[Bug libc/19144] daemon() fails to prevent reacquisition of controlling terminal

["mtk.manpages at gmail dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=19144

--- Comment #2 from Michael Kerrisk <mtk.manpages at gmail dot com> ---
Hi FLorian,

(In reply to Florian Weimer from comment #1)
> How risky is adding the second fork?  Would it otherwise change behavior?

Well:
* It would fix the bug that allows the caller to reacquire a controlling
terminal. It seems unlikely that any program would depend on that bug in order
to acquire a controlling tty.

* In a properly constructed System-V type daemon, the resulting process is
neither a session leader (which is the critical point) nor a process group
leader. This means (getpid() != getsid() && getpid() != getpgrp()).  But with
the baemon(3) in glibc, the following is true for the resulting process:
(getpid() == getsid() && getpid() == getpgrp()). It seems rather unlikely that
any program would depend on this fact though.

> By the way, the manual daemon(3) manual page talks about the âcalling
> process's current working directoryâ.  I think this is misleading because
> the function exits the calling process before changing the current directory.

I changed "calling process's" to "processes".

Cheers,

Michael

-- 
You are receiving this mail because:
You are on the CC list for the bug.