x86: make vcpu_destroy_pagetables() preemptible

... as it may take significant amounts of time.

The function, being moved to mm.c as the better home for it anyway, and
to avoid having to make a new helper function there non-static, is
given a "preemptible" parameter temporarily (until, in a subsequent
patch, its other caller is also being made capable of dealing with
preemption).

This is part of CVE-2013-1918 / XSA-45.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Tim Deegan <tim@xen.org>

xen/arch/x86/domain.c

@@ -71,8 +71,6 @@ void (*dead_idle) (void) __read_mostly = default_dead_idle;
 static void paravirt_ctxt_switch_from(struct vcpu *v);
 static void paravirt_ctxt_switch_to(struct vcpu *v);
 
-static void vcpu_destroy_pagetables(struct vcpu *v);
-
 static void default_idle(void)
 {
     local_irq_disable();
@@ -946,7 +944,7 @@ void arch_vcpu_reset(struct vcpu *v)
     if ( !is_hvm_vcpu(v) )
     {
         destroy_gdt(v);
-        vcpu_destroy_pagetables(v);
+        vcpu_destroy_pagetables(v, 0);
     }
     else
     {
@@ -1944,58 +1942,6 @@ static int relinquish_memory(
     return ret;
 }
 
-static void vcpu_destroy_pagetables(struct vcpu *v)
-{
-    struct domain *d = v->domain;
-    unsigned long pfn = pagetable_get_pfn(v->arch.guest_table);
-
-    if ( is_pv_32on64_vcpu(v) )
-    {
-        l4_pgentry_t *l4tab = map_domain_page(pfn);
-
-        pfn = l4e_get_pfn(*l4tab);
-
-        if ( pfn != 0 )
-        {
-            if ( paging_mode_refcounts(d) )
-                put_page(mfn_to_page(pfn));
-            else
-                put_page_and_type(mfn_to_page(pfn));
-        }
-
-        l4e_write(l4tab, l4e_empty());
-        unmap_domain_page(l4tab);
-
-        v->arch.cr3 = 0;
-        return;
-    }
-
-    if ( pfn != 0 )
-    {
-        if ( paging_mode_refcounts(d) )
-            put_page(mfn_to_page(pfn));
-        else
-            put_page_and_type(mfn_to_page(pfn));
-        v->arch.guest_table = pagetable_null();
-    }
-
-    /* Drop ref to guest_table_user (from MMUEXT_NEW_USER_BASEPTR) */
-    pfn = pagetable_get_pfn(v->arch.guest_table_user);
-    if ( pfn != 0 )
-    {
-        if ( !is_pv_32bit_vcpu(v) )
-        {
-            if ( paging_mode_refcounts(d) )
-                put_page(mfn_to_page(pfn));
-            else
-                put_page_and_type(mfn_to_page(pfn));
-        }
-        v->arch.guest_table_user = pagetable_null();
-    }
-
-    v->arch.cr3 = 0;
-}
-
 int domain_relinquish_resources(struct domain *d)
 {
     int ret;
@@ -2014,7 +1960,9 @@ int domain_relinquish_resources(struct domain *d)
         /* Drop the in-use references to page-table bases. */
         for_each_vcpu ( d, v )
         {
-            vcpu_destroy_pagetables(v);
+            ret = vcpu_destroy_pagetables(v, 1);
+            if ( ret )
+                return ret;
 
             unmap_vcpu_info(v);
         }

xen/arch/x86/mm.c

@@ -2579,6 +2579,79 @@ static void put_superpage(unsigned long mfn)
     return;
 }
 
+static int put_old_guest_table(struct vcpu *v)
+{
+    int rc;
+
+    if ( !v->arch.old_guest_table )
+        return 0;
+
+    switch ( rc = put_page_and_type_preemptible(v->arch.old_guest_table, 1) )
+    {
+    case -EINTR:
+    case -EAGAIN:
+        return -EAGAIN;
+    }
+
+    v->arch.old_guest_table = NULL;
+
+    return rc;
+}
+
+int vcpu_destroy_pagetables(struct vcpu *v, bool_t preemptible)
+{
+    unsigned long mfn = pagetable_get_pfn(v->arch.guest_table);
+    struct page_info *page;
+    l4_pgentry_t *l4tab = NULL;
+    int rc = put_old_guest_table(v);
+
+    if ( rc )
+        return rc;
+
+    if ( is_pv_32on64_vcpu(v) )
+    {
+        l4tab = map_domain_page(mfn);
+        mfn = l4e_get_pfn(*l4tab);
+    }
+
+    if ( mfn )
+    {
+        page = mfn_to_page(mfn);
+        if ( paging_mode_refcounts(v->domain) )
+            put_page(page);
+        else
+            rc = put_page_and_type_preemptible(page, preemptible);
+    }
+
+    if ( l4tab )
+    {
+        if ( !rc )
+            l4e_write(l4tab, l4e_empty());
+        unmap_domain_page(l4tab);
+    }
+    else if ( !rc )
+    {
+        v->arch.guest_table = pagetable_null();
+
+        /* Drop ref to guest_table_user (from MMUEXT_NEW_USER_BASEPTR) */
+        mfn = pagetable_get_pfn(v->arch.guest_table_user);
+        if ( mfn )
+        {
+            page = mfn_to_page(mfn);
+            if ( paging_mode_refcounts(v->domain) )
+                put_page(page);
+            else
+                rc = put_page_and_type_preemptible(page, preemptible);
+        }
+        if ( !rc )
+            v->arch.guest_table_user = pagetable_null();
+    }
+
+    v->arch.cr3 = 0;
+
+    return rc;
+}
+
 int new_guest_cr3(unsigned long mfn)
 {
     struct vcpu *curr = current;
@@ -2733,12 +2806,21 @@ long do_mmuext_op(
     unsigned int foreigndom)
 {
     struct mmuext_op op;
-    int rc = 0, i = 0, okay;
     unsigned long type;
-    unsigned int done = 0;
+    unsigned int i = 0, done = 0;
     struct vcpu *curr = current;
     struct domain *d = curr->domain;
     struct domain *pg_owner;
+    int okay, rc = put_old_guest_table(curr);
+
+    if ( unlikely(rc) )
+    {
+        if ( likely(rc == -EAGAIN) )
+            rc = hypercall_create_continuation(
+                     __HYPERVISOR_mmuext_op, "hihi", uops, count, pdone,
+                     foreigndom);
+        return rc;
+    }
 
     if ( unlikely(count & MMU_UPDATE_PREEMPTED) )
     {

xen/arch/x86/x86_64/compat/mm.c

@@ -365,7 +365,7 @@ int compat_mmuext_op(XEN_GUEST_HANDLE_PARAM(mmuext_op_compat_t) cmp_uops,
                                     : mcs->call.args[1];
                 unsigned int left = arg1 & ~MMU_UPDATE_PREEMPTED;
 
-                BUG_ON(left == arg1);
+                BUG_ON(left == arg1 && left != i);
                 BUG_ON(left > count);
                 guest_handle_add_offset(nat_ops, i - left);
                 guest_handle_subtract_offset(cmp_uops, left);

xen/include/asm-x86/domain.h

@@ -405,6 +405,7 @@ struct arch_vcpu
 
     pagetable_t guest_table_user;       /* (MFN) x86/64 user-space pagetable */
     pagetable_t guest_table;            /* (MFN) guest notion of cr3 */
+    struct page_info *old_guest_table;  /* partially destructed pagetable */
     /* guest_table holds a ref to the page, and also a type-count unless
      * shadow refcounts are in use */
     pagetable_t shadow_table[4];        /* (MFN) shadow(s) of guest */

xen/include/asm-x86/mm.h

@@ -556,6 +556,7 @@ void audit_domains(void);
 int new_guest_cr3(unsigned long pfn);
 void make_cr3(struct vcpu *v, unsigned long mfn);
 void update_cr3(struct vcpu *v);
+int vcpu_destroy_pagetables(struct vcpu *, bool_t preemptible);
 void propagate_page_fault(unsigned long addr, u16 error_code);
 void *do_page_walk(struct vcpu *v, unsigned long addr);