Difference between revisions of "Unsafe string replacement"
(40 intermediate revisions by 6 users not shown) | |||
Line 1: | Line 1: | ||
+ | {{info|<center>This is a [[programming language]] agnostic [[vulnerability]], effecting any [[application]] which incorrectly uses string replacement to [[sanitize]] data.</center>}}{{prereq|[[programming]]}} | ||
+ | |||
+ | __FORCETOC__ | ||
+ | <font size="-2">Special thanks to [[User:hatter|hatter]] for his contributions to this article.</font> | ||
=Overview= | =Overview= | ||
− | Unsafe | + | Unsafe string replacement occurs when a replacement call is used to remove a series of text longer than one character from a string, invoked only once, to sanitize it. Because string replacement ([http://php.net/manual/en/function.str-replace.php str_replace] in [[PHP]], =~ s/// in [[Perl]], etc) functions only do a single replacement, it is necessary to loop over them until all unsafe characters or strings are removed if you are replacing more than a single character. This also applies to replacements powered by regular expressions. |
+ | |||
+ | =Examples= | ||
− | == | + | ==[[PHP]]== |
A trivial example: | A trivial example: | ||
Line 19: | Line 25: | ||
}} | }} | ||
− | First an attacker may try a simple | + | First an attacker may try a simple directory transversal attack, using '../' to escape. The result: |
Safe filepath is 'etc/passwd' | Safe filepath is 'etc/passwd' | ||
Line 30: | Line 36: | ||
Safe filepath is '../../../../../../../../../../../etc/passwd' | Safe filepath is '../../../../../../../../../../../etc/passwd' | ||
− | [contents of /etc/passwd] | + | '''[contents of /etc/passwd]''' |
Even if '../' is replaced twice, it can be easily bypassed by using ......///. No matter how many times the replacement is made, the attacker simply needs to nest another layer. | Even if '../' is replaced twice, it can be easily bypassed by using ......///. No matter how many times the replacement is made, the attacker simply needs to nest another layer. | ||
Line 41: | Line 47: | ||
</source> | </source> | ||
}} | }} | ||
+ | Bypassed by '<<??' | ||
+ | |||
{{code | {{code | ||
|text= | |text= | ||
<source lang="php"> | <source lang="php"> | ||
− | str_replace('<script', '', $source); | + | str_replace(array('<script', '<img'), '', $source); |
</source> | </source> | ||
}} | }} | ||
+ | Bypassed by '<<imgscript>' | ||
+ | |||
{{code | {{code | ||
|text= | |text= | ||
Line 53: | Line 63: | ||
</source> | </source> | ||
}} | }} | ||
+ | Bypassed by 'file:/file:///' | ||
− | == | + | This also affects regular expressions: |
+ | {{code|text=<source lang="php"> | ||
+ | $string = preg_replace('@[.]{2}\/@','',$string); | ||
+ | </source>}} | ||
− | + | == [[Perl#Regular_Expressions|PCRE]] == | |
+ | Perl Example: | ||
+ | |||
+ | {{code|text=<source lang="perl"> | ||
+ | $string =~ s/[.]{2}\///g; | ||
+ | </source>}} | ||
+ | |||
+ | To test if perl compatible regular expressions are vulnerable to the attack, a small example perl script, ''rexpose.pl'', is attached: | ||
+ | {{code|text=<source lang="perl"> | ||
+ | #!/usr/bin/perl | ||
+ | use strict; | ||
+ | use Getopt::Std; | ||
+ | |||
+ | my %opts = (); | ||
+ | getopts('s:p:',\%opts); | ||
+ | |||
+ | my $string = $opts{s} if ($opts{s}) or usage(); | ||
+ | my $pattern; | ||
+ | $pattern = $opts{p} if ($opts{p}); | ||
+ | $pattern = '[.]{2}\/' unless (defined $pattern); | ||
+ | $string =~ s/$pattern//g; | ||
+ | print "$string\n"; | ||
+ | |||
+ | sub usage | ||
+ | { | ||
+ | print "rexpose.pl -s [string] [-p pattern]\n"; | ||
+ | print "Regex exposure will display string replacements using s/\$pattern//g. Pattern defaults to '[.]{2}\\/' to match '../'.\n"; | ||
+ | exit(0); | ||
+ | } | ||
+ | </source>}} | ||
+ | |||
+ | * First, a check is performed to make sure that the string is being stripped at all to being with: | ||
+ | [user@host ~]$ perl rexpose.pl -s ../../../../../../path/to/file | ||
+ | path/to/file | ||
+ | * To identify if the replacement is vulnerable, a check using '''....//''' in stead of '''../''' is performed: | ||
+ | [user@host ~]$ perl rexpose.pl -s ....//....//....//....//....//....//path/to/file | ||
+ | ../../../../../../path/to/file | ||
+ | |||
+ | == [[Python]] == | ||
+ | |||
+ | Python also suffers from this issue: | ||
+ | |||
+ | {{code|text=<source lang="python"> | ||
+ | >>> string = "....//....//....//" | ||
+ | >>> string.replace("../","") | ||
+ | '../../../' | ||
+ | </source>}} | ||
+ | |||
+ | Example with regular expressions: | ||
+ | |||
+ | {{code|text=<source lang="python"> | ||
+ | string = re.sub('[.]{2}\/','',string) | ||
+ | </source>}} | ||
+ | |||
+ | == [[Ruby]] == | ||
+ | |||
+ | A Ruby example: | ||
+ | |||
+ | {{code|text=<source lang="ruby"> | ||
+ | string =~ s/[.]{2}\///g | ||
+ | # OR | ||
+ | # string = string.gsub("../", "") | ||
+ | </source>}} | ||
+ | |||
+ | =Defense= | ||
+ | |||
+ | If one must use str_replace or PCRE for removal of a string (this does not apply to single bytes), it must be done using a loop or called recursively: | ||
+ | |||
+ | ==[[PHP]]== | ||
+ | * Loop | ||
{{code | {{code | ||
|text= | |text= | ||
<source lang="php"> | <source lang="php"> | ||
− | + | function safe_str_replace($search, $replace, $subject) { | |
− | $ | + | while(strstr($subject, $search) !== FALSE) { |
+ | $subject = str_replace($search, $replace, $subject); | ||
+ | } | ||
− | + | return $subject; | |
− | + | ||
− | + | ||
− | $ | + | |
} | } | ||
+ | </source> | ||
+ | }} | ||
− | + | * Recursion | |
− | + | {{code | |
− | + | |text= | |
+ | <source lang="php"> | ||
+ | function safe_str_replace($search, $replace, $subject) { | ||
+ | if (strstr($subject, $search) !== FALSE) { | ||
+ | return safe_str_replace($search,$replace,str_replace($search, $replace, $subject)); | ||
+ | } | ||
+ | return $subject; | ||
+ | } | ||
</source> | </source> | ||
}} | }} | ||
− | + | ==[[Perl#Regular_Expressions|PCRE]]== | |
+ | * When patching ''rexpose.pl'', line 12 contains the following line: | ||
+ | {{code|text=<source lang="perl"> | ||
+ | $string =~ s/$pattern//g; | ||
+ | </source>}} | ||
+ | * This can be fixed with a golfed until loop: | ||
+ | {{code|text=<source lang="perl"> | ||
+ | $string =~ s/$pattern//g until ($string !~ $pattern); | ||
+ | </source>}} | ||
+ | * This patch was tested: | ||
+ | [user@host ~]$ perl rexpose.pl -s ....//....//....//....//....//....//path/to/file | ||
+ | path/to/file | ||
+ | |||
+ | ==[[Python]]== | ||
+ | |||
+ | * Recursion | ||
+ | |||
+ | {{code|text=<source lang="python"> | ||
+ | def sanitize(str): | ||
+ | str = str.replace("../", "") | ||
+ | if "../" in str: | ||
+ | return sanitize(str) | ||
+ | return str | ||
+ | </source>}} | ||
+ | |||
+ | == [[Ruby]] == | ||
+ | Ruby sanitizing function: | ||
+ | {{code|text=<source lang="ruby"> | ||
+ | def safe_sub(pattern, replacement, string) | ||
+ | if (string ~ '/#{pattern}/') return safe_sub(pattern, replacement, string.gsub(pattern,replacement)) | ||
+ | return string | ||
+ | end | ||
+ | </source>}} | ||
+ | |||
+ | ==Whitelisting using PCRE== | ||
+ | It is always best to use whitelisting to sanitize input. This can be done by using regular expressions to test to be sure that the input contains valid data. For example, if the developer only wants to refer to a file in the same directory: | ||
+ | {{code|text=<source lang="perl"> | ||
+ | my $string = <>; | ||
+ | print $string if ($string ~ /^[\w\d\_\-]+[.][\w\d]{2,4}$/g); | ||
+ | </source>}} | ||
+ | |||
+ | |||
+ | = Auditing = | ||
+ | {{code|text=<source lang="bash"> | ||
+ | find -regextype posix-awk -regex ".*\.e?(rb|py|php)" -exec grep -HnC2 \\b\\(.*sub\|replace\\)\( '{}' \; &> string_replacement.txt | ||
+ | </source>}} | ||
+ | |||
+ | {{exploitation}}{{programming}} | ||
+ | {{social}} | ||
+ | [[Category:Secure programming]] |
Latest revision as of 05:51, 2 December 2012
Unsafe string replacement requires a basic understanding of programming |
Special thanks to hatter for his contributions to this article.
Contents
Overview
Unsafe string replacement occurs when a replacement call is used to remove a series of text longer than one character from a string, invoked only once, to sanitize it. Because string replacement (str_replace in PHP, =~ s/// in Perl, etc) functions only do a single replacement, it is necessary to loop over them until all unsafe characters or strings are removed if you are replacing more than a single character. This also applies to replacements powered by regular expressions.
Examples
PHP
A trivial example:
<?php $filepath = $_GET['file']; $safe_filepath = str_replace('../', '', $filepath); echo("Safe filepath is '" . $safe_filepath . "'<br />"); include($safe_filepath); ?> |
First an attacker may try a simple directory transversal attack, using '../' to escape. The result:
Safe filepath is 'etc/passwd'
No dice, the dangerous string ('../') is dutifully sanitized by str_replace. But, our attacker isn't going to give up yet, now armed with the knowledge that '../' is being filtered out, he may try:
test.php?file=....//....//....//....//....//....//....//....//....//....//....//....//....//etc/passwd
The result:
Safe filepath is '../../../../../../../../../../../etc/passwd' [contents of /etc/passwd]
Even if '../' is replaced twice, it can be easily bypassed by using ......///. No matter how many times the replacement is made, the attacker simply needs to nest another layer.
Other examples of unsafe uses of string replacement include:
str_replace('<?', '', $source); |
Bypassed by '<<??'
str_replace(array('<script', '<img'), '', $source); |
Bypassed by '<<imgscript>'
str_replace('file://', '', $source); |
Bypassed by 'file:/file:///'
This also affects regular expressions:
$string = preg_replace('@[.]{2}\/@','',$string); |
PCRE
Perl Example:
$string =~ s/[.]{2}\///g; |
To test if perl compatible regular expressions are vulnerable to the attack, a small example perl script, rexpose.pl, is attached:
#!/usr/bin/perl use strict; use Getopt::Std; my %opts = (); getopts('s:p:',\%opts); my $string = $opts{s} if ($opts{s}) or usage(); my $pattern; $pattern = $opts{p} if ($opts{p}); $pattern = '[.]{2}\/' unless (defined $pattern); $string =~ s/$pattern//g; print "$string\n"; sub usage { print "rexpose.pl -s [string] [-p pattern]\n"; print "Regex exposure will display string replacements using s/\$pattern//g. Pattern defaults to '[.]{2}\\/' to match '../'.\n"; exit(0); } |
- First, a check is performed to make sure that the string is being stripped at all to being with:
[user@host ~]$ perl rexpose.pl -s ../../../../../../path/to/file path/to/file
- To identify if the replacement is vulnerable, a check using ....// in stead of ../ is performed:
[user@host ~]$ perl rexpose.pl -s ....//....//....//....//....//....//path/to/file ../../../../../../path/to/file
Python
Python also suffers from this issue:
>>> string = "....//....//....//" >>> string.replace("../","") '../../../' |
Example with regular expressions:
string = re.sub('[.]{2}\/','',string) |
Ruby
A Ruby example:
string =~ s/[.]{2}\///g # OR # string = string.gsub("../", "") |
Defense
If one must use str_replace or PCRE for removal of a string (this does not apply to single bytes), it must be done using a loop or called recursively:
PHP
- Loop
function safe_str_replace($search, $replace, $subject) { while(strstr($subject, $search) !== FALSE) { $subject = str_replace($search, $replace, $subject); } return $subject; } |
- Recursion
function safe_str_replace($search, $replace, $subject) { if (strstr($subject, $search) !== FALSE) { return safe_str_replace($search,$replace,str_replace($search, $replace, $subject)); } return $subject; } |
PCRE
- When patching rexpose.pl, line 12 contains the following line:
$string =~ s/$pattern//g; |
- This can be fixed with a golfed until loop:
$string =~ s/$pattern//g until ($string !~ $pattern); |
- This patch was tested:
[user@host ~]$ perl rexpose.pl -s ....//....//....//....//....//....//path/to/file path/to/file
Python
- Recursion
def sanitize(str): str = str.replace("../", "") if "../" in str: return sanitize(str) return str |
Ruby
Ruby sanitizing function:
def safe_sub(pattern, replacement, string) if (string ~ '/#{pattern}/') return safe_sub(pattern, replacement, string.gsub(pattern,replacement)) return string end |
Whitelisting using PCRE
It is always best to use whitelisting to sanitize input. This can be done by using regular expressions to test to be sure that the input contains valid data. For example, if the developer only wants to refer to a file in the same directory:
my $string = <>; print $string if ($string ~ /^[\w\d\_\-]+[.][\w\d]{2,4}$/g); |
Auditing
find -regextype posix-awk -regex ".*\.e?(rb|py|php)" -exec grep -HnC2 \\b\\(.*sub\|replace\\)\( '{}' \; &> string_replacement.txt |
<center>