killing processes in an uninterruptible sleep

killing processes in an uninterruptible sleep

Post by stack.. » Mon, 26 Aug 2002 23:46:02



Hi NG!
Can someone give me a good reason for the fact that SIGKILL (supposed
to instantly kill an app as it can't be catched) can't kill a process
listed with status D in `ps ax` that is, an IO sleep? Why having
developed that signal if it can't kill everything it comes across?
Related question: Is it possible to force init to read the return code
of a zombie even if its parent process is still there but is stuck
e.g. in an IO sleep?
Thanks!
Bernd Haller.
 
 
 

killing processes in an uninterruptible sleep

Post by mjt » Tue, 27 Aug 2002 01:27:49


0 spat:

Quote:> Can someone give me a good reason for the fact that SIGKILL (supposed
> to instantly kill an app as it can't be catched) can't kill a process
> listed with status D in `ps ax` that is, an IO sleep? Why having
> developed that signal if it can't kill everything it comes across?

because that process is in a system call - it'll be killed once
the call returns.

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Michael J. Tobler: motorcyclist, surfer,  #    Black holes result
 skydiver, and author: "Inside Linux",     #   when God divides the
 "C++ HowTo", "C++ Unleashed"              #     universe by zero

 
 
 

killing processes in an uninterruptible sleep

Post by stack.. » Wed, 28 Aug 2002 05:26:25


Quote:> > Can someone give me a good reason for the fact that SIGKILL (supposed
> > to instantly kill an app as it can't be catched) can't kill a process
> > listed with status D in `ps ax` that is, an IO sleep? Why having
> > developed that signal if it can't kill everything it comes across?

> because that process is in a system call - it'll be killed once
> the call returns.

yes but if the system call is blocking or has a very long timeout,
there must be some sort of 'emergency break' to tell the kernel that
this or that process has gone mad. Just imagine a beta software or a
debugging session. The user should always have the last word, we're
not running windows here ;)
Bernd Haller
 
 
 

killing processes in an uninterruptible sleep

Post by mjt » Wed, 28 Aug 2002 10:02:34


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
NotDashEscaped: You need GnuPG to verify this message


>> > Can someone give me a good reason for the fact that SIGKILL (supposed
>> > to instantly kill an app as it can't be catched) can't kill a process
>> > listed with status D in `ps ax` that is, an IO sleep? Why having
>> > developed that signal if it can't kill everything it comes across?
>> because that process is in a system call - it'll be killed once
>> the call returns.

> yes but if the system call is blocking or has a very long timeout,
> there must be some sort of 'emergency break' to tell the kernel that
> this or that process has gone mad. Just imagine a beta software or a
> debugging session. The user should always have the last word, we're
> not running windows here ;)

... /usr/src/linux/...   :)

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Michael J. Tobler: motorcyclist, surfer,  #    Black holes result
 skydiver, and author: "Inside Linux",     #   when God divides the  
 "C++ HowTo", "C++ Unleashed"              #     universe by zero

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE9as9htTveLPAHcDIRAtSDAJ9w2A8g166Eva8f6dSGCpdAugg+wwCeKvcM
Qd1610VWfLKNcRefN2wJMSg=
=pBjq
-----END PGP SIGNATURE-----

 
 
 

1. Killing a process in "uninterruptible sleep"

A segmentation fault occured while trying to mount a Mac (HFS) formatted Zip
disk on a PC running linux 2.0.33. Even though it did not successfully mount,
I tried to umount it before removing the disk. The umount process hung and ps
shows the status of the process is "uninterruptible sleep".  

Now I can't do a soft reboot because reboot seems to want all processes killed
and even kill -9 as root will not kill the umount process. How can I recover
from this situation so that I can reboot gracefully? Or better yet, how do I
return to a normal situation where I can eject the disk? As it stands there is
no way to manually remove the Zip disk and the software method eject does not
seem to work either. (This is why I am trying to reboot) If I am forced to
power-down abruptly I will probably have to run fsck manually --- something
I'd prefer not having to do.

Any suggestions?

Thanks,
Alan

2. HOW ? do i insert spaces on each line of a file

3. How to kill a uninterruptible sleep process

4. growable disk partitions

5. How to "kill" uninterruptible sleeping process?

6. **** HELP where can I get these files?? ****

7. killing uninterruptible sleep PID

8. libqt.so.1 error in LinuxPPC 2000

9. Uninterruptible sleeping process

10. Uninterruptible sleep Process

11. Process in "uninterruptible sleep"?

12. Process falls into uninterruptible sleep

13. Uninterruptible sleep Process and Reboot