Re: [PATCH net 0/3] be2net fixes
From: Ben Hutchings <hidden>
Date: 2013-12-05 15:58:49
On Thu, 2013-12-05 at 01:28 -0500, David Miller wrote:
From: Somnath Kotur <redacted> Date: Thu, 5 Dec 2013 06:24:10 +0000quoted
quoted
This is absolutely not something an individual driver should decide to do, _EVERY_ driver has this issue.OK, Understood, guess you want this to ideally be handled somewhere up in the stack/framework?I think it's pointless to do at all.
It's not entirely pointless - if there is a lot of unwanted multicast traffic the change to all-multicast can substantially increase CPU load and the reason for this may not be obvious since there is no way to find out the real state of the hardware filter (or even the real state of the IFF_ALLMULTI flag). The same may be true if an interface is automatically switched into promiscuous mode, though switches can generally be assumed to filter unicast traffic properly. It seems like it would be useful to expose the real state of the hardware filter to userland - just not through log messages. Ben. -- Ben Hutchings, Staff Engineer, Solarflare Not speaking for my employer; that's the marketing department's job. They asked us to note that Solarflare product names are trademarked.