[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <a1380e68485ace5125031c1123896355936372d1.camel@phytec.de>
Date: Mon, 5 Jul 2021 07:42:01 +0000
From: Stefan Riedmüller <S.Riedmueller@...tec.de>
To: "sakari.ailus@...ux.intel.com" <sakari.ailus@...ux.intel.com>
CC: "mchehab@...nel.org" <mchehab@...nel.org>,
"laurent.pinchart@...asonboard.com"
<laurent.pinchart@...asonboard.com>,
"devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
"linux-media@...r.kernel.org" <linux-media@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"robh+dt@...nel.org" <robh+dt@...nel.org>,
"enrico.scholz@...ma-chemnitz.de" <enrico.scholz@...ma-chemnitz.de>
Subject: Re: [PATCH v3 1/6] media: mt9p031: Read back the real clock rate
Hi Sakari,
On Mon, 2021-07-05 at 10:13 +0300, Sakari Ailus wrote:
> Hi Enrico,
>
> On Fri, Jul 02, 2021 at 11:59:17AM +0200, Stefan Riedmueller wrote:
> > From: Enrico Scholz <enrico.scholz@...ma-chemnitz.de>
> >
> > The real and requested clock can differ and because it is used to
> > calculate PLL values, the real clock rate should be read.
>
> Do you have a system where this happens? That suggests there's a wrong
> value in DT.
The use case here is when the clock is supplied by one of the clock outputs of
a SOC which might not hit the requested frequency exactly due to internal PLL
configuration. So to get a better pixel clock the actual clock rate is read to
calculate the PLL parameters on the sensor. At least that's the idea.
Regards,
Stefan
>
> The preference nowadays is to rely on assigned-clock-rates, even though
> it's inherently somewhat unreliable, just as clk_set_rate(). This is an
> existing driver though. The old ones could be kept for compatibility with
> older DT binaries.
>
> > Signed-off-by: Enrico Scholz <enrico.scholz@...ma-chemnitz.de>
> > Signed-off-by: Stefan Riedmueller <s.riedmueller@...tec.de>
> > Reviewed-by: Laurent Pinchart <laurent.pinchart@...asonboard.com>
> > ---
> > drivers/media/i2c/mt9p031.c | 9 ++++++---
> > 1 file changed, 6 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/media/i2c/mt9p031.c b/drivers/media/i2c/mt9p031.c
> > index 77567341ec98..3eaaa8d44523 100644
> > --- a/drivers/media/i2c/mt9p031.c
> > +++ b/drivers/media/i2c/mt9p031.c
> > @@ -255,6 +255,7 @@ static int mt9p031_clk_setup(struct mt9p031 *mt9p031)
> >
> > struct i2c_client *client = v4l2_get_subdevdata(&mt9p031->subdev);
> > struct mt9p031_platform_data *pdata = mt9p031->pdata;
> > + unsigned long ext_freq;
> > int ret;
> >
> > mt9p031->clk = devm_clk_get(&client->dev, NULL);
> > @@ -265,13 +266,15 @@ static int mt9p031_clk_setup(struct mt9p031
> > *mt9p031)
> > if (ret < 0)
> > return ret;
> >
> > + ext_freq = clk_get_rate(mt9p031->clk);
> > +
> > /* If the external clock frequency is out of bounds for the PLL use
> > the
> > * pixel clock divider only and disable the PLL.
> > */
> > - if (pdata->ext_freq > limits.ext_clock_max) {
> > + if (ext_freq > limits.ext_clock_max) {
> > unsigned int div;
> >
> > - div = DIV_ROUND_UP(pdata->ext_freq, pdata->target_freq);
> > + div = DIV_ROUND_UP(ext_freq, pdata->target_freq);
> > div = roundup_pow_of_two(div) / 2;
> >
> > mt9p031->clk_div = min_t(unsigned int, div, 64);
> > @@ -280,7 +283,7 @@ static int mt9p031_clk_setup(struct mt9p031 *mt9p031)
> > return 0;
> > }
> >
> > - mt9p031->pll.ext_clock = pdata->ext_freq;
> > + mt9p031->pll.ext_clock = ext_freq;
> > mt9p031->pll.pix_clock = pdata->target_freq;
> > mt9p031->use_pll = true;
> >
Powered by blists - more mailing lists