On Thu, 2017-12-07 at 06:36 +0100, Oleksij Rempel wrote:

On 07.12.2017 00:11, Christoph Hellwig wrote:

quoted

quoted

 void (*pm_power_off_prepare)(void);
+EXPORT_SYMBOL(pm_power_off_prepare);

EXPORT_SYMBOL_GPL for something this deeply internal, please.

Ok,
probably all other symbols should be converted in this file in to
EXPORT_SYMBOL_GPL as well?

grep EXPORT_SYMBOL kernel/reboot.c
EXPORT_SYMBOL(cad_pid);
EXPORT_SYMBOL(pm_power_off_prepare);
EXPORT_SYMBOL_GPL(emergency_restart);
EXPORT_SYMBOL(register_reboot_notifier);
EXPORT_SYMBOL(unregister_reboot_notifier);
EXPORT_SYMBOL(devm_register_reboot_notifier);
EXPORT_SYMBOL(register_restart_handler);
EXPORT_SYMBOL(unregister_restart_handler);
EXPORT_SYMBOL_GPL(kernel_restart);
EXPORT_SYMBOL_GPL(kernel_halt);
EXPORT_SYMBOL_GPL(kernel_power_off);
EXPORT_SYMBOL_GPL(orderly_poweroff);
EXPORT_SYMBOL_GPL(orderly_reboot);

This call looks much more deeply internal than those other functions,
it's not like reboot_notifier and restart_handler.

It seems that the only user of pm_power_off_prepare right now is the
ACPI core and your patch uses it in a regulator driver. This looks
extremely strange. Maybe this is why imx maintainers are stalling on
this patch?

It might help if the ACPI/PM maintainers ack/review this part
explicitly, and the usage in part 5:

https://patchwork.kernel.org/patch/9799615/

--
Regards,
Leonard