Message ID | 1393878833-17667-1-git-send-email-vyasevic@redhat.com |
---|---|
State | Accepted, archived |
Delegated to: | David Miller |
Headers | show |
From: Vlad Yasevich <vyasevic@redhat.com> Date: Mon, 3 Mar 2014 15:33:53 -0500 > Macvlan currently inherits all of its features from the lower > device. When lower device disables offload support, this causes > macvlan to disable offload support as well. This causes > performance regression when using macvlan/macvtap in bridge > mode. > > It can be easily demonstrated by creating 2 namespaces using > macvlan in bridge mode and running netperf between them: > > MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 10.0.0.1 () port 0 AF_INET > Recv Send Send > Socket Socket Message Elapsed > Size Size Size Time Throughput > bytes bytes bytes secs. 10^6bits/sec > > 87380 16384 16384 20.00 1204.61 > > To restore the performance, we add software offload features > to the list of "always_on" features for macvlan. This way > when a namespace or a guest using macvtap initially sends a > packet, this packet will not be segmented at macvlan level. > It will only be segmented when macvlan sends the packet > to the lower device. > > MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 10.0.0.1 () port 0 AF_INET > Recv Send Send > Socket Socket Message Elapsed > Size Size Size Time Throughput > bytes bytes bytes secs. 10^6bits/sec > > 87380 16384 16384 20.00 5507.35 > > Fixes: 6acf54f1cf0a6747bac9fea26f34cfc5a9029523 (macvtap: Add support of packet capture on macvtap device.) > Fixes: 797f87f83b60685ff8a13fa0572d2f10393c50d3 (macvlan: fix netdev feature propagation from lower device) > CC: Florian Westphal <fw@strlen.de> > CC: Christian Borntraeger <borntraeger@de.ibm.com> > CC: Jason Wang <jasowang@redhat.com> > CC: Michael S. Tsirkin <mst@redhat.com> > Tested-by: Christian Borntraeger <borntraeger@de.ibm.com> > Signed-off-by: Vlad Yasevich <vyasevic@redhat.com> Applied, thanks Vlad. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 03/03/2014 12:33 PM, Vlad Yasevich wrote: > Macvlan currently inherits all of its features from the lower > device. When lower device disables offload support, this causes > macvlan to disable offload support as well. This causes > performance regression when using macvlan/macvtap in bridge > mode. > > It can be easily demonstrated by creating 2 namespaces using > macvlan in bridge mode and running netperf between them: > > MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 10.0.0.1 () port 0 AF_INET > Recv Send Send > Socket Socket Message Elapsed > Size Size Size Time Throughput > bytes bytes bytes secs. 10^6bits/sec > > 87380 16384 16384 20.00 1204.61 > > To restore the performance, we add software offload features > to the list of "always_on" features for macvlan. This way > when a namespace or a guest using macvtap initially sends a > packet, this packet will not be segmented at macvlan level. > It will only be segmented when macvlan sends the packet > to the lower device. If users specifically disable the offload features on the macvlan using ethtool, will they be turned off? If not, then I think that logic should be kept somehow? Thanks, Ben
On 03/03/2014 06:49 PM, Ben Greear wrote: > On 03/03/2014 12:33 PM, Vlad Yasevich wrote: >> Macvlan currently inherits all of its features from the lower >> device. When lower device disables offload support, this causes >> macvlan to disable offload support as well. This causes >> performance regression when using macvlan/macvtap in bridge >> mode. >> >> It can be easily demonstrated by creating 2 namespaces using >> macvlan in bridge mode and running netperf between them: >> >> MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 10.0.0.1 () port 0 AF_INET >> Recv Send Send >> Socket Socket Message Elapsed >> Size Size Size Time Throughput >> bytes bytes bytes secs. 10^6bits/sec >> >> 87380 16384 16384 20.00 1204.61 >> >> To restore the performance, we add software offload features >> to the list of "always_on" features for macvlan. This way >> when a namespace or a guest using macvtap initially sends a >> packet, this packet will not be segmented at macvlan level. >> It will only be segmented when macvlan sends the packet >> to the lower device. > > If users specifically disable the offload features on the macvlan using ethtool, > will they be turned off? If not, then I think that logic should > be kept somehow? Just looked and offloads on macvlan are fixed even before this patch, so a user can't disable them. Looks like the only way to disable them prior to this patch was to change the features of the lower device. However, that introduces an issue when running in bridged mode that this patch fixes. -vlad > > Thanks, > Ben > -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/net/macvlan.c b/drivers/net/macvlan.c index a5d2189..2625d2a 100644 --- a/drivers/net/macvlan.c +++ b/drivers/net/macvlan.c @@ -506,6 +506,9 @@ static int macvlan_change_mtu(struct net_device *dev, int new_mtu) static struct lock_class_key macvlan_netdev_xmit_lock_key; static struct lock_class_key macvlan_netdev_addr_lock_key; +#define ALWAYS_ON_FEATURES \ + (NETIF_F_SG | NETIF_F_GEN_CSUM | NETIF_F_GSO_SOFTWARE | NETIF_F_LLTX) + #define MACVLAN_FEATURES \ (NETIF_F_SG | NETIF_F_ALL_CSUM | NETIF_F_HIGHDMA | NETIF_F_FRAGLIST | \ NETIF_F_GSO | NETIF_F_TSO | NETIF_F_UFO | NETIF_F_GSO_ROBUST | \ @@ -539,7 +542,7 @@ static int macvlan_init(struct net_device *dev) dev->state = (dev->state & ~MACVLAN_STATE_MASK) | (lowerdev->state & MACVLAN_STATE_MASK); dev->features = lowerdev->features & MACVLAN_FEATURES; - dev->features |= NETIF_F_LLTX; + dev->features |= ALWAYS_ON_FEATURES; dev->gso_max_size = lowerdev->gso_max_size; dev->iflink = lowerdev->ifindex; dev->hard_header_len = lowerdev->hard_header_len; @@ -699,7 +702,7 @@ static netdev_features_t macvlan_fix_features(struct net_device *dev, features = netdev_increment_features(vlan->lowerdev->features, features, mask); - features |= NETIF_F_LLTX; + features |= ALWAYS_ON_FEATURES; return features; }