x86/EPT: work around hardware erratum setting A bit

Since commit 191b3f3344ee ("p2m/ept: enable PML in p2m-ept for
log-dirty"), the A and D bits of EPT paging entries are set
unconditionally, regardless of whether PML is enabled or not. This
causes a regression in Xen 4.6 on some processors due to Intel Errata
AVR41 -- HVM guests get severe memory corruption when the A bit is set
due to incorrect TLB flushing on mov to cr3. The erratum affects the
Atom C2000 family (Avoton).

To fix, do not set the A bit on this processor family.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Move feature suppression to feature detection code. Add command line
override.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Tested-by: Andrew Cooper <andrew.cooper3@citrix.com>

diff --git a/docs/misc/xen-command-line.markdown b/docs/misc/xen-command-line.markdown
index a2e427cb18d27c5295774f3387a3d3b3e6d54be3..a565c1b4b769ffad508733cee87695175e8abd0d 100644 (file)
--- a/docs/misc/xen-command-line.markdown
+++ b/docs/misc/xen-command-line.markdown
@@ -705,19 +705,28 @@ virtualization, to allow the L1 hypervisor to use EPT even if the L0 hypervisor
 does not provide VM\_ENTRY\_LOAD\_GUEST\_PAT.
 
 ### ept (Intel)
-> `= List of ( pml<boolean> )`
+> `= List of ( {no-}pml | {no-}ad )`
+
+Controls EPT related features.
+
+> Sub-options:
+
+> `pml`
 
 > Default: `false`
 
-Controls EPT related features. Currently only Page Modification Logging (PML) is
-the controllable feature as boolean type.
+>> PML is a new hardware feature in Intel's Broadwell Server and further
+>> platforms which reduces hypervisor overhead of log-dirty mechanism by
+>> automatically recording GPAs (guest physical addresses) when guest memory
+>> gets dirty, and therefore significantly reducing number of EPT violation
+>> caused by write protection of guest memory, which is a necessity to
+>> implement log-dirty mechanism before PML.
+
+> `ad`
+
+> Default: Hardware dependent
 
-PML is a new hardware feature in Intel's Broadwell Server and further platforms
-which reduces hypervisor overhead of log-dirty mechanism by automatically
-recording GPAs (guest physical addresses) when guest memory gets dirty, and
-therefore significantly reducing number of EPT violation caused by write
-protection of guest memory, which is a necessity to implement log-dirty
-mechanism before PML.
+>> Have hardware keep accessed/dirty (A/D) bits updated.
 
 ### gdb
 > `= <baud>[/<clock_hz>][,DPS[,<io-base>[,<irq>[,<port-bdf>[,<bridge-bdf>]]]] | pci | amt ] `

diff --git a/xen/arch/x86/hvm/vmx/vmcs.c b/xen/arch/x86/hvm/vmx/vmcs.c
index 15b136b5c459305fd4c8f0f8a753aa0f7dd910ac..3592a88847fb22d4e6d65dd68f576f5679512b1d 100644 (file)
--- a/xen/arch/x86/hvm/vmx/vmcs.c
+++ b/xen/arch/x86/hvm/vmx/vmcs.c
@@ -64,12 +64,14 @@ static unsigned int __read_mostly ple_window = 4096;
 integer_param("ple_window", ple_window);
 
 static bool_t __read_mostly opt_pml_enabled = 0;
+static s8 __read_mostly opt_ept_ad = -1;
 
 /*
  * The 'ept' parameter controls functionalities that depend on, or impact the
  * EPT mechanism. Optional comma separated value may contain:
  *
  *  pml                 Enable PML
+ *  ad                  Use A/D bits
  */
 static void __init parse_ept_param(char *s)
 {
@@ -87,6 +89,8 @@ static void __init parse_ept_param(char *s)
 
         if ( !strcmp(s, "pml") )
             opt_pml_enabled = val;
+        else if ( !strcmp(s, "ad") )
+            opt_ept_ad = val;
 
         s = ss + 1;
     } while ( ss );
@@ -268,6 +272,13 @@ static int vmx_init_vmcs_config(void)
     {
         rdmsrl(MSR_IA32_VMX_EPT_VPID_CAP, _vmx_ept_vpid_cap);
 
+        if ( !opt_ept_ad )
+            _vmx_ept_vpid_cap &= ~VMX_EPT_AD_BIT;
+        else if ( /* Work around Erratum AVR41 on Avoton processors. */
+                  boot_cpu_data.x86 == 6 && boot_cpu_data.x86_model == 0x4d &&
+                  opt_ept_ad < 0 )
+            _vmx_ept_vpid_cap &= ~VMX_EPT_AD_BIT;
+
         /*
          * Additional sanity checking before using EPT:
          * 1) the CPU we are running on must support EPT WB, as we will set

diff --git a/xen/arch/x86/mm/p2m-ept.c b/xen/arch/x86/mm/p2m-ept.c
index 0d87ebcf9b8d2593c52cbf05df50dd6a97e2f1cd..74ce9e0262d8ebefb0ac0c3ed112caa41d5b8202 100644 (file)
--- a/xen/arch/x86/mm/p2m-ept.c
+++ b/xen/arch/x86/mm/p2m-ept.c
@@ -130,14 +130,14 @@ static void ept_p2m_type_to_flags(struct p2m_domain *p2m, ept_entry_t *entry,
             break;
         case p2m_ram_rw:
             entry->r = entry->w = entry->x = 1;
-            entry->a = entry->d = 1;
+            entry->a = entry->d = !!cpu_has_vmx_ept_ad;
             break;
         case p2m_mmio_direct:
             entry->r = entry->x = 1;
             entry->w = !rangeset_contains_singleton(mmio_ro_ranges,
                                                     entry->mfn);
-            entry->a = 1;
-            entry->d = entry->w;
+            entry->a = !!cpu_has_vmx_ept_ad;
+            entry->d = entry->w && cpu_has_vmx_ept_ad;
             break;
         case p2m_ram_logdirty:
             entry->r = entry->x = 1;
@@ -152,7 +152,7 @@ static void ept_p2m_type_to_flags(struct p2m_domain *p2m, ept_entry_t *entry,
                 entry->w = 1;
             else
                 entry->w = 0;
-            entry->a = 1;
+            entry->a = !!cpu_has_vmx_ept_ad;
             /* For both PML or non-PML cases we clear D bit anyway */
             entry->d = 0;
             break;
@@ -160,20 +160,20 @@ static void ept_p2m_type_to_flags(struct p2m_domain *p2m, ept_entry_t *entry,
         case p2m_ram_shared:
             entry->r = entry->x = 1;
             entry->w = 0;
-            entry->a = 1;
+            entry->a = !!cpu_has_vmx_ept_ad;
             entry->d = 0;
             break;
         case p2m_grant_map_rw:
         case p2m_map_foreign:
             entry->r = entry->w = 1;
             entry->x = 0;
-            entry->a = entry->d = 1;
+            entry->a = entry->d = !!cpu_has_vmx_ept_ad;
             break;
         case p2m_grant_map_ro:
         case p2m_mmio_write_dm:
             entry->r = 1;
             entry->w = entry->x = 0;
-            entry->a = 1;
+            entry->a = !!cpu_has_vmx_ept_ad;
             entry->d = 0;
             break;
     }
@@ -233,7 +233,7 @@ static int ept_set_middle_entry(struct p2m_domain *p2m, ept_entry_t *ept_entry)
 
     ept_entry->r = ept_entry->w = ept_entry->x = 1;
     /* Manually set A bit to avoid overhead of MMU having to write it later. */
-    ept_entry->a = 1;
+    ept_entry->a = !!cpu_has_vmx_ept_ad;
 
     ept_entry->suppress_ve = 1;
 

diff --git a/xen/include/asm-x86/hvm/vmx/vmx.h b/xen/include/asm-x86/hvm/vmx/vmx.h
index 3fbfa44d58d0755b304ec4453bfee8b2612cffbd..2ed62f95b67e85bc8955a37995b2c01d399cb9d0 100644 (file)
--- a/xen/include/asm-x86/hvm/vmx/vmx.h
+++ b/xen/include/asm-x86/hvm/vmx/vmx.h
@@ -262,6 +262,7 @@ extern uint8_t posted_intr_vector;
     (vmx_ept_vpid_cap & VMX_EPT_SUPERPAGE_1GB)
 #define cpu_has_vmx_ept_2mb                     \
     (vmx_ept_vpid_cap & VMX_EPT_SUPERPAGE_2MB)
+#define cpu_has_vmx_ept_ad (vmx_ept_vpid_cap & VMX_EPT_AD_BIT)
 #define cpu_has_vmx_ept_invept_single_context   \
     (vmx_ept_vpid_cap & VMX_EPT_INVEPT_SINGLE_CONTEXT)