Best of the Newsgroups: processes in D state, can't kill with -9

If this isn't exactly what you wanted, please try our Search (there's a LOT of techy and non-techy stuff here about Linux, Unix, Mac OS X and just computers in general!):

From: Tony Lawrence <tony@aplawrence.com>
Subject: Re: Something stronger then kill -9
Date: Tue, 26 Nov 2002 15:34:57 -0500
References: <c673867.0211210922.1aabea69@posting.google.com> <slrnatq9ru.phc.avenj@cerberus.localhost> <arjbhk$5eu$1@bob.news.rcn.net> <j5rlra.ucb.ln@freenet.co.uk> <arufj5$j3u$1@omega-3.right.here> <dukvra.m31.ln@freenet.co.uk> <3DE3C6E2.CEBA9653@sympatico.ca> 

John-Paul Stewart wrote:
 > spike1@freenet.co.uk wrote:
 >
 >>rnichols@interaccess.com wrote:
 >>
 >>>In article <j5rlra.ucb.ln@freenet.co.uk>,  <spike1@freenet.co.uk> wrote:
 >>>:Morden <newspeak@nowhere.com> wrote:
 >>>:> I think it's normal for a process to be non-killable when waiting for
 >>>:> some I/O. If I/O the process's waiting for never happens the 
process can
 >>>:> not be killed. Try to find out what the process is waiting for.
 >>>:> Maybe there's something in /proc that can be used to find that out?
 >>>:
 >>>:Shame the process can't timeout when in such a state... locked D 
processes
 >>>:should be *eventually* recoverable. No hardware wait should require 
more
 >>>:than half an hour.
 >>
 >>>Here's one contrary data point.  An "ERASE" command to a DDS-2 tape
 >>>drive takes roughly 3 hours to complete.  That's with the "long" bit
 >>>set, which is what is compiled into the kernel's "st" driver.
 >>
 >>That's the entire action though isn't it? The process isn't in a "D" 
state
 >>all that time. It has to be actively erasing the tape for most of it.
 >
 >
 > I don't think the process is actually "actively erasing the
 > tape".  It is my understanding that the application simply
 > sends an "erase" command of some sort to the tape and waits
 > for it to complete, with the hardware doing the work.
 >
 > I'm not the one who made the comment about erasing a DDS-2
 > tape, but I can say this:  on a Travan tape, a seek done
 > with 'mt fsf 1' remains in state "D" until the seek
 > completes.  (The application is doing nothing while the
 > hardware does the seek.)  With ~8GB in a single tar archive,
 > that could take _hours_ to get there on my Travan drive.
 >
 > Some hardware does indeed require more than a half hour wait
 > in "D" state.

Of course.  But it still would be nice to be able to easily remove
something from the proc table when a human being knows full well that
the hardware is never going to provide whatever magic bits the driver is
expecting.

In a better world, every driver would have an ioctl that would spit out
some standardized info about what it is up to and a way to tell it to
give up and return to initial state 'cause I know better than it does.
Of course this takes more work than just that - it would also have to
give back some false info to whatever processes were waiting for
whatever it is supposed to provide, but that could be a list of
possibilities too.

Imagine this in rather fanciful pseudo code:

Hey, tape driver?

- What?  I'm waiting for register such and such to tell me the tape is
ready.

Give it up, it's not going to happen.

- Well, fine, but I've got pid 134567 waiting for data.  I could return
an error status, but if you don't want that, you need to give me a 512
byte block of something to work with.

Just the error, thanks.  Return "Not ready" from now until I tell you
different, OK?  Don't even look at the hardware, it's whacked.

- Sure thing.



-- 

Please note new phone number: (781) 784-7547

Tony Lawrence
Unix/Linux Support Tips, How-To's, Tests and more: http://aplawrence.com
Free Unix/Linux Consultants list: http://aplawrence.com/consultants.html

Take Control of the Mac Command Line with Terminal

Take Control: The Mac OS X Lexicon

Take Control of Screen Sharing in Leopard

Take Control of Permissions in Leopard

Jump to Comments

Many of the products and books I review are things I purchased for my own use. Some were given to me specifically for the purpose of reviewing them. I resell or can earn commissions from the sale of some of these items. Links within these pages may be affiliate links that pay me for referring you to them. That's mostly insignificant amounts of money; whenever it is not I have made my relationship plain. I also may own stock in companies mentioned here. If you have any question, please do feel free to contact me.

Specific links that take you to pages that allow you to purchase the item I reviewed are very likely to pay me a commission. Many of the books I review were given to me by the publishers specifically for the purpose of writing a review. These gifts and referral fees do not affect my opinions; I often give bad reviews anyway.

We use Google third-party advertising companies to serve ads when you visit our website. These companies may use information (not including your name, address, email address, or telephone number) about your visits to this and other websites in order to provide advertisements about goods and services of interest to you. If you would like more information about this practice and to know your choices about not having this information used by these companies, click here.