From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Walmsley Subject: Re: [PATCH v3 1/1] Registering sgx device and it's platform data Date: Wed, 31 Mar 2010 03:39:39 -0600 (MDT) Message-ID: References: <19F8576C6E063C45BE387C64729E7394044DEBF372@dbde02.ent.ti.com> <20100330094902.GG30597@nokia.com> Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Return-path: Received: from utopia.booyaka.com ([72.9.107.138]:44252 "EHLO utopia.booyaka.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933110Ab0CaJjk (ORCPT ); Wed, 31 Mar 2010 05:39:40 -0400 In-Reply-To: <20100330094902.GG30597@nokia.com> Sender: linux-omap-owner@vger.kernel.org List-Id: linux-omap@vger.kernel.org To: Felipe Balbi Cc: "Agarwal, Preshit" , "linux-omap@vger.kernel.org" , "Pais, Allen" Hi Felipe, On Tue, 30 Mar 2010, Felipe Balbi wrote: > On Tue, Mar 30, 2010 at 10:49:19AM +0200, ext Paul Walmsley wrote: > > I would suggest that you name this function be named something more > > specific, such as omap_sgx_set_bus_tput(), since the SGX driver may also > > need to set other PM constraints later, such as wakeup latency > > constraints. > > then it could just add the contraints to the same function instead of passing > another pointer to driver ?? It's hard to say without looking at the driver. Bus throughput constraints are only needed when the device is actively transferring data to/from the rest of the system. Wakeup latency constraints are only needed when the device can go idle and enter some low-power state, but also needs to be ready in some limited amount of time to service an interrupt or a wakeup. So, often it would make sense to set the two parameters at different points in the driver code. - Paul