Erik Trauschke <erik.trausc...@freenet.de> wrote: > I'm currently tinkering with sigtimedwait() to timeout a process I > forked by setting it up to listen for SIGCHLD. > > What strikes me as odd is that I have to define a bogus signal-handler > routine to make it work, otherwise no SIGCHLD is even received by the > parent process. > > I attached a simplified example to this mail. All it does is setting up > a signalset for SIGCHLD, fork a process and wait with sigtimedwait() for > the child to exit. Which should happen immediately. However, when > sigaction() with the bogus handler is not called, sigtimedwait() times > out because it receives no SIGCHLD (truss also shows no SIGCHLD raised) > > So is there anything wrong with the approach or am I just too stupid? > > Any hints or comments would be appreciated
Hi Erik. You are doing everything right. Solaris is doing the wrong thing in this case. I filed this bug report to address the problem: 6857047 sigwait malfunctions for system-generated SIGCHLD signals I expect to have it fixed soon, build snv_119 (or snv_120 at worst). In the meantime, setting up the bogus handler is a good workaround. Thanks for the test case. Roger Faulkner Sun Microsystems _______________________________________________ opensolaris-code mailing list opensolaris-code@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
