From owner-freebsd-stable@FreeBSD.ORG Tue Jan 9 20:50:13 2007 Return-Path: X-Original-To: freebsd-stable@freebsd.org Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 4D99016A40F; Tue, 9 Jan 2007 20:50:13 +0000 (UTC) (envelope-from robertw@ssginnovations.com) Received: from ssg1.ssginnovations.com (ssginnovations.com [205.145.135.135]) by mx1.freebsd.org (Postfix) with ESMTP id 2A8DB13C455; Tue, 9 Jan 2007 20:50:12 +0000 (UTC) (envelope-from robertw@ssginnovations.com) Received: from server1.ssgi.local (unknown [205.145.129.164]) by ssg1.ssginnovations.com (Mail Daemon) with ESMTP id 383D0406E; Tue, 9 Jan 2007 15:51:33 -0500 (EST) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Tue, 9 Jan 2007 15:50:06 -0500 X-MimeOLE: Produced By Microsoft Exchange V6.5 Message-ID: <85D4F2C294E8434CA0AF77574153268620304A@server1.ssgi.local> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: bge autoneg full-duplex modes missing on 6.2-RC2 (cause: IPMI patch) Thread-Index: Acc0BiJgR8vgOErpQTu3Othen4yjIwAEnfnQ From: "Robert Wojciechowski" To: Cc: freebsd-stable@freebsd.org Subject: RE: bge autoneg full-duplex modes missing on 6.2-RC2 (cause: IPMI patch) X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 09 Jan 2007 20:50:13 -0000 > I'm having problems on one of two identical servers which have dual > Broadcom BCM5704s onboard. The problem is that one of the servers is not > linking at 100baseTX-FDX on the bge0 interface no matter what I do > (forced, limited advertising on the switch, etc). >=20 > Upon further investigation I noticed this in dmesg: >=20 > bge0: mem 0xfe7f0000-0xfe7fffff > irq 49 at device 3.0 on pci2 > miibus1: on bge0 > brgphy0: on miibus1 > brgphy0: 10baseT, 100baseTX, 1000baseTX, 1000baseTX-FDX, auto > bge0: Ethernet address: 00:30:48:56:93:32 > bge1: mem 0xfe7e0000-0xfe7effff > irq 50 at device 3.1 on pci2 > miibus2: on bge1 > brgphy1: on miibus2 > brgphy1: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseTX, > 1000baseTX-FDX, auto > bge1: Ethernet address: 00:30:48:56:93:33 >=20 > Notice 10baseTX-FDX and 100baseTX-FDX are missing from brgphy0! The > other server doesn't have this problem at all and is running off the > same kernel and identical hardware. >=20 > This is on amd64 and 6.2-RC2 as of Jan 8th. Thanks! >=20 I may have found out why this is happening. I backed out Dan Ambrisko's bge IPMI patches but I forgot to reset my switch's negotiation configuration. I reset everything back to autoneg with clean RELENG_6_2 sources and everything links up as expected. After looking into a bit, it seems the problem is indeed related to the MII PHY media detection being flaky on bge0 (but operating correctly for bge1) when using the IPMI patch.=20 If I reboot over and over 10baseTX disappears and appears intermittently, as well as all the modes I'm looking for such as 100baseTX-FDX, albeit less often. Perhaps this is due to a race of some sort with the IPMI patch touching the PHY and the PHY media detection? I'm still running on the patch from http://www.ambrisko.com/doug/bge_ipmi_3.patch, modified slightly to patch cleanly on RELENG_6_2. I tried RELENG_6 but the system froze on boot while initializing bge and I'm not sure why (something I'll trace down after this). I see that it is also committed to HEAD but I haven't tried that version. Right now I temporarily patched mii_physubr.c to force it to advertise 100baseTX-FDX and it correctly links up. Any ideas? Thanks!! -- Robert