view common-session.c @ 1788:1fc0012b9c38

Fix handling of replies to global requests (#112) The current code assumes that all global requests want / need a reply. This isn't always true and the request itself indicates if it wants a reply or not. It causes a specific problem with [email protected] messages. These are sent by OpenSSH after authentication to inform the client of potential other host keys for the host. This can be used to add a new type of host key or to rotate host keys. The initial information message from the server is sent as a global request, but with want_reply set to false. This means that the server doesn't expect an answer to this message. Instead the client needs to send a prove request as a reply if it wants to receive proof of ownership for the host keys. The bug doesn't cause any current problems with due to how OpenSSH treats receiving the failure message. It instead treats it as a keepalive message and further ignores it. Arguably this is a protocol violation though of Dropbear and it is only accidental that it doesn't cause a problem with OpenSSH. The bug was found when adding host keys support to libssh, which is more strict protocol wise and treats the unexpected failure message an error, also see https://gitlab.com/libssh/libssh-mirror/-/merge_requests/145 for more information. The fix here is to honor the want_reply flag in the global request and to only send a reply if the other side expects a reply.
author Dirkjan Bussink <d.bussink@gmail.com>
date Thu, 10 Dec 2020 16:13:13 +0100
parents 36e77a51d5e4
children 1d86a58fb52d
line wrap: on
line source

/*
 * Dropbear - a SSH2 server
 * 
 * Copyright (c) Matt Johnston
 * All rights reserved.
 * 
 * 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 "includes.h"
#include "session.h"
#include "dbutil.h"
#include "packet.h"
#include "algo.h"
#include "buffer.h"
#include "dss.h"
#include "ssh.h"
#include "dbrandom.h"
#include "kex.h"
#include "channel.h"
#include "runopts.h"
#include "netio.h"

static void checktimeouts(void);
static long select_timeout(void);
static int ident_readln(int fd, char* buf, int count);
static void read_session_identification(void);

struct sshsession ses; /* GLOBAL */

/* called only at the start of a session, set up initial state */
void common_session_init(int sock_in, int sock_out) {
	time_t now;

#if DEBUG_TRACE
	debug_start_net();
#endif

	TRACE(("enter session_init"))

	ses.sock_in = sock_in;
	ses.sock_out = sock_out;
	ses.maxfd = MAX(sock_in, sock_out);

	if (sock_in >= 0) {
		setnonblocking(sock_in);
	}
	if (sock_out >= 0) {
		setnonblocking(sock_out);
	}

	ses.socket_prio = DROPBEAR_PRIO_DEFAULT;
	/* Sets it to lowdelay */
	update_channel_prio();

#if !DROPBEAR_SVR_MULTIUSER
	/* A sanity check to prevent an accidental configuration option
	   leaving multiuser systems exposed */
	errno = 0;
	getuid();
	if (errno != ENOSYS) {
		dropbear_exit("Non-multiuser Dropbear requires a non-multiuser kernel");
	}
#endif

	now = monotonic_now();
	ses.connect_time = now;
	ses.last_packet_time_keepalive_recv = now;
	ses.last_packet_time_idle = now;
	ses.last_packet_time_any_sent = 0;
	ses.last_packet_time_keepalive_sent = 0;
	
#if DROPBEAR_FUZZ
	if (!fuzz.fuzzing)
#endif
	{
	if (pipe(ses.signal_pipe) < 0) {
		dropbear_exit("Signal pipe failed");
	}
	setnonblocking(ses.signal_pipe[0]);
	setnonblocking(ses.signal_pipe[1]);
	ses.maxfd = MAX(ses.maxfd, ses.signal_pipe[0]);
	ses.maxfd = MAX(ses.maxfd, ses.signal_pipe[1]);
	}
	
	ses.writepayload = buf_new(TRANS_MAX_PAYLOAD_LEN);
	ses.transseq = 0;

	ses.readbuf = NULL;
	ses.payload = NULL;
	ses.recvseq = 0;

	initqueue(&ses.writequeue);

	ses.requirenext = SSH_MSG_KEXINIT;
	ses.dataallowed = 1; /* we can send data until we actually 
							send the SSH_MSG_KEXINIT */
	ses.ignorenext = 0;
	ses.lastpacket = 0;
	ses.reply_queue_head = NULL;
	ses.reply_queue_tail = NULL;

	/* set all the algos to none */
	ses.keys = (struct key_context*)m_malloc(sizeof(struct key_context));
	ses.newkeys = NULL;
	ses.keys->recv.algo_crypt = &dropbear_nocipher;
	ses.keys->trans.algo_crypt = &dropbear_nocipher;
	ses.keys->recv.crypt_mode = &dropbear_mode_none;
	ses.keys->trans.crypt_mode = &dropbear_mode_none;
	
	ses.keys->recv.algo_mac = &dropbear_nohash;
	ses.keys->trans.algo_mac = &dropbear_nohash;

	ses.keys->algo_kex = NULL;
	ses.keys->algo_hostkey = -1;
	ses.keys->recv.algo_comp = DROPBEAR_COMP_NONE;
	ses.keys->trans.algo_comp = DROPBEAR_COMP_NONE;

#ifndef DISABLE_ZLIB
	ses.keys->recv.zstream = NULL;
	ses.keys->trans.zstream = NULL;
#endif

	/* key exchange buffers */
	ses.session_id = NULL;
	ses.kexhashbuf = NULL;
	ses.transkexinit = NULL;
	ses.dh_K = NULL;
	ses.remoteident = NULL;

	ses.chantypes = NULL;

	ses.allowprivport = 0;

#if DROPBEAR_PLUGIN
        ses.plugin_session = NULL;
#endif

	TRACE(("leave session_init"))
}

void session_loop(void(*loophandler)(void)) {

	fd_set readfd, writefd;
	struct timeval timeout;
	int val;

	/* main loop, select()s for all sockets in use */
	for(;;) {
		const int writequeue_has_space = (ses.writequeue_len <= 2*TRANS_MAX_PAYLOAD_LEN);

		timeout.tv_sec = select_timeout();
		timeout.tv_usec = 0;
		DROPBEAR_FD_ZERO(&writefd);
		DROPBEAR_FD_ZERO(&readfd);

		dropbear_assert(ses.payload == NULL);

		/* We get woken up when signal handlers write to this pipe.
		   SIGCHLD in svr-chansession is the only one currently. */
#if DROPBEAR_FUZZ
		if (!fuzz.fuzzing) 
#endif
		{
		FD_SET(ses.signal_pipe[0], &readfd);
		}

		/* set up for channels which can be read/written */
		setchannelfds(&readfd, &writefd, writequeue_has_space);

		/* Pending connections to test */
		set_connect_fds(&writefd);

		/* We delay reading from the input socket during initial setup until
		after we have written out our initial KEXINIT packet (empty writequeue). 
		This means our initial packet can be in-flight while we're doing a blocking
		read for the remote ident.
		We also avoid reading from the socket if the writequeue is full, that avoids
		replies backing up */
		if (ses.sock_in != -1 
			&& (ses.remoteident || isempty(&ses.writequeue)) 
			&& writequeue_has_space) {
			FD_SET(ses.sock_in, &readfd);
		}

		/* Ordering is important, this test must occur after any other function
		might have queued packets (such as connection handlers) */
		if (ses.sock_out != -1 && !isempty(&ses.writequeue)) {
			FD_SET(ses.sock_out, &writefd);
		}

		val = select(ses.maxfd+1, &readfd, &writefd, NULL, &timeout);

		if (ses.exitflag) {
			dropbear_exit("Terminated by signal");
		}
		
		if (val < 0 && errno != EINTR) {
			dropbear_exit("Error in select");
		}

		if (val <= 0) {
			/* If we were interrupted or the select timed out, we still
			 * want to iterate over channels etc for reading, to handle
			 * server processes exiting etc. 
			 * We don't want to read/write FDs. */
			DROPBEAR_FD_ZERO(&writefd);
			DROPBEAR_FD_ZERO(&readfd);
		}
		
		/* We'll just empty out the pipe if required. We don't do
		any thing with the data, since the pipe's purpose is purely to
		wake up the select() above. */
		ses.channel_signal_pending = 0;
		if (FD_ISSET(ses.signal_pipe[0], &readfd)) {
			char x;
			TRACE(("signal pipe set"))
			while (read(ses.signal_pipe[0], &x, 1) > 0) {}
			ses.channel_signal_pending = 1;
		}

		/* check for auth timeout, rekeying required etc */
		checktimeouts();

		/* process session socket's incoming data */
		if (ses.sock_in != -1) {
			if (FD_ISSET(ses.sock_in, &readfd)) {
				if (!ses.remoteident) {
					/* blocking read of the version string */
					read_session_identification();
				} else {
					read_packet();
				}
			}
			
			/* Process the decrypted packet. After this, the read buffer
			 * will be ready for a new packet */
			if (ses.payload != NULL) {
				process_packet();
			}
		}

		/* if required, flush out any queued reply packets that
		were being held up during a KEX */
		maybe_flush_reply_queue();

		handle_connect_fds(&writefd);

		/* loop handler prior to channelio, in case the server loophandler closes
		channels on process exit */
		loophandler();

		/* process pipes etc for the channels, ses.dataallowed == 0
		 * during rekeying ) */
		channelio(&readfd, &writefd);

		/* process session socket's outgoing data */
		if (ses.sock_out != -1) {
			if (!isempty(&ses.writequeue)) {
				write_packet();
			}
		}

	} /* for(;;) */
	
	/* Not reached */
}

static void cleanup_buf(buffer **buf) {
	if (!*buf) {
		return;
	}
	buf_burn(*buf);
	buf_free(*buf);
	*buf = NULL;
}

/* clean up a session on exit */
void session_cleanup() {
	
	TRACE(("enter session_cleanup"))
	
	/* we can't cleanup if we don't know the session state */
	if (!ses.init_done) {
		TRACE(("leave session_cleanup: !ses.init_done"))
		return;
	}

	/* BEWARE of changing order of functions here. */

	/* Must be before extra_session_cleanup() */
	chancleanup();

	if (ses.extra_session_cleanup) {
		ses.extra_session_cleanup();
	}

	/* After these are freed most functions will fail */
#if DROPBEAR_CLEANUP
	/* listeners call cleanup functions, this should occur before
	other session state is freed. */
	remove_all_listeners();

	remove_connect_pending();

	while (!isempty(&ses.writequeue)) {
		buf_free(dequeue(&ses.writequeue));
	}

	m_free(ses.newkeys);
#ifndef DISABLE_ZLIB
	if (ses.keys->recv.zstream != NULL) {
		if (inflateEnd(ses.keys->recv.zstream) == Z_STREAM_ERROR) {
			dropbear_exit("Crypto error");
		}
		m_free(ses.keys->recv.zstream);
	}
#endif

	m_free(ses.remoteident);
	m_free(ses.authstate.pw_dir);
	m_free(ses.authstate.pw_name);
	m_free(ses.authstate.pw_shell);
	m_free(ses.authstate.pw_passwd);
	m_free(ses.authstate.username);
#endif

	cleanup_buf(&ses.session_id);
	cleanup_buf(&ses.hash);
	cleanup_buf(&ses.payload);
	cleanup_buf(&ses.readbuf);
	cleanup_buf(&ses.writepayload);
	cleanup_buf(&ses.kexhashbuf);
	cleanup_buf(&ses.transkexinit);
	if (ses.dh_K) {
		mp_clear(ses.dh_K);
	}
	m_free(ses.dh_K);

	m_burn(ses.keys, sizeof(struct key_context));
	m_free(ses.keys);

	TRACE(("leave session_cleanup"))
}

void send_session_identification() {
	buffer *writebuf = buf_new(strlen(LOCAL_IDENT "\r\n") + 1);
	buf_putbytes(writebuf, (const unsigned char *) LOCAL_IDENT "\r\n", strlen(LOCAL_IDENT "\r\n"));
	writebuf_enqueue(writebuf);
}

static void read_session_identification() {
	/* max length of 255 chars */
	char linebuf[256];
	int len = 0;
	char done = 0;
	int i;

	/* Servers may send other lines of data before sending the
	 * version string, client must be able to process such lines.
	 * If they send more than 50 lines, something is wrong */
	for (i = IS_DROPBEAR_CLIENT ? 50 : 1; i > 0; i--) {
		len = ident_readln(ses.sock_in, linebuf, sizeof(linebuf));

		if (len < 0 && errno != EINTR) {
			/* It failed */
			break;
		}

		if (len >= 4 && memcmp(linebuf, "SSH-", 4) == 0) {
			/* start of line matches */
			done = 1;
			break;
		}
	}

	if (!done) {
		TRACE(("error reading remote ident: %s\n", strerror(errno)))
		ses.remoteclosed();
	} else {
		/* linebuf is already null terminated */
		ses.remoteident = m_malloc(len);
		memcpy(ses.remoteident, linebuf, len);
	}

	/* Shall assume that 2.x will be backwards compatible. */
	if (strncmp(ses.remoteident, "SSH-2.", 6) != 0
			&& strncmp(ses.remoteident, "SSH-1.99-", 9) != 0) {
		dropbear_exit("Incompatible remote version '%s'", ses.remoteident);
	}

	TRACE(("remoteident: %s", ses.remoteident))

}

/* returns the length including null-terminating zero on success,
 * or -1 on failure */
static int ident_readln(int fd, char* buf, int count) {
	
	char in;
	int pos = 0;
	int num = 0;
	fd_set fds;
	struct timeval timeout;

	TRACE(("enter ident_readln"))

	if (count < 1) {
		return -1;
	}

	DROPBEAR_FD_ZERO(&fds);

	/* select since it's a non-blocking fd */
	
	/* leave space to null-terminate */
	while (pos < count-1) {

		FD_SET(fd, &fds);

		timeout.tv_sec = 1;
		timeout.tv_usec = 0;
		if (select(fd+1, &fds, NULL, NULL, &timeout) < 0) {
			if (errno == EINTR) {
				continue;
			}
			TRACE(("leave ident_readln: select error"))
			return -1;
		}

		checktimeouts();
		
		/* Have to go one byte at a time, since we don't want to read past
		 * the end, and have to somehow shove bytes back into the normal
		 * packet reader */
		if (FD_ISSET(fd, &fds)) {
			num = read(fd, &in, 1);
			/* a "\n" is a newline, "\r" we want to read in and keep going
			 * so that it won't be read as part of the next line */
			if (num < 0) {
				/* error */
				if (errno == EINTR) {
					continue; /* not a real error */
				}
				TRACE(("leave ident_readln: read error"))
				return -1;
			}
			if (num == 0) {
				/* EOF */
				TRACE(("leave ident_readln: EOF"))
				return -1;
			}

#if DROPBEAR_FUZZ
			fuzz_dump(&in, 1);
#endif

			if (in == '\n') {
				/* end of ident string */
				break;
			}
			/* we don't want to include '\r's */
			if (in != '\r') {
				buf[pos] = in;
				pos++;
			}
		}
	}

	buf[pos] = '\0';
	TRACE(("leave ident_readln: return %d", pos+1))
	return pos+1;
}

void ignore_recv_response() {
	/* Do nothing */
	TRACE(("Ignored msg_request_response"))
}

static void send_msg_keepalive() {
	time_t old_time_idle = ses.last_packet_time_idle;
	struct Channel *chan = get_any_ready_channel();

	CHECKCLEARTOWRITE();

	if (chan) {
		/* Channel requests are preferable, more implementations
		handle them than SSH_MSG_GLOBAL_REQUEST */
		TRACE(("keepalive channel request %d", chan->index))
		start_send_channel_request(chan, DROPBEAR_KEEPALIVE_STRING);
	} else {
		TRACE(("keepalive global request"))
		/* Some peers will reply with SSH_MSG_REQUEST_FAILURE, 
		some will reply with SSH_MSG_UNIMPLEMENTED, some will exit. */
		buf_putbyte(ses.writepayload, SSH_MSG_GLOBAL_REQUEST); 
		buf_putstring(ses.writepayload, DROPBEAR_KEEPALIVE_STRING,
			strlen(DROPBEAR_KEEPALIVE_STRING));
	}
	buf_putbyte(ses.writepayload, 1); /* want_reply */
	encrypt_packet();

	ses.last_packet_time_keepalive_sent = monotonic_now();

	/* keepalives shouldn't update idle timeout, reset it back */
	ses.last_packet_time_idle = old_time_idle;
}

/* Check all timeouts which are required. Currently these are the time for
 * user authentication, and the automatic rekeying. */
static void checktimeouts() {

	time_t now;
	now = monotonic_now();
	
	if (IS_DROPBEAR_SERVER && ses.connect_time != 0
		&& now - ses.connect_time >= AUTH_TIMEOUT) {
			dropbear_close("Timeout before auth");
	}

	/* we can't rekey if we haven't done remote ident exchange yet */
	if (ses.remoteident == NULL) {
		return;
	}

	if (!ses.kexstate.sentkexinit
			&& (now - ses.kexstate.lastkextime >= KEX_REKEY_TIMEOUT
			|| ses.kexstate.datarecv+ses.kexstate.datatrans >= KEX_REKEY_DATA)) {
		TRACE(("rekeying after timeout or max data reached"))
		send_msg_kexinit();
	}
	
	if (opts.keepalive_secs > 0 && ses.authstate.authdone) {
		/* Avoid sending keepalives prior to auth - those are
		not valid pre-auth packet types */

		/* Send keepalives if we've been idle */
		if (now - ses.last_packet_time_any_sent >= opts.keepalive_secs) {
			send_msg_keepalive();
		}

		/* Also send an explicit keepalive message to trigger a response
		if the remote end hasn't sent us anything */
		if (now - ses.last_packet_time_keepalive_recv >= opts.keepalive_secs
			&& now - ses.last_packet_time_keepalive_sent >= opts.keepalive_secs) {
			send_msg_keepalive();
		}

		if (now - ses.last_packet_time_keepalive_recv 
			>= opts.keepalive_secs * DEFAULT_KEEPALIVE_LIMIT) {
			dropbear_exit("Keepalive timeout");
		}
	}

	if (opts.idle_timeout_secs > 0 
			&& now - ses.last_packet_time_idle >= opts.idle_timeout_secs) {
		dropbear_close("Idle timeout");
	}
}

static void update_timeout(long limit, long now, long last_event, long * timeout) {
	TRACE2(("update_timeout limit %ld, now %ld, last %ld, timeout %ld",
		limit, now, last_event, *timeout))
	if (last_event > 0 && limit > 0) {
		*timeout = MIN(*timeout, last_event+limit-now);
		TRACE2(("new timeout %ld", *timeout))
	}
}

static long select_timeout() {
	/* determine the minimum timeout that might be required, so
	as to avoid waking when unneccessary */
	long timeout = KEX_REKEY_TIMEOUT;
	long now = monotonic_now();

	if (!ses.kexstate.sentkexinit) {
		update_timeout(KEX_REKEY_TIMEOUT, now, ses.kexstate.lastkextime, &timeout);
	}

	if (ses.authstate.authdone != 1 && IS_DROPBEAR_SERVER) {
		/* AUTH_TIMEOUT is only relevant before authdone */
		update_timeout(AUTH_TIMEOUT, now, ses.connect_time, &timeout);
	}

	if (ses.authstate.authdone) {
		update_timeout(opts.keepalive_secs, now, 
			MAX(ses.last_packet_time_keepalive_recv, ses.last_packet_time_keepalive_sent),
			&timeout);
	}

	update_timeout(opts.idle_timeout_secs, now, ses.last_packet_time_idle,
		&timeout);

	/* clamp negative timeouts to zero - event has already triggered */
	return MAX(timeout, 0);
}

const char* get_user_shell() {
	/* an empty shell should be interpreted as "/bin/sh" */
	if (ses.authstate.pw_shell[0] == '\0') {
		return "/bin/sh";
	} else {
		return ses.authstate.pw_shell;
	}
}
void fill_passwd(const char* username) {
	struct passwd *pw = NULL;
	if (ses.authstate.pw_name)
		m_free(ses.authstate.pw_name);
	if (ses.authstate.pw_dir)
		m_free(ses.authstate.pw_dir);
	if (ses.authstate.pw_shell)
		m_free(ses.authstate.pw_shell);
	if (ses.authstate.pw_passwd)
		m_free(ses.authstate.pw_passwd);

	pw = getpwnam(username);
	if (!pw) {
		return;
	}
	ses.authstate.pw_uid = pw->pw_uid;
	ses.authstate.pw_gid = pw->pw_gid;
	ses.authstate.pw_name = m_strdup(pw->pw_name);
	ses.authstate.pw_dir = m_strdup(pw->pw_dir);
	ses.authstate.pw_shell = m_strdup(pw->pw_shell);
	{
		char *passwd_crypt = pw->pw_passwd;
#ifdef HAVE_SHADOW_H
		/* get the shadow password if possible */
		struct spwd *spasswd = getspnam(ses.authstate.pw_name);
		if (spasswd && spasswd->sp_pwdp) {
			passwd_crypt = spasswd->sp_pwdp;
		}
#endif
		if (!passwd_crypt) {
			/* android supposedly returns NULL */
			passwd_crypt = "!!";
		}
		ses.authstate.pw_passwd = m_strdup(passwd_crypt);
	}
}

/* Called when channels are modified */
void update_channel_prio() {
	enum dropbear_prio new_prio;
	int any = 0;
	unsigned int i;

	TRACE(("update_channel_prio"))

	if (ses.sock_out < 0) {
		TRACE(("leave update_channel_prio: no socket"))
		return;
	}

	new_prio = DROPBEAR_PRIO_BULK;
	for (i = 0; i < ses.chansize; i++) {
		struct Channel *channel = ses.channels[i];
		if (!channel || channel->prio == DROPBEAR_CHANNEL_PRIO_EARLY) {
			if (channel && channel->prio == DROPBEAR_CHANNEL_PRIO_EARLY) {
				TRACE(("update_channel_prio: early %d", channel->index))
			}
			continue;
		}
		any = 1;
		if (channel->prio == DROPBEAR_CHANNEL_PRIO_INTERACTIVE)
		{
			TRACE(("update_channel_prio: lowdelay %d", channel->index))
			new_prio = DROPBEAR_PRIO_LOWDELAY;
			break;
		} else if (channel->prio == DROPBEAR_CHANNEL_PRIO_UNKNOWABLE
			&& new_prio == DROPBEAR_PRIO_BULK)
		{
			TRACE(("update_channel_prio: unknowable %d", channel->index))
			new_prio = DROPBEAR_PRIO_DEFAULT;
		}
	}

	if (any == 0) {
		/* lowdelay during setup */
		TRACE(("update_channel_prio: not any"))
		new_prio = DROPBEAR_PRIO_LOWDELAY;
	}

	if (new_prio != ses.socket_prio) {
		TRACE(("Dropbear priority transitioning %d -> %d", ses.socket_prio, new_prio))
		set_sock_priority(ses.sock_out, new_prio);
		ses.socket_prio = new_prio;
	}
}