On Wed, Jul 16, 2014 at 01:29:14PM +0200, Baptiste Daroussin wrote: > On Wed, Jul 16, 2014 at 12:45:12PM +0200, Mateusz Guzik wrote: > > On Wed, Jul 16, 2014 at 12:34:11PM +0200, Baptiste Daroussin wrote: > > > On Wed, Jul 16, 2014 at 12:29:08PM +0200, Mateusz Guzik wrote: > > > > On Wed, Jul 16, 2014 at 09:55:36AM +0000, Baptiste Daroussin wrote: > > > > > +static void > > > > > +usage(void) > > > > > +{ > > > > > + fprintf(stderr, "Usage: %s [--signal sig | -s sig] > > > > > [--preserve-status]" > > > > > + " [--kill-after time | -k time] [--foreground] <duration> > > > > > <command>" > > > > > + " <arg ...>\n", getprogname()); > > > > > + > > > > > > > > Missing newline at the begnning. > > > New line at the beginning? > > > > > > > > static void > > usage(void) > > { > > > > fprintf(...); > > Fixed > > > > > > > + exit(EX_USAGE); > > > > > +} > > > > > + > > > > > + switch(signo) { > > > > > + case 0: > > > > > > > > sig 0? I doubt it is ever delivered. > > > > > > one of the tests from the GNU testsuite is passing 0 signal > > > > I even checked. The kernel is not going to deliver signal 0, so checking > > for it in signal handler does not seem to make sense. > > > > Does the testcaes fail without it or something? > > > > Yes the test fails without it
Which test ? Kernel never delivers signal 0. There were some bugs in libthr which caused some signals to be reported as signal 0, they are supposedly fixed. The kill(0, pid) is defined as testing for the pid existence, but the target process is not interrupted.
pgpwD1uuuFkTI.pgp
Description: PGP signature