[2.6.37-rc1, patch] r8169: fix sleeping while holding spinlock...

[2.6.37-rc1, patch] r8169: fix sleeping while holding spinlock...

[Daniel J Blueman]

Recent changes to the r8169 driver cause it to call
device_set_wakeup_enable under spinlock, which may sleep.

Locking isn't necessary around the call to device_set_wakeup_enable,
so drop the spinlock before calling it, to address this.

Signed-off-by: Daniel J Blueman <daniel.blueman@gmail.com>

diff --git a/drivers/net/r8169.c b/drivers/net/r8169.c
index d88ce9f..894e7c7 100644
--- a/drivers/net/r8169.c
+++ b/drivers/net/r8169.c
@@ -846,10 +846,10 @@ static int rtl8169_set_wol(struct net_device
*dev, struct ethtool_wolinfo *wol)
 	else
 		tp->features &= ~RTL_FEATURE_WOL;
 	__rtl8169_set_wol(tp, wol->wolopts);
-	device_set_wakeup_enable(&tp->pci_dev->dev, wol->wolopts);
-
 	spin_unlock_irq(&tp->lock);

+	device_set_wakeup_enable(&tp->pci_dev->dev, wol->wolopts);
+
 	return 0;
 }

-- 
Daniel J Blueman

Re: [2.6.37-rc1, patch] r8169: fix sleeping while holding spinlock...

[Francois Romieu]

Daniel J Blueman <daniel.blueman@gmail.com> :
> Recent changes to the r8169 driver cause it to call
> device_set_wakeup_enable under spinlock, which may sleep.

The change is fine but the description is misleading : the r8169
driver has been issuing device_set_wakeup_enable under spinlock since
october 2008. device_set_wakeup_enable did not sleep until recently
(see f2dc0d1809ab7e0147c7e4ac837be1682f706538 for instance).

drivers/net/gianfar_ethtool.c::gfar_set_wol is broken as well when
CONFIG_PM is set.

-- 
Ueimor

Re: [2.6.37-rc1, patch v2] r8169: fix sleeping while holding spinlock...

[Daniel J Blueman]

Hi Francois,

On 1 November 2010 23:46, Francois Romieu <romieu@fr.zoreil.com> wrote:
> Daniel J Blueman <daniel.blueman@gmail.com> :
>> Recent changes to the r8169 driver cause it to call
>> device_set_wakeup_enable under spinlock, which may sleep.
>
> The change is fine but the description is misleading : the r8169
> driver has been issuing device_set_wakeup_enable under spinlock since
> october 2008. device_set_wakeup_enable did not sleep until recently
> (see f2dc0d1809ab7e0147c7e4ac837be1682f706538 for instance).

Good catch; I also find that only the gainfar is the other driver
needing fixing; I'll follow up with this.

Patch with updated description for David:

As device_set_wakeup_enable can now sleep, move the call to outside
the critical section.

Signed-off-by: Daniel J Blueman <daniel.blueman@gmail.com>

diff --git a/drivers/net/r8169.c b/drivers/net/r8169.c
index d88ce9f..894e7c7 100644
--- a/drivers/net/r8169.c
+++ b/drivers/net/r8169.c
@@ -846,10 +846,10 @@ static int rtl8169_set_wol(struct net_device
*dev, struct ethtool_wolinfo *wol)
 	else
 		tp->features &= ~RTL_FEATURE_WOL;
 	__rtl8169_set_wol(tp, wol->wolopts);
-	device_set_wakeup_enable(&tp->pci_dev->dev, wol->wolopts);
-
 	spin_unlock_irq(&tp->lock);

+	device_set_wakeup_enable(&tp->pci_dev->dev, wol->wolopts);
+
 	return 0;
 }

-- 
Daniel J Blueman

Re: [2.6.37-rc1, patch v2] r8169: fix sleeping while holding spinlock...

[Rafael J. Wysocki]

On Tuesday, November 02, 2010, Daniel J Blueman wrote:
> Hi Francois,
> 
> On 1 November 2010 23:46, Francois Romieu <romieu@fr.zoreil.com> wrote:
> > Daniel J Blueman <daniel.blueman@gmail.com> :
> >> Recent changes to the r8169 driver cause it to call
> >> device_set_wakeup_enable under spinlock, which may sleep.
> >
> > The change is fine but the description is misleading : the r8169
> > driver has been issuing device_set_wakeup_enable under spinlock since
> > october 2008. device_set_wakeup_enable did not sleep until recently
> > (see f2dc0d1809ab7e0147c7e4ac837be1682f706538 for instance).

Yes, sorry about that.  I overlooked the fact that these drivers called
device_set_wakeup_enable() under spinlocks.

> Good catch; I also find that only the gainfar is the other driver
> needing fixing; I'll follow up with this.

Thanks a lot for taking care of this!

> Patch with updated description for David:
> 
> As device_set_wakeup_enable can now sleep, move the call to outside
> the critical section.
> 
> Signed-off-by: Daniel J Blueman <daniel.blueman@gmail.com>

Acked-by: Rafael J. Wysocki <rjw@sisk.pl>

> diff --git a/drivers/net/r8169.c b/drivers/net/r8169.c
> index d88ce9f..894e7c7 100644
> --- a/drivers/net/r8169.c
> +++ b/drivers/net/r8169.c
> @@ -846,10 +846,10 @@ static int rtl8169_set_wol(struct net_device
> *dev, struct ethtool_wolinfo *wol)
>  	else
>  		tp->features &= ~RTL_FEATURE_WOL;
>  	__rtl8169_set_wol(tp, wol->wolopts);
> -	device_set_wakeup_enable(&tp->pci_dev->dev, wol->wolopts);
> -
>  	spin_unlock_irq(&tp->lock);
> 
> +	device_set_wakeup_enable(&tp->pci_dev->dev, wol->wolopts);
> +
>  	return 0;
>  }
> 
>