On Mon, Oct 30, 2023 at 07:43:50PM +0000, Jonathan Cameron wrote: > On Mon, 30 Oct 2023 19:41:12 +0000 > Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx> wrote: ... > Actually looking at usecase, why isn't the absence of an EXPORT in > the !CONFIG_PM_SLEEP path not a problem for bisection of this series given > you haven't yet protected the users? I'm not sure I got the issue you are trying to point out. Between first and last patches the main driver exports two things: the PM ops functions, which are _always_ been exported and PM ops structure, which is exported only when CONFIG_PM_SLEEP=y. Every converted user has pm_sleep_ptr() guard added so it shouldn't be a problem. What exactly did I miss? -- With Best Regards, Andy Shevchenko