On 16.02.2023 17:18, Stefano Garzarella wrote:

On Mon, Feb 06, 2023 at 06:58:24AM +0000, Arseniy Krasnov wrote:

quoted

Use pages of non-linear skb as buffers in virtio tx queue.

Signed-off-by: Arseniy Krasnov <redacted>
---
net/vmw_vsock/virtio_transport.c | 31 +++++++++++++++++++++++++------
1 file changed, 25 insertions(+), 6 deletions(-)

diff --git a/net/vmw_vsock/virtio_transport.c b/net/vmw_vsock/virtio_transport.c
index 28b5a8e8e094..b8a7d6dc9f46 100644
--- a/net/vmw_vsock/virtio_transport.c
+++ b/net/vmw_vsock/virtio_transport.c

@@ -100,7 +100,8 @@ virtio_transport_send_pkt_work(struct work_struct *work)

    vq = vsock->vqs[VSOCK_VQ_TX];

    for (;;) {
-        struct scatterlist hdr, buf, *sgs[2];
+        struct scatterlist *sgs[MAX_SKB_FRAGS + 1];
+        struct scatterlist bufs[MAX_SKB_FRAGS + 1];

+ 1 is for the header, right?
I'd add a comment just to be clear ;-)

quoted

        int ret, in_sg = 0, out_sg = 0;
        struct sk_buff *skb;
        bool reply;

@@ -111,12 +112,30 @@ virtio_transport_send_pkt_work(struct work_struct *work)

        virtio_transport_deliver_tap_pkt(skb);
        reply = virtio_vsock_skb_reply(skb);
+        sg_init_one(&bufs[0], virtio_vsock_hdr(skb), sizeof(*virtio_vsock_hdr(skb)));
+        sgs[out_sg++] = &bufs[0];
+
+        if (skb_is_nonlinear(skb)) {
+            int i;
+
+            for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
+                struct page *data_page = skb_shinfo(skb)->frags[i].bv_page;
+
+                /* We will use 'page_to_virt()' for userspace page here,
+                 * because virtio layer will call 'virt_to_phys()' later
+                 * to fill buffer descriptor. We don't touch memory at
+                 * "virtual" address of this page.
+                 */

IIUC data_page is a user page, so since we are exposing it to the host,
I think we should pin it.

Is data_page always a user page, or can it be a kernel page when skb is nonlinear?

By default it is user page, but...may be it is possible to send page of mapped
file with MAP_SHARED flags(i think it this case it will be page from page cache)

Thanks,
Stefano

quoted

+                sg_init_one(&bufs[i + 1],
+                        page_to_virt(data_page), PAGE_SIZE);
+                sgs[out_sg++] = &bufs[i + 1];
+            }
+        } else {
+            if (skb->len > 0) {
+                sg_init_one(&bufs[1], skb->data, skb->len);
+                sgs[out_sg++] = &bufs[1];
+            }

-        sg_init_one(&hdr, virtio_vsock_hdr(skb), sizeof(*virtio_vsock_hdr(skb)));
-        sgs[out_sg++] = &hdr;
-        if (skb->len > 0) {
-            sg_init_one(&buf, skb->data, skb->len);
-            sgs[out_sg++] = &buf;
        }

        ret = virtqueue_add_sgs(vq, sgs, out_sg, in_sg, skb, GFP_KERNEL);
-- 
2.25.1