summaryrefslogtreecommitdiffstats
path: root/src/libknot/xdp/protocols.h
blob: ee26e9726ebe207dd567242bc3595683fcf59bb2 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
/*  Copyright (C) 2023 CZ.NIC, z.s.p.o. <knot-dns@labs.nic.cz>

    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.

    This program 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 General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <https://www.gnu.org/licenses/>.
 */

#pragma once

#include <assert.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <string.h>

#include "libknot/endian.h"
#include "libknot/xdp/bpf-consts.h"
#include "libknot/xdp/msg.h"

/* Don't fragment flag. */
#define	IP_DF 0x4000

#define HDR_8021Q_LEN 4;

/*
 * Following prot_read_*() functions do not check sanity of parsed packet.
 * Broken packets have to be dropped by BPF filter prior getting here.
 */

inline static void *prot_read_udp(void *data, uint16_t *src_port, uint16_t *dst_port)
{
	const struct udphdr *udp = data;

	*src_port = udp->source;
	*dst_port = udp->dest;

	return data + sizeof(*udp);
}

enum {
	PROT_TCP_OPT_ENDOP = 0,
	PROT_TCP_OPT_NOOP  = 1,
	PROT_TCP_OPT_MSS   = 2,
	PROT_TCP_OPT_WSC   = 3, // window scale

	PROT_TCP_OPT_LEN_MSS = 4,
	PROT_TCP_OPT_LEN_WSC = 3,
};

inline static void *prot_read_tcp(void *data, knot_xdp_msg_t *msg, uint16_t *src_port, uint16_t *dst_port)
{
	const struct tcphdr *tcp = data;

	msg->flags |= KNOT_XDP_MSG_TCP;

	if (tcp->syn) {
		msg->flags |= KNOT_XDP_MSG_SYN;
	}
	if (tcp->ack) {
		msg->flags |= KNOT_XDP_MSG_ACK;
	}
	if (tcp->fin) {
		msg->flags |= KNOT_XDP_MSG_FIN;
	}
	if (tcp->rst) {
		msg->flags |= KNOT_XDP_MSG_RST;
	}

	msg->seqno = be32toh(tcp->seq);
	msg->ackno = be32toh(tcp->ack_seq);
	msg->win = be16toh(tcp->window);

	*src_port = tcp->source;
	*dst_port = tcp->dest;

	uint8_t *opts = data + sizeof(*tcp), *hdr_end = data + tcp->doff * 4;
	while (opts < hdr_end) {
		if (opts[0] == PROT_TCP_OPT_ENDOP || opts[0] == PROT_TCP_OPT_NOOP) {
			opts++;
			continue;
		}

		if (opts + 1 > hdr_end || opts + opts[1] > hdr_end || opts[1] < 2) {
			// Malformed option.
			break;
		}

		if (opts[0] == PROT_TCP_OPT_MSS && opts[1] == PROT_TCP_OPT_LEN_MSS) {
			msg->flags |= KNOT_XDP_MSG_MSS;
			memcpy(&msg->mss, &opts[2], sizeof(msg->mss));
			msg->mss = be16toh(msg->mss);
		}

		if (opts[0] == PROT_TCP_OPT_WSC && opts[1] == PROT_TCP_OPT_LEN_WSC) {
			msg->flags |= KNOT_XDP_MSG_WSC;
			msg->win_scale = opts[2];
		}

		opts += opts[1];
	}

	return hdr_end;
}

inline static void *prot_read_ipv4(void *data, knot_xdp_msg_t *msg, void **data_end)
{
	const struct iphdr *ip4 = data;

	// Conditions ensured by the BPF filter.
	assert(ip4->version == 4);
	assert(ip4->frag_off == 0 || ip4->frag_off == __constant_htons(IP_DF));
	// IPv4 header checksum is not verified!

	struct sockaddr_in *src = (struct sockaddr_in *)&msg->ip_from;
	struct sockaddr_in *dst = (struct sockaddr_in *)&msg->ip_to;
	memcpy(&src->sin_addr, &ip4->saddr, sizeof(src->sin_addr));
	memcpy(&dst->sin_addr, &ip4->daddr, sizeof(dst->sin_addr));
	src->sin_family = AF_INET;
	dst->sin_family = AF_INET;

	msg->ecn = (ip4->tos & 0x3);

	*data_end = data + be16toh(ip4->tot_len);
	data += ip4->ihl * 4;

	if (ip4->protocol == IPPROTO_TCP) {
		return prot_read_tcp(data, msg, &src->sin_port, &dst->sin_port);
	} else {
		assert(ip4->protocol == IPPROTO_UDP);
		return prot_read_udp(data, &src->sin_port, &dst->sin_port);
	}
}

inline static void *prot_read_ipv6(void *data, knot_xdp_msg_t *msg, void **data_end)
{
	const struct ipv6hdr *ip6 = data;

	msg->flags |= KNOT_XDP_MSG_IPV6;

	// Conditions ensured by the BPF filter.
	assert(ip6->version == 6);

	struct sockaddr_in6 *src = (struct sockaddr_in6 *)&msg->ip_from;
	struct sockaddr_in6 *dst = (struct sockaddr_in6 *)&msg->ip_to;
	memcpy(&src->sin6_addr, &ip6->saddr, sizeof(src->sin6_addr));
	memcpy(&dst->sin6_addr, &ip6->daddr, sizeof(dst->sin6_addr));
	src->sin6_family = AF_INET6;
	dst->sin6_family = AF_INET6;
	src->sin6_flowinfo = 0;
	dst->sin6_flowinfo = 0;
	// Scope ID is ignored.

	msg->ecn = ((ip6->flow_lbl[0] & 0x30) >> 4);

	data += sizeof(*ip6);
	*data_end = data + be16toh(ip6->payload_len);

	if (ip6->nexthdr == IPPROTO_TCP) {
		return prot_read_tcp(data, msg, &src->sin6_port, &dst->sin6_port);
	} else {
		assert(ip6->nexthdr == IPPROTO_UDP);
		return prot_read_udp(data, &src->sin6_port, &dst->sin6_port);
	}
}

inline static void *prot_read_eth(void *data, knot_xdp_msg_t *msg, void **data_end,
                                  const uint16_t *vlan_map, unsigned vlan_map_max)
{
	const struct ethhdr *eth = data;
	knot_xdp_info_t *info = data - KNOT_XDP_PKT_ALIGNMENT - sizeof(*info);

	memcpy(msg->eth_from, eth->h_source, ETH_ALEN);
	memcpy(msg->eth_to,   eth->h_dest,   ETH_ALEN);
	msg->flags = 0;

	if (eth->h_proto == __constant_htons(ETH_P_8021Q)) {
		if (info->out_if_index > 0 && info->out_if_index <= vlan_map_max) {
			assert(vlan_map);
			msg->vlan_tci = vlan_map[info->out_if_index];
		} else {
			memcpy(&msg->vlan_tci, data + sizeof(*eth), sizeof(msg->vlan_tci));
		}
		data += HDR_8021Q_LEN;
		eth = data;
	}

	data += sizeof(*eth);

	if (eth->h_proto == __constant_htons(ETH_P_IPV6)) {
		return prot_read_ipv6(data, msg, data_end);
	} else {
		assert(eth->h_proto == __constant_htons(ETH_P_IP));
		return prot_read_ipv4(data, msg, data_end);
	}
}

inline static size_t prot_write_hdrs_len(const knot_xdp_msg_t *msg)
{
	size_t res = sizeof(struct ethhdr) + sizeof(struct iphdr) + sizeof(struct udphdr);

	if (msg->vlan_tci != 0 || msg->flags & KNOT_XDP_MSG_VLAN) {
		res += HDR_8021Q_LEN;
	}

	if (msg->flags & KNOT_XDP_MSG_IPV6) {
		res += sizeof(struct ipv6hdr) - sizeof(struct iphdr);
	}

	if (msg->flags & KNOT_XDP_MSG_TCP) {
		res += sizeof(struct tcphdr) - sizeof(struct udphdr);

		if (msg->flags & KNOT_XDP_MSG_MSS) {
			res += PROT_TCP_OPT_LEN_MSS;
		}
		if (msg->flags & KNOT_XDP_MSG_WSC) {
			res += PROT_TCP_OPT_LEN_WSC + 1; // 1 == align
		}
	}

	return res;
}

/* Checksum endianness implementation notes for ipv4_checksum() and checksum().
 *
 * The basis for checksum is addition on big-endian 16-bit words, with bit 16 carrying
 * over to bit 0.  That can be viewed as first byte carrying to the second and the
 * second one carrying back to the first one, i.e. a symmetrical situation.
 * Therefore the result is the same even when arithmetics is done on little-endian (!)
 */

inline static void checksum(uint32_t *result, const void *_data, uint32_t _data_len)
{
	assert(!(_data_len & 1));
	const uint16_t *data = _data;
	uint32_t len = _data_len / 2;
	while (len-- > 0) {
		*result += *data++;
	}
}

inline static void checksum_uint16(uint32_t *result, uint16_t x)
{
	checksum(result, &x, sizeof(x));
}

inline static void checksum_payload(uint32_t *result, void *payload, size_t pay_len)
{
	if (pay_len & 1) {
		((uint8_t *)payload)[pay_len++] = 0;
	}
	checksum(result, payload, pay_len);
}

inline static uint16_t checksum_finish(uint32_t result, bool nonzero)
{
	while (result > 0xffff) {
		result = (result & 0xffff) + (result >> 16);
	}
	if (!nonzero || result != 0xffff) {
		result = ~result;
	}
	return result;
}

inline static void prot_write_udp(void *data, const knot_xdp_msg_t *msg, void *data_end,
                                  uint16_t src_port, uint16_t dst_port, uint32_t chksum)
{
	struct udphdr *udp = data;

	udp->len    = htobe16(data_end - data);
	udp->source = src_port;
	udp->dest   = dst_port;

	if (msg->flags & KNOT_XDP_MSG_IPV6) {
		udp->check = 0;
		checksum(&chksum, &udp->len, sizeof(udp->len));
		checksum_uint16(&chksum, htobe16(IPPROTO_UDP));
		checksum_payload(&chksum, data, data_end - data);
		udp->check = checksum_finish(chksum, true);
	} else {
		udp->check = 0; // UDP over IPv4 doesn't require checksum.
	}

	assert(data + sizeof(*udp) == msg->payload.iov_base);
}

inline static void prot_write_tcp(void *data, const knot_xdp_msg_t *msg, void *data_end,
                                  uint16_t src_port, uint16_t dst_port, uint32_t chksum,
                                  uint16_t mss)
{
	struct tcphdr *tcp = data;

	tcp->source  = src_port;
	tcp->dest    = dst_port;
	tcp->seq     = htobe32(msg->seqno);
	tcp->ack_seq = htobe32(msg->ackno);
	tcp->window  = htobe16(msg->win);
	tcp->check   = 0; // Temporarily initialize before checksum calculation.

	tcp->res1 = 0;
	tcp->urg = 0;
	tcp->ece = 0;
	tcp->cwr = 0;
	tcp->urg_ptr = 0;

	tcp->syn = ((msg->flags & KNOT_XDP_MSG_SYN) ? 1 : 0);
	tcp->ack = ((msg->flags & KNOT_XDP_MSG_ACK) ? 1 : 0);
	tcp->fin = ((msg->flags & KNOT_XDP_MSG_FIN) ? 1 : 0);
	tcp->rst = ((msg->flags & KNOT_XDP_MSG_RST) ? 1 : 0);

	uint8_t *hdr_end = data + sizeof(*tcp);
	if (msg->flags & KNOT_XDP_MSG_WSC) {
		hdr_end[0] = PROT_TCP_OPT_WSC;
		hdr_end[1] = PROT_TCP_OPT_LEN_WSC;
		hdr_end[2] = msg->win_scale;
		hdr_end += PROT_TCP_OPT_LEN_WSC;
		*hdr_end++ = PROT_TCP_OPT_NOOP; // align
	}
	if (msg->flags & KNOT_XDP_MSG_MSS) {
		mss = htobe16(mss);
		hdr_end[0] = PROT_TCP_OPT_MSS;
		hdr_end[1] = PROT_TCP_OPT_LEN_MSS;
		memcpy(&hdr_end[2], &mss, sizeof(mss));
		hdr_end += PROT_TCP_OPT_LEN_MSS;
	}

	tcp->psh = ((data_end - (void *)hdr_end > 0) ? 1 : 0);
	tcp->doff = (hdr_end - (uint8_t *)tcp) / 4;
	assert((hdr_end - (uint8_t *)tcp) % 4 == 0);

	checksum_uint16(&chksum, htobe16(IPPROTO_TCP));
	checksum_uint16(&chksum, htobe16(data_end - data));
	checksum_payload(&chksum, data, data_end - data);
	tcp->check = checksum_finish(chksum, false);

	assert(hdr_end == msg->payload.iov_base);
}

inline static uint16_t from32to16(uint32_t sum)
{
	sum = (sum & 0xffff) + (sum >> 16);
	sum = (sum & 0xffff) + (sum >> 16);
	return sum;
}

inline static uint16_t ipv4_checksum(const uint16_t *ipv4_hdr)
{
	uint32_t sum32 = 0;
	for (int i = 0; i < 10; ++i) {
		if (i != 5) {
			sum32 += ipv4_hdr[i];
		}
	}
	return ~from32to16(sum32);
}

inline static void prot_write_ipv4(void *data, const knot_xdp_msg_t *msg,
                                   void *data_end, uint16_t tcp_mss)
{
	struct iphdr *ip4 = data;

	ip4->version  = 4;
	ip4->ihl      = sizeof(*ip4) / 4;
	ip4->tos      = msg->ecn;
	ip4->tot_len  = htobe16(data_end - data);
	ip4->id       = 0;
	ip4->frag_off = 0;
	ip4->ttl      = IPDEFTTL;
	ip4->protocol = ((msg->flags & KNOT_XDP_MSG_TCP) ? IPPROTO_TCP : IPPROTO_UDP);

	const struct sockaddr_in *src = (const struct sockaddr_in *)&msg->ip_from;
	const struct sockaddr_in *dst = (const struct sockaddr_in *)&msg->ip_to;
	memcpy(&ip4->saddr, &src->sin_addr, sizeof(src->sin_addr));
	memcpy(&ip4->daddr, &dst->sin_addr, sizeof(dst->sin_addr));

	ip4->check = ipv4_checksum(data);

	data += sizeof(*ip4);

	if (msg->flags & KNOT_XDP_MSG_TCP) {
		uint32_t chk = 0;
		checksum(&chk, &src->sin_addr, sizeof(src->sin_addr));
		checksum(&chk, &dst->sin_addr, sizeof(dst->sin_addr));

		prot_write_tcp(data, msg, data_end, src->sin_port, dst->sin_port, chk, tcp_mss);
	} else {
		prot_write_udp(data, msg, data_end, src->sin_port, dst->sin_port, 0); // IPv4/UDP requires no checksum
	}
}

inline static void prot_write_ipv6(void *data, const knot_xdp_msg_t *msg,
                                   void *data_end, uint16_t tcp_mss)
{
	struct ipv6hdr *ip6 = data;

	ip6->version     = 6;
	ip6->priority    = 0;
	ip6->flow_lbl[0] = (msg->ecn << 4);
	ip6->payload_len = htobe16(data_end - data - sizeof(*ip6));
	ip6->nexthdr     = ((msg->flags & KNOT_XDP_MSG_TCP) ? IPPROTO_TCP : IPPROTO_UDP);
	ip6->hop_limit   = IPDEFTTL;

	memset(ip6->flow_lbl, 0, sizeof(ip6->flow_lbl));

	const struct sockaddr_in6 *src = (const struct sockaddr_in6 *)&msg->ip_from;
	const struct sockaddr_in6 *dst = (const struct sockaddr_in6 *)&msg->ip_to;
	memcpy(&ip6->saddr, &src->sin6_addr, sizeof(src->sin6_addr));
	memcpy(&ip6->daddr, &dst->sin6_addr, sizeof(dst->sin6_addr));

	data += sizeof(*ip6);

	uint32_t chk = 0;
	checksum(&chk, &src->sin6_addr, sizeof(src->sin6_addr));
	checksum(&chk, &dst->sin6_addr, sizeof(dst->sin6_addr));

	if (msg->flags & KNOT_XDP_MSG_TCP) {
		prot_write_tcp(data, msg, data_end, src->sin6_port, dst->sin6_port, chk, tcp_mss);
	} else {
		prot_write_udp(data, msg, data_end, src->sin6_port, dst->sin6_port, chk);
	}
}

inline static void prot_write_eth(void *data, const knot_xdp_msg_t *msg,
                                  void *data_end, uint16_t tcp_mss)
{
	struct ethhdr *eth = data;

	memcpy(eth->h_source, msg->eth_from, ETH_ALEN);
	memcpy(eth->h_dest,   msg->eth_to,   ETH_ALEN);

	if (msg->vlan_tci != 0) {
		eth->h_proto = __constant_htons(ETH_P_8021Q);
		memcpy(data + sizeof(*eth), &msg->vlan_tci, sizeof(msg->vlan_tci));
		data += HDR_8021Q_LEN;
		eth = data;
	}

	data += sizeof(*eth);

	if (msg->flags & KNOT_XDP_MSG_IPV6) {
		eth->h_proto = __constant_htons(ETH_P_IPV6);
		prot_write_ipv6(data, msg, data_end, tcp_mss);
	} else {
		eth->h_proto = __constant_htons(ETH_P_IP);
		prot_write_ipv4(data, msg, data_end, tcp_mss);
	}
}