Systemd/src/basic/io-util.c
Lennart Poettering 60d9771c59 core: rework how we flush incoming traffic when a socket unit goes down
Previously, we'd simply close and reopen the socket file descriptors. This is
problematic however, as we won't transition through the SOCKET_CHOWN state
then, and thus the file ownership won't be correct for the sockets.

Rework the flushing logic, and actually read any queued data from the sockets
for flushing, and accept any queued messages and disconnect them.
2016-05-06 13:29:26 +02:00

270 lines
7.2 KiB
C

/***
This file is part of systemd.
Copyright 2010 Lennart Poettering
systemd is free software; you can redistribute it and/or modify it
under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation; either version 2.1 of the License, or
(at your option) any later version.
systemd is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
#include <errno.h>
#include <limits.h>
#include <poll.h>
#include <stdio.h>
#include <time.h>
#include <unistd.h>
#include "io-util.h"
#include "time-util.h"
int flush_fd(int fd) {
struct pollfd pollfd = {
.fd = fd,
.events = POLLIN,
};
/* Read from the specified file descriptor, until POLLIN is not set anymore, throwing away everything
* read. Note that some file descriptors (notable IP sockets) will trigger POLLIN even when no data can be read
* (due to IP packet checksum mismatches), hence this function is only safe to be non-blocking if the fd used
* was set to non-blocking too. */
for (;;) {
char buf[LINE_MAX];
ssize_t l;
int r;
r = poll(&pollfd, 1, 0);
if (r < 0) {
if (errno == EINTR)
continue;
return -errno;
} else if (r == 0)
return 0;
l = read(fd, buf, sizeof(buf));
if (l < 0) {
if (errno == EINTR)
continue;
if (errno == EAGAIN)
return 0;
return -errno;
} else if (l == 0)
return 0;
}
}
ssize_t loop_read(int fd, void *buf, size_t nbytes, bool do_poll) {
uint8_t *p = buf;
ssize_t n = 0;
assert(fd >= 0);
assert(buf);
/* If called with nbytes == 0, let's call read() at least
* once, to validate the operation */
if (nbytes > (size_t) SSIZE_MAX)
return -EINVAL;
do {
ssize_t k;
k = read(fd, p, nbytes);
if (k < 0) {
if (errno == EINTR)
continue;
if (errno == EAGAIN && do_poll) {
/* We knowingly ignore any return value here,
* and expect that any error/EOF is reported
* via read() */
(void) fd_wait_for_event(fd, POLLIN, USEC_INFINITY);
continue;
}
return n > 0 ? n : -errno;
}
if (k == 0)
return n;
assert((size_t) k <= nbytes);
p += k;
nbytes -= k;
n += k;
} while (nbytes > 0);
return n;
}
int loop_read_exact(int fd, void *buf, size_t nbytes, bool do_poll) {
ssize_t n;
n = loop_read(fd, buf, nbytes, do_poll);
if (n < 0)
return (int) n;
if ((size_t) n != nbytes)
return -EIO;
return 0;
}
int loop_write(int fd, const void *buf, size_t nbytes, bool do_poll) {
const uint8_t *p = buf;
assert(fd >= 0);
assert(buf);
if (nbytes > (size_t) SSIZE_MAX)
return -EINVAL;
do {
ssize_t k;
k = write(fd, p, nbytes);
if (k < 0) {
if (errno == EINTR)
continue;
if (errno == EAGAIN && do_poll) {
/* We knowingly ignore any return value here,
* and expect that any error/EOF is reported
* via write() */
(void) fd_wait_for_event(fd, POLLOUT, USEC_INFINITY);
continue;
}
return -errno;
}
if (_unlikely_(nbytes > 0 && k == 0)) /* Can't really happen */
return -EIO;
assert((size_t) k <= nbytes);
p += k;
nbytes -= k;
} while (nbytes > 0);
return 0;
}
int pipe_eof(int fd) {
struct pollfd pollfd = {
.fd = fd,
.events = POLLIN|POLLHUP,
};
int r;
r = poll(&pollfd, 1, 0);
if (r < 0)
return -errno;
if (r == 0)
return 0;
return pollfd.revents & POLLHUP;
}
int fd_wait_for_event(int fd, int event, usec_t t) {
struct pollfd pollfd = {
.fd = fd,
.events = event,
};
struct timespec ts;
int r;
r = ppoll(&pollfd, 1, t == USEC_INFINITY ? NULL : timespec_store(&ts, t), NULL);
if (r < 0)
return -errno;
if (r == 0)
return 0;
return pollfd.revents;
}
static size_t nul_length(const uint8_t *p, size_t sz) {
size_t n = 0;
while (sz > 0) {
if (*p != 0)
break;
n++;
p++;
sz--;
}
return n;
}
ssize_t sparse_write(int fd, const void *p, size_t sz, size_t run_length) {
const uint8_t *q, *w, *e;
ssize_t l;
q = w = p;
e = q + sz;
while (q < e) {
size_t n;
n = nul_length(q, e - q);
/* If there are more than the specified run length of
* NUL bytes, or if this is the beginning or the end
* of the buffer, then seek instead of write */
if ((n > run_length) ||
(n > 0 && q == p) ||
(n > 0 && q + n >= e)) {
if (q > w) {
l = write(fd, w, q - w);
if (l < 0)
return -errno;
if (l != q -w)
return -EIO;
}
if (lseek(fd, n, SEEK_CUR) == (off_t) -1)
return -errno;
q += n;
w = q;
} else if (n > 0)
q += n;
else
q++;
}
if (q > w) {
l = write(fd, w, q - w);
if (l < 0)
return -errno;
if (l != q - w)
return -EIO;
}
return q - (const uint8_t*) p;
}