Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
In-Reply-To: <CAFr9PX=oLqQqvykiwOGAGg1H2CG0BTEqn0TuSrijodjxY52LxQ@mail.gmail.com>
References: <20201114135044.724385-1-daniel@0x0f.com> <20201114135044.724385-3-daniel@0x0f.com> <160843556037.1580929.13575770228476494246@swboyd.mtv.corp.google.com> <CAFr9PX=f=HyHBk4s3e_vnMDC53R1X18m7sKFCLMMjitPm+8oEQ@mail.gmail.com> <160848984935.1580929.936824086681978646@swboyd.mtv.corp.google.com> <CAFr9PX=oLqQqvykiwOGAGg1H2CG0BTEqn0TuSrijodjxY52LxQ@mail.gmail.com>
Subject: Re: [PATCH 2/6] dt-bindings: clk: mstar msc313 mpll binding description
From:   Stephen Boyd <sboyd@kernel.org>
Cc:     DTML <devicetree@vger.kernel.org>, linux-clk@vger.kernel.org,
        linux-arm-kernel <linux-arm-kernel@lists.infradead.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Willy Tarreau <w@1wt.eu>
To:     Daniel Palmer <daniel@0x0f.com>
Date:   Tue, 09 Feb 2021 18:29:30 -0800
Message-ID: <161292417080.418021.13416197091150418216@swboyd.mtv.corp.google.com>
User-Agent: alot/0.9.1
Precedence: bulk

Quoting Daniel Palmer (2020-12-21 00:51:56)
> Hi Stephen,
>=20
> On Mon, 21 Dec 2020 at 03:44, Stephen Boyd <sboyd@kernel.org> wrote:
> >
> > Quoting Daniel Palmer (2020-12-19 22:35:41)
> > > Hi Stephen,
> > >
> > > On Sun, 20 Dec 2020 at 12:39, Stephen Boyd <sboyd@kernel.org> wrote:
> > > > > +  clock-output-names:
> > > > > +    minItems: 8
> > > > > +    maxItems: 8
> > > > > +    description: |
> > > > > +      This should provide a name for the internal PLL clock and =
then
> > > > > +      a name for each of the divided outputs.
> > > >
> > > > Is this necessary?
> > >
> > > I found without the names specified in the dt probing of muxes that
> > > depend on the outputs but appear earlier didn't work.
> > > Also this same PLL layout seems to be used in some other places so
> > > eventually I was thinking this driver would get used for those PLLs
> > > with different output names.
> >
> > Still seems like it could be auto-generated based on dev_name() +
> > number.
>=20
> At one point I had something similar to that where the output names
> were generated at probe.
> Without the clock outputs listed in the device tree clock muxes that
> source clocks from the mpll couldn't probe properly as they couldn't
> look up all of their parents if they probed before the mpll.
> Maybe I'm doing something wrong there? I couldn't find a way to always
> resolve all of the parents or defer the probe of the muxes until the
> mpll clocks are registered.
>=20

The child clks should be using clk_parent_data to point to the parent
clks through DT. That way the name of the clk doesn't matter except for
debug purposes.