Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On Thu, Sep 29, 2016 at 11:15 AM, Eric Dumazet wrote: > On Thu, 2016-09-29 at 07:51 -0700, Eric Dumazet wrote: >> On Thu, 2016-09-29 at 10:08 -0400, Tom Herbert wrote: >> >> > It addresses the issue that Rick Jones pointed out was happening with >> > XPS. When packets are

Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On 09/29/2016 06:18 AM, Eric Dumazet wrote: Well, then what this patch series is solving ? You have a producer of packets running on 8 vcpus in a VM. Packets are exiting the VM and need to be queued on a mq NIC in the hypervisor. Flow X can be scheduled on any of these 8 vcpus, so XPS is

Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On Thu, 2016-09-29 at 07:51 -0700, Eric Dumazet wrote: > On Thu, 2016-09-29 at 10:08 -0400, Tom Herbert wrote: > > > It addresses the issue that Rick Jones pointed out was happening with > > XPS. When packets are sent for a flow that has no socket and XPS is > > enabled then each packet uses the

Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On Thu, 2016-09-29 at 10:08 -0400, Tom Herbert wrote: > It addresses the issue that Rick Jones pointed out was happening with > XPS. When packets are sent for a flow that has no socket and XPS is > enabled then each packet uses the XPS queue based on the running CPU. > Since the thread sending

Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On Thu, Sep 29, 2016 at 9:18 AM, Eric Dumazet wrote: > On Thu, 2016-09-29 at 08:53 -0400, Tom Herbert wrote: >> On Thu, Sep 29, 2016 at 12:54 AM, Eric Dumazet >> wrote: >> > On Wed, 2016-09-28 at 20:54 -0700, Tom Herbert wrote: >> >> xps_flows

Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On Thu, 2016-09-29 at 08:53 -0400, Tom Herbert wrote: > On Thu, Sep 29, 2016 at 12:54 AM, Eric Dumazet wrote: > > On Wed, 2016-09-28 at 20:54 -0700, Tom Herbert wrote: > >> xps_flows maintains a per device flow table that is indexed by the > >> skbuff hash. The table is

Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On Thu, Sep 29, 2016 at 12:54 AM, Eric Dumazet wrote: > On Wed, 2016-09-28 at 20:54 -0700, Tom Herbert wrote: >> xps_flows maintains a per device flow table that is indexed by the >> skbuff hash. The table is only consulted when there is no queue saved in >> a transmit

Re: [PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

On Wed, 2016-09-28 at 20:54 -0700, Tom Herbert wrote: > xps_flows maintains a per device flow table that is indexed by the > skbuff hash. The table is only consulted when there is no queue saved in > a transmit socket for an skbuff. > > Each entry in the flow table contains a queue index and a

[PATCH v2 net-next 4/5] xps_flows: XPS for packets that don't have a socket

xps_flows maintains a per device flow table that is indexed by the skbuff hash. The table is only consulted when there is no queue saved in a transmit socket for an skbuff. Each entry in the flow table contains a queue index and a queue pointer. The queue pointer is set when a queue is chosen