From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Tue, 11 Aug 2015 18:25:30 -0700 From: Stephen Boyd To: Linus Walleij Cc: Michael Turquette , linux-clk@vger.kernel.org Subject: Re: [PATCH] clk: versatile/impd1: switch order of CLCD clocks Message-ID: <20150812012530.GJ26614@codeaurora.org> References: <1438172615-11276-1-git-send-email-linus.walleij@linaro.org> <20150731065008.GU3159@codeaurora.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: List-ID: On 07/31, Linus Walleij wrote: > On Fri, Jul 31, 2015 at 8:50 AM, Stephen Boyd wrote: > > On 07/29, Linus Walleij wrote: > >> The CLCD clocks are registered in the wrong order: first the > >> APB clock, then the pixelclock. This should be the other way > >> around so the driver gets the pixelclock, not the APB clock, > >> when it asks for the first clock for the block. > >> > >> Signed-off-by: Linus Walleij > >> --- > > > > Is this because the consumer driver is calling clk_get() with a > > NULL connection id for the pixelclock and then a more specific > > connection id for the apb clk? > > Yes. Like most PrimeCells do. > > > Where's the consumer driver? > > The NULL (pixel) clock is taken by: > drivers/video/fbdev/amba-clcd.c > > The APB "apb_pclk" clock is taken by: > drivers/amba/bus.c > Ok. I'm still confused and concerned that putting the clocks into the list in a certain order fixes anything. I thought the purpose of a NULL clkdev connection ID was to make it so that we didn't match the apb_pclk lookup. Instead we find the second lookup for the device with the wildcard connection id and use that. If this fix is valid, did you want a fixes tag on this so it goes back to stable? -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project