On Mon, Mar 09, 2026 at 10:47:36AM +0100, Laurent Vivier wrote:
udp_vu_sock_recv() currently mixes two concerns: receiving data from the socket and managing virtqueue buffers (collecting, rewinding, releasing). This makes the function harder to reason about and couples socket I/O with virtqueue state.
Move all virtqueue operations, vu_collect(), vu_init_elem(), vu_queue_rewind(), vu_set_vnethdr(), and the queue-readiness check, into udp_vu_sock_to_tap(), which is the only caller. This turns udp_vu_sock_recv() into a pure socket receive function that simply reads into the provided iov array and adjusts its length.
Signed-off-by: Laurent Vivier
LGTM, although one note below.
--- udp_vu.c | 87 ++++++++++++++++++++++++++------------------------------ 1 file changed, 41 insertions(+), 46 deletions(-)
diff --git a/udp_vu.c b/udp_vu.c index a39254776099..27ae93de4420 100644 --- a/udp_vu.c +++ b/udp_vu.c @@ -58,73 +58,40 @@ static size_t udp_vu_hdrlen(bool v6)
/** * udp_vu_sock_recv() - Receive datagrams from socket into vhost-user buffers - * @c: Execution context * @iov: IO vector for the frame (modified on output) * @cnt: Number of IO vector entries (in/out) - * @vq: virtqueue to use to receive data * @s: Socket to receive from * @v6: Set for IPv6 connections * - * Return: size of received data, 0 if the datagram - * was discarded because the virtqueue is not ready, -1 on error + * Return: size of received data, -1 on error */ -static ssize_t udp_vu_sock_recv(const struct ctx *c, struct iovec *iov, - size_t *cnt, struct vu_virtq *vq, int s, - bool v6) +static ssize_t udp_vu_sock_recv(struct iovec *iov, size_t *cnt, int s, bool v6) { - const struct vu_dev *vdev = c->vdev; - struct msghdr msg = { 0 }; + struct iovec msg_iov[*cnt]; + struct msghdr msg = { 0 }; struct iov_tail payload; size_t hdrlen; ssize_t dlen; - int iov_cnt; - - ASSERT(!c->no_udp); - - if (!vu_queue_enabled(vq) || !vu_queue_started(vq)) { - debug("Got UDP packet, but RX virtqueue not usable yet"); - - if (recvmsg(s, &msg, MSG_DONTWAIT) < 0) - debug_perror("Failed to discard datagram"); - - *cnt = 0; - return 0; - }
/* compute L2 header length */ hdrlen = udp_vu_hdrlen(v6);
- vu_init_elem(elem, iov, *cnt); - - iov_cnt = vu_collect(vdev, vq, elem, ARRAY_SIZE(elem), - IP_MAX_MTU + ETH_HLEN + VNET_HLEN, NULL); - if (iov_cnt == 0) - return -1; - - payload = IOV_TAIL(iov, iov_cnt, hdrlen); + payload = IOV_TAIL(iov, *cnt, hdrlen);
- struct iovec msg_iov[payload.cnt]; msg.msg_iov = msg_iov; msg.msg_iovlen = iov_tail_clone(msg.msg_iov, payload.cnt, &payload);
/* read data from the socket */ dlen = recvmsg(s, &msg, 0); - if (dlen < 0) { - vu_queue_rewind(vq, iov_cnt); + if (dlen < 0) return -1; - }
/* Pad short frames to ETH_ZLEN */ if (ETH_ZLEN + VNET_HLEN > dlen + hdrlen) { - iov_memset(iov, iov_cnt, dlen + hdrlen, 0, + iov_memset(iov, *cnt, dlen + hdrlen, 0, ETH_ZLEN + VNET_HLEN - (dlen + hdrlen)); } - *cnt = iov_truncate(iov, iov_cnt, dlen + hdrlen); - - vu_set_vnethdr(iov[0].iov_base, *cnt); - - /* release unused buffers */ - vu_queue_rewind(vq, iov_cnt - *cnt); + *cnt = iov_truncate(iov, *cnt, dlen + hdrlen);
Padding the frame, then truncating that padding seems kind of odd.
return dlen; } @@ -212,20 +179,48 @@ void udp_vu_sock_to_tap(const struct ctx *c, int s, int n, flow_sidx_t tosidx) bool v6 = !(inany_v4(&toside->eaddr) && inany_v4(&toside->oaddr)); struct vu_dev *vdev = c->vdev; struct vu_virtq *vq = &vdev->vq[VHOST_USER_RX_QUEUE]; - struct iov_tail data; int i;
+ ASSERT(!c->no_udp); + + if (!vu_queue_enabled(vq) || !vu_queue_started(vq)) { + struct msghdr msg = { 0 }; + + debug("Got UDP packet, but RX virtqueue not usable yet"); + + for (i = 0; i < n; i++) { + if (recvmsg(s, &msg, MSG_DONTWAIT) < 0) + debug_perror("Failed to discard datagram"); + } + + return; + } + for (i = 0; i < n; i++) { size_t iov_cnt; ssize_t dlen; + int elem_cnt;
- iov_cnt = VIRTQUEUE_MAX_SIZE; - dlen = udp_vu_sock_recv(c, iov_vu, &iov_cnt, vq, s, v6); - if (dlen < 0) + vu_init_elem(elem, iov_vu, ARRAY_SIZE(elem)); + + elem_cnt = vu_collect(vdev, vq, elem, ARRAY_SIZE(elem), + IP_MAX_MTU + ETH_HLEN + VNET_HLEN, NULL); + if (elem_cnt == 0) break;
+ iov_cnt = elem_cnt; + dlen = udp_vu_sock_recv(iov_vu, &iov_cnt, s, v6); + if (dlen < 0) { + vu_queue_rewind(vq, elem_cnt); + break; + } + + /* release unused buffers */ + vu_queue_rewind(vq, elem_cnt - iov_cnt); + if (iov_cnt > 0) { - data = IOV_TAIL(iov_vu, iov_cnt, 0); + struct iov_tail data = IOV_TAIL(iov_vu, iov_cnt, 0); + vu_set_vnethdr(iov_vu[0].iov_base, iov_cnt); udp_vu_prepare(c, &data, toside, dlen); if (*c->pcap) { udp_vu_csum(toside, &data); -- 2.53.0
-- David Gibson (he or they) | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you, not the other way | around. http://www.ozlabs.org/~dgibson