On Mon, 2003-09-22 at 09:44, Andreas Winkelmann wrote:

Tom Nielsen wrote:

...

I've got this process that is defunct. I ran kill 7690 as root and nothing happens. Tried kill -l 7690, nothing. Here's my line:

tom-linux:/home/tom # ps aux | grep " Z " tom 7690 0.0 0.0 0 0 ? Z 08:59 0:00 [netstat] <defunct> root 7835 0.0 0.1 3544 572 pts/1 S 09:34 0:00 grep Z tom-linux:/home/tom # kill -l 7690 bash: kill: 7690: invalid signal specification

I'm sure I'm doing something wrong.

# kill -9 pid

is the hardest way. If that doesn't work you have to wait or reboot.

-- Andreas

Looks like I'll have to reboot. -9 didn't help. Thanks though. -- Tom Nielsen Neuro Logic Systems 805.389.5435 x18 www.neuro-logic.com

Alle 18:51, luned=EC 22 settembre 2003, js ha scritto:

...

You can't. It will remain until you reboot. They are essentially harmless until you've got thousands of them. It's just something you have to live with.

John S.

Tom Nielsen wrote:

...

I've got this process that is defunct. I ran kill 7690 as root and nothing happens. Tried kill -l 7690, nothing. Here's my line:

tom-linux:/home/tom # ps aux | grep " Z " tom 7690 0.0 0.0 0 0 ? Z 08:59 0:00 [netstat] <defunct> root 7835 0.0 0.1 3544 572 pts/1 S 09:34 0:00 grep Z tom-linux:/home/tom # kill -l 7690 bash: kill: 7690: invalid signal specification

I'm sure I'm doing something wrong.

Thanks, Tom

Killing them does not lead to success. Kill their parent until init inherit= =20 them and they will go to the deserved rest.

Praise

Just don't kill the parent if the parent is INIT. That would be bad. :-)

* Jim Norton <jrn@oregonhanggliding.com> [Sep 23. 2003 00:16]:

Just don't kill the parent if the parent is INIT. That would be bad. :-)

... and fortunately it's not possible to kill init :) -- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogical, with just a little bit more effort?" -- A. P. J.

* Jim Norton <jrn@oregonhanggliding.com> [Sep 23. 2003 01:54]:

...

...

Just don't kill the parent if the parent is INIT. That would be bad. :-)

... and fortunately it's not possible to kill init :)

So the kernel doesn't allow INIT to be killed by anything not of it's doing to prevent the system from being stopped if INIT was killed?

Well, the kill(2) syscall is just a program to send a signal to a process. So it's up to the program to implement proper signal handling. In the case of process 1 (init), there's the following note in kill(2): "It is impossible to send a signal to task number one, the init process, for which it has not installed a signal handler. This is done to assure the system is not brought down accidentally."

Does the kernel ever kill init itself ( for a reboot for example )? Or does it just enter a runlevel and sit and wait for the PC to reboot etc?

Well, what do you mean "if the kernel kills init"? At some point init tell the kernel it wants to reboot/shutdown, simply by doing all the preparation (nice shutdown of all processes etc.) and then when time permits, calls reboot(); It's good reading to take a stroll through the source-code for init and friends, if one wants to get acquainted with this stuff. -- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogical, with just a little bit more effort?" -- A. P. J.

* Alin Dobre <alin.dobre@ravantivirus.com> [Sep 23. 2003 08:23]:

Hi,

It works killall -1 init restarts the init process (for example the inittab file is re-read). (HUP) signal works, so "for which it has not installed a signal handler" is not true.

There's a signal handler for signal 1, which is HUP, yes. There's even for more, if you look in the init source code: /* * We got a signal (HUP PWR WINCH ALRM INT) */ static void signal_handler(int sig) { ADDSET(got_signals, sig); } What I quoted from the manual page was this: "It is impossible to send a signal to task number one, the init process, for which it has not installed a signal han­ dler. This is done to assure the system is not brought down accidentally." If you application does not install a signalhandler for SIGKILL, that does not mean your application is not killed, if we send kill -9 to it. In the case of init, nothing happens. -- Mads Martin Joergensen, http://mmj.dk "Why make things difficult, when it is possible to make them cryptic and totally illogical, with just a little bit more effort?" -- A. P. J.