we were partial to the soft ring cycle... I suppose that'd make mac the valkyrie :P

...of the Nibelungs?

hrm. cmn_err does not know about %z, how very disappointing.

do we ever move around physical pages of kernel text (ignoring that unload/loading of module could change things)?

(I'm thinking no -- at least also ignoring DR)

i'm picking back up the hyper-v stuff (since there's a bit more interest at work), so hopefully I can finished getting the gen2 support to a usable state

but one of the things right now is a very incestuous relationship between the hyperv and vmbus drivers

and the ultimate reason for that is that to make hypervisor calls, you have to give the HV a physical page address that the HV will overlay with the appropriate instructions (then you just jmp to that page's kernel va to make calls)... and right now both modules are sharing DMA-alloced memory for this (and other things)

but it's pretty much exactly have xen does, and we have a hypercall_page already in the kernel text for this exact same thing for xen guests (but always gets built)

so i'd like to just use that since it's serving the exact same purpose (and not really possible to conflict since there's only 1 possible HV immediately 'above' the guest)

sounds like reasonable idea:)

jbk: It seems OK, but it would be good to have something explicit and interlocking about the usage

some other global that says who is using it, and which prevents accidental use by something else