[PATCH kvmtool] virtio: Ensure virt_queue is always initialised

[PATCH kvmtool] virtio: Ensure virt_queue is always initialised

[Will Deacon]

Failing to initialise the virt_queue via virtio_init_device_vq() leaves,
amongst other things, the endianness unspecified. On arm/arm64 this
results in virtio_guest_to_host_uxx() treating the queue as big-endian
and trying to translate bogus addresses:

  Warning: unable to translate guest address 0x80b8249800000000 to host

Ensure the virt_queue is always initialised by the virtio device during
setup.

Cc: Marc Zyngier <maz@kernel.org>
Cc: Julien Thierry <julien.thierry.kdev@gmail.com>
Cc: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Will Deacon <will@kernel.org>
---
 virtio/balloon.c | 1 +
 virtio/rng.c     | 1 +
 virtio/scsi.c    | 1 +
 3 files changed, 3 insertions(+)

diff --git a/virtio/balloon.c b/virtio/balloon.c
index 15a9a46e77e0..0bd16703dfee 100644
--- a/virtio/balloon.c
+++ b/virtio/balloon.c
@@ -212,6 +212,7 @@ static int init_vq(struct kvm *kvm, void *dev, u32 vq, u32 page_size, u32 align,
 
 	thread_pool__init_job(&bdev->jobs[vq], kvm, virtio_bln_do_io, queue);
 	vring_init(&queue->vring, VIRTIO_BLN_QUEUE_SIZE, p, align);
+	virtio_init_device_vq(&bdev->vdev, queue);
 
 	return 0;
 }
diff --git a/virtio/rng.c b/virtio/rng.c
index 9dd757b7e6e9..78eaa64bda17 100644
--- a/virtio/rng.c
+++ b/virtio/rng.c
@@ -103,6 +103,7 @@ static int init_vq(struct kvm *kvm, void *dev, u32 vq, u32 page_size, u32 align,
 	job = &rdev->jobs[vq];
 
 	vring_init(&queue->vring, VIRTIO_RNG_QUEUE_SIZE, p, align);
+	virtio_init_device_vq(&rdev->vdev, queue);
 
 	*job = (struct rng_dev_job) {
 		.vq	= queue,
diff --git a/virtio/scsi.c b/virtio/scsi.c
index a72bb2a9a206..1ec78fe0945a 100644
--- a/virtio/scsi.c
+++ b/virtio/scsi.c
@@ -72,6 +72,7 @@ static int init_vq(struct kvm *kvm, void *dev, u32 vq, u32 page_size, u32 align,
 	p		= virtio_get_vq(kvm, queue->pfn, page_size);
 
 	vring_init(&queue->vring, VIRTIO_SCSI_QUEUE_SIZE, p, align);
+	virtio_init_device_vq(&sdev->vdev, queue);
 
 	if (sdev->vhost_fd == 0)
 		return 0;
-- 
2.23.0.700.g56cf767bdb-goog

Re: [PATCH kvmtool] virtio: Ensure virt_queue is always initialised

[Andre Przywara]

On Thu, 10 Oct 2019 15:28:52 +0100
Will Deacon <will@kernel.org> wrote:

Hi Will,

> Failing to initialise the virt_queue via virtio_init_device_vq() leaves,
> amongst other things, the endianness unspecified. On arm/arm64 this
> results in virtio_guest_to_host_uxx() treating the queue as big-endian
> and trying to translate bogus addresses:
> 
>   Warning: unable to translate guest address 0x80b8249800000000 to host

Ouch, a user! ;-)

> Ensure the virt_queue is always initialised by the virtio device during
> setup.

Indeed, this is also what the other virtio devices do.
Confirmed to fix rng and balloon.

Thanks for spotting this!

Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Tested-by: Andre Przywara <andre.przywara@arm.com>

Cheers,
Andre.

> 
> Cc: Marc Zyngier <maz@kernel.org>
> Cc: Julien Thierry <julien.thierry.kdev@gmail.com>
> Cc: Andre Przywara <andre.przywara@arm.com>
> Signed-off-by: Will Deacon <will@kernel.org>
> ---
>  virtio/balloon.c | 1 +
>  virtio/rng.c     | 1 +
>  virtio/scsi.c    | 1 +
>  3 files changed, 3 insertions(+)
> 
> diff --git a/virtio/balloon.c b/virtio/balloon.c
> index 15a9a46e77e0..0bd16703dfee 100644
> --- a/virtio/balloon.c
> +++ b/virtio/balloon.c
> @@ -212,6 +212,7 @@ static int init_vq(struct kvm *kvm, void *dev, u32 vq, u32 page_size, u32 align,
>  
>  	thread_pool__init_job(&bdev->jobs[vq], kvm, virtio_bln_do_io, queue);
>  	vring_init(&queue->vring, VIRTIO_BLN_QUEUE_SIZE, p, align);
> +	virtio_init_device_vq(&bdev->vdev, queue);
>  
>  	return 0;
>  }
> diff --git a/virtio/rng.c b/virtio/rng.c
> index 9dd757b7e6e9..78eaa64bda17 100644
> --- a/virtio/rng.c
> +++ b/virtio/rng.c
> @@ -103,6 +103,7 @@ static int init_vq(struct kvm *kvm, void *dev, u32 vq, u32 page_size, u32 align,
>  	job = &rdev->jobs[vq];
>  
>  	vring_init(&queue->vring, VIRTIO_RNG_QUEUE_SIZE, p, align);
> +	virtio_init_device_vq(&rdev->vdev, queue);
>  
>  	*job = (struct rng_dev_job) {
>  		.vq	= queue,
> diff --git a/virtio/scsi.c b/virtio/scsi.c
> index a72bb2a9a206..1ec78fe0945a 100644
> --- a/virtio/scsi.c
> +++ b/virtio/scsi.c
> @@ -72,6 +72,7 @@ static int init_vq(struct kvm *kvm, void *dev, u32 vq, u32 page_size, u32 align,
>  	p		= virtio_get_vq(kvm, queue->pfn, page_size);
>  
>  	vring_init(&queue->vring, VIRTIO_SCSI_QUEUE_SIZE, p, align);
> +	virtio_init_device_vq(&sdev->vdev, queue);
>  
>  	if (sdev->vhost_fd == 0)
>  		return 0;

Re: [PATCH kvmtool] virtio: Ensure virt_queue is always initialised

[Will Deacon]

On Fri, Oct 25, 2019 at 11:41:00AM +0100, Andre Przywara wrote:
> On Thu, 10 Oct 2019 15:28:52 +0100
> Will Deacon <will@kernel.org> wrote:
> > Failing to initialise the virt_queue via virtio_init_device_vq() leaves,
> > amongst other things, the endianness unspecified. On arm/arm64 this
> > results in virtio_guest_to_host_uxx() treating the queue as big-endian
> > and trying to translate bogus addresses:
> > 
> >   Warning: unable to translate guest address 0x80b8249800000000 to host
> 
> Ouch, a user! ;-)
> 
> > Ensure the virt_queue is always initialised by the virtio device during
> > setup.
> 
> Indeed, this is also what the other virtio devices do.
> Confirmed to fix rng and balloon.
> 
> Thanks for spotting this!
> 
> Reviewed-by: Andre Przywara <andre.przywara@arm.com>
> Tested-by: Andre Przywara <andre.przywara@arm.com>

Cheers, Andre. Now pushed with your tags.

Will