[IA64] Handle speculative vhpt walk

Since processor may support speculative VHPT walk,
The long format VHPT head entry needs to be disabled
before programming it.

Signed-off-by: Anthony Xu <anthony.xu@intel.com>

diff --git a/xen/arch/ia64/vmx/vmx_ivt.S b/xen/arch/ia64/vmx/vmx_ivt.S
index de2b0ba9eb796a83d02fa24f3b4b7149d7107d03..6c79cd53e544cd82304a39a9f772ce451ba5233d 100644 (file)
--- a/xen/arch/ia64/vmx/vmx_ivt.S
+++ b/xen/arch/ia64/vmx/vmx_ivt.S
@@ -168,11 +168,11 @@ vmx_itlb_loop:
     adds r16 = VLE_TITAG_OFFSET, r17
     adds r19 = VLE_CCHAIN_OFFSET, r17
     ;;
-    ld8 r22 = [r16]
+    ld8 r24 = [r16]
     ld8 r23 = [r19]
     ;;
     lfetch [r23]
-    cmp.eq  p6,p7 = r20, r22
+    cmp.eq  p6,p7 = r20, r24
     ;;
 (p7)mov r17 = r23;
 (p7)br.sptk vmx_itlb_loop
@@ -180,10 +180,12 @@ vmx_itlb_loop:
     ld8 r25 = [r17]
     ld8 r27 = [r18]
     ld8 r29 = [r28]
+    dep r22 = -1,r24,63,1    //set ti=1
     ;;
     st8 [r16] = r29, VLE_ITIR_OFFSET - VLE_TITAG_OFFSET
     st8 [r28] = r22, VLE_ITIR_OFFSET - VLE_TITAG_OFFSET
     extr.u r19 = r27, 56, 4
+    mf
     ;;
     ld8 r29 = [r16]
     ld8 r22 = [r28]
@@ -191,10 +193,11 @@ vmx_itlb_loop:
     dep r25 = r19, r25, 56, 4
     ;;
     st8 [r16] = r22
-    st8 [r28] = r29
+    st8 [r28] = r29, VLE_TITAG_OFFSET - VLE_ITIR_OFFSET
     st8 [r18] = r25
     st8 [r17] = r27
     ;;
+    st8.rel [r28] = r24
     itc.i r25
     dv_serialize_data
     mov r17=cr.isr
@@ -246,11 +249,11 @@ vmx_dtlb_loop:
     adds r16 = VLE_TITAG_OFFSET, r17
     adds r19 = VLE_CCHAIN_OFFSET, r17
     ;;
-    ld8 r22 = [r16]
+    ld8 r24 = [r16]
     ld8 r23 = [r19]
     ;;
     lfetch [r23]
-    cmp.eq  p6,p7 = r20, r22
+    cmp.eq  p6,p7 = r20, r24
     ;;
 (p7)mov r17 = r23;
 (p7)br.sptk vmx_dtlb_loop
@@ -258,10 +261,12 @@ vmx_dtlb_loop:
     ld8 r25 = [r17]
     ld8 r27 = [r18]
     ld8 r29 = [r28]
+    dep r22 = -1,r24,63,1    //set ti=1
     ;;
     st8 [r16] = r29, VLE_ITIR_OFFSET - VLE_TITAG_OFFSET
     st8 [r28] = r22, VLE_ITIR_OFFSET - VLE_TITAG_OFFSET
     extr.u r19 = r27, 56, 4
+    mf
     ;;
     ld8 r29 = [r16]
     ld8 r22 = [r28]
@@ -269,10 +274,11 @@ vmx_dtlb_loop:
     dep r25 = r19, r25, 56, 4
     ;;
     st8 [r16] = r22
-    st8 [r28] = r29
+    st8 [r28] = r29, VLE_TITAG_OFFSET - VLE_ITIR_OFFSET
     st8 [r18] = r25
     st8 [r17] = r27
-    ;;    
+    ;;
+    st8.rel [r28] = r24 
     itc.d r25
     dv_serialize_data
     mov r17=cr.isr

diff --git a/xen/arch/ia64/vmx/vtlb.c b/xen/arch/ia64/vmx/vtlb.c
index d9efe5dad14a9eb51d4264f9111ba703dbbd0b42..f6099197558a5fb8fd62965df40158ca99599a4b 100644 (file)
--- a/xen/arch/ia64/vmx/vtlb.c
+++ b/xen/arch/ia64/vmx/vtlb.c
@@ -141,7 +141,7 @@ static void thash_recycle_cch(thash_cb_t *hcb, thash_data_t *hash)
 
 static void vmx_vhpt_insert(thash_cb_t *hcb, u64 pte, u64 itir, u64 ifa)
 {
-    u64 tag ,len;
+    u64 tag;
     ia64_rr rr;
     thash_data_t *head, *cch;
     pte = pte & ~PAGE_FLAGS_RV_MASK;
@@ -155,14 +155,12 @@ static void vmx_vhpt_insert(thash_cb_t *hcb, u64 pte, u64 itir, u64 ifa)
         cch = cch->next;
     }
     if (cch) {
-        if (cch == head) {
-            len = head->len;
-        } else {
+        if (cch != head) {
             local_irq_disable();
             cch->page_flags = head->page_flags;
             cch->itir = head->itir;
             cch->etag  = head->etag;
-            len = head->len;
+            head->ti = 1;
             local_irq_enable();
         }
     }
@@ -175,16 +173,17 @@ static void vmx_vhpt_insert(thash_cb_t *hcb, u64 pte, u64 itir, u64 ifa)
         }
         local_irq_disable();
         *cch = *head;
+        head->ti = 1;
         head->next = cch;
-        len = cch->len+1;
+        head->len = cch->len + 1;
         cch->len = 0;
         local_irq_enable();
     }
-
+    //here head is invalid
+    wmb();
     head->page_flags=pte;
-    head->len = len;
     head->itir = rr.ps << 2;
-    head->etag=tag;
+    *(volatile unsigned long*)&head->etag = tag;
     return;
 }
 

diff --git a/xen/arch/ia64/xen/vhpt.c b/xen/arch/ia64/xen/vhpt.c
index 07d1810ee8a7a97e1d70d6180c3929334e7360f5..55eed6e5f7376fe9ba93887e161e60ab075686c5 100644 (file)
--- a/xen/arch/ia64/xen/vhpt.c
+++ b/xen/arch/ia64/xen/vhpt.c
@@ -78,11 +78,13 @@ void vhpt_insert (unsigned long vadr, unsigned long pte, unsigned long logps)
        struct vhpt_lf_entry *vlfe = (struct vhpt_lf_entry *)ia64_thash(vadr);
        unsigned long tag = ia64_ttag (vadr);
 
-       /* No need to first disable the entry, since VHPT is per LP
-          and VHPT is TR mapped.  */
+       /* Even though VHPT is per VCPU, still need to first disable the entry,
+        * because the processor may support speculative VHPT walk.  */
+       vlfe->ti_tag = INVALID_TI_TAG;
+       wmb();
        vlfe->itir = logps;
        vlfe->page_flags = pte | _PAGE_P;
-       vlfe->ti_tag = tag;
+       *(volatile unsigned long*)&vlfe->ti_tag = tag;
 }
 
 void vhpt_multiple_insert(unsigned long vaddr, unsigned long pte, unsigned long logps)