CareerCup

The question seems incomplete, and I would prefer to log the session id for ease of request.
However...

with userlog as (select ul.*,
                        rank() over (partition by ul.name, ul.event order by ul.name, ul.event, ul.event_time asc) as rn
                   from user_log ul
				   )
select u1.name
      ,u1.event_time login_time
      ,isnull(u2.event_time, getdate()) out_time
	  ,case when u2.event_time is null then 'online' else '-' end status
	  ,datediff(minute, u1.event_time, isnull(u2.event_time, getdate())) spend_time_in_m
	  ,datediff(minute, u1.event_time, isnull(u2.event_time, getdate()))/60 spend_time_in_h
	  ,(sum(datediff(minute, u1.event_time, isnull(u2.event_time, getdate()))) over (partition by u1.name order by u1.name))/60 as sum_spend_time
  from userlog u1 left join userlog u2 on u1.name = u2.name and u1.rn = u2.rn and u2.event = 'out'
 where u1.event = 'login'
 order by 1

select A,(la - c) as total_time_spend from (select A,B,c,lag(c,1) over (partition by A order by A) as la from t4 ) where la is not null;

or

with exp as ( select A,max(c) as logout,min(c) as login from t4 group by A)
SELECT A,(logout - login) as total_spent_time from exp;

or

with exp as (
select a,max(c) as logout,min(c) as login from t4 group by a order by a
)
select a, (logout - login) as total_time_spend from exp;

Here A is name, B is event, C is event time

This query gives the hours spent per user per day.

with cte as

( select name,event,time,row_number() over(partition by date(event_time), name order by event_time) as rnm from table)

select a.name,date(a.event_time),sum(datediff(hr,b.event_time,a.event_time)) from cte a join cte b on a.rnm=b.rnm-1 and a.event='login' group by a.name,date(a.event_time)


)

Following works if multiple login by student A

create table logins(name varchar(2),event varchar(10),time integer);
insert into logins values
('A','login',5),
('B','login',6),
('A','logout',7),
('B','logout',8),
('A','login',9),
('A','logout',10)

with tab as(
select name,event,time,lead(time)over(partition by name order by time) as timeout 
	from logins)
select name,time,timeout,timeout-time from tab where event='login';