Re: [Xen-devel] [PATCH RFC V7 4/5] xen, libxc: Request page fault injection via libxc

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 27.08.14 at 02:54, <kevin.tian@xxxxxxxxx> wrote:
> From code seems this info is a condition for PF injection, instead of
> recording VCPU faulting information. So it looks OK to be a per-domain
> structure, but the structure name 'fault_info' is too generic...

Not really, because it also makes assumptions about the guest OS.
Which I'd consider okay for a particular end product like the one
BitDefender is creating, but not for an interface. In particular here
the assumption is that you can inject a fault on an arbitrary vCPU
to achieve a certain goal, which certainly isn't valid without knowing
how the OS works.

But yes, with the naming changed to reflect the special purpose,
this might become less of an issue. Pretending (or making appear)
something to be general purpose when it isn't, however, would
only be setting us up for future problems.

An alternative might be to design the external interface in a way
being vCPU-centric, with a way to specify a "wildcard" to serve the
particular purpose here.

Jan


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