Using "**" as a pattern in Spring Security configuration for WebFlux creates a mismatch in pattern matching between Spring Security and Spring WebFlux, and the potential for a security bypass.

Spring Security, versions 5.7 prior to 5.7.5 and 5.6 prior to 5.6.9 could be susceptible to authorization rules bypass via forward or include dispatcher types. Specifically, an application is vulnerable when all of the following are true: The application expects that Spring Security applies security to forward and include dispatcher types. The application uses the AuthorizationFilter either manually or via the authorizeHttpRequests() method. The application configures the FilterChainProxy to apply to forward and/or include requests (e.g. spring.security.filter.dispatcher-types = request, error, async, forward, include). The application may forward or include the request to a higher privilege-secured endpoint.The application configures Spring Security to apply to every dispatcher type via authorizeHttpRequests().shouldFilterAllDispatcherTypes(true)

Spring Security, versions 5.7 prior to 5.7.5, and 5.6 prior to 5.6.9, and older unsupported versions could be susceptible to a privilege escalation under certain conditions. A malicious user or attacker can modify a request initiated by the Client (via the browser) to the Authorization Server which can lead to a privilege escalation on the subsequent approval. This scenario can happen if the Authorization Server responds with an OAuth2 Access Token Response containing an empty scope list (per RFC 6749, Section 5.1) on the subsequent request to the token endpoint to obtain the access token.

In spring security versions prior to 5.4.11+, 5.5.7+ , 5.6.4+ and older unsupported versions, RegexRequestMatcher can easily be misconfigured to be bypassed on some servlet containers. Applications using RegexRequestMatcher with `.` in the regular expression are possibly vulnerable to an authorization bypass.

Spring Security versions 5.5.x prior to 5.5.7, 5.6.x prior to 5.6.4, and earlier unsupported versions contain an integer overflow vulnerability. When using the BCrypt class with the maximum work factor (31), the encoder does not perform any salt rounds, due to an integer overflow error. The default settings are not affected by this CVE.