[Qemu-devel] [PATCH 0/2] balloon: add a feature bit to let Guest OS deflate virtio_balloon on OOM

[Qemu-devel] [PATCH 0/2] balloon: add a feature bit to let Guest OS deflate virtio_balloon on OOM

[Denis V. Lunev]

Excessive virtio_balloon inflation can cause invocation of OOM-killer,
when Linux is under severe memory pressure. Various mechanisms are
responsible for correct virtio_balloon memory management. Nevertheless it
is often the case that these control tools does not have enough time to
react on fast changing memory load. As a result OS runs out of memory and
invokes OOM-killer. The balancing of memory by use of the virtio balloon
should not cause the termination of processes while there are pages in the
balloon. Now there is no way for virtio balloon driver to free memory at
the last moment before some process get killed by OOM-killer.

This does not provide a security breach as balloon itself is running
inside Guest OS and is working in the cooperation with the host. Thus
some improvements from Guest side should be considered as normal.

To solve the problem, introduce a virtio_balloon callback which is
expected to be called from the oom notifier call chain in out_of_memory()
function. If virtio balloon could release some memory, it will make the
system to return and retry the allocation that forced the out of memory
killer to run.

This behavior should be enabled if and only if appropriate feature bit
is set on the device. It is off by default.

This functionality was recently merged into vanilla Linux (actually in
linux-next at the moment)

  commit 5a10b7dbf904bfe01bb9fcc6298f7df09eed77d5
  Author: Raushaniya Maksudova <rmaksudova@parallels.com>
  Date:   Mon Nov 10 09:36:29 2014 +1030

This patch adds respective control bits into QEMU. It introduces
deflate-on-oom option for baloon device which do the trick.

Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Anthony Liguori <aliguori@amazon.com>
CC: Michael S. Tsirkin <mst@redhat.com>

[Qemu-devel] [PATCH 1/2] balloon: call qdev_alias_all_properties for proxy dev in balloon class init

[Denis V. Lunev]

From: Raushaniya Maksudova <rmaksudova@parallels.com>

The idea is that all other virtio devices are calling this helper
to merge properties of the proxy device. This is the only difference
in between this helper and code in inside virtio_instance_init_common.
The patch should not cause any harm as property list in generic balloon
code is empty.

This also allows to avoid some dummy errors like fixed by this
    commit 91ba21208839643603e7f7fa5864723c3f371ebe
    Author: Gonglei <arei.gonglei@huawei.com>
    Date:   Tue Sep 30 14:10:35 2014 +0800
    virtio-balloon: fix virtio-balloon child refcount in transports

Signed-off-by: Denis V. Lunev <den@openvz.org>
Acked-by: Raushaniya Maksudova <rmaksudova@parallels.com>
CC: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: Christian Borntraeger <borntraeger@de.ibm.com>
CC: Anthony Liguori <aliguori@amazon.com>
CC: Michael S. Tsirkin <mst@redhat.com>
---
 hw/s390x/virtio-ccw.c  | 5 ++---
 hw/virtio/virtio-pci.c | 5 ++---
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c
index ea236c9..82da894 100644
--- a/hw/s390x/virtio-ccw.c
+++ b/hw/s390x/virtio-ccw.c
@@ -899,9 +899,8 @@ static void balloon_ccw_stats_set_poll_interval(Object *obj, struct Visitor *v,
 static void virtio_ccw_balloon_instance_init(Object *obj)
 {
     VirtIOBalloonCcw *dev = VIRTIO_BALLOON_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_ccw_stats_get_all, NULL, NULL, dev, NULL);
 
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index dde1d73..745324b 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -1316,9 +1316,8 @@ static void virtio_balloon_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_balloon_pci_instance_init(Object *obj)
 {
     VirtIOBalloonPCI *dev = VIRTIO_BALLOON_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_pci_stats_get_all, NULL, NULL, dev,
                         NULL);
-- 
1.9.1

[Qemu-devel] [PATCH 2/2] balloon: add a feature bit to let Guest OS deflate balloon on oom

[Denis V. Lunev]

From: Raushaniya Maksudova <rmaksudova@parallels.com>

Excessive virtio_balloon inflation can cause invocation of OOM-killer,
when Linux is under severe memory pressure. Various mechanisms are
responsible for correct virtio_balloon memory management. Nevertheless it
is often the case that these control tools does not have enough time to
react on fast changing memory load. As a result OS runs out of memory and
invokes OOM-killer. The balancing of memory by use of the virtio balloon
should not cause the termination of processes while there are pages in the
balloon. Now there is no way for virtio balloon driver to free memory at
the last moment before some process get killed by OOM-killer.

This does not provide a security breach as balloon itself is running
inside Guest OS and is working in the cooperation with the host. Thus
some improvements from Guest side should be considered as normal.

To solve the problem, introduce a virtio_balloon callback which is
expected to be called from the oom notifier call chain in out_of_memory()
function. If virtio balloon could release some memory, it will make the
system to return and retry the allocation that forced the out of memory
killer to run.

This behavior should be enabled if and only if appropriate feature bit
is set on the device. It is off by default.

This functionality was recently merged into vanilla Linux (actually in
linux-next at the moment)

  commit 5a10b7dbf904bfe01bb9fcc6298f7df09eed77d5
  Author: Raushaniya Maksudova <rmaksudova@parallels.com>
  Date:   Mon Nov 10 09:36:29 2014 +1030

This patch adds respective control bits into QEMU. It introduces
deflate-on-oom option for baloon device which do the trick.

Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Anthony Liguori <aliguori@amazon.com>
CC: Michael S. Tsirkin <mst@redhat.com>
---
 hw/virtio/virtio-balloon.c         | 7 +++++++
 include/hw/virtio/virtio-balloon.h | 2 ++
 qemu-options.hx                    | 6 +++++-
 3 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/hw/virtio/virtio-balloon.c b/hw/virtio/virtio-balloon.c
index 7bfbb75..9d145fa 100644
--- a/hw/virtio/virtio-balloon.c
+++ b/hw/virtio/virtio-balloon.c
@@ -305,7 +305,12 @@ static void virtio_balloon_set_config(VirtIODevice *vdev,
 
 static uint32_t virtio_balloon_get_features(VirtIODevice *vdev, uint32_t f)
 {
+    VirtIOBalloon *dev = VIRTIO_BALLOON(vdev);
     f |= (1 << VIRTIO_BALLOON_F_STATS_VQ);
+    if (dev->deflate_on_oom) {
+        f |= (1 << VIRTIO_BALLOON_F_DEFLATE_ON_OOM);
+    }
+
     return f;
 }
 
@@ -409,6 +414,7 @@ static void virtio_balloon_device_unrealize(DeviceState *dev, Error **errp)
 }
 
 static Property virtio_balloon_properties[] = {
+    DEFINE_PROP_BOOL("deflate-on-oom", VirtIOBalloon, deflate_on_oom, false),
     DEFINE_PROP_END_OF_LIST(),
 };
 
diff --git a/include/hw/virtio/virtio-balloon.h b/include/hw/virtio/virtio-balloon.h
index f863bfe..45cc55a 100644
--- a/include/hw/virtio/virtio-balloon.h
+++ b/include/hw/virtio/virtio-balloon.h
@@ -30,6 +30,7 @@
 /* The feature bitmap for virtio balloon */
 #define VIRTIO_BALLOON_F_MUST_TELL_HOST 0 /* Tell before reclaiming pages */
 #define VIRTIO_BALLOON_F_STATS_VQ 1       /* Memory stats virtqueue */
+#define VIRTIO_BALLOON_F_DEFLATE_ON_OOM 2 /* Deflate balloon on OOM */
 
 /* Size of a PFN in the balloon interface. */
 #define VIRTIO_BALLOON_PFN_SHIFT 12
@@ -67,6 +68,7 @@ typedef struct VirtIOBalloon {
     QEMUTimer *stats_timer;
     int64_t stats_last_update;
     int64_t stats_poll_interval;
+    bool deflate_on_oom;
 } VirtIOBalloon;
 
 #endif
diff --git a/qemu-options.hx b/qemu-options.hx
index da9851d..14ede0b 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -324,7 +324,8 @@ ETEXI
 DEF("balloon", HAS_ARG, QEMU_OPTION_balloon,
     "-balloon none   disable balloon device\n"
     "-balloon virtio[,addr=str]\n"
-    "                enable virtio balloon device (default)\n", QEMU_ARCH_ALL)
+    "                enable virtio balloon device (default)\n"
+    "               [,deflate-on-oom=on|off]\n", QEMU_ARCH_ALL)
 STEXI
 @item -balloon none
 @findex -balloon
@@ -332,6 +333,9 @@ Disable balloon device.
 @item -balloon virtio[,addr=@var{addr}]
 Enable virtio balloon device (default), optionally with PCI address
 @var{addr}.
+@item -balloon virtio[,deflate-on-oom=@var{deflate-on-oom}]
+@var{deflate-on-oom} is "on" or "off" and enables whether to let Guest OS
+to deflate virtio balloon on OOM. Default is off.
 ETEXI
 
 DEF("device", HAS_ARG, QEMU_OPTION_device,
-- 
1.9.1

Re: [Qemu-devel] [PATCH 2/2] balloon: add a feature bit to let Guest OS deflate balloon on oom

[Michael S. Tsirkin]

On Wed, Nov 26, 2014 at 01:11:25PM +0300, Denis V. Lunev wrote:
> From: Raushaniya Maksudova <rmaksudova@parallels.com>
> 
> Excessive virtio_balloon inflation can cause invocation of OOM-killer,
> when Linux is under severe memory pressure. Various mechanisms are
> responsible for correct virtio_balloon memory management. Nevertheless it
> is often the case that these control tools does not have enough time to
> react on fast changing memory load. As a result OS runs out of memory and
> invokes OOM-killer. The balancing of memory by use of the virtio balloon
> should not cause the termination of processes while there are pages in the
> balloon. Now there is no way for virtio balloon driver to free memory at
> the last moment before some process get killed by OOM-killer.
> 
> This does not provide a security breach as balloon itself is running
> inside Guest OS and is working in the cooperation with the host. Thus
> some improvements from Guest side should be considered as normal.
> 
> To solve the problem, introduce a virtio_balloon callback which is
> expected to be called from the oom notifier call chain in out_of_memory()
> function. If virtio balloon could release some memory, it will make the
> system to return and retry the allocation that forced the out of memory
> killer to run.
> 
> This behavior should be enabled if and only if appropriate feature bit
> is set on the device. It is off by default.
> 
> This functionality was recently merged into vanilla Linux (actually in
> linux-next at the moment)
> 
>   commit 5a10b7dbf904bfe01bb9fcc6298f7df09eed77d5
>   Author: Raushaniya Maksudova <rmaksudova@parallels.com>
>   Date:   Mon Nov 10 09:36:29 2014 +1030
> 
> This patch adds respective control bits into QEMU. It introduces
> deflate-on-oom option for baloon device which do the trick.
> 
> Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
> Signed-off-by: Denis V. Lunev <den@openvz.org>
> CC: Anthony Liguori <aliguori@amazon.com>
> CC: Michael S. Tsirkin <mst@redhat.com>
> ---
>  hw/virtio/virtio-balloon.c         | 7 +++++++
>  include/hw/virtio/virtio-balloon.h | 2 ++
>  qemu-options.hx                    | 6 +++++-
>  3 files changed, 14 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/virtio/virtio-balloon.c b/hw/virtio/virtio-balloon.c
> index 7bfbb75..9d145fa 100644
> --- a/hw/virtio/virtio-balloon.c
> +++ b/hw/virtio/virtio-balloon.c
> @@ -305,7 +305,12 @@ static void virtio_balloon_set_config(VirtIODevice *vdev,
>  
>  static uint32_t virtio_balloon_get_features(VirtIODevice *vdev, uint32_t f)
>  {
> +    VirtIOBalloon *dev = VIRTIO_BALLOON(vdev);
>      f |= (1 << VIRTIO_BALLOON_F_STATS_VQ);
> +    if (dev->deflate_on_oom) {
> +        f |= (1 << VIRTIO_BALLOON_F_DEFLATE_ON_OOM);
> +    }
> +
>      return f;
>  }
>  
> @@ -409,6 +414,7 @@ static void virtio_balloon_device_unrealize(DeviceState *dev, Error **errp)
>  }
>  
>  static Property virtio_balloon_properties[] = {
> +    DEFINE_PROP_BOOL("deflate-on-oom", VirtIOBalloon, deflate_on_oom, false),
>      DEFINE_PROP_END_OF_LIST(),
>  };
>  
> diff --git a/include/hw/virtio/virtio-balloon.h b/include/hw/virtio/virtio-balloon.h
> index f863bfe..45cc55a 100644
> --- a/include/hw/virtio/virtio-balloon.h
> +++ b/include/hw/virtio/virtio-balloon.h
> @@ -30,6 +30,7 @@
>  /* The feature bitmap for virtio balloon */
>  #define VIRTIO_BALLOON_F_MUST_TELL_HOST 0 /* Tell before reclaiming pages */
>  #define VIRTIO_BALLOON_F_STATS_VQ 1       /* Memory stats virtqueue */
> +#define VIRTIO_BALLOON_F_DEFLATE_ON_OOM 2 /* Deflate balloon on OOM */
>  
>  /* Size of a PFN in the balloon interface. */
>  #define VIRTIO_BALLOON_PFN_SHIFT 12
> @@ -67,6 +68,7 @@ typedef struct VirtIOBalloon {
>      QEMUTimer *stats_timer;
>      int64_t stats_last_update;
>      int64_t stats_poll_interval;
> +    bool deflate_on_oom;
>  } VirtIOBalloon;
>  
>  #endif

You don't need an extra bool, and open-coding.
Do it same as we do for other features please,
set bit in feature mask directly.

> diff --git a/qemu-options.hx b/qemu-options.hx
> index da9851d..14ede0b 100644
> --- a/qemu-options.hx
> +++ b/qemu-options.hx
> @@ -324,7 +324,8 @@ ETEXI
>  DEF("balloon", HAS_ARG, QEMU_OPTION_balloon,
>      "-balloon none   disable balloon device\n"
>      "-balloon virtio[,addr=str]\n"
> -    "                enable virtio balloon device (default)\n", QEMU_ARCH_ALL)
> +    "                enable virtio balloon device (default)\n"
> +    "               [,deflate-on-oom=on|off]\n", QEMU_ARCH_ALL)
>  STEXI
>  @item -balloon none
>  @findex -balloon
> @@ -332,6 +333,9 @@ Disable balloon device.
>  @item -balloon virtio[,addr=@var{addr}]
>  Enable virtio balloon device (default), optionally with PCI address
>  @var{addr}.
> +@item -balloon virtio[,deflate-on-oom=@var{deflate-on-oom}]
> +@var{deflate-on-oom} is "on" or "off" and enables whether to let Guest OS
> +to deflate virtio balloon on OOM. Default is off.
>  ETEXI
>  
>  DEF("device", HAS_ARG, QEMU_OPTION_device,

Please don't add stuff to legacy -balloon.
New -device is enough, you don't need to touch qemu-options.hx
for it.

> -- 
> 1.9.1

Re: [Qemu-devel] [PATCH 1/2] balloon: call qdev_alias_all_properties for proxy dev in balloon class init

[Cornelia Huck]

On Wed, 26 Nov 2014 13:11:24 +0300
"Denis V. Lunev" <den@openvz.org> wrote:

> From: Raushaniya Maksudova <rmaksudova@parallels.com>
> 
> The idea is that all other virtio devices are calling this helper
> to merge properties of the proxy device. This is the only difference
> in between this helper and code in inside virtio_instance_init_common.
> The patch should not cause any harm as property list in generic balloon
> code is empty.
> 
> This also allows to avoid some dummy errors like fixed by this
>     commit 91ba21208839643603e7f7fa5864723c3f371ebe
>     Author: Gonglei <arei.gonglei@huawei.com>
>     Date:   Tue Sep 30 14:10:35 2014 +0800
>     virtio-balloon: fix virtio-balloon child refcount in transports
> 
> Signed-off-by: Denis V. Lunev <den@openvz.org>
> Acked-by: Raushaniya Maksudova <rmaksudova@parallels.com>
> CC: Cornelia Huck <cornelia.huck@de.ibm.com>
> CC: Christian Borntraeger <borntraeger@de.ibm.com>
> CC: Anthony Liguori <aliguori@amazon.com>
> CC: Michael S. Tsirkin <mst@redhat.com>
> ---
>  hw/s390x/virtio-ccw.c  | 5 ++---
>  hw/virtio/virtio-pci.c | 5 ++---
>  2 files changed, 4 insertions(+), 6 deletions(-)

Shouldn't this have the sign-off of the author (rather than the ack) as
well?

Otherwise, looks sane.

Re: [Qemu-devel] [PATCH 1/2] balloon: call qdev_alias_all_properties for proxy dev in balloon class init

[Denis V. Lunev]

On 26/11/14 15:27, Cornelia Huck wrote:
> On Wed, 26 Nov 2014 13:11:24 +0300
> "Denis V. Lunev" <den@openvz.org> wrote:
>
>> From: Raushaniya Maksudova <rmaksudova@parallels.com>
>>
>> The idea is that all other virtio devices are calling this helper
>> to merge properties of the proxy device. This is the only difference
>> in between this helper and code in inside virtio_instance_init_common.
>> The patch should not cause any harm as property list in generic balloon
>> code is empty.
>>
>> This also allows to avoid some dummy errors like fixed by this
>>      commit 91ba21208839643603e7f7fa5864723c3f371ebe
>>      Author: Gonglei <arei.gonglei@huawei.com>
>>      Date:   Tue Sep 30 14:10:35 2014 +0800
>>      virtio-balloon: fix virtio-balloon child refcount in transports
>>
>> Signed-off-by: Denis V. Lunev <den@openvz.org>
>> Acked-by: Raushaniya Maksudova <rmaksudova@parallels.com>
>> CC: Cornelia Huck <cornelia.huck@de.ibm.com>
>> CC: Christian Borntraeger <borntraeger@de.ibm.com>
>> CC: Anthony Liguori <aliguori@amazon.com>
>> CC: Michael S. Tsirkin <mst@redhat.com>
>> ---
>>   hw/s390x/virtio-ccw.c  | 5 ++---
>>   hw/virtio/virtio-pci.c | 5 ++---
>>   2 files changed, 4 insertions(+), 6 deletions(-)
> Shouldn't this have the sign-off of the author (rather than the ack) as
> well?
>
> Otherwise, looks sane.
>
I am an original author of the patch and prepared it to Rushana
to pass command line options for her next patch. I'll fix this
line in the next submission.

Re: [Qemu-devel] [PATCH 2/2] balloon: add a feature bit to let Guest OS deflate balloon on oom

[Denis V. Lunev]

On 26/11/14 14:16, Michael S. Tsirkin wrote:
> On Wed, Nov 26, 2014 at 01:11:25PM +0300, Denis V. Lunev wrote:
>> From: Raushaniya Maksudova <rmaksudova@parallels.com>
>>
>> Excessive virtio_balloon inflation can cause invocation of OOM-killer,
>> when Linux is under severe memory pressure. Various mechanisms are
>> responsible for correct virtio_balloon memory management. Nevertheless it
>> is often the case that these control tools does not have enough time to
>> react on fast changing memory load. As a result OS runs out of memory and
>> invokes OOM-killer. The balancing of memory by use of the virtio balloon
>> should not cause the termination of processes while there are pages in the
>> balloon. Now there is no way for virtio balloon driver to free memory at
>> the last moment before some process get killed by OOM-killer.
>>
>> This does not provide a security breach as balloon itself is running
>> inside Guest OS and is working in the cooperation with the host. Thus
>> some improvements from Guest side should be considered as normal.
>>
>> To solve the problem, introduce a virtio_balloon callback which is
>> expected to be called from the oom notifier call chain in out_of_memory()
>> function. If virtio balloon could release some memory, it will make the
>> system to return and retry the allocation that forced the out of memory
>> killer to run.
>>
>> This behavior should be enabled if and only if appropriate feature bit
>> is set on the device. It is off by default.
>>
>> This functionality was recently merged into vanilla Linux (actually in
>> linux-next at the moment)
>>
>>    commit 5a10b7dbf904bfe01bb9fcc6298f7df09eed77d5
>>    Author: Raushaniya Maksudova <rmaksudova@parallels.com>
>>    Date:   Mon Nov 10 09:36:29 2014 +1030
>>
>> This patch adds respective control bits into QEMU. It introduces
>> deflate-on-oom option for baloon device which do the trick.
>>
>> Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
>> Signed-off-by: Denis V. Lunev <den@openvz.org>
>> CC: Anthony Liguori <aliguori@amazon.com>
>> CC: Michael S. Tsirkin <mst@redhat.com>
>> ---
>>   hw/virtio/virtio-balloon.c         | 7 +++++++
>>   include/hw/virtio/virtio-balloon.h | 2 ++
>>   qemu-options.hx                    | 6 +++++-
>>   3 files changed, 14 insertions(+), 1 deletion(-)
>>
>> diff --git a/hw/virtio/virtio-balloon.c b/hw/virtio/virtio-balloon.c
>> index 7bfbb75..9d145fa 100644
>> --- a/hw/virtio/virtio-balloon.c
>> +++ b/hw/virtio/virtio-balloon.c
>> @@ -305,7 +305,12 @@ static void virtio_balloon_set_config(VirtIODevice *vdev,
>>   
>>   static uint32_t virtio_balloon_get_features(VirtIODevice *vdev, uint32_t f)
>>   {
>> +    VirtIOBalloon *dev = VIRTIO_BALLOON(vdev);
>>       f |= (1 << VIRTIO_BALLOON_F_STATS_VQ);
>> +    if (dev->deflate_on_oom) {
>> +        f |= (1 << VIRTIO_BALLOON_F_DEFLATE_ON_OOM);
>> +    }
>> +
>>       return f;
>>   }
>>   
>> @@ -409,6 +414,7 @@ static void virtio_balloon_device_unrealize(DeviceState *dev, Error **errp)
>>   }
>>   
>>   static Property virtio_balloon_properties[] = {
>> +    DEFINE_PROP_BOOL("deflate-on-oom", VirtIOBalloon, deflate_on_oom, false),
>>       DEFINE_PROP_END_OF_LIST(),
>>   };
>>   
>> diff --git a/include/hw/virtio/virtio-balloon.h b/include/hw/virtio/virtio-balloon.h
>> index f863bfe..45cc55a 100644
>> --- a/include/hw/virtio/virtio-balloon.h
>> +++ b/include/hw/virtio/virtio-balloon.h
>> @@ -30,6 +30,7 @@
>>   /* The feature bitmap for virtio balloon */
>>   #define VIRTIO_BALLOON_F_MUST_TELL_HOST 0 /* Tell before reclaiming pages */
>>   #define VIRTIO_BALLOON_F_STATS_VQ 1       /* Memory stats virtqueue */
>> +#define VIRTIO_BALLOON_F_DEFLATE_ON_OOM 2 /* Deflate balloon on OOM */
>>   
>>   /* Size of a PFN in the balloon interface. */
>>   #define VIRTIO_BALLOON_PFN_SHIFT 12
>> @@ -67,6 +68,7 @@ typedef struct VirtIOBalloon {
>>       QEMUTimer *stats_timer;
>>       int64_t stats_last_update;
>>       int64_t stats_poll_interval;
>> +    bool deflate_on_oom;
>>   } VirtIOBalloon;
>>   
>>   #endif
> You don't need an extra bool, and open-coding.
> Do it same as we do for other features please,
> set bit in feature mask directly.
there is no host_features placeholder at this level.

We could add it and propagate proper bit in get_features
callback like
    return f | dev->host_features
or we have to move this stuff a little bit up into
CCW & PCI code like done in virtscsi with
   #define DEFINE_VIRTIO_SCSI_FEATURES(_state, _feature_field)

The first approach keeps bits in the same place,
the second follows current approach. If you prefer unification way,
I think that other devices could be re-written this way.

Any opinion?

>> diff --git a/qemu-options.hx b/qemu-options.hx
>> index da9851d..14ede0b 100644
>> --- a/qemu-options.hx
>> +++ b/qemu-options.hx
>> @@ -324,7 +324,8 @@ ETEXI
>>   DEF("balloon", HAS_ARG, QEMU_OPTION_balloon,
>>       "-balloon none   disable balloon device\n"
>>       "-balloon virtio[,addr=str]\n"
>> -    "                enable virtio balloon device (default)\n", QEMU_ARCH_ALL)
>> +    "                enable virtio balloon device (default)\n"
>> +    "               [,deflate-on-oom=on|off]\n", QEMU_ARCH_ALL)
>>   STEXI
>>   @item -balloon none
>>   @findex -balloon
>> @@ -332,6 +333,9 @@ Disable balloon device.
>>   @item -balloon virtio[,addr=@var{addr}]
>>   Enable virtio balloon device (default), optionally with PCI address
>>   @var{addr}.
>> +@item -balloon virtio[,deflate-on-oom=@var{deflate-on-oom}]
>> +@var{deflate-on-oom} is "on" or "off" and enables whether to let Guest OS
>> +to deflate virtio balloon on OOM. Default is off.
>>   ETEXI
>>   
>>   DEF("device", HAS_ARG, QEMU_OPTION_device,
> Please don't add stuff to legacy -balloon.
> New -device is enough, you don't need to touch qemu-options.hx
> for it.
>
sure

Re: [Qemu-devel] [PATCH 2/2] balloon: add a feature bit to let Guest OS deflate balloon on oom

[Michael S. Tsirkin]

On Thu, Nov 27, 2014 at 02:04:55PM +0300, Denis V. Lunev wrote:
> On 26/11/14 14:16, Michael S. Tsirkin wrote:
> >On Wed, Nov 26, 2014 at 01:11:25PM +0300, Denis V. Lunev wrote:
> >>From: Raushaniya Maksudova <rmaksudova@parallels.com>
> >>
> >>Excessive virtio_balloon inflation can cause invocation of OOM-killer,
> >>when Linux is under severe memory pressure. Various mechanisms are
> >>responsible for correct virtio_balloon memory management. Nevertheless it
> >>is often the case that these control tools does not have enough time to
> >>react on fast changing memory load. As a result OS runs out of memory and
> >>invokes OOM-killer. The balancing of memory by use of the virtio balloon
> >>should not cause the termination of processes while there are pages in the
> >>balloon. Now there is no way for virtio balloon driver to free memory at
> >>the last moment before some process get killed by OOM-killer.
> >>
> >>This does not provide a security breach as balloon itself is running
> >>inside Guest OS and is working in the cooperation with the host. Thus
> >>some improvements from Guest side should be considered as normal.
> >>
> >>To solve the problem, introduce a virtio_balloon callback which is
> >>expected to be called from the oom notifier call chain in out_of_memory()
> >>function. If virtio balloon could release some memory, it will make the
> >>system to return and retry the allocation that forced the out of memory
> >>killer to run.
> >>
> >>This behavior should be enabled if and only if appropriate feature bit
> >>is set on the device. It is off by default.
> >>
> >>This functionality was recently merged into vanilla Linux (actually in
> >>linux-next at the moment)
> >>
> >>   commit 5a10b7dbf904bfe01bb9fcc6298f7df09eed77d5
> >>   Author: Raushaniya Maksudova <rmaksudova@parallels.com>
> >>   Date:   Mon Nov 10 09:36:29 2014 +1030
> >>
> >>This patch adds respective control bits into QEMU. It introduces
> >>deflate-on-oom option for baloon device which do the trick.
> >>
> >>Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
> >>Signed-off-by: Denis V. Lunev <den@openvz.org>
> >>CC: Anthony Liguori <aliguori@amazon.com>
> >>CC: Michael S. Tsirkin <mst@redhat.com>
> >>---
> >>  hw/virtio/virtio-balloon.c         | 7 +++++++
> >>  include/hw/virtio/virtio-balloon.h | 2 ++
> >>  qemu-options.hx                    | 6 +++++-
> >>  3 files changed, 14 insertions(+), 1 deletion(-)
> >>
> >>diff --git a/hw/virtio/virtio-balloon.c b/hw/virtio/virtio-balloon.c
> >>index 7bfbb75..9d145fa 100644
> >>--- a/hw/virtio/virtio-balloon.c
> >>+++ b/hw/virtio/virtio-balloon.c
> >>@@ -305,7 +305,12 @@ static void virtio_balloon_set_config(VirtIODevice *vdev,
> >>  static uint32_t virtio_balloon_get_features(VirtIODevice *vdev, uint32_t f)
> >>  {
> >>+    VirtIOBalloon *dev = VIRTIO_BALLOON(vdev);
> >>      f |= (1 << VIRTIO_BALLOON_F_STATS_VQ);
> >>+    if (dev->deflate_on_oom) {
> >>+        f |= (1 << VIRTIO_BALLOON_F_DEFLATE_ON_OOM);
> >>+    }
> >>+
> >>      return f;
> >>  }
> >>@@ -409,6 +414,7 @@ static void virtio_balloon_device_unrealize(DeviceState *dev, Error **errp)
> >>  }
> >>  static Property virtio_balloon_properties[] = {
> >>+    DEFINE_PROP_BOOL("deflate-on-oom", VirtIOBalloon, deflate_on_oom, false),
> >>      DEFINE_PROP_END_OF_LIST(),
> >>  };
> >>diff --git a/include/hw/virtio/virtio-balloon.h b/include/hw/virtio/virtio-balloon.h
> >>index f863bfe..45cc55a 100644
> >>--- a/include/hw/virtio/virtio-balloon.h
> >>+++ b/include/hw/virtio/virtio-balloon.h
> >>@@ -30,6 +30,7 @@
> >>  /* The feature bitmap for virtio balloon */
> >>  #define VIRTIO_BALLOON_F_MUST_TELL_HOST 0 /* Tell before reclaiming pages */
> >>  #define VIRTIO_BALLOON_F_STATS_VQ 1       /* Memory stats virtqueue */
> >>+#define VIRTIO_BALLOON_F_DEFLATE_ON_OOM 2 /* Deflate balloon on OOM */
> >>  /* Size of a PFN in the balloon interface. */
> >>  #define VIRTIO_BALLOON_PFN_SHIFT 12
> >>@@ -67,6 +68,7 @@ typedef struct VirtIOBalloon {
> >>      QEMUTimer *stats_timer;
> >>      int64_t stats_last_update;
> >>      int64_t stats_poll_interval;
> >>+    bool deflate_on_oom;
> >>  } VirtIOBalloon;
> >>  #endif
> >You don't need an extra bool, and open-coding.
> >Do it same as we do for other features please,
> >set bit in feature mask directly.
> there is no host_features placeholder at this level.
> 
> We could add it and propagate proper bit in get_features
> callback like
>    return f | dev->host_features
> or we have to move this stuff a little bit up into
> CCW & PCI code like done in virtscsi with
>   #define DEFINE_VIRTIO_SCSI_FEATURES(_state, _feature_field)
> 
> The first approach keeps bits in the same place,
> the second follows current approach. If you prefer unification way,
> I think that other devices could be re-written this way.
> 
> Any opinion?

Balloon is the weird one out here.
Look at how feature bits are defined for e.g.
DEFINE_PROP_BIT("any_layout", _state, _field, VIRTIO_F_ANY_LAYOUT, true),

you should do something similar.


> >>diff --git a/qemu-options.hx b/qemu-options.hx
> >>index da9851d..14ede0b 100644
> >>--- a/qemu-options.hx
> >>+++ b/qemu-options.hx
> >>@@ -324,7 +324,8 @@ ETEXI
> >>  DEF("balloon", HAS_ARG, QEMU_OPTION_balloon,
> >>      "-balloon none   disable balloon device\n"
> >>      "-balloon virtio[,addr=str]\n"
> >>-    "                enable virtio balloon device (default)\n", QEMU_ARCH_ALL)
> >>+    "                enable virtio balloon device (default)\n"
> >>+    "               [,deflate-on-oom=on|off]\n", QEMU_ARCH_ALL)
> >>  STEXI
> >>  @item -balloon none
> >>  @findex -balloon
> >>@@ -332,6 +333,9 @@ Disable balloon device.
> >>  @item -balloon virtio[,addr=@var{addr}]
> >>  Enable virtio balloon device (default), optionally with PCI address
> >>  @var{addr}.
> >>+@item -balloon virtio[,deflate-on-oom=@var{deflate-on-oom}]
> >>+@var{deflate-on-oom} is "on" or "off" and enables whether to let Guest OS
> >>+to deflate virtio balloon on OOM. Default is off.
> >>  ETEXI
> >>  DEF("device", HAS_ARG, QEMU_OPTION_device,
> >Please don't add stuff to legacy -balloon.
> >New -device is enough, you don't need to touch qemu-options.hx
> >for it.
> >
> sure

[Qemu-devel] [PATCH 1/2] balloon: call qdev_alias_all_properties for proxy dev in balloon class init

[Denis V. Lunev]

The idea is that all other virtio devices are calling this helper
to merge properties of the proxy device. This is the only difference
in between this helper and code in inside virtio_instance_init_common.
The patch should not cause any harm as property list in generic balloon
code is empty.

This also allows to avoid some dummy errors like fixed by this
    commit 91ba21208839643603e7f7fa5864723c3f371ebe
    Author: Gonglei <arei.gonglei@huawei.com>
    Date:   Tue Sep 30 14:10:35 2014 +0800
    virtio-balloon: fix virtio-balloon child refcount in transports

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
Revieved-by: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: Christian Borntraeger <borntraeger@de.ibm.com>
CC: Anthony Liguori <aliguori@amazon.com>
CC: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
---
 hw/s390x/virtio-ccw.c  | 5 ++---
 hw/virtio/virtio-pci.c | 5 ++---
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c
index 3fee4aa..ffbb9c2 100644
--- a/hw/s390x/virtio-ccw.c
+++ b/hw/s390x/virtio-ccw.c
@@ -898,9 +898,8 @@ static void balloon_ccw_stats_set_poll_interval(Object *obj, struct Visitor *v,
 static void virtio_ccw_balloon_instance_init(Object *obj)
 {
     VirtIOBalloonCcw *dev = VIRTIO_BALLOON_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_ccw_stats_get_all, NULL, NULL, dev, NULL);
 
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index 6dd41b9..e7baf7b 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -1274,9 +1274,8 @@ static void virtio_balloon_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_balloon_pci_instance_init(Object *obj)
 {
     VirtIOBalloonPCI *dev = VIRTIO_BALLOON_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_pci_stats_get_all, NULL, NULL, dev,
                         NULL);
-- 
1.9.1

[Qemu-devel] [PATCH 1/2] balloon: call qdev_alias_all_properties for proxy dev in balloon class init

[Denis V. Lunev]

The idea is that all other virtio devices are calling this helper
to merge properties of the proxy device. This is the only difference
in between this helper and code in inside virtio_instance_init_common.
The patch should not cause any harm as property list in generic balloon
code is empty.

This also allows to avoid some dummy errors like fixed by this
    commit 91ba21208839643603e7f7fa5864723c3f371ebe
    Author: Gonglei <arei.gonglei@huawei.com>
    Date:   Tue Sep 30 14:10:35 2014 +0800
    virtio-balloon: fix virtio-balloon child refcount in transports

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
Revieved-by: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: Christian Borntraeger <borntraeger@de.ibm.com>
CC: Anthony Liguori <aliguori@amazon.com>
CC: Michael S. Tsirkin <mst@redhat.com>
---
 hw/s390x/virtio-ccw.c  | 5 ++---
 hw/virtio/virtio-pci.c | 5 ++---
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c
index 3fee4aa..ffbb9c2 100644
--- a/hw/s390x/virtio-ccw.c
+++ b/hw/s390x/virtio-ccw.c
@@ -898,9 +898,8 @@ static void balloon_ccw_stats_set_poll_interval(Object *obj, struct Visitor *v,
 static void virtio_ccw_balloon_instance_init(Object *obj)
 {
     VirtIOBalloonCcw *dev = VIRTIO_BALLOON_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_ccw_stats_get_all, NULL, NULL, dev, NULL);
 
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index 6dd41b9..e7baf7b 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -1274,9 +1274,8 @@ static void virtio_balloon_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_balloon_pci_instance_init(Object *obj)
 {
     VirtIOBalloonPCI *dev = VIRTIO_BALLOON_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_pci_stats_get_all, NULL, NULL, dev,
                         NULL);
-- 
1.9.1

[Qemu-devel] [PATCH 1/2] balloon: call qdev_alias_all_properties for proxy dev in balloon class init

[Denis V. Lunev]

The idea is that all other virtio devices are calling this helper
to merge properties of the proxy device. This is the only difference
in between this helper and code in inside virtio_instance_init_common.
The patch should not cause any harm as property list in generic balloon
code is empty.

This also allows to avoid some dummy errors like fixed by this
    commit 91ba21208839643603e7f7fa5864723c3f371ebe
    Author: Gonglei <arei.gonglei@huawei.com>
    Date:   Tue Sep 30 14:10:35 2014 +0800
    virtio-balloon: fix virtio-balloon child refcount in transports

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
Revieved-by: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: Christian Borntraeger <borntraeger@de.ibm.com>
CC: Anthony Liguori <aliguori@amazon.com>
CC: Michael S. Tsirkin <mst@redhat.com>
---
 hw/s390x/virtio-ccw.c  | 5 ++---
 hw/virtio/virtio-pci.c | 5 ++---
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c
index ea236c9..82da894 100644
--- a/hw/s390x/virtio-ccw.c
+++ b/hw/s390x/virtio-ccw.c
@@ -899,9 +899,8 @@ static void balloon_ccw_stats_set_poll_interval(Object *obj, struct Visitor *v,
 static void virtio_ccw_balloon_instance_init(Object *obj)
 {
     VirtIOBalloonCcw *dev = VIRTIO_BALLOON_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_ccw_stats_get_all, NULL, NULL, dev, NULL);
 
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index dde1d73..745324b 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -1316,9 +1316,8 @@ static void virtio_balloon_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_balloon_pci_instance_init(Object *obj)
 {
     VirtIOBalloonPCI *dev = VIRTIO_BALLOON_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_pci_stats_get_all, NULL, NULL, dev,
                         NULL);
-- 
1.9.1

[Qemu-devel] [PATCH 1/2] balloon: call qdev_alias_all_properties for proxy dev in balloon class init

[Denis V. Lunev]

The idea is that all other virtio devices are calling this helper
to merge properties of the proxy device. This is the only difference
in between this helper and code in inside virtio_instance_init_common.
The patch should not cause any harm as property list in generic balloon
code is empty.

This also allows to avoid some dummy errors like fixed by this
    commit 91ba21208839643603e7f7fa5864723c3f371ebe
    Author: Gonglei <arei.gonglei@huawei.com>
    Date:   Tue Sep 30 14:10:35 2014 +0800
    virtio-balloon: fix virtio-balloon child refcount in transports

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Raushaniya Maksudova <rmaksudova@parallels.com>
Revieved-by: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: Christian Borntraeger <borntraeger@de.ibm.com>
CC: Anthony Liguori <aliguori@amazon.com>
CC: Michael S. Tsirkin <mst@redhat.com>
---
 hw/s390x/virtio-ccw.c  | 5 ++---
 hw/virtio/virtio-pci.c | 5 ++---
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c
index ea236c9..82da894 100644
--- a/hw/s390x/virtio-ccw.c
+++ b/hw/s390x/virtio-ccw.c
@@ -899,9 +899,8 @@ static void balloon_ccw_stats_set_poll_interval(Object *obj, struct Visitor *v,
 static void virtio_ccw_balloon_instance_init(Object *obj)
 {
     VirtIOBalloonCcw *dev = VIRTIO_BALLOON_CCW(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_ccw_stats_get_all, NULL, NULL, dev, NULL);
 
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index dde1d73..745324b 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -1316,9 +1316,8 @@ static void virtio_balloon_pci_class_init(ObjectClass *klass, void *data)
 static void virtio_balloon_pci_instance_init(Object *obj)
 {
     VirtIOBalloonPCI *dev = VIRTIO_BALLOON_PCI(obj);
-    object_initialize(&dev->vdev, sizeof(dev->vdev), TYPE_VIRTIO_BALLOON);
-    object_property_add_child(obj, "virtio-backend", OBJECT(&dev->vdev), NULL);
-    object_unref(OBJECT(&dev->vdev));
+    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+                                TYPE_VIRTIO_BALLOON);
     object_property_add(obj, "guest-stats", "guest statistics",
                         balloon_pci_stats_get_all, NULL, NULL, dev,
                         NULL);
-- 
1.9.1