Re: [PATCH v3 01/11] usb: phy: Add APIs for runtime power management

From: Alan Stern <stern@rowland.harvard.edu>
To: Vivek Gautam <gautamvivek1987@gmail.com>
Cc: Felipe Balbi <balbi@ti.com>,
	Kishon Vijay Abraham I <kishon@ti.com>,
	Vivek Gautam <gautam.vivek@samsung.com>,
	<linux-usb@vger.kernel.org>, <linux-samsung-soc@vger.kernel.org>,
	<linux-omap@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
	<gregkh@linuxfoundation.org>, <sarah.a.sharp@linux.intel.com>,
	<rob.herring@calxeda.com>, <kgene.kim@samsung.com>,
	<dianders@chromium.org>, <t.figa@samsung.com>,
	<p.paneri@samsung.com>
Subject: Re: [PATCH v3 01/11] usb: phy: Add APIs for runtime power management
Date: Tue, 23 Apr 2013 14:12:29 -0400 (EDT)	[thread overview]
Message-ID: <Pine.LNX.4.44L0.1304231408370.1679-100000@iolanthe.rowland.org> (raw)
In-Reply-To: <CAFp+6iECumogLo9Tf+4pMqRhtkGHf1CqA52YjL1tACTRyGO7jw@mail.gmail.com>

On Tue, 23 Apr 2013, Vivek Gautam wrote:

> Hi,
> 
> 
> On Tue, Apr 23, 2013 at 10:23 PM, Alan Stern <stern@rowland.harvard.edu> wrote:
> > On Tue, 23 Apr 2013, Vivek Gautam wrote:
> >
> >> >> Alright, so here's my understanding:
> >> >>
> >> >> I suggested letting e.g. DWC3 enable the PHY's runtime_pm; Alan said
> >> >> that it could be done before that so that DWC3 sees an enabled PHY
> >> >> during probe.
> >> >
> >> > Basically right.  Help me to understand the overall situation a little
> >> > better:
> >> >
> >> >         What code registers the PHY initially?
> >>            PHY is added to global list by PHY drivers (like
> >> phy-samsung-usb2.c/phy-omap-usb2.c)
> >>            by usb_add_phy() API
> >
> > Then this routine should initialize the PHY.  The initialized state
> > could be either active or suspended, your choice.  Suspended would be
> > best, in case the PHY never gets used.
> 
> Fair enough.
> 
> >
> >> >         What routine does the DWC3 driver call to register itself
> >> >         as a consumer of the PHY?
> >>            I think DWC3 doesn't registers itself as consumer of PHY,
> >> rather it gets that PHY from
> >>            the list using devm_usb_get_phy()/devm_usb_get_phy_by_phandle() API.
> >>            DWC3 can now call PHY's initialization sequence using usb_phy_init().
> >>            So, before DWC3 initializes the PHY, PHYs should be in active state.
> >
> > Then usb_phy_init should make sure the PHY is in the active state.  If
> > usb_add_phy() left the PHY suspended, then this routine should call
> > pm_runtime_get_sync().
> 
> Right
> 
> >
> > After DWC3 (or any other driver) has acquired the PHY, it can call
> > pm_runtime_put/get() however it likes, so long as the calls balance
> > properly.  If the driver isn't runtime-PM aware then it won't use any
> > of these calls, and the PHY will remain active the entire time.
> 
> Alright, so DWC3 (or any other consumer of PHY) should do minimal to
> handle runtime state of PHYs; get() when accessing PHY and put() when it's done
> with it.

Yes.  The first operation will generally be a put, because
usb_phy_init() will leave the PHY in an active state.

> >> >         Likewise, what routine does it call to unregister itself?
> >>            Once DWC3's remove() is called PHYs are put.
> >
> > Is there a routine analogous to usb_phy_init() that gets called when
> > PHY is released?  That routine would do the opposite of usb_phy_init(),
> > putting the PHY back into its initialized state.
> 
> Yes, ofcourse there's a routine usb_phy_shutdown(). So this will be
> calling put_sync()
> to put PHYs back to its initialized state. Right ?

Correct.

Alan Stern