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

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 4 Apr 2013 12:26:16 +0300
From:	Felipe Balbi <balbi@...com>
To:	Vivek Gautam <gautamvivek1987@...il.com>
CC:	<balbi@...com>, Alan Stern <stern@...land.harvard.edu>,
	Kishon Vijay Abraham I <kishon@...com>,
	Vivek Gautam <gautam.vivek@...sung.com>,
	<linux-usb@...r.kernel.org>, <linux-samsung-soc@...r.kernel.org>,
	<linux-omap@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
	<gregkh@...uxfoundation.org>, <sarah.a.sharp@...ux.intel.com>,
	<rob.herring@...xeda.com>, <kgene.kim@...sung.com>,
	<dianders@...omium.org>, <t.figa@...sung.com>,
	<p.paneri@...sung.com>
Subject: Re: [PATCH v3 01/11] usb: phy: Add APIs for runtime power management

Hi,

On Thu, Apr 04, 2013 at 02:26:51PM +0530, Vivek Gautam wrote:
> >> > > Lets suppose DWC3 enables runtime_pm on USB 2 type phy,
> >> > > it will try to go into suspend state and thereby call runtime_suspend(), if any.
> >> > > And PHY will come to active state only when its consumer wakes it up,
> >> > > and this consumer is operational
> >> > > only when its related PHY is in fully functional state.
> >> > > So do we have a situation in which this PHY goes into low power state
> >> > > in its runtime_suspend(),
> >> > > resulting in non-detection of devices on further attach (since PHY is
> >> > > in low power state) ?
> >> > >
> >> > > Will the controller (like EHCI/OHCI) be functional now ?
> >> >
> >> > ehci/ohci need to cope with that by calling usb_phy_autopm_get_sync(),
> >> > right ? (so does DWC3 :-)
> >>
> >> Maybe you guys have already got this all figured out -- if so, feel
> >> free to ignore this email.
> >>
> >> Some subsystems handle this issue by calling pm_runtime_get_sync()
> >> before probing a driver and pm_runtime_put_sync() after unbinding the
> >> driver.  If the driver is runtime-PM-enabled, it then does its own
> >> put_sync near the end of its probe routine and get_sync in its release
> >> routine.
> >
> > sounds a bit 'fishy' to me... So a separate entity would call
> > pm_runtime_get_sync(), even when we don't have registered dev_pm_ops,
> > then drivers need to check if runtime_pm is enabled and call
> > pm_runtime_put*() conditionally before returning from probe(). One
> > remove, we might have another issue: device is already runtime_suspended
> > (due to e.g. autosuspend) when module is removed, a call to
> > pm_runtime_put_sync() will be unbalanced. No ?
> 
> May be i am misinterpreting !!
> If PHYs are runtime-PM enabled (PHY probe calls *runtime_enable*),
> then the consumers
> need to call pm_runtime_get_sync whever they want to access PHY.

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.

-- 
balbi

Download attachment "signature.asc" of type "application/pgp-signature" (837 bytes)