RE: [PATCH v2 2/3] phy: core: Drop unused runtime PM APIs

[Yoshihiro Shimoda]

> From: Ulf Hansson, Sent: Thursday, December 21, 2017 7:58 PM
> 
> On 21 December 2017 at 11:33, Yoshihiro Shimoda
> <yoshihiro.shimoda.uh@xxxxxxxxxxx> wrote:
> > Hi Ulf-san,
> >
> >> -----Original Message-----
> >> From: Ulf Hansson, Sent: Wednesday, December 20, 2017 11:09 PM
> > <snip>
> >> diff --git a/include/linux/phy/phy.h b/include/linux/phy/phy.h
> >> index b4298a1..050b620 100644
> >> --- a/include/linux/phy/phy.h
> >> +++ b/include/linux/phy/phy.h
> >> @@ -17,7 +17,6 @@
> >>  #include <linux/err.h>
> >>  #include <linux/of.h>
> >>  #include <linux/device.h>
> >> -#include <linux/pm_runtime.h>
> >
> > After I applied this patch, some ata and gpu drivers causes build error [1].
> > So, we should fix the drivers at first...
> 
> Huh, right, those drivers shouldn't be relying on the phy.h to include
> pm_runtime.h.
> 
> The easiest way at this point is to just put back "#include
> <linux/pm_runtime.h>" in phy.h, then we can deal with these problems
> separately. I do that in a re-spin soon.

I got it.

> BTW, I would be great if you could test this on the Renesas SoC to
> make sure it still fixes the problems (at least half of them I mean).

Sure. I put back the "#include <linux/pm_runtime.h>" in the phy.h and tested
the patches on my environment (r8a7795-salvator-x.dts with v4.15-rc4). And then,
the issue [1] disappeared.
[1]:
printed "Enabling runtime PM for inactive device (ee0a0200.usb-phy) with active children"
in resume timing.

So,

Tested-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@xxxxxxxxxxx>

Best regards,
Yoshihiro Shimoda