Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
MIME-Version: 1.0
References: <20220326102229.421718-1-tanure@linux.com> <20220326102229.421718-3-tanure@linux.com>
 <c789adcd-d072-bec2-a823-5f5993704365@baylibre.com>
In-Reply-To: <c789adcd-d072-bec2-a823-5f5993704365@baylibre.com>
Reply-To: tanure@linux.com
From:   Lucas Tanure <tanure@linux.com>
Date:   Fri, 8 Apr 2022 08:18:14 +0100
Message-ID: <CAJX_Q+3WHCKmg3Fq5jPTY=PNjow2HdLkgFVLd9GQRuGg_q96=Q@mail.gmail.com>
Subject: Re: [PATCH 2/3] i2c: meson: Use 50% duty cycle for I2C clock
To:     Neil Armstrong <narmstrong@baylibre.com>
Cc:     Kevin Hilman <khilman@baylibre.com>,
        Jerome Brunet <jbrunet@baylibre.com>,
        Martin Blumenstingl <martin.blumenstingl@googlemail.com>,
        linux-i2c@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        linux-amlogic@lists.infradead.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Apr 6, 2022 at 12:31 PM Neil Armstrong <narmstrong@baylibre.com> wrote:
>
> Hi,
>
> On 26/03/2022 11:22, Lucas Tanure wrote:
> > The duty cycle of 33% is less than the required
> > by the I2C specs for the LOW period of the SCL
> > clock.
> >
> > Move the duty cyle to 50% for 100Khz or lower
> > clocks, and (40% High SCL / 60% Low SCL) duty
> > cycle for clocks above 100Khz.
> >
> > Signed-off-by: Lucas Tanure <tanure@linux.com>
> > ---
> >   drivers/i2c/busses/i2c-meson.c | 45 +++++++++++++++++++++++++---------
> >   1 file changed, 33 insertions(+), 12 deletions(-)
> >
> > diff --git a/drivers/i2c/busses/i2c-meson.c b/drivers/i2c/busses/i2c-meson.c
> > index 4b4a5b2d77ab..b913ba20f06e 100644
> > --- a/drivers/i2c/busses/i2c-meson.c
> > +++ b/drivers/i2c/busses/i2c-meson.c
> > @@ -140,29 +140,50 @@ static void meson_i2c_add_token(struct meson_i2c *i2c, int token)
> >   static void meson_i2c_set_clk_div(struct meson_i2c *i2c, unsigned int freq)
> >   {
> >       unsigned long clk_rate = clk_get_rate(i2c->clk);
> > -     unsigned int div;
> > +     unsigned int div_h, div_l;
> >
> > -     div = DIV_ROUND_UP(clk_rate, freq);
> > -     div -= FILTER_DELAY;
> > -     div = DIV_ROUND_UP(div, i2c->data->div_factor);
> > +     if (freq <= 100000) {
>
> You should use I2C_MAX_STANDARD_MODE_FREQ instead here
>
> > +             div_h = DIV_ROUND_UP(clk_rate, freq);
> > +             div_l = DIV_ROUND_UP(div_h, 4);
> > +             div_h = DIV_ROUND_UP(div_h, 2) - FILTER_DELAY;
> > +     } else {
> > +     /* According to I2C-BUS Spec 2.1, in FAST-MODE, the minimum LOW period is 1.3uS, and
> > +      * minimum HIGH is least 0.6us.
> > +      * For 400000 freq, the period is 2.5us. To keep within the specs, give 40% of period to
> > +      * HIGH and 60% to LOW. This means HIGH at 1.0us and LOW 1.5us.
> > +      * The same applies for Fast-mode plus, where LOW is 0.5us and HIGH is 0.26us.
> > +      * Duty = H/(H + L) = 2/5
> > +      */
>
> Please move the comment before the if()
>
> > +             div_h = DIV_ROUND_UP(clk_rate * 2, freq * 5) - FILTER_DELAY;
> > +             div_l = DIV_ROUND_UP(clk_rate * 3, freq * 5 * 2);
> > +     }
> >
> >       /* clock divider has 12 bits */
> > -     if (div > GENMASK(11, 0)) {
> > +     if (div_h > GENMASK(11, 0)) {
> >               dev_err(i2c->dev, "requested bus frequency too low\n");
> > -             div = GENMASK(11, 0);
> > +             div_h = GENMASK(11, 0);
> > +     }
> > +     if (div_l > GENMASK(11, 0)) {
> > +             dev_err(i2c->dev, "requested bus frequency too low\n");
> > +             div_l = GENMASK(11, 0);
> >       }
> >
> >       meson_i2c_set_mask(i2c, REG_CTRL, REG_CTRL_CLKDIV_MASK,
> > -                        FIELD_PREP(REG_CTRL_CLKDIV_MASK, div & GENMASK(9, 0)));
> > +                        FIELD_PREP(REG_CTRL_CLKDIV_MASK, div_h & GENMASK(9, 0)));
> >
> >       meson_i2c_set_mask(i2c, REG_CTRL, REG_CTRL_CLKDIVEXT_MASK,
> > -                        FIELD_PREP(REG_CTRL_CLKDIVEXT_MASK, div >> 10));
> > +                        FIELD_PREP(REG_CTRL_CLKDIVEXT_MASK, div_h >> 10));
> > +
> > +
> > +     /* set SCL low delay */
> > +     meson_i2c_set_mask(i2c, REG_SLAVE_ADDR, REG_SLV_SCL_LOW_MASK,
> > +                        (div_l << REG_SLV_SCL_LOW_SHIFT) & REG_SLV_SCL_LOW_MASK);
>
> You could use FIELD_PREP() here
>
> >
> > -     /* Disable HIGH/LOW mode */
> > -     meson_i2c_set_mask(i2c, REG_SLAVE_ADDR, REG_SLV_SCL_LOW_EN, 0);
> > +     /* Enable HIGH/LOW mode */
> > +     meson_i2c_set_mask(i2c, REG_SLAVE_ADDR, REG_SLV_SCL_LOW_EN, REG_SLV_SCL_LOW_EN);
> >
> > -     dev_dbg(i2c->dev, "%s: clk %lu, freq %u, div %u\n", __func__,
> > -             clk_rate, freq, div);
> > +     dev_dbg(i2c->dev, "%s: clk %lu, freq %u, divh %u, divl %u\n", __func__,
> > +             clk_rate, freq, div_h, div_l);
> >   }
> >
> >   static void meson_i2c_get_data(struct meson_i2c *i2c, char *buf, int len)
>
> I looked at different amlogic downstream sources, and those match the recommended
> calculations.
>
> So with the legacy back for Meson6, it will be OK.
>
> Neil

Ok, I will do the modifications this weekend.
I only tested i2c3 and i2c_ao and will execute more measures and show
the results.

Thanks
Lucas