From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from coyote.quickmin.net (coyote.quickmin.net [217.14.112.24]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client CN "smtp.quickmin.de", Issuer "smtp.quickmin.de" (not verified)) by ozlabs.org (Postfix) with ESMTPS id 6B9B6B6F2B for ; Tue, 10 Nov 2009 23:47:41 +1100 (EST) Received: from rfietze.mit.telemotive.de (rfietze.mit.telemotive.de [192.168.5.33]) by alderan.mit.telemotive.de (Postfix) with ESMTP id 671A5C5053 for ; Tue, 10 Nov 2009 13:47:34 +0100 (CET) From: Roman Fietze To: linuxppc-dev@lists.ozlabs.org Subject: MPC5200B FEC_IEVENT_RFIFO_ERROR leading to Link is Down MIME-Version: 1.0 Message-ID: <200911101347.33837.roman.fietze@telemotive.de> Date: Tue, 10 Nov 2009 13:47:33 +0100 Content-Type: text/plain; charset="iso-8859-1" List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Hello, When having high traffic on the MPC5200B ATA (MWDMA2), FEC (100MBit) and LPC (MTD FLASH read) I will get the typical =46EC_IEVENT_RFIFO_ERROR's. Without the FLASH reads, that we use to stress the probably existing LPC arbiter or MPC5200B silicon bug when running UDMA2+LPC, and which should not happen using MWDMA2+LPC, I could not yet reproduce the =46EC_IEVENT_RFIFO_ERROR. Now comes the problem: since the PHY driver rework followed by John Bonesio's fix (rev 37ccd92f55c6c6) to avoid accessing the DIO bus at interrupt context, this now sometimes causes an irrecoverable link down (PHY: f0003000:00 - Link is Down). This patch is of course necessary to avoid a panic. I can reproduce that on the current DENX/master using an own board very close to the Lite5200B. This board runs fine under the old 2.4.25, of course also sometimes showing the FEC FIFO error. Any ideas? Roman =2D-=20 Roman Fietze Telemotive AG B=FCro M=FChlhausen Breitwiesen 73347 M=FChlhausen Tel.: +49(0)7335/18493-45 http://www.telemotive.de