diff mbox

[4/6] x86/HVM: limit flushing on cache attribute pinning adjustments

Message ID 56D8766F02000078000D8F4E@prv-mh.provo.novell.com (mailing list archive)
State New, archived
Headers show

Commit Message

Jan Beulich March 3, 2016, 4:37 p.m. UTC 
Avoid cache flush on EPT when removing a UC- range, since when used
this type gets converted to UC anyway (there's no UC- among the types
valid in MTRRs and hence EPT's emt field).

We might further wwant to consider only forcing write buffer flushes
when removing WC ranges.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
x86/HVM: limit flushing on cache attribute pinning adjustments

Avoid cache flush on EPT when removing a UC- range, since when used
this type gets converted to UC anyway (there's no UC- among the types
valid in MTRRs and hence EPT's emt field).

We might further wwant to consider only forcing write buffer flushes
when removing WC ranges.

Signed-off-by: Jan Beulich <jbeulich@suse.com>

--- a/xen/arch/x86/hvm/mtrr.c
+++ b/xen/arch/x86/hvm/mtrr.c
@@ -611,8 +611,21 @@ int32_t hvm_set_mem_pinned_cacheattr(
                 type = range->type;
                 call_rcu(&range->rcu, free_pinned_cacheattr_entry);
                 p2m_memory_type_changed(d);
-                if ( type != PAT_TYPE_UNCACHABLE )
+                switch ( type )
+                {
+                case PAT_TYPE_UC_MINUS:
+                    /*
+                     * For EPT we can also avoid the flush in this case;
+                     * see epte_get_entry_emt().
+                     */
+                    if ( hap_enabled(d) && cpu_has_vmx )
+                case PAT_TYPE_UNCACHABLE:
+                        break;
+                    /* fall through */
+                default:
                     flush_all(FLUSH_CACHE);
+                    break;
+                }
                 return 0;
             }
         rcu_read_unlock(&pinned_cacheattr_rcu_lock);

Comments

Andrew Cooper March 3, 2016, 5:38 p.m. UTC | #1 
On 03/03/16 16:37, Jan Beulich wrote:
> Avoid cache flush on EPT when removing a UC- range, since when used
> this type gets converted to UC anyway (there's no UC- among the types
> valid in MTRRs and hence EPT's emt field).
>
> We might further wwant to consider only forcing write buffer flushes
> when removing WC ranges.
>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
diff mbox

Patch

--- a/xen/arch/x86/hvm/mtrr.c
+++ b/xen/arch/x86/hvm/mtrr.c
@@ -611,8 +611,21 @@  int32_t hvm_set_mem_pinned_cacheattr(
                 type = range->type;
                 call_rcu(&range->rcu, free_pinned_cacheattr_entry);
                 p2m_memory_type_changed(d);
-                if ( type != PAT_TYPE_UNCACHABLE )
+                switch ( type )
+                {
+                case PAT_TYPE_UC_MINUS:
+                    /*
+                     * For EPT we can also avoid the flush in this case;
+                     * see epte_get_entry_emt().
+                     */
+                    if ( hap_enabled(d) && cpu_has_vmx )
+                case PAT_TYPE_UNCACHABLE:
+                        break;
+                    /* fall through */
+                default:
                     flush_all(FLUSH_CACHE);
+                    break;
+                }
                 return 0;
             }
         rcu_read_unlock(&pinned_cacheattr_rcu_lock);