projects / GitHub / moto-9609 / android_kernel_motorola_exynos9610.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: e980a07) | patch
macvtap: avoid hash calculating for single queue
authorJason Wang <jasowang@redhat.com>
Fri, 15 Jul 2016 07:46:30 +0000 (03:46 -0400)
committerDavid S. Miller <davem@davemloft.net>
Fri, 15 Jul 2016 21:40:39 +0000 (14:40 -0700)
commit1b16bf42d154c8fbbab2cccc419e2ba47d700849
treececa5157fc452af785e30c34259478f3038bc475tree | snapshot (tar.gz zip)
parente980a0764a16897294207eaf4556c3038768adb9commit | diff
macvtap: avoid hash calculating for single queue

We decide the rxq through calculating its hash which is not necessary
if we only have one rx queue. So this patch skip this and just return
queue 0. Test shows 22% improving on guest rx pps.

Before: 1201504 pkts/s
After:  1472731 pkts/s

Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/macvtap.c diff | blob | blame | history