Rusty Russell wrote:
> Chris: [RFC PATCH 12/33] Change __FIXADDR_TOP to leave room for the hypervisor.
> - Replace with dynamic (Geerd) patch, put in paravirt_ops structure.
>
I'm looking at this, and I'm not sure that there needs to be a void
(*set_fixmap_top)(unsigned long top) entry in paravirt ops. It seems to
me that the hypervisor's init code can call the normal set_fixmap_top()
function (which Geerd's patch introduces) with whatever appropriate
value it needs when it also installs the rest of the paravirt ops
pointers and/or patches. After all, this is something the hypervisor
needs to set in the kernel, rather than something the kernel needs to do
to the hypervisor.
Does that sound reasonable, or does there really need to be a paravirt
entry for this?
Also, since this is a pretty generic patch with no earlier dependencies,
we can push it near the front of the patch queue.
J
