Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Vitor Soares <Vitor.Soares@synopsys.com>
To:     Boris Brezillon <boris.brezillon@collabora.com>,
        Vitor Soares <Vitor.Soares@synopsys.com>
CC:     "linux-i3c@lists.infradead.org" <linux-i3c@lists.infradead.org>,
        "joao.pinto@synopsys.com" <Joao.Pinto@synopsys.com>,
        Boris Brezillon <bbrezillon@kernel.org>,
        "stable@vger.kernel.org" <stable@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: RE: [PATCH 1/3] i3c: fix i2c and i3c scl rate by bus mode
Thread-Topic: [PATCH 1/3] i3c: fix i2c and i3c scl rate by bus mode
Thread-Index: AQHU87sLhgtQ1+d000+4YAeSeMHzMaY+J6iAgACom0D//+7fAIAJjifg///0oYCAADajAP//8I4AAAU08HA=
Date:   Mon, 22 Apr 2019 19:59:46 +0000
Message-ID: <13D59CF9CEBAF94592A12E8AE55501350A61C658@DE02WEMBXB.internal.synopsys.com>
References: <cover.1555350118.git.vitor.soares@synopsys.com>
        <05fdeea79db83970e9ecb0d7045b4dd98f206f06.1555350118.git.vitor.soares@synopsys.com>
        <20190416075041.22f8e849@collabora.com>
        <13D59CF9CEBAF94592A12E8AE55501350A61596D@DE02WEMBXB.internal.synopsys.com>
        <20190416165250.0606e5a2@collabora.com>
        <13D59CF9CEBAF94592A12E8AE55501350A61B3B4@DE02WEMBXB.internal.synopsys.com>
        <20190422180715.40abe1b9@collabora.com>
        <13D59CF9CEBAF94592A12E8AE55501350A61C415@DE02WEMBXB.internal.synopsys.com>
 <20190422202732.0d539af3@collabora.com>
In-Reply-To: <20190422202732.0d539af3@collabora.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

From: Boris Brezillon <boris.brezillon@collabora.com>
Date: Mon, Apr 22, 2019 at 19:27:32

> On Mon, 22 Apr 2019 17:54:29 +0000
> Vitor Soares <Vitor.Soares@synopsys.com> wrote:
>=20
> > > > > > >      =20
> > > > > > > >  {
> > > > > > > >  	i3cbus->mode =3D mode;
> > > > > > > > =20
> > > > > > > > -	if (!i3cbus->scl_rate.i3c)
> > > > > > > > -		i3cbus->scl_rate.i3c =3D I3C_BUS_TYP_I3C_SCL_RATE;
> > > > > > > > -
> > > > > > > > -	if (!i3cbus->scl_rate.i2c) {
> > > > > > > > -		if (i3cbus->mode =3D=3D I3C_BUS_MODE_MIXED_SLOW)
> > > > > > > > -			i3cbus->scl_rate.i2c =3D I3C_BUS_I2C_FM_SCL_RATE;
> > > > > > > > -		else
> > > > > > > > -			i3cbus->scl_rate.i2c =3D I3C_BUS_I2C_FM_PLUS_SCL_RATE;
> > > > > > > > +	switch (i3cbus->mode) {
> > > > > > > > +	case I3C_BUS_MODE_PURE:
> > > > > > > > +		if (!i3cbus->scl_rate.i3c)
> > > > > > > > +			i3cbus->scl_rate.i3c =3D I3C_BUS_TYP_I3C_SCL_RATE;
> > > > > > > > +		break;
> > > > > > > > +	case I3C_BUS_MODE_MIXED_FAST:
> > > > > > > > +		if (!i3cbus->scl_rate.i3c)
> > > > > > > > +			i3cbus->scl_rate.i3c =3D I3C_BUS_TYP_I3C_SCL_RATE;
> > > > > > > > +		if (!i3cbus->scl_rate.i2c)
> > > > > > > > +			i3cbus->scl_rate.i2c =3D i2c_scl_rate;
> > > > > > > > +		break;
> > > > > > > > +	case I3C_BUS_MODE_MIXED_SLOW:
> > > > > > > > +		if (!i3cbus->scl_rate.i2c)
> > > > > > > > +			i3cbus->scl_rate.i2c =3D i2c_scl_rate;
> > > > > > > > +		i3cbus->scl_rate.i3c =3D i3cbus->scl_rate.i2c;     =20
> > > > > > >=20
> > > > > > > Maybe we should do
> > > > > > >=20
> > > > > > > 		if (!i3cbus->scl_rate.i3c ||
> > > > > > > 		    i3cbus->scl_rate.i3c > i3cbus->scl_rate.i2c)
> > > > > > > 			i3cbus->scl_rate.i3c =3D i3cbus->scl_rate.i2c;
> > > > > > > 					  =20
> > > > > > > Just in case the I3C rate forced by the user is lower than th=
e max I2C
> > > > > > > rate.     =20
> > > > > >=20
> > > > > > That was something that I considered but TBH it isn't a real us=
e case.   =20
> > > > >=20
> > > > > Add a WARN_ON() to at least catch such inconsistencies. And maybe=
 we
> > > > > should add a dev_warn() when the user-defined rates do not match
> > > > > the mode/LVR constraints. It's easy to do a mistake when writing =
a dts.   =20
> > > >=20
> > > > I think the WARN_ON() is too evasive on the screen and won't provid=
e the=20
> > > > information we want.
> > > > The dev_warn() should work perfectly here.
> > > >=20
> > > > 		if (i3cbus->scl_rate.i3c < i3cbus->scl_rate.i2c)
> > > > 			dev_warn(&i3cbus->cur_master->dev->dev,
> > > > 				     "%s: i3c-scl-hz lower then i2c-scl-hz\n",  __func__); =20
> > >=20
> > > Using dev_warn() sounds good, though I don't think you need the
> > > __func__ here. Also, please print the i2c/i3c rates in the message, a=
nd
> > > align the second line on the open parens.
> > >  =20
> > > > 		if (i3cbus->scl_rate.i2c !=3D I3C_BUS_I2C_FM_SCL_RATE ||
> > > > 		    i3cbus->scl_rate.i2c !=3D I3C_BUS_I2C_FM_PLUS_SCL_RATE)
> > > > 			dev_warn(&i3cbus->cur_master->dev->dev,
> > > > 				     "%s: i2c-scl-hz not defined according MIPI I3C spec\n",=20
> > > > 				     __func__); =20
> > >=20
> > > Is that really a problem? Having an i2c rate that is less than FM spe=
ed
> > > sounds like a valid case to me. =20
> >=20
> > I'm addressing the spec constrains.
>=20
> "Table 57 I3C Timing Requirements When Communicating With I2C Legacy
> Devices" says that freq can be between 0 and 400KHz when operating in
> slow(FM) mode.

That is the definition of Fast-mode:
"Fast-mode devices can receive and transmit at up to 400kbit/s" as per=20
i2c spec.
So this way the slaves can be connected to Standard Mode bus... you=20
already know that.

For the FM requirements in I3C bus, please refer to I3C FAQ Q2.2 and=20
table 56 of I3C v1.0 basic spec.

> Yes, maximum rate when not specified otherwise is
> 400Khz,=20

By default you will have always FM or FM+ due the LVR register be=20
mandatory.

> but the point of overloading the max I2C/I3C spec is to allow
> custom rates when the default/spec ones are not achievable, so I'm not
> sure complaining in that case is legitimate.
>=20

Well, if the users aren't able to achieve at least FM is because=20
something is not correct.

> We should definitely complain when one tries to set a maximum rate that
> is higher than what devices can do (i3cbus->scl_rate.i2c >
> max_i2c_scl_rate).
>=20

I can put another if() for such cases.

> Same goes for I3C communications, we shouldn't care when the forced rate
> is lower than what the bus is capable of, what's important is to
> complain when it's higher than what's supported.

The point is complain when scl_rate.i3c < scl_rate.i2c because it can be=20
error from user.

Best regards,
Vitor Soares