s:%ld", sshkey_ssh_name(ctx->keys[i]), l->path, l->linenum); ctx->keys_match[i] |= l->match; return 0; } /* This line contained a key that not offered by the server */ debug3_f("deprecated %s key at %s:%ld", sshkey_ssh_name(l->key), l->path, l->linenum); if ((tmp = recallocarray(ctx->old_keys, ctx->nold, ctx->nold + 1, sizeof(*ctx->old_keys))) == NULL) fatal_f("recallocarray failed nold = %zu", ctx->nold); ctx->old_keys = tmp; ctx->old_keys[ctx->nold++] = l->key; l->key = NULL; return 0; } /* callback to search for ctx->old_keys in known_hosts under other names */ static int hostkeys_check_old(struct hostkey_foreach_line *l, void *_ctx) { struct hostkeys_update_ctx *ctx = (struct hostkeys_update_ctx *)_ctx; size_t i; int hashed; /* only care about lines that *don't* match the active host spec */ if (l->status == HKF_STATUS_MATCHED || l->key == NULL) return 0; hashed = l->match & (HKF_MATCH_HOST_HASHED|HKF_MATCH_IP_HASHED); for (i = 0; i < ctx->nold; i++) { if (!sshkey_equal(l->key, ctx->old_keys[i])) continue; debug3_f("found deprecated %s key at %s:%ld as %s", sshkey_ssh_name(ctx->old_keys[i]), l->path, l->linenum, hashed ? "[HASHED]" : l->hosts); ctx->old_key_seen = 1; break; } return 0; } /* * Check known_hosts files for deprecated keys under other names. Returns 0 * on success or -1 on failure. Updates ctx->old_key_seen if deprecated keys * exist under names other than the active hostname/IP. */ static int check_old_keys_othernames(struct hostkeys_update_ctx *ctx) { size_t i; int r; debug2_f("checking for %zu deprecated keys", ctx->nold); for (i = 0; i < options.num_user_hostfiles; i++) { debug3_f("searching %s for %s / %s", options.user_hostfiles[i], ctx->host_str, ctx->ip_str ? ctx->ip_str : "(none)"); if ((r = hostkeys_foreach(options.user_hostfiles[i], hostkeys_check_old, ctx, ctx->host_str, ctx->ip_str, HKF_WANT_PARSE_KEY, 0)) != 0) { if (r == SSH_ERR_SYSTEM_ERROR && errno == ENOENT) { debug_f("hostkeys file %s does not exist", options.user_hostfiles[i]); continue; } error_fr(r, "hostkeys_foreach failed for %s", options.user_hostfiles[i]); return -1; } } return 0; } static void hostkey_change_preamble(LogLevel loglevel) { do_log2(loglevel, "The server has updated its host keys."); do_log2(loglevel, "These changes were verified by the server's " "existing trusted key."); } static void update_known_hosts(struct hostkeys_update_ctx *ctx) { int r, was_raw = 0, first = 1; int asking = options.update_hostkeys == SSH_UPDATE_HOSTKEYS_ASK; LogLevel loglevel = asking ? SYSLOG_LEVEL_INFO : SYSLOG_LEVEL_VERBOSE; char *fp, *response; size_t i; struct stat sb; for (i = 0; i < ctx->nkeys; i++) { if (!ctx->keys_verified[i]) continue; if ((fp = sshkey_fingerprint(ctx->keys[i], options.fingerprint_hash, SSH_FP_DEFAULT)) == NULL) fatal_f("sshkey_fingerprint failed"); if (first && asking) hostkey_change_preamble(loglevel); do_log2(loglevel, "Learned new hostkey: %s %s", sshkey_type(ctx->keys[i]), fp); first = 0; free(fp); } for (i = 0; i < ctx->nold; i++) { if ((fp = sshkey_fingerprint(ctx->old_keys[i], options.fingerprint_hash, SSH_FP_DEFAULT)) == NULL) fatal_f("sshkey_fingerprint failed"); if (first && asking) hostkey_change_preamble(loglevel); do_log2(loglevel, "Deprecating obsolete hostkey: %s %s", sshkey_type(ctx->old_keys[i]), fp); first = 0; free(fp); } if (options.update_hostkeys == SSH_UPDATE_HOSTKEYS_ASK) { if (get_saved_tio() != NULL) { leave_raw_mode(1); was_raw = 1; } response = NULL; for (i = 0; !quit_pending && i < 3; i++) { free(response); response = read_passphrase("Accept updated hostkeys? " "(yes/no): ", RP_ECHO); if (response != NULL && strcasecmp(response, "yes") == 0) break; else if (quit_pending || response == NULL || strcasecmp(response, "no") == 0) { options.update_hostkeys = 0; break; } else { do_log2(loglevel, "Please enter " "\"yes\" or \"no\""); } } if (quit_pending || i >= 3 || response == NULL) options.update_hostkeys = 0; free(response); if (was_raw) enter_raw_mode(1); } if (options.update_hostkeys == 0) return; /* * Now that all the keys are verified, we can go ahead and replace * them in known_hosts (assuming SSH_UPDATE_HOSTKEYS_ASK didn't * cancel the operation). */ for (i = 0; i < options.num_user_hostfiles; i++) { /* * NB. keys are only added to hostfiles[0], for the rest we * just delete the hostname entries. */ if (stat(options.user_hostfiles[i], &sb) != 0) { if (errno == ENOENT) { debug_f("known hosts file %s does not " "exist", options.user_hostfiles[i]); } else { error_f("known hosts file %s " "inaccessible: %s", options.user_hostfiles[i], strerror(errno)); } continue; } if ((r = hostfile_replace_entries(options.user_hostfiles[i], ctx->host_str, ctx->ip_str, i == 0 ? ctx->keys : NULL, i == 0 ? ctx->nkeys : 0, options.hash_known_hosts, 0, options.fingerprint_hash)) != 0) { error_fr(r, "hostfile_replace_entries failed for %s", options.user_hostfiles[i]); } } } static void client_global_hostkeys_prove_confirm(struct ssh *ssh, int type, u_int32_t seq, void *_ctx) { struct hostkeys_update_ctx *ctx = (struct hostkeys_update_ctx *)_ctx; size_t i, ndone; struct sshbuf *signdata; int r, plaintype; const u_char *sig; const char *rsa_kexalg = NULL; char *alg = NULL; size_t siglen; if (ctx->nnew == 0) fatal_f("ctx->nnew == 0"); /* sanity */ if (type != SSH2_MSG_REQUEST_SUCCESS) { error("Server failed to confirm ownership of " "private host keys"); hostkeys_update_ctx_free(ctx); return; } if (sshkey_type_plain(sshkey_type_from_name( ssh->kex->hostkey_alg)) == KEY_RSA) rsa_kexalg = ssh->kex->hostkey_alg; if ((signdata = sshbuf_new()) == NULL) fatal_f("sshbuf_new failed"); /* * Expect a signature for each of the ctx->nnew private keys we * haven't seen before. They will be in the same order as the * ctx->keys where the corresponding ctx->keys_match[i] == 0. */ for (ndone = i = 0; i < ctx->nkeys; i++) { if (ctx->keys_match[i]) continue; plaintype = sshkey_type_plain(ctx->keys[i]->type); /* Prepare data to be signed: session ID, unique string, key */ sshbuf_reset(signdata); if ( (r = sshbuf_put_cstring(signdata, "hostkeys-prove-00@openssh.com")) != 0 || (r = sshbuf_put_stringb(signdata, ssh->kex->session_id)) != 0 || (r = sshkey_puts(ctx->keys[i], signdata)) != 0) fatal_fr(r, "compose signdata"); /* Extract and verify signature */ if ((r = sshpkt_get_string_direct(ssh, &sig, &siglen)) != 0) { error_fr(r, "parse sig"); goto out; } if ((r = sshkey_get_sigtype(sig, siglen, &alg)) != 0) { error_fr(r, "server gave unintelligible signature " "for %s key %zu", sshkey_type(ctx->keys[i]), i); goto out; } /* * Special case for RSA keys: if a RSA hostkey was negotiated, * then use its signature type for verification of RSA hostkey * proofs. Otherwise, accept only RSA-SHA256/512 signatures. */ if (plaintype == KEY_RSA && rsa_kexalg == NULL && match_pattern_list(alg, HOSTKEY_PROOF_RSA_ALGS, 0) != 1) { debug_f("server used untrusted RSA signature algorithm " "%s for key %zu, disregarding", alg, i); free(alg); /* zap the key from the list */ sshkey_free(ctx->keys[i]); ctx->keys[i] = NULL; ndone++; continue; } debug3_f("verify %s key %zu using sigalg %s", sshkey_type(ctx->keys[i]), i, alg); free(alg); if ((r = sshkey_verify(ctx->keys[i], sig, siglen, sshbuf_ptr(signdata), sshbuf_len(signdata), plaintype == KEY_RSA ? rsa_kexalg : NULL, 0, NULL)) != 0) { error_fr(r, "server gave bad signature for %s key %zu", sshkey_type(ctx->keys[i]), i); goto out; } /* Key is good. Mark it as 'seen' */ ctx->keys_verified[i] = 1; ndone++; } /* Shouldn't happen */ if (ndone != ctx->nnew) fatal_f("ndone != ctx->nnew (%zu / %zu)", ndone, ctx->nnew); if ((r = sshpkt_get_end(ssh)) != 0) { error_f("protocol error"); goto out; } /* Make the edits to known_hosts */ update_known_hosts(ctx); out: hostkeys_update_ctx_free(ctx); hostkeys_update_complete = 1; client_repledge(); } /* * Returns non-zero if the key is accepted by HostkeyAlgorithms. * Made slightly less trivial by the multiple RSA signature algorithm names. */ static int key_accepted_by_hostkeyalgs(const struct sshkey *key) { const char *ktype = sshkey_ssh_name(key); const char *hostkeyalgs = options.hostkeyalgorithms; if (key->type == KEY_UNSPEC) return 0; if (key->type == KEY_RSA && (match_pattern_list("rsa-sha2-256", hostkeyalgs, 0) == 1 || match_pattern_list("rsa-sha2-512", hostkeyalgs, 0) == 1)) return 1; return match_pattern_list(ktype, hostkeyalgs, 0) == 1; } /* * Handle hostkeys-00@openssh.com global request to inform the client of all * the server's hostkeys. The keys are checked against the user's * HostkeyAlgorithms preference before they are accepted. */ static int client_input_hostkeys(struct ssh *ssh) { const u_char *blob = NULL; size_t i, len = 0; struct sshbuf *buf = NULL; struct sshkey *key = NULL, **tmp; int r, prove_sent = 0; char *fp; static int hostkeys_seen = 0; /* XXX use struct ssh */ extern struct sockaddr_storage hostaddr; /* XXX from ssh.c */ struct hostkeys_update_ctx *ctx = NULL; u_int want; if (hostkeys_seen) fatal_f("server already sent hostkeys"); if (!can_update_hostkeys()) return 1; hostkeys_seen = 1; ctx = xcalloc(1, sizeof(*ctx)); while (ssh_packet_remaining(ssh) > 0) { sshkey_free(key); key = NULL; if ((r = sshpkt_get_string_direct(ssh, &blob, &len)) != 0) { error_fr(r, "parse key"); goto out; } if ((r = sshkey_from_blob(blob, len, &key)) != 0) { do_log2_fr(r, r == SSH_ERR_KEY_TYPE_UNKNOWN ? SYSLOG_LEVEL_DEBUG1 : SYSLOG_LEVEL_ERROR, "convert key"); continue; } fp = sshkey_fingerprint(key, options.fingerprint_hash, SSH_FP_DEFAULT); debug3_f("received %s key %s", sshkey_type(key), fp); free(fp); if (!key_accepted_by_hostkeyalgs(key)) { debug3_f("%s key not permitted by " "HostkeyAlgorithms", sshkey_ssh_name(key)); continue; } /* Skip certs */ if (sshkey_is_cert(key)) { debug3_f("%s key is a certificate; skipping", sshkey_ssh_name(key)); continue; } /* Ensure keys are unique */ for (i = 0; i < ctx->nkeys; i++) { if (sshkey_equal(key, ctx->keys[i])) { error_f("received duplicated %s host key", sshkey_ssh_name(key)); goto out; } } /* Key is good, record it */ if ((tmp = recallocarray(ctx->keys, ctx->nkeys, ctx->nkeys + 1, sizeof(*ctx->keys))) == NULL) fatal_f("recallocarray failed nkeys = %zu", ctx->nkeys); ctx->keys = tmp; ctx->keys[ctx->nkeys++] = key; key = NULL; } if (ctx->nkeys == 0) { debug_f("server sent no hostkeys"); goto out; } if ((ctx->keys_match = calloc(ctx->nkeys, sizeof(*ctx->keys_match))) == NULL || (ctx->keys_verified = calloc(ctx->nkeys, sizeof(*ctx->keys_verified))) == NULL) fatal_f("calloc failed"); get_hostfile_hostname_ipaddr(host, options.check_host_ip ? (struct sockaddr *)&hostaddr : NULL, options.port, &ctx->host_str, options.check_host_ip ? &ctx->ip_str : NULL); /* Find which keys we already know about. */ for (i = 0; i < options.num_user_hostfiles; i++) { debug_f("searching %s for %s / %s", options.user_hostfiles[i], ctx->host_str, ctx->ip_str ? ctx->ip_str : "(none)"); if ((r = hostkeys_foreach(options.user_hostfiles[i], hostkeys_find, ctx, ctx->host_str, ctx->ip_str, HKF_WANT_PARSE_KEY, 0)) != 0) { if (r == SSH_ERR_SYSTEM_ERROR && errno == ENOENT) { debug_f("hostkeys file %s does not exist", options.user_hostfiles[i]); continue; } error_fr(r, "hostkeys_foreach failed for %s", options.user_hostfiles[i]); goto out; } } /* Figure out if we have any new keys to add */ ctx->nnew = ctx->nincomplete = 0; want = HKF_MATCH_HOST | ( options.check_host_ip ? HKF_MATCH_IP : 0); for (i = 0; i < ctx->nkeys; i++) { if (ctx->keys_match[i] == 0) ctx->nnew++; if ((ctx->keys_match[i] & want) != want) ctx->nincomplete++; } debug3_f("%zu server keys: %zu new, %zu retained, " "%zu incomplete match. %zu to remove", ctx->nkeys, ctx->nnew, ctx->nkeys - ctx->nnew - ctx->nincomplete, ctx->nincomplete, ctx->nold); if (ctx->nnew == 0 && ctx->nold == 0) { debug_f("no new or deprecated keys from server"); goto out; } /* Various reasons why we cannot proceed with the update */ if (ctx->complex_hostspec) { debug_f("CA/revocation marker, manual host list or wildcard " "host pattern found, skipping UserKnownHostsFile update"); goto out; } if (ctx->other_name_seen) { debug_f("host key found matching a different name/address, " "skipping UserKnownHostsFile update"); goto out; } /* * If removing keys, check whether they appear under different * names/addresses and refuse to proceed if they do. This avoids * cases such as hosts with multiple names becoming inconsistent * with regards to CheckHostIP entries. * XXX UpdateHostkeys=force to override this (and other) checks? */ if (ctx->nold != 0) { if (check_old_keys_othernames(ctx) != 0) goto out; /* error already logged */ if (ctx->old_key_seen) { debug_f("key(s) for %s%s%s exist under other names; " "skipping UserKnownHostsFile update", ctx->host_str, ctx->ip_str == NULL ? "" : ",", ctx->ip_str == NULL ? "" : ctx->ip_str); goto out; } } if (ctx->nnew == 0) { /* * We have some keys to remove or fix matching for. * We can proceed to do this without requiring a fresh proof * from the server. */ update_known_hosts(ctx); goto out; } /* * We have received previously-unseen keys from the server. * Ask the server to confirm ownership of the private halves. */ debug3_f("asking server to prove ownership for %zu keys", ctx->nnew); if ((r = sshpkt_start(ssh, SSH2_MSG_GLOBAL_REQUEST)) != 0 || (r = sshpkt_put_cstring(ssh, "hostkeys-prove-00@openssh.com")) != 0 || (r = sshpkt_put_u8(ssh, 1)) != 0) /* bool: want reply */ fatal_fr(r, "prepare hostkeys-prove"); if ((buf = sshbuf_new()) == NULL) fatal_f("sshbuf_new"); for (i = 0; i < ctx->nkeys; i++) { if (ctx->keys_match[i]) continue; sshbuf_reset(buf); if ((r = sshkey_putb(ctx->keys[i], buf)) != 0 || (r = sshpkt_put_stringb(ssh, buf)) != 0) fatal_fr(r, "assemble hostkeys-prove"); } if ((r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send hostkeys-prove"); client_register_global_confirm( client_global_hostkeys_prove_confirm, ctx); ctx = NULL; /* will be freed in callback */ prove_sent = 1; /* Success */ out: hostkeys_update_ctx_free(ctx); sshkey_free(key); sshbuf_free(buf); if (!prove_sent) { /* UpdateHostkeys handling completed */ hostkeys_update_complete = 1; client_repledge(); } /* * NB. Return success for all cases. The server doesn't need to know * what the client does with its hosts file. */ return 1; } static int client_input_global_request(int type, u_int32_t seq, struct ssh *ssh) { char *rtype; u_char want_reply; int r, success = 0; if ((r = sshpkt_get_cstring(ssh, &rtype, NULL)) != 0 || (r = sshpkt_get_u8(ssh, &want_reply)) != 0) goto out; debug("client_input_global_request: rtype %s want_reply %d", rtype, want_reply); if (strcmp(rtype, "hostkeys-00@openssh.com") == 0) success = client_input_hostkeys(ssh); if (want_reply) { if ((r = sshpkt_start(ssh, success ? SSH2_MSG_REQUEST_SUCCESS : SSH2_MSG_REQUEST_FAILURE)) != 0 || (r = sshpkt_send(ssh)) != 0 || (r = ssh_packet_write_wait(ssh)) != 0) goto out; } r = 0; out: free(rtype); return r; } static void client_send_env(struct ssh *ssh, int id, const char *name, const char *val) { int r; debug("channel %d: setting env %s = \"%s\"", id, name, val); channel_request_start(ssh, id, "env", 0); if ((r = sshpkt_put_cstring(ssh, name)) != 0 || (r = sshpkt_put_cstring(ssh, val)) != 0 || (r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send setenv"); } void client_session2_setup(struct ssh *ssh, int id, int want_tty, int want_subsystem, const char *term, struct termios *tiop, int in_fd, struct sshbuf *cmd, char **env) { size_t i, j, len; int matched, r; char *name, *val; Channel *c = NULL; debug2_f("id %d", id); if ((c = channel_lookup(ssh, id)) == NULL) fatal_f("channel %d: unknown channel", id); ssh_packet_set_interactive(ssh, want_tty, options.ip_qos_interactive, options.ip_qos_bulk); if (want_tty) { struct winsize ws; /* Store window size in the packet. */ if (ioctl(in_fd, TIOCGWINSZ, &ws) == -1) memset(&ws, 0, sizeof(ws)); channel_request_start(ssh, id, "pty-req", 1); client_expect_confirm(ssh, id, "PTY allocation", CONFIRM_TTY); if ((r = sshpkt_put_cstring(ssh, term != NULL ? term : "")) != 0 || (r = sshpkt_put_u32(ssh, (u_int)ws.ws_col)) != 0 || (r = sshpkt_put_u32(ssh, (u_int)ws.ws_row)) != 0 || (r = sshpkt_put_u32(ssh, (u_int)ws.ws_xpixel)) != 0 || (r = sshpkt_put_u32(ssh, (u_int)ws.ws_ypixel)) != 0) fatal_fr(r, "build pty-req"); if (tiop == NULL) tiop = get_saved_tio(); ssh_tty_make_modes(ssh, -1, tiop); if ((r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send pty-req"); /* XXX wait for reply */ c->client_tty = 1; } /* Transfer any environment variables from client to server */ if (options.num_send_env != 0 && env != NULL) { debug("Sending environment."); for (i = 0; env[i] != NULL; i++) { /* Split */ name = xstrdup(env[i]); if ((val = strchr(name, '=')) == NULL) { free(name); continue; } *val++ = '\0'; matched = 0; for (j = 0; j < options.num_send_env; j++) { if (match_pattern(name, options.send_env[j])) { matched = 1; break; } } if (!matched) { debug3("Ignored env %s", name); free(name); continue; } client_send_env(ssh, id, name, val); free(name); } } for (i = 0; i < options.num_setenv; i++) { /* Split */ name = xstrdup(options.setenv[i]); if ((val = strchr(name, '=')) == NULL) { free(name); continue; } *val++ = '\0'; client_send_env(ssh, id, name, val); free(name); } len = sshbuf_len(cmd); if (len > 0) { if (len > 900) len = 900; if (want_subsystem) { debug("Sending subsystem: %.*s", (int)len, (const u_char*)sshbuf_ptr(cmd)); channel_request_start(ssh, id, "subsystem", 1); client_expect_confirm(ssh, id, "subsystem", CONFIRM_CLOSE); } else { debug("Sending command: %.*s", (int)len, (const u_char*)sshbuf_ptr(cmd)); channel_request_start(ssh, id, "exec", 1); client_expect_confirm(ssh, id, "exec", CONFIRM_CLOSE); } if ((r = sshpkt_put_stringb(ssh, cmd)) != 0 || (r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send command"); } else { channel_request_start(ssh, id, "shell", 1); client_expect_confirm(ssh, id, "shell", CONFIRM_CLOSE); if ((r = sshpkt_send(ssh)) != 0) fatal_fr(r, "send shell"); } session_setup_complete = 1; client_repledge(); } static void client_init_dispatch(struct ssh *ssh) { ssh_dispatch_init(ssh, &dispatch_protocol_error); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_CLOSE, &channel_input_oclose); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_DATA, &channel_input_data); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_EOF, &channel_input_ieof); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_EXTENDED_DATA, &channel_input_extended_data); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_OPEN, &client_input_channel_open); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_OPEN_CONFIRMATION, &channel_input_open_confirmation); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_OPEN_FAILURE, &channel_input_open_failure); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_REQUEST, &client_input_channel_req); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_WINDOW_ADJUST, &channel_input_window_adjust); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_SUCCESS, &channel_input_status_confirm); ssh_dispatch_set(ssh, SSH2_MSG_CHANNEL_FAILURE, &channel_input_status_confirm); ssh_dispatch_set(ssh, SSH2_MSG_GLOBAL_REQUEST, &client_input_global_request); /* rekeying */ ssh_dispatch_set(ssh, SSH2_MSG_KEXINIT, &kex_input_kexinit); /* global request reply messages */ ssh_dispatch_set(ssh, SSH2_MSG_REQUEST_FAILURE, &client_global_request_reply); ssh_dispatch_set(ssh, SSH2_MSG_REQUEST_SUCCESS, &client_global_request_reply); } void client_stop_mux(void) { if (options.control_path != NULL && muxserver_sock != -1) unlink(options.control_path); /* * If we are in persist mode, or don't have a shell, signal that we * should close when all active channels are closed. */ if (options.control_persist || options.session_type == SESSION_TYPE_NONE) { session_closed = 1; setproctitle("[stopped mux]"); } } /* client specific fatal cleanup */ void cleanup_exit(int i) { leave_raw_mode(options.request_tty == REQUEST_TTY_FORCE); if (options.control_path != NULL && muxserver_sock != -1) unlink(options.control_path); ssh_kill_proxy_command(); _exit(i); }