[Patch][SVM]Disable intercepting CR3 writes when nested paging is enabled

[Wei Huang <wei.huang2@amd.com>]

[-- Attachment #1: Type: text/plain, Size: 220 bytes --]

This patch disables intercepting CR3 writes when nested paging is 
enabled. For applications which cause excessive CR3 accesses, the patch 
can increases their performance.

Signed-off-by: Wei Huang <wei.huang2@amd.com>

[-- Attachment #2: npt_disable_intercept_cr3_write.txt --]
[-- Type: text/plain, Size: 1294 bytes --]

diff -r c31edc59ae5d xen/arch/x86/hvm/svm/svm.c
--- a/xen/arch/x86/hvm/svm/svm.c	Sat Jul 25 19:31:35 2009 -0500
+++ b/xen/arch/x86/hvm/svm/svm.c	Sat Jul 25 19:31:51 2009 -0500
@@ -1299,6 +1299,8 @@ asmlinkage void svm_vmexit_handler(struc
     eventinj_t eventinj;
     int inst_len, rc;
 
+    if ( paging_mode_hap(v->domain) )
+        v->arch.hvm_vcpu.guest_cr[3] = v->arch.hvm_vcpu.hw_cr[3] = vmcb->cr3;
     /*
      * Before doing anything else, we need to sync up the VLAPIC's TPR with
      * SVM's vTPR. It's OK if the guest doesn't touch CR8 (e.g. 32-bit Windows)
diff -r c31edc59ae5d xen/arch/x86/hvm/svm/vmcb.c
--- a/xen/arch/x86/hvm/svm/vmcb.c	Sat Jul 25 19:31:35 2009 -0500
+++ b/xen/arch/x86/hvm/svm/vmcb.c	Sat Jul 25 19:31:51 2009 -0500
@@ -236,10 +236,10 @@ static int construct_vmcb(struct vcpu *v
         vmcb->h_cr3 = pagetable_get_paddr(v->domain->arch.phys_table);
 
         /*
-         * No point in intercepting CR3 reads, because the hardware will return
-         * the guest version anyway.
+         * No point in intercepting CR3 reads/writes
          */
         vmcb->cr_intercepts &= ~CR_INTERCEPT_CR3_READ;
+        vmcb->cr_intercepts &= ~CR_INTERCEPT_CR3_WRITE;
 
         /*
          * No point in intercepting INVLPG if we don't have shadow pagetables

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel