Displaying 20 results from an estimated 46 matches for "xdp1".
Did you mean:
xdp
2018 Dec 14
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...irectly.
> > >
> > > Note, this was only done when device IOTLB is not enabled. We could
> > > use similar method to optimize it in the future.
> > >
> > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > >
> > > Before: ~5.0Mpps
> > > After: ~6.1Mpps
> > >
> > > Signed-off-by: Jason Wang<jasowang at redhat.com>
> > > ---
> > > drivers/vhost/vhost.c | 178 ++++++++++++++++++...
2018 Dec 14
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...irectly.
> > >
> > > Note, this was only done when device IOTLB is not enabled. We could
> > > use similar method to optimize it in the future.
> > >
> > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > >
> > > Before: ~5.0Mpps
> > > After: ~6.1Mpps
> > >
> > > Signed-off-by: Jason Wang<jasowang at redhat.com>
> > > ---
> > > drivers/vhost/vhost.c | 178 ++++++++++++++++++...
2017 Mar 22
2
[PATCH net-next 7/8] vhost_net: try batch dequing from skb array
...tilization and spinlock
> touching for each packet. This patch tries to batch them by calling
> batch dequeuing helpers explicitly on the exported skb array and pass
> the skb back through msg_control for underlayer socket to finish the
> userspace copying.
>
> Tests were done by XDP1:
> - small buffer:
> Before: 1.88Mpps
> After : 2.25Mpps (+19.6%)
> - mergeable buffer:
> Before: 1.83Mpps
> After : 2.10Mpps (+14.7%)
>
> Signed-off-by: Jason Wang <jasowang at redhat.com>
> ---
> drivers/vhost/net.c | 64 ++++++++++++++++++++++++++++++...
2017 Mar 22
2
[PATCH net-next 7/8] vhost_net: try batch dequing from skb array
...tilization and spinlock
> touching for each packet. This patch tries to batch them by calling
> batch dequeuing helpers explicitly on the exported skb array and pass
> the skb back through msg_control for underlayer socket to finish the
> userspace copying.
>
> Tests were done by XDP1:
> - small buffer:
> Before: 1.88Mpps
> After : 2.25Mpps (+19.6%)
> - mergeable buffer:
> Before: 1.83Mpps
> After : 2.10Mpps (+14.7%)
>
> Signed-off-by: Jason Wang <jasowang at redhat.com>
> ---
> drivers/vhost/net.c | 64 ++++++++++++++++++++++++++++++...
2018 Dec 13
5
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...s are modified to use pointers to
> access the metadata directly.
>
> Note, this was only done when device IOTLB is not enabled. We could
> use similar method to optimize it in the future.
>
> Tests shows about ~24% improvement on TX PPS when using virtio-user +
> vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
>
> Before: ~5.0Mpps
> After: ~6.1Mpps
>
> Signed-off-by: Jason Wang <jasowang at redhat.com>
> ---
> drivers/vhost/vhost.c | 178 ++++++++++++++++++++++++++++++++++++++++++
> drivers/vhost/vhost.h | 11 +++
> 2...
2018 Dec 13
5
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...s are modified to use pointers to
> access the metadata directly.
>
> Note, this was only done when device IOTLB is not enabled. We could
> use similar method to optimize it in the future.
>
> Tests shows about ~24% improvement on TX PPS when using virtio-user +
> vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
>
> Before: ~5.0Mpps
> After: ~6.1Mpps
>
> Signed-off-by: Jason Wang <jasowang at redhat.com>
> ---
> drivers/vhost/vhost.c | 178 ++++++++++++++++++++++++++++++++++++++++++
> drivers/vhost/vhost.h | 11 +++
> 2...
2018 Dec 24
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...Note, this was only done when device IOTLB is not enabled. We could
> > > > > use similar method to optimize it in the future.
> > > > >
> > > > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > > > >
> > > > > Before: ~5.0Mpps
> > > > > After: ~6.1Mpps
> > > > >
> > > > > Signed-off-by: Jason Wang<jasowang at redhat.com>
> > > > > ---
>...
2018 Dec 24
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...Note, this was only done when device IOTLB is not enabled. We could
> > > > > use similar method to optimize it in the future.
> > > > >
> > > > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > > > >
> > > > > Before: ~5.0Mpps
> > > > > After: ~6.1Mpps
> > > > >
> > > > > Signed-off-by: Jason Wang<jasowang at redhat.com>
> > > > > ---
>...
2017 Mar 23
0
[PATCH net-next 7/8] vhost_net: try batch dequing from skb array
...>> touching for each packet. This patch tries to batch them by calling
>> batch dequeuing helpers explicitly on the exported skb array and pass
>> the skb back through msg_control for underlayer socket to finish the
>> userspace copying.
>>
>> Tests were done by XDP1:
>> - small buffer:
>> Before: 1.88Mpps
>> After : 2.25Mpps (+19.6%)
>> - mergeable buffer:
>> Before: 1.83Mpps
>> After : 2.10Mpps (+14.7%)
>>
>> Signed-off-by: Jason Wang <jasowang at redhat.com>
>> ---
>> drivers/vho...
2018 Dec 14
0
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...ters to
>> access the metadata directly.
>>
>> Note, this was only done when device IOTLB is not enabled. We could
>> use similar method to optimize it in the future.
>>
>> Tests shows about ~24% improvement on TX PPS when using virtio-user +
>> vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
>>
>> Before: ~5.0Mpps
>> After: ~6.1Mpps
>>
>> Signed-off-by: Jason Wang<jasowang at redhat.com>
>> ---
>> drivers/vhost/vhost.c | 178 ++++++++++++++++++++++++++++++++++++++++++
>> drivers/...
2018 Dec 25
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...LB is not enabled. We could
> > > > > > > use similar method to optimize it in the future.
> > > > > > >
> > > > > > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > > > > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > > > > > >
> > > > > > > Before: ~5.0Mpps
> > > > > > > After: ~6.1Mpps
> > > > > > >
> > > > > > > Signed-off-by: Jason Wang<jasowang...
2018 Dec 25
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...LB is not enabled. We could
> > > > > > > use similar method to optimize it in the future.
> > > > > > >
> > > > > > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > > > > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > > > > > >
> > > > > > > Before: ~5.0Mpps
> > > > > > > After: ~6.1Mpps
> > > > > > >
> > > > > > > Signed-off-by: Jason Wang<jasowang...
2018 Dec 26
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...t; > > > > > > use similar method to optimize it in the future.
> > > > > > > > >
> > > > > > > > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > > > > > > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > > > > > > > >
> > > > > > > > > Before: ~5.0Mpps
> > > > > > > > > After: ~6.1Mpps
> > > > > > > > >
> > > > > > >...
2018 Dec 26
2
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...t; > > > > > > use similar method to optimize it in the future.
> > > > > > > > >
> > > > > > > > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > > > > > > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > > > > > > > >
> > > > > > > > > Before: ~5.0Mpps
> > > > > > > > > After: ~6.1Mpps
> > > > > > > > >
> > > > > > >...
2018 Dec 24
0
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...gt;>>>
>>>> Note, this was only done when device IOTLB is not enabled. We could
>>>> use similar method to optimize it in the future.
>>>>
>>>> Tests shows about ~24% improvement on TX PPS when using virtio-user +
>>>> vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
>>>>
>>>> Before: ~5.0Mpps
>>>> After: ~6.1Mpps
>>>>
>>>> Signed-off-by: Jason Wang<jasowang at redhat.com>
>>>> ---
>>>> drivers/vhost/vhost.c | 178 +++++...
2017 Mar 21
0
[PATCH net-next 7/8] vhost_net: try batch dequing from skb array
...ent because of the bad cache utilization and spinlock
touching for each packet. This patch tries to batch them by calling
batch dequeuing helpers explicitly on the exported skb array and pass
the skb back through msg_control for underlayer socket to finish the
userspace copying.
Tests were done by XDP1:
- small buffer:
Before: 1.88Mpps
After : 2.25Mpps (+19.6%)
- mergeable buffer:
Before: 1.83Mpps
After : 2.10Mpps (+14.7%)
Signed-off-by: Jason Wang <jasowang at redhat.com>
---
drivers/vhost/net.c | 64 +++++++++++++++++++++++++++++++++++++++++++++++++----
1 file changed, 60 insert...
2018 Dec 30
1
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...t; use similar method to optimize it in the future.
> > > > > > > > > > >
> > > > > > > > > > > Tests shows about ~24% improvement on TX PPS when using virtio-user +
> > > > > > > > > > > vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
> > > > > > > > > > >
> > > > > > > > > > > Before: ~5.0Mpps
> > > > > > > > > > > After: ~6.1Mpps
> > > > > > > > > >...
2018 Dec 25
0
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...Note, this was only done when device IOTLB is not enabled. We could
>>>>>> use similar method to optimize it in the future.
>>>>>>
>>>>>> Tests shows about ~24% improvement on TX PPS when using virtio-user +
>>>>>> vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
>>>>>>
>>>>>> Before: ~5.0Mpps
>>>>>> After: ~6.1Mpps
>>>>>>
>>>>>> Signed-off-by: Jason Wang<jasowang at redhat.com>
>>>>>> ---
>&...
2018 Dec 27
0
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...ould
>>>>>>>>>> use similar method to optimize it in the future.
>>>>>>>>>>
>>>>>>>>>> Tests shows about ~24% improvement on TX PPS when using virtio-user +
>>>>>>>>>> vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
>>>>>>>>>>
>>>>>>>>>> Before: ~5.0Mpps
>>>>>>>>>> After: ~6.1Mpps
>>>>>>>>>>
>>>>>>>>>> Signed-off...
2018 Dec 13
0
[PATCH net-next 3/3] vhost: access vq metadata through kernel virtual address
...those mappings and memory accessors are modified to use pointers to
access the metadata directly.
Note, this was only done when device IOTLB is not enabled. We could
use similar method to optimize it in the future.
Tests shows about ~24% improvement on TX PPS when using virtio-user +
vhost_net + xdp1 on TAP (CONFIG_HARDENED_USERCOPY is not enabled):
Before: ~5.0Mpps
After: ~6.1Mpps
Signed-off-by: Jason Wang <jasowang at redhat.com>
---
drivers/vhost/vhost.c | 178 ++++++++++++++++++++++++++++++++++++++++++
drivers/vhost/vhost.h | 11 +++
2 files changed, 189 insertions(+)
diff --git...