forcible kill?

paulwvanc@yahoo.ca

hi

I have one process that I can't kill with "kill"
it's locked up my /dev/dsp.
Is there a "stronger" "kill"?

Todd H.

paulwvanc@yahoo.ca writes:

> hi
>
> I have one process that I can't kill with "kill"
> it's locked up my /dev/dsp.
> Is there a "stronger" "kill"?

kill -9 <pid>



--
Todd H.
http://www.toddh.net/

Michel Bardiaux

On 13 Dec 2006 02:43:40 -0600, comphelp@toddh.net (Todd H.) wrote:

>paulwvanc@yahoo.ca writes:
>
>> hi
>>
>> I have one process that I can't kill with "kill"
>> it's locked up my /dev/dsp.
>> Is there a "stronger" "kill"?
>
>kill -9 <pid>

I began using Unix in 1986, and even then it was a FAQ: how to
'kill-I-mean-it'. 20 years later, we still dont have that: kill -9
will kill the process but may leave a zombie if the parent is hung,
and some resources may not be released until the zombie has been
cleared. Which may force you to kill the parent. Which may not be
good!

Bill Marcum

On Wed, 13 Dec 2006 12:31:53 +0100, Michel Bardiaux
<mbardiaux@mediaxim.be> wrote:
> On 13 Dec 2006 02:43:40 -0600, comphelp@toddh.net (Todd H.) wrote:
>
>>paulwvanc@yahoo.ca writes:
>>
>>> hi
>>>
>>> I have one process that I can't kill with "kill"
>>> it's locked up my /dev/dsp.
>>> Is there a "stronger" "kill"?
>>
>>kill -9 <pid>
>
> I began using Unix in 1986, and even then it was a FAQ: how to
> 'kill-I-mean-it'. 20 years later, we still dont have that: kill -9
> will kill the process but may leave a zombie if the parent is hung,
> and some resources may not be released until the zombie has been
> cleared. Which may force you to kill the parent. Which may not be
> good!

The advice that I have often read is: if kill -9 is necessary, then
remove the defective binary file.


--
Brandy-and-water spoils two good things.
-- Charles Lamb

Andrew Smallshaw

On 2006-12-13, Michel Bardiaux <mbardiaux@mediaxim.be> wrote:
>
> I began using Unix in 1986, and even then it was a FAQ: how to
> 'kill-I-mean-it'. 20 years later, we still dont have that: kill -9
> will kill the process but may leave a zombie if the parent is hung,
> and some resources may not be released until the zombie has been
> cleared. Which may force you to kill the parent. Which may not be
> good!

Having zombies left lying around isn't really so much of a problem
- leave them there and they won't do much harm. When a process
goes zombie all memory and open files that it was using are released
immediately. The only thing left is an entry in the process table
holding the process's exit code.

--
Andrew Smallshaw
andrews@sdf.lonestar.org

Barry Margolin

In article <j0pvn25klp62pf9470jd864s1hbilaos9b@4ax.com>,
Michel Bardiaux <mbardiaux@mediaxim.be> wrote:

> On 13 Dec 2006 02:43:40 -0600, comphelp@toddh.net (Todd H.) wrote:
>
> >paulwvanc@yahoo.ca writes:
> >
> >> hi
> >>
> >> I have one process that I can't kill with "kill"
> >> it's locked up my /dev/dsp.
> >> Is there a "stronger" "kill"?
> >
> >kill -9 <pid>
>
> I began using Unix in 1986, and even then it was a FAQ: how to
> 'kill-I-mean-it'. 20 years later, we still dont have that: kill -9
> will kill the process but may leave a zombie if the parent is hung,
> and some resources may not be released until the zombie has been
> cleared. Which may force you to kill the parent. Which may not be
> good!

Zombies never consume any resources except a process table slot that
holds the termination status. A process doesn't become a zombie until
all its resources have been released and the only thing it's hanging
around for is its parent to call one of the wait() functionx.

--
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***
*** PLEASE don't copy me on replies, I'll read them in the group ***

Michael Tosch

Bill Marcum wrote:
> On Wed, 13 Dec 2006 12:31:53 +0100, Michel Bardiaux
> <mbardiaux@mediaxim.be> wrote:
>> On 13 Dec 2006 02:43:40 -0600, comphelp@toddh.net (Todd H.) wrote:
>>
>>> paulwvanc@yahoo.ca writes:
>>>
>>>> hi
>>>>
>>>> I have one process that I can't kill with "kill"
>>>> it's locked up my /dev/dsp.
>>>> Is there a "stronger" "kill"?
>>> kill -9 <pid>
>> I began using Unix in 1986, and even then it was a FAQ: how to
>> 'kill-I-mean-it'. 20 years later, we still dont have that: kill -9
>> will kill the process but may leave a zombie if the parent is hung,
>> and some resources may not be released until the zombie has been
>> cleared. Which may force you to kill the parent. Which may not be
>> good!
>
> The advice that I have often read is: if kill -9 is necessary, then
> remove the defective binary file.
>

This is certainly a good advice :-)

But the program can hang because of another reason: a hung I/O device.
In this case a "kill -9" works if the kernel device driver supports
an interrupt, otherwise even "kill -9" won't work.

The culprit then is either a faulty kernel device driver or the
"hardware" (which can be virtual i.e. represented by another
software layer).

--
Michael Tosch @ hp : com