Re: [RFC v2 5/7] virtio: unmask F_NEXT flag in desc_extra

From: Guo Zhi
Date: Wed Aug 17 2022 - 23:15:15 EST




----- Original Message -----
> From: "Xuan Zhuo" <xuanzhuo@xxxxxxxxxxxxxxxxx>
> To: "Guo Zhi" <qtxuning1999@xxxxxxxxxxx>
> Cc: "netdev" <netdev@xxxxxxxxxxxxxxx>, "linux-kernel" <linux-kernel@xxxxxxxxxxxxxxx>, "kvm list" <kvm@xxxxxxxxxxxxxxx>,
> "virtualization" <virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx>, "Guo Zhi" <qtxuning1999@xxxxxxxxxxx>, "eperezma"
> <eperezma@xxxxxxxxxx>, "jasowang" <jasowang@xxxxxxxxxx>, "sgarzare" <sgarzare@xxxxxxxxxx>, "Michael Tsirkin"
> <mst@xxxxxxxxxx>
> Sent: Thursday, August 18, 2022 11:05:37 AM
> Subject: Re: [RFC v2 5/7] virtio: unmask F_NEXT flag in desc_extra

> On Wed, 17 Aug 2022 21:57:16 +0800, Guo Zhi <qtxuning1999@xxxxxxxxxxx> wrote:
>> We didn't unmask F_NEXT flag in desc_extra in the end of a chain,
>> unmask it so that we can access desc_extra to get next information.
>
> I think we should state the purpose of this.
>
I have to unmask F_NEXT flag in desc_extra in this series, because if in order
negotiated, the driver has to iterate the descriptor chain to get chain length
from desc_extra. (The reason why we should use desc_extra is that descs may be
changed by malicious hypervisors, https://lkml.org/lkml/2022/7/26/224).

>>
>> Signed-off-by: Guo Zhi <qtxuning1999@xxxxxxxxxxx>
>> ---
>> drivers/virtio/virtio_ring.c | 6 ++++--
>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
>> index a5ec724c01d8..1c1b3fa376a2 100644
>> --- a/drivers/virtio/virtio_ring.c
>> +++ b/drivers/virtio/virtio_ring.c
>> @@ -567,7 +567,7 @@ static inline int virtqueue_add_split(struct virtqueue *_vq,
>> }
>> /* Last one doesn't continue. */
>> desc[prev].flags &= cpu_to_virtio16(_vq->vdev, ~VRING_DESC_F_NEXT);
>> - if (!indirect && vq->use_dma_api)
>> + if (!indirect)
>> vq->split.desc_extra[prev & (vq->split.vring.num - 1)].flags &=
>> ~VRING_DESC_F_NEXT;
>>
>> @@ -584,6 +584,8 @@ static inline int virtqueue_add_split(struct virtqueue *_vq,
>> total_sg * sizeof(struct vring_desc),
>> VRING_DESC_F_INDIRECT,
>> false);
>> + vq->split.desc_extra[head & (vq->split.vring.num - 1)].flags &=
>> + ~VRING_DESC_F_NEXT;
>
> This seems unnecessary.
>
>> }
>>
>> /* We're using some buffers from the free list. */
>> @@ -693,7 +695,7 @@ static void detach_buf_split(struct vring_virtqueue *vq,
>> unsigned int head,
>> /* Put back on free list: unmap first-level descriptors and find end */
>> i = head;
>>
>> - while (vq->split.vring.desc[i].flags & nextflag) {
>> + while (vq->split.desc_extra[i].flags & nextflag) {
>
> nextflag is __virtio16
>
> You can use VRING_DESC_F_NEXT directly.
>
> Thanks.

Sorry for the mistake, I will fix it.

>
>> vring_unmap_one_split(vq, i);
>> i = vq->split.desc_extra[i].next;
>> vq->vq.num_free++;
>> --
>> 2.17.1
>>