Seems to me that the issue is not that the MTU isn't being honoured, but that the MTU should be checked *before* TSO assembly and not *after*. Assembly should happen outside the VM if the VM has enabled it on the interface internally, and clearly the incoming segments (not packets, at this point) may be longer than the MTU.
-- You received this bug notification because you are a member of qemu- devel-ml, which is subscribed to QEMU. https://bugs.launchpad.net/bugs/1297487 Title: MTU not honored in virtio vnet Status in QEMU: New Status in Salt: New Status in “qemu-kvm” package in Ubuntu: Confirmed Bug description: I am observing a potential regression/different behavior between rel. 14.04 (dev branch) and release 13.04. My hardware is: Cisco UCS blade B200-M3 and the network adapter card: Cisco UCS VIC 1240. lsb_release -rd Description: Ubuntu Trusty Tahr (development branch) Release: 14.04 uname -a Linux konan2 3.13.0-19-generic #40-Ubuntu SMP Mon Mar 24 02:36:06 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux The problem: After starting a kvm with virtio interfaces I am passing HTTP traffic via an external network traffic simulator. The tool is sending a TCP packet of 3481B, because the tool MTU is set to 1400B, it splits the packets into 3 TCP segments. When the 3 segments are received at the host eth1 interface, the host (ubuntu 14.04) reassembles the TCP packets into a larger packet (GRO), then passes the packet up on vnet1. At this point, because vnet1 MTU is 1500B, it is supposed to re-segment the packet and pass the 3 segments up to the VM. But it passes the big 3481B packet instead. This behavior did not happen when I tried the same scenario in release 13.04 I can disable this behavior by disabling TSO (TCP segment offloading in the vnet), but I did not have to do this in rel. 13.04 and I feel the MTU is not honored as it should be with rel. 14.04. ip link show | grep eth1 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq master br1 state UP mode DEFAULT group default qlen 1000 ip link show | grep vnet1 16: vnet1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master br1 state UNKNOWN mode DEFAULT group default qlen 500 I am attaching two tcpdump/pcap traces that show a TCP transaction passing on vnet1 when TSO is on and when TSO is off. Please see: - vnet1_tso_on.pcap - vnet1_tso_off.pcap in attachment. I noticed there was a driver upgrade in rel. 14.04: in 14.04: ethtool -i eth1 driver: enic version: 2.1.1.50 firmware-version: 2.1(3a) bus-info: 0000:07:00.0 supports-statistics: yes supports-test: no supports-eeprom-access: no supports-register-dump: no supports-priv-flags: no ethtool -i vnet1 driver: tun version: 1.6 firmware-version: bus-info: tap supports-statistics: no supports-test: no supports-eeprom-access: no supports-register-dump: no supports-priv-flags: no ethtool -k vnet1 Features for vnet1: rx-checksumming: off [fixed] tx-checksumming: on tx-checksum-ipv4: off [fixed] tx-checksum-ip-generic: on tx-checksum-ipv6: off [fixed] tx-checksum-fcoe-crc: off [fixed] tx-checksum-sctp: off [fixed] scatter-gather: on tx-scatter-gather: on tx-scatter-gather-fraglist: on tcp-segmentation-offload: off tx-tcp-segmentation: off tx-tcp-ecn-segmentation: off tx-tcp6-segmentation: off udp-fragmentation-offload: on generic-segmentation-offload: on generic-receive-offload: on large-receive-offload: off [fixed] rx-vlan-offload: off [fixed] tx-vlan-offload: on ntuple-filters: off [fixed] receive-hashing: off [fixed] highdma: off [fixed] rx-vlan-filter: off [fixed] vlan-challenged: off [fixed] tx-lockless: off [fixed] netns-local: off [fixed] tx-gso-robust: off [fixed] tx-fcoe-segmentation: off [fixed] tx-gre-segmentation: off [fixed] tx-ipip-segmentation: off [fixed] tx-sit-segmentation: off [fixed] tx-udp_tnl-segmentation: off [fixed] tx-mpls-segmentation: off [fixed] fcoe-mtu: off [fixed] tx-nocache-copy: on loopback: off [fixed] rx-fcs: off [fixed] rx-all: off [fixed] tx-vlan-stag-hw-insert: on rx-vlan-stag-hw-parse: off [fixed] rx-vlan-stag-filter: off [fixed] l2-fwd-offload: off [fixed] in 13.04 : ethtool -i eth1 driver: enic version: 2.1.1.39 firmware-version: 2.1(3a) bus-info: 0000:07:00.0 supports-statistics: yes supports-test: no supports-eeprom-access: no supports-register-dump: no supports-priv-flags: no ethtool -i vnet1 driver: tun version: 1.6 firmware-version: bus-info: tap supports-statistics: no supports-test: no supports-eeprom-access: no supports-register-dump: no supports-priv-flags: no ethtool -k vnet1 Features for vnet1: rx-checksumming: off [fixed] tx-checksumming: on tx-checksum-ipv4: off [fixed] tx-checksum-ip-generic: on tx-checksum-ipv6: off [fixed] tx-checksum-fcoe-crc: off [fixed] tx-checksum-sctp: off [fixed] scatter-gather: on tx-scatter-gather: on tx-scatter-gather-fraglist: on tcp-segmentation-offload: on tx-tcp-segmentation: on tx-tcp-ecn-segmentation: on tx-tcp6-segmentation: on udp-fragmentation-offload: on generic-segmentation-offload: on generic-receive-offload: on large-receive-offload: off [fixed] rx-vlan-offload: off [fixed] tx-vlan-offload: off [fixed] ntuple-filters: off [fixed] receive-hashing: off [fixed] highdma: off [fixed] rx-vlan-filter: off [fixed] vlan-challenged: off [fixed] tx-lockless: off [fixed] netns-local: off [fixed] tx-gso-robust: off [fixed] tx-fcoe-segmentation: off [fixed] fcoe-mtu: off [fixed] To manage notifications about this bug go to: https://bugs.launchpad.net/qemu/+bug/1297487/+subscriptions
