Don't close sockets that we're using for sessions

[dwelch-r7]

Spotted an issue where when targeting a single host with multiple user/pass combinations that if a session was created then the next login attempt would close the socket being used by the newly created session. I've fixed this by explicitly setting the socket to nil in the login scanner.
There were also some issues with the error handling where we were trying to access a clients socket incorrectly so I've addressed that in here too by passing the socket in as part of the result object so all error handling is the same across implementations.

Verification Steps

• boot up msfconsole
• create a set of user/pass combos
• use each of the login modules with the set of credentials
• ensure a session is generated for each of mysql_login, postgresql_login, mssql_login and smb_login
• interact with the session(s) to make sure they remain valid after more attempted logins

[adfoster-r7]

Needs a rebase 👍

[cgranleese-r7]

Tested against:

• postgres:16.1
• mcr.microsoft.com/mssql/server:2022-preview-ubuntu-22.04
• mariadb:11.2.2
• Windows 10

Everything seemed to be working as expected 👍

Before

After

[cgranleese-r7]

Have a question, in the description you mention:

I've fixed this by explicitly setting the socket to nil in the login scanner.

But it seems that was only done in the login scanner for MySQL and was already present in SMB. Just wondering if the description is outdated or was it not required for MSSQL or Postgres?

[cgranleese-r7]

Looks good to me 👍

[cgranleese-r7]

Release Notes

This PR fixes an issue were the socket would be closed if targeting a single host with multiple user_file/pass_file module option combinations. This was caused when a session was successfully opened but then the next login attempt would close the socket being used by the newly created session.