[solved] dwmrc script keeps running

[psychedup]

I've recently been getting sucked into the world of dwm and other suckless.org apps. Having a little issue with my custom .dwmrc script though, probably a dumb mistake but hopefully somebody here can help.

I'm using lightdm to log in, and it seems to call a script in /etc/X11/Sessions/dwm:

[The Main Man]

Why dont' you kill a script before logging out ?

[AlexJGreen]

_

[psychedup]

Thanks, your suggestions helped. i also looked at this blog post.

I tried using a trap but it seems that dwm doesn't send the script any signals when it quits, the trap never gets triggered. So I realized (as in your suggestion), the script itself is going to have to monitor and shut down itself and its child processes.

From that blog I learned that when my script is run, the PID of the process that called it, gets assigned as its PPID. But when that parent process quits, my process's PPID changes to 1. So I just have to monitor the PPID and if it's 1, kill all of my child processes and myself.

So I added this loop to the end of my dwmrc script and it seems to be working:

[Hu]

That is a bad way to detect that your process has become an orphan. Instead, you can consult $PPID, a magic variable that obtains the parent process ID easily.

Second, you could avoid the problem by instructing the kernel to terminate the children automatically. For each child, as you create it, set its parent-death-signal to SIGTERM. When the parent terminates, the kernel will automatically SIGTERM the children. To do this, you can use setpriv --pdeathsig SIGTERM command & instead of a bare command &. Once this is done, just exit your script when you are an orphan, and your children will be terminated automatically. You could even be clever and reexecute yourself with a pdeathsig so that the death of your script's parent kills your script.