From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from gate.crashing.org (gate.crashing.org [63.228.1.57]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTP id 39ED5DDF22 for ; Fri, 20 Jul 2007 03:08:38 +1000 (EST) In-Reply-To: <1184795678.25235.269.camel@localhost.localdomain> References: <1184722754.25235.183.camel@localhost.localdomain> <5F6EB5F7-C5AE-437F-9DF1-22FE38D7A841@kernel.crashing.org> <1184795678.25235.269.camel@localhost.localdomain> Mime-Version: 1.0 (Apple Message framework v752.2) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: From: Segher Boessenkool Subject: Re: [PATCH] Add StorCenter DTS first draft. Date: Thu, 19 Jul 2007 19:07:59 +0200 To: Benjamin Herrenschmidt Cc: linuxppc-dev@ozlabs.org, Jon Loeliger List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , >>>> + /* Following required by dtc but not used */ >>>> + i-cache-line-size = <0>; >>>> + d-cache-line-size = <0>; >>>> + i-cache-size = <4000>; >>>> + d-cache-size = <4000>; >>>> + }; >>> >>> The 32 bits kernel may not be using those yet, but it will. 64 bits >>> does >>> already and I plan to merge those bits at one point. >> >> Hrm, what does it use it for? Are you going to require >> all other defined CPU properties as well (like the PowerPC >> binding does)? > > Cache line size is used by the kernel on ppc64 for things like > clearing > memory (to get the stride between subsequent dcbz) or flushing the > cache :-) It's also passed on to userland. > > If it's absent from the device-tree, we default to the values in the > cputable, but if you're going to put the properties in the tree, don't > put a 0 in there. As it is, the day I make the 64 bits code common, > your > DT will break unless I special case "0". So your plan is to prefer the device tree over the cputable, and maybe even deprecate the cputable? I was under the impression that we were going the exact opposite way (we don't have many of required properties in CPU nodes in most DTS files). Either way is fine with me, just choose one; but picking a little from the device tree and the rest from the cputable doesn't sound useful to me, and it would be a lovely maintenance problem. Segher