4 * Copyright (c) 2003-2008 Fabrice Bellard
5 * Copyright (c) 2009 Red Hat, Inc.
7 * Permission is hereby granted, free of charge, to any person obtaining a copy
8 * of this software and associated documentation files (the "Software"), to deal
9 * in the Software without restriction, including without limitation the rights
10 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
11 * copies of the Software, and to permit persons to whom the Software is
12 * furnished to do so, subject to the following conditions:
14 * The above copyright notice and this permission notice shall be included in
15 * all copies or substantial portions of the Software.
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
18 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
19 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
20 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
21 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
22 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
28 #include "config-host.h"
31 #include <sys/ioctl.h>
34 #include <sys/socket.h>
39 #include "qemu-char.h"
40 #include "qemu-common.h"
41 #include "qemu-error.h"
43 #include "net/tap-linux.h"
45 /* Maximum GSO packet size (64k) plus plenty of room for
46 * the ethernet and virtio_net headers
48 #define TAP_BUFSIZE (4096 + 65536)
50 typedef struct TAPState
{
53 char down_script
[1024];
54 char down_script_arg
[128];
55 uint8_t buf
[TAP_BUFSIZE
];
56 unsigned int read_poll
: 1;
57 unsigned int write_poll
: 1;
58 unsigned int has_vnet_hdr
: 1;
59 unsigned int using_vnet_hdr
: 1;
60 unsigned int has_ufo
: 1;
63 static int launch_script(const char *setup_script
, const char *ifname
, int fd
);
65 static int tap_can_send(void *opaque
);
66 static void tap_send(void *opaque
);
67 static void tap_writable(void *opaque
);
69 static void tap_update_fd_handler(TAPState
*s
)
71 qemu_set_fd_handler2(s
->fd
,
72 s
->read_poll
? tap_can_send
: NULL
,
73 s
->read_poll
? tap_send
: NULL
,
74 s
->write_poll
? tap_writable
: NULL
,
78 static void tap_read_poll(TAPState
*s
, int enable
)
80 s
->read_poll
= !!enable
;
81 tap_update_fd_handler(s
);
84 static void tap_write_poll(TAPState
*s
, int enable
)
86 s
->write_poll
= !!enable
;
87 tap_update_fd_handler(s
);
90 static void tap_writable(void *opaque
)
96 qemu_flush_queued_packets(&s
->nc
);
99 static ssize_t
tap_write_packet(TAPState
*s
, const struct iovec
*iov
, int iovcnt
)
104 len
= writev(s
->fd
, iov
, iovcnt
);
105 } while (len
== -1 && errno
== EINTR
);
107 if (len
== -1 && errno
== EAGAIN
) {
108 tap_write_poll(s
, 1);
115 static ssize_t
tap_receive_iov(VLANClientState
*nc
, const struct iovec
*iov
,
118 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
119 const struct iovec
*iovp
= iov
;
120 struct iovec iov_copy
[iovcnt
+ 1];
121 struct virtio_net_hdr hdr
= { 0, };
123 if (s
->has_vnet_hdr
&& !s
->using_vnet_hdr
) {
124 iov_copy
[0].iov_base
= &hdr
;
125 iov_copy
[0].iov_len
= sizeof(hdr
);
126 memcpy(&iov_copy
[1], iov
, iovcnt
* sizeof(*iov
));
131 return tap_write_packet(s
, iovp
, iovcnt
);
134 static ssize_t
tap_receive_raw(VLANClientState
*nc
, const uint8_t *buf
, size_t size
)
136 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
139 struct virtio_net_hdr hdr
= { 0, };
141 if (s
->has_vnet_hdr
) {
142 iov
[iovcnt
].iov_base
= &hdr
;
143 iov
[iovcnt
].iov_len
= sizeof(hdr
);
147 iov
[iovcnt
].iov_base
= (char *)buf
;
148 iov
[iovcnt
].iov_len
= size
;
151 return tap_write_packet(s
, iov
, iovcnt
);
154 static ssize_t
tap_receive(VLANClientState
*nc
, const uint8_t *buf
, size_t size
)
156 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
159 if (s
->has_vnet_hdr
&& !s
->using_vnet_hdr
) {
160 return tap_receive_raw(nc
, buf
, size
);
163 iov
[0].iov_base
= (char *)buf
;
164 iov
[0].iov_len
= size
;
166 return tap_write_packet(s
, iov
, 1);
169 static int tap_can_send(void *opaque
)
171 TAPState
*s
= opaque
;
173 return qemu_can_send_packet(&s
->nc
);
177 ssize_t
tap_read_packet(int tapfd
, uint8_t *buf
, int maxlen
)
179 return read(tapfd
, buf
, maxlen
);
183 static void tap_send_completed(VLANClientState
*nc
, ssize_t len
)
185 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
189 static void tap_send(void *opaque
)
191 TAPState
*s
= opaque
;
195 uint8_t *buf
= s
->buf
;
197 size
= tap_read_packet(s
->fd
, s
->buf
, sizeof(s
->buf
));
202 if (s
->has_vnet_hdr
&& !s
->using_vnet_hdr
) {
203 buf
+= sizeof(struct virtio_net_hdr
);
204 size
-= sizeof(struct virtio_net_hdr
);
207 size
= qemu_send_packet_async(&s
->nc
, buf
, size
, tap_send_completed
);
211 } while (size
> 0 && qemu_can_send_packet(&s
->nc
));
214 int tap_has_ufo(VLANClientState
*nc
)
216 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
218 assert(nc
->info
->type
== NET_CLIENT_TYPE_TAP
);
223 int tap_has_vnet_hdr(VLANClientState
*nc
)
225 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
227 assert(nc
->info
->type
== NET_CLIENT_TYPE_TAP
);
229 return s
->has_vnet_hdr
;
232 void tap_using_vnet_hdr(VLANClientState
*nc
, int using_vnet_hdr
)
234 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
236 using_vnet_hdr
= using_vnet_hdr
!= 0;
238 assert(nc
->info
->type
== NET_CLIENT_TYPE_TAP
);
239 assert(s
->has_vnet_hdr
== using_vnet_hdr
);
241 s
->using_vnet_hdr
= using_vnet_hdr
;
244 void tap_set_offload(VLANClientState
*nc
, int csum
, int tso4
,
245 int tso6
, int ecn
, int ufo
)
247 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
249 return tap_fd_set_offload(s
->fd
, csum
, tso4
, tso6
, ecn
, ufo
);
252 static void tap_cleanup(VLANClientState
*nc
)
254 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
256 qemu_purge_queued_packets(nc
);
258 if (s
->down_script
[0])
259 launch_script(s
->down_script
, s
->down_script_arg
, s
->fd
);
262 tap_write_poll(s
, 0);
266 static void tap_poll(VLANClientState
*nc
, bool enable
)
268 TAPState
*s
= DO_UPCAST(TAPState
, nc
, nc
);
269 tap_read_poll(s
, enable
);
270 tap_write_poll(s
, enable
);
275 static NetClientInfo net_tap_info
= {
276 .type
= NET_CLIENT_TYPE_TAP
,
277 .size
= sizeof(TAPState
),
278 .receive
= tap_receive
,
279 .receive_raw
= tap_receive_raw
,
280 .receive_iov
= tap_receive_iov
,
282 .cleanup
= tap_cleanup
,
285 static TAPState
*net_tap_fd_init(VLANState
*vlan
,
294 nc
= qemu_new_net_client(&net_tap_info
, vlan
, NULL
, model
, name
);
296 s
= DO_UPCAST(TAPState
, nc
, nc
);
299 s
->has_vnet_hdr
= vnet_hdr
!= 0;
300 s
->using_vnet_hdr
= 0;
301 s
->has_ufo
= tap_probe_has_ufo(s
->fd
);
302 tap_set_offload(&s
->nc
, 0, 0, 0, 0, 0);
307 static int launch_script(const char *setup_script
, const char *ifname
, int fd
)
309 sigset_t oldmask
, mask
;
315 sigaddset(&mask
, SIGCHLD
);
316 sigprocmask(SIG_BLOCK
, &mask
, &oldmask
);
318 /* try to launch network script */
321 int open_max
= sysconf(_SC_OPEN_MAX
), i
;
323 for (i
= 0; i
< open_max
; i
++) {
324 if (i
!= STDIN_FILENO
&&
325 i
!= STDOUT_FILENO
&&
326 i
!= STDERR_FILENO
&&
332 *parg
++ = (char *)setup_script
;
333 *parg
++ = (char *)ifname
;
335 execv(setup_script
, args
);
337 } else if (pid
> 0) {
338 while (waitpid(pid
, &status
, 0) != pid
) {
341 sigprocmask(SIG_SETMASK
, &oldmask
, NULL
);
343 if (WIFEXITED(status
) && WEXITSTATUS(status
) == 0) {
347 fprintf(stderr
, "%s: could not launch network script\n", setup_script
);
351 static int net_tap_init(QemuOpts
*opts
, int *vnet_hdr
)
353 int fd
, vnet_hdr_required
;
354 char ifname
[128] = {0,};
355 const char *setup_script
;
357 if (qemu_opt_get(opts
, "ifname")) {
358 pstrcpy(ifname
, sizeof(ifname
), qemu_opt_get(opts
, "ifname"));
361 *vnet_hdr
= qemu_opt_get_bool(opts
, "vnet_hdr", 1);
362 if (qemu_opt_get(opts
, "vnet_hdr")) {
363 vnet_hdr_required
= *vnet_hdr
;
365 vnet_hdr_required
= 0;
368 TFR(fd
= tap_open(ifname
, sizeof(ifname
), vnet_hdr
, vnet_hdr_required
));
373 setup_script
= qemu_opt_get(opts
, "script");
375 setup_script
[0] != '\0' &&
376 strcmp(setup_script
, "no") != 0 &&
377 launch_script(setup_script
, ifname
, fd
)) {
382 qemu_opt_set(opts
, "ifname", ifname
);
387 int net_init_tap(QemuOpts
*opts
, Monitor
*mon
, const char *name
, VLANState
*vlan
)
390 int fd
, vnet_hdr
= 0;
392 if (qemu_opt_get(opts
, "fd")) {
393 if (qemu_opt_get(opts
, "ifname") ||
394 qemu_opt_get(opts
, "script") ||
395 qemu_opt_get(opts
, "downscript") ||
396 qemu_opt_get(opts
, "vnet_hdr")) {
397 qemu_error("ifname=, script=, downscript= and vnet_hdr= is invalid with fd=\n");
401 fd
= net_handle_fd_param(mon
, qemu_opt_get(opts
, "fd"));
406 fcntl(fd
, F_SETFL
, O_NONBLOCK
);
408 vnet_hdr
= tap_probe_vnet_hdr(fd
);
410 if (!qemu_opt_get(opts
, "script")) {
411 qemu_opt_set(opts
, "script", DEFAULT_NETWORK_SCRIPT
);
414 if (!qemu_opt_get(opts
, "downscript")) {
415 qemu_opt_set(opts
, "downscript", DEFAULT_NETWORK_DOWN_SCRIPT
);
418 fd
= net_tap_init(opts
, &vnet_hdr
);
424 s
= net_tap_fd_init(vlan
, "tap", name
, fd
, vnet_hdr
);
430 if (tap_set_sndbuf(s
->fd
, opts
) < 0) {
434 if (qemu_opt_get(opts
, "fd")) {
435 snprintf(s
->nc
.info_str
, sizeof(s
->nc
.info_str
), "fd=%d", fd
);
437 const char *ifname
, *script
, *downscript
;
439 ifname
= qemu_opt_get(opts
, "ifname");
440 script
= qemu_opt_get(opts
, "script");
441 downscript
= qemu_opt_get(opts
, "downscript");
443 snprintf(s
->nc
.info_str
, sizeof(s
->nc
.info_str
),
444 "ifname=%s,script=%s,downscript=%s",
445 ifname
, script
, downscript
);
447 if (strcmp(downscript
, "no") != 0) {
448 snprintf(s
->down_script
, sizeof(s
->down_script
), "%s", downscript
);
449 snprintf(s
->down_script_arg
, sizeof(s
->down_script_arg
), "%s", ifname
);