Re: [Xen-ia64-devel] [PATCH 4/9] [P2M/VP step 1] xen: domain0 builder change

2006-04-25 Thread Tristan Gingold
Le Mardi 25 Avril 2006 06:12, Isaku Yamahata a écrit :
  4 / 9
Two possible problems (we may fix them later):
* should ACPI be copied instead of mapped ?
  At least some ACPI tables must be duplicated to other IO domains.

* We must be sure that only physical memory  4GB is mapped below 4 GB in 
domains.  This seems to be already handled in xen/x86, but we must be sure it 
is also handled in xen/ia64.

Tristan.




___
Xen-ia64-devel mailing list
Xen-ia64-devel@lists.xensource.com
http://lists.xensource.com/xen-ia64-devel


Re: [Xen-ia64-devel] [PATCH 4/9] [P2M/VP step 1] xen: domain0 builder change

2006-04-25 Thread Isaku Yamahata

Hi Tristan. Thank you for your comment.

On Tue, Apr 25, 2006 at 05:25:10PM +0200, Tristan Gingold wrote:
 Le Mardi 25 Avril 2006 06:12, Isaku Yamahata a écrit :
   4 / 9
 Two possible problems (we may fix them later):
 * should ACPI be copied instead of mapped ?
   At least some ACPI tables must be duplicated to other IO domains.

I agree with you that customized ACPI tables is needed for IO domains.
However I prefer read-only mapping of these pages to domain for domain0

I haven't implemented read-only grant page mapping yet.
(I plan to do it after the mergeing effort)
Once it is implemented, it can be used for read only mapping of
ACPI-related pages.


 * We must be sure that only physical memory  4GB is mapped below 4 GB in 
 domains.  This seems to be already handled in xen/x86, but we must be sure it 
 is also handled in xen/ia64.

Does alloc_domheap_page() handles it?
#define alloc_domheap_page(d) (alloc_domheap_pages(d,0,0))
It passes flags = 0.
Or do you mention about other things?

-- 
yamahata

___
Xen-ia64-devel mailing list
Xen-ia64-devel@lists.xensource.com
http://lists.xensource.com/xen-ia64-devel