Russell King - ARM Linux <li...@arm.linux.org.uk> writes: > On Mon, Jul 13, 2009 at 02:27:01PM -0700, Kevin Hilman wrote: >> Well, it's not actually this patch that started the problem, it was >> the platform_data updates where clock names were added as part of the >> new platform_data struct shared between platform code and ASoC. >> >> My original proposal was to pass a clock pointer, not a clock name, >> but that was not done, so... >> >> I've pushed an update my 'temp/asoc' branch with the patch below which >> converts to pasing clock pointers instead of clock names. > > Also wrong. Why do you want to pass a clock name _or_ a clock pointer? > Why not follow the clk API and derive it from the struct device.
This is how it is currently is in davinci mainline. There was some objection to this on the ASoC side when support for a new DaVinci SoC was added, so things were changed to pass clock name strings. I need to better understand from ASoC folks why the original model wasn't working so we try to keep the current mainline code, or possibly use clk_add_alias() if needed. Kevin _______________________________________________ Davinci-linux-open-source mailing list Davinci-linux-open-source@linux.davincidsp.com http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
