[Qemu-devel] [PATCH v2] virtio-rng: process pending requests when driver is ready

[Qemu-devel] [PATCH v2] virtio-rng: process pending requests when driver is ready

[Pankaj Gupta]

 virtio-rng device causing old guest kernels(2.6.32) to hang on latest qemu.
 The driver attempts to read from the virtio-rng device too early in it's
 initialization. Qemu detects guest is not ready and returns, resulting in
 hang.

 Fix is to handle pending request when guest is running and driver status is
 set to 'VIRTIO_CONFIG_S_DRIVER_OK'.

Reported-by: Sergio lopez <slopezpa@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Pankaj Gupta <pagupta@redhat.com>
---
v1->v2: Update new status value early only for virtio-rng (Stefan)

 hw/virtio/virtio-rng.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/hw/virtio/virtio-rng.c b/hw/virtio/virtio-rng.c
index 289bbcac03..855f1b41d1 100644
--- a/hw/virtio/virtio-rng.c
+++ b/hw/virtio/virtio-rng.c
@@ -156,6 +156,19 @@ static void check_rate_limit(void *opaque)
     vrng->activate_timer = true;
 }
 
+static void virtio_rng_set_status(VirtIODevice *vdev, uint8_t status)
+{
+    VirtIORNG *vrng = VIRTIO_RNG(vdev);
+
+    if (!vdev->vm_running) {
+        return;
+    }
+    vdev->status = status;
+
+    /* Something changed, try to process buffers */
+    virtio_rng_process(vrng);
+}
+
 static void virtio_rng_device_realize(DeviceState *dev, Error **errp)
 {
     VirtIODevice *vdev = VIRTIO_DEVICE(dev);
@@ -261,6 +274,7 @@ static void virtio_rng_class_init(ObjectClass *klass, void *data)
     vdc->realize = virtio_rng_device_realize;
     vdc->unrealize = virtio_rng_device_unrealize;
     vdc->get_features = get_features;
+    vdc->set_status = virtio_rng_set_status;
 }
 
 static const TypeInfo virtio_rng_info = {
-- 
2.14.3

Re: [Qemu-devel] [PATCH v2] virtio-rng: process pending requests when driver is ready

[Daniel P. Berrangé]

On Wed, Jun 27, 2018 at 04:55:20PM +0530, Pankaj Gupta wrote:
>  virtio-rng device causing old guest kernels(2.6.32) to hang on latest qemu.
>  The driver attempts to read from the virtio-rng device too early in it's
>  initialization. Qemu detects guest is not ready and returns, resulting in
>  hang.

Presumably this all worked previously at some point, so do you know what
change in QEMU has caused a regression in this regard.

> 
>  Fix is to handle pending request when guest is running and driver status is
>  set to 'VIRTIO_CONFIG_S_DRIVER_OK'.
> 
> Reported-by: Sergio lopez <slopezpa@redhat.com>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> Signed-off-by: Pankaj Gupta <pagupta@redhat.com>
> ---
> v1->v2: Update new status value early only for virtio-rng (Stefan)
> 
>  hw/virtio/virtio-rng.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/hw/virtio/virtio-rng.c b/hw/virtio/virtio-rng.c
> index 289bbcac03..855f1b41d1 100644
> --- a/hw/virtio/virtio-rng.c
> +++ b/hw/virtio/virtio-rng.c
> @@ -156,6 +156,19 @@ static void check_rate_limit(void *opaque)
>      vrng->activate_timer = true;
>  }
>  
> +static void virtio_rng_set_status(VirtIODevice *vdev, uint8_t status)
> +{
> +    VirtIORNG *vrng = VIRTIO_RNG(vdev);
> +
> +    if (!vdev->vm_running) {
> +        return;
> +    }
> +    vdev->status = status;
> +
> +    /* Something changed, try to process buffers */
> +    virtio_rng_process(vrng);
> +}
> +
>  static void virtio_rng_device_realize(DeviceState *dev, Error **errp)
>  {
>      VirtIODevice *vdev = VIRTIO_DEVICE(dev);
> @@ -261,6 +274,7 @@ static void virtio_rng_class_init(ObjectClass *klass, void *data)
>      vdc->realize = virtio_rng_device_realize;
>      vdc->unrealize = virtio_rng_device_unrealize;
>      vdc->get_features = get_features;
> +    vdc->set_status = virtio_rng_set_status;
>  }
>  
>  static const TypeInfo virtio_rng_info = {
> -- 
> 2.14.3
> 
> 

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|

Re: [Qemu-devel] [PATCH v2] virtio-rng: process pending requests when driver is ready

[Pankaj Gupta]

> 
> On Wed, Jun 27, 2018 at 04:55:20PM +0530, Pankaj Gupta wrote:
> >  virtio-rng device causing old guest kernels(2.6.32) to hang on latest
> >  qemu.
> >  The driver attempts to read from the virtio-rng device too early in it's
> >  initialization. Qemu detects guest is not ready and returns, resulting in
> >  hang.
> 
> Presumably this all worked previously at some point, so do you know what
> change in QEMU has caused a regression in this regard.

Latest qemu(2.x) has this issue with older guest kernel(2.6.32) versions.
But it does work properly with upstream guest kernel because the way virtio-rng
driver is reworked, it sends request to virtio-rng driver when(before) there is 
read call. 

If guest driver is ready when it notifies qemu this works fine. There are number
of changes in guest virtio/driver code infrastructure as well as qemu side, I don't 
know exact commit after which this stopped working for older guest kernels. 
 
Thanks,
Pankaj

> 
> > 
> >  Fix is to handle pending request when guest is running and driver status
> >  is
> >  set to 'VIRTIO_CONFIG_S_DRIVER_OK'.
> > 
> > Reported-by: Sergio lopez <slopezpa@redhat.com>
> > Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> > Signed-off-by: Pankaj Gupta <pagupta@redhat.com>
> > ---
> > v1->v2: Update new status value early only for virtio-rng (Stefan)
> > 
> >  hw/virtio/virtio-rng.c | 14 ++++++++++++++
> >  1 file changed, 14 insertions(+)
> > 
> > diff --git a/hw/virtio/virtio-rng.c b/hw/virtio/virtio-rng.c
> > index 289bbcac03..855f1b41d1 100644
> > --- a/hw/virtio/virtio-rng.c
> > +++ b/hw/virtio/virtio-rng.c
> > @@ -156,6 +156,19 @@ static void check_rate_limit(void *opaque)
> >      vrng->activate_timer = true;
> >  }
> >  
> > +static void virtio_rng_set_status(VirtIODevice *vdev, uint8_t status)
> > +{
> > +    VirtIORNG *vrng = VIRTIO_RNG(vdev);
> > +
> > +    if (!vdev->vm_running) {
> > +        return;
> > +    }
> > +    vdev->status = status;
> > +
> > +    /* Something changed, try to process buffers */
> > +    virtio_rng_process(vrng);
> > +}
> > +
> >  static void virtio_rng_device_realize(DeviceState *dev, Error **errp)
> >  {
> >      VirtIODevice *vdev = VIRTIO_DEVICE(dev);
> > @@ -261,6 +274,7 @@ static void virtio_rng_class_init(ObjectClass *klass,
> > void *data)
> >      vdc->realize = virtio_rng_device_realize;
> >      vdc->unrealize = virtio_rng_device_unrealize;
> >      vdc->get_features = get_features;
> > +    vdc->set_status = virtio_rng_set_status;
> >  }
> >  
> >  static const TypeInfo virtio_rng_info = {
> > --
> > 2.14.3
> > 
> > 
> 
> Regards,
> Daniel
> --
> |: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange
> |:|
> |: https://libvirt.org         -o-            https://fstop138.berrange.com
> |:|
> |: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange
> |:|
> 
> 

Re: [Qemu-devel] [PATCH v2] virtio-rng: process pending requests when driver is ready

[Stefan Hajnoczi]

[-- Attachment #1: Type: text/plain, Size: 395 bytes --]

On Wed, Jun 27, 2018 at 04:55:20PM +0530, Pankaj Gupta wrote:
> +static void virtio_rng_set_status(VirtIODevice *vdev, uint8_t status)
> +{
> +    VirtIORNG *vrng = VIRTIO_RNG(vdev);
> +
> +    if (!vdev->vm_running) {
> +        return;
> +    }
> +    vdev->status = status;

Please add a comment to explain this, otherwise someone might think this
is unnecessary since virtio.c also does it.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 455 bytes --]

Re: [Qemu-devel] [PATCH v2] virtio-rng: process pending requests when driver is ready

[Pankaj Gupta]

> 
> On Wed, Jun 27, 2018 at 04:55:20PM +0530, Pankaj Gupta wrote:
> > +static void virtio_rng_set_status(VirtIODevice *vdev, uint8_t status)
> > +{
> > +    VirtIORNG *vrng = VIRTIO_RNG(vdev);
> > +
> > +    if (!vdev->vm_running) {
> > +        return;
> > +    }
> > +    vdev->status = status;
> 
> Please add a comment to explain this, otherwise someone might think this
> is unnecessary since virtio.c also does it.

Sure. Will add a comment and send v3.

Thanks,
Pankaj