🚀 go-pugleaf

dmm0t@rincewind.mech.Virginia.EDU (David Meyer) writes:

>In article <1992Nov19.191653.29086@schbbs.mot.com>  
>sdunbar@mailbox.fwrdc.rtsg.mot.com (Steven T. Dunbar) writes:
>> I'm trying to do some accounting of my log in time.
>> 
>> I wrote a login script like this:
>> #!/bin/csh
>> echo 'Login :' `/bin/date` >> ~/.worklog
>> 
>> and I wrote a logout script like this:
>> #!/bin/csh
>> echo 'Logout :' `/bin/date` >> ~/.worklog
>> 
>> I then did (as root):
>> 
>> dwrite loginwindow LoginHook /usr/local/bin/loginscript
>> dwrite loginwindow LogoutHook /usr/local/bin/logoutscript
>> 
>> I logged out of the console, used 'exit' to reload the 
>> loginwindow, logged in, and it doesn't work.

>loginwindow gets run by root, not by the user logging in.  Hence,
>loginscript and logoutscript also run as root, so you probably
>have /.worklog (since ~root == /) with the information you're
>looking for.

>I ran into this problem when I wanted to empty the trashcan on logout;
>I couldn't figure out how to communicate the login ID of the person
>exiting to the LogoutHook script.  FYI, I got around that by just
>emptying everybody's trashcan whenever anyone logs out.  It's an ugly
>hack.

The $1 argument to LoginHook is the userid (name) of the user who is
logging in.  I do not think that LogoutHook gets this information
passed to it as well (but would be happy to be proven wrong!  It
would make life simpler).  This is under 2.x, I haven't tried Hooks
in 3.0 yet.
-- 
David Lemson                                                (217) 244-1205
University of Illinois NeXT Campus Consultant / CCSO NeXT Lab System Admin
Internet : lemson@uiuc.edu                UUCP :...!uiucuxc!uiucux1!lemson 
NeXTMail accepted                                  BITNET : LEMSON@UIUCVMD

Argghhh.... root runs shellscripts!! DON'T DON'T DON'T! A shellscript
that is run by root *is* a threat to the security of your system and 
possibly of many other systems as well. Especially when it is supposed
to run inattendedly.

Besides, your programs look like a lot of redundancy to me. Every Login
and Logout gets logged (with time, connecting host etc.) in
/usr/adm/wtmp. There is also information about shutdowns, reboots and
crashes in that file. 

You get at least two unix programs to evaluate that file: last, ac. For
details let Librarian.app show you the according man pages.

You can also search the archives (with archie) for a little piece of C
code called 'rdwtmp.c' which gives you yet another view of the login
data in /usr/adm/wtmp.

Cheers
- A.
-- 
	 lytras@ifi.unizh.ch   		|   Apostolos Lytras
	 lytras@avalon.physik.unizh.ch	|   Informatik Club der Uni
	 lytras@amiga.physik.unizh.ch  	|   Zuerich, SysAdmin

In an article, lemson@ux1.cso.uiuc.edu (David Lemson) writes:
> 
> The $1 argument to LoginHook is the userid (name) of the user who is
> logging in.  I do not think that LogoutHook gets this information
> passed to it as well (but would be happy to be proven wrong!  It
> would make life simpler).  This is under 2.x, I haven't tried Hooks
> in 3.0 yet.

In 2.2, LogoutHook does get the user name passed as $1 (argv[1]).  Further,
the LogoutHook _is_ run as the user...if you want to use it to write
a privileged file, the program will have to be setuid.  Another
poster says that these hooks are not needed since logins are automatically
put in /usr/adm/wtmp.  This is NOT true for console logins (at least in
2.2), so you do need to write your own hooks to handle it.  As stated, 
though, shell scripts are probably not the best idea (but that argument is 
usually directed against _setuid_ shell scripts, not just ordinary shell 
scripts - there are tons of ordinary shell scripts routinely run unattended 
by root.)

I can post the tiny LoginHook and LogoutHook C programs I use to handle
wtmp if anyone cares...

Wade Blomgren
wade@hobbes.ucsd.edu