usb: dwc3: of-simple: reset host controller at suspend/resume

[Felipe Balbi <balbi@kernel.org>]

Hi,

Enric Balletbo Serra <eballetbo@gmail.com> writes:
> Hi,
> Missatge de Ezequiel Garcia <ezequiel@collabora.com> del dia dt., 10
> de jul. 2018 a les 0:10:
>>
>> Sigh, now actually Cc devicetree folks
>>
>> On Mon, 2018-07-09 at 19:04 -0300, Ezequiel Garcia wrote:
>> > (Cc devicetree folks)
>> >
>> > Hi Enric,
>> >
>> > Thanks for the patch.
>> >
>> > On Mon, 2018-07-09 at 17:08 +0200, Enric Balletbo i Serra wrote:
>> > > If we power off the SoC logic rail in S3, we can find that the
>> > > Type-C
>> > > PHY can't initialize correctly after system resume. We need to
>> > > toggle
>> > > the USB3-OTG reset before trying to initialize the PHY, or else it
>> > > times out.
>> > >
>> > >     phy phy-ff800000.phy.9: phy poweron failed --> -110
>> > >     dwc3 fe900000.dwc3: failed to initialize core
>> > >     dwc3: probe of fe900000.dwc3 failed with error -110
>> > >
>> > > Note that the RK3399 TRM suggests that we should keep the whole
>> > > usb3
>> > > controller in reset for the duration of the Type-C PHY
>> > > initialization.
>> > > However, it's hard to assert the reset in the current framework of
>> > > reset. We're still skeptical about that, and we haven't yet found a
>> > > case where this seems to have mattered. This approach is much
>> > > easier,
>> > > it
>> > > simply holds the USB3-OTG reset while device is supended.
>> > >
>> > > The dwc3 core is going to reinitialize the controller at
>> > > suspend/resume
>> > > anyway (including a "soft reset"), so it should be safe to do this.
>> > >
>> > > Signed-off-by: Enric Balletbo i Serra <enric.balletbo@collabora.com
>> > > >
>> > > ---
>> > > Dear all,
>> > >
>> > > Now that the usb3-phy otg port support for rk3399 has been merged
>> > > [1]
>> > > we
>> > > found that suspend/resume is broken. The problem is well known for
>> > > ChromeOS kernels, they solved it in a similar way adding a reset
>> > > pulse on
>> > > resume in the specific usb glue layer (dwc3-rockchip). In mainline,
>> > > though, we use the dwc3-of-simple glue layer instead of a specific
>> > > layer
>> > > for rockchip. The patch is based on the Brian Norris work but
>> > > slightly
>> > > different, it holds the reset while device is suspended. It was
>> > > tested
>> > > on a Samsung Chromebook Plus with usbc docking station attached by
>> > > doing
>> > > different suspend/resume cycles and checking no usb devices has
>> > > been
>> > > lost.
>> > >
>> > > I am not sure this is the better way to solve this but I did not
>> > > find
>> > > any other way, and, as I am not sure this can be generic, the reset
>> > > is only
>> > > done on rockchip platforms.
>> > >
>> >
>> > I don't really understand why there are per-platform hacks in glue
>> > drivers, instead of having per-platform glue drivers, or some other
>> > pluggable hooks.
>> >
>> > > Best regards,
>> > >  Enric
>> > >
>> > > [1] bfdca1736ea76345071bbc5607d18928e54909ac ('arm64: dts:
>> > > rockchip:
>> > > add
>> > >     usb3-phy otg-port support for rk3399')
>> > >
>> > >  drivers/usb/dwc3/dwc3-of-simple.c | 21 +++++++++++++++++++++
>> > >  1 file changed, 21 insertions(+)
>> > >
>> > > diff --git a/drivers/usb/dwc3/dwc3-of-simple.c
>> > > b/drivers/usb/dwc3/dwc3-of-simple.c
>> > > index dbeff5e6ad14..1d1ece99ed94 100644
>> > > --- a/drivers/usb/dwc3/dwc3-of-simple.c
>> > > +++ b/drivers/usb/dwc3/dwc3-of-simple.c
>> > > @@ -201,9 +201,30 @@ static int
>> > > dwc3_of_simple_runtime_resume(struct
>> > > device *dev)
>> > >
>> > >     return 0;
>> > >  }
>> > > +
>> > > +static int dwc3_of_simple_suspend(struct device *dev)
>> > > +{
>> > > +   struct dwc3_of_simple *simple = dev_get_drvdata(dev);
>> > > +
>> > > +   if (of_device_is_compatible(dev->of_node,
>> > > "rockchip,rk3399-
>> > > dwc3"))
>> > >
>> >
>> > Instead of calling of_device_is_compatible in each suspend/resume,
>> > which seems quite expensive, how about having a per-device boolean
>> > 'needs_reset' or something like that?
>> >
> Yep, not sure how much, but probably use a boolean will be faster. I
> am also wondering if we can remove the of_device_is_compatible call
> and do the reset on all platforms. This will need lots of tests on
> different platforms, of course. i'd like to hear maintainers feedback
> here.

ideally, the reset would be unconditional, but in practice, probably
it's handled differently by different implementations.

I'd be fine with a single of_device_is_compatible() call during probe to
set a driver flag. Another possibility would be to use the data field of
of_device_id to pass that flag.