[PATCH vhost 03/17] virtio_ring: packed: structure the indirect desc table

[Xuan Zhuo]

On Wed, 31 Jan 2024 17:12:10 +0800, Jason Wang <jasowang at redhat.com> wrote:
> On Tue, Jan 30, 2024 at 7:42 PM Xuan Zhuo <xuanzhuo at linux.alibaba.com> wrote:
> >
> > This commit structure the indirect desc table.
> > Then we can get the desc num directly when doing unmap.
> >
> > And save the dma info to the struct, then the indirect
> > will not use the dma fields of the desc_extra. The subsequent
> > commits will make the dma fields are optional. But for
> > the indirect case, we must record the dma info.
> >
> > Signed-off-by: Xuan Zhuo <xuanzhuo at linux.alibaba.com>
> > ---
> >  drivers/virtio/virtio_ring.c | 63 ++++++++++++++++++++----------------
> >  1 file changed, 35 insertions(+), 28 deletions(-)
> >
> > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
> > index 7280a1706cca..dd03bc5a81fe 100644
> > --- a/drivers/virtio/virtio_ring.c
> > +++ b/drivers/virtio/virtio_ring.c
> > @@ -72,9 +72,16 @@ struct vring_desc_state_split {
> >         struct vring_desc *indir_desc;  /* Indirect descriptor, if any. */
> >  };
> >
> > +struct vring_packed_desc_indir {
> > +       dma_addr_t addr;                /* Descriptor Array DMA addr. */
> > +       u32 len;                        /* Descriptor Array length. */
> > +       u32 num;
> > +       struct vring_packed_desc desc[];
> > +};
> > +
> >  struct vring_desc_state_packed {
> >         void *data;                     /* Data for callback. */
> > -       struct vring_packed_desc *indir_desc; /* Indirect descriptor, if any. */
> > +       struct vring_packed_desc_indir *indir_desc; /* Indirect descriptor, if any. */
> >         u16 num;                        /* Descriptor list length. */
> >         u16 last;                       /* The last desc state in a list. */
> >  };
> > @@ -1249,10 +1256,13 @@ static void vring_unmap_desc_packed(const struct vring_virtqueue *vq,
> >                        DMA_FROM_DEVICE : DMA_TO_DEVICE);
> >  }
> >
> > -static struct vring_packed_desc *alloc_indirect_packed(unsigned int total_sg,
> > +static struct vring_packed_desc_indir *alloc_indirect_packed(unsigned int total_sg,
> >                                                        gfp_t gfp)
> >  {
> > -       struct vring_packed_desc *desc;
> > +       struct vring_packed_desc_indir *in_desc;
> > +       u32 size;
> > +
> > +       size = struct_size(in_desc, desc, total_sg);
> >
> >         /*
> >          * We require lowmem mappings for the descriptors because
> > @@ -1261,9 +1271,10 @@ static struct vring_packed_desc *alloc_indirect_packed(unsigned int total_sg,
> >          */
> >         gfp &= ~__GFP_HIGHMEM;
> >
> > -       desc = kmalloc_array(total_sg, sizeof(struct vring_packed_desc), gfp);
> >
> > -       return desc;
> > +       in_desc = kmalloc(size, gfp);
> > +
> > +       return in_desc;
> >  }
> >
> >  static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
> > @@ -1274,6 +1285,7 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
> >                                          void *data,
> >                                          gfp_t gfp)
> >  {
> > +       struct vring_packed_desc_indir *in_desc;
> >         struct vring_packed_desc *desc;
> >         struct scatterlist *sg;
> >         unsigned int i, n, err_idx;
> > @@ -1281,10 +1293,12 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
> >         dma_addr_t addr;
> >
> >         head = vq->packed.next_avail_idx;
> > -       desc = alloc_indirect_packed(total_sg, gfp);
> > -       if (!desc)
> > +       in_desc = alloc_indirect_packed(total_sg, gfp);
> > +       if (!in_desc)
> >                 return -ENOMEM;
> >
> > +       desc = in_desc->desc;
> > +
> >         if (unlikely(vq->vq.num_free < 1)) {
> >                 pr_debug("Can't add buf len 1 - avail = 0\n");
> >                 kfree(desc);
> > @@ -1321,17 +1335,15 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
> >                 goto unmap_release;
> >         }
> >
> > +       in_desc->num = i;
> > +       in_desc->addr = addr;
> > +       in_desc->len = total_sg * sizeof(struct vring_packed_desc);
>
> It looks to me if we don't use dma_api we don't even need these steps?

YES


>
> > +
> >         vq->packed.vring.desc[head].addr = cpu_to_le64(addr);
> >         vq->packed.vring.desc[head].len = cpu_to_le32(total_sg *
> >                                 sizeof(struct vring_packed_desc));
> >         vq->packed.vring.desc[head].id = cpu_to_le16(id);
> >
> > -       if (vring_need_unmap_buffer(vq)) {
> > -               vq->packed.desc_extra[id].addr = addr;
> > -               vq->packed.desc_extra[id].len = total_sg *
> > -                               sizeof(struct vring_packed_desc);
> > -       }
> > -
> >         vq->packed.desc_extra[id].flags = VRING_DESC_F_INDIRECT |
> >                 vq->packed.avail_used_flags;
> >
> > @@ -1362,7 +1374,7 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
> >         /* Store token and indirect buffer state. */
> >         vq->packed.desc_state[id].num = 1;
> >         vq->packed.desc_state[id].data = data;
> > -       vq->packed.desc_state[id].indir_desc = desc;
> > +       vq->packed.desc_state[id].indir_desc = in_desc;
> >         vq->packed.desc_state[id].last = id;
> >
> >         vq->num_added += 1;
> > @@ -1381,7 +1393,7 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq,
> >                 vring_unmap_desc_packed(vq, &desc[i]);
> >
> >  free_desc:
> > -       kfree(desc);
> > +       kfree(in_desc);
> >
> >         END_USE(vq);
> >         return -ENOMEM;
> > @@ -1595,7 +1607,6 @@ static void detach_buf_packed(struct vring_virtqueue *vq,
> >                               unsigned int id, void **ctx)
> >  {
> >         struct vring_desc_state_packed *state = NULL;
> > -       struct vring_packed_desc *desc;
> >         unsigned int i, curr;
> >         u16 flags;
> >
> > @@ -1621,28 +1632,24 @@ static void detach_buf_packed(struct vring_virtqueue *vq,
> >
> >                 if (ctx)
> >                         *ctx = state->indir_desc;
> > +
>
> Unnecessary changes.


Could you say more?
You do not like this patch?

Thanks.



>
> Thanks
>