>>> On 12.05.15 at 14:37, wrote:
> On 12/05/15 12:37, Jan Beulich wrote:
> On 12.05.15 at 13:01, wrote:
>>> On 05/05/15 13:34, Jan Beulich wrote:
>>> On 30.04.15 at 15:28, wrote:
> @@ -1430,6 +1456,17 @@ gnttab_setup_table(
> gt = d->grant_table;
> write_lock(>->loc
On 12/05/15 12:37, Jan Beulich wrote:
On 12.05.15 at 13:01, wrote:
>> On 05/05/15 13:34, Jan Beulich wrote:
>> On 30.04.15 at 15:28, wrote:
@@ -1430,6 +1456,17 @@ gnttab_setup_table(
gt = d->grant_table;
write_lock(>->lock);
+/* Tracking of mapped
>>> On 12.05.15 at 13:01, wrote:
> On 05/05/15 13:34, Jan Beulich wrote:
> On 30.04.15 at 15:28, wrote:
>>> @@ -1430,6 +1456,17 @@ gnttab_setup_table(
>>> gt = d->grant_table;
>>> write_lock(>->lock);
>>>
>>> +/* Tracking of mapped foreign frames table */
>>> +if ( (gt->ma
On 05/05/15 13:34, Jan Beulich wrote:
On 30.04.15 at 15:28, wrote:
>> 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 hand
>>> On 30.04.15 at 15:28, wrote:
> 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
On 30/04/15 16:12, Tim Deegan wrote:
> At 14:28 +0100 on 30 Apr (1430404125), David Vrabel wrote:
>> 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 obtainin
At 14:28 +0100 on 30 Apr (1430404125), David Vrabel wrote:
> 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
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