>>> On 24.04.15 at 12:21, wrote:
> On 24/04/15 10:50, Jan Beulich wrote:
> On 24.04.15 at 11:09, wrote:
>>> On 23/04/15 17:11, Jan Beulich wrote:
>>> On 22.04.15 at 18:00, wrote:
> --- a/xen/include/xen/grant_table.h
> +++ b/xen/include/xen/grant_table.h
> @@ -60,6 +60,8 @@ s
On 24/04/15 10:50, Jan Beulich wrote:
On 24.04.15 at 11:09, wrote:
>> On 23/04/15 17:11, Jan Beulich wrote:
>> On 22.04.15 at 18:00, wrote:
--- a/xen/include/xen/grant_table.h
+++ b/xen/include/xen/grant_table.h
@@ -60,6 +60,8 @@ struct grant_mapping {
u32 r
On 24/04/15 10:50, Jan Beulich wrote:
On 24.04.15 at 11:09, wrote:
>> On 23/04/15 17:11, Jan Beulich wrote:
>> On 22.04.15 at 18:00, wrote:
--- a/xen/include/xen/grant_table.h
+++ b/xen/include/xen/grant_table.h
@@ -60,6 +60,8 @@ struct grant_mapping {
u32 r
>>> On 24.04.15 at 11:09, wrote:
> On 23/04/15 17:11, Jan Beulich wrote:
> On 22.04.15 at 18:00, wrote:
>>> --- a/xen/include/xen/grant_table.h
>>> +++ b/xen/include/xen/grant_table.h
>>> @@ -60,6 +60,8 @@ struct grant_mapping {
>>> u32 ref; /* grant ref */
>>> u16
On 23/04/15 17:11, Jan Beulich wrote:
On 22.04.15 at 18:00, wrote:
>> static inline int
>> get_maptrack_handle(
>> struct grant_table *lgt)
>> {
>> +struct vcpu *v = current;
>> int i;
>> grant_handle_thandle;
>> struct grant_mappi
>>> On 23.04.15 at 18:29, wrote:
> On 23/04/15 17:11, Jan Beulich wrote:
> On 22.04.15 at 18:00, wrote:
>>> -return handle;
>>> +v->maptrack_limit = new_mt_limit;
>>> +
>>> +return __get_maptrack_handle(lgt, v);
>>
>> With the lock dropped, nothing guarantees this to succeed, whi
On 23/04/15 17:11, Jan Beulich wrote:
On 22.04.15 at 18:00, wrote:
>>
>> +v->maptrack_head = lgt->maptrack_pages * MAPTRACK_PER_PAGE;
>> +if (v->maptrack_tail == MAPTRACK_TAIL)
>> +{
>> +v->maptrack_tail = (lgt->maptrack_pages * MAPTRACK_PER_PAGE)
>> ++ MAPTR
>>> On 22.04.15 at 18:00, wrote:
> static inline int
> get_maptrack_handle(
> struct grant_table *lgt)
> {
> +struct vcpu *v = current;
> int i;
> grant_handle_thandle;
> struct grant_mapping *new_mt;
> unsigned int new_mt
From: Malcolm Crossley
Performance analysis of aggregate network throughput with many VMs
shows that performance is signficantly limited by contention on the
maptrack lock when obtaining/releasing maptrack handles from the free
list.
Instead of a single free list use a per-VCPU list. This avoids
