[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [RFC v2 1/6] xen/arm: Save and restore support with hvm context hypercalls



On 04/17/2014 10:06 AM, Julien Grall wrote:
Hello Wei,

On 04/15/2014 10:05 PM, Wei Huang wrote:
+static int hvm_vtimer_load_ctxt(struct domain *d, hvm_domain_context_t *h)
+{
+    int vcpuid;
+    struct hvm_hw_timer ctxt;
+    struct vcpu *v;
+    struct vtimer *t = NULL;
+
+    /* Which vcpu is this? */
+    vcpuid = hvm_load_instance(h);
+
+    if ( vcpuid >= d->max_vcpus || (v = d->vcpu[vcpuid]) == NULL )
+    {
+        dprintk(XENLOG_G_ERR, "HVM restore: dom%u has no vcpu%u\n",
+                d->domain_id, vcpuid);
+        return -EINVAL;
+    }
+
+    if ( hvm_load_entry(TIMER, h, &ctxt) != 0 )
+        return -EINVAL;
+
+    if ( ctxt.type == TIMER_TYPE_VIRT )
+    {
+        t = &v->arch.virt_timer;
+        d->arch.virt_timer_base.offset = ctxt.vtb_offset;
+    }
+    else
+    {
+        t = &v->arch.phys_timer;
+        d->arch.phys_timer_base.offset = ctxt.vtb_offset;

I thought a bit more about the {phys,virt}_timer_base.offset.

When you are migrating a guest, this offset will be invalidated. This
offset is used to get a relative offset from the Xen timer counter.

Agreed.
That also made me think the context switch in Xen for the timer looks
wrong to me.

When a guest VCPU is context switch, the Xen timer counter continue to
run. But not CVAL, so the timer_base.offset will drift a bit. It will
result by setting a wrong timer via set_timer in Xen.

I need to examine the code more carefully. But by skimming through vtimer.c, it looks like this is the case: the missing ticks were not compensated.
Did I miss something?



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.