The original daemon continues to run and host the existing sessions. Your existing connection will not be killed (it is unrelated to TCP timeouts as suggested by some comments).
You can verify this by
- Print your process table tree
- Find your sshd and its children
- restart sshd
- ssh to the server in a new window (do not close your existing connection)
- Print your process table tree again
- Find your sshd and its children and validate the process ids are identical.
- Find the new
sshd
daemon process running
If sshd did indeed kill your ssh session, then you would lose your bash shell no matter how "fast" it restarts compared to TCP timeout configurations.