On Thu, Dec 20, 2012 at 06:37:26AM +0000, Liu, Chuansheng wrote:

> Meanwhile, is it useful to add one warning there for that case?
> After all, in probing, set the bias to _STANDBY even idle_bias_off == 1, and calling get_runtime_sync(), it
> will let the code more obscure. So giving a warning there to indicate the driver:
> it is not suggested that in probing, set the bias to _STANDBY even idle_bias_off == 1.

Probably, send a patch please.  Like I say it is possible to start off
in _STANDBY providing the driver grabs the runtime PM reference too but
I can't think of any reason for doing that so the warning seems sensible.