> On Mon, Nov 22, 1999 at 10:10:16PM +0100, Pavel Machek wrote:
> > Ioctls should be banned. They make network transparent remote exec
> > ...
>
> plan9 solved this with ctl files. i don't see why linux shouldn't move
> the same way in order to reap the benefits of distributable computing.
And not only distributable computing. Ioctl lacks enforced interface
and a model to back up this interface up. Ioctl is slow (a syscall
for each ioctl is required). Ioctl is not safe (even with
{device,tag,type,size} semantics for request). And finally, ioctl's
are not portable.
I'm not saying they should banned right away, but that they should be
considered for replacement in the near future with something like
plan9 ctl files or /dev/sg type of thing.
> how do systems like mosix work around ioctl problems?
I thought mosix does not have ioctl problem, since they don't move
the processes, but allocate them in close proximity (for an ioctl)
to the resources.
Pjotr.
-
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/