> > No. I have a very standard server with a 6x86L 133 MHz on an Intel VX
> > motherboard that crashes every 5 minutes if I enable the SUSP_HLT
> > option, which is by far the most inocuous option of all.
>
> Oh, I wasn't aware it was adversely affecting anyone.
>
> > So, no. Sorry :-(
>
> Perhaps an option for HLT could be available if CONFIG_EXPIRIMENTAL is
> set? In my day job we sell systems with Linux running under the hood.
There have been other problems with hlt in the past, hence the
kernel has a boot option no-hlt which disables the use of hlt in
the idle loop. In may be preferable to simply enable SUSP_HLT
and document the fact that a system that crashes when idle may
need to be booted with no-hlt - regardless of techie details
like CPU, motherboard chipset and all that crap.
> Well, I have and I haven't. :) I sent you a private mail earlier about it
> (should have it by now; if not, let me know). It seems that with the
> NO_LOCK option set, you can indeed defend against the CPU loop exploit.
> But, if you try it (and then kill it, for example), the systems becomes
> unstable for X. If I never run the loop exploit, X is fine. If I run it,
> I can't safely use X (communicator in particular) until I reboot. Does
> this even sound possible?
It sounds weird. With NO_LOCK in effect the only thing that
the loop does that other code might not is back to back
read-modify-write cycles... Uh oh...
Mike
-- .----------------------------------------------------------------------. | Mike Jagdis | Internet: mailto:mike@roan.co.uk | | Roan Technology Ltd. | | | 54A Peach Street, Wokingham | Telephone: +44 118 989 0403 | | RG40 1XG, ENGLAND | Fax: +44 118 989 1195 | `----------------------------------------------------------------------'