Re: [Xen-devel] [PATCH] hvm: Use main memory for video memory. [Was: xend, 8MB video memory and ballooning]

[Keir Fraser <keir.fraser@xxxxxxxxxxxxx>]

On 31/7/08 15:57, "Samuel Thibault" <samuel.thibault@xxxxxxxxxxxxx> wrote:

>> I'd've hoped it would just work... does the extra page of allocation
>> during restore fail?
> 
> It does work, but I guess it replaces the mapping of the "magic" page
> with the mapping of the allocated page, so that shared_info wouldn't
> work any more.

PV drivers will add_to_physmap() these pages again during resume, which will
map the correct pages (shared_info and grant_table, notably) on top of the
RAM pages that save/restore will leave behind. things should work fine then
and the displaced RAM pages *should* get freed immediately. Even if they
don't they will on domain destruction and it's only a few pages...

>> I'm thinking that the grant tables will be mapped in the I/O hole too, and
>> that is from 'Xen heap' just like shared_info is.
> 
> Urgl, hadn't thought about those, I guess the PV drivers would
> re-etablish them on resume anyway?

Yep, as above.

>> So special-casing one and not the other sure is odd.
> 
> Actually, shouldn't it be up to the hypervisor to not return all these
> from the SHADOW_OP_PEEK operation?

Well, maybe. :-) But shadow code doesn't special-case these Xen-heap pages I
think. And doing so could be a pain, and maybe cause other issues?

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel