x86/PV: check GDT/LDT limits during emulation

Accesses beyond the LDT limit originating from emulation would trigger
the ASSERT() in pv_map_ldt_shadow_page(). On production builds such
accesses would cause an attempt to promote the touched page (offset from
the present LDT base address) to a segment descriptor one. If this
happens to succeed, guest user mode would be able to elevate its
privileges to that of the guest kernel. This is particularly easy when
there's no LDT at all, in which case the LDT base stored internally to
Xen is simply zero.

Also adjust the ASSERT() that was triggering: It was off by one to
begin with, and for production builds we also better use
ASSERT_UNREACHABLE() instead with suitable recovery code afterwards.

This is XSA-298.

Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: 93021cbe880a8013691a48d0febef8ed7d3e3ebd
master date: 2019-10-31 16:08:16 +0100

diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c
index f10d823f8e051062eaa9a88ced8c85e8a2c02d95..bdc26821543c9ce83514011f01b8065406e0647e 100644 (file)
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -712,6 +712,18 @@ int map_ldt_shadow_page(unsigned int off)
 
     BUG_ON(unlikely(in_irq()));
 
+    /*
+     * Prior limit checking should guarantee this property.  NB. This is
+     * safe as updates to the LDT can only be made by MMUEXT_SET_LDT to the
+     * current vcpu, and vcpu_reset() will block until this vcpu has been
+     * descheduled before continuing.
+     */
+    if ( unlikely((off >> 3) >= v->arch.pv_vcpu.ldt_ents) )
+    {
+        ASSERT_UNREACHABLE();
+        return 0;
+    }
+
     if ( is_pv_32bit_domain(d) )
         gva = (u32)gva;
     guest_get_eff_kern_l1e(v, gva, &l1e);

diff --git a/xen/arch/x86/traps.c b/xen/arch/x86/traps.c
index f3c87053eda2fda5a9e383c00cd270b630df4da9..53301574e2e55980d9c3987446ef9766a9e8eeda 100644 (file)
--- a/xen/arch/x86/traps.c
+++ b/xen/arch/x86/traps.c
@@ -1639,7 +1639,14 @@ static int read_descriptor(unsigned int sel,
 {
     struct desc_struct desc;
 
-    if ( sel < 4)
+    if ( sel < 4 ||
+         /*
+          * Don't apply the GDT limit here, as the selector may be a Xen
+          * provided one. __get_user() will fail (without taking further
+          * action) for ones falling in the gap between guest populated
+          * and Xen ones.
+          */
+         ((sel & 4) && (sel >> 3) >= v->arch.pv_vcpu.ldt_ents) )
         desc.b = desc.a = 0;
     else if ( __get_user(desc,
                          (const struct desc_struct *)(!(sel & 4)
@@ -1698,7 +1705,13 @@ static int read_gate_descriptor(unsigned int gate_sel,
         (!(gate_sel & 4) ? GDT_VIRT_START(v) : LDT_VIRT_START(v))
         + (gate_sel >> 3);
     if ( (gate_sel < 4) ||
-         ((gate_sel >= FIRST_RESERVED_GDT_BYTE) && !(gate_sel & 4)) ||
+         /*
+          * We're interested in call gates only, which occupy a single
+          * seg_desc_t for 32-bit and a consecutive pair of them for 64-bit.
+          */
+         ((gate_sel >> 3) + !is_pv_32bit_vcpu(v) >=
+          (gate_sel & 4 ? v->arch.pv_vcpu.ldt_ents
+                        : v->arch.pv_vcpu.gdt_ents)) ||
          __get_user(desc, pdesc) )
         return 0;
 
@@ -1717,7 +1730,7 @@ static int read_gate_descriptor(unsigned int gate_sel,
     if ( !is_pv_32bit_vcpu(v) )
     {
         if ( (*ar & 0x1f00) != 0x0c00 ||
-             (gate_sel >= FIRST_RESERVED_GDT_BYTE - 8 && !(gate_sel & 4)) ||
+             /* Limit check done above already. */
              __get_user(desc, pdesc + 1) ||
              (desc.b & 0x1f00) )
             return 0;