Re: [PATCH 1/2] PM / sleep: Return RPM_SUSPENDED to keep devices in runtime-suspended

[Chen Yu <yu.c.chen@intel.com>]

Hi,
On Wed, Sep 28, 2016 at 01:46:10PM +0200, Rafael J. Wysocki wrote:
> On Wed, Sep 28, 2016 at 5:26 AM, Chen Yu <yu.c.chen@intel.com> wrote:
> > Currently if the ->prepare() callback of a device returns a positive number,
> > the PM core will regard that as an indication that it may leave the
> > device runtime-suspended. However it would be more convenient to define
> > this positive number then makes it more maintainable. Consider there might be
> > already some device drivers using different positive values, this patch
> > prints a warning if the positive value is other than RPM_SUSPENDED, and hoping
> > driver developers would adjust their return values to RPM_SUSPENDED, then
> > at last we can modify the code to only enable this feature for values return
> > of RPM_SUSPENDED rather than arbitrary positive value.
> >
> > Suggested-by: Lee Jones <lee.jones@linaro.org>
> > Suggested-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > Signed-off-by: Chen Yu <yu.c.chen@intel.com>
> > ---
> >  }
> 
> No.
> 
> (1) RPM_SUSPENDED has a specific meaning to the runtime PM framework,
> so please don't overload it.
> 
> (2) Define a new symbol (e.g. DPM_DIRECT_COMPLETE), but allow drivers
> to return positive numbers different from that.
> 
OK.
> That may be useful if someone wants to do "return a > b" or "return
> count", where "direct complete" is to be used for all values of count
> different from 0.
> 
> The issue here is that "1" in your previous patch looked arbitrary, so
> you're addressing this by defining a symbol to use instead.
> 
OK.
> And BTW, there are places that return "1" already from their
> ->prepare(), so this patch would have generated a bunch of
> false-positives.
> 
OK, will send a new version. Thanks!
> Thanks,
> Rafael
Yu