Re: atomicity

jens@pinguin.conetix.de
Sun, 6 Dec 1998 16:59:21 +0100


On Sat, Dec 05, 1998 at 12:24:10AM -0500, David Feuer wrote:

> Is there a way for a program to make a section of code atomic? I think
> it would be useful to allow root programs (or, for added security, root

I don't think so.

> a) the scheduler will not interrupt them
> b) they cannot receive _any_ signals, including SIGKILL

That means even root gives up some control over the system. I like to think
that root = system god, and if root cannot do something (i.e. kill a
process, delete a file, change something, etc) this something is broken.

With atomic user space programs you are effectively giving up control, like
in NT where you can (e.g.) chown <myself> <file> as root, but never chown
<someotherluser> <file> - even not as root. The philosophy behind this is
that users should be able to note if someone, even admins, fumbled with
their data - but I still think that keeping the admin from doing something
is a limitation that shouldn't exist.

> it more possible to write such things as background defragmenters,

you don't need one with ext2fs, it has its own (kind of).

> So the defragmenter flipped around parts of the disk without anybody
> else caring. With the defragger running as an idle-process only when
> the system load is low, it can effectively keep the disk at near-zero
> fragmentation without serious slow-down.

hack this into the file system, not into a user space program. Or maybe help
the guys/gals hacking on JFS features for ext3fs. :-)

SIGKILL is like pulling the plug, and ANY code should be capable of coping
with that.

-- 
_ciao, Jens_______________________________ http://www.pinguin.conetix.de
    cat /dev/boiler/water | tea | sieve > /cup
    mount -t hdev /dev/human/mouth01 /mouth ; cat /cup >/mouth/gulp

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