On Thu, May 05, 2022 at 09:45:49AM -0700, Matthias Kaehlcke wrote:
> On Thu, May 05, 2022 at 08:56:18AM +0530, Pavan Kondeti wrote:
> > On Wed, May 04, 2022 at 10:46:30AM -0700, Matthias Kaehlcke wrote:
> > > On Wed, Apr 20, 2022 at 12:41:06AM +0530, Sandeep Maheswaram wrote:
> > > > During suspend read the status of all port and set hs phy mode
> > > > based on current speed. Use this hs phy mode to configure wakeup
> > > > interrupts in qcom glue driver.
> > > >
> > > > Check wakeup-source property for dwc3 core node to set the
> > > > wakeup capability. Drop the device_init_wakeup call from
> > > > runtime suspend and resume.
> > > >
> > > > Also check during suspend if any wakeup capable devices are
> > > > connected to the controller (directly or through hubs), if there
> > > > are none set a flag to indicate that the PHY is powered
> > > > down during suspend.
> > > >
> > > > Signed-off-by: Sandeep Maheswaram <quic_c_sanm@xxxxxxxxxxx>
> > > > ---
> > > > v14:
> > > > Used device_children_wakeup_capable instead of usb_wakeup_enabled_descendants.
> > > >
> > > > v13:
> > > > Changed dwc3_set_phy_speed_mode to dwc3_check_phy_speed_mode.
> > > > Removed device_init_wakeup calls from dwc3_runtime_suspend and dwc3_runtime_resume
> > > > as we have a new dt property wakeup-source.
> > > >
> > > >
> > > > drivers/usb/dwc3/core.c | 33 ++++++++++++++++++++-------------
> > > > drivers/usb/dwc3/core.h | 4 ++++
> > > > drivers/usb/dwc3/host.c | 24 ++++++++++++++++++++++++
> > > > 3 files changed, 48 insertions(+), 13 deletions(-)
> > > >
> > > > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
> > > > index 1170b80..898aa66 100644
> > > > --- a/drivers/usb/dwc3/core.c
> > > > +++ b/drivers/usb/dwc3/core.c
> > > > @@ -32,6 +32,7 @@
> > > > #include <linux/usb/gadget.h>
> > > > #include <linux/usb/of.h>
> > > > #include <linux/usb/otg.h>
> > > > +#include <linux/usb/hcd.h>
> > > >
> > > > #include "core.h"
> > > > #include "gadget.h"
> > > > @@ -1723,6 +1724,7 @@ static int dwc3_probe(struct platform_device *pdev)
> > > >
> > > > platform_set_drvdata(pdev, dwc);
> > > > dwc3_cache_hwparams(dwc);
> > > > + device_init_wakeup(&pdev->dev, of_property_read_bool(dev->of_node, "wakeup-source"));
> > > >
> > > > spin_lock_init(&dwc->lock);
> > > > mutex_init(&dwc->mutex);
> > > > @@ -1865,6 +1867,7 @@ static int dwc3_suspend_common(struct dwc3 *dwc, pm_message_t msg)
> > > > {
> > > > unsigned long flags;
> > > > u32 reg;
> > > > + struct usb_hcd *hcd = platform_get_drvdata(dwc->xhci);
> > > >
> > > > switch (dwc->current_dr_role) {
> > > > case DWC3_GCTL_PRTCAP_DEVICE:
> > > > @@ -1877,10 +1880,7 @@ 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)) {
> > > > - dwc3_core_exit(dwc);
> > > > - break;
> > > > - }
> > > > + dwc3_check_phy_speed_mode(dwc);
> > > >
> > > > /* Let controller to suspend HSPHY before PHY driver suspends */
> > > > if (dwc->dis_u2_susphy_quirk ||
> > > > @@ -1896,6 +1896,16 @@ static int dwc3_suspend_common(struct dwc3 *dwc, pm_message_t msg)
> > > >
> > > > phy_pm_runtime_put_sync(dwc->usb2_generic_phy);
> > > > phy_pm_runtime_put_sync(dwc->usb3_generic_phy);
> > > > +
> > > > + if (!PMSG_IS_AUTO(msg)) {
> > > > + if (device_may_wakeup(dwc->dev) &&
> > > > + device_children_wakeup_capable(&hcd->self.root_hub->dev)) {
> > > > + dwc->phy_power_off = false;
> > > > + } else {
> > > > + dwc->phy_power_off = true;
> > > > + dwc3_core_exit(dwc);
> > >
> > > I found that shutting the PHYs down during suspend leads to high power
> > > consumption of a downstream hub (about 80mW vs 15mW when the PHYs are
> > > not shut down).
> > >
> > > It would be interesting to know if this also impacts other non-hub
> > > peripherals. Unfortunately I can't test that, the hub on my system is
> > > soldered to the board.
> > >
> > > I understand that shutting the PHYs down might be beneficial in terms
> > > of power on some systems, however on those I'm looking at we'd strongly
> > > prefer to save the 65mW of power consumed by the hub, rather than
> > > whatever smaller amount of power that is saved by powering down the
> > > PHYs.
> > >
> > > Could we introduce a sysfs attribute (or some other sort of knob) to
> > > allow the admin to configure whether the PHYs should remain on or off
> > > during suspend? That is assuming that it is actually desirable to power
> > > them off on some systems.
> >
> > The result may vary across SoCs also. The current proposal is to keep PHY
> > powered during system suspend if any of the downstream USB devices are enabled
> > for wakeup. This also includes USB2/USB3 root hub. If one wants to keep PHY
> > always powered on even when no device is attached, they can do so by enabling
> > wakeup (echo enabled > /sys/bus/usb/devices/usbX/power/wakeup). This is anyway
> > needed if you want to detect a peripheral attach during system suspend.
>
> My concern is that it is not evident for an admin what causes the high power
> consumption of the USB client (if they detect/localize it in the first place),
> and even less that wakeup needs to be enabled to mitigate it.
>
> Why can't we just put the PHYs in suspend, rather than taking the controller
> down completely during suspend?
Agreed and I also have the same question.
I don't know the background on why DWC3 chooses to power down the PHY(s)
during system suspend. Probably it is beneficial in some board designs.
Atleast this patch series provides a way to wakeup the USB from system
suspend, which also can be used not to power down the PHY(s). If all the users
of DWC3 agree that powering down the PHY is bad, then we can do something
about it.
Thanks,
Pavan
