[Xen-devel] [PATCH 2/6] x86/EPT: don't walk entire page tables when changing types on a range

["Jan Beulich" <JBeulich@xxxxxxxx>]

This requires a new P2M backend hook and a little bit of extra care and
accounting in the generic function.

Note that even on leaf entries we must not immediately set the new
type (in an attempt to avoid the EPT_MISCONFIG VM exits), since the
global accounting in p2m_change_type_range() gets intentionally done
only after updating page tables (or else the update there would
conflict with the function's own use of p2m_is_logdirty_range()), and
the correct type can only be calculated with that in place.

Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>

--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -730,17 +730,33 @@ void p2m_change_type_range(struct domain
 {
     p2m_access_t a;
     p2m_type_t pt;
-    unsigned long gfn;
+    unsigned long gfn = start;
     mfn_t mfn;
     struct p2m_domain *p2m = p2m_get_hostp2m(d);
     int rc = 0;
 
-    BUG_ON(p2m_is_grant(ot) || p2m_is_grant(nt));
+    ASSERT(ot != nt);
+    ASSERT(p2m_is_changeable(ot) && p2m_is_changeable(nt));
 
     p2m_lock(p2m);
     p2m->defer_nested_flush = 1;
 
-    for ( gfn = start; gfn < end; )
+    if ( unlikely(end > p2m->max_mapped_pfn) )
+    {
+        if ( !gfn )
+        {
+            p2m->change_entry_type_global(p2m, ot, nt);
+            gfn = end;
+        }
+        end = p2m->max_mapped_pfn + 1;
+    }
+
+    if ( gfn < end && p2m->change_entry_type_range )
+    {
+        rc = p2m->change_entry_type_range(p2m, ot, nt, gfn, end - 1);
+        gfn = end;
+    }
+    while ( !rc && gfn < end )
     {
         unsigned int order;
 
@@ -769,12 +785,18 @@ void p2m_change_type_range(struct domain
                 order = PAGE_ORDER_4K;
         }
         if ( pt == ot )
-            p2m_set_entry(p2m, gfn, mfn, order, nt, a);
+            rc = p2m_set_entry(p2m, gfn, mfn, order, nt, a);
         gfn += 1UL << order;
         gfn &= -1UL << order;
         if ( !gfn )
             break;
     }
+    if ( rc )
+    {
+        printk(XENLOG_G_ERR "Error %d changing Dom%d GFNs [%lx,%lx] from %d to 
%d\n",
+               rc, d->domain_id, start, end - 1, ot, nt);
+        domain_crash(d);
+    }
 
     switch ( nt )
     {
--- a/xen/arch/x86/mm/p2m-ept.c
+++ b/xen/arch/x86/mm/p2m-ept.c
@@ -295,6 +295,70 @@ static bool_t ept_invalidate_emt(mfn_t m
     return changed;
 }
 
+static int ept_invalidate_emt_range(struct p2m_domain *p2m,
+                                    unsigned int target,
+                                    unsigned long first_gfn,
+                                    unsigned long last_gfn)
+{
+    ept_entry_t *table;
+    unsigned long gfn_remainder = first_gfn;
+    unsigned int i, index;
+    int rc = 0, ret = GUEST_TABLE_MAP_FAILED;
+
+    table = map_domain_page(pagetable_get_pfn(p2m_get_pagetable(p2m)));
+    for ( i = ept_get_wl(&p2m->ept); i > target; --i )
+    {
+        ret = ept_next_level(p2m, 1, &table, &gfn_remainder, i);
+        if ( ret == GUEST_TABLE_MAP_FAILED )
+            goto out;
+        if ( ret != GUEST_TABLE_NORMAL_PAGE )
+            break;
+    }
+
+    if ( i > target )
+    {
+        /* We need to split the original page. */
+        ept_entry_t split_ept_entry;
+
+        index = gfn_remainder >> (i * EPT_TABLE_ORDER);
+        split_ept_entry = atomic_read_ept_entry(&table[index]);
+        ASSERT(is_epte_superpage(&split_ept_entry));
+        if ( !ept_split_super_page(p2m, &split_ept_entry, i, target) )
+        {
+            ept_free_entry(p2m, &split_ept_entry, i);
+            rc = -ENOMEM;
+            goto out;
+        }
+        atomic_write_ept_entry(&table[index], split_ept_entry);
+
+        for ( ; i > target; --i )
+            if ( !ept_next_level(p2m, 1, &table, &gfn_remainder, i) )
+                break;
+        ASSERT(i == target);
+    }
+
+    index = gfn_remainder >> (i * EPT_TABLE_ORDER);
+    i = (last_gfn >> (i * EPT_TABLE_ORDER)) & (EPT_PAGETABLE_ENTRIES - 1);
+    for ( ; index <= i; ++index )
+    {
+        ept_entry_t e = atomic_read_ept_entry(&table[index]);
+
+        if ( is_epte_valid(&e) && is_epte_present(&e) &&
+             (e.emt != MTRR_NUM_TYPES || !e.recalc) )
+        {
+            e.emt = MTRR_NUM_TYPES;
+            e.recalc = 1;
+            atomic_write_ept_entry(&table[index], e);
+            rc = 1;
+        }
+    }
+
+ out:
+    unmap_domain_page(table);
+
+    return rc;
+}
+
 static int resolve_misconfig(struct p2m_domain *p2m, unsigned long gfn)
 {
     struct ept_data *ept = &p2m->ept;
@@ -806,6 +870,53 @@ static void ept_change_entry_type_global
         ept_sync_domain(p2m);
 }
 
+static int ept_change_entry_type_range(struct p2m_domain *p2m,
+                                       p2m_type_t ot, p2m_type_t nt,
+                                       unsigned long first_gfn,
+                                       unsigned long last_gfn)
+{
+    unsigned int i, wl = ept_get_wl(&p2m->ept);
+    unsigned long mask = (1 << EPT_TABLE_ORDER) - 1;
+    int rc = 0, sync = 0;
+
+    if ( !ept_get_asr(&p2m->ept) )
+        return -EINVAL;
+
+    for ( i = 0; i <= wl; )
+    {
+        if ( first_gfn & mask )
+        {
+            unsigned long end_gfn = min(first_gfn | mask, last_gfn);
+
+            rc = ept_invalidate_emt_range(p2m, i, first_gfn, end_gfn);
+            sync |= rc;
+            if ( rc < 0 || end_gfn >= last_gfn )
+                break;
+            first_gfn = end_gfn + 1;
+        }
+        else if ( (last_gfn & mask) != mask )
+        {
+            unsigned long start_gfn = max(first_gfn, last_gfn & ~mask);
+
+            rc = ept_invalidate_emt_range(p2m, i, start_gfn, last_gfn);
+            sync |= rc;
+            if ( rc < 0 || start_gfn <= first_gfn )
+                break;
+            last_gfn = start_gfn - 1;
+        }
+        else
+        {
+            ++i;
+            mask |= mask << EPT_TABLE_ORDER;
+        }
+    }
+
+    if ( sync )
+        ept_sync_domain(p2m);
+
+    return rc < 0 ? rc : 0;
+}
+
 static void ept_memory_type_changed(struct p2m_domain *p2m)
 {
     unsigned long mfn = ept_get_asr(&p2m->ept);
@@ -854,6 +965,7 @@ int ept_p2m_init(struct p2m_domain *p2m)
     p2m->set_entry = ept_set_entry;
     p2m->get_entry = ept_get_entry;
     p2m->change_entry_type_global = ept_change_entry_type_global;
+    p2m->change_entry_type_range = ept_change_entry_type_range;
     p2m->memory_type_changed = ept_memory_type_changed;
     p2m->audit_p2m = NULL;
 
--- a/xen/include/asm-x86/p2m.h
+++ b/xen/include/asm-x86/p2m.h
@@ -243,6 +243,10 @@ struct p2m_domain {
     void               (*change_entry_type_global)(struct p2m_domain *p2m,
                                                    p2m_type_t ot,
                                                    p2m_type_t nt);
+    int                (*change_entry_type_range)(struct p2m_domain *p2m,
+                                                  p2m_type_t ot, p2m_type_t nt,
+                                                  unsigned long first_gfn,
+                                                  unsigned long last_gfn);
     void               (*memory_type_changed)(struct p2m_domain *p2m);
     
     void               (*write_p2m_entry)(struct p2m_domain *p2m,

Attachment: EPT-implement-cetr.patch
Description: Text document

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel