Date: Fri, 1 Nov 1996 22:47:24 +0100 (MET) From: J Wunsch <j@uriah.heep.sax.de> To: freebsd-hackers@freebsd.org (FreeBSD hackers) Subject: Re: Zombie processes Message-ID: <199611012147.WAA23522@uriah.heep.sax.de> In-Reply-To: <199611011827.LAA28085@phaeton.artisoft.com> from Terry Lambert at "Nov 1, 96 11:27:06 am"
next in thread | previous in thread | raw e-mail | index | archive | help
As Terry Lambert wrote: > > I call this intent of confusion, at least. It's not helpful to the > > one who's been asking the question in the first place, either. > > What about the innocent developer, whose code works on all other > platforms? He was too innocent then. AFAIK, this never worked on any BSD system before. (SA_NOCLDWAIT vs. SIG_IGN) > Well, that's an implementation detail... it only bears on the amount > of work required to do it, not whether or not it should be done. Not really. I've already told you that the real work is not required for implementing the SA_ flag, but rather for releasing the old process' address space completely. It's much more convenient to do this from the context of another process. It turns out to be the best solution to reparent the zombie to init, and send init the SIGCHLD. This sounds hacky, but all other ideas i've tried were even more icky. > We have a lot of non-POSIX historical behaviours. This is another > (though POSIX systems implement it as well) which is worth emulating. Nothing to `emulate', but something to `implement'. -- cheers, J"org joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE Never trust an operating system you don't have sources for. ;-)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199611012147.WAA23522>