[PATCH 6/9] ARM: pxa: support pxa95x

[Haojian Zhuang]

On Tue, Nov 9, 2010 at 1:00 AM, Eric Miao <eric.y.miao at gmail.com> wrote:
> On Mon, Nov 8, 2010 at 8:37 PM, Haojian Zhuang
> <haojian.zhuang at marvell.com> wrote:
>> The core of PXA955 is PJ4. Add new PJ4 support. And add new macro
>> CONFIG_PXA95x. Support PXA955 on Marvell Saarb handheld platform.
>>
>> Signed-off-by: Haojian Zhuang <haojian.zhuang at marvell.com>
>> Cc: Eric Miao <eric.y.miao at gmail.com>
>> ---
>> +config MACH_SAARB
>> +       bool "PXA955 Handheld Platform (aka SAARB)"
>> +       depends on !(CPU_XSCALE || CPU_XSC3)
>> +       select CPU_PXA955
>> +
>
> It would be better to separte the introduction of saarb board support
> into another patch.
>
OK

>> +config CPU_PXA955
>> +       bool
>> +       select PXA95x
>>        help
>> -         PXA950 (codename Tavor-PV2)
>> +         PXA955 (codename MG1)
>
> Wondering how much difference is the PXA955 from PXA95x.
>
Actually, I prefer to call it MG1 and MGx directly. Since the next
chip will be called PXA96x, but it's the same family of PXA955.

>
> The above changes seem to be moved their positions in the file, and is
> not necessary, could you check again?
>
Sure. I'll check it.

>>
>>  /* pxa2xx-spi platform-device ID equals respective SSP platform-device ID + 1.
>>  * See comment in arch/arm/mach-pxa/ssp.c::ssp_probe() */
>> diff --git a/arch/arm/mach-pxa/generic.c b/arch/arm/mach-pxa/generic.c
>> index 0ff785e..3c9e426 100644
>> --- a/arch/arm/mach-pxa/generic.c
>> +++ b/arch/arm/mach-pxa/generic.c
>> @@ -37,8 +37,10 @@ void clear_reset_status(unsigned int mask)
>>                pxa2xx_clear_reset_status(mask);
>>        else if (cpu_is_pxa3xx())
>>                pxa3xx_clear_reset_status(mask);
>> -       else
>> +       else if (cpu_is_pxa93x())
>>                pxa93x_clear_reset_status(mask);
>> +       else
>> +               pxa95x_clear_reset_status(mask);
>>  }
>
> I think it's better to have 'clear_arsr_reset_status' in generic.c and make
> that common for all pxa3xx/pxa93x/pxa95x instead of introducing the same
> code all around?
>
OK.
>>
>> diff --git a/arch/arm/mach-pxa/pxa93x.c b/arch/arm/mach-pxa/pxa93x.c
>> index b3cc821..a2d51cb 100644
>> --- a/arch/arm/mach-pxa/pxa93x.c
>> +++ b/arch/arm/mach-pxa/pxa93x.c
>> @@ -316,21 +316,21 @@ static struct clk clk_dummy = {
>>        .ops            = &clk_dummy_ops,
>>  };
>>
>> -static DEFINE_PXA9_CK(pxa93x_lcd, LCD, &clk_pxa93x_hsio_ops);
>> -static DEFINE_PXA9_CK(pxa93x_camera, CAMERA, &clk_pxa93x_hsio_ops);
>> -static DEFINE_PXA9_CKEN(pxa93x_ffuart, FFUART, 14857000, 1);
>> -static DEFINE_PXA9_CKEN(pxa93x_btuart, BTUART, 14857000, 1);
>> -static DEFINE_PXA9_CKEN(pxa93x_stuart, STUART, 14857000, 1);
>> -static DEFINE_PXA9_CKEN(pxa93x_i2c, I2C, 32842000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_keypad, KEYPAD, 32768, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_ssp1, SSP1, 13000000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_ssp2, SSP2, 13000000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_ssp3, SSP3, 13000000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_ssp4, SSP4, 13000000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_pwm0, PWM0, 13000000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_pwm1, PWM1, 13000000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_mmc1, MMC1, 19500000, 0);
>> -static DEFINE_PXA9_CKEN(pxa93x_mmc2, MMC2, 19500000, 0);
>> +static DEFINE_PXA93_CK(pxa93x_lcd, LCD, &clk_pxa93x_hsio_ops);
>> +static DEFINE_PXA93_CK(pxa93x_camera, CAMERA, &clk_pxa93x_hsio_ops);
>> +static DEFINE_PXA93_CKEN(pxa93x_ffuart, FFUART, 14857000, 1);
>> +static DEFINE_PXA93_CKEN(pxa93x_btuart, BTUART, 14857000, 1);
>> +static DEFINE_PXA93_CKEN(pxa93x_stuart, STUART, 14857000, 1);
>> +static DEFINE_PXA93_CKEN(pxa93x_i2c, I2C, 32842000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_keypad, KEYPAD, 32768, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_ssp1, SSP1, 13000000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_ssp2, SSP2, 13000000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_ssp3, SSP3, 13000000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_ssp4, SSP4, 13000000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_pwm0, PWM0, 13000000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_pwm1, PWM1, 13000000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_mmc1, MMC1, 19500000, 0);
>> +static DEFINE_PXA93_CKEN(pxa93x_mmc2, MMC2, 19500000, 0);
>
> Maybe this change could be done in earlier pxa93x separation from pxa3xx patch?
>
OK.
>> diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
>> index a0a2928..674aba5 100644
>> --- a/arch/arm/mm/Kconfig
>> +++ b/arch/arm/mm/Kconfig
>> @@ -382,6 +382,13 @@ config CPU_FEROCEON_OLD_ID
>>          for which the CPU ID is equal to the ARM926 ID.
>>          Relevant for Feroceon-1850 and early Feroceon-2850.
>>
>> +# Marvell PJ4
>> +config CPU_PJ4
>> +       bool
>> +       select CPU_V7
>> +       select CPU_32v6K
>> +       select ARM_THUMBEE
>
> I doubt we need to introduce CONFIG_CPU_PJ4 at this stage, as it's fully
> compatible so far with ARMv7. CPU_32v6K will be selected by CPU_V7 except
> for OMAP2, and ARM_THUMBEE depends on CPU_V7 and is optional (selectable).
>
We still need CPU_PJ4 because of dependancy of iwmmx and tauros cache.
PXA955 need both iwmmx and tauros cache.