From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mailout2.w1.samsung.com ([210.118.77.12]:39489 "EHLO mailout2.w1.samsung.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750767AbdJCGDI (ORCPT ); Tue, 3 Oct 2017 02:03:08 -0400 Subject: Re: [PATCH 0/4] Fix problems with Exynos4412 ISP clocks To: Krzysztof Kozlowski Cc: linux-clk@vger.kernel.org, linux-samsung-soc@vger.kernel.org, Sylwester Nawrocki , Chanwoo Choi , Inki Dae , Bartlomiej Zolnierkiewicz From: Marek Szyprowski Message-id: Date: Tue, 03 Oct 2017 08:03:02 +0200 MIME-version: 1.0 In-reply-to: <20171002173631.lgkqfrtbacyzqeml@kozik-lap> Content-type: text/plain; charset="utf-8"; format="flowed" References: <20171002104759.25944-1-m.szyprowski@samsung.com> <20171002173631.lgkqfrtbacyzqeml@kozik-lap> Sender: linux-clk-owner@vger.kernel.org List-ID: Hi Krzysztof, On 2017-10-02 19:36, Krzysztof Kozlowski wrote: > On Mon, Oct 02, 2017 at 12:47:55PM +0200, Marek Szyprowski wrote: >> Exynos4412 ISP clock controller is located in the SOC area, which belongs >> to ISP power domain. This was not properly handled by the current >> Exynos4-clk driver. This patchset instantiates a separate clock driver >> for those clocks, updates all clients of ISP clocks and ensures that >> the driver is properly integrated in ISP power domin using runtime PM >> feature of the clock framework. >> >> This finally solves all the mysterious freezes in accessing ISP clocks >> when ISP power domain is disabled. >> >> The last patch breaks support for old dtbs. It can be applied when all >> boards are updated. Exynos4412 ISP subsystem is only used by Trats2 >> boards, for which kernel is updated always together with the dtb file, >> so the last patch can be applied to the next kernel release after merging >> the DTS patch. > I am fine with this approach. However I see that third patch (dts) > depends on previous so it will also wait one cycle. arm-soc folks for > some time do not accept code mixed with dts. > > Unless you can split dts into two patches: adding new clock controller > and removing old. In such case the new clock could be applied in > parallel to clk code changes. DTS patch cannot be split into two patches without breaking ISP support, because adding a new clock controller requires changes in the clients and register resource size of the old one, so either old or new one has to be fully defined. If arm-soc cannot pull mixed branches, then I'm okay with postponing DTS changes by one release. > ... Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland