From mboxrd@z Thu Jan 1 00:00:00 1970 From: sboyd@codeaurora.org (Stephen Boyd) Date: Thu, 8 Oct 2015 11:09:36 -0700 Subject: [PATCH] clk: qcom: Drop calls to qcom_cc_remove() In-Reply-To: <56163F97.1060604@codeaurora.org> References: <1444288678-24594-1-git-send-email-sboyd@codeaurora.org> <56163F97.1060604@codeaurora.org> Message-ID: <20151008180936.GG26883@codeaurora.org> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 10/08, Rajendra Nayak wrote: > On 10/08/2015 12:47 PM, Stephen Boyd wrote: > > Now that qcom_cc_remove() is a nop, drop calls to > > qcom_cc_remove() and any empty driver remove functions. > > > > Signed-off-by: Stephen Boyd > > --- > > > > And here's the driver part. I left 8960_remove in place to make your patch > > easier to apply. Time to sleep. > > I was trying to test this out, but looks like none of our clock drivers > support a clean remove and reprobe (with or without this patch) > > The problem seems to be the fixed rate clocks that get registered at probe, > never unregistered, causing the subsequent probe to fail. Yeah the gcc drivers are not so good there. We should probably fix those too. The mmcc drivers support a clean remove and reprobe though. So I tested it on msm8960 with these patches and everything looks good. Removing gcc is sort of annoying because it provides the clock for the serial console. -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project