Re: [REPOST][REPOST][REPOST] Killing processes in D state

Bill Davidsen (davidsen@tmr.com)
Fri, 6 Jun 2003 12:36:15 -0400 (EDT)


On Mon, 2 Jun 2003, Roy Sigurd Karlsbakk wrote:

> this discussion has been up a few times, but I want it again. I _REALLY_ want
> to be able to kill processes in D state. I am aware of that this is not good,
> but compared to other parts of the linux kernel, there's quite a lot suicidal
> stuff there already, so why not. There is, for instance, in 2.5, the
> possibility to forcably remove loaded modules - FAR worse than merely killing
> a userspace process in D state.
>
> So please, dear kernel people, Free Thy Users From Those Terrible Unreasonable
> Reboots.

Do you understand why processes in D state are unkillable? What do you
think you would gain by killing them, since the memory couldn't be safely
reused? Just not seeing the process in ps?

I can see wanting to avoid having processes in D state, but once you get
such a thing I don't see why the kernel should have a lot of code added to
hide a problem. A reboot *is* a way to kill a process in D state.

In general most of the process should page out, so you only lose some swap
space. I don't get processes in this state, so I can't really check the
RSS and verify that.

Tell the developers how you get the processes in that state and ask for
that to be prevented, and you will find a lot more support.

-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/