1
0
mirror of https://passt.top/passt synced 2025-01-22 04:05:22 +00:00
passt/tcp_vu.c
Stefano Brivio a8f4fc481c tcp: Mask EPOLLIN altogether if we're blocked waiting on an ACK from the guest
There are pretty much two cases of the (misnomer) STALLED: in one
case, we could send more data to the guest if it becomes available,
and in another case, we can't, because we filled the window.

If, in this second case, we keep EPOLLIN enabled, but never read from
the socket, we get short but CPU-annoying storms of EPOLLIN events,
upon which we reschedule the ACK timeout handler, never read from the
socket, go back to epoll_wait(), and so on:

  timerfd_settime(76, 0, {it_interval={tv_sec=0, tv_nsec=0}, it_value={tv_sec=2, tv_nsec=0}}, NULL) = 0
  epoll_wait(3, [{events=EPOLLIN, data={u32=10497, u64=38654716161}}], 8, 1000) = 1
  timerfd_settime(76, 0, {it_interval={tv_sec=0, tv_nsec=0}, it_value={tv_sec=2, tv_nsec=0}}, NULL) = 0
  epoll_wait(3, [{events=EPOLLIN, data={u32=10497, u64=38654716161}}], 8, 1000) = 1
  timerfd_settime(76, 0, {it_interval={tv_sec=0, tv_nsec=0}, it_value={tv_sec=2, tv_nsec=0}}, NULL) = 0
  epoll_wait(3, [{events=EPOLLIN, data={u32=10497, u64=38654716161}}], 8, 1000) = 1

also known as:

  29.1517: Flow 2 (TCP connection): timer expires in 2.000s
  29.1517: Flow 2 (TCP connection): timer expires in 2.000s
  29.1517: Flow 2 (TCP connection): timer expires in 2.000s

which, for some reason, becomes very visible with muvm and aria2c
downloading from a server nearby in parallel chunks.

That's because EPOLLIN isn't cleared if we don't read from the socket,
and even with EPOLLET, epoll_wait() will repeatedly wake us up until
we actually read something.

In this case, we don't want to subscribe to EPOLLIN at all: all we're
waiting for is an ACK segment from the guest. Differentiate this case
with a new connection flag, ACK_FROM_TAP_BLOCKS, which doesn't just
indicate that we're waiting for an ACK from the guest
(ACK_FROM_TAP_DUE), but also that we're blocked waiting for it.

If this flag is set before we set STALLED, EPOLLIN will be masked
while we set EPOLLET because of STALLED. Whenever we clear STALLED,
we also clear this flag.

This is definitely not elegant, but it's a minimal fix.

We can probably simplify this at a later point by having a category
of connection flags directly corresponding to epoll flags, and
dropping STALLED altogether, or, perhaps, always using EPOLLET (but
we need a mechanism to re-check sockets for pending data if we can't
temporarily write to the guest).

I suspect that this might also be implied in
https://github.com/containers/podman/issues/23686, hence the Link:
tag. It doesn't necessarily mean I'm fixing it (I can't reproduce
that).

Link: https://github.com/containers/podman/issues/23686
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2025-01-16 21:15:33 +01:00

471 lines
11 KiB
C

// SPDX-License-Identifier: GPL-2.0-or-later
/* tcp_vu.c - TCP L2 vhost-user management functions
*
* Copyright Red Hat
* Author: Laurent Vivier <lvivier@redhat.com>
*/
#include <errno.h>
#include <stddef.h>
#include <stdint.h>
#include <netinet/ip.h>
#include <netinet/tcp.h>
#include <sys/socket.h>
#include <netinet/if_ether.h>
#include <linux/virtio_net.h>
#include "util.h"
#include "ip.h"
#include "passt.h"
#include "siphash.h"
#include "inany.h"
#include "vhost_user.h"
#include "tcp.h"
#include "pcap.h"
#include "flow.h"
#include "tcp_conn.h"
#include "flow_table.h"
#include "tcp_vu.h"
#include "tap.h"
#include "tcp_internal.h"
#include "checksum.h"
#include "vu_common.h"
#include <time.h>
static struct iovec iov_vu[VIRTQUEUE_MAX_SIZE + 1];
static struct vu_virtq_element elem[VIRTQUEUE_MAX_SIZE];
static int head[VIRTQUEUE_MAX_SIZE + 1];
static int head_cnt;
/**
* tcp_vu_hdrlen() - return the size of the header in level 2 frame (TCP)
* @v6: Set for IPv6 packet
*
* Return: Return the size of the header
*/
static size_t tcp_vu_hdrlen(bool v6)
{
size_t hdrlen;
hdrlen = sizeof(struct virtio_net_hdr_mrg_rxbuf) +
sizeof(struct ethhdr) + sizeof(struct tcphdr);
if (v6)
hdrlen += sizeof(struct ipv6hdr);
else
hdrlen += sizeof(struct iphdr);
return hdrlen;
}
/**
* tcp_vu_send_flag() - Send segment with flags to vhost-user (no payload)
* @c: Execution context
* @conn: Connection pointer
* @flags: TCP flags: if not set, send segment only if ACK is due
*
* Return: negative error code on connection reset, 0 otherwise
*/
int tcp_vu_send_flag(const struct ctx *c, struct tcp_tap_conn *conn, int flags)
{
struct vu_dev *vdev = c->vdev;
struct vu_virtq *vq = &vdev->vq[VHOST_USER_RX_QUEUE];
size_t optlen, hdrlen;
struct vu_virtq_element flags_elem[2];
struct ipv6hdr *ip6h = NULL;
struct iphdr *ip4h = NULL;
struct iovec flags_iov[2];
struct tcp_syn_opts *opts;
struct iov_tail payload;
struct tcphdr *th;
struct ethhdr *eh;
uint32_t seq;
int elem_cnt;
int nb_ack;
int ret;
hdrlen = tcp_vu_hdrlen(CONN_V6(conn));
vu_set_element(&flags_elem[0], NULL, &flags_iov[0]);
elem_cnt = vu_collect(vdev, vq, &flags_elem[0], 1,
hdrlen + sizeof(struct tcp_syn_opts), NULL);
if (elem_cnt != 1)
return -1;
ASSERT(flags_elem[0].in_sg[0].iov_len >=
hdrlen + sizeof(struct tcp_syn_opts));
vu_set_vnethdr(vdev, flags_elem[0].in_sg[0].iov_base, 1);
eh = vu_eth(flags_elem[0].in_sg[0].iov_base);
memcpy(eh->h_dest, c->guest_mac, sizeof(eh->h_dest));
memcpy(eh->h_source, c->our_tap_mac, sizeof(eh->h_source));
if (CONN_V4(conn)) {
eh->h_proto = htons(ETH_P_IP);
ip4h = vu_ip(flags_elem[0].in_sg[0].iov_base);
*ip4h = (struct iphdr)L2_BUF_IP4_INIT(IPPROTO_TCP);
th = vu_payloadv4(flags_elem[0].in_sg[0].iov_base);
} else {
eh->h_proto = htons(ETH_P_IPV6);
ip6h = vu_ip(flags_elem[0].in_sg[0].iov_base);
*ip6h = (struct ipv6hdr)L2_BUF_IP6_INIT(IPPROTO_TCP);
th = vu_payloadv6(flags_elem[0].in_sg[0].iov_base);
}
memset(th, 0, sizeof(*th));
th->doff = sizeof(*th) / 4;
th->ack = 1;
seq = conn->seq_to_tap;
opts = (struct tcp_syn_opts *)(th + 1);
ret = tcp_prepare_flags(c, conn, flags, th, opts, &optlen);
if (ret <= 0) {
vu_queue_rewind(vq, 1);
return ret;
}
flags_elem[0].in_sg[0].iov_len = hdrlen + optlen;
payload = IOV_TAIL(flags_elem[0].in_sg, 1, hdrlen);
tcp_fill_headers(conn, NULL, ip4h, ip6h, th, &payload,
NULL, seq, !*c->pcap);
if (*c->pcap) {
pcap_iov(&flags_elem[0].in_sg[0], 1,
sizeof(struct virtio_net_hdr_mrg_rxbuf));
}
nb_ack = 1;
if (flags & DUP_ACK) {
vu_set_element(&flags_elem[1], NULL, &flags_iov[1]);
elem_cnt = vu_collect(vdev, vq, &flags_elem[1], 1,
flags_elem[0].in_sg[0].iov_len, NULL);
if (elem_cnt == 1 &&
flags_elem[1].in_sg[0].iov_len >=
flags_elem[0].in_sg[0].iov_len) {
memcpy(flags_elem[1].in_sg[0].iov_base,
flags_elem[0].in_sg[0].iov_base,
flags_elem[0].in_sg[0].iov_len);
nb_ack++;
if (*c->pcap) {
pcap_iov(&flags_elem[1].in_sg[0], 1,
sizeof(struct virtio_net_hdr_mrg_rxbuf));
}
}
}
vu_flush(vdev, vq, flags_elem, nb_ack);
return 0;
}
/** tcp_vu_sock_recv() - Receive datastream from socket into vhost-user buffers
* @c: Execution context
* @conn: Connection pointer
* @v6: Set for IPv6 connections
* @already_sent: Number of bytes already sent
* @fillsize: Maximum bytes to fill in guest-side receiving window
* @iov_cnt: number of iov (output)
*
* Return: Number of iov entries used to store the data or negative error code
*/
static ssize_t tcp_vu_sock_recv(const struct ctx *c,
const struct tcp_tap_conn *conn, bool v6,
uint32_t already_sent, size_t fillsize,
int *iov_cnt)
{
struct vu_dev *vdev = c->vdev;
struct vu_virtq *vq = &vdev->vq[VHOST_USER_RX_QUEUE];
struct msghdr mh_sock = { 0 };
uint16_t mss = MSS_GET(conn);
int s = conn->sock;
ssize_t ret, len;
size_t hdrlen;
int elem_cnt;
int i;
*iov_cnt = 0;
hdrlen = tcp_vu_hdrlen(v6);
vu_init_elem(elem, &iov_vu[1], VIRTQUEUE_MAX_SIZE);
elem_cnt = 0;
head_cnt = 0;
while (fillsize > 0 && elem_cnt < VIRTQUEUE_MAX_SIZE) {
struct iovec *iov;
size_t frame_size, dlen;
int cnt;
cnt = vu_collect(vdev, vq, &elem[elem_cnt],
VIRTQUEUE_MAX_SIZE - elem_cnt,
MIN(mss, fillsize) + hdrlen, &frame_size);
if (cnt == 0)
break;
dlen = frame_size - hdrlen;
/* reserve space for headers in iov */
iov = &elem[elem_cnt].in_sg[0];
ASSERT(iov->iov_len >= hdrlen);
iov->iov_base = (char *)iov->iov_base + hdrlen;
iov->iov_len -= hdrlen;
head[head_cnt++] = elem_cnt;
fillsize -= dlen;
elem_cnt += cnt;
}
if (peek_offset_cap) {
mh_sock.msg_iov = iov_vu + 1;
mh_sock.msg_iovlen = elem_cnt;
} else {
iov_vu[0].iov_base = tcp_buf_discard;
iov_vu[0].iov_len = already_sent;
mh_sock.msg_iov = iov_vu;
mh_sock.msg_iovlen = elem_cnt + 1;
}
do
ret = recvmsg(s, &mh_sock, MSG_PEEK);
while (ret < 0 && errno == EINTR);
if (ret < 0) {
vu_queue_rewind(vq, elem_cnt);
return -errno;
}
if (!peek_offset_cap)
ret -= already_sent;
/* adjust iov number and length of the last iov */
len = ret;
for (i = 0; len && i < elem_cnt; i++) {
struct iovec *iov = &elem[i].in_sg[0];
if (iov->iov_len > (size_t)len)
iov->iov_len = len;
len -= iov->iov_len;
}
/* adjust head count */
while (head_cnt > 0 && head[head_cnt - 1] > i)
head_cnt--;
/* mark end of array */
head[head_cnt] = i;
*iov_cnt = i;
/* release unused buffers */
vu_queue_rewind(vq, elem_cnt - i);
/* restore space for headers in iov */
for (i = 0; i < head_cnt; i++) {
struct iovec *iov = &elem[head[i]].in_sg[0];
iov->iov_base = (char *)iov->iov_base - hdrlen;
iov->iov_len += hdrlen;
}
return ret;
}
/**
* tcp_vu_prepare() - Prepare the frame header
* @c: Execution context
* @conn: Connection pointer
* @iov: Pointer to the array of IO vectors
* @iov_cnt: Number of entries in @iov
* @check: Checksum, if already known
* @no_tcp_csum: Do not set TCP checksum
*/
static void tcp_vu_prepare(const struct ctx *c, struct tcp_tap_conn *conn,
struct iovec *iov, size_t iov_cnt,
const uint16_t **check, bool no_tcp_csum)
{
const struct flowside *toside = TAPFLOW(conn);
bool v6 = !(inany_v4(&toside->eaddr) && inany_v4(&toside->oaddr));
size_t hdrlen = tcp_vu_hdrlen(v6);
struct iov_tail payload = IOV_TAIL(iov, iov_cnt, hdrlen);
char *base = iov[0].iov_base;
struct ipv6hdr *ip6h = NULL;
struct iphdr *ip4h = NULL;
struct tcphdr *th;
struct ethhdr *eh;
/* we guess the first iovec provided by the guest can embed
* all the headers needed by L2 frame
*/
ASSERT(iov[0].iov_len >= hdrlen);
eh = vu_eth(base);
memcpy(eh->h_dest, c->guest_mac, sizeof(eh->h_dest));
memcpy(eh->h_source, c->our_tap_mac, sizeof(eh->h_source));
/* initialize header */
if (!v6) {
eh->h_proto = htons(ETH_P_IP);
ip4h = vu_ip(base);
*ip4h = (struct iphdr)L2_BUF_IP4_INIT(IPPROTO_TCP);
th = vu_payloadv4(base);
} else {
eh->h_proto = htons(ETH_P_IPV6);
ip6h = vu_ip(base);
*ip6h = (struct ipv6hdr)L2_BUF_IP6_INIT(IPPROTO_TCP);
th = vu_payloadv6(base);
}
memset(th, 0, sizeof(*th));
th->doff = sizeof(*th) / 4;
th->ack = 1;
tcp_fill_headers(conn, NULL, ip4h, ip6h, th, &payload,
*check, conn->seq_to_tap, no_tcp_csum);
if (ip4h)
*check = &ip4h->check;
}
/**
* tcp_vu_data_from_sock() - Handle new data from socket, queue to vhost-user,
* in window
* @c: Execution context
* @conn: Connection pointer
*
* Return: Negative on connection reset, 0 otherwise
*/
int tcp_vu_data_from_sock(const struct ctx *c, struct tcp_tap_conn *conn)
{
uint32_t wnd_scaled = conn->wnd_from_tap << conn->ws_from_tap;
struct vu_dev *vdev = c->vdev;
struct vu_virtq *vq = &vdev->vq[VHOST_USER_RX_QUEUE];
ssize_t len, previous_dlen;
size_t hdrlen, fillsize;
int v6 = CONN_V6(conn);
uint32_t already_sent;
const uint16_t *check;
int i, iov_cnt;
if (!vu_queue_enabled(vq) || !vu_queue_started(vq)) {
debug("Got packet, but RX virtqueue not usable yet");
return 0;
}
already_sent = conn->seq_to_tap - conn->seq_ack_from_tap;
if (SEQ_LT(already_sent, 0)) {
/* RFC 761, section 2.1. */
flow_trace(conn, "ACK sequence gap: ACK for %u, sent: %u",
conn->seq_ack_from_tap, conn->seq_to_tap);
conn->seq_to_tap = conn->seq_ack_from_tap;
already_sent = 0;
if (tcp_set_peek_offset(conn->sock, 0)) {
tcp_rst(c, conn);
return -1;
}
}
if (!wnd_scaled || already_sent >= wnd_scaled) {
conn_flag(c, conn, ACK_FROM_TAP_BLOCKS);
conn_flag(c, conn, STALLED);
conn_flag(c, conn, ACK_FROM_TAP_DUE);
return 0;
}
/* Set up buffer descriptors we'll fill completely and partially. */
fillsize = wnd_scaled - already_sent;
/* collect the buffers from vhost-user and fill them with the
* data from the socket
*/
len = tcp_vu_sock_recv(c, conn, v6, already_sent, fillsize, &iov_cnt);
if (len < 0) {
if (len != -EAGAIN && len != -EWOULDBLOCK) {
tcp_rst(c, conn);
return len;
}
if (already_sent) /* No new data and EAGAIN: set EPOLLET */
conn_flag(c, conn, STALLED);
return 0;
}
if (!len) {
if (already_sent) {
conn_flag(c, conn, STALLED);
} else if ((conn->events & (SOCK_FIN_RCVD | TAP_FIN_SENT)) ==
SOCK_FIN_RCVD) {
int ret = tcp_vu_send_flag(c, conn, FIN | ACK);
if (ret) {
tcp_rst(c, conn);
return ret;
}
conn_event(c, conn, TAP_FIN_SENT);
}
return 0;
}
conn_flag(c, conn, ~ACK_FROM_TAP_BLOCKS);
conn_flag(c, conn, ~STALLED);
/* Likely, some new data was acked too. */
tcp_update_seqack_wnd(c, conn, false, NULL);
/* initialize headers */
/* iov_vu is an array of buffers and the buffer size can be
* smaller than the frame size we want to use but with
* num_buffer we can merge several virtio iov buffers in one packet
* we need only to set the packet headers in the first iov and
* num_buffer to the number of iov entries
*/
hdrlen = tcp_vu_hdrlen(v6);
for (i = 0, previous_dlen = -1, check = NULL; i < head_cnt; i++) {
struct iovec *iov = &elem[head[i]].in_sg[0];
int buf_cnt = head[i + 1] - head[i];
ssize_t dlen = iov_size(iov, buf_cnt) - hdrlen;
vu_set_vnethdr(vdev, iov->iov_base, buf_cnt);
/* The IPv4 header checksum varies only with dlen */
if (previous_dlen != dlen)
check = NULL;
previous_dlen = dlen;
tcp_vu_prepare(c, conn, iov, buf_cnt, &check, !*c->pcap);
if (*c->pcap) {
pcap_iov(iov, buf_cnt,
sizeof(struct virtio_net_hdr_mrg_rxbuf));
}
conn->seq_to_tap += dlen;
}
/* send packets */
vu_flush(vdev, vq, elem, iov_cnt);
conn_flag(c, conn, ACK_FROM_TAP_DUE);
return 0;
}