<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN">
<html><body>
<p>On 18.02.2012 13:10, Adis Nezirovic wrote:</p>
<blockquote type="cite" style="padding-left:5px; border-left:#1010ff 2px solid; margin-left:5px; width:100%"><!-- html ignored --><!-- head ignored --><!-- meta ignored -->
<pre>Can we add the following line to udev rules (fixup_udev_rules.patch):
+KERNEL=="vhost_net", GROUP="@GROUP@", MODE="0660"
so that both, /dev/kvm and /dev/vhost_net devices are group-owned by
kvm group (or @GROUP@)
The defaults for /dev/vhost_net are root:root 600
The idea is that one can run qemu-system-x86_64 as ordinary user, and
use vhost_net acceleration using netdev syntax:
-netdev type=tap,ifname=tap0,id=guest0,script=no,downscript=no,vhost=on \
-device virtio-net-pci,netdev=guest0,mac=XX:YY:XX:YY:XX:YY
tap0 device is precreated in rc.bridge script (tunctl -g kvm -t tap0),
so nothing requires root privileges for running VM.
Best regards,
Adis
_______________________________________________
SlackBuilds-users mailing list
<a href="mailto:SlackBuilds-users@slackbuilds.org">SlackBuilds-users@slackbuilds.org</a>
<a href="http://lists.slackbuilds.org/mailman/listinfo/slackbuilds-users">http://lists.slackbuilds.org/mailman/listinfo/slackbuilds-users</a>
Archives - <a href="http://lists.slackbuilds.org/pipermail/slackbuilds-users/">http://lists.slackbuilds.org/pipermail/slackbuilds-users/</a>
FAQ - <a href="http://slackbuilds.org/faq/">http://slackbuilds.org/faq/</a>
</pre>
</blockquote>
<p>I had some time over the weekend to check out qdev and vhost_net and I like what I've found so far. I reconfigured a non-critical, production KVM (slackware-13.37) to run vhost_net and the network latency on pings is 50% of the non-vhost_net KVMs. I'll keep the KVM running in this config and watch performance and stability. So far I support this recommendation and will add it to the next qemu-kvm SlackBuild if my KVM survives my soak test.</p>
<p>Thanks for the suggestion!</p>
<p>Stu</p>
<div> </div>
</body></html>