MIME-Version: 1.0
References: <20211126115652.1134230-1-kai.heng.feng@canonical.com> <YaD9oubkHPzHtSY4@rowland.harvard.edu>
In-Reply-To: <YaD9oubkHPzHtSY4@rowland.harvard.edu>
From:   Kai-Heng Feng <kai.heng.feng@canonical.com>
Date:   Tue, 21 Dec 2021 11:35:25 +0800
Message-ID: <CAAd53p6uXBa2zms1M0aqvRhVWq-kNtw2LvgE5ai94NWMDi5rYQ@mail.gmail.com>
Subject: Re: [PATCH v2] usb: core: Avoid doing warm reset on disconnect event
To:     Alan Stern <stern@rowland.harvard.edu>
Cc:     gregkh@linuxfoundation.org, mathias.nyman@linux.intel.com,
        Thinh Nguyen <Thinh.Nguyen@synopsys.com>,
        Andrew Lunn <andrew@lunn.ch>, Rajat Jain <rajatja@google.com>,
        Chris Chiu <chris.chiu@canonical.com>,
        linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

(


On Fri, Nov 26, 2021 at 11:30 PM Alan Stern <stern@rowland.harvard.edu> wro=
te:
>
> On Fri, Nov 26, 2021 at 07:56:51PM +0800, Kai-Heng Feng wrote:
> > Unplugging USB device may cause an incorrect warm reset loop:
> > [  143.039019] xhci_hcd 0000:00:14.0: Port change event, 2-3, id 19, po=
rtsc: 0x4202c0
> > [  143.039025] xhci_hcd 0000:00:14.0: handle_port_status: starting usb2=
 port polling.
> > [  143.039051] hub 2-0:1.0: state 7 ports 10 chg 0000 evt 0008
> > [  143.039058] xhci_hcd 0000:00:14.0: Get port status 2-3 read: 0x4202c=
0, return 0x4102c0
> > [  143.039092] xhci_hcd 0000:00:14.0: clear port3 connect change, ports=
c: 0x4002c0
> > [  143.039096] usb usb2-port3: link state change
> > [  143.039099] xhci_hcd 0000:00:14.0: clear port3 link state change, po=
rtsc: 0x2c0
> > [  143.039101] usb usb2-port3: do warm reset
> > [  143.096736] xhci_hcd 0000:00:14.0: Get port status 2-3 read: 0x2b0, =
return 0x2b0
> > [  143.096751] usb usb2-port3: not warm reset yet, waiting 50ms
> > [  143.131500] xhci_hcd 0000:00:14.0: Can't queue urb, port error, link=
 inactive
> > [  143.138260] xhci_hcd 0000:00:14.0: Port change event, 2-3, id 19, po=
rtsc: 0x2802a0
> > [  143.138263] xhci_hcd 0000:00:14.0: handle_port_status: starting usb2=
 port polling.
> > [  143.160756] xhci_hcd 0000:00:14.0: Get port status 2-3 read: 0x2802a=
0, return 0x3002a0
> > [  143.160798] usb usb2-port3: not warm reset yet, waiting 200ms
> >
> > The warm reset is due to its PLS is in eSS.Inactive state. However, USB
> > 3.2 spec table 10-13 mentions "Ports can be disabled by either a fault
> > condition (disconnect event or other fault condition)", xHCI 1.2 spec
> > table 5-27 also states that "This flag shall automatically be cleared t=
o
> > =E2=80=980=E2=80=99 by a disconnect event or other fault condition." on=
 PED.
> >
> > So use CSC =3D 0 and PED =3D 0 as indication that device is disconnecti=
ng to
> > avoid doing warm reset.
> >
> > Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> > ---
> > v2:
> >  - Change the variable type to bool.
> >
> >  drivers/usb/core/hub.c | 6 +++++-
> >  1 file changed, 5 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
> > index a9a04ea967019..4f081df70ecf2 100644
> > --- a/drivers/usb/core/hub.c
> > +++ b/drivers/usb/core/hub.c
> > @@ -5564,6 +5564,7 @@ static void port_event(struct usb_hub *hub, int p=
ort1)
> >               __must_hold(&port_dev->status_lock)
> >  {
> >       int connect_change;
> > +     bool disconnect =3D false;
> >       struct usb_port *port_dev =3D hub->ports[port1 - 1];
> >       struct usb_device *udev =3D port_dev->child;
> >       struct usb_device *hdev =3D hub->hdev;
> > @@ -5579,6 +5580,9 @@ static void port_event(struct usb_hub *hub, int p=
ort1)
> >       if (portchange & USB_PORT_STAT_C_CONNECTION) {
> >               usb_clear_port_feature(hdev, port1, USB_PORT_FEAT_C_CONNE=
CTION);
> >               connect_change =3D 1;
> > +             if (!(portstatus & USB_PORT_STAT_CONNECTION) &&
> > +                 !(portstatus & USB_PORT_STAT_ENABLE))
> > +                     disconnect =3D true;
> >       }
>
> This looks a little strange.  Can there ever be a situation where
> PORT_STAT_CONNECTION is off and PORT_STAT_ENABLE is on?  (It's not allowe=
d in
> USB-2.)

Right, the spec only states PORT_STAT_ENABLE =3D 0. Will change that in
next revision.

>
> >       if (portchange & USB_PORT_STAT_C_ENABLE) {
> > @@ -5647,7 +5651,7 @@ static void port_event(struct usb_hub *hub, int p=
ort1)
> >        * Warm reset a USB3 protocol port if it's in
> >        * SS.Inactive state.
> >        */
> > -     if (hub_port_warm_reset_required(hub, port1, portstatus)) {
> > +     if (hub_port_warm_reset_required(hub, port1, portstatus) && !disc=
onnect) {
> >               dev_dbg(&port_dev->dev, "do warm reset\n");
> >               if (!udev || !(portstatus & USB_PORT_STAT_CONNECTION)
> >                               || udev->state =3D=3D USB_STATE_NOTATTACH=
ED) {
>
> Why is it correct to skip doing a warm reset on a disconnected port here,=
 but not
> correct to skip doing a warm reset on a disconnected port in all the othe=
r places
> where hub_port_warm_reset_required() gets called?

Can a disconnect event happens to other places other than port_event()?

Kai-Heng

>
> Alan Stern