[Xen-devel] [PATCH] [HVM] fix HVM restore hang

["Zhai, Edwin" <edwin.zhai@xxxxxxxxx>]

r18383 mark video memory as ram, and make all valid pages migrated,
including vlapic page (0xFEE00), and share page(0xFFFFF).
An extra memory population for lapic page would override previous
mapping then cause HVM guest with vlapic acceleration hang.

This patch from Keir makes mmio page as invalid, so xc_domain_save
skip vlapic page.

Share page is safe, as the mapping is not set up until after normal
guest memory is restored.


--
best rgds,
edwin

Index: hv/xen/arch/x86/mm.c
===================================================================
--- hv.orig/xen/arch/x86/mm.c
+++ hv/xen/arch/x86/mm.c
@@ -1542,6 +1542,7 @@ static int mod_l1_entry(l1_pgentry_t *pl
     unsigned long mfn;
     struct page_info *l1pg = mfn_to_page(gl1mfn);
     int rc = 1;
+    p2m_type_t p2mt;
 
     page_lock(l1pg);
 
@@ -1558,8 +1559,8 @@ static int mod_l1_entry(l1_pgentry_t *pl
     if ( l1e_get_flags(nl1e) & _PAGE_PRESENT )
     {
         /* Translate foreign guest addresses. */
-        mfn = gmfn_to_mfn(FOREIGNDOM, l1e_get_pfn(nl1e));
-        if ( unlikely(mfn == INVALID_MFN) )
+        mfn = gfn_to_mfn(FOREIGNDOM, l1e_get_pfn(nl1e), &p2mt);
+        if ( !p2m_is_ram(p2mt) || unlikely(mfn == INVALID_MFN) )
             return page_unlock(l1pg), 0;
         ASSERT((mfn & ~(PADDR_MASK >> PAGE_SHIFT)) == 0);
         nl1e = l1e_from_pfn(mfn, l1e_get_flags(nl1e));

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