1/*
2 * QEMU network structures definitions and helper functions
3 *
4 * Copyright (c) 2012 Ravello Systems LTD (http://ravellosystems.com)
5 *
6 * Developed by Daynix Computing LTD (http://www.daynix.com)
7 *
8 * Portions developed by Free Software Foundation, Inc
9 * Copyright (C) 1991-1997, 2001, 2003, 2006 Free Software Foundation, Inc.
10 * See netinet/ip6.h and netinet/in.h (GNU C Library)
11 *
12 * Portions developed by Igor Kovalenko
13 * Copyright (c) 2006 Igor Kovalenko
14 * See hw/rtl8139.c (QEMU)
15 *
16 * Authors:
17 * Dmitry Fleytman <dmitry@daynix.com>
18 * Tamir Shomer <tamirs@daynix.com>
19 * Yan Vugenfirer <yan@daynix.com>
20 *
21 * This work is licensed under the terms of the GNU GPL, version 2 or later.
22 * See the COPYING file in the top-level directory.
23 *
24 */
25
26#ifndef QEMU_ETH_H
27#define QEMU_ETH_H
28
29#include "qemu/bswap.h"
30#include "qemu/iov.h"
31
32#define ETH_ALEN 6
33#define ETH_HLEN 14
34
35struct eth_header {
36 uint8_t h_dest[ETH_ALEN]; /* destination eth addr */
37 uint8_t h_source[ETH_ALEN]; /* source ether addr */
38 uint16_t h_proto; /* packet type ID field */
39};
40
41struct vlan_header {
42 uint16_t h_tci; /* priority and VLAN ID */
43 uint16_t h_proto; /* encapsulated protocol */
44};
45
46struct ip_header {
47 uint8_t ip_ver_len; /* version and header length */
48 uint8_t ip_tos; /* type of service */
49 uint16_t ip_len; /* total length */
50 uint16_t ip_id; /* identification */
51 uint16_t ip_off; /* fragment offset field */
52 uint8_t ip_ttl; /* time to live */
53 uint8_t ip_p; /* protocol */
54 uint16_t ip_sum; /* checksum */
55 uint32_t ip_src, ip_dst; /* source and destination address */
56};
57
58typedef struct tcp_header {
59 uint16_t th_sport; /* source port */
60 uint16_t th_dport; /* destination port */
61 uint32_t th_seq; /* sequence number */
62 uint32_t th_ack; /* acknowledgment number */
63 uint16_t th_offset_flags; /* data offset, reserved 6 bits, */
64 /* TCP protocol flags */
65 uint16_t th_win; /* window */
66 uint16_t th_sum; /* checksum */
67 uint16_t th_urp; /* urgent pointer */
68} tcp_header;
69
70#define TCP_FLAGS_ONLY(flags) ((flags) & 0x3f)
71
72#define TCP_HEADER_FLAGS(tcp) \
73 TCP_FLAGS_ONLY(be16_to_cpu((tcp)->th_offset_flags))
74
75#define TCP_FLAG_ACK 0x10
76
77#define TCP_HEADER_DATA_OFFSET(tcp) \
78 (((be16_to_cpu((tcp)->th_offset_flags) >> 12) & 0xf) << 2)
79
80typedef struct udp_header {
81 uint16_t uh_sport; /* source port */
82 uint16_t uh_dport; /* destination port */
83 uint16_t uh_ulen; /* udp length */
84 uint16_t uh_sum; /* udp checksum */
85} udp_header;
86
87typedef struct ip_pseudo_header {
88 uint32_t ip_src;
89 uint32_t ip_dst;
90 uint8_t zeros;
91 uint8_t ip_proto;
92 uint16_t ip_payload;
93} ip_pseudo_header;
94
95/* IPv6 address */
96struct in6_address {
97 union {
98 uint8_t __u6_addr8[16];
99 } __in6_u;
100};
101
102struct ip6_header {
103 union {
104 struct ip6_hdrctl {
105 uint32_t ip6_un1_flow; /* 4 bits version, 8 bits TC,
106 20 bits flow-ID */
107 uint16_t ip6_un1_plen; /* payload length */
108 uint8_t ip6_un1_nxt; /* next header */
109 uint8_t ip6_un1_hlim; /* hop limit */
110 } ip6_un1;
111 uint8_t ip6_un2_vfc; /* 4 bits version, top 4 bits tclass */
112 struct ip6_ecn_access {
113 uint8_t ip6_un3_vfc; /* 4 bits version, top 4 bits tclass */
114 uint8_t ip6_un3_ecn; /* 2 bits ECN, top 6 bits payload length */
115 } ip6_un3;
116 } ip6_ctlun;
117 struct in6_address ip6_src; /* source address */
118 struct in6_address ip6_dst; /* destination address */
119};
120
121typedef struct ip6_pseudo_header {
122 struct in6_address ip6_src;
123 struct in6_address ip6_dst;
124 uint32_t len;
125 uint8_t zero[3];
126 uint8_t next_hdr;
127} ip6_pseudo_header;
128
129struct ip6_ext_hdr {
130 uint8_t ip6r_nxt; /* next header */
131 uint8_t ip6r_len; /* length in units of 8 octets */
132};
133
134struct ip6_ext_hdr_routing {
135 uint8_t nxt;
136 uint8_t len;
137 uint8_t rtype;
138 uint8_t segleft;
139 uint8_t rsvd[4];
140};
141
142struct ip6_option_hdr {
143#define IP6_OPT_PAD1 (0x00)
144#define IP6_OPT_HOME (0xC9)
145 uint8_t type;
146 uint8_t len;
147};
148
149struct udp_hdr {
150 uint16_t uh_sport; /* source port */
151 uint16_t uh_dport; /* destination port */
152 uint16_t uh_ulen; /* udp length */
153 uint16_t uh_sum; /* udp checksum */
154};
155
156struct tcp_hdr {
157 u_short th_sport; /* source port */
158 u_short th_dport; /* destination port */
159 uint32_t th_seq; /* sequence number */
160 uint32_t th_ack; /* acknowledgment number */
161#ifdef HOST_WORDS_BIGENDIAN
162 u_char th_off : 4, /* data offset */
163 th_x2:4; /* (unused) */
164#else
165 u_char th_x2 : 4, /* (unused) */
166 th_off:4; /* data offset */
167#endif
168
169#define TH_ELN 0x1 /* explicit loss notification */
170#define TH_ECN 0x2 /* explicit congestion notification */
171#define TH_FS 0x4 /* fast start */
172
173 u_char th_flags;
174#define TH_FIN 0x01
175#define TH_SYN 0x02
176#define TH_RST 0x04
177#define TH_PUSH 0x08
178#define TH_ACK 0x10
179#define TH_URG 0x20
180#define TH_ECE 0x40
181#define TH_CWR 0x80
182 u_short th_win; /* window */
183 u_short th_sum; /* checksum */
184 u_short th_urp; /* urgent pointer */
185};
186
187#define ip6_nxt ip6_ctlun.ip6_un1.ip6_un1_nxt
188#define ip6_ecn_acc ip6_ctlun.ip6_un3.ip6_un3_ecn
189
190#define PKT_GET_ETH_HDR(p) \
191 ((struct eth_header *)(p))
192#define PKT_GET_VLAN_HDR(p) \
193 ((struct vlan_header *) (((uint8_t *)(p)) + sizeof(struct eth_header)))
194#define PKT_GET_DVLAN_HDR(p) \
195 (PKT_GET_VLAN_HDR(p) + 1)
196#define PKT_GET_IP_HDR(p) \
197 ((struct ip_header *)(((uint8_t *)(p)) + eth_get_l2_hdr_length(p)))
198#define IP_HDR_GET_LEN(p) \
199 ((ldub_p(p + offsetof(struct ip_header, ip_ver_len)) & 0x0F) << 2)
200#define IP_HDR_GET_P(p) \
201 (ldub_p(p + offsetof(struct ip_header, ip_p)))
202#define PKT_GET_IP_HDR_LEN(p) \
203 (IP_HDR_GET_LEN(PKT_GET_IP_HDR(p)))
204#define PKT_GET_IP6_HDR(p) \
205 ((struct ip6_header *) (((uint8_t *)(p)) + eth_get_l2_hdr_length(p)))
206#define IP_HEADER_VERSION(ip) \
207 (((ip)->ip_ver_len >> 4) & 0xf)
208#define IP4_IS_FRAGMENT(ip) \
209 ((be16_to_cpu((ip)->ip_off) & (IP_OFFMASK | IP_MF)) != 0)
210
211#define ETH_P_IP (0x0800) /* Internet Protocol packet */
212#define ETH_P_ARP (0x0806) /* Address Resolution packet */
213#define ETH_P_IPV6 (0x86dd)
214#define ETH_P_VLAN (0x8100)
215#define ETH_P_DVLAN (0x88a8)
216#define ETH_P_NCSI (0x88f8)
217#define ETH_P_UNKNOWN (0xffff)
218#define VLAN_VID_MASK 0x0fff
219#define IP_HEADER_VERSION_4 (4)
220#define IP_HEADER_VERSION_6 (6)
221#define IP_PROTO_TCP (6)
222#define IP_PROTO_UDP (17)
223#define IPTOS_ECN_MASK 0x03
224#define IPTOS_ECN(x) ((x) & IPTOS_ECN_MASK)
225#define IPTOS_ECN_CE 0x03
226#define IP6_ECN_MASK 0xC0
227#define IP6_ECN(x) ((x) & IP6_ECN_MASK)
228#define IP6_ECN_CE 0xC0
229#define IP4_DONT_FRAGMENT_FLAG (1 << 14)
230
231#define IS_SPECIAL_VLAN_ID(x) \
232 (((x) == 0) || ((x) == 0xFFF))
233
234#define ETH_MAX_L2_HDR_LEN \
235 (sizeof(struct eth_header) + 2 * sizeof(struct vlan_header))
236
237#define ETH_MAX_IP4_HDR_LEN (60)
238#define ETH_MAX_IP_DGRAM_LEN (0xFFFF)
239
240#define IP_FRAG_UNIT_SIZE (8)
241#define IP_FRAG_ALIGN_SIZE(x) ((x) & ~0x7)
242#define IP_RF 0x8000 /* reserved fragment flag */
243#define IP_DF 0x4000 /* don't fragment flag */
244#define IP_MF 0x2000 /* more fragments flag */
245#define IP_OFFMASK 0x1fff /* mask for fragmenting bits */
246
247#define IP6_EXT_GRANULARITY (8) /* Size granularity for
248 IPv6 extension headers */
249
250/* IP6 extension header types */
251#define IP6_HOP_BY_HOP (0)
252#define IP6_ROUTING (43)
253#define IP6_FRAGMENT (44)
254#define IP6_ESP (50)
255#define IP6_AUTHENTICATION (51)
256#define IP6_NONE (59)
257#define IP6_DESTINATON (60)
258#define IP6_MOBILITY (135)
259
260static inline int is_multicast_ether_addr(const uint8_t *addr)
261{
262 return 0x01 & addr[0];
263}
264
265static inline int is_broadcast_ether_addr(const uint8_t *addr)
266{
267 return (addr[0] & addr[1] & addr[2] & addr[3] & addr[4] & addr[5]) == 0xff;
268}
269
270static inline int is_unicast_ether_addr(const uint8_t *addr)
271{
272 return !is_multicast_ether_addr(addr);
273}
274
275typedef enum {
276 ETH_PKT_UCAST = 0xAABBCC00,
277 ETH_PKT_BCAST,
278 ETH_PKT_MCAST
279} eth_pkt_types_e;
280
281static inline eth_pkt_types_e
282get_eth_packet_type(const struct eth_header *ehdr)
283{
284 if (is_broadcast_ether_addr(ehdr->h_dest)) {
285 return ETH_PKT_BCAST;
286 } else if (is_multicast_ether_addr(ehdr->h_dest)) {
287 return ETH_PKT_MCAST;
288 } else { /* unicast */
289 return ETH_PKT_UCAST;
290 }
291}
292
293static inline uint32_t
294eth_get_l2_hdr_length(const void *p)
295{
296 uint16_t proto = be16_to_cpu(PKT_GET_ETH_HDR(p)->h_proto);
297 struct vlan_header *hvlan = PKT_GET_VLAN_HDR(p);
298 switch (proto) {
299 case ETH_P_VLAN:
300 return sizeof(struct eth_header) + sizeof(struct vlan_header);
301 case ETH_P_DVLAN:
302 if (be16_to_cpu(hvlan->h_proto) == ETH_P_VLAN) {
303 return sizeof(struct eth_header) + 2 * sizeof(struct vlan_header);
304 } else {
305 return sizeof(struct eth_header) + sizeof(struct vlan_header);
306 }
307 default:
308 return sizeof(struct eth_header);
309 }
310}
311
312static inline uint32_t
313eth_get_l2_hdr_length_iov(const struct iovec *iov, int iovcnt)
314{
315 uint8_t p[sizeof(struct eth_header) + sizeof(struct vlan_header)];
316 size_t copied = iov_to_buf(iov, iovcnt, 0, p, ARRAY_SIZE(p));
317
318 if (copied < ARRAY_SIZE(p)) {
319 return copied;
320 }
321
322 return eth_get_l2_hdr_length(p);
323}
324
325static inline uint16_t
326eth_get_pkt_tci(const void *p)
327{
328 uint16_t proto = be16_to_cpu(PKT_GET_ETH_HDR(p)->h_proto);
329 struct vlan_header *hvlan = PKT_GET_VLAN_HDR(p);
330 switch (proto) {
331 case ETH_P_VLAN:
332 case ETH_P_DVLAN:
333 return be16_to_cpu(hvlan->h_tci);
334 default:
335 return 0;
336 }
337}
338
339size_t
340eth_strip_vlan(const struct iovec *iov, int iovcnt, size_t iovoff,
341 uint8_t *new_ehdr_buf,
342 uint16_t *payload_offset, uint16_t *tci);
343
344size_t
345eth_strip_vlan_ex(const struct iovec *iov, int iovcnt, size_t iovoff,
346 uint16_t vet, uint8_t *new_ehdr_buf,
347 uint16_t *payload_offset, uint16_t *tci);
348
349uint16_t
350eth_get_l3_proto(const struct iovec *l2hdr_iov, int iovcnt, size_t l2hdr_len);
351
352void eth_setup_vlan_headers_ex(struct eth_header *ehdr, uint16_t vlan_tag,
353 uint16_t vlan_ethtype, bool *is_new);
354
355static inline void
356eth_setup_vlan_headers(struct eth_header *ehdr, uint16_t vlan_tag,
357 bool *is_new)
358{
359 eth_setup_vlan_headers_ex(ehdr, vlan_tag, ETH_P_VLAN, is_new);
360}
361
362
363uint8_t eth_get_gso_type(uint16_t l3_proto, uint8_t *l3_hdr, uint8_t l4proto);
364
365typedef struct eth_ip6_hdr_info_st {
366 uint8_t l4proto;
367 size_t full_hdr_len;
368 struct ip6_header ip6_hdr;
369 bool has_ext_hdrs;
370 bool rss_ex_src_valid;
371 struct in6_address rss_ex_src;
372 bool rss_ex_dst_valid;
373 struct in6_address rss_ex_dst;
374 bool fragment;
375} eth_ip6_hdr_info;
376
377typedef struct eth_ip4_hdr_info_st {
378 struct ip_header ip4_hdr;
379 bool fragment;
380} eth_ip4_hdr_info;
381
382typedef struct eth_l4_hdr_info_st {
383 union {
384 struct tcp_header tcp;
385 struct udp_header udp;
386 } hdr;
387
388 bool has_tcp_data;
389} eth_l4_hdr_info;
390
391void eth_get_protocols(const struct iovec *iov, int iovcnt,
392 bool *isip4, bool *isip6,
393 bool *isudp, bool *istcp,
394 size_t *l3hdr_off,
395 size_t *l4hdr_off,
396 size_t *l5hdr_off,
397 eth_ip6_hdr_info *ip6hdr_info,
398 eth_ip4_hdr_info *ip4hdr_info,
399 eth_l4_hdr_info *l4hdr_info);
400
401void eth_setup_ip4_fragmentation(const void *l2hdr, size_t l2hdr_len,
402 void *l3hdr, size_t l3hdr_len,
403 size_t l3payload_len,
404 size_t frag_offset, bool more_frags);
405
406void
407eth_fix_ip4_checksum(void *l3hdr, size_t l3hdr_len);
408
409uint32_t
410eth_calc_ip4_pseudo_hdr_csum(struct ip_header *iphdr,
411 uint16_t csl,
412 uint32_t *cso);
413
414uint32_t
415eth_calc_ip6_pseudo_hdr_csum(struct ip6_header *iphdr,
416 uint16_t csl,
417 uint8_t l4_proto,
418 uint32_t *cso);
419
420bool
421eth_parse_ipv6_hdr(const struct iovec *pkt, int pkt_frags,
422 size_t ip6hdr_off, eth_ip6_hdr_info *info);
423
424#endif
425