4 * Copyright (c) 2004-2008 Fabrice Bellard
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to deal
8 * in the Software without restriction, including without limitation the rights
9 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10 * copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
24 #include "qemu/osdep.h"
25 #include "qemu-common.h"
26 #include "qemu/timer.h"
27 #include "qemu/error-report.h"
28 #include "chardev/char-fe.h"
29 #include "migration/register.h"
32 #include "qemu/cutils.h"
38 /* host loopback address */
39 struct in_addr loopback_addr
;
40 /* host loopback network mask */
41 unsigned long loopback_mask
;
43 /* emulated hosts use the MAC addr 52:55:IP:IP:IP:IP */
44 static const uint8_t special_ethaddr
[ETH_ALEN
] = {
45 0x52, 0x55, 0x00, 0x00, 0x00, 0x00
50 static QTAILQ_HEAD(slirp_instances
, Slirp
) slirp_instances
=
51 QTAILQ_HEAD_INITIALIZER(slirp_instances
);
53 static struct in_addr dns_addr
;
55 static struct in6_addr dns6_addr
;
57 static u_int dns_addr_time
;
59 static u_int dns6_addr_time
;
62 #define TIMEOUT_FAST 2 /* milliseconds */
63 #define TIMEOUT_SLOW 499 /* milliseconds */
64 /* for the aging of certain requests like DNS */
65 #define TIMEOUT_DEFAULT 1000 /* milliseconds */
69 int get_dns_addr(struct in_addr
*pdns_addr
)
71 FIXED_INFO
*FixedInfo
=NULL
;
74 IP_ADDR_STRING
*pIPAddr
;
75 struct in_addr tmp_addr
;
77 if (dns_addr
.s_addr
!= 0 && (curtime
- dns_addr_time
) < TIMEOUT_DEFAULT
) {
78 *pdns_addr
= dns_addr
;
82 FixedInfo
= (FIXED_INFO
*)GlobalAlloc(GPTR
, sizeof(FIXED_INFO
));
83 BufLen
= sizeof(FIXED_INFO
);
85 if (ERROR_BUFFER_OVERFLOW
== GetNetworkParams(FixedInfo
, &BufLen
)) {
87 GlobalFree(FixedInfo
);
90 FixedInfo
= GlobalAlloc(GPTR
, BufLen
);
93 if ((ret
= GetNetworkParams(FixedInfo
, &BufLen
)) != ERROR_SUCCESS
) {
94 printf("GetNetworkParams failed. ret = %08x\n", (u_int
)ret
);
96 GlobalFree(FixedInfo
);
102 pIPAddr
= &(FixedInfo
->DnsServerList
);
103 inet_aton(pIPAddr
->IpAddress
.String
, &tmp_addr
);
104 *pdns_addr
= tmp_addr
;
106 dns_addr_time
= curtime
;
108 GlobalFree(FixedInfo
);
114 int get_dns6_addr(struct in6_addr
*pdns6_addr
, uint32_t *scope_id
)
119 static void winsock_cleanup(void)
126 static int get_dns_addr_cached(void *pdns_addr
, void *cached_addr
,
128 struct stat
*cached_stat
, u_int
*cached_time
)
130 struct stat old_stat
;
131 if (curtime
- *cached_time
< TIMEOUT_DEFAULT
) {
132 memcpy(pdns_addr
, cached_addr
, addrlen
);
135 old_stat
= *cached_stat
;
136 if (stat("/etc/resolv.conf", cached_stat
) != 0) {
139 if (cached_stat
->st_dev
== old_stat
.st_dev
140 && cached_stat
->st_ino
== old_stat
.st_ino
141 && cached_stat
->st_size
== old_stat
.st_size
142 && cached_stat
->st_mtime
== old_stat
.st_mtime
) {
143 memcpy(pdns_addr
, cached_addr
, addrlen
);
149 static int get_dns_addr_resolv_conf(int af
, void *pdns_addr
, void *cached_addr
,
150 socklen_t addrlen
, uint32_t *scope_id
,
157 void *tmp_addr
= alloca(addrlen
);
160 f
= fopen("/etc/resolv.conf", "r");
165 fprintf(stderr
, "IP address of your DNS(s): ");
167 while (fgets(buff
, 512, f
) != NULL
) {
168 if (sscanf(buff
, "nameserver%*[ \t]%256s", buff2
) == 1) {
169 char *c
= strchr(buff2
, '%');
171 if_index
= if_nametoindex(c
+ 1);
177 if (!inet_pton(af
, buff2
, tmp_addr
)) {
180 /* If it's the first one, set it to dns_addr */
182 memcpy(pdns_addr
, tmp_addr
, addrlen
);
183 memcpy(cached_addr
, tmp_addr
, addrlen
);
185 *scope_id
= if_index
;
187 *cached_time
= curtime
;
191 fprintf(stderr
, ", ");
195 fprintf(stderr
, "(more)");
201 char s
[INET6_ADDRSTRLEN
];
202 const char *res
= inet_ntop(af
, tmp_addr
, s
, sizeof(s
));
204 res
= "(string conversion error)";
206 fprintf(stderr
, "%s", res
);
217 int get_dns_addr(struct in_addr
*pdns_addr
)
219 static struct stat dns_addr_stat
;
221 if (dns_addr
.s_addr
!= 0) {
223 ret
= get_dns_addr_cached(pdns_addr
, &dns_addr
, sizeof(dns_addr
),
224 &dns_addr_stat
, &dns_addr_time
);
229 return get_dns_addr_resolv_conf(AF_INET
, pdns_addr
, &dns_addr
,
230 sizeof(dns_addr
), NULL
, &dns_addr_time
);
233 int get_dns6_addr(struct in6_addr
*pdns6_addr
, uint32_t *scope_id
)
235 static struct stat dns6_addr_stat
;
237 if (!in6_zero(&dns6_addr
)) {
239 ret
= get_dns_addr_cached(pdns6_addr
, &dns6_addr
, sizeof(dns6_addr
),
240 &dns6_addr_stat
, &dns6_addr_time
);
245 return get_dns_addr_resolv_conf(AF_INET6
, pdns6_addr
, &dns6_addr
,
247 scope_id
, &dns6_addr_time
);
252 static void slirp_init_once(void)
254 static int initialized
;
265 WSAStartup(MAKEWORD(2,0), &Data
);
266 atexit(winsock_cleanup
);
269 loopback_addr
.s_addr
= htonl(INADDR_LOOPBACK
);
270 loopback_mask
= htonl(IN_CLASSA_NET
);
273 static void slirp_state_save(QEMUFile
*f
, void *opaque
);
274 static int slirp_state_load(QEMUFile
*f
, void *opaque
, int version_id
);
276 static SaveVMHandlers savevm_slirp_state
= {
277 .save_state
= slirp_state_save
,
278 .load_state
= slirp_state_load
,
281 Slirp
*slirp_init(int restricted
, bool in_enabled
, struct in_addr vnetwork
,
282 struct in_addr vnetmask
, struct in_addr vhost
,
284 struct in6_addr vprefix_addr6
, uint8_t vprefix_len
,
285 struct in6_addr vhost6
, const char *vhostname
,
286 const char *tftp_path
, const char *bootfile
,
287 struct in_addr vdhcp_start
, struct in_addr vnameserver
,
288 struct in6_addr vnameserver6
, const char **vdnssearch
,
291 Slirp
*slirp
= g_malloc0(sizeof(Slirp
));
295 slirp
->grand
= g_rand_new();
296 slirp
->restricted
= restricted
;
298 slirp
->in_enabled
= in_enabled
;
299 slirp
->in6_enabled
= in6_enabled
;
305 /* Initialise mbufs *after* setting the MTU */
308 slirp
->vnetwork_addr
= vnetwork
;
309 slirp
->vnetwork_mask
= vnetmask
;
310 slirp
->vhost_addr
= vhost
;
311 slirp
->vprefix_addr6
= vprefix_addr6
;
312 slirp
->vprefix_len
= vprefix_len
;
313 slirp
->vhost_addr6
= vhost6
;
315 pstrcpy(slirp
->client_hostname
, sizeof(slirp
->client_hostname
),
318 slirp
->tftp_prefix
= g_strdup(tftp_path
);
319 slirp
->bootp_filename
= g_strdup(bootfile
);
320 slirp
->vdhcp_startaddr
= vdhcp_start
;
321 slirp
->vnameserver_addr
= vnameserver
;
322 slirp
->vnameserver_addr6
= vnameserver6
;
325 translate_dnssearch(slirp
, vdnssearch
);
328 slirp
->opaque
= opaque
;
330 register_savevm_live(NULL
, "slirp", 0, 4, &savevm_slirp_state
, slirp
);
332 QTAILQ_INSERT_TAIL(&slirp_instances
, slirp
, entry
);
337 void slirp_cleanup(Slirp
*slirp
)
339 QTAILQ_REMOVE(&slirp_instances
, slirp
, entry
);
341 unregister_savevm(NULL
, "slirp", slirp
);
347 g_rand_free(slirp
->grand
);
349 g_free(slirp
->vdnssearch
);
350 g_free(slirp
->tftp_prefix
);
351 g_free(slirp
->bootp_filename
);
355 #define CONN_CANFSEND(so) (((so)->so_state & (SS_FCANTSENDMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
356 #define CONN_CANFRCV(so) (((so)->so_state & (SS_FCANTRCVMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
358 static void slirp_update_timeout(uint32_t *timeout
)
363 if (*timeout
<= TIMEOUT_FAST
) {
367 t
= MIN(1000, *timeout
);
369 /* If we have tcp timeout with slirp, then we will fill @timeout with
370 * more precise value.
372 QTAILQ_FOREACH(slirp
, &slirp_instances
, entry
) {
373 if (slirp
->time_fasttimo
) {
374 *timeout
= TIMEOUT_FAST
;
377 if (slirp
->do_slowtimo
) {
378 t
= MIN(TIMEOUT_SLOW
, t
);
384 void slirp_pollfds_fill(GArray
*pollfds
, uint32_t *timeout
)
387 struct socket
*so
, *so_next
;
389 if (QTAILQ_EMPTY(&slirp_instances
)) {
397 QTAILQ_FOREACH(slirp
, &slirp_instances
, entry
) {
399 * *_slowtimo needs calling if there are IP fragments
400 * in the fragment queue, or there are TCP connections active
402 slirp
->do_slowtimo
= ((slirp
->tcb
.so_next
!= &slirp
->tcb
) ||
403 (&slirp
->ipq
.ip_link
!= slirp
->ipq
.ip_link
.next
));
405 for (so
= slirp
->tcb
.so_next
; so
!= &slirp
->tcb
;
409 so_next
= so
->so_next
;
411 so
->pollfds_idx
= -1;
414 * See if we need a tcp_fasttimo
416 if (slirp
->time_fasttimo
== 0 &&
417 so
->so_tcpcb
->t_flags
& TF_DELACK
) {
418 slirp
->time_fasttimo
= curtime
; /* Flag when want a fasttimo */
422 * NOFDREF can include still connecting to local-host,
423 * newly socreated() sockets etc. Don't want to select these.
425 if (so
->so_state
& SS_NOFDREF
|| so
->s
== -1) {
430 * Set for reading sockets which are accepting
432 if (so
->so_state
& SS_FACCEPTCONN
) {
435 .events
= G_IO_IN
| G_IO_HUP
| G_IO_ERR
,
437 so
->pollfds_idx
= pollfds
->len
;
438 g_array_append_val(pollfds
, pfd
);
443 * Set for writing sockets which are connecting
445 if (so
->so_state
& SS_ISFCONNECTING
) {
448 .events
= G_IO_OUT
| G_IO_ERR
,
450 so
->pollfds_idx
= pollfds
->len
;
451 g_array_append_val(pollfds
, pfd
);
456 * Set for writing if we are connected, can send more, and
457 * we have something to send
459 if (CONN_CANFSEND(so
) && so
->so_rcv
.sb_cc
) {
460 events
|= G_IO_OUT
| G_IO_ERR
;
464 * Set for reading (and urgent data) if we are connected, can
465 * receive more, and we have room for it XXX /2 ?
467 if (CONN_CANFRCV(so
) &&
468 (so
->so_snd
.sb_cc
< (so
->so_snd
.sb_datalen
/2))) {
469 events
|= G_IO_IN
| G_IO_HUP
| G_IO_ERR
| G_IO_PRI
;
477 so
->pollfds_idx
= pollfds
->len
;
478 g_array_append_val(pollfds
, pfd
);
485 for (so
= slirp
->udb
.so_next
; so
!= &slirp
->udb
;
487 so_next
= so
->so_next
;
489 so
->pollfds_idx
= -1;
492 * See if it's timed out
495 if (so
->so_expire
<= curtime
) {
499 slirp
->do_slowtimo
= true; /* Let socket expire */
504 * When UDP packets are received from over the
505 * link, they're sendto()'d straight away, so
506 * no need for setting for writing
507 * Limit the number of packets queued by this session
508 * to 4. Note that even though we try and limit this
509 * to 4 packets, the session could have more queued
510 * if the packets needed to be fragmented
513 if ((so
->so_state
& SS_ISFCONNECTED
) && so
->so_queued
<= 4) {
516 .events
= G_IO_IN
| G_IO_HUP
| G_IO_ERR
,
518 so
->pollfds_idx
= pollfds
->len
;
519 g_array_append_val(pollfds
, pfd
);
526 for (so
= slirp
->icmp
.so_next
; so
!= &slirp
->icmp
;
528 so_next
= so
->so_next
;
530 so
->pollfds_idx
= -1;
533 * See if it's timed out
536 if (so
->so_expire
<= curtime
) {
540 slirp
->do_slowtimo
= true; /* Let socket expire */
544 if (so
->so_state
& SS_ISFCONNECTED
) {
547 .events
= G_IO_IN
| G_IO_HUP
| G_IO_ERR
,
549 so
->pollfds_idx
= pollfds
->len
;
550 g_array_append_val(pollfds
, pfd
);
554 slirp_update_timeout(timeout
);
557 void slirp_pollfds_poll(GArray
*pollfds
, int select_error
)
560 struct socket
*so
, *so_next
;
563 if (QTAILQ_EMPTY(&slirp_instances
)) {
567 curtime
= qemu_clock_get_ms(QEMU_CLOCK_REALTIME
);
569 QTAILQ_FOREACH(slirp
, &slirp_instances
, entry
) {
571 * See if anything has timed out
573 if (slirp
->time_fasttimo
&&
574 ((curtime
- slirp
->time_fasttimo
) >= TIMEOUT_FAST
)) {
576 slirp
->time_fasttimo
= 0;
578 if (slirp
->do_slowtimo
&&
579 ((curtime
- slirp
->last_slowtimo
) >= TIMEOUT_SLOW
)) {
582 slirp
->last_slowtimo
= curtime
;
592 for (so
= slirp
->tcb
.so_next
; so
!= &slirp
->tcb
;
596 so_next
= so
->so_next
;
599 if (so
->pollfds_idx
!= -1) {
600 revents
= g_array_index(pollfds
, GPollFD
,
601 so
->pollfds_idx
).revents
;
604 if (so
->so_state
& SS_NOFDREF
|| so
->s
== -1) {
610 * This will soread as well, so no need to
611 * test for G_IO_IN below if this succeeds
613 if (revents
& G_IO_PRI
) {
616 /* Socket error might have resulted in the socket being
617 * removed, do not try to do anything more with it. */
622 * Check sockets for reading
624 else if (revents
& (G_IO_IN
| G_IO_HUP
| G_IO_ERR
)) {
626 * Check for incoming connections
628 if (so
->so_state
& SS_FACCEPTCONN
) {
634 /* Output it if we read something */
636 tcp_output(sototcpcb(so
));
639 /* Socket error might have resulted in the socket being
640 * removed, do not try to do anything more with it. */
646 * Check sockets for writing
648 if (!(so
->so_state
& SS_NOFDREF
) &&
649 (revents
& (G_IO_OUT
| G_IO_ERR
))) {
651 * Check for non-blocking, still-connecting sockets
653 if (so
->so_state
& SS_ISFCONNECTING
) {
655 so
->so_state
&= ~SS_ISFCONNECTING
;
657 ret
= send(so
->s
, (const void *) &ret
, 0, 0);
659 /* XXXXX Must fix, zero bytes is a NOP */
660 if (errno
== EAGAIN
|| errno
== EWOULDBLOCK
||
661 errno
== EINPROGRESS
|| errno
== ENOTCONN
) {
666 so
->so_state
&= SS_PERSISTENT_MASK
;
667 so
->so_state
|= SS_NOFDREF
;
669 /* else so->so_state &= ~SS_ISFCONNECTING; */
674 tcp_input((struct mbuf
*)NULL
, sizeof(struct ip
), so
,
681 * XXXXX If we wrote something (a lot), there
682 * could be a need for a window update.
683 * In the worst case, the remote will send
684 * a window probe to get things going again
689 * Probe a still-connecting, non-blocking socket
690 * to check if it's still alive
693 if (so
->so_state
& SS_ISFCONNECTING
) {
694 ret
= qemu_recv(so
->s
, &ret
, 0, 0);
698 if (errno
== EAGAIN
|| errno
== EWOULDBLOCK
||
699 errno
== EINPROGRESS
|| errno
== ENOTCONN
) {
700 continue; /* Still connecting, continue */
704 so
->so_state
&= SS_PERSISTENT_MASK
;
705 so
->so_state
|= SS_NOFDREF
;
707 /* tcp_input will take care of it */
709 ret
= send(so
->s
, &ret
, 0, 0);
712 if (errno
== EAGAIN
|| errno
== EWOULDBLOCK
||
713 errno
== EINPROGRESS
|| errno
== ENOTCONN
) {
717 so
->so_state
&= SS_PERSISTENT_MASK
;
718 so
->so_state
|= SS_NOFDREF
;
720 so
->so_state
&= ~SS_ISFCONNECTING
;
724 tcp_input((struct mbuf
*)NULL
, sizeof(struct ip
), so
,
726 } /* SS_ISFCONNECTING */
732 * Incoming packets are sent straight away, they're not buffered.
733 * Incoming UDP data isn't buffered either.
735 for (so
= slirp
->udb
.so_next
; so
!= &slirp
->udb
;
739 so_next
= so
->so_next
;
742 if (so
->pollfds_idx
!= -1) {
743 revents
= g_array_index(pollfds
, GPollFD
,
744 so
->pollfds_idx
).revents
;
748 (revents
& (G_IO_IN
| G_IO_HUP
| G_IO_ERR
))) {
754 * Check incoming ICMP relies.
756 for (so
= slirp
->icmp
.so_next
; so
!= &slirp
->icmp
;
760 so_next
= so
->so_next
;
763 if (so
->pollfds_idx
!= -1) {
764 revents
= g_array_index(pollfds
, GPollFD
,
765 so
->pollfds_idx
).revents
;
769 (revents
& (G_IO_IN
| G_IO_HUP
| G_IO_ERR
))) {
779 static void arp_input(Slirp
*slirp
, const uint8_t *pkt
, int pkt_len
)
781 struct slirp_arphdr
*ah
= (struct slirp_arphdr
*)(pkt
+ ETH_HLEN
);
782 uint8_t arp_reply
[MAX(ETH_HLEN
+ sizeof(struct slirp_arphdr
), 64)];
783 struct ethhdr
*reh
= (struct ethhdr
*)arp_reply
;
784 struct slirp_arphdr
*rah
= (struct slirp_arphdr
*)(arp_reply
+ ETH_HLEN
);
786 struct ex_list
*ex_ptr
;
788 if (!slirp
->in_enabled
) {
792 ar_op
= ntohs(ah
->ar_op
);
795 if (ah
->ar_tip
== ah
->ar_sip
) {
797 arp_table_add(slirp
, ah
->ar_sip
, ah
->ar_sha
);
801 if ((ah
->ar_tip
& slirp
->vnetwork_mask
.s_addr
) ==
802 slirp
->vnetwork_addr
.s_addr
) {
803 if (ah
->ar_tip
== slirp
->vnameserver_addr
.s_addr
||
804 ah
->ar_tip
== slirp
->vhost_addr
.s_addr
)
806 for (ex_ptr
= slirp
->exec_list
; ex_ptr
; ex_ptr
= ex_ptr
->ex_next
) {
807 if (ex_ptr
->ex_addr
.s_addr
== ah
->ar_tip
)
812 memset(arp_reply
, 0, sizeof(arp_reply
));
814 arp_table_add(slirp
, ah
->ar_sip
, ah
->ar_sha
);
816 /* ARP request for alias/dns mac address */
817 memcpy(reh
->h_dest
, pkt
+ ETH_ALEN
, ETH_ALEN
);
818 memcpy(reh
->h_source
, special_ethaddr
, ETH_ALEN
- 4);
819 memcpy(&reh
->h_source
[2], &ah
->ar_tip
, 4);
820 reh
->h_proto
= htons(ETH_P_ARP
);
822 rah
->ar_hrd
= htons(1);
823 rah
->ar_pro
= htons(ETH_P_IP
);
824 rah
->ar_hln
= ETH_ALEN
;
826 rah
->ar_op
= htons(ARPOP_REPLY
);
827 memcpy(rah
->ar_sha
, reh
->h_source
, ETH_ALEN
);
828 rah
->ar_sip
= ah
->ar_tip
;
829 memcpy(rah
->ar_tha
, ah
->ar_sha
, ETH_ALEN
);
830 rah
->ar_tip
= ah
->ar_sip
;
831 slirp_output(slirp
->opaque
, arp_reply
, sizeof(arp_reply
));
835 arp_table_add(slirp
, ah
->ar_sip
, ah
->ar_sha
);
842 void slirp_input(Slirp
*slirp
, const uint8_t *pkt
, int pkt_len
)
847 if (pkt_len
< ETH_HLEN
)
850 proto
= ntohs(*(uint16_t *)(pkt
+ 12));
853 arp_input(slirp
, pkt
, pkt_len
);
860 /* Note: we add 2 to align the IP header on 4 bytes,
861 * and add the margin for the tcpiphdr overhead */
862 if (M_FREEROOM(m
) < pkt_len
+ TCPIPHDR_DELTA
+ 2) {
863 m_inc(m
, pkt_len
+ TCPIPHDR_DELTA
+ 2);
865 m
->m_len
= pkt_len
+ TCPIPHDR_DELTA
+ 2;
866 memcpy(m
->m_data
+ TCPIPHDR_DELTA
+ 2, pkt
, pkt_len
);
868 m
->m_data
+= TCPIPHDR_DELTA
+ 2 + ETH_HLEN
;
869 m
->m_len
-= TCPIPHDR_DELTA
+ 2 + ETH_HLEN
;
871 if (proto
== ETH_P_IP
) {
873 } else if (proto
== ETH_P_IPV6
) {
879 ncsi_input(slirp
, pkt
, pkt_len
);
887 /* Prepare the IPv4 packet to be sent to the ethernet device. Returns 1 if no
888 * packet should be sent, 0 if the packet must be re-queued, 2 if the packet
891 static int if_encap4(Slirp
*slirp
, struct mbuf
*ifm
, struct ethhdr
*eh
,
892 uint8_t ethaddr
[ETH_ALEN
])
894 const struct ip
*iph
= (const struct ip
*)ifm
->m_data
;
896 if (iph
->ip_dst
.s_addr
== 0) {
897 /* 0.0.0.0 can not be a destination address, something went wrong,
898 * avoid making it worse */
901 if (!arp_table_search(slirp
, iph
->ip_dst
.s_addr
, ethaddr
)) {
902 uint8_t arp_req
[ETH_HLEN
+ sizeof(struct slirp_arphdr
)];
903 struct ethhdr
*reh
= (struct ethhdr
*)arp_req
;
904 struct slirp_arphdr
*rah
= (struct slirp_arphdr
*)(arp_req
+ ETH_HLEN
);
906 if (!ifm
->resolution_requested
) {
907 /* If the client addr is not known, send an ARP request */
908 memset(reh
->h_dest
, 0xff, ETH_ALEN
);
909 memcpy(reh
->h_source
, special_ethaddr
, ETH_ALEN
- 4);
910 memcpy(&reh
->h_source
[2], &slirp
->vhost_addr
, 4);
911 reh
->h_proto
= htons(ETH_P_ARP
);
912 rah
->ar_hrd
= htons(1);
913 rah
->ar_pro
= htons(ETH_P_IP
);
914 rah
->ar_hln
= ETH_ALEN
;
916 rah
->ar_op
= htons(ARPOP_REQUEST
);
919 memcpy(rah
->ar_sha
, special_ethaddr
, ETH_ALEN
- 4);
920 memcpy(&rah
->ar_sha
[2], &slirp
->vhost_addr
, 4);
923 rah
->ar_sip
= slirp
->vhost_addr
.s_addr
;
925 /* target hw addr (none) */
926 memset(rah
->ar_tha
, 0, ETH_ALEN
);
929 rah
->ar_tip
= iph
->ip_dst
.s_addr
;
930 slirp
->client_ipaddr
= iph
->ip_dst
;
931 slirp_output(slirp
->opaque
, arp_req
, sizeof(arp_req
));
932 ifm
->resolution_requested
= true;
934 /* Expire request and drop outgoing packet after 1 second */
935 ifm
->expiration_date
= qemu_clock_get_ns(QEMU_CLOCK_REALTIME
) + 1000000000ULL;
939 memcpy(eh
->h_source
, special_ethaddr
, ETH_ALEN
- 4);
940 /* XXX: not correct */
941 memcpy(&eh
->h_source
[2], &slirp
->vhost_addr
, 4);
942 eh
->h_proto
= htons(ETH_P_IP
);
949 /* Prepare the IPv6 packet to be sent to the ethernet device. Returns 1 if no
950 * packet should be sent, 0 if the packet must be re-queued, 2 if the packet
953 static int if_encap6(Slirp
*slirp
, struct mbuf
*ifm
, struct ethhdr
*eh
,
954 uint8_t ethaddr
[ETH_ALEN
])
956 const struct ip6
*ip6h
= mtod(ifm
, const struct ip6
*);
957 if (!ndp_table_search(slirp
, ip6h
->ip_dst
, ethaddr
)) {
958 if (!ifm
->resolution_requested
) {
959 ndp_send_ns(slirp
, ip6h
->ip_dst
);
960 ifm
->resolution_requested
= true;
961 ifm
->expiration_date
=
962 qemu_clock_get_ns(QEMU_CLOCK_REALTIME
) + 1000000000ULL;
966 eh
->h_proto
= htons(ETH_P_IPV6
);
967 in6_compute_ethaddr(ip6h
->ip_src
, eh
->h_source
);
974 /* Output the IP packet to the ethernet device. Returns 0 if the packet must be
977 int if_encap(Slirp
*slirp
, struct mbuf
*ifm
)
980 struct ethhdr
*eh
= (struct ethhdr
*)buf
;
981 uint8_t ethaddr
[ETH_ALEN
];
982 const struct ip
*iph
= (const struct ip
*)ifm
->m_data
;
985 if (ifm
->m_len
+ ETH_HLEN
> sizeof(buf
)) {
991 ret
= if_encap4(slirp
, ifm
, eh
, ethaddr
);
998 ret
= if_encap6(slirp
, ifm
, eh
, ethaddr
);
1005 g_assert_not_reached();
1009 memcpy(eh
->h_dest
, ethaddr
, ETH_ALEN
);
1010 DEBUG_ARGS((dfd
, " src = %02x:%02x:%02x:%02x:%02x:%02x\n",
1011 eh
->h_source
[0], eh
->h_source
[1], eh
->h_source
[2],
1012 eh
->h_source
[3], eh
->h_source
[4], eh
->h_source
[5]));
1013 DEBUG_ARGS((dfd
, " dst = %02x:%02x:%02x:%02x:%02x:%02x\n",
1014 eh
->h_dest
[0], eh
->h_dest
[1], eh
->h_dest
[2],
1015 eh
->h_dest
[3], eh
->h_dest
[4], eh
->h_dest
[5]));
1016 memcpy(buf
+ sizeof(struct ethhdr
), ifm
->m_data
, ifm
->m_len
);
1017 slirp_output(slirp
->opaque
, buf
, ifm
->m_len
+ ETH_HLEN
);
1021 /* Drop host forwarding rule, return 0 if found. */
1022 int slirp_remove_hostfwd(Slirp
*slirp
, int is_udp
, struct in_addr host_addr
,
1026 struct socket
*head
= (is_udp
? &slirp
->udb
: &slirp
->tcb
);
1027 struct sockaddr_in addr
;
1028 int port
= htons(host_port
);
1031 for (so
= head
->so_next
; so
!= head
; so
= so
->so_next
) {
1032 addr_len
= sizeof(addr
);
1033 if ((so
->so_state
& SS_HOSTFWD
) &&
1034 getsockname(so
->s
, (struct sockaddr
*)&addr
, &addr_len
) == 0 &&
1035 addr
.sin_addr
.s_addr
== host_addr
.s_addr
&&
1036 addr
.sin_port
== port
) {
1046 int slirp_add_hostfwd(Slirp
*slirp
, int is_udp
, struct in_addr host_addr
,
1047 int host_port
, struct in_addr guest_addr
, int guest_port
)
1049 if (!guest_addr
.s_addr
) {
1050 guest_addr
= slirp
->vdhcp_startaddr
;
1053 if (!udp_listen(slirp
, host_addr
.s_addr
, htons(host_port
),
1054 guest_addr
.s_addr
, htons(guest_port
), SS_HOSTFWD
))
1057 if (!tcp_listen(slirp
, host_addr
.s_addr
, htons(host_port
),
1058 guest_addr
.s_addr
, htons(guest_port
), SS_HOSTFWD
))
1064 int slirp_add_exec(Slirp
*slirp
, int do_pty
, const void *args
,
1065 struct in_addr
*guest_addr
, int guest_port
)
1067 if (!guest_addr
->s_addr
) {
1068 guest_addr
->s_addr
= slirp
->vnetwork_addr
.s_addr
|
1069 (htonl(0x0204) & ~slirp
->vnetwork_mask
.s_addr
);
1071 if ((guest_addr
->s_addr
& slirp
->vnetwork_mask
.s_addr
) !=
1072 slirp
->vnetwork_addr
.s_addr
||
1073 guest_addr
->s_addr
== slirp
->vhost_addr
.s_addr
||
1074 guest_addr
->s_addr
== slirp
->vnameserver_addr
.s_addr
) {
1077 return add_exec(&slirp
->exec_list
, do_pty
, (char *)args
, *guest_addr
,
1081 ssize_t
slirp_send(struct socket
*so
, const void *buf
, size_t len
, int flags
)
1083 if (so
->s
== -1 && so
->extra
) {
1084 /* XXX this blocks entire thread. Rewrite to use
1085 * qemu_chr_fe_write and background I/O callbacks */
1086 qemu_chr_fe_write_all(so
->extra
, buf
, len
);
1090 return send(so
->s
, buf
, len
, flags
);
1093 static struct socket
*
1094 slirp_find_ctl_socket(Slirp
*slirp
, struct in_addr guest_addr
, int guest_port
)
1098 for (so
= slirp
->tcb
.so_next
; so
!= &slirp
->tcb
; so
= so
->so_next
) {
1099 if (so
->so_faddr
.s_addr
== guest_addr
.s_addr
&&
1100 htons(so
->so_fport
) == guest_port
) {
1107 size_t slirp_socket_can_recv(Slirp
*slirp
, struct in_addr guest_addr
,
1110 struct iovec iov
[2];
1113 so
= slirp_find_ctl_socket(slirp
, guest_addr
, guest_port
);
1115 if (!so
|| so
->so_state
& SS_NOFDREF
) {
1119 if (!CONN_CANFRCV(so
) || so
->so_snd
.sb_cc
>= (so
->so_snd
.sb_datalen
/2)) {
1123 return sopreprbuf(so
, iov
, NULL
);
1126 void slirp_socket_recv(Slirp
*slirp
, struct in_addr guest_addr
, int guest_port
,
1127 const uint8_t *buf
, int size
)
1130 struct socket
*so
= slirp_find_ctl_socket(slirp
, guest_addr
, guest_port
);
1135 ret
= soreadbuf(so
, (const char *)buf
, size
);
1138 tcp_output(sototcpcb(so
));
1141 static int slirp_tcp_post_load(void *opaque
, int version
)
1143 tcp_template((struct tcpcb
*)opaque
);
1148 static const VMStateDescription vmstate_slirp_tcp
= {
1149 .name
= "slirp-tcp",
1151 .post_load
= slirp_tcp_post_load
,
1152 .fields
= (VMStateField
[]) {
1153 VMSTATE_INT16(t_state
, struct tcpcb
),
1154 VMSTATE_INT16_ARRAY(t_timer
, struct tcpcb
, TCPT_NTIMERS
),
1155 VMSTATE_INT16(t_rxtshift
, struct tcpcb
),
1156 VMSTATE_INT16(t_rxtcur
, struct tcpcb
),
1157 VMSTATE_INT16(t_dupacks
, struct tcpcb
),
1158 VMSTATE_UINT16(t_maxseg
, struct tcpcb
),
1159 VMSTATE_UINT8(t_force
, struct tcpcb
),
1160 VMSTATE_UINT16(t_flags
, struct tcpcb
),
1161 VMSTATE_UINT32(snd_una
, struct tcpcb
),
1162 VMSTATE_UINT32(snd_nxt
, struct tcpcb
),
1163 VMSTATE_UINT32(snd_up
, struct tcpcb
),
1164 VMSTATE_UINT32(snd_wl1
, struct tcpcb
),
1165 VMSTATE_UINT32(snd_wl2
, struct tcpcb
),
1166 VMSTATE_UINT32(iss
, struct tcpcb
),
1167 VMSTATE_UINT32(snd_wnd
, struct tcpcb
),
1168 VMSTATE_UINT32(rcv_wnd
, struct tcpcb
),
1169 VMSTATE_UINT32(rcv_nxt
, struct tcpcb
),
1170 VMSTATE_UINT32(rcv_up
, struct tcpcb
),
1171 VMSTATE_UINT32(irs
, struct tcpcb
),
1172 VMSTATE_UINT32(rcv_adv
, struct tcpcb
),
1173 VMSTATE_UINT32(snd_max
, struct tcpcb
),
1174 VMSTATE_UINT32(snd_cwnd
, struct tcpcb
),
1175 VMSTATE_UINT32(snd_ssthresh
, struct tcpcb
),
1176 VMSTATE_INT16(t_idle
, struct tcpcb
),
1177 VMSTATE_INT16(t_rtt
, struct tcpcb
),
1178 VMSTATE_UINT32(t_rtseq
, struct tcpcb
),
1179 VMSTATE_INT16(t_srtt
, struct tcpcb
),
1180 VMSTATE_INT16(t_rttvar
, struct tcpcb
),
1181 VMSTATE_UINT16(t_rttmin
, struct tcpcb
),
1182 VMSTATE_UINT32(max_sndwnd
, struct tcpcb
),
1183 VMSTATE_UINT8(t_oobflags
, struct tcpcb
),
1184 VMSTATE_UINT8(t_iobc
, struct tcpcb
),
1185 VMSTATE_INT16(t_softerror
, struct tcpcb
),
1186 VMSTATE_UINT8(snd_scale
, struct tcpcb
),
1187 VMSTATE_UINT8(rcv_scale
, struct tcpcb
),
1188 VMSTATE_UINT8(request_r_scale
, struct tcpcb
),
1189 VMSTATE_UINT8(requested_s_scale
, struct tcpcb
),
1190 VMSTATE_UINT32(ts_recent
, struct tcpcb
),
1191 VMSTATE_UINT32(ts_recent_age
, struct tcpcb
),
1192 VMSTATE_UINT32(last_ack_sent
, struct tcpcb
),
1193 VMSTATE_END_OF_LIST()
1197 /* The sbuf has a pair of pointers that are migrated as offsets;
1198 * we calculate the offsets and restore the pointers using
1199 * pre_save/post_load on a tmp structure.
1202 struct sbuf
*parent
;
1203 uint32_t roff
, woff
;
1206 static int sbuf_tmp_pre_save(void *opaque
)
1208 struct sbuf_tmp
*tmp
= opaque
;
1209 tmp
->woff
= tmp
->parent
->sb_wptr
- tmp
->parent
->sb_data
;
1210 tmp
->roff
= tmp
->parent
->sb_rptr
- tmp
->parent
->sb_data
;
1215 static int sbuf_tmp_post_load(void *opaque
, int version
)
1217 struct sbuf_tmp
*tmp
= opaque
;
1218 uint32_t requested_len
= tmp
->parent
->sb_datalen
;
1220 /* Allocate the buffer space used by the field after the tmp */
1221 sbreserve(tmp
->parent
, tmp
->parent
->sb_datalen
);
1223 if (tmp
->parent
->sb_datalen
!= requested_len
) {
1226 if (tmp
->woff
>= requested_len
||
1227 tmp
->roff
>= requested_len
) {
1228 error_report("invalid sbuf offsets r/w=%u/%u len=%u",
1229 tmp
->roff
, tmp
->woff
, requested_len
);
1233 tmp
->parent
->sb_wptr
= tmp
->parent
->sb_data
+ tmp
->woff
;
1234 tmp
->parent
->sb_rptr
= tmp
->parent
->sb_data
+ tmp
->roff
;
1240 static const VMStateDescription vmstate_slirp_sbuf_tmp
= {
1241 .name
= "slirp-sbuf-tmp",
1242 .post_load
= sbuf_tmp_post_load
,
1243 .pre_save
= sbuf_tmp_pre_save
,
1245 .fields
= (VMStateField
[]) {
1246 VMSTATE_UINT32(woff
, struct sbuf_tmp
),
1247 VMSTATE_UINT32(roff
, struct sbuf_tmp
),
1248 VMSTATE_END_OF_LIST()
1252 static const VMStateDescription vmstate_slirp_sbuf
= {
1253 .name
= "slirp-sbuf",
1255 .fields
= (VMStateField
[]) {
1256 VMSTATE_UINT32(sb_cc
, struct sbuf
),
1257 VMSTATE_UINT32(sb_datalen
, struct sbuf
),
1258 VMSTATE_WITH_TMP(struct sbuf
, struct sbuf_tmp
, vmstate_slirp_sbuf_tmp
),
1259 VMSTATE_VBUFFER_UINT32(sb_data
, struct sbuf
, 0, NULL
, sb_datalen
),
1260 VMSTATE_END_OF_LIST()
1264 static bool slirp_older_than_v4(void *opaque
, int version_id
)
1266 return version_id
< 4;
1269 static bool slirp_family_inet(void *opaque
, int version_id
)
1271 union slirp_sockaddr
*ssa
= (union slirp_sockaddr
*)opaque
;
1272 return ssa
->ss
.ss_family
== AF_INET
;
1275 static int slirp_socket_pre_load(void *opaque
)
1277 struct socket
*so
= opaque
;
1278 if (tcp_attach(so
) < 0) {
1281 /* Older versions don't load these fields */
1282 so
->so_ffamily
= AF_INET
;
1283 so
->so_lfamily
= AF_INET
;
1288 #define VMSTATE_SIN4_ADDR(f, s, t) VMSTATE_UINT32_TEST(f, s, t)
1290 /* Win uses u_long rather than uint32_t - but it's still 32bits long */
1291 #define VMSTATE_SIN4_ADDR(f, s, t) VMSTATE_SINGLE_TEST(f, s, t, 0, \
1292 vmstate_info_uint32, u_long)
1295 /* The OS provided ss_family field isn't that portable; it's size
1296 * and type varies (16/8 bit, signed, unsigned)
1297 * and the values it contains aren't fully portable.
1299 typedef struct SS_FamilyTmpStruct
{
1300 union slirp_sockaddr
*parent
;
1301 uint16_t portable_family
;
1302 } SS_FamilyTmpStruct
;
1304 #define SS_FAMILY_MIG_IPV4 2 /* Linux, BSD, Win... */
1305 #define SS_FAMILY_MIG_IPV6 10 /* Linux */
1306 #define SS_FAMILY_MIG_OTHER 0xffff
1308 static int ss_family_pre_save(void *opaque
)
1310 SS_FamilyTmpStruct
*tss
= opaque
;
1312 tss
->portable_family
= SS_FAMILY_MIG_OTHER
;
1314 if (tss
->parent
->ss
.ss_family
== AF_INET
) {
1315 tss
->portable_family
= SS_FAMILY_MIG_IPV4
;
1316 } else if (tss
->parent
->ss
.ss_family
== AF_INET6
) {
1317 tss
->portable_family
= SS_FAMILY_MIG_IPV6
;
1323 static int ss_family_post_load(void *opaque
, int version_id
)
1325 SS_FamilyTmpStruct
*tss
= opaque
;
1327 switch (tss
->portable_family
) {
1328 case SS_FAMILY_MIG_IPV4
:
1329 tss
->parent
->ss
.ss_family
= AF_INET
;
1331 case SS_FAMILY_MIG_IPV6
:
1332 case 23: /* compatibility: AF_INET6 from mingw */
1333 case 28: /* compatibility: AF_INET6 from FreeBSD sys/socket.h */
1334 tss
->parent
->ss
.ss_family
= AF_INET6
;
1337 error_report("invalid ss_family type %x", tss
->portable_family
);
1344 static const VMStateDescription vmstate_slirp_ss_family
= {
1345 .name
= "slirp-socket-addr/ss_family",
1346 .pre_save
= ss_family_pre_save
,
1347 .post_load
= ss_family_post_load
,
1348 .fields
= (VMStateField
[]) {
1349 VMSTATE_UINT16(portable_family
, SS_FamilyTmpStruct
),
1350 VMSTATE_END_OF_LIST()
1354 static const VMStateDescription vmstate_slirp_socket_addr
= {
1355 .name
= "slirp-socket-addr",
1357 .fields
= (VMStateField
[]) {
1358 VMSTATE_WITH_TMP(union slirp_sockaddr
, SS_FamilyTmpStruct
,
1359 vmstate_slirp_ss_family
),
1360 VMSTATE_SIN4_ADDR(sin
.sin_addr
.s_addr
, union slirp_sockaddr
,
1362 VMSTATE_UINT16_TEST(sin
.sin_port
, union slirp_sockaddr
,
1366 /* Untested: Needs checking by someone with IPv6 test */
1367 VMSTATE_BUFFER_TEST(sin6
.sin6_addr
, union slirp_sockaddr
,
1368 slirp_family_inet6
),
1369 VMSTATE_UINT16_TEST(sin6
.sin6_port
, union slirp_sockaddr
,
1370 slirp_family_inet6
),
1371 VMSTATE_UINT32_TEST(sin6
.sin6_flowinfo
, union slirp_sockaddr
,
1372 slirp_family_inet6
),
1373 VMSTATE_UINT32_TEST(sin6
.sin6_scope_id
, union slirp_sockaddr
,
1374 slirp_family_inet6
),
1377 VMSTATE_END_OF_LIST()
1381 static const VMStateDescription vmstate_slirp_socket
= {
1382 .name
= "slirp-socket",
1384 .pre_load
= slirp_socket_pre_load
,
1385 .fields
= (VMStateField
[]) {
1386 VMSTATE_UINT32(so_urgc
, struct socket
),
1387 /* Pre-v4 versions */
1388 VMSTATE_SIN4_ADDR(so_faddr
.s_addr
, struct socket
,
1389 slirp_older_than_v4
),
1390 VMSTATE_SIN4_ADDR(so_laddr
.s_addr
, struct socket
,
1391 slirp_older_than_v4
),
1392 VMSTATE_UINT16_TEST(so_fport
, struct socket
, slirp_older_than_v4
),
1393 VMSTATE_UINT16_TEST(so_lport
, struct socket
, slirp_older_than_v4
),
1395 VMSTATE_STRUCT(fhost
, struct socket
, 4, vmstate_slirp_socket_addr
,
1396 union slirp_sockaddr
),
1397 VMSTATE_STRUCT(lhost
, struct socket
, 4, vmstate_slirp_socket_addr
,
1398 union slirp_sockaddr
),
1400 VMSTATE_UINT8(so_iptos
, struct socket
),
1401 VMSTATE_UINT8(so_emu
, struct socket
),
1402 VMSTATE_UINT8(so_type
, struct socket
),
1403 VMSTATE_INT32(so_state
, struct socket
),
1404 VMSTATE_STRUCT(so_rcv
, struct socket
, 0, vmstate_slirp_sbuf
,
1406 VMSTATE_STRUCT(so_snd
, struct socket
, 0, vmstate_slirp_sbuf
,
1408 VMSTATE_STRUCT_POINTER(so_tcpcb
, struct socket
, vmstate_slirp_tcp
,
1410 VMSTATE_END_OF_LIST()
1414 static const VMStateDescription vmstate_slirp_bootp_client
= {
1415 .name
= "slirp_bootpclient",
1416 .fields
= (VMStateField
[]) {
1417 VMSTATE_UINT16(allocated
, BOOTPClient
),
1418 VMSTATE_BUFFER(macaddr
, BOOTPClient
),
1419 VMSTATE_END_OF_LIST()
1423 static const VMStateDescription vmstate_slirp
= {
1426 .fields
= (VMStateField
[]) {
1427 VMSTATE_UINT16_V(ip_id
, Slirp
, 2),
1428 VMSTATE_STRUCT_ARRAY(bootp_clients
, Slirp
, NB_BOOTP_CLIENTS
, 3,
1429 vmstate_slirp_bootp_client
, BOOTPClient
),
1430 VMSTATE_END_OF_LIST()
1434 static void slirp_state_save(QEMUFile
*f
, void *opaque
)
1436 Slirp
*slirp
= opaque
;
1437 struct ex_list
*ex_ptr
;
1439 for (ex_ptr
= slirp
->exec_list
; ex_ptr
; ex_ptr
= ex_ptr
->ex_next
)
1440 if (ex_ptr
->ex_pty
== 3) {
1442 so
= slirp_find_ctl_socket(slirp
, ex_ptr
->ex_addr
,
1443 ntohs(ex_ptr
->ex_fport
));
1447 qemu_put_byte(f
, 42);
1448 vmstate_save_state(f
, &vmstate_slirp_socket
, so
, NULL
);
1450 qemu_put_byte(f
, 0);
1452 vmstate_save_state(f
, &vmstate_slirp
, slirp
, NULL
);
1456 static int slirp_state_load(QEMUFile
*f
, void *opaque
, int version_id
)
1458 Slirp
*slirp
= opaque
;
1459 struct ex_list
*ex_ptr
;
1461 while (qemu_get_byte(f
)) {
1463 struct socket
*so
= socreate(slirp
);
1468 ret
= vmstate_load_state(f
, &vmstate_slirp_socket
, so
, version_id
);
1473 if ((so
->so_faddr
.s_addr
& slirp
->vnetwork_mask
.s_addr
) !=
1474 slirp
->vnetwork_addr
.s_addr
) {
1477 for (ex_ptr
= slirp
->exec_list
; ex_ptr
; ex_ptr
= ex_ptr
->ex_next
) {
1478 if (ex_ptr
->ex_pty
== 3 &&
1479 so
->so_faddr
.s_addr
== ex_ptr
->ex_addr
.s_addr
&&
1480 so
->so_fport
== ex_ptr
->ex_fport
) {
1487 so
->extra
= (void *)ex_ptr
->ex_exec
;
1490 return vmstate_load_state(f
, &vmstate_slirp
, slirp
, version_id
);