linux-kernel - Re: [PATCH] usb: dwc3: keep PHYs disabled during suspend

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:   Wed, 24 Aug 2022 09:41:49 -0700
From:   Matthias Kaehlcke <mka@...omium.org>
To:     Johan Hovold <johan@...nel.org>
Cc:     Johan Hovold <johan+linaro@...nel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Felipe Balbi <balbi@...nel.org>,
        Andy Gross <agross@...nel.org>,
        Bjorn Andersson <andersson@...nel.org>,
        Konrad Dybcio <konrad.dybcio@...ainline.org>,
        Krishna Kurapati <quic_kriskura@...cinc.com>,
        Pavankumar Kondeti <quic_pkondeti@...cinc.com>,
        linux-arm-msm@...r.kernel.org, linux-usb@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] usb: dwc3: keep PHYs disabled during suspend

On Wed, Aug 24, 2022 at 10:27:44AM +0200, Johan Hovold wrote:
> On Tue, Aug 23, 2022 at 11:33:18AM -0700, Matthias Kaehlcke wrote:
> > Hi Johan,
> > 
> > On Tue, Aug 23, 2022 at 02:40:47PM +0200, Johan Hovold wrote:
> > > Commit 649f5c842ba3 ("usb: dwc3: core: Host wake up support from system
> > > suspend") started leaving the PHYs enabled during suspend for
> > > wakeup-capable controllers even though it turns out this had nothing to
> > > do with wakeup.
> > > 
> > > Rather, the wakeup capability flag was (ab-)used as a proxy to configure
> > > the suspend behaviour in an attempt to reduce power leakage on some
> > > platforms.
> > > 
> > > Stop abusing the wakeup configuration and restore the 5.19 behaviour of
> > > keeping the PHYs powered off during suspend. If needed, a dedicated
> > > mechanism for configuring the PHY power state during suspend can be
> > > added later.
> > > 
> > > Fixes: 649f5c842ba3 ("usb: dwc3: core: Host wake up support from system suspend")
> > > Link: https://lore.kernel.org/r/Yuv7AM/5jtO/pgcm@google.com
> > > Signed-off-by: Johan Hovold <johan+linaro@...nel.org>
> > > ---
> > >  drivers/usb/dwc3/core.c      | 4 ++--
> > >  drivers/usb/dwc3/dwc3-qcom.c | 1 -
> > >  2 files changed, 2 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
> > > index 8c8e32651473..0cdb6be720e1 100644
> > > --- a/drivers/usb/dwc3/core.c
> > > +++ b/drivers/usb/dwc3/core.c
> > > @@ -1983,7 +1983,7 @@ static int dwc3_suspend_common(struct dwc3 *dwc, pm_message_t msg)
> > >  		dwc3_core_exit(dwc);
> > >  		break;
> > >  	case DWC3_GCTL_PRTCAP_HOST:
> > > -		if (!PMSG_IS_AUTO(msg) && !device_may_wakeup(dwc->dev)) {
> > > +		if (!PMSG_IS_AUTO(msg)) {
> > 
> > My assumption was that the PHYs need to be powered for wakeup to work, but
> > apparently that isn't the case, wakeup still works on sc7x80 with this part
> > of this patch.
> 
> Thanks for confirming.
> 
> > >  			dwc3_core_exit(dwc);
> > >  			break;
> > >  		}
> > > @@ -2044,7 +2044,7 @@ static int dwc3_resume_common(struct dwc3 *dwc, pm_message_t msg)
> > >  		spin_unlock_irqrestore(&dwc->lock, flags);
> > >  		break;
> > >  	case DWC3_GCTL_PRTCAP_HOST:
> > > -		if (!PMSG_IS_AUTO(msg) && !device_may_wakeup(dwc->dev)) {
> > > +		if (!PMSG_IS_AUTO(msg)) {
> > >  			ret = dwc3_core_init_for_resume(dwc);
> > >  			if (ret)
> > >  				return ret;
> > > diff --git a/drivers/usb/dwc3/dwc3-qcom.c b/drivers/usb/dwc3/dwc3-qcom.c
> > > index 9a94b1ab8f7a..9995395baa12 100644
> > > --- a/drivers/usb/dwc3/dwc3-qcom.c
> > > +++ b/drivers/usb/dwc3/dwc3-qcom.c
> > > @@ -904,7 +904,6 @@ static int dwc3_qcom_probe(struct platform_device *pdev)
> > >  
> > >  	wakeup_source = of_property_read_bool(dev->of_node, "wakeup-source");
> > >  	device_init_wakeup(&pdev->dev, wakeup_source);
> > > -	device_init_wakeup(&qcom->dwc3->dev, wakeup_source);
> > 
> > Surprisingly this part breaks wakeup on sc7x80, with the above removal
> > of the device_may_wakeup() checks it is not clear to me why wakeup needs
> > to be enabled for the core.
> 
> I can't explain that behaviour either. This change doesn't affect the
> wakeup_path flag and genpd, and notably wakeup still works here with
> sc8280xp.
> 
> Could it be some Chromium user-space issue in that it expects all
> devices on the wakeup path to be wakeup capable? Note that the
> xhci-plat driver (e.g. for the descendant xhci-hcd.1.auto device)
> unconditionally sets the wakeup-capable flag (but leaves it disabled by
> default).
> 
> I guess we could do something similar for the dwc3 core device, but we'd
> need to figure out if and why that is at all needed first.
> 
> Can you verify that the wakeup source (e.g. keyboard) you're using still
> has power/wakeup set to "enabled"?

I confirmed that the wakeup flag of the wakeup source is still enabled when
the wakeup source suspends.

It turns out the dwc3 core wakeup flags are evaluated by:

	int usb_phy_roothub_suspend(struct device *controller_dev,
	                            struct usb_phy_roothub *phy_roothub)
	{
		usb_phy_roothub_power_off(phy_roothub);

		/* keep the PHYs initialized so the device can wake up the system */
		if (device_may_wakeup(controller_dev))
			return 0;

		return usb_phy_roothub_exit(phy_roothub);
	}

'controller_dev' is the dwc3 core. The root hub is un-initialized when wakeup is
disabled. That causes wakeup to fail, and also happens to be the/one cause of
the high power consumption of an onboard USB hub that I mentioned earlier in
another thread.