From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from zipcode.az.mvista.com (unknown [65.200.49.156]) by ozlabs.org (Postfix) with ESMTP id CA0CCDDF21 for ; Sat, 10 Feb 2007 04:31:22 +1100 (EST) Date: Fri, 9 Feb 2007 10:31:49 -0700 From: "Mark A. Greer" To: Grant Likely Subject: Re: [RFC] mpc5200 device tree bindings refinement Message-ID: <20070209173149.GA23362@mag.az.mvista.com> References: <11710062092006-git-send-email-grant.likely@secretlab.ca> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <11710062092006-git-send-email-grant.likely@secretlab.ca> Cc: linuxppc-dev@ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Fri, Feb 09, 2007 at 12:30:09AM -0700, Grant Likely wrote: Hi Grant, > Much needed refinement of mpc5200 device tree binding specifications. > > Short list: > - Specify reqiured 'cell-index' property to identify between multiple SOC > devices of the same type. (Useful for arbitrating shared register access) > +Important child node properties > +name type description > +---- ---- ----------- > +cell-index int When multiple devices are present, is the > + index of the device in the hardware (ie. There > + are 6 PSC on the 5200 numbered PSC1 to PSC6) > + PSC1 has 'cell-index = <0>' > + PSC4 has 'cell-index = <3>' > + Dale Farnsworth and I have the exact same problem with the Marvell bridges (e.g., 2 MPSC ctlrs use different bits in the same reg). Yesterday we talked about it for a while and came up with essentially the same solution as you. The only difference is that we called the property "register-set". I think that is a better name because a) "cell-index" sounds like an index into a DT cell which it isn't, and b) its more descriptive (IMHO) because its really saying what set of regs or bits in a reg are for that ctlr. Comments? Mark