From mboxrd@z Thu Jan 1 00:00:00 1970 From: "David S. Miller" Subject: Re: RFC/PATCH capture qdisc requeue event in stats Date: Tue, 28 Sep 2004 17:03:15 -0700 Sender: netdev-bounce@oss.sgi.com Message-ID: <20040928170315.760b9851.davem@davemloft.net> References: <1093799632.1073.410.camel@jzny.localdomain> <1096413009.27967.1.camel@localhost.localdomain> <20040928161838.3d19a17d.davem@davemloft.net> <1096416101.27967.10.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: hadi@cyberus.ca, netdev@oss.sgi.com Return-path: To: Stephen Hemminger In-Reply-To: <1096416101.27967.10.camel@localhost.localdomain> Errors-to: netdev-bounce@oss.sgi.com List-Id: netdev.vger.kernel.org On Tue, 28 Sep 2004 17:01:41 -0700 Stephen Hemminger wrote: > On Tue, 2004-09-28 at 16:18 -0700, David S. Miller wrote: > > On Tue, 28 Sep 2004 16:10:09 -0700 > > Stephen Hemminger wrote: > > > > > On Sun, 2004-08-29 at 13:13 -0400, jamal wrote: > > > > > > > > The requeue event is useful in finding out when a device is overloaded > > > > on the egress (bus or bandwidth). > > > > Atached patch introduces this. I would have used the overlimit bits > > > > but at the moment thats being used for different semantical reasons. > > > > I have not done extensive testing on it. > > > > > > > > Opinions welcome - If all is good, Dave please apply. > > > > > > Dave, what happened to this? I put the stuff into iproute2 but the > > > requeue stat never made it into 2.6. Is it a bad idea? > > > > Yes, API breaker. > > Well it seems to work for me: > kernel > New Old > tc New Ok Ok(0) > Old Ok(0) Ok > > Because tc correctly handles the returned TLV size. Because tc has all the length checking stuff added to it, other applications using netlink might not. The whole world is not 'tc'. I'm just rehashing how the most recent thread between Jamal and myself ended about this topic.