2016-11-02 01:24:56

by Stephen Boyd

Subject: Re: [PATCH v3 4/7] clk: qcom: ipq4019: Added all the frequencies for apps cpu

On 09/21, Abhishek Sahu wrote:
> The APPS CPU clock does not contain all the frequencies in its
> frequency table so this patch adds the same.
>
> Signed-off-by: Abhishek Sahu <[email protected]>
> ---
> drivers/clk/qcom/gcc-ipq4019.c | 12 +++++++++++-
> 1 file changed, 11 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/clk/qcom/gcc-ipq4019.c b/drivers/clk/qcom/gcc-ipq4019.c
> index 211c68c..160e0cf 100644
> --- a/drivers/clk/qcom/gcc-ipq4019.c
> +++ b/drivers/clk/qcom/gcc-ipq4019.c
> @@ -565,10 +565,20 @@ static struct clk_rcg2 sdcc1_apps_clk_src = {
> };
>
> static const struct freq_tbl ftbl_gcc_apps_clk[] = {
> - F(48000000, P_XO, 1, 0, 0),
> + F(48000000, P_XO, 1, 0, 0),
> F(200000000, P_FEPLL200, 1, 0, 0),
> + F(380000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(409000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(444000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(484000000, P_DDRPLLAPSS, 1, 0, 0),
> F(500000000, P_FEPLL500, 1, 0, 0),
> + F(507000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(532000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(560000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(592000000, P_DDRPLLAPSS, 1, 0, 0),
> F(626000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(666000000, P_DDRPLLAPSS, 1, 0, 0),
> + F(710000000, P_DDRPLLAPSS, 1, 0, 0),
> { }
> };

Can't we have the determine_rate callback know the speeds of the
"fixed" PLLs and use those first if the rate hits exactly? And
then if that doesn't happen go try ddrpllapps and set the rate on
it? I'm hoping we can get rid of this frequency table.

--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

2016-11-24 12:46:07

by Abhishek Sahu

Subject: Re: [PATCH v3 4/7] clk: qcom: ipq4019: Added all the frequencies for apps cpu

On 2016-11-02 06:54, Stephen Boyd wrote:
> On 09/21, Abhishek Sahu wrote:
>> The APPS CPU clock does not contain all the frequencies in its
>> frequency table so this patch adds the same.
>>
>> Signed-off-by: Abhishek Sahu <[email protected]>
>> ---
>> drivers/clk/qcom/gcc-ipq4019.c | 12 +++++++++++-
>> 1 file changed, 11 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/clk/qcom/gcc-ipq4019.c
>> b/drivers/clk/qcom/gcc-ipq4019.c
>> index 211c68c..160e0cf 100644
>> --- a/drivers/clk/qcom/gcc-ipq4019.c
>> +++ b/drivers/clk/qcom/gcc-ipq4019.c
>> @@ -565,10 +565,20 @@ static struct clk_rcg2 sdcc1_apps_clk_src = {
>> };
>>
>> static const struct freq_tbl ftbl_gcc_apps_clk[] = {
>> - F(48000000, P_XO, 1, 0, 0),
>> + F(48000000, P_XO, 1, 0, 0),
>> F(200000000, P_FEPLL200, 1, 0, 0),
>> + F(380000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(409000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(444000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(484000000, P_DDRPLLAPSS, 1, 0, 0),
>> F(500000000, P_FEPLL500, 1, 0, 0),
>> + F(507000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(532000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(560000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(592000000, P_DDRPLLAPSS, 1, 0, 0),
>> F(626000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(666000000, P_DDRPLLAPSS, 1, 0, 0),
>> + F(710000000, P_DDRPLLAPSS, 1, 0, 0),
>> { }
>> };
>
> Can't we have the determine_rate callback know the speeds of the
> "fixed" PLLs and use those first if the rate hits exactly? And
> then if that doesn't happen go try ddrpllapps and set the rate on
> it? I'm hoping we can get rid of this frequency table.
This clock is being registered with QCOM clk_rcg2 operations which
already has determine_rate callback based on this frequency table.
Currently all the frequencies are being generated without HID
divider but in future, we can have some frequency which will use
dividers also.
--
Abhishek Sahu