Tinyauth is an authentication and authorization server. Prior to 5.1.0, Tinyauth exposes a remotely observable timing difference between authentication attempts for existing and nonexistent local usernames. internal/controller/user_controller.go loginHandler and internal/middleware/context_middleware.go basicAuth return quickly after internal/service/auth_service.go reports a missing user, while an existing user causes bcrypt password verification work. Repeated measurements can therefore disclose valid usernames and support targeted credential attacks. This issue is fixed in version 5.1.0.

Affected: Go: github.com/tinyauthapp/tinyauth

### Summary It's possible to enumerate users through a timing oracle. In other words: I can easily check if a username exists or not by observing the timing differences between logins.

### PoC Setup a tinyauth server with a local user. It can be over the network.

Try to log in with the local user, using an incorrect password: there is a noticeable delay. You know that the user exists.

Now try to log in with a username that does not exist, using an incorrect password: it will complete almost immediately. You know that the user does not exist.

### Expected vs actual behavior

Login attempts should take roughly the same amount of time when the user exists vs when the user does not exist. Right now, nonexistent user attempts are way faster, meaning if your login attempt is slow, then the username exists for sure.

### Details

Existing usernames will take an extra ~50 milliseconds to respond to login attempts, while non-existing usernames will take only ~50 microseconds (1000x less time) to return an incorrect password response.

Even taking network traffic into consideration, it is trivial to check whether a local user exists or not.

=== Existing user, wrong password === #1: 50.52ms #2: 46.24ms #3: 43.57ms

=== Nonexistent user === #1: 43.03µs #2: 48.45µs #3: 59.33µs

### Suggested fix

This can be solved by checking a dummy password hash when a user is not found before returning a query, this will mimic the exact same delay irrelevant of hardware capabilities.…