Newsgroups: comp.unix.questions Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!think.com!mintaka!bloom-picayune.mit.edu!athena.mit.edu!jik From: jik@athena.mit.edu (Jonathan I. Kamens) Subject: Re: "Unkillable" processes In-Reply-To: oli@odbffm.incom.de's message of Sat, 20 Apr 91 14:56:33 GMT Message-ID: Sender: news@athena.mit.edu (News system) Organization: Massachusetts Institute of Technology References: <1991Apr20.145633.5639@odbffm.incom.de> Date: Mon, 22 Apr 91 06:38:39 GMT Lines: 30 In article <1991Apr20.145633.5639@odbffm.incom.de> oli@odbffm.incom.de (Oliver Boehmer) writes: A kill -9 xxx only has the effect, that the PPID changes to init, and after this any kill -9 is ignored. The problem is that the process is hung inside a device driver inside the kernel, ignoring signals. This usually indicates a somewhat badly written device driver -- the author of the device driver decided that some condition was urgent enough and would go by quickly enough that the device driver could hang the process while waiting for it to complete, and the author was wrong. :-) Why? How can I get rid of these processes w/o shutting down the system? Well, check to see if the device (i.e. the hardware device -- tape drive or whatever) the process is accessing is offline or something -- that may be what the device driver is waiting on. If it isn't, try taking it offline and putting it back online :-). If that doesn't work, then you can adb the running kernel (if you have adb) and fiddle with the process to get the device driver to relinquish its hold on the process. Unless you are very good at mucking with running kernels, this is probably a bad idea. If fiddling with the device doesn't work, and you can't/won't fiddle with the kernel, then rebooting is usually your only alternative. Jonathan Kamens USnail: MIT Project Athena 11 Ashford Terrace jik@Athena.MIT.EDU Allston, MA 02134 Office: 617-253-8085 Home: 617-782-0710