On Thu, Jul 06, 2023 at 09:25:26AM +0200, Amadeusz Sławiński wrote: > On 7/6/2023 8:14 AM, Johan Hovold wrote: > > In short, it is not correct to use dev_err_probe() here as this is not a > > probe function. > > > > dev_err_probe() is tied to driver core and will specifically allocate > > and associate an error message with the struct device on probe > > deferrals, which is later freed when the struct device is bound to a > > driver (or released). > I guess you mean call to: device_set_deferred_probe_reason(dev, &vaf); > perhaps functionality could be extended to allow to skip this call and > just do prints? Or just add separate dev_err_defer function without this > step, although it would be best if they could share parts of code. Feel free to suggest adding such a function if you think it's worthwhile. It doesn't exist today it should not be a prerequisite for suppressing these error messages. Johan