simplesamlphp/saml2-legacy

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

There's a signature confusion attack in the HTTPRedirect binding. An attacker with any signed SAMLResponse via the HTTP-Redirect binding can cause the application to accept an unsigned message.

I believe that it exists for v4 only. I have not yet developed a PoC.

V5 is well designed and instead builds the signed query from the same message that will be consumed.

Details

What is verified

The data['SignedQuery'] is the string that will be verified by the public key.

It is defined here: https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L178-L217

THe code will iterate through each parameter name. Notably, sigQuery is overridden each time when processing, making the last of SAMLRequest/SAMLResponse used for sigQuery.

For example, given:

SAMLRequest=a&SAMLResponse=idpsigned

SAMLResponse=idpsigned will be set as sigQuery, then later verified

What is actually processed

Processing uses SAMLRequest parameter value first, (if it exists) then SAMLResponse:

https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L104-L113

Given this, the contents that are processed might not be the same as the data that is actually verified.

Exploiting

Suppose an attacker has a signed HTTP Redirect binding from IdP, say a signed logout response. :

SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature

Then an attacker can append SAMLRequest in front:

SAMLRequest=unverifieddata&SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature=..

SimpleSAMLPhp will only verify the SAMLResponse, but will actually use the SAMLRequest contents. The impact here is increased because there's no checks that SAMLRequest actually contains a Request, it could instead contain an Response, which allows the attacker to effectively impersonate any user within the SP.

IdPs

Microsoft Azure AD/Entra (and likely ADFS) signs the LogoutResponse via this SimpleSign format in HTTP Redirect binding. If an attacker logs out of Entra, they will be able to extract a valid Signature.

Attached is an HTTP Request when an I initiated a SLO request from the service provider to the IdP (entra). Then IdP POSTed this SAMLResponse with HTTP Redirect binding signature, via the user browser to the SP. It should be possible to carry out the described attack with this.

https://webhook.site/c6038292-6ef5-46ac-973d-d7c25520ec48/logout?SAMLResponse=fVJNa%2bMwEP0rRndZtmw5tnAMy%2fYSaC9N6aGXIsmjVMTRGI9M%2bvObdeihsPQ4w7x5HzM9mcs060c84ZqegWaMBNnhYc%2fejS1UW1TAnVU7XldK8s7JkcvOd60Db3zTsewVFgoY90zmBcsORCscIiUT061VyJqXJS%2fbl7LRUmrZ5mXdvLHsASiFaNKG%2fEhpJi3EFewH4jmnkEC4pqha2UnegFe8bozj3a4a%2bbhzUilZgKtbMW2yb7TxW%2foL7lkM9hTC2XnEOPvZXjDECb2N1lh7mvBsp%2bnsErDs8zJF0lsEe7YuUaOhQDqaC5BOTh%2f%2fPD3qmzE9L5jQ4cSGfrO43KG%2fgwwRLP8ssuHbIiXKryGOeKU8QhLSVN7WteejV8Bru%2bt4WynFbwE3bdVV5ahG0Ys759Dfj3VMJq30s%2fqLI2SvZlrhd020Tevj6hwQMTH04udS8b%2bHGL4A&Signature=Z%2f7gIPv7Gkgvqtwo0bzgXyum9IjHMfP0zTYuNbl%2fBUGlQ%2fU%2bbOZGZJ6Rk9wLUyvNQ5XlZRxZrfESNA%2bn0CVyIedsg9GxQKTi7VqPTJFJqEIP1BZaEpYYP3%2f6sFfLxfTMKecJoQdxnDE5Malte1hMj2UujWnLXOnp0CgO%2f%2fU2K52SoGckIzNDRB%2fJ6%2fysTn%2bDjBrmgdro%2fgdTyby9%2f3vm8dzY8pUkRCgMjlimShrZxr5U33wQvwPLIXlDgActr91RUtWKE0k8sy%2brshrK9DKLPo8AdTLk7NYhjSWdF7OG7uqgEeEo470tacqQuA09E0qDh8CWS%2bycLJijiGYWVyQa4Q%3d%3d&SigAlg=http%3a%2f%2fwww.w3.org%2f2001%2f04%2fxmldsig-more%23rsa-sha256

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

There's a signature confusion attack in the HTTPRedirect binding. An attacker with any signed SAMLResponse via the HTTP-Redirect binding can cause the application to accept an unsigned message.

I believe that it exists for v4 only. I have not yet developed a PoC.

V5 is well designed and instead builds the signed query from the same message that will be consumed.

Details

What is verified

The data['SignedQuery'] is the string that will be verified by the public key.

It is defined here: https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L178-L217

THe code will iterate through each parameter name. Notably, sigQuery is overridden each time when processing, making the last of SAMLRequest/SAMLResponse used for sigQuery.

For example, given:

SAMLRequest=a&SAMLResponse=idpsigned

SAMLResponse=idpsigned will be set as sigQuery, then later verified

What is actually processed

Processing uses SAMLRequest parameter value first, (if it exists) then SAMLResponse:

https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L104-L113

Given this, the contents that are processed might not be the same as the data that is actually verified.

Exploiting

Suppose an attacker has a signed HTTP Redirect binding from IdP, say a signed logout response. :

SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature

Then an attacker can append SAMLRequest in front:

SAMLRequest=unverifieddata&SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature=..

SimpleSAMLPhp will only verify the SAMLResponse, but will actually use the SAMLRequest contents. The impact here is increased because there's no checks that SAMLRequest actually contains a Request, it could instead contain an Response, which allows the attacker to effectively impersonate any user within the SP.

IdPs

Microsoft Azure AD/Entra (and likely ADFS) signs the LogoutResponse via this SimpleSign format in HTTP Redirect binding. If an attacker logs out of Entra, they will be able to extract a valid Signature.

Attached is an HTTP Request when an I initiated a SLO request from the service provider to the IdP (entra). Then IdP POSTed this SAMLResponse with HTTP Redirect binding signature, via the user browser to the SP. It should be possible to carry out the described attack with this.

https://webhook.site/c6038292-6ef5-46ac-973d-d7c25520ec48/logout?SAMLResponse=fVJNa%2bMwEP0rRndZtmw5tnAMy%2fYSaC9N6aGXIsmjVMTRGI9M%2bvObdeihsPQ4w7x5HzM9mcs060c84ZqegWaMBNnhYc%2fejS1UW1TAnVU7XldK8s7JkcvOd60Db3zTsewVFgoY90zmBcsORCscIiUT061VyJqXJS%2fbl7LRUmrZ5mXdvLHsASiFaNKG%2fEhpJi3EFewH4jmnkEC4pqha2UnegFe8bozj3a4a%2bbhzUilZgKtbMW2yb7TxW%2foL7lkM9hTC2XnEOPvZXjDECb2N1lh7mvBsp%2bnsErDs8zJF0lsEe7YuUaOhQDqaC5BOTh%2f%2fPD3qmzE9L5jQ4cSGfrO43KG%2fgwwRLP8ssuHbIiXKryGOeKU8QhLSVN7WteejV8Bru%2bt4WynFbwE3bdVV5ahG0Ys759Dfj3VMJq30s%2fqLI2SvZlrhd020Tevj6hwQMTH04udS8b%2bHGL4A&Signature=Z%2f7gIPv7Gkgvqtwo0bzgXyum9IjHMfP0zTYuNbl%2fBUGlQ%2fU%2bbOZGZJ6Rk9wLUyvNQ5XlZRxZrfESNA%2bn0CVyIedsg9GxQKTi7VqPTJFJqEIP1BZaEpYYP3%2f6sFfLxfTMKecJoQdxnDE5Malte1hMj2UujWnLXOnp0CgO%2f%2fU2K52SoGckIzNDRB%2fJ6%2fysTn%2bDjBrmgdro%2fgdTyby9%2f3vm8dzY8pUkRCgMjlimShrZxr5U33wQvwPLIXlDgActr91RUtWKE0k8sy%2brshrK9DKLPo8AdTLk7NYhjSWdF7OG7uqgEeEo470tacqQuA09E0qDh8CWS%2bycLJijiGYWVyQa4Q%3d%3d&SigAlg=http%3a%2f%2fwww.w3.org%2f2001%2f04%2fxmldsig-more%23rsa-sha256

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

There's a signature confusion attack in the HTTPRedirect binding. An attacker with any signed SAMLResponse via the HTTP-Redirect binding can cause the application to accept an unsigned message.

I believe that it exists for v4 only. I have not yet developed a PoC.

V5 is well designed and instead builds the signed query from the same message that will be consumed.

Details

What is verified

The data['SignedQuery'] is the string that will be verified by the public key.

It is defined here: https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L178-L217

THe code will iterate through each parameter name. Notably, sigQuery is overridden each time when processing, making the last of SAMLRequest/SAMLResponse used for sigQuery.

For example, given:

SAMLRequest=a&SAMLResponse=idpsigned

SAMLResponse=idpsigned will be set as sigQuery, then later verified

What is actually processed

Processing uses SAMLRequest parameter value first, (if it exists) then SAMLResponse:

https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L104-L113

Given this, the contents that are processed might not be the same as the data that is actually verified.

Exploiting

Suppose an attacker has a signed HTTP Redirect binding from IdP, say a signed logout response. :

SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature

Then an attacker can append SAMLRequest in front:

SAMLRequest=unverifieddata&SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature=..

SimpleSAMLPhp will only verify the SAMLResponse, but will actually use the SAMLRequest contents. The impact here is increased because there's no checks that SAMLRequest actually contains a Request, it could instead contain an Response, which allows the attacker to effectively impersonate any user within the SP.

IdPs

Microsoft Azure AD/Entra (and likely ADFS) signs the LogoutResponse via this SimpleSign format in HTTP Redirect binding. If an attacker logs out of Entra, they will be able to extract a valid Signature.

Attached is an HTTP Request when an I initiated a SLO request from the service provider to the IdP (entra). Then IdP POSTed this SAMLResponse with HTTP Redirect binding signature, via the user browser to the SP. It should be possible to carry out the described attack with this.

https://webhook.site/c6038292-6ef5-46ac-973d-d7c25520ec48/logout?SAMLResponse=fVJNa%2bMwEP0rRndZtmw5tnAMy%2fYSaC9N6aGXIsmjVMTRGI9M%2bvObdeihsPQ4w7x5HzM9mcs060c84ZqegWaMBNnhYc%2fejS1UW1TAnVU7XldK8s7JkcvOd60Db3zTsewVFgoY90zmBcsORCscIiUT061VyJqXJS%2fbl7LRUmrZ5mXdvLHsASiFaNKG%2fEhpJi3EFewH4jmnkEC4pqha2UnegFe8bozj3a4a%2bbhzUilZgKtbMW2yb7TxW%2foL7lkM9hTC2XnEOPvZXjDECb2N1lh7mvBsp%2bnsErDs8zJF0lsEe7YuUaOhQDqaC5BOTh%2f%2fPD3qmzE9L5jQ4cSGfrO43KG%2fgwwRLP8ssuHbIiXKryGOeKU8QhLSVN7WteejV8Bru%2bt4WynFbwE3bdVV5ahG0Ys759Dfj3VMJq30s%2fqLI2SvZlrhd020Tevj6hwQMTH04udS8b%2bHGL4A&Signature=Z%2f7gIPv7Gkgvqtwo0bzgXyum9IjHMfP0zTYuNbl%2fBUGlQ%2fU%2bbOZGZJ6Rk9wLUyvNQ5XlZRxZrfESNA%2bn0CVyIedsg9GxQKTi7VqPTJFJqEIP1BZaEpYYP3%2f6sFfLxfTMKecJoQdxnDE5Malte1hMj2UujWnLXOnp0CgO%2f%2fU2K52SoGckIzNDRB%2fJ6%2fysTn%2bDjBrmgdro%2fgdTyby9%2f3vm8dzY8pUkRCgMjlimShrZxr5U33wQvwPLIXlDgActr91RUtWKE0k8sy%2brshrK9DKLPo8AdTLk7NYhjSWdF7OG7uqgEeEo470tacqQuA09E0qDh8CWS%2bycLJijiGYWVyQa4Q%3d%3d&SigAlg=http%3a%2f%2fwww.w3.org%2f2001%2f04%2fxmldsig-more%23rsa-sha256

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

There's a signature confusion attack in the HTTPRedirect binding. An attacker with any signed SAMLResponse via the HTTP-Redirect binding can cause the application to accept an unsigned message.

I believe that it exists for v4 only. I have not yet developed a PoC.

V5 is well designed and instead builds the signed query from the same message that will be consumed.

Details

What is verified

The data['SignedQuery'] is the string that will be verified by the public key.

It is defined here: https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L178-L217

THe code will iterate through each parameter name. Notably, sigQuery is overridden each time when processing, making the last of SAMLRequest/SAMLResponse used for sigQuery.

For example, given:

SAMLRequest=a&SAMLResponse=idpsigned

SAMLResponse=idpsigned will be set as sigQuery, then later verified

What is actually processed

Processing uses SAMLRequest parameter value first, (if it exists) then SAMLResponse:

https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L104-L113

Given this, the contents that are processed might not be the same as the data that is actually verified.

Exploiting

Suppose an attacker has a signed HTTP Redirect binding from IdP, say a signed logout response. :

SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature

Then an attacker can append SAMLRequest in front:

SAMLRequest=unverifieddata&SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature=..

SimpleSAMLPhp will only verify the SAMLResponse, but will actually use the SAMLRequest contents. The impact here is increased because there's no checks that SAMLRequest actually contains a Request, it could instead contain an Response, which allows the attacker to effectively impersonate any user within the SP.

IdPs

Microsoft Azure AD/Entra (and likely ADFS) signs the LogoutResponse via this SimpleSign format in HTTP Redirect binding. If an attacker logs out of Entra, they will be able to extract a valid Signature.

Attached is an HTTP Request when an I initiated a SLO request from the service provider to the IdP (entra). Then IdP POSTed this SAMLResponse with HTTP Redirect binding signature, via the user browser to the SP. It should be possible to carry out the described attack with this.

https://webhook.site/c6038292-6ef5-46ac-973d-d7c25520ec48/logout?SAMLResponse=fVJNa%2bMwEP0rRndZtmw5tnAMy%2fYSaC9N6aGXIsmjVMTRGI9M%2bvObdeihsPQ4w7x5HzM9mcs060c84ZqegWaMBNnhYc%2fejS1UW1TAnVU7XldK8s7JkcvOd60Db3zTsewVFgoY90zmBcsORCscIiUT061VyJqXJS%2fbl7LRUmrZ5mXdvLHsASiFaNKG%2fEhpJi3EFewH4jmnkEC4pqha2UnegFe8bozj3a4a%2bbhzUilZgKtbMW2yb7TxW%2foL7lkM9hTC2XnEOPvZXjDECb2N1lh7mvBsp%2bnsErDs8zJF0lsEe7YuUaOhQDqaC5BOTh%2f%2fPD3qmzE9L5jQ4cSGfrO43KG%2fgwwRLP8ssuHbIiXKryGOeKU8QhLSVN7WteejV8Bru%2bt4WynFbwE3bdVV5ahG0Ys759Dfj3VMJq30s%2fqLI2SvZlrhd020Tevj6hwQMTH04udS8b%2bHGL4A&Signature=Z%2f7gIPv7Gkgvqtwo0bzgXyum9IjHMfP0zTYuNbl%2fBUGlQ%2fU%2bbOZGZJ6Rk9wLUyvNQ5XlZRxZrfESNA%2bn0CVyIedsg9GxQKTi7VqPTJFJqEIP1BZaEpYYP3%2f6sFfLxfTMKecJoQdxnDE5Malte1hMj2UujWnLXOnp0CgO%2f%2fU2K52SoGckIzNDRB%2fJ6%2fysTn%2bDjBrmgdro%2fgdTyby9%2f3vm8dzY8pUkRCgMjlimShrZxr5U33wQvwPLIXlDgActr91RUtWKE0k8sy%2brshrK9DKLPo8AdTLk7NYhjSWdF7OG7uqgEeEo470tacqQuA09E0qDh8CWS%2bycLJijiGYWVyQa4Q%3d%3d&SigAlg=http%3a%2f%2fwww.w3.org%2f2001%2f04%2fxmldsig-more%23rsa-sha256

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.

Summary

SimpleSAMLphp's HTTP-Artifact receive path can treat an unsigned embedded SAML Response as cryptographically valid for the wrong IdP.

In the HTTPArtifact::receive() flow, the SOAP ArtifactResponse receives a TLS-based validator from SOAPClient::addSSLValidator(). The embedded SAML Response then receives a validator that delegates signature validation to that outer ArtifactResponse. Later, the SP validates the embedded Response against metadata selected from the embedded response issuer, not necessarily the artifact issuer.

The critical issue is that SOAPClient::validateSSL() returns normally when the TLS public key does not match the key currently being validated. SAML2\Message::validate() treats any validator call that does not throw an exception as successful. As a result, an ArtifactResponse obtained from one IdP can validate an unsigned embedded SAML Response that claims to be issued by a different IdP.

In a multi-IdP/federation deployment where a malicious or lower-trust IdP can issue an HTTP-Artifact response to an SP, this can allow the attacker to authenticate to the SP as arbitrary users from a higher-trust victim IdP.

Impact

A malicious or lower-trust IdP in the same SP/federation trust set can authenticate to the SP as users from another IdP when HTTP-Artifact is used. The attacker can choose assertion attributes, NameID, and session data in the forged unsigned assertion.

This is an authentication bypass and identity-provider impersonation issue. In realistic federations, the security boundary between IdPs matters: a compromised or low-assurance IdP should not be able to mint identities for a high-assurance IdP.

Summary

There's a signature confusion attack in the HTTPRedirect binding. An attacker with any signed SAMLResponse via the HTTP-Redirect binding can cause the application to accept an unsigned message.

I believe that it exists for v4 only. I have not yet developed a PoC.

V5 is well designed and instead builds the signed query from the same message that will be consumed.

Details

What is verified

The data['SignedQuery'] is the string that will be verified by the public key.

It is defined here: https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L178-L217

THe code will iterate through each parameter name. Notably, sigQuery is overridden each time when processing, making the last of SAMLRequest/SAMLResponse used for sigQuery.

For example, given:

SAMLRequest=a&SAMLResponse=idpsigned

SAMLResponse=idpsigned will be set as sigQuery, then later verified

What is actually processed

Processing uses SAMLRequest parameter value first, (if it exists) then SAMLResponse:

https://github.com/simplesamlphp/saml2/blob/9545abd0d9d48388f2fa00469c5c1e0294f0303e/src/SAML2/HTTPRedirect.php#L104-L113

Given this, the contents that are processed might not be the same as the data that is actually verified.

Exploiting

Suppose an attacker has a signed HTTP Redirect binding from IdP, say a signed logout response. :

SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature

Then an attacker can append SAMLRequest in front:

SAMLRequest=unverifieddata&SAMLResponse=idpsigned&RelayState=...&SigAlg=...&Signature=..

SimpleSAMLPhp will only verify the SAMLResponse, but will actually use the SAMLRequest contents. The impact here is increased because there's no checks that SAMLRequest actually contains a Request, it could instead contain an Response, which allows the attacker to effectively impersonate any user within the SP.

IdPs

Microsoft Azure AD/Entra (and likely ADFS) signs the LogoutResponse via this SimpleSign format in HTTP Redirect binding. If an attacker logs out of Entra, they will be able to extract a valid Signature.

Attached is an HTTP Request when an I initiated a SLO request from the service provider to the IdP (entra). Then IdP POSTed this SAMLResponse with HTTP Redirect binding signature, via the user browser to the SP. It should be possible to carry out the described attack with this.

https://webhook.site/c6038292-6ef5-46ac-973d-d7c25520ec48/logout?SAMLResponse=fVJNa%2bMwEP0rRndZtmw5tnAMy%2fYSaC9N6aGXIsmjVMTRGI9M%2bvObdeihsPQ4w7x5HzM9mcs060c84ZqegWaMBNnhYc%2fejS1UW1TAnVU7XldK8s7JkcvOd60Db3zTsewVFgoY90zmBcsORCscIiUT061VyJqXJS%2fbl7LRUmrZ5mXdvLHsASiFaNKG%2fEhpJi3EFewH4jmnkEC4pqha2UnegFe8bozj3a4a%2bbhzUilZgKtbMW2yb7TxW%2foL7lkM9hTC2XnEOPvZXjDECb2N1lh7mvBsp%2bnsErDs8zJF0lsEe7YuUaOhQDqaC5BOTh%2f%2fPD3qmzE9L5jQ4cSGfrO43KG%2fgwwRLP8ssuHbIiXKryGOeKU8QhLSVN7WteejV8Bru%2bt4WynFbwE3bdVV5ahG0Ys759Dfj3VMJq30s%2fqLI2SvZlrhd020Tevj6hwQMTH04udS8b%2bHGL4A&Signature=Z%2f7gIPv7Gkgvqtwo0bzgXyum9IjHMfP0zTYuNbl%2fBUGlQ%2fU%2bbOZGZJ6Rk9wLUyvNQ5XlZRxZrfESNA%2bn0CVyIedsg9GxQKTi7VqPTJFJqEIP1BZaEpYYP3%2f6sFfLxfTMKecJoQdxnDE5Malte1hMj2UujWnLXOnp0CgO%2f%2fU2K52SoGckIzNDRB%2fJ6%2fysTn%2bDjBrmgdro%2fgdTyby9%2f3vm8dzY8pUkRCgMjlimShrZxr5U33wQvwPLIXlDgActr91RUtWKE0k8sy%2brshrK9DKLPo8AdTLk7NYhjSWdF7OG7uqgEeEo470tacqQuA09E0qDh8CWS%2bycLJijiGYWVyQa4Q%3d%3d&SigAlg=http%3a%2f%2fwww.w3.org%2f2001%2f04%2fxmldsig-more%23rsa-sha256

Summary

When loading an (untrusted) XML document, for example the SAMLResponse, it's possible to induce an XXE.

$options is defined as: https://github.com/simplesamlphp/saml2/blob/717c0adc4877ebd58428637e5626345e59fa0109/src/SAML2/DOMDocumentFactory.php#L41 including the DTDLoad option, which allows an attacker to read file contents from local file system OR internal network.

While there is the NONET option, an attacker can simply bypass if by using PHP filters: php://filter/convert.base64-encode/resource=http://URL OR FILE

From there an attacker can induce network connections and steal the targeted file OOB (haven't fully tested this).

RCE may be possible with the php://expect or php://phar wrappers, but this hasn't been tested.

Note: The mitigation here: https://github.com/simplesamlphp/saml2/blob/717c0adc4877ebd58428637e5626345e59fa0109/src/SAML2/DOMDocumentFactory.php#L63-L69 Comes too late, as the XML has already been loaded into a document. Mitigation:

Remove the LIBXML_DTDLOAD | LIBXML_DTDATTR options. Additionally, as a defense in depth measure, check if there is the string: <!DOCTYPE inside the XML before parsing it. (This is not a complete fix because someone may be able to exploit some parser differentials, to load a DOCTYPE, maybe through spacing like: <! DOCTYPE)

Summary

This library turned out to be vulnerable to Denial-of-Service attacks using XPath transforms. A mitigation has been put in place to restrict the number of transforms and to restrict transforms to only the transform-algorithms mentioned in the SAML 2.0 Core Specifications (and specifically refuse XPath transforms).

Impact

An attacker is able to send specially crafted messages to any entity relying on SimpleSAMLphp (or directly on this SAML2-library) to be able to perform a Denial-of-Service attack.