What to do about nologin(8)?

[Tim Kientzle]

John Baldwin wrote:
> 
> My point (sigh) is that doing system("logger") has the same problem set as 
> making nologin dynamic ...

No, it doesn't.  Not if you make nologin static and
have it create a fresh environment before running
any external programs.  This would also be considerably
more compact than statically linking in the logging functions.

> Also, personally, I would rather have nologin be static than fix the one 
> known case of login -p and just hope no other cases pop up in the future.  
> Call me paranoid. :)

Armoring nologin(8) is insufficient.

In particular, as David Schultz pointed out, there are a lot
of home-grown nologin scripts out there that are potentially
vulnerable regardless of what we do with the "official"
nologin program.

Tim Kientzle