Thread (40 messages) 40 messages, 6 authors, 2022-10-05

Re: [virtio-dev] RE: [PATCH v5 2/2] virtio-net: use mtu size as buffer length for big packets

From: Si-Wei Liu <hidden>
Date: 2022-09-07 22:58:41
Also in: virtualization


On 9/7/2022 3:11 PM, Parav Pandit wrote:
quoted
From: Si-Wei Liu <redacted>
Sent: Wednesday, September 7, 2022 5:40 PM


On 9/7/2022 12:51 PM, Parav Pandit wrote:
quoted
quoted
And I'd like commit log to include results of perf testing
- with indirect feature on
Which device do you suggest using for this test?
You may use software vhost-net backend with and without fix to compare.
Since this driver fix effectively lowers down the buffer size for the
indirect=on case as well,
Do you have sample example for this?
ip link add link ens300f1 name macvtap1 address $mac type macvtap mode 
bridge
ifindex1=`cat /sys/class/net/macvtap1/ifindex`

qemu-system-x86_64 \
   ...
   -netdev type=tap,id=hnet1,vhost=on,fd=25,vhostfd=27 \
   -device 
virtio-net-pci,netdev=hnet1,id=vnet1,mac=$mac,host_mtu=9000,mrg_rxbuf=off,guest_tso4=off,guest_tso6=off,guest_ecn=off,guest_ufo=off,indirect_desc=on 
\
   25<>/dev/tap${ifindex1} 27<>/dev/vhost-net

indirect_desc=on is not actually required as it's on by default if not 
specified.

-Siwei
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help