Compound Element ID: 352 (Compound Element Variant: Composite)
Status: Draft
Description
Description Summary
The web application does not, or can not, sufficiently verify whether a well-formed, valid, consistent request was intentionally provided by the user who submitted the request.
Extended Description
When a web server is designed to receive a request from a client without any mechanism for verifying that it was intentionally sent, then it might be possible for an attacker to trick a client into making an unintentional request to the web server which will be treated as an authentic request. This can be done via a URL, image load, XMLHttpRequest, etc. and can result in exposure of data or unintended code execution.
The consequences will vary depending on the nature of the
functionality that is vulnerable to CSRF. An attacker could effectively
perform any operations as the victim. If the victim is an administrator
or privileged user, the consequences may include obtaining complete
control over the web application - deleting or stealing data,
uninstalling the product, or using it to launch other attacks against
all of the product's users. Because the attacker has the identity of the
victim, the scope of CSRF is limited only by the victim's privileges.
Likelihood of Exploit
Medium to High
Detection Methods
Manual Analysis
This weakness can be detected using tools and techniques that require
manual (human) analysis, such as penetration testing, threat modeling,
and interactive tools that allow the tester to record and modify an
active session.
Specifically, manual analysis can be useful for finding this weakness,
and for minimizing false positives assuming an understanding of business
logic. However, it might not achieve desired code coverage within
limited time constraints. For black-box analysis, if credentials are not
known for privileged accounts, then the most security-critical portions
of the application may not receive sufficient attention.
Consider using OWASP CSRFTester to identify potential issues and aid
in manual analysis.
Effectiveness: High
These may be more effective than strictly automated techniques. This
is especially the case with weaknesses that are related to design and
business rules.
Automated Static Analysis
CSRF is currently difficult to detect reliably using automated
techniques. This is because each application has its own implicit
security policy that dictates which requests can be influenced by an
outsider and automatically performed on behalf of a user, versus which
requests require strong confidence that the user intends to make the
request. For example, a keyword search of the public portion of a web
site is typically expected to be encoded within a link that can be
launched automatically when the user clicks on the link.
Effectiveness: Limited
Demonstrative Examples
Example 1
This example PHP code attempts to secure the form submission process
by validating that the user submitting the form has a valid session. A CSRF
attack would not be prevented by this countermeasure because the attacker
forges a request through the user's web browser in which a valid session
already exists.
The following HTML is intended to allow a user to update a
profile.
echo "Your profile has been successfully updated.";
}
This code may look protected since it checks for a valid session.
However, CSRF attacks can be staged from virtually any tag or HTML
construct, including image tags, links, embed or object tags, or other
attributes that load background images.
The attacker can then host code that will silently change the username
and email address of any user that visits the page while remaining
logged in to the target web application. The code might be an
innocent-looking web page such as:
Notice how the form contains hidden fields, so when it is loaded into
the browser, the user will not notice it. Because SendAttack() is
defined in the body's onload attribute, it will be automatically called
when the victim loads the web page.
Assuming that the user is already logged in to victim.example.com,
profile.php will see that a valid user session has been established,
then update the email address to the attacker's own address. At this
stage, the user's identity has been compromised, and messages sent
through this profile could be sent to the attacker's address.
web interface allows password changes or stopping
a virtual machine via CSRF
Potential Mitigations
Phase: Architecture and Design
Strategy: Libraries or Frameworks
Use a vetted library or framework that does not allow this weakness to
occur or provides constructs that make this weakness easier to
avoid.
For example, use anti-CSRF packages such as the OWASP CSRFGuard. [R.352.3]
Another example is the ESAPI Session Management control, which includes a component for CSRF. [R.352.9]
Phase: Implementation
Ensure that the application is free of cross-site scripting issues (CWE-79), because most CSRF defenses can be bypassed using attacker-controlled script.
Phase: Architecture and Design
Generate a unique nonce for each form, place the nonce into the form, and verify the nonce upon receipt of the form. Be sure that the nonce is not predictable (CWE-330). [R.352.5]
Note that this can be bypassed using XSS (CWE-79).
Phase: Architecture and Design
Identify especially dangerous operations. When the user performs a
dangerous operation, send a separate confirmation request to ensure that
the user intended to perform that operation.
Note that this can be bypassed using XSS (CWE-79).
Phase: Architecture and Design
Use the "double-submitted cookie" method as described by Felten and
Zeller:
When a user visits a site, the site should generate a pseudorandom
value and set it as a cookie on the user's machine. The site should
require every form submission to include this value as a form value and
also as a cookie value. When a POST request is sent to the site, the
request should only be considered valid if the form value and the cookie
value are the same.
Because of the same-origin policy, an attacker cannot read or modify
the value stored in the cookie. To successfully submit a form on behalf
of the user, the attacker would have to correctly guess the pseudorandom
value. If the pseudorandom value is cryptographically strong, this will
be prohibitively difficult.
This technique requires Javascript, so it may not work for browsers that have Javascript disabled. [R.352.4]
Note that this can probably be bypassed using XSS (CWE-79), or when using web technologies that enable the attacker to read raw headers from HTTP requests.
Phase: Architecture and Design
Do not use the GET method for any request that triggers a state
change.
Phase: Implementation
Check the HTTP Referer header to see if the request originated from an
expected page. This could break legitimate functionality, because users
or proxies may have disabled sending the Referer for privacy
reasons.
Note that this can be bypassed using XSS (CWE-79). An attacker could use XSS to generate a spoofed Referer, or to generate a malicious request from a page whose Referer would be allowed.
[R.352.1] [REF-17] Michael Howard, David LeBlanc
and John Viega. "24 Deadly Sins of Software Security". "Sin 2: Web-Server Related Vulnerabilities (XSS, XSRF, and
Response Splitting)." Page 37. McGraw-Hill. 2010.
The software stores security-critical state information about its users, or the software itself, in a location that is accessible to unauthorized actors.
Extended Description
If an attacker can modify the state information without detection, then it could be used to perform unauthorized actions or access unexpectedresources, since the application programmer does not expect that the state can be changed.
State information can be stored in various locations such as a cookie, in a hidden web form field, input parameter or argument, an environment variable, a database record, within a settings file, etc. All of these locations have the potential to be modified by an attacker. When this state information is used to control security or determine resource usage, then it may create a vulnerability. For example, an application may perform authentication, then save the state in an "authenticated=true" cookie. An attacker may simply create this cookie in order to bypass the authentication.
Time of Introduction
Architecture and Design
Implementation
Applicable Platforms
Languages
All
Technology Classes
Web-Server: (Often)
Common Consequences
Scope
Effect
Access Control
Technical Impact: Bypass protection
mechanism; Gain privileges / assume
identity
An attacker could potentially modify the state in malicious ways. If
the state is related to the privileges or level of authentication that
the user has, then state modification might allow the user to bypass
authentication or elevate privileges.
Confidentiality
Technical Impact: Read application
data
The state variables may contain sensitive information that should not
be known by the client.
Availability
Technical Impact: DoS: crash / exit /
restart
By modifying state variables, the attacker could violate the
application's expectations for the contents of the state, leading to a
denial of service due to an unexpected error condition.
Likelihood of Exploit
High
Enabling Factors for Exploitation
An application maintains its own state and/or user state (i.e. application
is stateful).
State information can be affected by the user of an application through
some means other than the legitimate state transitions (e.g. logging into
the system, purchasing an item, making a payment, etc.)
An application does not have means to detect state tampering and behave in
a fail safe manner.
Demonstrative Examples
Example 1
In the following example, an authentication flag is read from a
browser cookie, thus allowing for external control of user state
data.
(Bad Code)
Example
Language: Java
Cookie[] cookies = request.getCookies();
for (int i =0; i< cookies.length; i++) {
Cookie c = cookies[i];
if (c.getName().equals("authenticated") &&
Boolean.TRUE.equals(c.getValue())) {
authenticated = true;
}
}
Example 2
The following code uses input from an HTTP request to create a file name. The programmer has not considered the possibility that an attacker could provide a file name such as "../../tomcat/conf/server.xml", which causes the application to delete one of its own configuration files (CWE-22).
File rFile = new File("/usr/local/apfr/reports/" + rName);
...
rFile.delete();
Example 3
The following code uses input from a configuration file to determine
which file to open and echo back to the user. If the program runs with
privileges and malicious users can change the configuration file, they can
use the program to read any file on the system that ends with the extension
.txt.
(Bad Code)
Example
Language: Java
fis = new FileInputStream(cfg.getProperty("sub")+".txt");
amt = fis.read(arr);
out.println(arr);
Example 4
This program is intended to execute a command that lists the
contents of a restricted directory, then performs other actions. Assume that
it runs with setuid privileges in order to bypass the permissions check by
the operating system.
(Bad Code)
Example
Language: C
#define DIR "/restricted/directory"
char cmd[500];
sprintf(cmd, "ls -l %480s", DIR);
/* Raise privileges to those needed for accessing DIR.
*/
RaisePrivileges(...);
system(cmd);
DropPrivileges(...);
...
This code may look harmless at first, since both the directory and the
command are set to fixed values that the attacker can't control. The
attacker can only see the contents for DIR, which is the intended
program behavior. Finally, the programmer is also careful to limit the
code that executes with raised privileges.
However, because the program does not modify the PATH environment
variable, the following attack would work:
The user sets the PATH to reference a directory under that user's
control, such as "/my/dir/".
The user creates a malicious program called "ls", and puts that
program in /my/dir
The user executes the program.
When system() is executed, the shell consults the PATH to find the
ls program
The program finds the malicious program, "/my/dir/ls". It doesn't
find "/bin/ls" because PATH does not contain "/bin/".
The program executes the malicious program with the raised
privileges.
Example 5
This code prints all of the running processes belonging to the
current user.
(Bad Code)
Example
Language: PHP
//assume getCurrentUser() returns a username that is
guaranteed to be alphanumeric (CWE-78)
$userName = getCurrentUser();
$command = 'ps aux | grep ' . $userName;
system($command);
This program is also vulnerable to a PATH based attack (CWE-426), as an attacker may be able to create malicious versions of the ps or grep commands. While the program does not explicitly raise privileges to run the system commands, the PHP interpreter may by default be running with higher privileges than users.
Example 6
The following code segment implements a basic server that uses the
"ls" program to perform a directory listing of the directory that is listed
in the "HOMEDIR" environment variable. The code intends to allow the user to
specify an alternate "LANG" environment variable. This causes "ls" to
customize its output based on a given language, which is an important
capability when supporting internationalization.
(Bad Code)
Example
Language: Perl
$ENV{"HOMEDIR"} = "/home/mydir/public/";
my $stream = AcceptUntrustedInputStream();
while (<$stream>) {
chomp;
if (/^ENV ([\w\_]+) (.*)/) {
$ENV{$1} = $2;
}
elsif (/^QUIT/) { ... }
elsif (/^LIST/) {
open($fh, "/bin/ls -l $ENV{HOMEDIR}|");
while (<$fh>) {
SendOutput($stream, "FILEINFO: $_");
}
close($fh);
}
}
The programmer takes care to call a specific "ls" program and sets the HOMEDIR to a fixed value. However, an attacker can use a command such as "ENV HOMEDIR /secret/directory" to specify an alternate directory, enabling a path traversal attack (CWE-22). At the same time, other attacks are enabled as well, such as OS command injection (CWE-78) by setting HOMEDIR to a value such as "/tmp; rm -rf /". In this case, the programmer never intends for HOMEDIR to be modified, so input validation for HOMEDIR is not the solution. A partial solution would be a whitelist that only allows the LANG variable to be specified in the ENV command. Alternately, assuming this is an authenticated user, the language could be stored in a local file so that no ENV command at all would be needed.
While this example may not appear realistic, this type of problem
shows up in code fairly frequently. See CVE-1999-0073 in the observed
examples for a real-world example with similar behaviors.
Server allows client to specify the search path,
which can be modified to point to a program that the client has
uploaded.
Potential Mitigations
Phase: Architecture and Design
Understand all the potential locations that are accessible to
attackers. For example, some programmers assume that cookies and hidden
form fields cannot be modified by an attacker, or they may not consider
that environment variables can be modified before a privileged program
is invoked.
Phase: Architecture and Design
Strategy: Identify and Reduce Attack Surface
Store state information and sensitive data on the server side only.
Ensure that the system definitively and unambiguously keeps track of
its own state and user state and has rules defined for legitimate state
transitions. Do not allow any application user to affect state directly
in any way other than through legitimate actions leading to state
transitions.
If information must be stored on the client, do not do so without encryption and integrity checking, or otherwise having a mechanism on the server side to catch tampering. Use a message authentication code (MAC) algorithm, such as Hash Message Authentication Code (HMAC) [R.642.2]. Apply this against the state or sensitive data that you have to expose, which can guarantee the integrity of the data - i.e., that the data has not been modified. Ensure that you use an algorithm with a strong hash function (CWE-328).
Phase: Architecture and Design
Store state information on the server side only. Ensure that the
system definitively and unambiguously keeps track of its own state and
user state and has rules defined for legitimate state transitions. Do
not allow any application user to affect state directly in any way other
than through legitimate actions leading to state transitions.
Phase: Architecture and Design
Strategy: Libraries or Frameworks
Use a vetted library or framework that does not allow this weakness to
occur or provides constructs that make this weakness easier to
avoid.
With a stateless protocol such as HTTP, use some frameworks can
maintain the state for you.
Examples include ASP.NET View State and the OWASP ESAPI Session
Management feature.
Be careful of language features that provide state support, since
these might be provided as a convenience to the programmer and may not
be considering security.
Phase: Architecture and Design
For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.
Phases: Operation; Implementation
Strategy: Environment Hardening
When using PHP, configure the application so that it does not use register_globals. During implementation, develop the application so that it does not rely on this feature, but be wary of implementing a register_globals emulation that is subject to weaknesses such as CWE-95, CWE-621, and similar issues.
Phase: Testing
Use automated static analysis tools that target this type of weakness.
Many modern techniques use data flow analysis to minimize the number of
false positives. This is not a perfect solution, since 100% accuracy and
coverage are not feasible.
Phase: Testing
Use dynamic tools and techniques that interact with the software using
large test suites with many diverse inputs, such as fuzz testing
(fuzzing), robustness testing, and fault injection. The software's
operation may slow down, but it should not become unstable, crash, or
generate incorrect results.
Phase: Testing
Use tools and techniques that require manual (human) analysis, such as
penetration testing, threat modeling, and interactive tools that allow
the tester to record and modify an active session. These may be more
effective than strictly automated techniques. This is especially the
case with weaknesses that are related to design and business
rules.
[R.642.3] [REF-17] Michael Howard, David LeBlanc
and John Viega. "24 Deadly Sins of Software Security". "Sin 4: Use of Magic URLs, Predictable Cookies, and Hidden
Form Fields." Page 75. McGraw-Hill. 2010.
According to WASC, "Insufficient Session Expiration is when a web site permits an attacker to reuse old session credentials or session IDs for authorization."
Time of Introduction
Architecture and Design
Implementation
Common Consequences
Scope
Effect
Access Control
Technical Impact: Bypass protection
mechanism
Demonstrative Examples
Example 1
The following snippet was taken from a J2EE web.xml deployment
descriptor in which the session-timeout parameter is explicitly defined (the
default value depends on the container). In this case the value is set to
-1, which means that a session will never expire.
(Bad Code)
Example
Language: Java
<web-app>
[...snipped...]
<session-config>
<session-timeout>-1</session-timeout>
</session-config>
</web-app>
Potential Mitigations
Phase: Implementation
Set sessions/credentials expiration date.
Other Notes
The lack of proper session expiration may improve the likely success of
certain attacks. For example, an attacker may intercept a session ID,
possibly via a network sniffer or Cross-site Scripting attack. Although
short session expiration times do not help if a stolen token is immediately
used, they will protect against ongoing replaying of the session ID. In
another scenario, a user might access a web site from a shared computer
(such as at a library, Internet cafe, or open work environment).
Insufficient Session Expiration could allow an attacker to use the browser's
back button to access web pages previously accessed by the victim.
product does not sufficiently distinguish external
HTML from internal, potentially dangerous HTML, allowing bypass using
special strings in the page title. Overlaps special
elements.
Unintended Proxy or Intermediary ('Confused Deputy')
Definition in a New Window
Weakness ID: 441 (Weakness Class)
Status: Draft
Description
Description Summary
The software receives a request, message, or directive from an upstream component, but the software does not sufficiently preserve the original source of the request before forwarding the request to an external actor that is outside of the software's control sphere. This causes the software to appear to be the source of the request, leading it to act as a proxy or other intermediary between the upstream component and the external actor.
Extended Description
If an attacker cannot directly contact a target, but the software has access to the target, then the attacker can send a request to the software and have it be forwarded from the target. The request would appear to be coming from the software's system, not the attacker's system. As a result, the attacker can bypass access controls (such as firewalls) or hide the source of malicious requests, since the requests would not be coming directly from the attacker.
Since proxy functionality and message-forwarding often serve a legitimate purpose, this issue only becomes a vulnerability when:
The software runs with different privileges or on a different system, or otherwise has different levels of access than the upstream component;
The attacker is prevented from making the request directly to the target; and
The attacker can create a request that the proxy does not explicitly intend to be forwarded on the behalf of the requester. Such a request might point to an unexpected hostname, port number, or service. Or, the request might be sent to an allowed service, but the request could contain disallowed directives, commands, or resources.
Alternate Terms
Confused Deputy:
This weakness is sometimes referred to as the "Confused deputy"
problem, in which an attacker misused the authority of one victim (the
"confused deputy") when targeting another victim.
Time of Introduction
Architecture and Design
Applicable Platforms
Languages
Language-independent
Common Consequences
Scope
Effect
Non-Repudiation
Access Control
Technical Impact: Gain privileges / assume
identity; Hide activities
FTP bounce attack. The design of the protocol
allows an attacker to modify the PORT command to cause the FTP server to
connect to other machines besides the
attacker's.
This weakness has a chaining relationship with CWE-668 (Exposure of Resource to Wrong Sphere) because the proxy effectively provides the attacker with access to the target's resources that the attacker cannot directly obtain.
Theoretical Notes
It could be argued that the "confused deputy" is a fundamental aspect of
most vulnerabilities that require an active attacker. Even for common
implementation issues such as buffer overflows, SQL injection, OS command
injection, and path traversal, the vulnerable program already has the
authorization to run code or access files. The vulnerability arises when the
attacker causes the program to run unexpected code or access unexpected
files.
Description
