On Wed, Aug 10, 2022 at 02:29:55PM +0300, Matti Vaittinen wrote:
> A few regulator consumer drivers seem to be just getting a regulator,
> enabling it and registering a devm-action to disable the regulator at
> the driver detach and then forget about it.
>
> We can simplify this a bit by adding a devm-helper for this pattern.
> Add devm_regulator_get_enable() and devm_regulator_get_enable_optional()
I'm really not keen on the idea of a devm managed enable, it's too prone
to bugs when someone gets round to implementing runtime PM.
Hi deee Ho Mark,
Long time no chat. Glad that you had the time to check this series :)
On 8/10/22 14:56, Mark Brown wrote:
> On Wed, Aug 10, 2022 at 02:29:55PM +0300, Matti Vaittinen wrote:
>> A few regulator consumer drivers seem to be just getting a regulator,
>> enabling it and registering a devm-action to disable the regulator at
>> the driver detach and then forget about it.
>>
>> We can simplify this a bit by adding a devm-helper for this pattern.
>> Add devm_regulator_get_enable() and devm_regulator_get_enable_optional()
>
> I'm really not keen on the idea of a devm managed enable, it's too prone
> to bugs when someone gets round to implementing runtime PM.
I see. And I agree the devm-based regulator disable can cause problems
when combined with manual disable/enable.
In order to tackle the issue the suggested API does not return handle to
the regulators - it really just provides the "get'n enable, then forget"
solution. The consumers who use the suggested API to "devm get'n enable"
will have had time manually controlling the regulator afterwards as they
will not get the handle. I would almost claim that the pattern we
nowadays see (devm_get, enable, add_action_or_reset(disable())) is more
error prone as users seem to in many case be storing the regulator
handle w/o any comment about the automated disable at detach.
--
Matti Vaittinen
Linux kernel developer at ROHM Semiconductors
Oulu Finland
~~ When things go utterly wrong vim users can always type :help! ~~
Discuss - Estimate - Plan - Report and finally accomplish this:
void do_work(int time) __attribute__ ((const));
On Wed, Aug 10, 2022 at 03:19:05PM +0300, Matti Vaittinen wrote:
> In order to tackle the issue the suggested API does not return handle to the
> regulators - it really just provides the "get'n enable, then forget"
> solution. The consumers who use the suggested API to "devm get'n enable"
> will have had time manually controlling the regulator afterwards as they
> will not get the handle. I would almost claim that the pattern we nowadays
> see (devm_get, enable, add_action_or_reset(disable())) is more error prone
> as users seem to in many case be storing the regulator handle w/o any
> comment about the automated disable at detach.
Hrm, right - that does help with that case. However we do need a bulk
version since that's an obvious problem case.
On 8/10/22 18:16, Mark Brown wrote:
> On Wed, Aug 10, 2022 at 03:19:05PM +0300, Matti Vaittinen wrote:
>
>> In order to tackle the issue the suggested API does not return handle to the
>> regulators - it really just provides the "get'n enable, then forget"
>> solution. The consumers who use the suggested API to "devm get'n enable"
>> will have had time manually controlling the regulator afterwards as they
>> will not get the handle. I would almost claim that the pattern we nowadays
>> see (devm_get, enable, add_action_or_reset(disable())) is more error prone
>> as users seem to in many case be storing the regulator handle w/o any
>> comment about the automated disable at detach.
>
> Hrm, right - that does help with that case. However we do need a bulk
> version since that's an obvious problem case.
I'll take a look at the bulk APIs and add them if they're not too
complex. I'm a bit short on time as I was told I should be doing
something we can show to a customer ;)
As a result of this discussion - not returning the handle to struct
regulator * sounds like a safest option here. I'll drop the RFC when I
respin this (hopefully with the bulk-APIs and a few more converted drivers)
Thanks for the input!
Best Regards
Matti Vaittinen
--
Matti Vaittinen
Linux kernel developer at ROHM Semiconductors
Oulu Finland
~~ When things go utterly wrong vim users can always type :help! ~~
Discuss - Estimate - Plan - Report and finally accomplish this:
void do_work(int time) __attribute__ ((const));