Re: [PATCH 2/2] PM / OPP: check for existing OPP list when initialising from device tree

[Sudeep KarkadaNagesha <Sudeep.KarkadaNagesha@arm.com>]

On 01/05/13 16:04, Nishanth Menon wrote:
> On 12:11-20130501, Sudeep.KarkadaNagesha@arm.com wrote:
>> From: Sudeep KarkadaNagesha <sudeep.karkadanagesha@arm.com>
>>
>> CPUs are registered as devices and their OPPs can be initialised from
>> the device tree. Whenever CPUs can be hotplugged out, the corresponding
>> cpu devices are not removed. As a result all their OPPs remain intact
>> even when they are offlined.
>>
>> But when they are hotplugged back-in, the cpufreq along with other cpu
>> related subsystem gets re-initialised. Since its almost same as secondary
>> cpu being brought up, no special consideration is taken in the hotplug
>> path. As a result of this the cpufreq will try to initialise the OPPs
>> again though the cpu device already contains the OPPs.
>>
>> This patch checks if there exist an OPP list associated with the device,
>> before attempting to initialise it.
>>
>> Signed-off-by: Sudeep KarkadaNagesha <sudeep.karkadanagesha@arm.com>
>> ---
>>  drivers/base/power/opp.c |    5 +++++
>>  1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/base/power/opp.c b/drivers/base/power/opp.c
>> index 4dfdc01..66d52d2 100644
>> --- a/drivers/base/power/opp.c
>> +++ b/drivers/base/power/opp.c
>> @@ -706,6 +706,11 @@ static int of_init_opp_table_from_ofnode(struct device *dev,
>>  	const __be32 *val;
>>  	int nr;
>>  
>> +	/* Check for existing list for 'dev' */
>> +	dev_opp = find_device_opp(dev);
>> +	if (!IS_ERR(dev_opp))
>> +		return 0; /* Device OPP already initialized */
>> +
> It gets a little touchy here -> the normal expectation is for the OPP
> entries to be populated onetime at boot.
> For example - driver bug where same device was attempted twice Vs the
> usecase you mention here - how'd we differentiate between the two?

Do we really need to differentiate ? How about returning -EEXIST ?