"Ronciak, John" <[email protected]> writes:

> If you want to do this in your product you are by no means prevented
> from doing so.  I've been saying this all along.

Removing promiscuous mode support from the PF driver and enabling it in
the VF driver cannot do harm by misconfiguration. In a scenario where
VFs are driven inside VMs and the hypervisor is mainline Linux, it
cannot be accidentally enabled.

If a hypervisor vendor chooses to implement promiscuous mode support in
his PF driver, he obviously has to think about policy. Think of it as a
checkbox in the GUI saying "Give this VM the ability to monitor _all_
network traffic (bad things may happen)". Of course, he can get the
default wrong, but there is always _some_ way to shoot yourself in the
foot. The difference compared to the situation now is that the
hypervisor vendor does not have to tell his customers to use a special
kernel with his patches, which of course is not compatible to some other
vendor's patches. The customer can use whatever he likes. Everybody
wins.

My suggestion in a nutshell:

 - Establish a common protocol for promiscuous mode support by pushing
   the respective changes to igbvf into Linux.

 - Don't implement it in igb.

Regards, Julian


------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev 
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel&#174; Ethernet, visit 
http://communities.intel.com/community/wired

Reply via email to