Can't kill hung remote CP copy

[Andy Blanchard]

On 14 April 2012 20:06, Joe Zeff <joe at zeff.us> wrote:

> I'd think that there'd have to be a special command, requiring root, to do
> it for exactly that reason.  There's no way you can safely automate that
> decision, and it's probably best if the average user doesn't have direct
> access to it; if nothing else, having to use either su or sudo along with a
> password (and possibly a confirmation dialog) would remind people that this
> isn't something you do lightly.  How it gets done, alas, is beyond me
> because I don't know enough about such things.


I'm not an expert at this low a level of the kernel either, but I think
that it would, in theory, be possible for the kernel to forcibly unload a
program, then try to cleanly close any open file handles and similar "tidy
up" operations.

That still leaves the problem of dealing the actual device which caused the
lockup in the first place, which is now in an unknown state. Depending on
what it is you might want to "offline" it until the next reboot, "reset" it
(rmmod, insmod), or even perform an immediate shutdown of the system.  In
many cases the kernel just isn't going to know the correct course of
action, especially if the device is remote or a more unusual peripheral.

Could get messy, and who is the user going to blame for *their* incorrect
call?  I'd assume that the current status quo is the safest bet with
regards to potential data loss.

-- 
Andy

*The only person to have all his work done by Friday was Robinson Crusoe*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.fedoraproject.org/pipermail/users/attachments/20120414/ad6a5a44/attachment.html>