Messages in this thread | | | Date | Mon, 01 Mar 2010 17:43:37 -0800 | From | Jeremy Fitzhardinge <> | Subject | Re: [Xen-devel] [PATCH 3/7] xen/hvm: Xen PV extension of HVM initialization |
| |
On 03/01/2010 05:38 PM, Sheng Yang wrote: >> Why use this? Why not just set the domain type to HVM, and leave the >> "status" field as a bitset of available paravirtualizations? >> > A annoy thing in pv drivers is that it would test if the domain type is _NOT_ > XEN_NATIVE. So set the domain to XEN_HVM_DOMAIN would result in PV driver > initialization then probably panic. Maybe we can do something to PV drivers, > as patch 6 and a part of patch 7. >
In that case, change them to specifically test for PV until they're ready to deal with an HVM environment.
> > Part of pv_ops are overlapped. I would try if a new file would bring much > duplicate. >
It shouldn't do; un-static things if necessary.
>>> + .align PAGE_SIZE_asm >>> +ENTRY(shared_info_page) >>> + .skip PAGE_SIZE_asm >>> +.popsection >>> >> Why does this need to be defined in asm? Can't it be either allocated >> or defined in C? >> > I think we need a aligned page, as hypercall page. >
That can be declared in C with the __page_aligned_bss annotation. And couldn't you dynamically allocate it with the bootmem allocator?
>> I think "status" is a misnomer here. Isn't it specifically a set of PV >> features which are active? >> > Could you give a suggestion of the name? I am not a native English speaker... >
xen_hvm_pv_features.
J
| |