Commit | Line | Data |
a0d0e21e |
1 | =head1 NAME |
2 | |
3 | perlsec - Perl security |
4 | |
5 | =head1 DESCRIPTION |
6 | |
425e5e39 |
7 | Perl is designed to make it easy to program securely even when running |
8 | with extra privileges, like setuid or setgid programs. Unlike most |
54310121 |
9 | command line shells, which are based on multiple substitution passes on |
425e5e39 |
10 | each line of the script, Perl uses a more conventional evaluation scheme |
11 | with fewer hidden snags. Additionally, because the language has more |
54310121 |
12 | builtin functionality, it can rely less upon external (and possibly |
425e5e39 |
13 | untrustworthy) programs to accomplish its purposes. |
a0d0e21e |
14 | |
425e5e39 |
15 | Perl automatically enables a set of special security checks, called I<taint |
16 | mode>, when it detects its program running with differing real and effective |
17 | user or group IDs. The setuid bit in Unix permissions is mode 04000, the |
18 | setgid bit mode 02000; either or both may be set. You can also enable taint |
5f05dabc |
19 | mode explicitly by using the B<-T> command line flag. This flag is |
425e5e39 |
20 | I<strongly> suggested for server programs and any program run on behalf of |
fb73857a |
21 | someone else, such as a CGI script. Once taint mode is on, it's on for |
22 | the remainder of your script. |
a0d0e21e |
23 | |
1e422769 |
24 | While in this mode, Perl takes special precautions called I<taint |
25 | checks> to prevent both obvious and subtle traps. Some of these checks |
26 | are reasonably simple, such as verifying that path directories aren't |
27 | writable by others; careful programmers have always used checks like |
28 | these. Other checks, however, are best supported by the language itself, |
fb73857a |
29 | and it is these checks especially that contribute to making a set-id Perl |
425e5e39 |
30 | program more secure than the corresponding C program. |
31 | |
fb73857a |
32 | You may not use data derived from outside your program to affect |
33 | something else outside your program--at least, not by accident. All |
34 | command line arguments, environment variables, locale information (see |
35 | L<perllocale>), results of certain system calls (readdir, readlink, |
36 | the gecos field of getpw* calls), and all file input are marked as |
37 | "tainted". Tainted data may not be used directly or indirectly in any |
38 | command that invokes a sub-shell, nor in any command that modifies |
39 | files, directories, or processes. Any variable set |
40 | to a value derived from tainted data will itself be tainted, |
41 | even if it is logically impossible for the tainted data |
42 | to alter the variable. Because taintedness is associated with each |
a034a98d |
43 | scalar value, some elements of an array can be tainted and others not. |
a0d0e21e |
44 | |
a0d0e21e |
45 | For example: |
46 | |
425e5e39 |
47 | $arg = shift; # $arg is tainted |
48 | $hid = $arg, 'bar'; # $hid is also tainted |
49 | $line = <>; # Tainted |
8ebc5c01 |
50 | $line = <STDIN>; # Also tainted |
51 | open FOO, "/home/me/bar" or die $!; |
52 | $line = <FOO>; # Still tainted |
a0d0e21e |
53 | $path = $ENV{'PATH'}; # Tainted, but see below |
425e5e39 |
54 | $data = 'abc'; # Not tainted |
a0d0e21e |
55 | |
425e5e39 |
56 | system "echo $arg"; # Insecure |
57 | system "/bin/echo", $arg; # Secure (doesn't use sh) |
58 | system "echo $hid"; # Insecure |
59 | system "echo $data"; # Insecure until PATH set |
a0d0e21e |
60 | |
425e5e39 |
61 | $path = $ENV{'PATH'}; # $path now tainted |
a0d0e21e |
62 | |
54310121 |
63 | $ENV{'PATH'} = '/bin:/usr/bin'; |
c90c0ff4 |
64 | delete @ENV{'IFS', 'CDPATH', 'ENV', 'BASH_ENV'}; |
a0d0e21e |
65 | |
425e5e39 |
66 | $path = $ENV{'PATH'}; # $path now NOT tainted |
67 | system "echo $data"; # Is secure now! |
a0d0e21e |
68 | |
425e5e39 |
69 | open(FOO, "< $arg"); # OK - read-only file |
70 | open(FOO, "> $arg"); # Not OK - trying to write |
a0d0e21e |
71 | |
425e5e39 |
72 | open(FOO,"echo $arg|"); # Not OK, but... |
73 | open(FOO,"-|") |
74 | or exec 'echo', $arg; # OK |
a0d0e21e |
75 | |
425e5e39 |
76 | $shout = `echo $arg`; # Insecure, $shout now tainted |
a0d0e21e |
77 | |
425e5e39 |
78 | unlink $data, $arg; # Insecure |
79 | umask $arg; # Insecure |
a0d0e21e |
80 | |
425e5e39 |
81 | exec "echo $arg"; # Insecure |
82 | exec "echo", $arg; # Secure (doesn't use the shell) |
83 | exec "sh", '-c', $arg; # Considered secure, alas! |
a0d0e21e |
84 | |
7bac28a0 |
85 | @files = <*.c>; # Always insecure (uses csh) |
86 | @files = glob('*.c'); # Always insecure (uses csh) |
87 | |
a0d0e21e |
88 | If you try to do something insecure, you will get a fatal error saying |
89 | something like "Insecure dependency" or "Insecure PATH". Note that you |
425e5e39 |
90 | can still write an insecure B<system> or B<exec>, but only by explicitly |
54310121 |
91 | doing something like the last example above. |
425e5e39 |
92 | |
93 | =head2 Laundering and Detecting Tainted Data |
94 | |
95 | To test whether a variable contains tainted data, and whose use would thus |
fb73857a |
96 | trigger an "Insecure dependency" message, check your nearby CPAN mirror |
97 | for the F<Taint.pm> module, which should become available around November |
98 | 1997. Or you may be able to use the following I<is_tainted()> function. |
425e5e39 |
99 | |
100 | sub is_tainted { |
54310121 |
101 | return ! eval { |
102 | join('',@_), kill 0; |
103 | 1; |
425e5e39 |
104 | }; |
105 | } |
106 | |
107 | This function makes use of the fact that the presence of tainted data |
108 | anywhere within an expression renders the entire expression tainted. It |
109 | would be inefficient for every operator to test every argument for |
110 | taintedness. Instead, the slightly more efficient and conservative |
111 | approach is used that if any tainted value has been accessed within the |
112 | same expression, the whole expression is considered tainted. |
113 | |
5f05dabc |
114 | But testing for taintedness gets you only so far. Sometimes you have just |
425e5e39 |
115 | to clear your data's taintedness. The only way to bypass the tainting |
54310121 |
116 | mechanism is by referencing subpatterns from a regular expression match. |
425e5e39 |
117 | Perl presumes that if you reference a substring using $1, $2, etc., that |
118 | you knew what you were doing when you wrote the pattern. That means using |
119 | a bit of thought--don't just blindly untaint anything, or you defeat the |
a034a98d |
120 | entire mechanism. It's better to verify that the variable has only good |
121 | characters (for certain values of "good") rather than checking whether it |
122 | has any bad characters. That's because it's far too easy to miss bad |
123 | characters that you never thought of. |
425e5e39 |
124 | |
125 | Here's a test to make sure that the data contains nothing but "word" |
126 | characters (alphabetics, numerics, and underscores), a hyphen, an at sign, |
127 | or a dot. |
128 | |
54310121 |
129 | if ($data =~ /^([-\@\w.]+)$/) { |
425e5e39 |
130 | $data = $1; # $data now untainted |
131 | } else { |
132 | die "Bad data in $data"; # log this somewhere |
133 | } |
134 | |
5f05dabc |
135 | This is fairly secure because C</\w+/> doesn't normally match shell |
425e5e39 |
136 | metacharacters, nor are dot, dash, or at going to mean something special |
137 | to the shell. Use of C</.+/> would have been insecure in theory because |
138 | it lets everything through, but Perl doesn't check for that. The lesson |
139 | is that when untainting, you must be exceedingly careful with your patterns. |
140 | Laundering data using regular expression is the I<ONLY> mechanism for |
141 | untainting dirty data, unless you use the strategy detailed below to fork |
142 | a child of lesser privilege. |
143 | |
a034a98d |
144 | The example does not untaint $data if C<use locale> is in effect, |
145 | because the characters matched by C<\w> are determined by the locale. |
146 | Perl considers that locale definitions are untrustworthy because they |
147 | contain data from outside the program. If you are writing a |
148 | locale-aware program, and want to launder data with a regular expression |
149 | containing C<\w>, put C<no locale> ahead of the expression in the same |
150 | block. See L<perllocale/SECURITY> for further discussion and examples. |
151 | |
3a52c276 |
152 | =head2 Switches On the "#!" Line |
153 | |
154 | When you make a script executable, in order to make it usable as a |
155 | command, the system will pass switches to perl from the script's #! |
54310121 |
156 | line. Perl checks that any command line switches given to a setuid |
3a52c276 |
157 | (or setgid) script actually match the ones set on the #! line. Some |
54310121 |
158 | Unix and Unix-like environments impose a one-switch limit on the #! |
3a52c276 |
159 | line, so you may need to use something like C<-wU> instead of C<-w -U> |
54310121 |
160 | under such systems. (This issue should arise only in Unix or |
161 | Unix-like environments that support #! and setuid or setgid scripts.) |
3a52c276 |
162 | |
425e5e39 |
163 | =head2 Cleaning Up Your Path |
164 | |
1fef88e7 |
165 | For "Insecure C<$ENV{PATH}>" messages, you need to set C<$ENV{'PATH'}> to a |
1e422769 |
166 | known value, and each directory in the path must be non-writable by others |
167 | than its owner and group. You may be surprised to get this message even |
168 | if the pathname to your executable is fully qualified. This is I<not> |
169 | generated because you didn't supply a full path to the program; instead, |
170 | it's generated because you never set your PATH environment variable, or |
171 | you didn't set it to something that was safe. Because Perl can't |
172 | guarantee that the executable in question isn't itself going to turn |
173 | around and execute some other program that is dependent on your PATH, it |
54310121 |
174 | makes sure you set the PATH. |
a0d0e21e |
175 | |
176 | It's also possible to get into trouble with other operations that don't |
177 | care whether they use tainted values. Make judicious use of the file |
178 | tests in dealing with any user-supplied filenames. When possible, do |
fb73857a |
179 | opens and such B<after> properly dropping any special user (or group!) |
180 | privileges. Perl doesn't prevent you from opening tainted filenames for reading, |
a0d0e21e |
181 | so be careful what you print out. The tainting mechanism is intended to |
182 | prevent stupid mistakes, not to remove the need for thought. |
183 | |
425e5e39 |
184 | Perl does not call the shell to expand wild cards when you pass B<system> |
185 | and B<exec> explicit parameter lists instead of strings with possible shell |
186 | wildcards in them. Unfortunately, the B<open>, B<glob>, and |
54310121 |
187 | backtick functions provide no such alternate calling convention, so more |
188 | subterfuge will be required. |
425e5e39 |
189 | |
190 | Perl provides a reasonably safe way to open a file or pipe from a setuid |
191 | or setgid program: just create a child process with reduced privilege who |
192 | does the dirty work for you. First, fork a child using the special |
193 | B<open> syntax that connects the parent and child by a pipe. Now the |
194 | child resets its ID set and any other per-process attributes, like |
195 | environment variables, umasks, current working directories, back to the |
196 | originals or known safe values. Then the child process, which no longer |
197 | has any special permissions, does the B<open> or other system call. |
198 | Finally, the child passes the data it managed to access back to the |
5f05dabc |
199 | parent. Because the file or pipe was opened in the child while running |
425e5e39 |
200 | under less privilege than the parent, it's not apt to be tricked into |
201 | doing something it shouldn't. |
202 | |
54310121 |
203 | Here's a way to do backticks reasonably safely. Notice how the B<exec> is |
425e5e39 |
204 | not called with a string that the shell could expand. This is by far the |
205 | best way to call something that might be subjected to shell escapes: just |
fb73857a |
206 | never call the shell at all. |
cb1a09d0 |
207 | |
54310121 |
208 | use English; |
fb73857a |
209 | die "Can't fork: $!" unless defined $pid = open(KID, "-|"); |
cb1a09d0 |
210 | if ($pid) { # parent |
211 | while (<KID>) { |
212 | # do something |
425e5e39 |
213 | } |
cb1a09d0 |
214 | close KID; |
215 | } else { |
fb73857a |
216 | my @temp = ($EUID, $EGID); |
425e5e39 |
217 | $EUID = $UID; |
218 | $EGID = $GID; # XXX: initgroups() not called |
fb73857a |
219 | # Make sure privs are really gone |
220 | ($EUID, $EGID) = @temp; |
221 | die "Can't drop privileges" unless |
222 | $UID == $EUID and |
223 | $GID eq $EGID; # String test |
425e5e39 |
224 | $ENV{PATH} = "/bin:/usr/bin"; |
fb73857a |
225 | exec 'myprog', 'arg1', 'arg2' or |
425e5e39 |
226 | die "can't exec myprog: $!"; |
227 | } |
228 | |
fb73857a |
229 | A similar strategy would work for wildcard expansion via C<glob>, although |
230 | you can use C<readdir> instead. |
425e5e39 |
231 | |
232 | Taint checking is most useful when although you trust yourself not to have |
233 | written a program to give away the farm, you don't necessarily trust those |
234 | who end up using it not to try to trick it into doing something bad. This |
fb73857a |
235 | is the kind of security checking that's useful for set-id programs and |
425e5e39 |
236 | programs launched on someone else's behalf, like CGI programs. |
237 | |
238 | This is quite different, however, from not even trusting the writer of the |
239 | code not to try to do something evil. That's the kind of trust needed |
240 | when someone hands you a program you've never seen before and says, "Here, |
241 | run this." For that kind of safety, check out the Safe module, |
242 | included standard in the Perl distribution. This module allows the |
243 | programmer to set up special compartments in which all system operations |
244 | are trapped and namespace access is carefully controlled. |
245 | |
246 | =head2 Security Bugs |
247 | |
248 | Beyond the obvious problems that stem from giving special privileges to |
fb73857a |
249 | systems as flexible as scripts, on many versions of Unix, set-id scripts |
425e5e39 |
250 | are inherently insecure right from the start. The problem is a race |
251 | condition in the kernel. Between the time the kernel opens the file to |
fb73857a |
252 | see which interpreter to run and when the (now-set-id) interpreter turns |
425e5e39 |
253 | around and reopens the file to interpret it, the file in question may have |
254 | changed, especially if you have symbolic links on your system. |
255 | |
256 | Fortunately, sometimes this kernel "feature" can be disabled. |
257 | Unfortunately, there are two ways to disable it. The system can simply |
fb73857a |
258 | outlaw scripts with any set-id bit set, which doesn't help much. |
259 | Alternately, it can simply ignore the set-id bits on scripts. If the |
425e5e39 |
260 | latter is true, Perl can emulate the setuid and setgid mechanism when it |
261 | notices the otherwise useless setuid/gid bits on Perl scripts. It does |
262 | this via a special executable called B<suidperl> that is automatically |
54310121 |
263 | invoked for you if it's needed. |
425e5e39 |
264 | |
fb73857a |
265 | However, if the kernel set-id script feature isn't disabled, Perl will |
266 | complain loudly that your set-id script is insecure. You'll need to |
267 | either disable the kernel set-id script feature, or put a C wrapper around |
425e5e39 |
268 | the script. A C wrapper is just a compiled program that does nothing |
269 | except call your Perl program. Compiled programs are not subject to the |
fb73857a |
270 | kernel bug that plagues set-id scripts. Here's a simple wrapper, written |
425e5e39 |
271 | in C: |
272 | |
273 | #define REAL_PATH "/path/to/script" |
54310121 |
274 | main(ac, av) |
425e5e39 |
275 | char **av; |
276 | { |
277 | execv(REAL_PATH, av); |
54310121 |
278 | } |
cb1a09d0 |
279 | |
54310121 |
280 | Compile this wrapper into a binary executable and then make I<it> rather |
281 | than your script setuid or setgid. |
425e5e39 |
282 | |
283 | See the program B<wrapsuid> in the F<eg> directory of your Perl |
284 | distribution for a convenient way to do this automatically for all your |
285 | setuid Perl programs. It moves setuid scripts into files with the same |
286 | name plus a leading dot, and then compiles a wrapper like the one above |
287 | for each of them. |
288 | |
289 | In recent years, vendors have begun to supply systems free of this |
290 | inherent security bug. On such systems, when the kernel passes the name |
fb73857a |
291 | of the set-id script to open to the interpreter, rather than using a |
425e5e39 |
292 | pathname subject to meddling, it instead passes I</dev/fd/3>. This is a |
293 | special file already opened on the script, so that there can be no race |
294 | condition for evil scripts to exploit. On these systems, Perl should be |
295 | compiled with C<-DSETUID_SCRIPTS_ARE_SECURE_NOW>. The B<Configure> |
296 | program that builds Perl tries to figure this out for itself, so you |
297 | should never have to specify this yourself. Most modern releases of |
298 | SysVr4 and BSD 4.4 use this approach to avoid the kernel race condition. |
299 | |
300 | Prior to release 5.003 of Perl, a bug in the code of B<suidperl> could |
301 | introduce a security hole in systems compiled with strict POSIX |
302 | compliance. |
68dc0745 |
303 | |
304 | =head2 Protecting Your Programs |
305 | |
306 | There are a number of ways to hide the source to your Perl programs, |
307 | with varying levels of "security". |
308 | |
309 | First of all, however, you I<can't> take away read permission, because |
310 | the source code has to be readable in order to be compiled and |
311 | interpreted. (That doesn't mean that a CGI script's source is |
312 | readable by people on the web, though.) So you have to leave the |
313 | permissions at the socially friendly 0755 level. |
314 | |
315 | Some people regard this as a security problem. If your program does |
316 | insecure things, and relies on people not knowing how to exploit those |
317 | insecurities, it is not secure. It is often possible for someone to |
318 | determine the insecure things and exploit them without viewing the |
319 | source. Security through obscurity, the name for hiding your bugs |
320 | instead of fixing them, is little security indeed. |
321 | |
322 | You can try using encryption via source filters (Filter::* from CPAN). |
323 | But crackers might be able to decrypt it. You can try using the |
54310121 |
324 | byte code compiler and interpreter described below, but crackers might |
68dc0745 |
325 | be able to de-compile it. You can try using the native-code compiler |
326 | described below, but crackers might be able to disassemble it. These |
327 | pose varying degrees of difficulty to people wanting to get at your |
328 | code, but none can definitively conceal it (this is true of every |
329 | language, not just Perl). |
330 | |
331 | If you're concerned about people profiting from your code, then the |
332 | bottom line is that nothing but a restrictive licence will give you |
333 | legal security. License your software and pepper it with threatening |
334 | statements like "This is unpublished proprietary software of XYZ Corp. |
335 | Your access to it does not give you permission to use it blah blah |
336 | blah." You should see a lawyer to be sure your licence's wording will |
337 | stand up in court. |