[PATCH 2/3] firmware: arm_scmi: Add support for marking certain frequencies as boost

[Sibi Sankar]

On 1/31/24 21:38, Sudeep Holla wrote:
> On Wed, Jan 31, 2024 at 03:29:43PM +0100, Pierre Gondois wrote:
>> Hello Sibi,
>>
>> On 1/17/24 12:04, Sibi Sankar wrote:
>>> All opps above the sustained level/frequency are treated as boost, so mark
>>> them accordingly.
>>>

Sudeep/Pierre,

Thanks for taking time to review the series.

>>> Suggested-by: Sudeep Holla <sudeep.holla at arm.com>
>>> Signed-off-by: Sibi Sankar <quic_sibis at quicinc.com>
>>> ---
>>>    drivers/firmware/arm_scmi/perf.c | 11 ++++++++++-
>>>    1 file changed, 10 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/firmware/arm_scmi/perf.c b/drivers/firmware/arm_scmi/perf.c
>>> index e286f04ee6e3..d3fb8c804b3d 100644
>>> --- a/drivers/firmware/arm_scmi/perf.c
>>> +++ b/drivers/firmware/arm_scmi/perf.c
>>> @@ -811,7 +811,7 @@ static int scmi_dvfs_device_opps_add(const struct scmi_protocol_handle *ph,
>>>    				     struct device *dev, u32 domain)
>>>    {
>>>    	int idx, ret;
>>> -	unsigned long freq;
>>> +	unsigned long freq, sustained_freq;
>>>    	struct dev_pm_opp_data data = {};
>>>    	struct perf_dom_info *dom;
>>> @@ -819,12 +819,21 @@ static int scmi_dvfs_device_opps_add(const struct scmi_protocol_handle *ph,
>>>    	if (IS_ERR(dom))
>>>    		return PTR_ERR(dom);
>>> +	if (!dom->level_indexing_mode)
>>> +		sustained_freq = dom->sustained_perf_level * dom->mult_factor;
>>> +	else
>>> +		sustained_freq = dom->sustained_freq_khz * dom->mult_factor;
>>> +
>>>    	for (idx = 0; idx < dom->opp_count; idx++) {
>>>    		if (!dom->level_indexing_mode)
>>>    			freq = dom->opp[idx].perf * dom->mult_factor;
>>>    		else
>>>    			freq = dom->opp[idx].indicative_freq * dom->mult_factor;
>>> +		/* All opps above the sustained level/frequency are treated as boost */
>>> +		if (sustained_freq && freq > sustained_freq)
>>
>> It seems the sustained_freq is not optional since SCMI v1.0,
>> is it necessary to check that (sustained_freq != 0) ?
>>
> 
> Technically correct, we don't have to. But since day 1, we checked and
> handled 0 for perf_level specifically to avoid division by zero. I am
> just worried if there are any platforms in the wild with these values as
> 0. We can start without the check and add it if someone complains perhaps ?

sure will drop the check in the re-spin.

-Sibi

>