Received-SPF: pass (google.com: domain of linux-kernel+bounces-96463-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Precedence: bulk
MIME-Version: 1.0
References: <CAEkJfYO6jRVC8Tfrd_R=cjO0hguhrV31fDPrLrNOOHocDkPoAA@mail.gmail.com>
 <0afd0885-1650-47db-b2b3-43b597352e6b@rowland.harvard.edu>
In-Reply-To: <0afd0885-1650-47db-b2b3-43b597352e6b@rowland.harvard.edu>
From: Sam Sun <samsun1006219@gmail.com>
Date: Fri, 8 Mar 2024 10:48:07 +0800
Message-ID: <CAEkJfYPAvVbN5qzAWZchaF25WehCWU4FEYH8VJBkBadLgY-rFQ@mail.gmail.com>
Subject: Re: [Bug] INFO: task hung in hub_activate
To: Alan Stern <stern@rowland.harvard.edu>
Cc: linux-kernel@vger.kernel.org, linux-usb@vger.kernel.org, 
	"xrivendell7@gmail.com" <xrivendell7@gmail.com>, Greg KH <gregkh@linuxfoundation.org>, 
	hgajjar@de.adit-jv.com, quic_ugoswami@quicinc.com, stanley_chang@realtek.com, 
	heikki.krogerus@linux.intel.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Mar 8, 2024 at 4:35=E2=80=AFAM Alan Stern <stern@rowland.harvard.ed=
u> wrote:
>
> On Mon, Mar 04, 2024 at 08:10:02PM +0800, Sam Sun wrote:
> > Dear developers and maintainers,
> >
> > We encountered a task hung in function hub_activate(). It was reported
> > before by Syzbot several years ago
> > (https://groups.google.com/g/syzkaller-lts-bugs/c/_komEgHj03Y/m/rbcVKyL=
XBwAJ),
> > but no repro at that time. We have a C repro this time and kernel
> > config is attached to this email. The bug report is listed below.
>
> > If you have any questions, please contact us.
> >
> > Reported by Yue Sun <samsun1006219@gmail.com>
> > Reported by xingwei lee <xrivendell7@gmail.com>
>
> Yue:
>
> Can you try testing the patch below?  Thanks.
>
> Alan Stern
>
>
>
> Index: usb-devel/drivers/usb/core/sysfs.c
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> --- usb-devel.orig/drivers/usb/core/sysfs.c
> +++ usb-devel/drivers/usb/core/sysfs.c
> @@ -1168,14 +1168,24 @@ static ssize_t interface_authorized_stor
>  {
>         struct usb_interface *intf =3D to_usb_interface(dev);
>         bool val;
> +       struct kernfs_node *kn;
>
>         if (kstrtobool(buf, &val) !=3D 0)
>                 return -EINVAL;
>
> -       if (val)
> +       if (val) {
>                 usb_authorize_interface(intf);
> -       else
> -               usb_deauthorize_interface(intf);
> +       } else {
> +               /*
> +                * Prevent deadlock if another process is concurrently
> +                * trying to unregister intf or its parent device.
> +                */
> +               kn =3D sysfs_break_active_protection(&dev->kobj, &attr->a=
ttr);
> +               if (kn) {
> +                       usb_deauthorize_interface(intf);
> +                       sysfs_unbreak_active_protection(kn);
> +               }
> +       }
>
>         return count;
>  }

Hi Alan,

I applied this patch on the latest linux git commit and tested several
times. There is no more task hung report, so I think it works. Should
I try this on usb-dev branch?

Best,
Yue