DeadHead@cup.portal.com (Bruce M Ong) (06/01/90)
I am sure there is a very easy way to do this, but I just havent figured this one out: How can a parent find out if one of its children is still alive or not without 1) going through fifos, 2) going through messages 3) going through shared memory 4) going through sockets I thought the parent could use the getpgrp() call to find out the child's group id and in turn find out if the child is still alive or not. But that didnt seem to work. I dont want the parent to wait for the child to die, either... Please e-mail, thanks! deadhead@cup.portal.com
toma@ozdaltx.UUCP (Tom Armistead) (06/02/90)
In article <30408@cup.portal.com>, DeadHead@cup.portal.com (Bruce M Ong) writes: > I am sure there is a very easy way to do this, but I just havent > figured this one out: > > How can a parent find out if one of its children is still alive > or not without ..... If you have the process id of the child you can use kill(2) to find out if it is still there. i.e.: if( kill( child_pid, 0 ) == -1 ) puts( "The child is dead" ); Signal 0 is called the NULL signal and is used to validate a process id, no signal is actually sent to the destination process. Will this do it for you? Tom -- ------------------------------- {uunet,smu}!sulaco!ozdaltx!toma (Tom Armistead @ Garland, Texas) {uunet,smu}!sulaco!ozdaltx!swsrv1!toma
jrl@cbnews.att.com (john.lupien) (06/06/90)
In article <6334@ozdaltx.UUCP> toma@ozdaltx.UUCP (Tom Armistead) writes: >In article <30408@cup.portal.com>, DeadHead@cup.portal.com (Bruce M Ong) writes: >> I am sure there is a very easy way to do this, but I just havent >> figured this one out: >> How can a parent find out if one of its children is still alive >> or not without >If you have the process id of the child you can use kill(2) to find out if >it is still there. >i.e.: if( kill( child_pid, 0 ) == -1 ) > puts( "The child is dead" ); >Signal 0 is called the NULL signal and is used to validate a process id, no >signal is actually sent to the destination process. This is nice in that, with a small change, you can use it to find out if processes belonging to other uid's are alive, provided you know their pid. If kill(child_pid, 0) returns other than -1, the process exists and you have permission to kill (007?) If it returns -1 and an errno of ESRCH, the pid does not exist, otherwise you typically get errno of EPERM. -- -John Lupien mvuxr!jrl jrl@mvuxr.att.com
matt@group-w.uchicago.edu (Matt Crawford) (06/06/90)
) DeadHead@cup.portal.com (Bruce M Ong) writes:
) > How can a parent find out if one of its children is still alive
) > or not without
Tom Armistead, wizard-wannabe, answers:
) i.e.: if( kill( child_pid, 0 ) == -1 )
) puts( "The child is dead" );
overlooking the possibility that the child exited many minutes ago and
the pid is now being used by some unrelated process.
________________________________________________________
Matt Crawford matt@oddjob.uchicago.edusmb@ulysses.att.com (Steven Bellovin) (06/07/90)
In article <1046@gargoyle.uchicago.edu>, matt@group-w.uchicago.edu (Matt Crawford) writes: } ) DeadHead@cup.portal.com (Bruce M Ong) writes: } ) > How can a parent find out if one of its children is still alive } ) > or not without } } Tom Armistead, wizard-wannabe, answers: } ) i.e.: if( kill( child_pid, 0 ) == -1 ) } ) puts( "The child is dead" ); } } overlooking the possibility that the child exited many minutes ago and } the pid is now being used by some unrelated process. You can look at errno to determine if the kill() failed because of a uid mismatch. That will yield a lot of info if root isn't doing the checking.
meissner@osf.org (Michael Meissner) (06/07/90)
In article <13071@ulysses.att.com> smb@ulysses.att.com (Steven Bellovin) writes: | Path: paperboy!think!snorkelwacker!usc!ucsd!ucbvax!ulysses!ulysses.att.com!smb | From: smb@ulysses.att.com (Steven Bellovin) | Newsgroups: comp.unix.wizards | Date: 6 Jun 90 19:22:28 GMT | References: <30408@cup.portal.com> <6334@ozdaltx.UUCP> <1046@gargoyle.uchicago.edu> | Sender: netnews@ulysses.att.com | Distribution: na | Lines: 15 | | In article <1046@gargoyle.uchicago.edu>, matt@group-w.uchicago.edu (Matt Crawford) writes: | } ) DeadHead@cup.portal.com (Bruce M Ong) writes: | } ) > How can a parent find out if one of its children is still alive | } ) > or not without | } | } Tom Armistead, wizard-wannabe, answers: | } ) i.e.: if( kill( child_pid, 0 ) == -1 ) | } ) puts( "The child is dead" ); | } | } overlooking the possibility that the child exited many minutes ago and | } the pid is now being used by some unrelated process. | | You can look at errno to determine if the kill() failed because of | a uid mismatch. That will yield a lot of info if root isn't doing | the checking. Unless of course, the newly created pid is running one of your cron jobs, a different job created by your job control shell, on a different window owned by you, or somebody running nethack which is setuid to your account. Kill only checks for the effective UID being the same. The above methods are ways there can be a process with the same PID as a dearly departed child process, and have the same EUID as the parent proces, but not be the process the parent wants to deal with. I'm sure I've missed a few ways to create such processes. -- Michael Meissner email: meissner@osf.org phone: 617-621-8861 Open Software Foundation, 11 Cambridge Center, Cambridge, MA Catproof is an oxymoron, Childproof is nearly so
tcl@cs.brown.edu (Tom Lawrence) (06/07/90)
In article <MEISSNER.90Jun6191114@curley.osf.org> meissner@osf.org (Michael Meissner) writes: >Unless of course, the newly created pid is running one of your cron >jobs, a different job created by your job control shell, on a >different window owned by you, or somebody running nethack which is >setuid to your account. Kill only checks for the effective UID being >the same. > >The above methods are ways there can be a process with the same PID as >a dearly departed child process, and have the same EUID as the parent >proces, but not be the process the parent wants to deal with. I'm >sure I've missed a few ways to create such processes. I'd like to point out that pid numbers aren't recycled for a while. It may take quite a long time on a machine which is relatively unused. I think the best way to reliably check the status of a child process is to catch the SIGCHLD signal, which is delivered when the status of a child process changes. (this signal is normally ignored). The handler routine can either explicitely do whatever you want to do when the child dies, or set some global flag which the main thread will later check. * Tom Lawrence A is for apathy * * tcl@cs.brown.edu B is for boredom * * C is for... oh the hell with it *
friedl@mtndew.UUCP (Stephen J. Friedl) (06/07/90)
In article <42180@brunix.UUCP>, tcl@cs.brown.edu (Tom Lawrence) writes: > I'd like to point out that pid numbers aren't recycled for a while. It > may take quite a long time on a machine which is relatively unused. Be careful with this one -- the 3B15 running Sys V Rel 3 recycles PIDs very quickly (certainly not in a circular manner) and it can surprise you -- it surprised me. Other machines might recycle soon too. Steve -- Stephen J. Friedl, KA8CMY / Software Consultant / Tustin, CA / 3B2-kind-of-guy +1 714 544 6561 / friedl@mtndew.Tustin.CA.US / {uunet,attmail}!mtndew!friedl "I will defend to your death my right to my opinion" - me