David Vrabel
2013-Oct-07 12:55 UTC
[PATCHv1 net] xen-netback: transition to CLOSED when removing a VIF
From: David Vrabel <david.vrabel@citrix.com>
If a guest is destroyed without transitioning its frontend to CLOSED,
the domain becomes a zombie as netback was not grant unmapping the
shared rings.
When removing a VIF, transition the backend to CLOSED so the VIF is
disconnected if necessary (which will unmap the shared rings etc).
This fixes a regression introduced by
279f438e36c0a70b23b86d2090aeec50155034a9 (xen-netback: Don''t destroy
the netdev until the vif is shut down).
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Cc: Paul Durrant <Paul.Durrant@citrix.com>
---
drivers/net/xen-netback/xenbus.c | 4 ++++
1 files changed, 4 insertions(+), 0 deletions(-)
diff --git a/drivers/net/xen-netback/xenbus.c b/drivers/net/xen-netback/xenbus.c
index b45bce2..1b08d87 100644
--- a/drivers/net/xen-netback/xenbus.c
+++ b/drivers/net/xen-netback/xenbus.c
@@ -39,11 +39,15 @@ static int connect_rings(struct backend_info *);
static void connect(struct backend_info *);
static void backend_create_xenvif(struct backend_info *be);
static void unregister_hotplug_status_watch(struct backend_info *be);
+static void set_backend_state(struct backend_info *be,
+ enum xenbus_state state);
static int netback_remove(struct xenbus_device *dev)
{
struct backend_info *be = dev_get_drvdata(&dev->dev);
+ set_backend_state(be, XenbusStateClosed);
+
unregister_hotplug_status_watch(be);
if (be->vif) {
kobject_uevent(&dev->dev.kobj, KOBJ_OFFLINE);
--
1.7.2.5
Wei Liu
2013-Oct-07 13:43 UTC
Re: [PATCHv1 net] xen-netback: transition to CLOSED when removing a VIF
On Mon, Oct 07, 2013 at 01:55:19PM +0100, David Vrabel wrote:> From: David Vrabel <david.vrabel@citrix.com> > > If a guest is destroyed without transitioning its frontend to CLOSED, > the domain becomes a zombie as netback was not grant unmapping the > shared rings. > > When removing a VIF, transition the backend to CLOSED so the VIF is > disconnected if necessary (which will unmap the shared rings etc). > > This fixes a regression introduced by > 279f438e36c0a70b23b86d2090aeec50155034a9 (xen-netback: Don''t destroy > the netdev until the vif is shut down). >Is this regression solely caused by 279f438e36c or caused by both ea732dff5c and 279f438e36c? I ask because you make use of the new state machine introduced in ea732dff5c. Or are you simply using the new state machine to fix the regression instead of going back to old code? Wei.> Signed-off-by: David Vrabel <david.vrabel@citrix.com> > Cc: Ian Campbell <ian.campbell@citrix.com> > Cc: Wei Liu <wei.liu2@citrix.com> > Cc: Paul Durrant <Paul.Durrant@citrix.com> > --- > drivers/net/xen-netback/xenbus.c | 4 ++++ > 1 files changed, 4 insertions(+), 0 deletions(-) > > diff --git a/drivers/net/xen-netback/xenbus.c b/drivers/net/xen-netback/xenbus.c > index b45bce2..1b08d87 100644 > --- a/drivers/net/xen-netback/xenbus.c > +++ b/drivers/net/xen-netback/xenbus.c > @@ -39,11 +39,15 @@ static int connect_rings(struct backend_info *); > static void connect(struct backend_info *); > static void backend_create_xenvif(struct backend_info *be); > static void unregister_hotplug_status_watch(struct backend_info *be); > +static void set_backend_state(struct backend_info *be, > + enum xenbus_state state); > > static int netback_remove(struct xenbus_device *dev) > { > struct backend_info *be = dev_get_drvdata(&dev->dev); > > + set_backend_state(be, XenbusStateClosed); > + > unregister_hotplug_status_watch(be); > if (be->vif) { > kobject_uevent(&dev->dev.kobj, KOBJ_OFFLINE); > -- > 1.7.2.5
David Vrabel
2013-Oct-07 13:57 UTC
Re: [PATCHv1 net] xen-netback: transition to CLOSED when removing a VIF
On 07/10/13 14:43, Wei Liu wrote:> On Mon, Oct 07, 2013 at 01:55:19PM +0100, David Vrabel wrote: >> From: David Vrabel <david.vrabel@citrix.com> >> >> If a guest is destroyed without transitioning its frontend to CLOSED, >> the domain becomes a zombie as netback was not grant unmapping the >> shared rings. >> >> When removing a VIF, transition the backend to CLOSED so the VIF is >> disconnected if necessary (which will unmap the shared rings etc). >> >> This fixes a regression introduced by >> 279f438e36c0a70b23b86d2090aeec50155034a9 (xen-netback: Don''t destroy >> the netdev until the vif is shut down). >> > > Is this regression solely caused by 279f438e36c or caused by both > ea732dff5c and 279f438e36c? I ask because you make use of the new state > machine introduced in ea732dff5c. Or are you simply using the new state > machine to fix the regression instead of going back to old code?I bisected it to 279f438. I''m just using the handy new state machine to fix it. David
Wei Liu
2013-Oct-07 14:15 UTC
Re: [PATCHv1 net] xen-netback: transition to CLOSED when removing a VIF
On Mon, Oct 07, 2013 at 02:57:37PM +0100, David Vrabel wrote:> On 07/10/13 14:43, Wei Liu wrote: > > On Mon, Oct 07, 2013 at 01:55:19PM +0100, David Vrabel wrote: > >> From: David Vrabel <david.vrabel@citrix.com> > >> > >> If a guest is destroyed without transitioning its frontend to CLOSED, > >> the domain becomes a zombie as netback was not grant unmapping the > >> shared rings. > >> > >> When removing a VIF, transition the backend to CLOSED so the VIF is > >> disconnected if necessary (which will unmap the shared rings etc). > >> > >> This fixes a regression introduced by > >> 279f438e36c0a70b23b86d2090aeec50155034a9 (xen-netback: Don''t destroy > >> the netdev until the vif is shut down). > >> > > > > Is this regression solely caused by 279f438e36c or caused by both > > ea732dff5c and 279f438e36c? I ask because you make use of the new state > > machine introduced in ea732dff5c. Or are you simply using the new state > > machine to fix the regression instead of going back to old code? > > I bisected it to 279f438. I''m just using the handy new state machine to > fix it. >Thanks for the explanation. Acked-by: Wei Liu <wei.liu2@citrix.com> Wei.> David
Paul Durrant
2013-Oct-07 14:28 UTC
Re: [PATCHv1 net] xen-netback: transition to CLOSED when removing a VIF
> -----Original Message----- > From: David Vrabel [mailto:david.vrabel@citrix.com] > Sent: 07 October 2013 13:55 > To: xen-devel@lists.xen.org > Cc: David Vrabel; Konrad Rzeszutek Wilk; Boris Ostrovsky; > netdev@vger.kernel.org; Ian Campbell; Wei Liu; Paul Durrant > Subject: [PATCHv1 net] xen-netback: transition to CLOSED when removing a > VIF > > From: David Vrabel <david.vrabel@citrix.com> > > If a guest is destroyed without transitioning its frontend to CLOSED, > the domain becomes a zombie as netback was not grant unmapping the > shared rings. > > When removing a VIF, transition the backend to CLOSED so the VIF is > disconnected if necessary (which will unmap the shared rings etc). > > This fixes a regression introduced by > 279f438e36c0a70b23b86d2090aeec50155034a9 (xen-netback: Don''t destroy > the netdev until the vif is shut down). > > Signed-off-by: David Vrabel <david.vrabel@citrix.com> > Cc: Ian Campbell <ian.campbell@citrix.com> > Cc: Wei Liu <wei.liu2@citrix.com> > Cc: Paul Durrant <Paul.Durrant@citrix.com> > --- > drivers/net/xen-netback/xenbus.c | 4 ++++ > 1 files changed, 4 insertions(+), 0 deletions(-) > > diff --git a/drivers/net/xen-netback/xenbus.c b/drivers/net/xen- > netback/xenbus.c > index b45bce2..1b08d87 100644 > --- a/drivers/net/xen-netback/xenbus.c > +++ b/drivers/net/xen-netback/xenbus.c > @@ -39,11 +39,15 @@ static int connect_rings(struct backend_info *); > static void connect(struct backend_info *); > static void backend_create_xenvif(struct backend_info *be); > static void unregister_hotplug_status_watch(struct backend_info *be); > +static void set_backend_state(struct backend_info *be, > + enum xenbus_state state); > > static int netback_remove(struct xenbus_device *dev) > { > struct backend_info *be = dev_get_drvdata(&dev->dev); > > + set_backend_state(be, XenbusStateClosed); > + > unregister_hotplug_status_watch(be); > if (be->vif) { > kobject_uevent(&dev->dev.kobj, KOBJ_OFFLINE); > -- > 1.7.2.5Looks good to me. Reviewed-by: Paul Durrant <paul.durrant@citrix.com> Paul
David Miller
2013-Oct-09 01:34 UTC
Re: [PATCHv1 net] xen-netback: transition to CLOSED when removing a VIF
From: David Vrabel <david.vrabel@citrix.com> Date: Mon, 7 Oct 2013 13:55:19 +0100> From: David Vrabel <david.vrabel@citrix.com> > > If a guest is destroyed without transitioning its frontend to CLOSED, > the domain becomes a zombie as netback was not grant unmapping the > shared rings. > > When removing a VIF, transition the backend to CLOSED so the VIF is > disconnected if necessary (which will unmap the shared rings etc). > > This fixes a regression introduced by > 279f438e36c0a70b23b86d2090aeec50155034a9 (xen-netback: Don''t destroy > the netdev until the vif is shut down). > > Signed-off-by: David Vrabel <david.vrabel@citrix.com>Applied, thanks.