[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <d84ea240-5bb7-1b77-b287-c2bcca971201@kernel.org>
Date: Fri, 23 Oct 2020 13:42:25 +0200
From: Sylwester Nawrocki <snawrocki@...nel.org>
To: Krzysztof Kozlowski <krzk@...nel.org>,
Tomasz Figa <tomasz.figa@...il.com>, linux-clk@...r.kernel.org
Cc: Sylwester Nawrocki <s.nawrocki@...sung.com>,
Chanwoo Choi <cw00.choi@...sung.com>,
Michael Turquette <mturquette@...libre.com>,
Stephen Boyd <sboyd@...nel.org>, Kukjin Kim <kgene@...nel.org>,
linux-samsung-soc@...r.kernel.org,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
Marek Szyprowski <m.szyprowski@...sung.com>
Subject: Re: [PATCH 2/2] clk: samsung: exynos-clkout: convert to module driver
On 10/1/20 18:56, Krzysztof Kozlowski wrote:
> The Exynos clkout driver depends on board input clock (typically XXTI or
> XUSBXTI), however on Exynos4 boards these clocks were modeled as part of
> SoC clocks (Exynos4 clocks driver). Obviously this is not proper, but
> correcting it would break DT backward compatibility.
>
> Both drivers - clkout and Exynos4 clocks - register the clock providers
> with CLK_OF_DECLARE/OF_DECLARE_1 so their order is fragile (in the
> Makefile clkout is behind Exynos4 clock). It will work only if the
> Exynos4 clock driver comes up before clkout.
>
> A change in DTS adding input clock reference to Exynos4 clocks input
> PLL, see reverted commit eaf2d2f6895d ("ARM: dts: exynos: add input
> clock to CMU in Exynos4412 Odroid"), caused probe reorder: the clkout
> appeared before Exynos4 clock provider. Since clkout depends on Exynos4
> clocks and does not support deferred probe, this did not work and caused
> later failure of usb3503 USB hub probe which needs clkout:
>
> [ 5.007442] usb3503 0-0008: unable to request refclk (-517)
>
> The Exynos clkout driver is not a critical/core clock so there is
> actually no problem in instantiating it later, as a regular module.
> This removes specific probe ordering and adds support for probe
> deferral.
The patch looks good to me, I have tested it on Trats2, where CLKOUT
provides master clock for the audio codec.
Tested-by: Sylwester Nawrocki <s.nawrocki@...sung.com>
With the debug print removed feel free to apply it through your tree.
Reviewed-by: Sylwester Nawrocki <s.nawrocki@...sung.com>
--
Regards,
Sylwester
Powered by blists - more mailing lists