Re: [RFC PATCH v1] usb: core: add sysfs entry for usb device state

[Roy Luo]

On Thu, May 25, 2023 at 11:02 AM Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On Thu, May 25, 2023 at 05:38:18PM +0000, Roy Luo wrote:
> > Expose usb device state to userland as the information is useful in
> > detecting non-compliant setups and diagnosing enumeration failures.
> > For example:
> > - End-to-end signal integrity issues: the device would fail port reset
> >   repeatedly and thus be stuck in POWERED state.
> > - Charge-only cables (missing D+/D- lines): the device would never enter
> >   POWERED state as the HC would not see any pullup.
> >
> > What's the status quo?
> > We do have error logs such as "Cannot enable. Maybe the USB cable is bad?"
> > to flag potential setup issues, but there's no good way to expose them to
> > userspace.
> >
> > Why add a sysfs entry in struct usb_port instead of struct usb_device?
> > The struct usb_device is not device_add() to the system until it's in
> > ADDRESS state hence we would miss the first two states. The struct
> > usb_port is a better place to keep the information because its life
> > cycle is longer than the struct usb_device that is attached to the port.
> >
> > Signed-off-by: Roy Luo <royluo@xxxxxxxxxx>
> > ---
>
> > diff --git a/drivers/usb/core/hub.h b/drivers/usb/core/hub.h
> > index e23833562e4f..110143568c77 100644
> > --- a/drivers/usb/core/hub.h
> > +++ b/drivers/usb/core/hub.h
> > @@ -84,8 +84,10 @@ struct usb_hub {
> >   * @peer: related usb2 and usb3 ports (share the same connector)
> >   * @req: default pm qos request for hubs without port power control
> >   * @connect_type: port's connect type
> > + * @state: device state of the usb device attached to the port
>
> This member is essentially a duplicate of the .child member of the
> usb_port structure.  That is, it points to the .state member of the
> child device instead of to the child device itself, but this is pretty
> much the same thing.  You could replace *(port_dev->state) with
> port_dev->child->state.
>
Alan, thanks for the quick response!
Yes, port_dev->state is indeed the same as port_dev->child->state. However,
I still add port_dev->state because port_dev->child won't be assigned until
the corresponding usb_device is in ADDRESS state.
I wish I can assign get port_dev->child assigned earlier, but I think
the current design - assign port_dev->child and device_add() after ADDRESS
state - also makes sense because there are many ways that the enumeration
could fail in the early stage. By adding port_dev->state, I can link
usb_device->state to usb_port as soon as the usb_device is created to get
around the limitation of port_dev->child.
I would be very happy to hear other ideas.


> > diff --git a/drivers/usb/core/port.c b/drivers/usb/core/port.c
> > index 06a8f1f84f6f..7f3430170115 100644
> > --- a/drivers/usb/core/port.c
> > +++ b/drivers/usb/core/port.c
> > @@ -160,6 +160,19 @@ static ssize_t connect_type_show(struct device *dev,
> >  }
> >  static DEVICE_ATTR_RO(connect_type);
> >
> > +static ssize_t state_show(struct device *dev,
> > +                       struct device_attribute *attr, char *buf)
> > +{
> > +     struct usb_port *port_dev = to_usb_port(dev);
> > +     enum usb_device_state state = USB_STATE_NOTATTACHED;
> > +
> > +     if (port_dev->state)
> > +             state = *port_dev->state;
> > +
> > +     return sprintf(buf, "%s\n",  usb_state_string(state));
>
> This races with device addition and removal (and with device state
> changes).  To prevent these races, you have to hold the
> device_state_lock spinlock while accessing the child device and its
> state.
>
> Unfortunately that spinlock is private to hub.c, so you will have to
> make it public before you can use it here.
>
> Alan Stern