From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nate Case Subject: Re: [PATCH] PHYLIB: Add 1000Base-X support for Broadcom bcm5482 Date: Thu, 15 May 2008 16:03:55 -0500 Message-ID: <1210885435.13845.512.camel@localhost.localdomain> References: <1210695378.13845.362.camel@localhost.localdomain> <1210804791.13845.423.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: netdev To: "Maciej W. Rozycki" Return-path: Received: from xes-mad.com ([216.165.139.214]:50773 "EHLO xes-mad.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754947AbYEOVEB (ORCPT ); Thu, 15 May 2008 17:04:01 -0400 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On Thu, 2008-05-15 at 00:15 +0100, Maciej W. Rozycki wrote: > OK. One note on this occasion: please keep the registers sorted by the > index. I missed it with the original review, but the additional registers > at 0x15, 0x17 and 0x18 (and the values within) should be placed between > MII_BCM54XX_ESR at 0x11 and MII_BCM54XX_ISR at 0x1a. Fixed this in patch v3 (coming up shortly). > I was a bit worried how it plays with actually trying to force different, > perhaps incompatible, link parameters. I didn't try forcing speeds from userspace, but the worst I can envision happening is that after forcing 10mbit/s or half-duplex mode (for example) it would still report 1000mbit/s FD. The forcing itself would probably not report an error, though you could argue that it should. My understanding is that the hardware will take the writes but ignore those registers in 1000Base-X mode. -- Nate Case