Xen devel - Sep 2008 - is_mapped() in xc_domain_save()

Hi all,

I think I have a pretty good grasp on domain save/migration code in
xc_domain_save() in Xen 3.2.x, except for the what the is_mapped()
macro is doing in following:

/* Hypercall interfaces operate in PFNs for HVM guests
* and MFNs for PV guests */
if ( hvm )
    pfn_type[batch] = n;
else
    pfn_type[batch] = pfn_to_mfn(n);

if ( !is_mapped(pfn_type[batch]) )
{
    /*
    ** not currently in psuedo-physical map -- set bit
    ** in to_fix since we must send this page in last_iter
    ** unless its sent sooner anyhow, or it never enters
    ** pseudo-physical map (e.g. for ballooned down doms)
    */
    set_bit(n, to_fix);
    continue;
}

This chunk of code makes sense for saving PV domains.  The is_mapped()
function checks the MSB of the mfn returned by pfn_to_mfn() to make
sure the pfn is actually a part of the pseudo-physical map.  And if
it''s not, we mark that pfn as to_fix for later in case it later
becomes part of the pseudo-physical map.

But for HVM domains, it doesn''t make any sense to me.  The is_mapped()
is being called on a pfn, not mfn, and it doesn''t seem like it would
ever return false.  Is it just a piece of irrelevant code for HVM
domains?

Thanks,
Mike

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

On 28/9/08 04:31, "Mike Sun" <msun@gatech.edu> wrote:
> But for HVM domains, it doesn''t make any sense to me.  The
is_mapped()
> is being called on a pfn, not mfn, and it doesn''t seem like it
would
> ever return false.  Is it just a piece of irrelevant code for HVM
> domains?
Yes, irrelevant but harmless.

 -- Keir



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

Thanks for clearing that up.  And thanks for all the recent responses;
really appreciate it.

Mike

On Mon, Sep 29, 2008 at 4:41 AM, Keir Fraser <keir.fraser@eu.citrix.com>
wrote:
>
>
>
> On 28/9/08 04:31, "Mike Sun" <msun@gatech.edu> wrote:
>
>> But for HVM domains, it doesn''t make any sense to me.  The
is_mapped()
>> is being called on a pfn, not mfn, and it doesn''t seem like it
would
>> ever return false.  Is it just a piece of irrelevant code for HVM
>> domains?
>
> Yes, irrelevant but harmless.
>
>  -- Keir
>
>
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel