Re: [PATCH v2 2/3] rcar-csi2: Update start procedure for H3 ES2

["Niklas Söderlund" <niklas.soderlund@ragnatech.se>]

Hi Laurent,

Thanks for your feedback.

On 2019-03-11 11:27:25 +0200, Laurent Pinchart wrote:
> Hi Niklas,
> 
> Thank you for the patch.
> 
> On Sat, Mar 09, 2019 at 12:57:01AM +0100, Niklas Söderlund wrote:
> > Latest information from hardware engineers reveals that H3 ES2 and ES3
> > behave differently when working with link speeds bellow 250 Mpbs.
> > Add a SoC match for H3 ES2.* and use the correct startup sequence.
> > 
> > Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> > Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
> > ---
> >  drivers/media/platform/rcar-vin/rcar-csi2.c | 39 ++++++++++++++++++---
> >  1 file changed, 35 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> > index 6be81d4839f35a0e..07d5c8c66b7cd382 100644
> > --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> > +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> > @@ -914,6 +914,25 @@ static int rcsi2_init_phtw_h3_v3h_m3n(struct rcar_csi2 *priv, unsigned int mbps)
> >  	return rcsi2_phtw_write_array(priv, step2);
> >  }
> >  
> > +static int rcsi2_init_phtw_h3es2(struct rcar_csi2 *priv, unsigned int mbps)
> > +{
> > +	static const struct phtw_value step1[] = {
> > +		{ .data = 0xcc, .code = 0xe2 },
> > +		{ .data = 0x01, .code = 0xe3 },
> > +		{ .data = 0x11, .code = 0xe4 },
> > +		{ .data = 0x01, .code = 0xe5 },
> > +		{ .data = 0x10, .code = 0x04 },
> > +		{ .data = 0x38, .code = 0x08 },
> > +		{ .data = 0x01, .code = 0x00 },
> > +		{ .data = 0x4b, .code = 0xac },
> > +		{ .data = 0x03, .code = 0x00 },
> > +		{ .data = 0x80, .code = 0x07 },
> > +		{ /* sentinel */ },
> > +	};
> > +
> > +	return rcsi2_phtw_write_array(priv, step1);
> 
> Another option would have been to condition the mbps check in
> rcsi2_init_phtw_h3_v3h_m3n() to the ES version, which would save a bit
> of memory as we could remove the above table, but we would need to add a
> field to the rcar_csi2_info structure so we may not save much in the
> end.

I agree this would not be my preferred solution.

> 
> I wonder, however, if you could move the step1 and step2 tables out of
> rcsi2_init_phtw_h3_v3h_m3n() and reuse them here, or possibly create a
> __rcsi2_init_phtw_h3_v3h_m3n() with two rcsi2_init_phtw_h3_v3h_m3n() and
> wrappers rcsi2_init_phtw_h3es2() wrappers.

This feels like a nicer solution, will use it in next version. Thanks 
for the good suggestion.

> 
> > +}
> > +
> >  static int rcsi2_init_phtw_v3m_e3(struct rcar_csi2 *priv, unsigned int mbps)
> >  {
> >  	return rcsi2_phtw_write_mbps(priv, mbps, phtw_mbps_v3m_e3, 0x44);
> > @@ -976,6 +995,14 @@ static const struct rcar_csi2_info rcar_csi2_info_r8a7795es1 = {
> >  	.num_channels = 4,
> >  };
> >  
> > +static const struct rcar_csi2_info rcar_csi2_info_r8a7795es2 = {
> > +	.init_phtw = rcsi2_init_phtw_h3es2,
> > +	.hsfreqrange = hsfreqrange_h3_v3h_m3n,
> > +	.csi0clkfreqrange = 0x20,
> > +	.num_channels = 4,
> > +	.clear_ulps = true,
> > +};
> > +
> >  static const struct rcar_csi2_info rcar_csi2_info_r8a7796 = {
> >  	.hsfreqrange = hsfreqrange_m3w_h3es1,
> >  	.num_channels = 4,
> > @@ -1041,11 +1068,15 @@ static const struct of_device_id rcar_csi2_of_table[] = {
> >  };
> >  MODULE_DEVICE_TABLE(of, rcar_csi2_of_table);
> >  
> > -static const struct soc_device_attribute r8a7795es1[] = {
> > +static const struct soc_device_attribute r8a7795[] = {
> >  	{
> >  		.soc_id = "r8a7795", .revision = "ES1.*",
> >  		.data = &rcar_csi2_info_r8a7795es1,
> >  	},
> > +	{
> > +		.soc_id = "r8a7795", .revision = "ES2.*",
> > +		.data = &rcar_csi2_info_r8a7795es2,
> > +	},
> >  	{ /* sentinel */ },
> >  };
> >  
> > @@ -1063,10 +1094,10 @@ static int rcsi2_probe(struct platform_device *pdev)
> >  	priv->info = of_device_get_match_data(&pdev->dev);
> >  
> >  	/*
> > -	 * r8a7795 ES1.x behaves differently than the ES2.0+ but doesn't
> > -	 * have it's own compatible string.
> > +	 * The different ES versions of r8a7795 (H3) behave differently but
> > +	 * share the same compatible string.
> >  	 */
> > -	attr = soc_device_match(r8a7795es1);
> > +	attr = soc_device_match(r8a7795);
> >  	if (attr)
> >  		priv->info = attr->data;
> >  
> 
> -- 
> Regards,
> 
> Laurent Pinchart

-- 
Regards,
Niklas Söderlund