Message-ID: <19ce51bd94685e2bed6cbc20467e5705.sboyd@kernel.org>
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
In-Reply-To: <d2519ebc-8f71-1cf0-9ec8-a65a7f094853@loongson.cn>
References: <20221129034157.15036-1-zhuyinbo@loongson.cn> <31c690a347f858a477bbba9c838984ed.sboyd@kernel.org> <4b5fd886-57ce-01ef-8224-432898b7fb1c@loongson.cn> <8332a1cf44b01f06bdd5db9dc5d7f387.sboyd@kernel.org> <01ee3dc6-a868-fd2b-93aa-11e6bdfcc9df@loongson.cn> <9e8952c9415973dc7276185e3cdf5ae7.sboyd@kernel.org> <d92223a0-6d4c-33ea-1473-3d40bdd0ad9e@loongson.cn> <834da7dc-bb5d-3427-43e5-938e40a2d180@loongson.cn> <6c497d2d70d215a86be178fc08546f4d.sboyd@kernel.org> <d2519ebc-8f71-1cf0-9ec8-a65a7f094853@loongson.cn>
Subject: Re: [PATCH v10 2/4] clk: clk-loongson2: add clock controller driver support
From:   Stephen Boyd <sboyd@kernel.org>
To:     Huacai Chen <chenhuacai@kernel.org>,
        Jianmin Lv <lvjianmin@loongson.cn>,
        Jiaxun Yang <jiaxun.yang@flygoat.com>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Michael Turquette <mturquette@baylibre.com>,
        Rob Herring <robh+dt@kernel.org>,
        WANG Xuerui <kernel@xen0n.name>,
        Yang Li <yang.lee@linux.alibaba.com>,
        devicetree@vger.kernel.org, linux-clk@vger.kernel.org,
        linux-kernel@vger.kernel.org, liupeibao@loongson.cn,
        loongarch@lists.linux.dev, wanghongliang@loongson.cn,
        zhuyinbo <zhuyinbo@loongson.cn>
Date:   Fri, 24 Feb 2023 11:10:14 -0800
User-Agent: alot/0.10
Precedence: bulk

Quoting zhuyinbo (2023-02-22 18:16:49)
>=20
> =E5=9C=A8 2023/2/23 =E4=B8=8A=E5=8D=885:00, Stephen Boyd =E5=86=99=E9=81=
=93:
> > Quoting zhuyinbo (2023-02-21 21:10:14)
> >>>>> yes, the use of dts doesn't require the use of CLK_OF_DECLARE and c=
an
> >>>>> use platform_driver_register
> >>>>>
> >>>>> but my drvier not use platform_driver_register to register clk and =
use
> >>>>> CLK_OF_DECLARE to match of_clk_init.
> >>>> of_clk_init() is there to register clks that are needed for early in=
it,
> >>>> i.e. the clockevent/clocksource or the root interrupt controller
> >>>> (irqchip). Otherwise, it isn't necessary to register clks via
> >>>> of_clk_init().
> >>> okay, I got it.
> >> and,=C2=A0 the time driver=C2=A0 get clock by CCF that ask loongson2 c=
lock driver
> >> use CLK_OF_DECLARE
> >>
> >> to match of_clk_init.=C2=A0=C2=A0 because=C2=A0 the timer_probe=C2=A0 =
is very early and the
> >> timer driver was use TIMER_OF_DECLARE
> >>
> >> to match time_probe.
> >>
> > If you have a time driver that gets clks, register those early with
> > CLK_OF_DECLARE_DRIVER() and then have a platform driver for the rest of
> > the clk tree that registers clks later.
>=20
> okay, I got it.=C2=A0 and this series patch I will use platform driver.=20
> later, if the loongson2 time driver use the
>=20
> CLK_OF_DECLARE style I will make a alone time clk driver for it that=20
> use=C2=A0 CLK_OF_DECLARE_DRIVER()
>=20

It can be the same file if you want. But then it can't be a module. Up
to you what you want to do there.