Re: [PATCH v2 5/5] arm64: dts: renesas: r9a07g043u11-smarc-cru-csi-ov5645: Reduce I2C frequency

[Sakari Ailus <sakari.ailus@linux.intel.com>]

Hi folks,

On Fri, Feb 09, 2024 at 02:36:22PM +0000, Biju Das wrote:
> Hi Geert,
> 
> > -----Original Message-----
> > From: Biju Das
> > Sent: Tuesday, January 30, 2024 2:15 PM
> > Subject: RE: [PATCH v2 5/5] arm64: dts: renesas: r9a07g043u11-smarc-cru-
> > csi-ov5645: Reduce I2C frequency
> > 
> > Hi Geert,
> > 
> > > -----Original Message-----
> > > From: Biju Das
> > > Sent: Friday, January 26, 2024 3:57 PM
> > > Subject: RE: [PATCH v2 5/5] arm64: dts: renesas:
> > > r9a07g043u11-smarc-cru-
> > > csi-ov5645: Reduce I2C frequency
> > >
> > > Hi Geert,
> > >
> > > Thanks for the feedback.
> > >
> > > > -----Original Message-----
> > > > From: Geert Uytterhoeven <geert@linux-m68k.org>
> > > > Sent: Friday, January 26, 2024 1:53 PM
> > > > Subject: Re: [PATCH v2 5/5] arm64: dts: renesas:
> > > > r9a07g043u11-smarc-cru-
> > > > csi-ov5645: Reduce I2C frequency
> > > >
> > > > Hi Biju,
> > > >
> > > > On Fri, Jan 26, 2024 at 2:31 PM Biju Das
> > > > <biju.das.jz@bp.renesas.com>
> > > > wrote:
> > > > > Reduce i2c freq from 400->100 kHz on RZ/G2UL SMARC EVK as the
> > > > > captured image is not proper with the sensor configuration over I2C.
> > > > >
> > > > > Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
> > > >
> > > > Thanks for your patch!
> > > >
> > > > > ---
> > > > > a/arch/arm64/boot/dts/renesas/r9a07g043u11-smarc-cru-csi-ov5645.dt
> > > > > so
> > > > > +++ b/arch/arm64/boot/dts/renesas/r9a07g043u11-smarc-cru-csi-ov5645.
> > > > > +++ dt
> > > > > +++ so
> > > > > @@ -19,3 +19,7 @@ &ov5645 {
> > > > >         enable-gpios = <&pinctrl RZG2L_GPIO(4, 4) GPIO_ACTIVE_HIGH>;
> > > > >         reset-gpios = <&pinctrl RZG2L_GPIO(0, 1) GPIO_ACTIVE_LOW>;
> > > > > };
> > > > > +
> > > > > +&i2c0 {
> > > > > +       clock-frequency = <100000>; };
> > > >
> > > > Is this a limitation of one of the I2C devices on the bus, or a PCB
> > > > design issue?
> > >
> > > Currently versa3 clock generator connected to the same bus and it
> > > works ok with 400kHz clock. Maybe it is stressed not that much
> > > compared to OV5645 sensor configuration.
> > >
> > > At the moment with 400kHz I2C bus clock, Camera capture is not working
> > > properly on RZ/G2UL, but with same bus frequency the same works fine
> > > on RZ/{G2L,G2LC,V2L}.
> > > There may be some hardware differences which is causing this issue.
> > >
> > > >
> > > > Doesn't this need a Fixes tag?
> > >
> > > I can create a new patch updating bus frequency as 100kHz and add
> > > fixes tag.
> > > After this I will drop this patch as it no longer needed.
> > >
> > > Please let me know.
> > 
> > + media
> > 
> > Adding a delay after Software reset makes it to work at 400kHz with
> > RZ/G2UL SMARC EVK.
> > 
> > So not sure we need to add delay after software reset?
> > 
> > Now after OV5645 gpio reset, then there is 20msec delay and then again we
> > are issuing software reset and there is no delay after this for this
> > software reset.
> > 
> > diff --git a/drivers/media/i2c/ov5645.c b/drivers/media/i2c/ov5645.c index
> > a26ac11c989d..d67a5e23fe5a 100644
> > --- a/drivers/media/i2c/ov5645.c
> > +++ b/drivers/media/i2c/ov5645.c
> > @@ -622,11 +622,19 @@ static int ov5645_set_register_array(struct ov5645
> > *ov5645,  {
> >         unsigned int i;
> >         int ret;
> > 
> >         for (i = 0; i < num_settings; ++i, ++settings) {
> >                 ret = ov5645_write_reg(ov5645, settings->reg, settings-
> > >val);
> >                 if (ret < 0)
> >                         return ret;
> > +
> > +               if (settings->reg == OV5645_SYSTEM_CTRL0)
> > +                       fsleep(1000);
> > 
> 
> This issue seen on RZ/G2L SMARC EVK as well. My testing on G2L family shows
> we need to add delay to make OV5645 to work @400kHZ.

Typically there's a delay before the chip is accessible over I²C after
resetting it. It's a bit open whether this one needs it, very probably it
does. It'd be nicer nonetheless to do this outside the register list and
instead use a separate ov5645_write_reg() call.

Probably the first write is redundant. The second write resets the device.
0x80 should be sufficient value for that.

-- 
Regards,

Sakari Ailus