在 2022/3/24 下午4:44, Xuan Zhuo 写道:
On Tue, 22 Mar 2022 14:30:29 +0800, Jason Wang <jasowang@xxxxxxxxxx> wrote:
在 2022/3/14 下午5:34, Xuan Zhuo 写道:
virtio ring split supports resize.
Only after the new vring is successfully allocated based on the new num,
we will release the old vring. In any case, an error is returned,
indicating that the vring still points to the old vring. In the case of
an error, we will re-initialize the state of the vring to ensure that
the vring can be used.
In addition, vring_align, may_reduce_num are necessary for reallocating
vring, so they are retained for creating vq.
Signed-off-by: Xuan Zhuo <xuanzhuo@xxxxxxxxxxxxxxxxx>
---
drivers/virtio/virtio_ring.c | 62 ++++++++++++++++++++++++++++++++++++
1 file changed, 62 insertions(+)
diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
index 81bbfd65411e..a15869514146 100644
--- a/drivers/virtio/virtio_ring.c
+++ b/drivers/virtio/virtio_ring.c
@@ -139,6 +139,12 @@ struct vring_virtqueue {
/* DMA address and size information */
dma_addr_t queue_dma_addr;
size_t queue_size_in_bytes;
+
+ /* The parameters for creating vrings are reserved for
+ * creating new vrings when enabling reset queue.
+ */
+ u32 vring_align;
+ bool may_reduce_num;
} split;
/* Available for packed ring */
@@ -198,6 +204,16 @@ struct vring_virtqueue {
#endif
};
+static void __vring_free(struct virtqueue *_vq);
+static void __vring_virtqueue_init_split(struct vring_virtqueue *vq,
+ struct virtio_device *vdev);
+static void __vring_virtqueue_attach_split(struct vring_virtqueue *vq,
+ struct vring vring,
+ struct vring_desc_state_split *desc_state,
+ struct vring_desc_extra *desc_extra);
+static int __vring_alloc_state_extra_split(u32 num,
+ struct vring_desc_state_split **desc_state,
+ struct vring_desc_extra **desc_extra);
/*
* Helpers.
@@ -991,6 +1007,8 @@ static struct virtqueue *vring_create_virtqueue_split(
return NULL;
}
+ to_vvq(vq)->split.vring_align = vring_align;
+ to_vvq(vq)->split.may_reduce_num = may_reduce_num;
to_vvq(vq)->split.queue_dma_addr = dma_addr;
to_vvq(vq)->split.queue_size_in_bytes = queue_size_in_bytes;
to_vvq(vq)->we_own_ring = true;
@@ -998,6 +1016,50 @@ static struct virtqueue *vring_create_virtqueue_split(
return vq;
}
+static int virtqueue_resize_split(struct virtqueue *_vq, u32 num)
+{
+ struct vring_virtqueue *vq = to_vvq(_vq);
+ struct virtio_device *vdev = _vq->vdev;
+ struct vring_desc_state_split *state;
+ struct vring_desc_extra *extra;
+ size_t queue_size_in_bytes;
+ dma_addr_t dma_addr;
+ struct vring vring;
+ int err = -ENOMEM;
+ void *queue;
+
+ BUG_ON(!vq->we_own_ring);
I don't see any checks in virtqueue_resize(). So I think it's better to
either
1) return -EINVAL here
or
2) add a check in virtqueue_resize and fail there
+
+ queue = vring_alloc_queue_split(vdev, &dma_addr, &num,
+ vq->split.vring_align,
+ vq->weak_barriers,
+ vq->split.may_reduce_num);
+ if (!queue)
+ goto init;
+
+ queue_size_in_bytes = vring_size(num, vq->split.vring_align);
+
+ err = __vring_alloc_state_extra_split(num, &state, &extra);
+ if (err) {
+ vring_free_queue(vdev, queue_size_in_bytes, queue, dma_addr);
+ goto init;
+ }
+
+ __vring_free(&vq->vq);
+
+ vring_init(&vring, num, queue, vq->split.vring_align);
+ __vring_virtqueue_attach_split(vq, vring, state, extra);
I wonder if we need a symmetric virtqueue_resize_detach() internal helper.
I think __vring_free() is somewhat similar to what you said about
virtqueue_resize_detach() .
So from what I'm understanding the code, the key point for attaching is:
vq->split.vring = vring;
But this is not what __vring_free() did, it just free the resources.
+ vq->split.queue_dma_addr = dma_addr;
+ vq->split.queue_size_in_bytes = queue_size_in_bytes;
+
+ err = 0;
+
+init:
+ __vring_virtqueue_init_split(vq, vdev);
+ vq->we_own_ring = true;
Then we can leave this unchanged.
I think you mean "vq->we_own_ring = true";
The reason for modifying we_own_ring alone is that in the general process of
creating a ring, __vring_virtqueue_init_split is called in
__vring_new_virtqueue. At this time, we_own_ring is false.
vring_create_virtqueue_split will change it to true. So after calling
__vring_virtqueue_init_split alone, we_own_ring is false.
I think it's possible to wrap __vring_virtqueue_init_split() again
static void vring_virtqueue_init_split(struct vring_virtqueue *vq,
struct virtio_device *vdev)
{
__vring_virtqueue_init_split(vq, vdev);
vq->we_own_ring = true;
}
Is this what you want?
Nope, I meant there are some transports that allocate the vring by
themselves, we can't resize those vring.
See the callers of vring_new_virtqueue()
Thanks
Thanks.
Thanks
+ return err;
+}
+
/*
* Packed ring specific functions - *_packed().
_______________________________________________
Virtualization mailing list
Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linuxfoundation.org/mailman/listinfo/virtualization
