[1/2] dt-bindings: usb: Add binding for onboard USB hubs

Onboard USB hubs need to be powered and may require initiaization of
other resources (like GPIOs or clocks) to work properly. This adds
a device tree binding for these hubs.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
---

 .../bindings/usb/onboard_usb_hub.yaml         | 70 +++++++++++++++++++
 1 file changed, 70 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml

Hi,

I just noticed that building this results in a compilation error, due
to a missing brace. I tested this patch, but with another patch on top
with debug logs, the other patch adds the brace.

I'll still hold off a bit before sending v2, for if others have
comments.

On Mon, Sep 14, 2020 at 11:27:49AM -0700, Matthias Kaehlcke wrote:

> diff --git a/drivers/usb/misc/onboard_usb_hub.c b/drivers/usb/misc/onboard_usb_hub.c
> new file mode 100644
> index 000000000000..e5a816d0b124
> --- /dev/null
> +++ b/drivers/usb/misc/onboard_usb_hub.c
>
> +static int onboard_hub_suspend(struct platform_device *pdev, pm_message_t msg)
> +{
> +	struct onboard_hub *hub = dev_get_drvdata(&pdev->dev);
> +	int rc = 0;
> +
> +	if (!hub->cfg.power_off_in_suspend)
> +		return 0;
> +
> +	hub->has_wakeup_capable_descendants = false;
> +
> +	if (hub->cfg.wakeup_source) {
> +		struct udev_node *node;
> +
> +		mutex_lock(&hub->lock);
> +
> +		list_for_each_entry(node, &hub->udev_list, list) {
> +			if (usb_wakeup_enabled_descendants(node->udev)) {
> +				hub->has_wakeup_capable_descendants = true;
> +				break;

missing brace here:	}

> +		}
> +
> +		mutex_unlock(&hub->lock);
> +	}
> +
> +	if (!hub->has_wakeup_capable_descendants)
> +		rc = onboard_hub_power_off(hub);
> +
> +	return rc;
> +}

Hi Alan,

thanks for your feedback!

On Mon, Sep 14, 2020 at 04:14:03PM -0400, Alan Stern wrote:
> On Mon, Sep 14, 2020 at 11:27:49AM -0700, Matthias Kaehlcke wrote:
> > The main issue this driver addresses is that a USB hub needs to be
> > powered before it can be discovered. For onboard hubs this is often
> > solved by supplying the hub with an 'always-on' regulator, which is
> > kind of a hack. Some onboard hubs may require further initialization
> > steps, like changing the state of a GPIO or enabling a clock, which
> > requires further hacks. This driver creates a platform device
> > representing the hub which performs the necessary initialization.
> > Currently it only supports switching on a single regulator, support
> > for multiple regulators or other actions can be added as needed.
> > Different initialization sequences can be supported based on the
> > compatible string.
> > 
> > Besides performing the initialization the driver can be configured
> > to power the hub off during system suspend. This can help to extend
> > battery life on battery powered devices, which have no requirements
> > to keep the hub powered during suspend. The driver can also be
> > configured to leave the hub powered when a wakeup capable USB device
> > is connected when suspending, and keeping it powered otherwise.
> > 
> > Technically the driver consists of two drivers, the platform driver
> > described above and a very thin USB driver that subclasses the
> > generic hub driver.
> 
> Actually it subclasses the generic usb device driver, not the hub 
> driver.

ok, I'll change it in the next version.

> >  The purpose of this driver is to provide the
> > platform driver with the USB devices corresponding to the hub(s)
> > (a hub controller may provide multiple 'logical' hubs, e.g. one
> > to support USB 2.0 and another for USB 3.x).
> > 
> > Co-developed-by: Ravi Chandra Sadineni <ravisadineni@chromium.org>
> > Signed-off-by: Ravi Chandra Sadineni <ravisadineni@chromium.org>
> > Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
> > ---
> > This is an evolution of '[RFC] USB: misc: Add usb_hub_pwr driver'
> > (https://lore.kernel.org/patchwork/patch/1299239/).
> > 
> > Changes in v1:
> > - renamed the driver to 'onboard_usb_hub'
> > - single file for platform and USB driver
> > - USB hub devices register with the platform device
> >   - the DT includes a phandle of the platform device
> > - the platform device now controls when power is turned off
> > - the USB driver became a very thin subclass of the generic hub
> >   driver
> > - enabled autosuspend support
> 
> See https://marc.info/?l=linux-usb&m=159914635920888&w=2 and the 
> accompanying submissions.  You'll probably want to include those updates 
> in your driver.

Thanks for the pointer! I'll change the driver to use pm_ptr as suggested.

> > > +	hub->cfg.power_off_in_suspend =

> of_property_read_bool(dev->of_node, "power-off-in-suspend");

> > > +	hub->cfg.wakeup_source = of_property_read_bool(dev->of_node,

> > > +"wakeup-source");

> >

> > Do you really need these two properties? If the device (and its

> > children if existed) has wakeup enabled, you keep power in suspend,

> > otherwise, you could close it, any exceptions?

> 

> That would work for my use case, but I'm not sure it's a universally good

> configuration.

> 

> I don't have a specific USB device in mind, but you could have a device that

> shouldn't lose it's context during suspend or keep operating autonomously (e.g.

> a sensor with a large buffer collecting samples). Not sure if something like this

> exists in the real though.

> 

> I'm not an expert, but it seems there are USB controllers with wakeup support

> which is always enabled. A board with such a controller then couldn't have a

> policy to power down the hub regardless of wakeup capable devices being

> connected.

> 


Whether or not it is a wakeup_source, it could get through its or its children's /sys/../power/wakeup
value, you have already used usb_wakeup_enabled_descendants to know it. If the onboard HUB
needs to reflect wakeup signal, it should not power off its regulator.

For another property power-off-in-suspend, I think it is also a user option, but not a hardware feature.

If (wakeup-source || ! power-off-in-suspend)
	power off;
else
	keep power;

Peter

On Mon, Sep 14, 2020 at 11:27:48AM -0700, Matthias Kaehlcke wrote:
> Onboard USB hubs need to be powered and may require initiaization of
> other resources (like GPIOs or clocks) to work properly. This adds
> a device tree binding for these hubs.

We already have bindings for these. 2 in fact as I2C controlled hubs are 
often described under the I2C bus.

> 
> Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
> ---
> 
>  .../bindings/usb/onboard_usb_hub.yaml         | 70 +++++++++++++++++++
>  1 file changed, 70 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> 
> diff --git a/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml b/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> new file mode 100644
> index 000000000000..f82d8f459eed
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> @@ -0,0 +1,70 @@
> +# SPDX-License-Identifier: GPL-2.0-only or BSD-2-Clause
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/usb/onboard_usb_hub.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Binding for onboard USB hubs
> +
> +maintainers:
> +  - Matthias Kaehlcke <mka@chromium.org>
> +
> +allOf:
> +  - $ref: /schemas/usb/onboard_usb_hub.yaml#
> +
> +properties:
> +  compatible:
> +    enum:
> +      - onboard-usb-hub
> +      - realtek,rts5411
> +
> +  power-off-in-suspend:
> +    description:
> +      The hub should be powered off during system suspend. When the
> +      "wakeup-source" property is also provided the hub is only powered
> +      off during suspend when no wakeup capable descendants are connected.
> +    type: boolean
> +
> +  vdd-supply:
> +    description:
> +      phandle to the regulator that provides power to the hub.
> +
> +  wakeup-source:
> +    description:
> +      Wakeup capable USB devices connected to this hub can be used as
> +      wakeup source.
> +    type: boolean
> +
> +required:
> +  - compatible
> +  - vdd-supply
> +
> +examples:
> +  - |
> +    usb_hub: usb-hub {
> +        compatible = "realtek,rts5411", "onboard-usb-hub";
> +        vdd-supply = <&pp3300_hub>;
> +        power-off-in-suspend;
> +        wakeup-source;

This is the hub device?

> +    };
> +
> +    &usb_1_dwc3 {
> +	dr_mode = "host";
> +	#address-cells = <1>;
> +	#size-cells = <0>;
> +
> +	/* 2.0 hub on port 1 */
> +	hub@1 {
> +		compatible = "usbbda,5411";
> +		reg = <1>;
> +		hub = <&usb_hub>;

Or this node is?

> +	};
> +
> +	/* 3.0 hub on port 2 */
> +	hub@2 {
> +		compatible = "usbbda,411";
> +		reg = <2>;
> +		hub = <&usb_hub>;

Or this node is?

The hub node belongs here.

If you really have it connected to 2 upstream ports, then just do 
one node with 'reg = <1 2>;'.

Rob

> +	};
> +
> +...
> -- 
> 2.28.0.618.gf4bc123cb7-goog
>

Hi Peter,

On Tue, Sep 15, 2020 at 07:05:38AM +0000, Peter Chen wrote:
>   

> > > > +	hub->cfg.power_off_in_suspend =

> > of_property_read_bool(dev->of_node, "power-off-in-suspend");

> > > > +	hub->cfg.wakeup_source = of_property_read_bool(dev->of_node,

> > > > +"wakeup-source");

> > >

> > > Do you really need these two properties? If the device (and its

> > > children if existed) has wakeup enabled, you keep power in suspend,

> > > otherwise, you could close it, any exceptions?

> > 

> > That would work for my use case, but I'm not sure it's a universally good

> > configuration.

> > 

> > I don't have a specific USB device in mind, but you could have a device that

> > shouldn't lose it's context during suspend or keep operating autonomously (e.g.

> > a sensor with a large buffer collecting samples). Not sure if something like this

> > exists in the real though.

> > 

> > I'm not an expert, but it seems there are USB controllers with wakeup support

> > which is always enabled. A board with such a controller then couldn't have a

> > policy to power down the hub regardless of wakeup capable devices being

> > connected.

> > 

> 

> Whether or not it is a wakeup_source, it could get through its or its children's

> /sys/../power/wakeup value, you have already used usb_wakeup_enabled_descendants

> to know it.


I conceptually agree, but in practice there are some conflicting details:

wakeup for the hubs on my system is by default disabled, yet USB wakeup works
regardless, so the flag doesn't really provide useful information. I guess we
could still use it if there is no better way, but it doesn't seem ideal.

Similar for udev->bus->controller, according to sysfs it doesn't even have wakeup
support. Please let me know if there is a reliable way to check if wakeup is
enabled on the controller of a device.

> If the onboard HUB needs to reflect wakeup signal, it should not power off its regulator.

> 

> For another property power-off-in-suspend, I think it is also a user option,

> but not a hardware feature.


Ok, I think you are suggesting a sysfs attribute instead of a DT property, that
sounds good to me.

Hi Rob,

On Tue, Sep 15, 2020 at 08:21:45AM -0600, Rob Herring wrote:
> On Mon, Sep 14, 2020 at 11:27:48AM -0700, Matthias Kaehlcke wrote:
> > Onboard USB hubs need to be powered and may require initiaization of
> > other resources (like GPIOs or clocks) to work properly. This adds
> > a device tree binding for these hubs.
> 
> We already have bindings for these. 2 in fact as I2C controlled hubs are 
> often described under the I2C bus.

Yes, these are I2C controlled hubs, which need hub specific drivers. This
driver is for hubs without an additional bus that share similar
initialization requirements and can benefit from common functionality.

> > Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
> > ---
> > 
> >  .../bindings/usb/onboard_usb_hub.yaml         | 70 +++++++++++++++++++
> >  1 file changed, 70 insertions(+)
> >  create mode 100644 Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> > 
> > diff --git a/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml b/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> > new file mode 100644
> > index 000000000000..f82d8f459eed
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> > @@ -0,0 +1,70 @@
> > +# SPDX-License-Identifier: GPL-2.0-only or BSD-2-Clause
> > +%YAML 1.2
> > +---
> > +$id: http://devicetree.org/schemas/usb/onboard_usb_hub.yaml#
> > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > +
> > +title: Binding for onboard USB hubs
> > +
> > +maintainers:
> > +  - Matthias Kaehlcke <mka@chromium.org>
> > +
> > +allOf:
> > +  - $ref: /schemas/usb/onboard_usb_hub.yaml#
> > +
> > +properties:
> > +  compatible:
> > +    enum:
> > +      - onboard-usb-hub
> > +      - realtek,rts5411
> > +
> > +  power-off-in-suspend:
> > +    description:
> > +      The hub should be powered off during system suspend. When the
> > +      "wakeup-source" property is also provided the hub is only powered
> > +      off during suspend when no wakeup capable descendants are connected.
> > +    type: boolean
> > +
> > +  vdd-supply:
> > +    description:
> > +      phandle to the regulator that provides power to the hub.
> > +
> > +  wakeup-source:
> > +    description:
> > +      Wakeup capable USB devices connected to this hub can be used as
> > +      wakeup source.
> > +    type: boolean
> > +
> > +required:
> > +  - compatible
> > +  - vdd-supply
> > +
> > +examples:
> > +  - |
> > +    usb_hub: usb-hub {
> > +        compatible = "realtek,rts5411", "onboard-usb-hub";
> > +        vdd-supply = <&pp3300_hub>;
> > +        power-off-in-suspend;
> > +        wakeup-source;
> 
> This is the hub device?

This is the physical hub device on the platform bus, which is the
equivalent to this entry for a usb2512b hub on an I2C bus:

    usb2512b@2c {
        compatible = "microchip,usb2512b";
	reg = <0x2c>;
	reset-gpios = <&gpio1 4 GPIO_ACTIVE_LOW>;
    };

(source: Documentation/devicetree/bindings/usb/usb251xb.txt)

It doesn't have an I2C, SPI or other bus, hence the platform bus is
used.

> > +    };
> > +
> > +    &usb_1_dwc3 {
> > +	dr_mode = "host";
> > +	#address-cells = <1>;
> > +	#size-cells = <0>;
> > +
> > +	/* 2.0 hub on port 1 */
> > +	hub@1 {
> > +		compatible = "usbbda,5411";
> > +		reg = <1>;
> > +		hub = <&usb_hub>;
> 
> Or this node is?

It is the USB 2.0 part of the hub. The device is instantiated by
Linux even without this node, but the system associates the node
with the device, which suggests it 'exists'.

The usb2512b mentioned above implicitly also has a node here, it just
isn't specified since the USB controller autodetects it.

> > +	};
> > +
> > +	/* 3.0 hub on port 2 */
> > +	hub@2 {
> > +		compatible = "usbbda,411";
> > +		reg = <2>;
> > +		hub = <&usb_hub>;
> 
> Or this node is?

It is the USB 3.0 part of the hub.

> The hub node belongs here.

The platform device isn't probed when the node is inside the USB
controller node. I haven't investigated why that's the case.

> If you really have it connected to 2 upstream ports, then just do
> one node with 'reg = <1 2>;'.

Yes, it is connected to two upstream ports. The platform driver needs a
reference to both/all hubs, to be able to determine whether to keep the
hub powered during system suspend or not.

Technically the hub with product id 0x5411 is connected to port 1 and the
one with product id 0x411 to port 2, so I would say the above is more
accurate than pretending one of the hubs is connected to both ports.

I would argue that the two hub nodes are similar to a SDIO BT/WiFi combo,
where you have one chip/module with multiple functions. The DT has entries
for both functions, even though they reside in the same chip and share the
same bus.

On Tue, Sep 15, 2020 at 04:03:45PM -0700, Matthias Kaehlcke wrote:
> Hi Peter,

> 

> On Tue, Sep 15, 2020 at 07:05:38AM +0000, Peter Chen wrote:


> > Whether or not it is a wakeup_source, it could get through its or its children's

> > /sys/../power/wakeup value, you have already used usb_wakeup_enabled_descendants

> > to know it.

> 

> I conceptually agree, but in practice there are some conflicting details:

> 

> wakeup for the hubs on my system is by default disabled, yet USB wakeup works

> regardless, so the flag doesn't really provide useful information. I guess we

> could still use it if there is no better way, but it doesn't seem ideal.


The wakeup setting for USB hubs affects only the following events: port 
connect, port disconnect, and port overcurrent.  It does not refer to 
forwarding wakeup requests from downstream USB devices; that is always 
enabled.  So maybe your wakeup flag really is accurate and you didn't 
realize it.

> Similar for udev->bus->controller, according to sysfs it doesn't even have wakeup

> support. Please let me know if there is a reliable way to check if wakeup is

> enabled on the controller of a device.


The host controller's sysfs wakeup setting should always be correct.  If 
it isn't, that indicates there is a bug in the host controller driver or 
the corresponding platform-specific code.  What driver does your system 
use?

Alan Stern

On 20-09-15 16:03:45, Matthias Kaehlcke wrote:
> Hi Peter,
> 
> On Tue, Sep 15, 2020 at 07:05:38AM +0000, Peter Chen wrote:
> >   
> > > > > +	hub->cfg.power_off_in_suspend =
> > > of_property_read_bool(dev->of_node, "power-off-in-suspend");
> > > > > +	hub->cfg.wakeup_source = of_property_read_bool(dev->of_node,
> > > > > +"wakeup-source");
> > > >
> > > > Do you really need these two properties? If the device (and its
> > > > children if existed) has wakeup enabled, you keep power in suspend,
> > > > otherwise, you could close it, any exceptions?
> > > 
> > > That would work for my use case, but I'm not sure it's a universally good
> > > configuration.
> > > 
> > > I don't have a specific USB device in mind, but you could have a device that
> > > shouldn't lose it's context during suspend or keep operating autonomously (e.g.
> > > a sensor with a large buffer collecting samples). Not sure if something like this
> > > exists in the real though.
> > > 
> > > I'm not an expert, but it seems there are USB controllers with wakeup support
> > > which is always enabled. A board with such a controller then couldn't have a
> > > policy to power down the hub regardless of wakeup capable devices being
> > > connected.
> > > 
> > 
> > Whether or not it is a wakeup_source, it could get through its or its children's
> > /sys/../power/wakeup value, you have already used usb_wakeup_enabled_descendants
> > to know it.
> 
> I conceptually agree, but in practice there are some conflicting details:
> 
> wakeup for the hubs on my system is by default disabled, yet USB wakeup works
> regardless, so the flag doesn't really provide useful information. I guess we
> could still use it if there is no better way, but it doesn't seem ideal.
> 
> Similar for udev->bus->controller, according to sysfs it doesn't even have wakeup
> support. Please let me know if there is a reliable way to check if wakeup is
> enabled on the controller of a device.

Then, how could your code work, you use usb_wakeup_enabled_descendants
to get if HUB or the descendants under the HUB has wakeup enabled?

If you use dwc3, you need to enable xhci-plat.c's wakeup entry if your
system needs xHCI connect/disconnect wakeup event. I have one pending
patch to do it:

https://www.spinics.net/lists/linux-usb/msg199406.html

> 
> > If the onboard HUB needs to reflect wakeup signal, it should not power off its regulator.
> > 
> > For another property power-off-in-suspend, I think it is also a user option,
> > but not a hardware feature.
> 
> Ok, I think you are suggesting a sysfs attribute instead of a DT property, that
> sounds good to me.

Yes.

Hi Peter,

On Wed, Sep 16, 2020 at 08:19:07AM +0000, Peter Chen wrote:
> On 20-09-15 16:03:45, Matthias Kaehlcke wrote:
> > Hi Peter,
> > 
> > On Tue, Sep 15, 2020 at 07:05:38AM +0000, Peter Chen wrote:
> > >   
> > > > > > +	hub->cfg.power_off_in_suspend =
> > > > of_property_read_bool(dev->of_node, "power-off-in-suspend");
> > > > > > +	hub->cfg.wakeup_source = of_property_read_bool(dev->of_node,
> > > > > > +"wakeup-source");
> > > > >
> > > > > Do you really need these two properties? If the device (and its
> > > > > children if existed) has wakeup enabled, you keep power in suspend,
> > > > > otherwise, you could close it, any exceptions?
> > > > 
> > > > That would work for my use case, but I'm not sure it's a universally good
> > > > configuration.
> > > > 
> > > > I don't have a specific USB device in mind, but you could have a device that
> > > > shouldn't lose it's context during suspend or keep operating autonomously (e.g.
> > > > a sensor with a large buffer collecting samples). Not sure if something like this
> > > > exists in the real though.
> > > > 
> > > > I'm not an expert, but it seems there are USB controllers with wakeup support
> > > > which is always enabled. A board with such a controller then couldn't have a
> > > > policy to power down the hub regardless of wakeup capable devices being
> > > > connected.
> > > > 
> > > 
> > > Whether or not it is a wakeup_source, it could get through its or its children's
> > > /sys/../power/wakeup value, you have already used usb_wakeup_enabled_descendants
> > > to know it.
> > 
> > I conceptually agree, but in practice there are some conflicting details:
> > 
> > wakeup for the hubs on my system is by default disabled, yet USB wakeup works
> > regardless, so the flag doesn't really provide useful information. I guess we
> > could still use it if there is no better way, but it doesn't seem ideal.
> > 
> > Similar for udev->bus->controller, according to sysfs it doesn't even have wakeup
> > support. Please let me know if there is a reliable way to check if wakeup is
> > enabled on the controller of a device.
> 
> Then, how could your code work, you use usb_wakeup_enabled_descendants
> to get if HUB or the descendants under the HUB has wakeup enabled?

Doing just that would not allow to switch the hub off when wakeup enabled
descendants are connected, which might be desirable in some configurations.

> If you use dwc3, you need to enable xhci-plat.c's wakeup entry if your
> system needs xHCI connect/disconnect wakeup event. I have one pending
> patch to do it:
> 
> https://www.spinics.net/lists/linux-usb/msg199406.html

Thanks, my system has indeed a dwc3(-qcom) controller, your patch adds
the missing wakeup entry to sysfs. So it seems your patch should solve
my problem (sharp timing!), however you mention specifically the 'xHCI
connect/disconnect wakeup event', so I wonder if the xHCI wakeup flag
isn't applicable to other wakeup events. I know the dwc3-qcom platform
device has its own wakeup flag. The driver currently enables wakeup
interrupts unconditionally, I sent a patch to change that
(https://lore.kernel.org/patchwork/patch/1305894/), however I now wonder
if it should evaluate the xHCI wakeup flag instead of its own.

Thanks

Matthias

On Tue, Sep 15, 2020 at 10:14:21PM -0400, Alan Stern wrote:
> On Tue, Sep 15, 2020 at 04:03:45PM -0700, Matthias Kaehlcke wrote:
> > Hi Peter,
> > 
> > On Tue, Sep 15, 2020 at 07:05:38AM +0000, Peter Chen wrote:
> 
> > > Whether or not it is a wakeup_source, it could get through its or its children's
> > > /sys/../power/wakeup value, you have already used usb_wakeup_enabled_descendants
> > > to know it.
> > 
> > I conceptually agree, but in practice there are some conflicting details:
> > 
> > wakeup for the hubs on my system is by default disabled, yet USB wakeup works
> > regardless, so the flag doesn't really provide useful information. I guess we
> > could still use it if there is no better way, but it doesn't seem ideal.
> 
> The wakeup setting for USB hubs affects only the following events: port 
> connect, port disconnect, and port overcurrent.  It does not refer to 
> forwarding wakeup requests from downstream USB devices; that is always 
> enabled.  So maybe your wakeup flag really is accurate and you didn't 
> realize it.

Thanks for the clarification!

> > Similar for udev->bus->controller, according to sysfs it doesn't even have wakeup
> > support. Please let me know if there is a reliable way to check if wakeup is
> > enabled on the controller of a device.
> 
> The host controller's sysfs wakeup setting should always be correct.  If 
> it isn't, that indicates there is a bug in the host controller driver or 
> the corresponding platform-specific code.

Good to know :)

> What driver does your system use?

The driver is dwc3-qcom, Peter pointed me to a patch he recently sent to add
the missing wakeup entry (https://patchwork.kernel.org/patch/11717835/). It
seems that should solve the problem, except for some confusion on my side
about the wakeup flag of the xHCI device vs. that of the platform device
(details in my reply to Peter).

On Thu, Sep 17, 2020 at 12:27:29AM +0000, Peter Chen wrote:
> On 20-09-16 12:16:07, Matthias Kaehlcke wrote:
> > Hi Peter,
> > 
> > On Wed, Sep 16, 2020 at 08:19:07AM +0000, Peter Chen wrote:
> > > On 20-09-15 16:03:45, Matthias Kaehlcke wrote:
> > > > Hi Peter,
> > > > 
> > > > On Tue, Sep 15, 2020 at 07:05:38AM +0000, Peter Chen wrote:
> > > > >   
> > > > > > > > +	hub->cfg.power_off_in_suspend =
> > > > > > of_property_read_bool(dev->of_node, "power-off-in-suspend");
> > > > > > > > +	hub->cfg.wakeup_source = of_property_read_bool(dev->of_node,
> > > > > > > > +"wakeup-source");
> > > > > > >
> > > > > > > Do you really need these two properties? If the device (and its
> > > > > > > children if existed) has wakeup enabled, you keep power in suspend,
> > > > > > > otherwise, you could close it, any exceptions?
> > > > > > 
> > > > > > That would work for my use case, but I'm not sure it's a universally good
> > > > > > configuration.
> > > > > > 
> > > > > > I don't have a specific USB device in mind, but you could have a device that
> > > > > > shouldn't lose it's context during suspend or keep operating autonomously (e.g.
> > > > > > a sensor with a large buffer collecting samples). Not sure if something like this
> > > > > > exists in the real though.
> > > > > > 
> > > > > > I'm not an expert, but it seems there are USB controllers with wakeup support
> > > > > > which is always enabled. A board with such a controller then couldn't have a
> > > > > > policy to power down the hub regardless of wakeup capable devices being
> > > > > > connected.
> > > > > > 
> > > > > 
> > > > > Whether or not it is a wakeup_source, it could get through its or its children's
> > > > > /sys/../power/wakeup value, you have already used usb_wakeup_enabled_descendants
> > > > > to know it.
> > > > 
> > > > I conceptually agree, but in practice there are some conflicting details:
> > > > 
> > > > wakeup for the hubs on my system is by default disabled, yet USB wakeup works
> > > > regardless, so the flag doesn't really provide useful information. I guess we
> > > > could still use it if there is no better way, but it doesn't seem ideal.
> > > > 
> > > > Similar for udev->bus->controller, according to sysfs it doesn't even have wakeup
> > > > support. Please let me know if there is a reliable way to check if wakeup is
> > > > enabled on the controller of a device.
> > > 
> > > Then, how could your code work, you use usb_wakeup_enabled_descendants
> > > to get if HUB or the descendants under the HUB has wakeup enabled?
> > 
> > Doing just that would not allow to switch the hub off when wakeup enabled
> > descendants are connected, which might be desirable in some configurations.
> > 
> > > If you use dwc3, you need to enable xhci-plat.c's wakeup entry if your
> > > system needs xHCI connect/disconnect wakeup event. I have one pending
> > > patch to do it:
> > > 
> > > https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.spinics.net%2Flists%2Flinux-usb%2Fmsg199406.html&data=02%7C01%7Cpeter.chen%40nxp.com%7C02c4cc75e26a47d0224d08d85a74f945%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C637358805725394858&sdata=cjZhSmQiXVJoLsN5PjFACsLwsikH%2BeRTztPhsckJFNs%3D&reserved=0
> > 
> > Thanks, my system has indeed a dwc3(-qcom) controller, your patch adds
> > the missing wakeup entry to sysfs. So it seems your patch should solve
> > my problem (sharp timing!), however you mention specifically the 'xHCI
> > connect/disconnect wakeup event', so I wonder if the xHCI wakeup flag
> > isn't applicable to other wakeup events. I know the dwc3-qcom platform
> > device has its own wakeup flag. The driver currently enables wakeup
> > interrupts unconditionally, I sent a patch to change that
> > (https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.kernel.org%2Fpatchwork%2Fpatch%2F1305894%2F&data=02%7C01%7Cpeter.chen%40nxp.com%7C02c4cc75e26a47d0224d08d85a74f945%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C637358805725394858&sdata=6IjiiHJql%2FW4vzDla9q3qdfiiOzOQy1Vk7ryUhKOOTc%3D&reserved=0), however I now wonder
> > if it should evaluate the xHCI wakeup flag instead of its own.
> > 
> 
> You may need both (glue & xhci), it depends on system design, and
> usually, these two kinds of wakeup setting isn't conflict.

Ok, thanks. So if I understand correctly the onboard hub driver should
check the wakeup state of the xHCI to determine if remote wakeup is
enabled for the controller (after all it doesn't know anything about
the platform device). Wakeup might not work properly if it is disabled
for the platform device, but it's the responsability of the board
software/config to make sure it is enabled (possibly this could be done
by making the dwc3-qcom driver understand the 'wakeup-source' property,
as the xhci-mtk driver does).

+Florian

On Tue, Sep 15, 2020 at 6:00 PM Matthias Kaehlcke <mka@chromium.org> wrote:
>
> Hi Rob,
>
> On Tue, Sep 15, 2020 at 08:21:45AM -0600, Rob Herring wrote:
> > On Mon, Sep 14, 2020 at 11:27:48AM -0700, Matthias Kaehlcke wrote:
> > > Onboard USB hubs need to be powered and may require initiaization of
> > > other resources (like GPIOs or clocks) to work properly. This adds
> > > a device tree binding for these hubs.
> >
> > We already have bindings for these. 2 in fact as I2C controlled hubs are
> > often described under the I2C bus.
>
> Yes, these are I2C controlled hubs, which need hub specific drivers. This
> driver is for hubs without an additional bus that share similar
> initialization requirements and can benefit from common functionality.

Yes, as I said, there's already 2 ways to do this. The second is
defining the USB bus under the USB host. I'm sure there's some
examples in the tree.

> > > Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
> > > ---
> > >
> > >  .../bindings/usb/onboard_usb_hub.yaml         | 70 +++++++++++++++++++
> > >  1 file changed, 70 insertions(+)
> > >  create mode 100644 Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> > >
> > > diff --git a/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml b/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> > > new file mode 100644
> > > index 000000000000..f82d8f459eed
> > > --- /dev/null
> > > +++ b/Documentation/devicetree/bindings/usb/onboard_usb_hub.yaml
> > > @@ -0,0 +1,70 @@
> > > +# SPDX-License-Identifier: GPL-2.0-only or BSD-2-Clause
> > > +%YAML 1.2
> > > +---
> > > +$id: http://devicetree.org/schemas/usb/onboard_usb_hub.yaml#
> > > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > > +
> > > +title: Binding for onboard USB hubs
> > > +
> > > +maintainers:
> > > +  - Matthias Kaehlcke <mka@chromium.org>
> > > +
> > > +allOf:
> > > +  - $ref: /schemas/usb/onboard_usb_hub.yaml#
> > > +
> > > +properties:
> > > +  compatible:
> > > +    enum:
> > > +      - onboard-usb-hub
> > > +      - realtek,rts5411
> > > +
> > > +  power-off-in-suspend:
> > > +    description:
> > > +      The hub should be powered off during system suspend. When the
> > > +      "wakeup-source" property is also provided the hub is only powered
> > > +      off during suspend when no wakeup capable descendants are connected.
> > > +    type: boolean
> > > +
> > > +  vdd-supply:
> > > +    description:
> > > +      phandle to the regulator that provides power to the hub.
> > > +
> > > +  wakeup-source:
> > > +    description:
> > > +      Wakeup capable USB devices connected to this hub can be used as
> > > +      wakeup source.
> > > +    type: boolean
> > > +
> > > +required:
> > > +  - compatible
> > > +  - vdd-supply
> > > +
> > > +examples:
> > > +  - |
> > > +    usb_hub: usb-hub {
> > > +        compatible = "realtek,rts5411", "onboard-usb-hub";
> > > +        vdd-supply = <&pp3300_hub>;
> > > +        power-off-in-suspend;
> > > +        wakeup-source;
> >
> > This is the hub device?
>
> This is the physical hub device on the platform bus, which is the

How is a USB hub connected to the 'platform bus'? There's no such
thing as 'platform bus' in DT.

> equivalent to this entry for a usb2512b hub on an I2C bus:
>
>     usb2512b@2c {
>         compatible = "microchip,usb2512b";
>         reg = <0x2c>;
>         reset-gpios = <&gpio1 4 GPIO_ACTIVE_LOW>;
>     };
>
> (source: Documentation/devicetree/bindings/usb/usb251xb.txt)
>
> It doesn't have an I2C, SPI or other bus, hence the platform bus is
> used.
>
> > > +    };
> > > +
> > > +    &usb_1_dwc3 {
> > > +   dr_mode = "host";
> > > +   #address-cells = <1>;
> > > +   #size-cells = <0>;
> > > +
> > > +   /* 2.0 hub on port 1 */
> > > +   hub@1 {
> > > +           compatible = "usbbda,5411";
> > > +           reg = <1>;
> > > +           hub = <&usb_hub>;
> >
> > Or this node is?
>
> It is the USB 2.0 part of the hub. The device is instantiated by
> Linux even without this node, but the system associates the node
> with the device, which suggests it 'exists'.
>
> The usb2512b mentioned above implicitly also has a node here, it just
> isn't specified since the USB controller autodetects it.

Like other probe-able buses, we describe the devices in DT when they
have extra resources/config which are not probe-able.

> > > +   };
> > > +
> > > +   /* 3.0 hub on port 2 */
> > > +   hub@2 {
> > > +           compatible = "usbbda,411";
> > > +           reg = <2>;
> > > +           hub = <&usb_hub>;
> >
> > Or this node is?
>
> It is the USB 3.0 part of the hub.
>
> > The hub node belongs here.
>
> The platform device isn't probed when the node is inside the USB
> controller node. I haven't investigated why that's the case.

It shouldn't be a platform device, but associating a device_node with
the usb device. I think at least that is there in the kernel.

Though if you need setup to happen before the device is probe-able
which appears to be the case here, then that is a common problem which
isn't really a solved issue. There's the mmc-pwrseq stuff, but I don't
want to see a repeat of that. There was an #armlinux irc discussion I
had with Florian on this just 2 days ago[1]. If the only thing you
have to configure is 'reset-gpios', then you probably can do that
generically scanning all the child USB DT devices and deassert reset.
But as soon as you have device specific things and need specific
ordering and timing, then you'll need device specific code to handle
it.

>
> > If you really have it connected to 2 upstream ports, then just do
> > one node with 'reg = <1 2>;'.
>
> Yes, it is connected to two upstream ports. The platform driver needs a
> reference to both/all hubs, to be able to determine whether to keep the
> hub powered during system suspend or not.
>
> Technically the hub with product id 0x5411 is connected to port 1 and the
> one with product id 0x411 to port 2, so I would say the above is more
> accurate than pretending one of the hubs is connected to both ports.
>
> I would argue that the two hub nodes are similar to a SDIO BT/WiFi combo,
> where you have one chip/module with multiple functions. The DT has entries
> for both functions, even though they reside in the same chip and share the
> same bus.

That has generally worked because little is shared between BT and WiFi
and what is shared is refcounted (clks for example). That could work
here, we just have to be careful.

Rob

[1] https://pastebin.com/iepyfe9c