Received-SPF: pass (google.com: domain of linux-kernel+bounces-33621-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240116193542.711482-1-tmenninger@purestorage.com>
 <04d22048-737a-4281-a43f-b125ebe0c896@lunn.ch> <CAO-L_44YVi0HDk4gC9QijMZrYNGoKtfH7qsXOwtDwM4VrFRDHw@mail.gmail.com>
 <da87ce82-7337-4be4-a2af-bd2136626c56@lunn.ch> <CAO-L_46kqBrDdYP7p3He0cBF1OP7TJKnhYK1NR_gMZf2n_928A@mail.gmail.com>
 <20240122123349.cxx2i2kzrhuqnasp@skbuf> <1aab2398-2fe9-40b6-aa5b-34dde946668a@lunn.ch>
 <20240122151251.sl6fzxmfi2f6tokf@skbuf>
In-Reply-To: <20240122151251.sl6fzxmfi2f6tokf@skbuf>
From: Tim Menninger <tmenninger@purestorage.com>
Date: Mon, 22 Jan 2024 07:46:06 -0800
Message-ID: <CAO-L_45_nZ24pvycdahEy0OP2tZjxCw40_o6HE-_C4jmsX3b8g@mail.gmail.com>
Subject: Re: [PATCH] net: dsa: mv88e6xxx: Make *_c45 callbacks agree with
 phy_*_c45 callbacks
To: Vladimir Oltean <olteanv@gmail.com>
Cc: Andrew Lunn <andrew@lunn.ch>, f.fainelli@gmail.com, davem@davemloft.net, 
	edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, 
	netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Mon, Jan 22, 2024 at 7:12=E2=80=AFAM Vladimir Oltean <olteanv@gmail.com>=
 wrote:
>
> On Mon, Jan 22, 2024 at 03:30:20PM +0100, Andrew Lunn wrote:
> > On Mon, Jan 22, 2024 at 02:33:49PM +0200, Vladimir Oltean wrote:
> > > On Tue, Jan 16, 2024 at 05:51:13PM -0800, Tim Menninger wrote:
> > > > My impression is still that the read_c45 function should agree with=
 the
> > > > phy_read_c45 function, but that isn't a hill I care to die on if yo=
u still
> > > > think otherwise. Thoughts?
> > >
> > > FWIW, Tim's approach is consistent with what drivers/net/mdio/mdio-mu=
x.c does.
> > >
> > >             if (parent_bus->read)
> > >                     cb->mii_bus->read =3D mdio_mux_read;
> > >             if (parent_bus->write)
> > >                     cb->mii_bus->write =3D mdio_mux_write;
> > >             if (parent_bus->read_c45)
> > >                     cb->mii_bus->read_c45 =3D mdio_mux_read_c45;
> > >             if (parent_bus->write_c45)
> > >                     cb->mii_bus->write_c45 =3D mdio_mux_write_c45;
> > >
> > > My only objection to his patch (apart from the commit message which
> > > should indeed be more detailed) is that I would have preferred the sa=
me
> > > "if" syntax rather than the use of a ternary operator with NULL.
> >
> > I agree it could be fixed this way. But what i don't like about the
> > current code is how C22 and C45 do different things with error
> > codes. Since the current code is trying to use an error code, i would
> > prefer to fix that error code handling, rather than swap to a
> > different way to indicate its not supported.
> >
> >         Andrew
>
> You did write in commit da099a7fb13d ("net: phy: Remove probe_capabilitie=
s")
> that the MDIO bus API is now this: "Deciding if to probe of PHYs using
> C45 is now determine by if the bus provides the C45 read method."
>
> Do you not agree that Tim's approach is the more straightforward
> solution overall to skip C45 PHY probing, given this API, both code wise
> and runtime wise? Are there downsides to it?
>
> I have no objection to the C22 vs C45 error code handling inconsistency.
> It can be improved, sure. But it also does not matter here, if we agree
> that this problem can be sorted out in a more straightforward way with
> no negative consequences.
>
> I sort of don't understand the desire to have the smallest patch in
> terms of lines of code, when the end result will end up being suboptimal
> compared to something with just a little more lines (1 vs 4).


Andrew, would you feel differently if I added to the patch the same
logic for C22 ops? Perhaps that symmetry should have existed
in the initial patch, e.g.

    bus->read =3D chip->info->ops->phy_read
        ? mv88e6xxx_mdio_read : NULL;
    bus->write =3D chip->info->ops->phy_write
        ? mv88e6xxx_mdio_write : NULL;
    bus->read_c45 =3D chip->info->ops->phy_read_c45
        ? mv88e6xxx_mdio_read_c45 : NULL;
    bus->write_c45 =3D chip->info->ops->phy_write_c45
        ? mv88e6xxx_mdio_write_c45 : NULL;

Vladimir, as far as style I have no objections moving to straightlined
if's. I most prefer to follow the convention the rest of the code follows
and can change my patch accordingly.