Message-ID: <Z8g6nWDe2cjumixt@kekkonen.localdomain>
Date: Wed, 5 Mar 2025 11:50:53 +0000
From: Sakari Ailus <sakari.ailus@...ux.intel.com>
To: Kieran Bingham <kieran.bingham@...asonboard.com>
Cc: Shravan.Chippa@...rochip.com, mchehab@...nel.org,
	linux-media@...r.kernel.org, linux-kernel@...r.kernel.org,
	Conor.Dooley@...rochip.com, Valentina.FernandezAlanis@...rochip.com,
	Praveen.Kumar@...rochip.com
Subject: Re: [PATCH V7 4/4] media: i2c: imx334: add modes for 720p and 480p
 resolutions

Hi Kieran, Shravan,

On Wed, Mar 05, 2025 at 10:45:42AM +0000, Kieran Bingham wrote:
> Quoting Shravan.Chippa@...rochip.com (2025-03-05 10:22:12)
> > Hi Kieran
> > 
> > > -----Original Message-----
> > > From: Kieran Bingham <kieran.bingham@...asonboard.com>
> > > Sent: Wednesday, March 5, 2025 3:05 PM
> > > To: mchehab@...nel.org; sakari.ailus@...ux.intel.com; shravan Chippa -
> > > I35088 <Shravan.Chippa@...rochip.com>
> > > Cc: linux-media@...r.kernel.org; linux-kernel@...r.kernel.org; Conor Dooley
> > > - M52691 <Conor.Dooley@...rochip.com>; Valentina Fernandez Alanis -
> > > M63239 <Valentina.FernandezAlanis@...rochip.com>; Praveen Kumar -
> > > I30718 <Praveen.Kumar@...rochip.com>; shravan Chippa - I35088
> > > <Shravan.Chippa@...rochip.com>
> > > Subject: Re: [PATCH V7 4/4] media: i2c: imx334: add modes for 720p and 480p
> > > resolutions
> > > 
> > > EXTERNAL EMAIL: Do not click links or open attachments unless you know the
> > > content is safe
> > > 
> > > Quoting shravan kumar (2025-03-05 05:14:42)
> > > > From: Shravan Chippa <shravan.chippa@...rochip.com>
> > > >
> > > > Added support for 1280x720@30 and 640x480@30 resolutions
> > > >
> > > > Signed-off-by: Shravan Chippa <shravan.chippa@...rochip.com>
> > > > ---
> > > >  drivers/media/i2c/imx334.c | 66
> > > > ++++++++++++++++++++++++++++++++++++++
> > > >  1 file changed, 66 insertions(+)
> > > >
> > > > diff --git a/drivers/media/i2c/imx334.c b/drivers/media/i2c/imx334.c
> > > > index a7c0bd38c9b8..8cd1eecd0143 100644
> > > > --- a/drivers/media/i2c/imx334.c
> > > > +++ b/drivers/media/i2c/imx334.c
> > > > @@ -314,6 +314,46 @@ static const struct imx334_reg
> > > common_mode_regs[] = {
> > > >         {0x3002, 0x00},
> > > >  };
> > > >
> > > > +/* Sensor mode registers for 640x480@...ps */ static const struct
> > > > +imx334_reg mode_640x480_regs[] = {
> > > > +       {0x302c, 0x70},
> > > > +       {0x302d, 0x06},
> > > 
> > > These two are a single 16 bit register HTRIMMING_START = 1648
> > > 
> > > > +       {0x302e, 0x80},
> > > > +       {0x302f, 0x02},
> > > 
> > > These two are a single 16 bit register HNUM = 640
> > > 
> > > > +       {0x3074, 0x48},
> > > > +       {0x3075, 0x07},
> > > 
> > > These two are a single 16 bit (well, 12 bit value) AREA3_ST_ADR_1 = 1864
> > > 
> > > > +       {0x308e, 0x49},
> > > > +       {0x308f, 0x07},
> > > 
> > > These two are a single 16 bit register AREA3_ST_ADR_2 = 1865
> > > 
> > > > +       {0x3076, 0xe0},
> > > > +       {0x3077, 0x01},
> > > 
> > > These two are a single 16 bit register AREA3_WIDTH_1 = 480
> > > 
> > > > +       {0x3090, 0xe0},
> > > > +       {0x3091, 0x01},
> > > 
> > > These two are a single 16 bit register AREA3_WIDTH_2 = 480
> > > 
> > > > +       {0x3308, 0xe0},
> > > > +       {0x3309, 0x01},
> > > 
> > > These two are a single 16 bit register Y_OUT_SIZE
> > > 
> > > Don't you think
> > >         { Y_OUT_SIZE, 480 },
> > > 
> > > Is so much more readable and easier to comprehend and maintain?
> > > 
> > > 
> > > > +       {0x30d8, 0x30},
> > > > +       {0x30d9, 0x0b},
> > > 
> > > These two are a single 16 bit register UNREAD_ED_ADR = 2864
> > > 
> > > > +};
> > > 
> > > I'm still sad that we can all know the names of all these registers and yet this
> > > is writing new tables of hex values.
> > 
> > Do you want me use call like bellow API with register names:
> > CCI_REG16_LE(0x30d8);
> > cci_write();
> > cci_multi_reg_write();
> > devm_cci_regmap_init_i2c();
>  
> 
> Yes please, I would want that very much. I'm not very good at reading
> and storing hex values in my head! That's why I broke down the above
> registers to strings and decimal values.

I agree, it'd be good to do these while changing the driver now. I think it
could be done after adding the new modes, now that the patches already
exist.

> 
> 
> The discussions at
> https://lore.kernel.org/all/PH0PR11MB5611FD22CF6E12F7226FA9C081E12@PH0PR11MB5611.namprd11.prod.outlook.com/
> reports the full datasheet, and I stated:
> 
> 
> > > > This is an enormous amount of duplicated data that could be factored
> > > > out.
> > > >
> > > > These are also /very/ common against the existing mode register
> > > > tables too.
> > > >
> > > > I think several things need to happen in this driver:
> > > >
> > > >  1. It should be converted to use the CCI helpers.
> > > >  2. Whereever identifiable, the register names should be used
> > > > instead of
> > > >     just the addresses.
> > > >  3. The common factors of these tables should be de-duplicated.
> > > >
> > > > In your additions you only have differences in the following
> > > > registers from those entire tables:
> 
> You replied with
> 
> > >
> > > I will try to optimize camera resolution array register value usage by
> > > writing common register array values.
> 
> which you have done (point 3), and is great progress in the series.
> 
> Further down in the thread I stated:
> 
> 
> > > >  4. And ideally - the differences which determine the modes should
> > > > be
> > > >     factored out to calculations so that we are not writing out
> > > > large
> > > >     tables just to write a parameterised frame size.
> > > >
> > > >
> > > > I would beleive that at least steps 1 and 3 would be achievable, 2
> > > > and 4 would depend upon access to the datasheet.
> > > >
> 
> 
> I still believe steps 1 is important to this development. You have done
> step 3 I think. And now both step 2 and (later) step 4 are possible as we
> have the datasheet.

In this case the datasheet doesn't appear to be documenting the PLL.

-- 
Regards,

Sakari Ailus

Powered by blists - more mailing lists