On Wed, 25 Mar 2020, Qais Yousef wrote:

Thanks for all the hints Alan.

I think I figured it out, the below patch seems to fix it for me. Looking
at other drivers resume functions it seems we're missing the
pm_runtime_disable()->set_active()->enable() dance. Doing that fixes the
warning and the dev_err() in driver/base/power.

Ah, yes.  This should have been added years ago; guess I forgot.  :-(

I don't see xhci-plat.c doing that, I wonder if it needs it too.

I'm not well versed about the details and the rules here. So my fix could be
a hack, though it does seem the right thing to do.

I wonder why the power core doesn't handle this transparently..

Initially, we didn't want the PM core to do this automatically because
we thought some devices might want to remain runtime-suspended
following a system resume, and only the device driver would know what 
to do.

Raphael, now that we have the direct_complete mechanism, can we revisit 
this?  Should the PM core automatically call pm_runtime_set_active() if 
dev->power.direct_complete isn't set?  Perhaps in device_resume_early() 
prior to the pm_runtime_enable() call?

It's possible we discussed this and decided against it at the time when 
direct_complete was added, but if so I don't remember what was said.

Alan Stern