I have written a replacement for shell standard shell. So when a user logs in they are presented with my custom shell app.

My problem is remote logins (i.e. a user that logs in via SSH). If they log in remotely and are using the custom shell, but the SSH process suddenly dies (for whatever reason) the custom shell app. doesn't end, it becomes an orphan (i.e. the init process becomes the parent).

Now, when I use a standard shell and the SSH process dies suddenly, the shell dies right away.

Can anybody explain this? I'd like the custom shell to die when it's parent process (usually SSH) dies. I'd prefer to do this without threads.

Thoughts?

Thanks.

Most likely it'll be in the difference between how the getty process spawns the login shell and how it spawns the login shell replacement. If it's not that then you might have coded something keeping it from dying properly. In both cases posting details should clear things up real quick.

You were right about: "you might have coded something keeping it from dying properly."
When I was building this shell app. I didn't want people killing it by sending it term/int/etc signals so I wrote a signal handler to ignore a bunch of signals. I went through the signals I was ignoring one by one and found the one that was keeping it alive: HUP. I thought it was odd that HUP was the one keeping it alive (I figured maybe SIGSTOP or something).

I no longer ignore HUP and it seems to be dying when the SSH session dies.

Thanks.

Moved: This thread is more suitable in <Programming> and has been moved accordingly to help your thread/question get the exposure it deserves.

As an FYI, SIGHUP stands for SIGnal Hang UP. It was used in the days of dial-in modems, etc., to indicate that the other end had "hung up".