netdev - Re: Re: [PATCH 02/11] drivers: usb: host: Fix deadlock in oxu_bus

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <4170b816.3f4a8.18003e801f8.Coremail.duoming@zju.edu.cn>
Date:   Thu, 7 Apr 2022 20:02:58 +0800 (GMT+08:00)
From:   duoming@....edu.cn
To:     "Oliver Neukum" <oneukum@...e.com>
Cc:     linux-kernel@...r.kernel.org, chris@...kel.net, jcmvbkbc@...il.com,
        mustafa.ismail@...el.com, shiraz.saleem@...el.com, jgg@...pe.ca,
        wg@...ndegger.com, mkl@...gutronix.de, davem@...emloft.net,
        kuba@...nel.org, pabeni@...hat.com, jes@...ined-monkey.org,
        gregkh@...uxfoundation.org, jirislaby@...nel.org,
        alexander.deucher@....com, linux-xtensa@...ux-xtensa.org,
        linux-rdma@...r.kernel.org, linux-can@...r.kernel.org,
        netdev@...r.kernel.org, linux-hippi@...site.dk,
        linux-staging@...ts.linux.dev, linux-serial@...r.kernel.org,
        linux-usb@...r.kernel.org, linma@....edu.cn
Subject: Re: Re: [PATCH 02/11] drivers: usb: host: Fix deadlock in
 oxu_bus_suspend()

Hello,

On Thu, 7 Apr 2022 10:01:43 +0200 Oliver Neukum wrote:

> On 07.04.22 08:33, Duoming Zhou wrote:
> > There is a deadlock in oxu_bus_suspend(), which is shown below:
> >
> >    (Thread 1)              |      (Thread 2)
> >                            | timer_action()
> > oxu_bus_suspend()          |  mod_timer()
> >  spin_lock_irq() //(1)     |  (wait a time)
> >  ...                       | oxu_watchdog()
> >  del_timer_sync()          |  spin_lock_irq() //(2)
> >  (wait timer to stop)      |  ...
> >
> > We hold oxu->lock in position (1) of thread 1, and use
> > del_timer_sync() to wait timer to stop, but timer handler
> > also need oxu->lock in position (2) of thread 2. As a result,
> > oxu_bus_suspend() will block forever.
> >
> > This patch extracts del_timer_sync() from the protection of
> > spin_lock_irq(), which could let timer handler to obtain
> > the needed lock.
> Good catch.
> > Signed-off-by: Duoming Zhou <duoming@....edu.cn>
> > ---
> >  drivers/usb/host/oxu210hp-hcd.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/usb/host/oxu210hp-hcd.c b/drivers/usb/host/oxu210hp-hcd.c
> > index b741670525e..ee403df3309 100644
> > --- a/drivers/usb/host/oxu210hp-hcd.c
> > +++ b/drivers/usb/host/oxu210hp-hcd.c
> > @@ -3909,8 +3909,10 @@ static int oxu_bus_suspend(struct usb_hcd *hcd)
> >  		}
> >  	}
> >  
> > +	spin_unlock_irq(&oxu->lock);
> >  	/* turn off now-idle HC */
> >  	del_timer_sync(&oxu->watchdog);
> > +	spin_lock_irq(&oxu->lock);
> >  	ehci_halt(oxu);
> >  	hcd->state = HC_STATE_SUSPENDED;
> >  
> 
> What is the lock protecting at that stage? Why not just drop it earlier.

I think there is a race condition between oxu_bus_suspend() and oxu_stop(),
so I think we could not drop the oxu->lock earlier.

               (Thread 1)               |         (Thread 2)
 oxu_bus_suspend()                      | oxu_stop()
                                        | 
 hcd->state = HC_STATE_SUSPENDED;       |  spin_lock_irq(&oxu->lock);
 ...                                    | 
 writel(mask, &oxu->regs->intr_enable); |  ...
                                        |  writel(0, &oxu->regs->intr_enable);
 readl(&oxu->regs->intr_enable);        |

The oxu->regs->intr_enable is set to 0 in oxu_stop(), and the readl() in
oxu_bus_suspend() will read the wrong value.

Thanks a lot for your time and advice. If you have questions, welcome to ask me.

Best regards,
Duoming Zhou