* Uwe Kleine-König <u.kleine-koenig@xxxxxxxxxxxxxx> [230413 07:07]:
> On Thu, Apr 13, 2023 at 09:39:15AM +0300, Tony Lindgren wrote:
> > Oh OK. Care to clarify a bit why we are not allowed to return errors
> > on remove though? Are we getting rid of the return value for remove?
> > Sorry if I'm not following the cunning plan here :)
>
> Yes, that's the plan. If you look at the caller of the remove functions
> (before 5c5a7680e67ba6fbbb5f4d79fa41485450c1985c):
>
> static void platform_remove(struct device *_dev)
> {
> struct platform_driver *drv = to_platform_driver(_dev->driver);
> struct platform_device *dev = to_platform_device(_dev);
>
> if (drv->remove) {
> int ret = drv->remove(dev);
>
> if (ret)
> dev_warn(_dev, "remove callback returned a non-zero value. This will be ignored.\n");
> }
> dev_pm_domain_detach(_dev, true);
> }
>
> you see it's pointless to return an error value. But the prototype
> seduces driver authors to do it yielding to error that can easily
> prevented if .remove returns void. See also
> 5c5a7680e67ba6fbbb5f4d79fa41485450c1985c for some background and details
> of the quest.
OK thanks. So maybe check the pm_runtime_get_sync() and on error do
pm_runtime_put_noidle(), or pm_runtime_resume_and_get(). Both ways
are fine for me, maybe you already figured it out.
Regards,
Tony
