Re: [PATCH Remus v1 5/8] tools/libxc: defer the setting of HVM_PARAM_IDENT_PT

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 07/05/15 14:59, Hongyang Yang wrote:
>
>
> On 05/07/2015 06:35 PM, Andrew Cooper wrote:
>> On 07/05/15 07:37, Yang Hongyang wrote:
>>> Set the hvm param HVM_PARAM_IDENT_PT second time will fail.
>>> So defer the setting of HVM_PARAM_IDENT_PT.
>>>
>>> Signed-off-by: Yang Hongyang <yanghy@cn.fujitsu.com>
>>
>> I would argue that this is a Xen bug, not a migration bug.  There is
>> nothing conceptually wrong with setting this param multiple times.
>
> I noticed that in legacy migration, this was set multiple times, so
> this patch is only a workaround that will avoid the fail in my test.
> will try your hypervisor patch tomorrow.

Try this version, which has been rebased over Pauls' cleanup in this
area which has just been committed to staging.

~Andrew

diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 371fd33..941e9ea 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -5710,12 +5710,13 @@ static int hvmop_set_param(
             rc = -EINVAL;
         break;
     case HVM_PARAM_IDENT_PT:
-        rc = -EINVAL;
-        if ( d->arch.hvm_domain.params[a.index] != 0 )
-            break;
-
         rc = 0;
-        if ( !paging_mode_hap(d) )
+        d->arch.hvm_domain.params[a.index] = a.value;
+        /*
+         * Only actually required for VT-x lacking unrestricted_guest
+         * capabilities.  Short circuit the pause if possible.
+         */
+        if ( !paging_mode_hap(d) || !cpu_has_vmx )
             break;
 
         /*
@@ -5729,7 +5730,6 @@ static int hvmop_set_param(
 
         rc = 0;
         domain_pause(d);
-        d->arch.hvm_domain.params[a.index] = a.value;
         for_each_vcpu ( d, v )
             paging_update_cr3(v);
         domain_unpause(d);