Generated from doc/security.xml; do not edit directly.

neon is intended to be secure against a specific threat model: use of a malicious HTTP server. Under this threat model, a range of attacks are possible against a client when the user (or application) can be tricked into accessing an HTTP server which is controlled by an attacker. This section documents various types of possible attack and describes what mitigation is used in neon.

Arguments passed to the neon API are trusted application input and are not validated for correctness or potential security impact. Use of untrusted API input is explicitly outside the threat model described here.

neon uses fixed resource limits to prevent the following attacks:

• memory/CPU consumption attack using an unbounded number of response header fields

• memory consumption attack using an unbounded length of individual response header lines (or continuation headers)

• memory consumption attack against the PROPFIND code using an unbounded number of properties (propstat elements) per resource

• memory consumption attack against the PROPFIND code using an unbounded CDATA element in a "flat property" value

Memory consumption attacks against applications based on neon by use of unbounded response length are also possible, but must be mitigated at application level. Memory consumption in neon while reading response bodies is fixed and is not proportional to the response length.

Test cases for all the above attacks are present in the neon test suite.

When using a connection secured by SSL/TLS, it is necessary for clients to verify that the X.509 certificate presented by the server matches the server's expected identity. The algorithm required for this purpose is described in RFC 9110 and RFC 6125, and is implemented by neon in the following manner:

1. the host argument passed to ne_session_create() is the expected identity of the server

2. if the host argument is an IP literal (e.g. "198.51.100.42" or "[2001:db8::42]"), it is compared only to any iPAddress subjectAltName values present

3. otherwise, the host parameter is treated as a DNS hostname, and is compared with any dNSName subjectAltName values if present; if none match the hostname is compared with the most specific commonName attribute in the Subject name.

In the case where a server certificate is presented that does not match the expected identity (or is otherwise not trusted), neon will fail the handshake by default. This behaviour can be overridden by the use of a callback installed using ne_ssl_set_verify(), which allows the application to present the certificate details to a user for manual/off-line verification, if possible.

Test cases for the correctness of the implementation of the identity verification algorithm are present in the neon test suite.

Where error messages (as returned by ne_get_error()) contain data supplied by the server, the untrusted data is sanitised to remove both control characters and non-ASCII characters. This prevents any attacks where such error messages are exposed to the user and can potentially distort the presentation of the interface (for example, through the use of a carriage return character in a text user interface).

Authentication credentials can be compromised by a "downgrade attack" by an active attacker; for example, where a MITM presents a Basic authentication challenge in place of the server's Digest challenge. neon mitigates these attacks by allowing the application (and hence, user) to specify that only a specific set of authentication protocols is permitted.

When using Basic authentication, neon applies the scoping rules from RFC 7617 Section 2.2 to limit reuse of cached credentials within a session.

neon supports the Digest and Negotiate authentication schemes, which both allow authentication of users without passing credentials in cleartext over the wire.

When using Digest authentication, neon uses hash algorithm implementations from the configured SSL/TLS toolkit where possible, or falls back to a bundled MD5 implementation where SSL/TLS is not supported. If available, the SSL/TLS toolkit is also used to generate random cnonce values. The userhash field is supported for username privacy (this depends on server-side enablement). The full range of hash algorithms specified in RFC 7616 Section 6.1 is supported if configured using OpenSSL 1.1.0 or later.