From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [GIT] Networking Date: Thu, 02 May 2013 13:55:42 -0400 (EDT) Message-ID: <20130502.135542.1332946768988319212.davem@davemloft.net> References: <1367512467.4000.7.camel@dcbw.foobar.com> <20130502.133947.387694205526879673.davem@davemloft.net> Mime-Version: 1.0 Content-Type: Text/Plain; charset=euc-kr Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: dcbw@redhat.com, torvalds@linux-foundation.org, hayeswang@realtek.com, akpm@linux-foundation.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org To: mirqus@gmail.com Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org =46rom: Micha=A9=A9 Miros=A9=A9aw Date: Thu, 2 May 2013 19:47:25 +0200 > 2013/5/2 David Miller : >> From: Dan Williams >> Date: Thu, 02 May 2013 11:34:27 -0500 >> >>> On Wed, 2013-05-01 at 18:28 -0700, Linus Torvalds wrote: >>>> On Wed, May 1, 2013 at 1:47 PM, David Miller = wrote: >>>> > >>>> > Highlights (1721 non-merge commits, this has to be a record of s= ome >>>> > sort): >>>> >>>> Lowlight: it completely breaks my machine with r8169 ethernet. In= the >>>> networkmanager applet, it claims no cable connection, which is a b= it >>>> odd, because (a) it works with an older kernel and (b) the kernel >>>> messages actually say >>>> >>>> r8169 0000:01:00.0 eth0: link up >>> >>> NM calls ETHTOOL_GLINK and if that returns success, NM expects the >>> driver to support carrier detection. NM then listens to netlink fo= r >>> device flags changes, and uses IFF_LOWER_UP to determine carrier on= /off >>> state. NM does not use NETIF_F_VLAN_CHALLENGED or any of the other >>> ETHTOOL_GFEATURES flags yet, but will in the future. >>> >>> Is the link status accurately reflected by /sys/class/net/eth0/carr= ier ? >> Something cares about the .../eth0/flags value because with the bit >> ordering different for bits 10 and above things break. >=20 > Are you sure it's "flags"? /sys/class/net/*/features were removed som= e > time ago, and flags don't depend on NETIF_F_*. Oh, it's flags, which is printed in raw hex by net/core/net-sysfs.c: NETDEVICE_SHOW(flags, fmt_hex);