DragonFly On-Line Manual Pages


SECURITY(7)	  DragonFly Miscellaneous Information Manual	   SECURITY(7)

NAME

security -- introduction to security under DragonFly

DESCRIPTION

Security is a function that begins and ends with the system administra- tor. While all BSD multi-user systems have some inherent security, the job of building and maintaining additional security mechanisms to keep users `honest' is probably one of the single largest undertakings of the sysadmin. Machines are only as secure as you make them, and security concerns are ever competing with the human necessity for convenience. UNIX systems, in general, are capable of running a huge number of simul- taneous processes and many of these processes operate as servers - mean- ing that external entities can connect and talk to them. As yesterday's mini-computers and mainframes become today's desktops, and as computers become networked and internetworked, security becomes an ever bigger issue. Security is best implemented through a layered onion approach. In a nut- shell, what you want to do is to create as many layers of security as are convenient and then carefully monitor the system for intrusions. You do not want to overbuild your security or you will interfere with the detec- tion side, and detection is one of the single most important aspects of any security mechanism. For example, it makes little sense to set the schg flags (see chflags(1)) on every system binary because while this may temporarily protect the binaries, it prevents a hacker who has broken in from making an easily detectable change that may result in your security mechanisms not detecting the hacker at all. System security also pertains to dealing with various forms of attack, including attacks that attempt to crash or otherwise make a system unus- able but do not attempt to break root. Security concerns can be split up into several categories: 1. Denial of service attacks 2. User account compromises 3. Root compromise through accessible servers 4. Root compromise via user accounts 5. Backdoor creation A denial of service attack is an action that deprives the machine of needed resources. Typically, D.O.S. attacks are brute-force mechanisms that attempt to crash or otherwise make a machine unusable by overwhelm- ing its servers or network stack. Some D.O.S. attacks try to take advan- tages of bugs in the networking stack to crash a machine with a single packet. The latter can only be fixed by applying a bug fix to the ker- nel. Attacks on servers can often be fixed by properly specifying options to limit the load the servers incur on the system under adverse conditions. Brute-force network attacks are harder to deal with. A spoofed-packet attack, for example, is nearly impossible to stop short of cutting your system off from the Internet. It may not be able to take your machine down, but it can fill up Internet pipe. A user account compromise is even more common than a D.O.S. attack. Many sysadmins still run standard telnetd, rlogind, rshd, and ftpd servers on their machines. These servers, by default, do not operate over encrypted connections. The result is that if you have any moderate-sized user base, one or more of your users logging into your system from a remote location (which is the most common and convenient way to login to a sys- tem) will have his or her password sniffed. The attentive system admin will analyze his remote access logs looking for suspicious source addresses even for successful logins. One must always assume that once an attacker has access to a user account, the attacker can break root. However, the reality is that in a well secured and maintained system, access to a user account does not necessarily give the attacker access to root. The distinction is impor- tant because without access to root the attacker cannot generally hide his tracks and may, at best, be able to do nothing more than mess with the user's files or crash the machine. User account compromises are very common because users tend not to take the precautions that sysadmins take. System administrators must keep in mind that there are potentially many ways to break root on a machine. The attacker may know the root pass- word, the attacker may find a bug in a root-run server and be able to break root over a network connection to that server, or the attacker may know of a bug in an suid-root program that allows the attacker to break root once he has broken into a user's account. If an attacker has found a way to break root on a machine, the attacker may not have a need to install a backdoor. Many of the root holes found and closed to date involve a considerable amount of work by the hacker to cleanup after him- self, so most hackers do install backdoors. This gives you a convenient way to detect the hacker. Making it impossible for a hacker to install a backdoor may actually be detrimental to your security because it will not close off the hole the hacker found to break in the first place. Security remedies should always be implemented with a multi-layered `onion peel' approach and can be categorized as follows: 1. Securing root and staff accounts 2. Securing root - root-run servers and suid/sgid binaries 3. Securing user accounts 4. Securing the password file 5. Securing the kernel core, raw devices, and filesystems 6. Quick detection of inappropriate changes made to the system 7. Paranoia

SECURING THE ROOT ACCOUNT AND SECURING STAFF ACCOUNTS

Don't bother securing staff accounts if you haven't secured the root account. Most systems have a password assigned to the root account. The first thing you do is assume that the password is `always' compromised. This does not mean that you should remove the password. The password is almost always necessary for console access to the machine. What it does mean is that you should not make it possible to use the password outside of the console or possibly even with a su(1) command. For example, make sure that your pty's are specified as being unsecure in the `/etc/ttys' file so that direct root logins via telnet or rlogin are disallowed. If using other login services such as sshd, make sure that direct root logins are disabled there as well. Consider every access method - ser- vices such as ftp often fall through the cracks. Direct root logins should only be allowed via the system console. Of course, as a sysadmin you have to be able to get to root, so we open up a few holes. But we make sure these holes require additional password verification to operate. One way to make root accessible is to add appropriate staff accounts to the wheel group (in /etc/group). The staff members placed in the wheel group are allowed to `su' to root. You should never give staff members native wheel access by putting them in the wheel group in their password entry. Staff accounts should be placed in a `staff' group, and then added to the wheel group via the `/etc/group' file. Only those staff members who actually need to have root access should be placed in the wheel group. It is also possible, when using an authentication method such as kerberos, to use kerberos's `.k5login' file in the root account to allow a ksu(1) to root without having to place anyone at all in the wheel group. This may be the better solution since the wheel mechanism still allows an intruder to break root if the intruder has gotten hold of your password file and can break into a staff account. While having the wheel mechanism is better than having nothing at all, it isn't necessarily the safest option. An indirect way to secure the root account is to secure your staff accounts by using an alternative login access method and *'ing out the crypted password for the staff accounts. This way an intruder may be able to steal the password file but will not be able to break into any staff accounts (or, indirectly, root, even if root has a crypted password associated with it). Staff members get into their staff accounts through a secure login mechanism such as kerberos(8) or ssh(1) using a pri- vate/public key pair. When you use something like kerberos you generally must secure the machines which run the kerberos servers and your desktop workstation. When you use a public/private key pair with ssh, you must generally secure the machine you are logging in FROM (typically your workstation), but you can also add an additional layer of protection to the key pair by password protecting the keypair when you create it with ssh-keygen(1). Being able to *-out the passwords for staff accounts also guarantees that staff members can only login through secure access meth- ods that you have setup. You can thus force all staff members to use secure, encrypted connections for all their sessions which closes an important hole used by many intruders: That of sniffing the network from an unrelated, less secure machine. The more indirect security mechanisms also assume that you are logging in from a more restrictive server to a less restrictive server. For exam- ple, if your main box is running all sorts of servers, your workstation shouldn't be running any. In order for your workstation to be reasonably secure you should run as few servers as possible, up to and including no servers at all, and you should run a password-protected screen blanker. Of course, given physical access to a workstation an attacker can break any sort of security you put on it. This is definitely a problem that you should consider but you should also consider the fact that the vast majority of break-ins occur remotely, over a network, from people who do not have physical access to your workstation or servers. Using something like kerberos also gives you the ability to disable or change the password for a staff account in one place and have it immedi- ately affect all the machines the staff member may have an account on. If a staff member's account gets compromised, the ability to instantly change his password on all machines should not be underrated. With dis- crete passwords, changing a password on N machines can be a mess. You can also impose re-passwording restrictions with kerberos: not only can a kerberos ticket be made to timeout after a while, but the kerberos sys- tem can require that the user choose a new password after a certain period of time (say, once a month). SECURING ROOT - ROOT-RUN SERVERS AND SUID/SGID BINARIES The prudent sysadmin only runs the servers he needs to, no more, no less. Be aware that third party servers are often the most bug-prone. For example, running an old version of imapd or popper is like giving a uni- versal root ticket out to the entire world. Never run a server that you have not checked out carefully. Many servers do not need to be run as root. For example, the ntalk, comsat, and finger daemons can be run in special user `sandboxes'. A sandbox isn't perfect unless you go to a large amount of trouble, but the onion approach to security still stands: If someone is able to break in through a server running in a sandbox, they still have to break out of the sandbox. The more layers the attacker must break through, the lower the likelihood of his success. Root holes have historically been found in virtually every server ever run as root, including basic system servers. If you are running a machine through which people only login via sshd and never login via tel- netd or rshd or rlogind, then turn off those services! DragonFly now defaults to running ntalkd, comsat, and finger in a sand- box. Another program which may be a candidate for running in a sandbox is named(8). The default rc.conf includes the arguments necessary to run named in a sandbox in a commented-out form. Depending on whether you are installing a new system or upgrading an existing system, the special user accounts used by these sandboxes may not be installed. The prudent sysadmin would research and implement sandboxes for servers whenever pos- sible. There are a number of other servers that typically do not run in sand- boxes: sendmail, popper, imapd, ftpd, and others. There are alternatives to some of these, but installing them may require more work than you are willing to put (the convenience factor strikes again). You may have to run these servers as root and rely on other mechanisms to detect break- ins that might occur through them. The other big potential root hole in a system are the suid-root and sgid binaries installed on the system. Most of these binaries, such as rlogin, reside in /bin, /sbin, /usr/bin, or /usr/sbin. While nothing is 100% safe, the system-default suid and sgid binaries can be considered reasonably safe. Still, root holes are occasionally found in these bina- ries. A root hole was found in Xlib in 1998 that made xterm (which is typically suid) vulnerable. It is better to be safe than sorry and the prudent sysadmin will restrict suid binaries that only staff should run to a special group that only staff can access, and get rid of (chmod 000) any suid binaries that nobody uses. A server with no display generally does not need an xterm binary. Sgid binaries can be almost as dangerous. If an intruder can break an sgid-kmem binary the intruder might be able to read /dev/kmem and thus read the crypted password file, potentially compromising any passworded account. Alternatively an intruder who breaks group kmem can monitor keystrokes sent through pty's, including pty's used by users who login through secure methods. An intruder that breaks the tty group can write to almost any user's tty. If a user is running a terminal program or emulator with a keyboard-simulation fea- ture, the intruder can potentially generate a data stream that causes the user's terminal to echo a command, which is then run as that user.

SECURING USER ACCOUNTS

User accounts are usually the most difficult to secure. While you can impose Draconian access restrictions on your staff and *-out their pass- words, you may not be able to do so with any general user accounts you might have. If you do have sufficient control then you may win out and be able to secure the user accounts properly. If not, you simply have to be more vigilant in your monitoring of those accounts. Use of ssh and kerberos for user accounts is more problematic due to the extra adminis- tration and technical support required, but still a very good solution compared to a crypted password file.

SECURING THE PASSWORD FILE

The only sure-fire way is to *-out as many passwords as you can and use ssh or kerberos for access to those accounts. Even though the crypted password file (/etc/spwd.db) can only be read by root, it may be possible for an intruder to obtain read access to that file even if the attacker cannot obtain root-write access. Your security scripts should always check for and report changes to the password file (see `Checking file integrity' below).

SECURING THE KERNEL CORE, RAW DEVICES, AND FILESYSTEMS

If an attacker breaks root he can do just about anything, but there are certain conveniences. For example, most modern kernels have a packet sniffing device driver built in. Under DragonFly it is called the `bpf' device. An intruder will commonly attempt to run a packet sniffer on a compromised machine. You do not need to give the intruder the capability and most systems should not have the bpf device compiled in. But even if you turn off the bpf device, you still have /dev/mem and /dev/kmem to worry about. For that matter, the intruder can still write to raw disk devices. Also, there is another kernel feature called the module loader, kldload(8). An enterprising intruder can use a KLD module to install his own bpf device or other sniffing device on a running ker- nel. To avoid these problems you have to run the kernel at a higher secure level, at least securelevel 1. The securelevel can be set with a sysctl on the kern.securelevel variable. Once you have set the securelevel to 1, write access to raw devices will be denied and special chflags flags, such as `schg', will be enforced. You must also ensure that the `schg' flag is set on critical startup binaries, directories, and script files - everything that gets run up to the point where the securelevel is set. This might be overdoing it, and upgrading the system is much more difficult when you operate at a higher secure level. You may compromise and run the system at a higher secure level but not set the schg flag for every system file and directory under the sun. Another possibility is to simply mount / and /usr read-only. It should be noted that being too draconian in what you attempt to protect may prevent the all-important detection of an intrusion. CHECKING FILE INTEGRITY: BINARIES, CONFIG FILES, ETC When it comes right down to it, you can only protect your core system configuration and control files so much before the convenience factor rears its ugly head. For example, using chflags to set the schg bit on most of the files in / and /usr is probably counterproductive because while it may protect the files, it also closes a detection window. The last layer of your security onion is perhaps the most important - detec- tion. The rest of your security is pretty much useless (or, worse, presents you with a false sense of safety) if you cannot detect potential incursions. Half the job of the onion is to slow down the attacker rather than stop him in order to give the detection side of the equation a chance to catch him in the act. The best way to detect an incursion is to look for modified, missing, or unexpected files. The best way to look for modified files is from another (often centralized) limited-access system. Writing your security scripts on the extra-secure limited-access system makes them mostly invisible to potential hackers, and this is important. In order to take maximum advantage you generally have to give the limited-access box sig- nificant access to the other machines in the business, usually either by doing a read-only NFS export of the other machines to the limited-access box, or by setting up ssh keypairs to allow the limit-access box to ssh to the other machines. Except for its network traffic, NFS is the least visible method - allowing you to monitor the filesystems on each client box virtually undetected. If your limited-access server is connected to the client boxes through a switch, the NFS method is often the better choice. If your limited-access server is connected to the client boxes through a hub or through several layers of routing, the NFS method may be too insecure (network-wise) and using ssh may be the better choice even with the audit-trail tracks that ssh lays. Once you give a limit-access box at least read access to the client sys- tems it is supposed to monitor, you must write scripts to do the actual monitoring. Given an NFS mount, you can write scripts out of simple sys- tem utilities such as find(1) and md5(1). It is best to physically md5 the client-box files boxes at least once a day, and to test control files such as those found in /etc, /usr/local/etc and /usr/pkg/etc even more often. When mismatches are found relative to the base md5 information the limited-access machine knows is valid, it should scream at a sysadmin to go check it out. A good security script will also check for inappro- priate suid binaries and for new or deleted files on system partitions such as / and /usr When using ssh rather than NFS, writing the security script is much more difficult. You essentially have to scp the scripts to the client box in order to run them, making them visible, and for safety you also need to scp the binaries (such as find) that those scripts use. The ssh daemon on the client box may already be compromised. All in all, using ssh may be necessary when running over unsecure links, but it's also a lot harder to deal with. A good security script will also check for changes to user and staff mem- bers access configuration files: .rhosts, .shosts, .ssh/authorized_keys and so forth... files that might fall outside the purview of the MD5 check. If you have a huge amount of user disk space it may take too long to run through every file on those partitions. In this case, setting mount flags to disallow suid binaries and devices on those partitions is a good idea. The `nodev' and `nosuid' options (see mount(8)) are what you want to look into. I would scan them anyway at least once a week, since the object of this layer is to detect a break-in whether or not the breakin is effective. Process accounting (see accton(8)) is a relatively low-overhead feature of the operating system which I recommend using as a post-break-in evalu- ation mechanism. It is especially useful in tracking down how an intruder has actually broken into a system, assuming the file is still intact after the break-in occurs. Finally, security scripts should process the log files and the logs them- selves should be generated in as secure a manner as possible - remote syslog can be very useful. An intruder tries to cover his tracks, and log files are critical to the sysadmin trying to track down the time and method of the initial break-in. One way to keep a permanent record of the log files is to run the system console to a serial port and collect the information on a continuing basis through a secure machine monitoring the consoles.

PARANOIA

A little paranoia never hurts. As a rule, a sysadmin can add any number of security features as long as they do not affect convenience, and can add security features that do affect convenience with some added thought. Even more importantly, a security administrator should mix it up a bit - if you use recommendations such as those given by this manual page verba- tim, you give away your methodologies to the prospective hacker who also has access to this manual page. SPECIAL SECTION ON D.O.S. ATTACKS This section covers Denial of Service attacks. A DOS attack is typically a packet attack. While there isn't much you can do about modern spoofed packet attacks that saturate your network, you can generally limit the damage by ensuring that the attacks cannot take down your servers. 1. Limiting server forks 2. Limiting springboard attacks (ICMP response attacks, ping broadcast, etc...) 3. Kernel Route Cache A common D.O.S. attack is against a forking server that attempts to cause the server to eat processes, file descriptors, and memory until the machine dies. Inetd (see inetd(8)) has several options to limit this sort of attack. It should be noted that while it is possible to prevent a machine from going down it is not generally possible to prevent a ser- vice from being disrupted by the attack. Read the inetd manual page carefully and pay specific attention to the -c, -C, and -R options. Note that spoofed-IP attacks will circumvent the -C option to inetd, so typi- cally a combination of options must be used. Some standalone servers have self-fork-limitation parameters. Sendmail has its -OMaxDaemonChildren option which tends to work much bet- ter than trying to use sendmail's load limiting options due to the load lag. You should specify a MaxDaemonChildren parameter when you start sendmail high enough to handle your expected load but no so high that the computer cannot handle that number of sendmails without falling on its face. It is also prudent to run sendmail in queued mode (-ODeliveryMode=queued) and to run the daemon (sendmail -bd) separate from the queue-runs (sendmail -q15m). If you still want realtime deliv- ery you can run the queue at a much lower interval, such as -q1m, but be sure to specify a reasonable MaxDaemonChildren option for that sendmail to prevent cascade failures. Syslogd can be attacked directly and it is strongly recommended that you use the -s option whenever possible, and the -a option otherwise. You should also be fairly careful with connect-back services such as tcp- wrapper's reverse-identd, which can be attacked directly. You generally do not want to use the reverse-ident feature of tcpwrappers for this rea- son. It is a very good idea to protect internal services from external access by firewalling them off at your border routers. The idea here is to pre- vent saturation attacks from outside your LAN, not so much to protect internal services from network-based root compromise. Always configure an exclusive firewall, i.e. `firewall everything *except* ports A, B, C, D, and M-Z'. This way you can firewall off all of your low ports except for certain specific services such as named (if you are primary for a zone), ntalkd, sendmail, and other internet-accessible services. If you try to configure the firewall the other way - as an inclusive or permis- sive firewall, there is a good chance that you will forget to `close' a couple of services or that you will add a new internal service and forget to update the firewall. You can still open up the high-numbered port range on the firewall to allow permissive-like operation without compro- mising your low ports. Also take note that DragonFly allows you to con- trol the range of port numbers used for dynamic binding via the various net.inet.ip.portrange sysctl's (sysctl -a | fgrep portrange), which can also ease the complexity of your firewall's configuration. I usually use a normal first/last range of 4000 to 5000, and a hiport range of 49152 to 65535, then block everything under 4000 off in my firewall (except for certain specific internet-accessible ports, of course). Another common D.O.S. attack is called a springboard attack - to attack a server in a manner that causes the server to generate responses which then overload the server, the local network, or some other machine. The most common attack of this nature is the ICMP PING BROADCAST attack. The attacker spoofs ping packets sent to your LAN's broadcast address with the source IP address set to the actual machine they wish to attack. If your border routers are not configured to stomp on ping's to broadcast addresses, your LAN winds up generating sufficient responses to the spoofed source address to saturate the victim, especially when the attacker uses the same trick on several dozen broadcast addresses over several dozen different networks at once. Broadcast attacks of over a hundred and twenty megabits have been measured. A second common spring- board attack is against the ICMP error reporting system. By constructing packets that generate ICMP error responses, an attacker can saturate a server's incoming network and cause the server to saturate its outgoing network with ICMP responses. This type of attack can also crash the server by running it out of mbuf's, especially if the server cannot drain the ICMP responses it generates fast enough. The DragonFly kernel has a new kernel compile option called ICMP_BANDLIM which limits the effective- ness of these sorts of attacks. The last major class of springboard attacks is related to certain internal inetd services such as the udp echo service. An attacker simply spoofs a UDP packet with the source address being server A's echo port, and the destination address being server B's echo port, where server A and B are both on your LAN. The two servers then bounce this one packet back and forth between each other. The attacker can overload both servers and their LANs simply by injecting a few packets in this manner. Similar problems exist with the internal chargen port. A competent sysadmin will turn off all of these inetd- internal test services. Spoofed packet attacks may also be used to overload the kernel route cache. Refer to the net.inet.ip.rtexpire, net.inet.ip.rtminexpire, and net.inet.ip.rtmaxcache sysctl parameters. A spoofed packet attack that uses a random source IP will cause the kernel to generate a temporary cached route in the route table, viewable with `netstat -rna | fgrep W3'. These routes typically timeout in 1600 seconds or so. If the kernel detects that the cached route table has gotten too big it will dynami- cally reduce the rtexpire but will never decrease it to less than rtmin- expire. There are two problems: (1) The kernel does not react quickly enough when a lightly loaded server is suddenly attacked, and (2) The rtminexpire is not low enough for the kernel to survive a sustained attack. If your servers are connected to the internet via a T3 or better it may be prudent to manually override both rtexpire and rtminexpire via sysctl(8). Never set either parameter to zero (unless you want to crash the machine :-)). Setting both parameters to 2 seconds should be suffi- cient to protect the route table from attack.

ACCESS ISSUES WITH KERBEROS AND SSH

There are a few issues with both kerberos and ssh that need to be addressed if you intend to use them. Kerberos V is an excellent authen- tication protocol but the kerberized telnet and rlogin suck rocks. There are bugs that make them unsuitable for dealing with binary streams. Also, by default kerberos does not encrypt a session unless you use the -x option. Ssh encrypts everything by default. Ssh works quite well in every respect except when it is set up to forward encryption keys. What this means is that if you have a secure worksta- tion holding keys that give you access to the rest of the system, and you ssh to an unsecure machine, your keys becomes exposed. The actual keys themselves are not exposed, but ssh installs a forwarding port for the duration of your login and if a hacker has broken root on the unsecure machine he can utilize that port to use your keys to gain access to any other machine that your keys unlock. We recommend that you use ssh in combination with kerberos whenever pos- sible for staff logins. Ssh can be compiled with kerberos support. This reduces your reliance on potentially exposable ssh keys while at the same time protecting passwords via kerberos. Ssh keys should only be used for automated tasks from secure machines (something that kerberos is unsuited to). We also recommend that you either turn off key-forwarding in the ssh configuration, or that you make use of the from=IP/DOMAIN option that ssh allows in its authorized_keys file to make the key only usable to entities logging in from specific machines.

SEE ALSO

chflags(1), find(1), md5(1), netstat(1), openssl(1), ssh(1), xdm(1) (x11/xdm), group(5), ttys(5), firewall(7), accton(8), init(8), kerberos(8), sshd(8), sysctl(8), syslogd(8), vipw(8)

HISTORY

The security manual page was originally written by Matthew Dillon and first appeared in FreeBSD 3.1, December 1998. DragonFly 3.7 September 18, 1999 DragonFly 3.7