httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chuck Murcko <ch...@telebase.com>
Subject Re: 1.0.2++
Date Sun, 04 Feb 1996 18:55:49 GMT
Rob Hartill liltingly intones:
> 
>  
> > Shouldn't (and doesn't) matter; int * is the proper way. This is the
> > culprit, though. Have I just discovered that IIRIX and BSDI are non-
> > POSIX conformant signal behavior? Or should the kill_never be just_wait?
> > Doing that reaps 'em fine in the waitpid() a few lines below this.
> 
> Why not see what the patch authro has to say. In the meantime it
> should be rejected for the next release.
> 
OK, I'll send out some mail, then. The combination of my (borrowed)
POSIX book and the man pages tells me this:

The premise of the patch is spot on.

Return code of 0 from waitpid() means child process status is not available
yet.

These are the children that I suggest should be marked just_wait in the
first waitpid() loop, so they can be reaped later. kill_never doesn't allow
cleanup after server push scripts, for example. This is the cause of the
zombies I was seeing. I'd wager that any child marked kill_never while
still runnning will end up zombified when it terminates normally.

I'll get this resolved before vote time.

chuck
Chuck Murcko	Telebase Systems, Inc.	Wayne PA	chuck@telebase.com
And now, on a lighter note:
An authority is a person who can tell you more about something than you
really care to know.

Mime
View raw message