[Xen-changelog] [xen-unstable] [IA64] allow __assign_domain_page() to assign real MMIO page over _PAGE_IO.

[Xen patchbot-unstable <patchbot-unstable@xxxxxxxxxxxxxxxxxxx>]

# HG changeset patch
# User Isaku Yamahata <yamahata@xxxxxxxxxxxxx>
# Date 1224815889 -32400
# Node ID 7ad8c47f5c4b5ef43ec71cb946c39b08eeac3d18
# Parent  02c8733e2d91557a98b6964bfb7e136f0e3924d4
[IA64] allow __assign_domain_page() to assign real MMIO page over _PAGE_IO.

When VTD is enabled on HVM domain,
P2M entry may change from _PAGE_IO to real MMIO page.

Signed-off-by: Anthony Xu <anthony.xu@xxxxxxxxx>
---
 xen/arch/ia64/xen/mm.c |    8 ++++++++
 1 files changed, 8 insertions(+)

diff -r 02c8733e2d91 -r 7ad8c47f5c4b xen/arch/ia64/xen/mm.c
--- a/xen/arch/ia64/xen/mm.c    Wed Oct 22 17:20:15 2008 +0900
+++ b/xen/arch/ia64/xen/mm.c    Fri Oct 24 11:38:09 2008 +0900
@@ -917,10 +917,18 @@ __assign_domain_page(struct domain *d,
 
     old_pte = __pte(0);
     new_pte = pfn_pte(physaddr >> PAGE_SHIFT, __pgprot(prot));
+ again_hvm_page_io:
     ret_pte = ptep_cmpxchg_rel(&d->arch.mm, mpaddr, pte, old_pte, new_pte);
     if (pte_val(ret_pte) == pte_val(old_pte)) {
         smp_mb();
         return 0;
+    }
+    /* in HVM guest, when VTD is enabled,
+     * P2M entry may change from _PAGE_IO type to real MMIO page 
+     */
+    if(VMX_DOMAIN(d->vcpu[0]) && (pte_val(ret_pte) & _PAGE_IO)) {
+        old_pte = ret_pte;
+        goto again_hvm_page_io;
     }
 
     // dom0 tries to map real machine's I/O region, but failed.

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