[PATCH] ARM: imx: powerdown PLL before changing of its setting

Jiada Wang jiada_wang at mentor.com
Tue May 20 00:00:46 PDT 2014


Hi Dirk and Fabio

On 05/19/2014 11:17 PM, Dirk Behme wrote:
> Hi Fabio,
>
> On 19.05.2014 19:31, Fabio Estevam wrote:
>> Hi Jiada,
>>
>> On Wed, May 14, 2014 at 6:36 AM, Jiada Wang <jiada_wang at mentor.com>
>> wrote:
>>> According to i.MX6 user manual,  "Before changing the PLL setting,
>>> power it down. Power up the PLL after the change"
>>> But currently PLL setting is being updated without power-down
>>> it first. This may end up with improper output/gitches which
>>> prevents furture operation.
>>>
>>> Signed-off-by: Jiada Wang <jiada_wang at mentor.com>
>>> ---
>>>   arch/arm/mach-imx/clk-pllv3.c | 72
>>> +++++++++++++++++++++++++++++++++++++++----
>>>   1 file changed, 66 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/arch/arm/mach-imx/clk-pllv3.c
>>> b/arch/arm/mach-imx/clk-pllv3.c
>>> index 6136405..f77edd6 100644
>>> --- a/arch/arm/mach-imx/clk-pllv3.c
>>> +++ b/arch/arm/mach-imx/clk-pllv3.c
>>> @@ -67,6 +67,42 @@ static int clk_pllv3_wait_lock(struct clk_pllv3 *pll)
>>>          return readl_relaxed(pll->base) & BM_PLL_LOCK ? 0 : -ETIMEDOUT;
>>>   }
>>>
>>> +static void clk_pllv3_powerdown(struct clk_pllv3 *pll)
>>> +{
>>> +       u32 val;
>>> +
>>> +       val = readl_relaxed(pll->base);
>>> +       val |= BM_PLL_BYPASS;
>>> +       writel_relaxed(val, pll->base);
>>> +
>>> +       if (pll->powerup_set)
>>> +               val &= ~BM_PLL_POWER;
>>> +       else
>>> +               val |= BM_PLL_POWER;
>>> +       writel_relaxed(val, pll->base);
>>> +}
>>> +
>>> +static int clk_pllv3_powerrestore(struct clk_pllv3 *pll, u32 power,
>>> u32 bypass)
>>> +{
>>> +       u32 val;
>>> +       int ret;
>>> +
>>> +       val = readl_relaxed(pll->base);
>>> +       val &= ~BM_PLL_POWER;
>>
>> Shouldn't this be:
>>        if (pll->powerup_set)
>>               val |= BM_PLL_POWER;
>>         else
>>               val &= ~BM_PLL_POWER;
>>
>
> Jiada will correct me if I'm wrong, but to my understanding the job of
> this function is to *restore* the previously configured power state (not
> to set any particular one).
>
> The previously configured power state is passed in 'u32 power'. So we
> just mask out the POWER bit here to be able to restore the old power
> state, independent if it is 0 or 1. So I would think
>
> +    val = readl_relaxed(pll->base);
> +    val &= ~BM_PLL_POWER;
> +    val |= power;
> +    writel_relaxed(val, pll->base);
>
> is correct here.
>
> Best regards
>
> Dirk
>

Yes, this function restores power state as well as bypass state,
as set_rate() may be called before prepare(),
so we have no idea about power and bypass state here.

Thanks,
Jiada



More information about the linux-arm-kernel mailing list