Re: [Xen-devel] Boot time trap handling

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 12.05.14 at 14:53, <andrew.cooper3@xxxxxxxxxx> wrote:
> One point I forgot to say was that I plan to leave the reserved
> exception vectors with non-present descriptors.  This way we get a fatal
> GPF indicating the exact vector, rather than a PCPU lockup optionally
> followed by an NMI watchdog timeout when another pcpu times out waiting
> for the locked up pcpu to rendezvous in the time calibration code.

This I'm not sure about. Replacing the simplistic handler is a must of
course, but I would tend towards a generic exception handler wired
up to fatal_trap() rather than having to decode the #GP error code
to understand what's going on.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel