static int
nl_sock_send__(struct nl_sock *sock, const struct ofpbuf *msg,
uint32_t nlmsg_seq, bool wait)
{
struct nlmsghdr *nlmsg = nl_msg_nlmsghdr(msg);
int error;
nlmsg->nlmsg_len = ofpbuf_size(msg);
nlmsg->nlmsg_seq = nlmsg_seq;
nlmsg->nlmsg_pid = sock->pid;
do {
int retval;
#ifdef _WIN32
bool result;
DWORD last_error = 0;
result = WriteFile(sock->handle, ofpbuf_data(msg), ofpbuf_size(msg),
&retval, NULL);
last_error = GetLastError();
if (last_error != ERROR_SUCCESS && !result) {
retval = -1;
errno = EAGAIN;
}
#else
retval = send(sock->fd, ofpbuf_data(msg), ofpbuf_size(msg), wait ? 0 : MSG_DONTWAIT);
#endif
error = retval < 0 ? errno : 0;
} while (error == EINTR);
log_nlmsg(__func__, error, ofpbuf_data(msg), ofpbuf_size(msg), sock->protocol);
if (!error) {
COVERAGE_INC(netlink_sent);
}
return error;
}
static int
nl_sock_send__(struct nl_sock *sock, const struct ofpbuf *msg,
uint32_t nlmsg_seq, bool wait)
{
struct nlmsghdr *nlmsg = nl_msg_nlmsghdr(msg);
int error;
nlmsg->nlmsg_len = msg->size;
nlmsg->nlmsg_seq = nlmsg_seq;
nlmsg->nlmsg_pid = sock->pid;
do {
int retval;
retval = send(sock->fd, msg->data, msg->size, wait ? 0 : MSG_DONTWAIT);
error = retval < 0 ? errno : 0;
} while (error == EINTR);
log_nlmsg(__func__, error, msg->data, msg->size, sock->protocol);
if (!error) {
COVERAGE_INC(netlink_sent);
}
return error;
}
static int
nl_sock_transact_multiple__(struct nl_sock *sock,
struct nl_transaction **transactions, size_t n,
size_t *done)
{
uint64_t tmp_reply_stub[1024 / 8];
struct nl_transaction tmp_txn;
struct ofpbuf tmp_reply;
uint32_t base_seq;
struct iovec iovs[MAX_IOVS];
struct msghdr msg;
int error;
int i;
base_seq = nl_sock_allocate_seq(sock, n);
*done = 0;
for (i = 0; i < n; i++) {
struct nl_transaction *txn = transactions[i];
struct nlmsghdr *nlmsg = nl_msg_nlmsghdr(txn->request);
nlmsg->nlmsg_len = txn->request->size;
nlmsg->nlmsg_seq = base_seq + i;
nlmsg->nlmsg_pid = sock->pid;
iovs[i].iov_base = txn->request->data;
iovs[i].iov_len = txn->request->size;
}
memset(&msg, 0, sizeof msg);
msg.msg_iov = iovs;
msg.msg_iovlen = n;
do {
error = sendmsg(sock->fd, &msg, 0) < 0 ? errno : 0;
} while (error == EINTR);
for (i = 0; i < n; i++) {
struct nl_transaction *txn = transactions[i];
log_nlmsg(__func__, error, txn->request->data, txn->request->size,
sock->protocol);
}
if (!error) {
COVERAGE_ADD(netlink_sent, n);
}
if (error) {
return error;
}
ofpbuf_use_stub(&tmp_reply, tmp_reply_stub, sizeof tmp_reply_stub);
tmp_txn.request = NULL;
tmp_txn.reply = &tmp_reply;
tmp_txn.error = 0;
while (n > 0) {
struct nl_transaction *buf_txn, *txn;
uint32_t seq;
/* Find a transaction whose buffer we can use for receiving a reply.
* If no such transaction is left, use tmp_txn. */
buf_txn = &tmp_txn;
for (i = 0; i < n; i++) {
if (transactions[i]->reply) {
buf_txn = transactions[i];
break;
}
}
/* Receive a reply. */
error = nl_sock_recv__(sock, buf_txn->reply, false);
if (error) {
if (error == EAGAIN) {
nl_sock_record_errors__(transactions, n, 0);
*done += n;
error = 0;
}
break;
}
/* Match the reply up with a transaction. */
seq = nl_msg_nlmsghdr(buf_txn->reply)->nlmsg_seq;
if (seq < base_seq || seq >= base_seq + n) {
VLOG_DBG_RL(&rl, "ignoring unexpected seq %#"PRIx32, seq);
continue;
}
i = seq - base_seq;
txn = transactions[i];
/* Fill in the results for 'txn'. */
if (nl_msg_nlmsgerr(buf_txn->reply, &txn->error)) {
if (txn->reply) {
ofpbuf_clear(txn->reply);
}
if (txn->error) {
VLOG_DBG_RL(&rl, "received NAK error=%d (%s)",
error, ovs_strerror(txn->error));
}
} else {
txn->error = 0;
if (txn->reply && txn != buf_txn) {
/* Swap buffers. */
struct ofpbuf *reply = buf_txn->reply;
buf_txn->reply = txn->reply;
txn->reply = reply;
}
}
/* Fill in the results for transactions before 'txn'. (We have to do
* this after the results for 'txn' itself because of the buffer swap
* above.) */
nl_sock_record_errors__(transactions, i, 0);
/* Advance. */
*done += i + 1;
transactions += i + 1;
n -= i + 1;
base_seq += i + 1;
}
ofpbuf_uninit(&tmp_reply);
return error;
}
static int
nl_sock_recv__(struct nl_sock *sock, struct ofpbuf *buf, bool wait)
{
/* We can't accurately predict the size of the data to be received. The
* caller is supposed to have allocated enough space in 'buf' to handle the
* "typical" case. To handle exceptions, we make available enough space in
* 'tail' to allow Netlink messages to be up to 64 kB long (a reasonable
* figure since that's the maximum length of a Netlink attribute). */
struct nlmsghdr *nlmsghdr;
uint8_t tail[65536];
struct iovec iov[2];
struct msghdr msg;
ssize_t retval;
ovs_assert(buf->allocated >= sizeof *nlmsghdr);
ofpbuf_clear(buf);
iov[0].iov_base = buf->base;
iov[0].iov_len = buf->allocated;
iov[1].iov_base = tail;
iov[1].iov_len = sizeof tail;
memset(&msg, 0, sizeof msg);
msg.msg_iov = iov;
msg.msg_iovlen = 2;
do {
retval = recvmsg(sock->fd, &msg, wait ? 0 : MSG_DONTWAIT);
} while (retval < 0 && errno == EINTR);
if (retval < 0) {
int error = errno;
if (error == ENOBUFS) {
/* Socket receive buffer overflow dropped one or more messages that
* the kernel tried to send to us. */
COVERAGE_INC(netlink_overflow);
}
return error;
}
if (msg.msg_flags & MSG_TRUNC) {
VLOG_ERR_RL(&rl, "truncated message (longer than %zu bytes)",
sizeof tail);
return E2BIG;
}
nlmsghdr = buf->data;
if (retval < sizeof *nlmsghdr
|| nlmsghdr->nlmsg_len < sizeof *nlmsghdr
|| nlmsghdr->nlmsg_len > retval) {
VLOG_ERR_RL(&rl, "received invalid nlmsg (%zd bytes < %zu)",
retval, sizeof *nlmsghdr);
return EPROTO;
}
buf->size = MIN(retval, buf->allocated);
if (retval > buf->allocated) {
COVERAGE_INC(netlink_recv_jumbo);
ofpbuf_put(buf, tail, retval - buf->allocated);
}
log_nlmsg(__func__, 0, buf->data, buf->size, sock->protocol);
COVERAGE_INC(netlink_received);
return 0;
}
static int
nl_sock_recv__(struct nl_sock *sock, struct ofpbuf *buf, bool wait)
{
/* We can't accurately predict the size of the data to be received. The
* caller is supposed to have allocated enough space in 'buf' to handle the
* "typical" case. To handle exceptions, we make available enough space in
* 'tail' to allow Netlink messages to be up to 64 kB long (a reasonable
* figure since that's the maximum length of a Netlink attribute). */
struct nlmsghdr *nlmsghdr;
#ifdef _WIN32
#define MAX_STACK_LENGTH 81920
uint8_t tail[MAX_STACK_LENGTH];
#else
uint8_t tail[65536];
#endif
struct iovec iov[2];
struct msghdr msg;
ssize_t retval;
int error;
ovs_assert(buf->allocated >= sizeof *nlmsghdr);
ofpbuf_clear(buf);
iov[0].iov_base = ofpbuf_base(buf);
iov[0].iov_len = buf->allocated;
iov[1].iov_base = tail;
iov[1].iov_len = sizeof tail;
memset(&msg, 0, sizeof msg);
msg.msg_iov = iov;
msg.msg_iovlen = 2;
/* Receive a Netlink message from the kernel.
*
* This works around a kernel bug in which the kernel returns an error code
* as if it were the number of bytes read. It doesn't actually modify
* anything in the receive buffer in that case, so we can initialize the
* Netlink header with an impossible message length and then, upon success,
* check whether it changed. */
nlmsghdr = ofpbuf_base(buf);
do {
nlmsghdr->nlmsg_len = UINT32_MAX;
#ifdef _WIN32
boolean result = false;
DWORD last_error = 0;
result = ReadFile(sock->handle, tail, MAX_STACK_LENGTH, &retval, NULL);
last_error = GetLastError();
if (last_error != ERROR_SUCCESS && !result) {
retval = -1;
errno = EAGAIN;
} else {
ofpbuf_put(buf, tail, retval);
}
#else
retval = recvmsg(sock->fd, &msg, wait ? 0 : MSG_DONTWAIT);
#endif
error = (retval < 0 ? errno
: retval == 0 ? ECONNRESET /* not possible? */
: nlmsghdr->nlmsg_len != UINT32_MAX ? 0
: retval);
} while (error == EINTR);
if (error) {
if (error == ENOBUFS) {
/* Socket receive buffer overflow dropped one or more messages that
* the kernel tried to send to us. */
COVERAGE_INC(netlink_overflow);
}
return error;
}
if (msg.msg_flags & MSG_TRUNC) {
VLOG_ERR_RL(&rl, "truncated message (longer than %"PRIuSIZE" bytes)",
sizeof tail);
return E2BIG;
}
if (retval < sizeof *nlmsghdr
|| nlmsghdr->nlmsg_len < sizeof *nlmsghdr
|| nlmsghdr->nlmsg_len > retval) {
VLOG_ERR_RL(&rl, "received invalid nlmsg (%"PRIuSIZE" bytes < %"PRIuSIZE")",
retval, sizeof *nlmsghdr);
return EPROTO;
}
#ifndef _WIN32
ofpbuf_set_size(buf, MIN(retval, buf->allocated));
if (retval > buf->allocated) {
COVERAGE_INC(netlink_recv_jumbo);
ofpbuf_put(buf, tail, retval - buf->allocated);
}
#endif
log_nlmsg(__func__, 0, ofpbuf_data(buf), ofpbuf_size(buf), sock->protocol);
COVERAGE_INC(netlink_received);
return 0;
}
