From mboxrd@z Thu Jan 1 00:00:00 1970 From: Francois Romieu Subject: Re: NIC driver r8168 with r8169 for RTL8111/8168B and DGE-528T together Date: Thu, 21 Jul 2011 12:22:59 +0200 Message-ID: <20110721102259.GA10481@electric-eye.fr.zoreil.com> References: <4E26B198.4020606@telkomsa.net> <20110720122206.73898673@nehalam.ftrdhcpuser.net> <4E2752A2.3030208@telkomsa.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Stephen Hemminger , netdev@vger.kernel.org To: Danie Wessels Return-path: Received: from violet.fr.zoreil.com ([92.243.8.30]:40292 "EHLO violet.fr.zoreil.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752003Ab1GUKiP (ORCPT ); Thu, 21 Jul 2011 06:38:15 -0400 Content-Disposition: inline In-Reply-To: <4E2752A2.3030208@telkomsa.net> Sender: netdev-owner@vger.kernel.org List-ID: Danie Wessels : [...] > .> Remove the 8168 PCI IDs from the r8169 driver and you should be set. > gr8. I can give that a try...8^0 (some more hints....?) Search for rtl8169_pci_tbl and remove the 0x8168 line : the kernel r8169 driver will stop being used on the 8168s. > > .>It ought to be supported by the kernel r8169 driver. > > See bugs listed on Ubuntu: > https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/141343 Executive summary : a. started in 2007 b. includes "kernel r8169 driver does not work, tried Realtek's r8168, happy now" c. includes b. + "...wait a few days, Oops" d. covers different devices (8168b, 8168c, plain 8169). e. people still using old kernel (see post 2.6.26 77332894c21165404496c56763d7df6c15c4bb09 in #38 then mention of 2.6.22, 2.6.24) f. usual "this is the same bug" (#141343 is the same bug as #141343 e. one (1) dmesg attached to the whole thread (74 messages). Tons of lspci, not a single XID. > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/221499 Trend (?) : recent kernels help. XID and up-to-date reports would be welcome if they can be streamlined to a standard (for me :o) ) kernel. See below. > https://bugs.launchpad.net/linux/+bug/347711 I wish I could filter out comments by author. Flat PR get really, really messy with time. > And 0n Bug #347711 > linux-kernel-bugs #12411 > Duplicates of this bug > Bug #76489 > Bug #347670 Parts of reports look the same but it is more a bag of mixed bugs than a duplicated one. On the bright side I only have to pick one to fix something. :o/ > .> Which problem(s) do you have with it ? > No communication to outside devices. I can not ping my router > through it but can ping its IP. So it is receiving packets either in promiscuous mode or (and) as long as its peer knows its MAC address ? No Tx at all (it may help if you can capture traffic on the remote end) or just a few packets before it stops (and spits a NETDEV WATCHDOG message) ? > What can I do now or where should I report it or how can I help? > RTFM 4 rules.d = where? - Where (general) Here and/or kernel.org bugzilla. Cc: good. Private email: bad. You may Cc: Hayes as well. If you go for bugzilla, fill Product/Component as "Drivers/Network" - What (general) If it is not a regression - i.e. it does not fall in the "stopped working" bucket - always try last Linus's -rc or David Miller's "next" branch. I may have a bit more pending sauce but the "next" branch usually is a good starting point (especially as fixing regressions in it asap is a rather high priority task). "next" is waiting for you at : git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next-2.6.git Comparing Realtek's driver and kernel r8169 one is welcome. BIOS lan options and reboot from a different OS sometimes make a difference (its always nice to be sure than the damn thing _can_ work). Include: - dmesg output (unabbreviated, explicit). The XID line is a mess but it allows me to triage the bugs. - ethtool -d ethX. Be it a single line or a (nowadays partial) registers dump. - lspci -v/-tv - brand/motherboard identification (wrt your report, I already have most of the needed material) Ping me when you want a status update or feel things are sidetracked. - What (specific) One of Ubuntu's report suggests that you are experiencing a regression. If so it would be nice to bissect it. In the short run I can not do more as I must push something out for a mac address change problem with the 8168evl. -- Ueimor