From mboxrd@z Thu Jan 1 00:00:00 1970 From: broonie@opensource.wolfsonmicro.com (Mark Brown) Date: Tue, 5 Oct 2010 09:54:28 -0700 Subject: [alsa-devel] [PATCH 10/10] ASoC: SAMSUNG: Add Machine driver for S/PDIF PCM audio In-Reply-To: References: <1286191550-22197-1-git-send-email-sw.youn@samsung.com> <1286194583-19937-1-git-send-email-sw.youn@samsung.com> <20101004224159.GA4972@opensource.wolfsonmicro.com> <20101005044304.GA26509@opensource.wolfsonmicro.com> <20101005055937.GA19989@opensource.wolfsonmicro.com> Message-ID: <20101005165427.GB20555@opensource.wolfsonmicro.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Tue, Oct 05, 2010 at 04:09:47PM +0900, Jassi Brar wrote: > On Tue, Oct 5, 2010 at 2:59 PM, Mark Brown > > Well, there's two approaches the driver can take: one is to change the > > EPLL to deliver the clock rates which allow maximum flexibility, the > > other is to constrain the clock rates offered to applications based on > > the frequencies that can be generated from the current EPLL setup. > For SMDKs(our reference platform) I would like to change EPLL in order > to be able to verify accuracy and provide all capabilities of the controllers. That's why I'm saying provide an option to turn this behaviour on - I agree that it's useful to be able to show the feature, I just think it's going to catch users out if they're not warned that it's happening somehow. > >> (our priority is accuracy of each IP's functioning rather than having > >> all parts of > >> SMDK playing nice with each other on such h/w design issues) > > That'd be the problem, kind of - it does mean people can get burned when > > they pick up the BSP code and use it as a reference for their systems. > I expect people to replace/modify/inspect board specific code before they > run the BSP on their systems. IMO reference platforms are not good for > integrated-testing. To be fair the users doing this the individual drivers do look OK by themselves - there's only a problem when it comes to system integration and that's not going to be obvious unless someone has reviewed the drivers while understanding the clock tree for the chip as a whole.