[U-Boot,v1,1/4] arm, imx6: add mxc_iomux_set_gpr_register function

add mxc_iomux_set_gpr_register fucntion

Signed-off-by: Heiko Schocher <hs@denx.de>
Cc: Stefano Babic <sbabic@denx.de>
---
 arch/arm/imx-common/iomux-v3.c             | 15 +++++++++++++++
 arch/arm/include/asm/imx-common/iomux-v3.h |  2 ++
 2 files changed, 17 insertions(+)

Hi Heiko,

On 28/05/2014 12:16, Heiko Schocher wrote:
> add mxc_iomux_set_gpr_register fucntion
> 
> Signed-off-by: Heiko Schocher <hs@denx.de>
> Cc: Stefano Babic <sbabic@denx.de>
> ---
>  arch/arm/imx-common/iomux-v3.c             | 15 +++++++++++++++
>  arch/arm/include/asm/imx-common/iomux-v3.h |  2 ++
>  2 files changed, 17 insertions(+)
> 
> diff --git a/arch/arm/imx-common/iomux-v3.c b/arch/arm/imx-common/iomux-v3.c
> index 6e46ea8..8739b4c 100644
> --- a/arch/arm/imx-common/iomux-v3.c
> +++ b/arch/arm/imx-common/iomux-v3.c
> @@ -63,3 +63,18 @@ void imx_iomux_v3_setup_multiple_pads(iomux_v3_cfg_t const *pad_list,
>  	for (i = 0; i < count; i++)
>  		imx_iomux_v3_setup_pad(*p++);
>  }
> +
> +void mxc_iomux_set_gpr_register(int group, int start_bit, int num_bits,
> +				int value)
> +{
> +	int i = 0;
> +	u32 reg;
> +	reg = __raw_readl(base + group * 4);
> +	while (num_bits) {
> +		reg &= ~(1<<(start_bit + i));
> +		i++;
> +		num_bits--;
> +	}
> +	reg |= (value << start_bit);
> +	__raw_writel(reg, base + group * 4);
> +}

It is not clear to me why we need to add this or better, I want to avoid
having twice quite the same code doing the same ;-).

We have already a field for gpr:
struct iomuxc {
        u32 gpr[14];
        u32 omux[5];
        /* mux and pad registers */
};

+	reg = __raw_readl(base + group * 4);

this could not work when Freescale introduce a new variant with a slight
different layout. Thanks to Fabio, we know this is already reality:

	http://patchwork.ozlabs.org/patch/368336/

Your code *does* not work with the solox.

Generally, we access register and we use setbits_ and clrsetbits_ to
modify single bits. What is the advantage here ?

> diff --git a/arch/arm/include/asm/imx-common/iomux-v3.h b/arch/arm/include/asm/imx-common/iomux-v3.h
> index cca920b..4c83617 100644
> --- a/arch/arm/include/asm/imx-common/iomux-v3.h
> +++ b/arch/arm/include/asm/imx-common/iomux-v3.h
> @@ -174,5 +174,7 @@ typedef u64 iomux_v3_cfg_t;
>  void imx_iomux_v3_setup_pad(iomux_v3_cfg_t pad);
>  void imx_iomux_v3_setup_multiple_pads(iomux_v3_cfg_t const *pad_list,
>  				     unsigned count);
> +void mxc_iomux_set_gpr_register(int group, int start_bit, int num_bits,
> +				int value);
>  
>  #endif	/* __MACH_IOMUX_V3_H__*/
> 

Best regards,
Stefano Babic

Hello Stefano,

Am 11.07.2014 13:10, schrieb Stefano Babic:
> Hi Heiko,
>
> On 28/05/2014 12:16, Heiko Schocher wrote:
>> add mxc_iomux_set_gpr_register fucntion
>>
>> Signed-off-by: Heiko Schocher<hs@denx.de>
>> Cc: Stefano Babic<sbabic@denx.de>
>> ---
>>   arch/arm/imx-common/iomux-v3.c             | 15 +++++++++++++++
>>   arch/arm/include/asm/imx-common/iomux-v3.h |  2 ++
>>   2 files changed, 17 insertions(+)
>>
>> diff --git a/arch/arm/imx-common/iomux-v3.c b/arch/arm/imx-common/iomux-v3.c
>> index 6e46ea8..8739b4c 100644
>> --- a/arch/arm/imx-common/iomux-v3.c
>> +++ b/arch/arm/imx-common/iomux-v3.c
>> @@ -63,3 +63,18 @@ void imx_iomux_v3_setup_multiple_pads(iomux_v3_cfg_t const *pad_list,
>>   	for (i = 0; i<  count; i++)
>>   		imx_iomux_v3_setup_pad(*p++);
>>   }
>> +
>> +void mxc_iomux_set_gpr_register(int group, int start_bit, int num_bits,
>> +				int value)
>> +{
>> +	int i = 0;
>> +	u32 reg;
>> +	reg = __raw_readl(base + group * 4);
>> +	while (num_bits) {
>> +		reg&= ~(1<<(start_bit + i));
>> +		i++;
>> +		num_bits--;
>> +	}
>> +	reg |= (value<<  start_bit);
>> +	__raw_writel(reg, base + group * 4);
>> +}
>
> It is not clear to me why we need to add this or better, I want to avoid
> having twice quite the same code doing the same ;-).
>
> We have already a field for gpr:
> struct iomuxc {
>          u32 gpr[14];
>          u32 omux[5];
>          /* mux and pad registers */
> };

Ah!

> +	reg = __raw_readl(base + group * 4);
>
> this could not work when Freescale introduce a new variant with a slight
> different layout. Thanks to Fabio, we know this is already reality:
>
> 	http://patchwork.ozlabs.org/patch/368336/
>
> Your code *does* not work with the solox.
>
> Generally, we access register and we use setbits_ and clrsetbits_ to
> modify single bits. What is the advantage here ?

None, patch not longer needed. I remove it from the next patchserie.

Thanks!

bye,
Heiko