Xen devel - Oct 2012 - [PATCH 0/5] drivers: xen frontend devices should handle missed backend CLOSING

Subsystem maintainers, you can either pick up the relevant driver patch
or ack it to go via Konrad''s Xen tree.

The series makes all the Xen frontend drivers handle the backend
transitioning to CLOSED without the frontend having previously seen
the backend in the CLOSING state.

Backends shouldn''t do this but some do.  e.g., if the host is
XenServer and the toolstack decides to do a forced shutdown of a VBD,
then the blkfront may miss the CLOSING transition and the /dev/xvdX
device will not be destroyed which prevents it being reused.

I have seen systems that ended up in this state but it''s not clear if
this was the actual cause.  However, I think in general it''s a good
thing to thing to improve the handling of unexpected state
transitions.

David

From: David Vrabel <david.vrabel@citrix.com>

Backend drivers shouldn''t transistion to CLOSED unless the frontend is
CLOSED.  If a backend does transition to CLOSED too soon then the
frontend may not see the CLOSING state and will not properly shutdown.

So, treat an unexpected backend CLOSED state the same as CLOSING.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
Cc: netdev@vger.kernel.org
Cc: "David S. Miller" <davem@davemloft.net>
---
 drivers/net/xen-netfront.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/drivers/net/xen-netfront.c b/drivers/net/xen-netfront.c
index caa0110..40cde51 100644
--- a/drivers/net/xen-netfront.c
+++ b/drivers/net/xen-netfront.c
@@ -1701,7 +1701,6 @@ static void netback_changed(struct xenbus_device *dev,
 	case XenbusStateReconfiguring:
 	case XenbusStateReconfigured:
 	case XenbusStateUnknown:
-	case XenbusStateClosed:
 		break;
 
 	case XenbusStateInitWait:
@@ -1716,6 +1715,10 @@ static void netback_changed(struct xenbus_device *dev,
 		netdev_notify_peers(netdev);
 		break;
 
+	case XenbusStateClosed:
+		if (dev->state == XenbusStateClosed)
+			break;
+		/* Missed the backend''s CLOSING state -- fallthrough */
 	case XenbusStateClosing:
 		xenbus_frontend_closed(dev);
 		break;
-- 
1.7.2.5

From: David Vrabel <david.vrabel@citrix.com>

Backend drivers shouldn''t transistion to CLOSED unless the frontend is
CLOSED.  If a backend does transition to CLOSED too soon then the
frontend may not see the CLOSING state and will not properly shutdown.

So, treat an unexpected backend CLOSED state the same as CLOSING.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
Cc: Jens Axboe <axboe@kernel.dk>
---
 drivers/block/xen-blkfront.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index 007db89..5d162b5 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -1340,13 +1340,16 @@ static void blkback_changed(struct xenbus_device *dev,
 	case XenbusStateReconfiguring:
 	case XenbusStateReconfigured:
 	case XenbusStateUnknown:
-	case XenbusStateClosed:
 		break;
 
 	case XenbusStateConnected:
 		blkfront_connect(info);
 		break;
 
+	case XenbusStateClosed:
+		if (dev->state == XenbusStateClosed)
+			break;
+		/* Missed the backend''s Closing state -- fallthrough */
 	case XenbusStateClosing:
 		blkfront_closing(info);
 		break;
-- 
1.7.2.5

From: David Vrabel <david.vrabel@citrix.com>

Backend drivers shouldn''t transistion to CLOSED unless the frontend is
CLOSED.  If a backend does transition to CLOSED too soon then the
frontend may not see the CLOSING state and will not properly shutdown.

So, treat an unexpected backend CLOSED state the same as CLOSING.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
Cc: linux-pci@vger.kernel.org
Cc: Bjorn Helgaas <bhelgaas@google.com>
---
 drivers/pci/xen-pcifront.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/drivers/pci/xen-pcifront.c b/drivers/pci/xen-pcifront.c
index 0aab85a..a0c7312 100644
--- a/drivers/pci/xen-pcifront.c
+++ b/drivers/pci/xen-pcifront.c
@@ -1068,13 +1068,16 @@ static void __init_refok pcifront_backend_changed(struct
xenbus_device *xdev,
 	case XenbusStateInitialising:
 	case XenbusStateInitWait:
 	case XenbusStateInitialised:
-	case XenbusStateClosed:
 		break;
 
 	case XenbusStateConnected:
 		pcifront_try_connect(pdev);
 		break;
 
+	case XenbusStateClosed:
+		if (xdev->state == XenbusStateClosed)
+			break;
+		/* Missed the backend''s CLOSING state -- fallthrough */
 	case XenbusStateClosing:
 		dev_warn(&xdev->dev, "backend going away!\n");
 		pcifront_try_disconnect(pdev);
-- 
1.7.2.5

From: David Vrabel <david.vrabel@citrix.com>

Backend drivers shouldn''t transistion to CLOSED unless the frontend is
CLOSED.  If a backend does transition to CLOSED too soon then the
frontend may not see the CLOSING state and will not properly shutdown.

So, treat an unexpected backend CLOSED state the same as CLOSING.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
Cc: linux-fbdev@vger.kernel.org
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
---
 drivers/video/xen-fbfront.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/drivers/video/xen-fbfront.c b/drivers/video/xen-fbfront.c
index b7f5173..917bb56 100644
--- a/drivers/video/xen-fbfront.c
+++ b/drivers/video/xen-fbfront.c
@@ -641,7 +641,6 @@ static void xenfb_backend_changed(struct xenbus_device *dev,
 	case XenbusStateReconfiguring:
 	case XenbusStateReconfigured:
 	case XenbusStateUnknown:
-	case XenbusStateClosed:
 		break;
 
 	case XenbusStateInitWait:
@@ -670,6 +669,10 @@ InitWait:
 		info->feature_resize = val;
 		break;
 
+	case XenbusStateClosed:
+		if (dev->state == XenbusStateClosed)
+			break;
+		/* Missed the backend''s CLOSING state -- fallthrough */
 	case XenbusStateClosing:
 		xenbus_frontend_closed(dev);
 		break;
-- 
1.7.2.5

From: David Vrabel <david.vrabel@citrix.com>

Backend drivers shouldn''t transistion to CLOSED unless the frontend is
CLOSED.  If a backend does transition to CLOSED too soon then the
frontend may not see the CLOSING state and will not properly shutdown.

So, treat an unexpected backend CLOSED state the same as CLOSING.

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
---
Cc: linux-input@vger.kernel.org
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
 drivers/input/misc/xen-kbdfront.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/drivers/input/misc/xen-kbdfront.c
b/drivers/input/misc/xen-kbdfront.c
index 02ca868..6f7d990 100644
--- a/drivers/input/misc/xen-kbdfront.c
+++ b/drivers/input/misc/xen-kbdfront.c
@@ -311,7 +311,6 @@ static void xenkbd_backend_changed(struct xenbus_device
*dev,
 	case XenbusStateReconfiguring:
 	case XenbusStateReconfigured:
 	case XenbusStateUnknown:
-	case XenbusStateClosed:
 		break;
 
 	case XenbusStateInitWait:
@@ -350,6 +349,10 @@ InitWait:
 
 		break;
 
+	case XenbusStateClosed:
+		if (dev->state == XenbusStateClosed)
+			break;
+		/* Missed the backend''s CLOSING state -- fallthrough */
 	case XenbusStateClosing:
 		xenbus_frontend_closed(dev);
 		break;
-- 
1.7.2.5

On Thu, Oct 18, 2012 at 11:03:36AM +0100, David Vrabel
wrote:
> From: David Vrabel <david.vrabel@citrix.com>
> 
> Backend drivers shouldn''t transistion to CLOSED unless the
frontend is
> CLOSED.  If a backend does transition to CLOSED too soon then the
> frontend may not see the CLOSING state and will not properly shutdown.
> 
> So, treat an unexpected backend CLOSED state the same as CLOSING.
> 
> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> ---
> Cc: linux-pci@vger.kernel.org
> Cc: Bjorn Helgaas <bhelgaas@google.com>
Bjorn, do you want me to prep a git pull with this patch
or can I have your Ack to put it my tree and have it part of my
git pull to Linus?

Thx.
> ---
>  drivers/pci/xen-pcifront.c |    5 ++++-
>  1 files changed, 4 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/pci/xen-pcifront.c b/drivers/pci/xen-pcifront.c
> index 0aab85a..a0c7312 100644
> --- a/drivers/pci/xen-pcifront.c
> +++ b/drivers/pci/xen-pcifront.c
> @@ -1068,13 +1068,16 @@ static void __init_refok
pcifront_backend_changed(struct xenbus_device *xdev,
>  	case XenbusStateInitialising:
>  	case XenbusStateInitWait:
>  	case XenbusStateInitialised:
> -	case XenbusStateClosed:
>  		break;
>  
>  	case XenbusStateConnected:
>  		pcifront_try_connect(pdev);
>  		break;
>  
> +	case XenbusStateClosed:
> +		if (xdev->state == XenbusStateClosed)
> +			break;
> +		/* Missed the backend''s CLOSING state -- fallthrough */
>  	case XenbusStateClosing:
>  		dev_warn(&xdev->dev, "backend going away!\n");
>  		pcifront_try_disconnect(pdev);
> -- 
> 1.7.2.5

On Thu, Oct 18, 2012 at 11:03:37AM +0100, David Vrabel
wrote:
> From: David Vrabel <david.vrabel@citrix.com>
> 
> Backend drivers shouldn''t transistion to CLOSED unless the
frontend is
> CLOSED.  If a backend does transition to CLOSED too soon then the
> frontend may not see the CLOSING state and will not properly shutdown.
> 
> So, treat an unexpected backend CLOSED state the same as CLOSING.
> 
> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> ---
> Cc: linux-fbdev@vger.kernel.org
> Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Hey Florian,

Should I prep a git pull for you with this or would it be OK
if I just have your Ack to put this in my git pull for Linus?

Thanks!
> ---
>  drivers/video/xen-fbfront.c |    5 ++++-
>  1 files changed, 4 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/video/xen-fbfront.c b/drivers/video/xen-fbfront.c
> index b7f5173..917bb56 100644
> --- a/drivers/video/xen-fbfront.c
> +++ b/drivers/video/xen-fbfront.c
> @@ -641,7 +641,6 @@ static void xenfb_backend_changed(struct xenbus_device
*dev,
>  	case XenbusStateReconfiguring:
>  	case XenbusStateReconfigured:
>  	case XenbusStateUnknown:
> -	case XenbusStateClosed:
>  		break;
>  
>  	case XenbusStateInitWait:
> @@ -670,6 +669,10 @@ InitWait:
>  		info->feature_resize = val;
>  		break;
>  
> +	case XenbusStateClosed:
> +		if (dev->state == XenbusStateClosed)
> +			break;
> +		/* Missed the backend''s CLOSING state -- fallthrough */
>  	case XenbusStateClosing:
>  		xenbus_frontend_closed(dev);
>  		break;
> -- 
> 1.7.2.5

On Thu, Oct 18, 2012 at 11:03:38AM +0100, David Vrabel
wrote:
> From: David Vrabel <david.vrabel@citrix.com>
> 
> Backend drivers shouldn''t transistion to CLOSED unless the
frontend is
> CLOSED.  If a backend does transition to CLOSED too soon then the
> frontend may not see the CLOSING state and will not properly shutdown.
> 
> So, treat an unexpected backend CLOSED state the same as CLOSING.
> 
> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> ---
> Cc: linux-input@vger.kernel.org
> Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Hey Dmitry,

Should I prep a git pull for you for this or are you OK giving
an Ack for me to put this patch in my git pull for Linus?

Thx.
> ---
>  drivers/input/misc/xen-kbdfront.c |    5 ++++-
>  1 files changed, 4 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/input/misc/xen-kbdfront.c
b/drivers/input/misc/xen-kbdfront.c
> index 02ca868..6f7d990 100644
> --- a/drivers/input/misc/xen-kbdfront.c
> +++ b/drivers/input/misc/xen-kbdfront.c
> @@ -311,7 +311,6 @@ static void xenkbd_backend_changed(struct xenbus_device
*dev,
>  	case XenbusStateReconfiguring:
>  	case XenbusStateReconfigured:
>  	case XenbusStateUnknown:
> -	case XenbusStateClosed:
>  		break;
>  
>  	case XenbusStateInitWait:
> @@ -350,6 +349,10 @@ InitWait:
>  
>  		break;
>  
> +	case XenbusStateClosed:
> +		if (dev->state == XenbusStateClosed)
> +			break;
> +		/* Missed the backend''s CLOSING state -- fallthrough */
>  	case XenbusStateClosing:
>  		xenbus_frontend_closed(dev);
>  		break;
> -- 
> 1.7.2.5

On Fri, Oct 19, 2012 at 09:00:59AM -0400, Konrad Rzeszutek Wilk
wrote:
> On Thu, Oct 18, 2012 at 11:03:38AM +0100, David Vrabel wrote:
> > From: David Vrabel <david.vrabel@citrix.com>
> > 
> > Backend drivers shouldn''t transistion to CLOSED unless the
frontend is
> > CLOSED.  If a backend does transition to CLOSED too soon then the
> > frontend may not see the CLOSING state and will not properly shutdown.
> > 
> > So, treat an unexpected backend CLOSED state the same as CLOSING.
> > 
> > Signed-off-by: David Vrabel <david.vrabel@citrix.com>
> > Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> > ---
> > Cc: linux-input@vger.kernel.org
> > Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> 
> Hey Dmitry,
> 
> Should I prep a git pull for you for this or are you OK giving
> an Ack for me to put this patch in my git pull for Linus?
Sure, please merge with the rest through your tree.

Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Thanks!
> 
> Thx.
> > ---
> >  drivers/input/misc/xen-kbdfront.c |    5 ++++-
> >  1 files changed, 4 insertions(+), 1 deletions(-)
> > 
> > diff --git a/drivers/input/misc/xen-kbdfront.c
b/drivers/input/misc/xen-kbdfront.c
> > index 02ca868..6f7d990 100644
> > --- a/drivers/input/misc/xen-kbdfront.c
> > +++ b/drivers/input/misc/xen-kbdfront.c
> > @@ -311,7 +311,6 @@ static void xenkbd_backend_changed(struct
xenbus_device *dev,
> >  	case XenbusStateReconfiguring:
> >  	case XenbusStateReconfigured:
> >  	case XenbusStateUnknown:
> > -	case XenbusStateClosed:
> >  		break;
> >  
> >  	case XenbusStateInitWait:
> > @@ -350,6 +349,10 @@ InitWait:
> >  
> >  		break;
> >  
> > +	case XenbusStateClosed:
> > +		if (dev->state == XenbusStateClosed)
> > +			break;
> > +		/* Missed the backend''s CLOSING state -- fallthrough */
> >  	case XenbusStateClosing:
> >  		xenbus_frontend_closed(dev);
> >  		break;
> > -- 
> > 1.7.2.5
-- 
Dmitry

Hi Konrad,

On 10/19/2012 01:00 PM, Konrad Rzeszutek Wilk wrote:
> On Thu, Oct 18, 2012 at 11:03:37AM +0100, David Vrabel wrote:
>> From: David Vrabel <david.vrabel@citrix.com>
>>
>> Backend drivers shouldn''t transistion to CLOSED unless the
frontend is
>> CLOSED.  If a backend does transition to CLOSED too soon then the
>> frontend may not see the CLOSING state and will not properly shutdown.
>>
>> So, treat an unexpected backend CLOSED state the same as CLOSING.
>>
>> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
>> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>> ---
>> Cc: linux-fbdev@vger.kernel.org
>> Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
> 
> Hey Florian,
> 
> Should I prep a git pull for you with this or would it be OK
> if I just have your Ack to put this in my git pull for Linus?
Feel free to take it and add
Acked-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>


Best regards,

Florian Tobias Schandinat
> 
> Thanks!
>> ---
>>  drivers/video/xen-fbfront.c |    5 ++++-
>>  1 files changed, 4 insertions(+), 1 deletions(-)
>>
>> diff --git a/drivers/video/xen-fbfront.c b/drivers/video/xen-fbfront.c
>> index b7f5173..917bb56 100644
>> --- a/drivers/video/xen-fbfront.c
>> +++ b/drivers/video/xen-fbfront.c
>> @@ -641,7 +641,6 @@ static void xenfb_backend_changed(struct
xenbus_device *dev,
>>  	case XenbusStateReconfiguring:
>>  	case XenbusStateReconfigured:
>>  	case XenbusStateUnknown:
>> -	case XenbusStateClosed:
>>  		break;
>>  
>>  	case XenbusStateInitWait:
>> @@ -670,6 +669,10 @@ InitWait:
>>  		info->feature_resize = val;
>>  		break;
>>  
>> +	case XenbusStateClosed:
>> +		if (dev->state == XenbusStateClosed)
>> +			break;
>> +		/* Missed the backend''s CLOSING state -- fallthrough */
>>  	case XenbusStateClosing:
>>  		xenbus_frontend_closed(dev);
>>  		break;
>> -- 
>> 1.7.2.5
>

On Fri, Oct 19, 2012 at 6:59 AM, Konrad Rzeszutek Wilk
<konrad.wilk@oracle.com> wrote:
> On Thu, Oct 18, 2012 at 11:03:36AM +0100, David Vrabel wrote:
>> From: David Vrabel <david.vrabel@citrix.com>
>>
>> Backend drivers shouldn''t transistion to CLOSED unless the
frontend is
>> CLOSED.  If a backend does transition to CLOSED too soon then the
>> frontend may not see the CLOSING state and will not properly shutdown.
>>
>> So, treat an unexpected backend CLOSED state the same as CLOSING.
>>
>> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
>> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>> ---
>> Cc: linux-pci@vger.kernel.org
>> Cc: Bjorn Helgaas <bhelgaas@google.com>
>
> Bjorn, do you want me to prep a git pull with this patch
> or can I have your Ack to put it my tree and have it part of my
> git pull to Linus?
Sorry, I missed this.  I can put it in my -next branch for the v3.8
merge window.  Would that work for you?
>> ---
>>  drivers/pci/xen-pcifront.c |    5 ++++-
>>  1 files changed, 4 insertions(+), 1 deletions(-)
>>
>> diff --git a/drivers/pci/xen-pcifront.c b/drivers/pci/xen-pcifront.c
>> index 0aab85a..a0c7312 100644
>> --- a/drivers/pci/xen-pcifront.c
>> +++ b/drivers/pci/xen-pcifront.c
>> @@ -1068,13 +1068,16 @@ static void __init_refok
pcifront_backend_changed(struct xenbus_device *xdev,
>>       case XenbusStateInitialising:
>>       case XenbusStateInitWait:
>>       case XenbusStateInitialised:
>> -     case XenbusStateClosed:
>>               break;
>>
>>       case XenbusStateConnected:
>>               pcifront_try_connect(pdev);
>>               break;
>>
>> +     case XenbusStateClosed:
>> +             if (xdev->state == XenbusStateClosed)
>> +                     break;
>> +             /* Missed the backend''s CLOSING state --
fallthrough */
>>       case XenbusStateClosing:
>>               dev_warn(&xdev->dev, "backend going
away!\n");
>>               pcifront_try_disconnect(pdev);
>> --
>> 1.7.2.5

On Fri, Nov 30, 2012 at 11:41 AM, Bjorn Helgaas <bhelgaas@google.com>
wrote:
> On Fri, Oct 19, 2012 at 6:59 AM, Konrad Rzeszutek Wilk
> <konrad.wilk@oracle.com> wrote:
>> On Thu, Oct 18, 2012 at 11:03:36AM +0100, David Vrabel wrote:
>>> From: David Vrabel <david.vrabel@citrix.com>
>>>
>>> Backend drivers shouldn''t transistion to CLOSED unless the
frontend is
>>> CLOSED.  If a backend does transition to CLOSED too soon then the
>>> frontend may not see the CLOSING state and will not properly
shutdown.
>>>
>>> So, treat an unexpected backend CLOSED state the same as CLOSING.
>>>
>>> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
>>> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>> ---
>>> Cc: linux-pci@vger.kernel.org
>>> Cc: Bjorn Helgaas <bhelgaas@google.com>
>>
>> Bjorn, do you want me to prep a git pull with this patch
>> or can I have your Ack to put it my tree and have it part of my
>> git pull to Linus?
>
> Sorry, I missed this.  I can put it in my -next branch for the v3.8
> merge window.  Would that work for you?
I put this in my -next branch, so we''ll at least have a chance of
making a linux-next cycle before v3.7 pops.
>>> ---
>>>  drivers/pci/xen-pcifront.c |    5 ++++-
>>>  1 files changed, 4 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/drivers/pci/xen-pcifront.c
b/drivers/pci/xen-pcifront.c
>>> index 0aab85a..a0c7312 100644
>>> --- a/drivers/pci/xen-pcifront.c
>>> +++ b/drivers/pci/xen-pcifront.c
>>> @@ -1068,13 +1068,16 @@ static void __init_refok
pcifront_backend_changed(struct xenbus_device *xdev,
>>>       case XenbusStateInitialising:
>>>       case XenbusStateInitWait:
>>>       case XenbusStateInitialised:
>>> -     case XenbusStateClosed:
>>>               break;
>>>
>>>       case XenbusStateConnected:
>>>               pcifront_try_connect(pdev);
>>>               break;
>>>
>>> +     case XenbusStateClosed:
>>> +             if (xdev->state == XenbusStateClosed)
>>> +                     break;
>>> +             /* Missed the backend''s CLOSING state --
fallthrough */
>>>       case XenbusStateClosing:
>>>               dev_warn(&xdev->dev, "backend going
away!\n");
>>>               pcifront_try_disconnect(pdev);
>>> --
>>> 1.7.2.5

On Fri, Nov 30, 2012 at 02:42:14PM -0700, Bjorn Helgaas
wrote:
> On Fri, Nov 30, 2012 at 11:41 AM, Bjorn Helgaas <bhelgaas@google.com>
wrote:
> > On Fri, Oct 19, 2012 at 6:59 AM, Konrad Rzeszutek Wilk
> > <konrad.wilk@oracle.com> wrote:
> >> On Thu, Oct 18, 2012 at 11:03:36AM +0100, David Vrabel wrote:
> >>> From: David Vrabel <david.vrabel@citrix.com>
> >>>
> >>> Backend drivers shouldn''t transistion to CLOSED
unless the frontend is
> >>> CLOSED.  If a backend does transition to CLOSED too soon then
the
> >>> frontend may not see the CLOSING state and will not properly
shutdown.
> >>>
> >>> So, treat an unexpected backend CLOSED state the same as
CLOSING.
> >>>
> >>> Signed-off-by: David Vrabel <david.vrabel@citrix.com>
> >>> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> >>> ---
> >>> Cc: linux-pci@vger.kernel.org
> >>> Cc: Bjorn Helgaas <bhelgaas@google.com>
> >>
> >> Bjorn, do you want me to prep a git pull with this patch
> >> or can I have your Ack to put it my tree and have it part of my
> >> git pull to Linus?
> >
> > Sorry, I missed this.  I can put it in my -next branch for the v3.8
> > merge window.  Would that work for you?
> 
> I put this in my -next branch, so we''ll at least have a chance of
> making a linux-next cycle before v3.7 pops.
Great. Thx!
> 
> >>> ---
> >>>  drivers/pci/xen-pcifront.c |    5 ++++-
> >>>  1 files changed, 4 insertions(+), 1 deletions(-)
> >>>
> >>> diff --git a/drivers/pci/xen-pcifront.c
b/drivers/pci/xen-pcifront.c
> >>> index 0aab85a..a0c7312 100644
> >>> --- a/drivers/pci/xen-pcifront.c
> >>> +++ b/drivers/pci/xen-pcifront.c
> >>> @@ -1068,13 +1068,16 @@ static void __init_refok
pcifront_backend_changed(struct xenbus_device *xdev,
> >>>       case XenbusStateInitialising:
> >>>       case XenbusStateInitWait:
> >>>       case XenbusStateInitialised:
> >>> -     case XenbusStateClosed:
> >>>               break;
> >>>
> >>>       case XenbusStateConnected:
> >>>               pcifront_try_connect(pdev);
> >>>               break;
> >>>
> >>> +     case XenbusStateClosed:
> >>> +             if (xdev->state == XenbusStateClosed)
> >>> +                     break;
> >>> +             /* Missed the backend''s CLOSING state
-- fallthrough */
> >>>       case XenbusStateClosing:
> >>>               dev_warn(&xdev->dev, "backend going
away!\n");
> >>>               pcifront_try_disconnect(pdev);
> >>> --
> >>> 1.7.2.5