On 14.11.2023 14:53, Guenter Roeck wrote: > On Mon, Nov 13, 2023 at 08:20:49PM +0100, Heiner Kallweit wrote: >> Use pm_sleep_ptr/DEFINE_SIMPLE_DEV_PM_OPS to simplify the code. >> >> Note: Apparently conditional compiling based on CONFIG_PM >> wasn't optimal, it should have been CONFIG_PM_SLEEP. > > Maybe that is apparent to you, but not to me, and I assume it won't > be apparent to others either. That warrants a real explanation, > not just "apparently". > See e.g. pm_generic_suspend(), it's NULL if CONFIG_PM_SLEEP isn't defined. Another hint is the following from pm.h: #ifdef CONFIG_PM_SLEEP #define SET_SYSTEM_SLEEP_PM_OPS(suspend_fn, resume_fn) \ SYSTEM_SLEEP_PM_OPS(suspend_fn, resume_fn) #else #define SET_SYSTEM_SLEEP_PM_OPS(suspend_fn, resume_fn) #endif CONFIG_PM_SLEEP implies CONFIG_PM, but we can have a setup where CONFIG_PM is defined but CONFIG_PM_SLEEP is not. In this case jc42_suspend/resume would be defined but not used by the PM core. > Guenter Heiner
