@ -1,3 +1,10 @@
Changes in wait4 0.3.0
* Use Linux Socket Filtering to filter events in the kernel.
* Add (theoretical) support for BSD kevent(2).
* Fix Netlink initialization (thanks to "neologix", linuxfr.org)
Changes in wait4 0.2.1
* Add the -t, --timeout option.
@ -21,6 +21,14 @@ privileged process can bind to an AF_NETLINK address; root privileges
are dropped immediately after the bind(2) call.
BSD kevent support
Starting from version 0.3.0, wait4 should be able to make use of the
kevent(2) call found on BSD systems, thus avoiding the signal-based
fallback method on those systems. Note, however, that this support
is currently theoretical only, as it has not been tested yet.
wait4 is distributed under the terms of the GNU General Public License,
@ -34,13 +34,24 @@ Aborts waiting after the specified
The ability to report the termination status of the waited
process depends on
.BR netlink (7).
On systems without netlink, a fallback
depends on the
.BR netlink (7)
interface or the
.BR kevent (2)
system call to fulfill its function properly. On systems
which do not have any of these, a fallback
.BR kill (2)-based
mechanism is used, which does not allow to retrieve the
termination status; the exit status is then always 0.
mechanism is used, which suffers from the following
limitations: it does not allow to retrieve the termination
status, and thus
will always return 0; and it can miss the termination of a
process, if its PID gets reused too quickly by a new process.
.BR netlink (7),
the number of processes to wait for cannot be greater than 251.
.SH REPORTING BUGS