summaryrefslogtreecommitdiff
path: root/vendor/github.com/vishvananda/netlink/conntrack_linux.go
blob: 4bff0dcbab03c5473f1ffdd35e79012becff79ef (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
package netlink

import (
	"bytes"
	"encoding/binary"
	"errors"
	"fmt"
	"net"

	"github.com/vishvananda/netlink/nl"
	"golang.org/x/sys/unix"
)

// ConntrackTableType Conntrack table for the netlink operation
type ConntrackTableType uint8

const (
	// ConntrackTable Conntrack table
	// https://github.com/torvalds/linux/blob/master/include/uapi/linux/netfilter/nfnetlink.h -> #define NFNL_SUBSYS_CTNETLINK		 1
	ConntrackTable = 1
	// ConntrackExpectTable Conntrack expect table
	// https://github.com/torvalds/linux/blob/master/include/uapi/linux/netfilter/nfnetlink.h -> #define NFNL_SUBSYS_CTNETLINK_EXP 2
	ConntrackExpectTable = 2
)

const (
	// backward compatibility with golang 1.6 which does not have io.SeekCurrent
	seekCurrent = 1
)

// InetFamily Family type
type InetFamily uint8

//  -L [table] [options]          List conntrack or expectation table
//  -G [table] parameters         Get conntrack or expectation

//  -I [table] parameters         Create a conntrack or expectation
//  -U [table] parameters         Update a conntrack
//  -E [table] [options]          Show events

//  -C [table]                    Show counter
//  -S                            Show statistics

// ConntrackTableList returns the flow list of a table of a specific family
// conntrack -L [table] [options]          List conntrack or expectation table
func ConntrackTableList(table ConntrackTableType, family InetFamily) ([]*ConntrackFlow, error) {
	return pkgHandle.ConntrackTableList(table, family)
}

// ConntrackTableFlush flushes all the flows of a specified table
// conntrack -F [table]            Flush table
// The flush operation applies to all the family types
func ConntrackTableFlush(table ConntrackTableType) error {
	return pkgHandle.ConntrackTableFlush(table)
}

// ConntrackDeleteFilter deletes entries on the specified table on the base of the filter
// conntrack -D [table] parameters         Delete conntrack or expectation
func ConntrackDeleteFilter(table ConntrackTableType, family InetFamily, filter CustomConntrackFilter) (uint, error) {
	return pkgHandle.ConntrackDeleteFilter(table, family, filter)
}

// ConntrackTableList returns the flow list of a table of a specific family using the netlink handle passed
// conntrack -L [table] [options]          List conntrack or expectation table
func (h *Handle) ConntrackTableList(table ConntrackTableType, family InetFamily) ([]*ConntrackFlow, error) {
	res, err := h.dumpConntrackTable(table, family)
	if err != nil {
		return nil, err
	}

	// Deserialize all the flows
	var result []*ConntrackFlow
	for _, dataRaw := range res {
		result = append(result, parseRawData(dataRaw))
	}

	return result, nil
}

// ConntrackTableFlush flushes all the flows of a specified table using the netlink handle passed
// conntrack -F [table]            Flush table
// The flush operation applies to all the family types
func (h *Handle) ConntrackTableFlush(table ConntrackTableType) error {
	req := h.newConntrackRequest(table, unix.AF_INET, nl.IPCTNL_MSG_CT_DELETE, unix.NLM_F_ACK)
	_, err := req.Execute(unix.NETLINK_NETFILTER, 0)
	return err
}

// ConntrackDeleteFilter deletes entries on the specified table on the base of the filter using the netlink handle passed
// conntrack -D [table] parameters         Delete conntrack or expectation
func (h *Handle) ConntrackDeleteFilter(table ConntrackTableType, family InetFamily, filter CustomConntrackFilter) (uint, error) {
	res, err := h.dumpConntrackTable(table, family)
	if err != nil {
		return 0, err
	}

	var matched uint
	for _, dataRaw := range res {
		flow := parseRawData(dataRaw)
		if match := filter.MatchConntrackFlow(flow); match {
			req2 := h.newConntrackRequest(table, family, nl.IPCTNL_MSG_CT_DELETE, unix.NLM_F_ACK)
			// skip the first 4 byte that are the netfilter header, the newConntrackRequest is adding it already
			req2.AddRawData(dataRaw[4:])
			req2.Execute(unix.NETLINK_NETFILTER, 0)
			matched++
		}
	}

	return matched, nil
}

func (h *Handle) newConntrackRequest(table ConntrackTableType, family InetFamily, operation, flags int) *nl.NetlinkRequest {
	// Create the Netlink request object
	req := h.newNetlinkRequest((int(table)<<8)|operation, flags)
	// Add the netfilter header
	msg := &nl.Nfgenmsg{
		NfgenFamily: uint8(family),
		Version:     nl.NFNETLINK_V0,
		ResId:       0,
	}
	req.AddData(msg)
	return req
}

func (h *Handle) dumpConntrackTable(table ConntrackTableType, family InetFamily) ([][]byte, error) {
	req := h.newConntrackRequest(table, family, nl.IPCTNL_MSG_CT_GET, unix.NLM_F_DUMP)
	return req.Execute(unix.NETLINK_NETFILTER, 0)
}

// The full conntrack flow structure is very complicated and can be found in the file:
// http://git.netfilter.org/libnetfilter_conntrack/tree/include/internal/object.h
// For the time being, the structure below allows to parse and extract the base information of a flow
type ipTuple struct {
	Bytes    uint64
	DstIP    net.IP
	DstPort  uint16
	Packets  uint64
	Protocol uint8
	SrcIP    net.IP
	SrcPort  uint16
}

type ConntrackFlow struct {
	FamilyType uint8
	Forward    ipTuple
	Reverse    ipTuple
	Mark       uint32
}

func (s *ConntrackFlow) String() string {
	// conntrack cmd output:
	// udp      17 src=127.0.0.1 dst=127.0.0.1 sport=4001 dport=1234 packets=5 bytes=532 [UNREPLIED] src=127.0.0.1 dst=127.0.0.1 sport=1234 dport=4001 packets=10 bytes=1078 mark=0
	return fmt.Sprintf("%s\t%d src=%s dst=%s sport=%d dport=%d packets=%d bytes=%d\tsrc=%s dst=%s sport=%d dport=%d packets=%d bytes=%d mark=%d",
		nl.L4ProtoMap[s.Forward.Protocol], s.Forward.Protocol,
		s.Forward.SrcIP.String(), s.Forward.DstIP.String(), s.Forward.SrcPort, s.Forward.DstPort, s.Forward.Packets, s.Forward.Bytes,
		s.Reverse.SrcIP.String(), s.Reverse.DstIP.String(), s.Reverse.SrcPort, s.Reverse.DstPort, s.Reverse.Packets, s.Reverse.Bytes,
		s.Mark)
}

// This method parse the ip tuple structure
// The message structure is the following:
// <len, [CTA_IP_V4_SRC|CTA_IP_V6_SRC], 16 bytes for the IP>
// <len, [CTA_IP_V4_DST|CTA_IP_V6_DST], 16 bytes for the IP>
// <len, NLA_F_NESTED|nl.CTA_TUPLE_PROTO, 1 byte for the protocol, 3 bytes of padding>
// <len, CTA_PROTO_SRC_PORT, 2 bytes for the source port, 2 bytes of padding>
// <len, CTA_PROTO_DST_PORT, 2 bytes for the source port, 2 bytes of padding>
func parseIpTuple(reader *bytes.Reader, tpl *ipTuple) uint8 {
	for i := 0; i < 2; i++ {
		_, t, _, v := parseNfAttrTLV(reader)
		switch t {
		case nl.CTA_IP_V4_SRC, nl.CTA_IP_V6_SRC:
			tpl.SrcIP = v
		case nl.CTA_IP_V4_DST, nl.CTA_IP_V6_DST:
			tpl.DstIP = v
		}
	}
	// Skip the next 4 bytes  nl.NLA_F_NESTED|nl.CTA_TUPLE_PROTO
	reader.Seek(4, seekCurrent)
	_, t, _, v := parseNfAttrTLV(reader)
	if t == nl.CTA_PROTO_NUM {
		tpl.Protocol = uint8(v[0])
	}
	// Skip some padding 3 bytes
	reader.Seek(3, seekCurrent)
	for i := 0; i < 2; i++ {
		_, t, _ := parseNfAttrTL(reader)
		switch t {
		case nl.CTA_PROTO_SRC_PORT:
			parseBERaw16(reader, &tpl.SrcPort)
		case nl.CTA_PROTO_DST_PORT:
			parseBERaw16(reader, &tpl.DstPort)
		}
		// Skip some padding 2 byte
		reader.Seek(2, seekCurrent)
	}
	return tpl.Protocol
}

func parseNfAttrTLV(r *bytes.Reader) (isNested bool, attrType, len uint16, value []byte) {
	isNested, attrType, len = parseNfAttrTL(r)

	value = make([]byte, len)
	binary.Read(r, binary.BigEndian, &value)
	return isNested, attrType, len, value
}

func parseNfAttrTL(r *bytes.Reader) (isNested bool, attrType, len uint16) {
	binary.Read(r, nl.NativeEndian(), &len)
	len -= nl.SizeofNfattr

	binary.Read(r, nl.NativeEndian(), &attrType)
	isNested = (attrType & nl.NLA_F_NESTED) == nl.NLA_F_NESTED
	attrType = attrType & (nl.NLA_F_NESTED - 1)

	return isNested, attrType, len
}

func parseBERaw16(r *bytes.Reader, v *uint16) {
	binary.Read(r, binary.BigEndian, v)
}

func parseBERaw32(r *bytes.Reader, v *uint32) {
	binary.Read(r, binary.BigEndian, v)
}

func parseBERaw64(r *bytes.Reader, v *uint64) {
	binary.Read(r, binary.BigEndian, v)
}

func parseByteAndPacketCounters(r *bytes.Reader) (bytes, packets uint64) {
	for i := 0; i < 2; i++ {
		switch _, t, _ := parseNfAttrTL(r); t {
		case nl.CTA_COUNTERS_BYTES:
			parseBERaw64(r, &bytes)
		case nl.CTA_COUNTERS_PACKETS:
			parseBERaw64(r, &packets)
		default:
			return
		}
	}
	return
}

func parseConnectionMark(r *bytes.Reader) (mark uint32) {
	parseBERaw32(r, &mark)
	return
}

func parseRawData(data []byte) *ConntrackFlow {
	s := &ConntrackFlow{}
	// First there is the Nfgenmsg header
	// consume only the family field
	reader := bytes.NewReader(data)
	binary.Read(reader, nl.NativeEndian(), &s.FamilyType)

	// skip rest of the Netfilter header
	reader.Seek(3, seekCurrent)
	// The message structure is the following:
	// <len, NLA_F_NESTED|CTA_TUPLE_ORIG> 4 bytes
	// <len, NLA_F_NESTED|CTA_TUPLE_IP> 4 bytes
	// flow information of the forward flow
	// <len, NLA_F_NESTED|CTA_TUPLE_REPLY> 4 bytes
	// <len, NLA_F_NESTED|CTA_TUPLE_IP> 4 bytes
	// flow information of the reverse flow
	for reader.Len() > 0 {
		if nested, t, l := parseNfAttrTL(reader); nested {
			switch t {
			case nl.CTA_TUPLE_ORIG:
				if nested, t, _ = parseNfAttrTL(reader); nested && t == nl.CTA_TUPLE_IP {
					parseIpTuple(reader, &s.Forward)
				}
			case nl.CTA_TUPLE_REPLY:
				if nested, t, _ = parseNfAttrTL(reader); nested && t == nl.CTA_TUPLE_IP {
					parseIpTuple(reader, &s.Reverse)
				} else {
					// Header not recognized skip it
					reader.Seek(int64(l), seekCurrent)
				}
			case nl.CTA_COUNTERS_ORIG:
				s.Forward.Bytes, s.Forward.Packets = parseByteAndPacketCounters(reader)
			case nl.CTA_COUNTERS_REPLY:
				s.Reverse.Bytes, s.Reverse.Packets = parseByteAndPacketCounters(reader)
			}
		} else {
			switch t {
			case nl.CTA_MARK:
				s.Mark = parseConnectionMark(reader)
			}
		}
	}
	return s
}

// Conntrack parameters and options:
//   -n, --src-nat ip                      source NAT ip
//   -g, --dst-nat ip                      destination NAT ip
//   -j, --any-nat ip                      source or destination NAT ip
//   -m, --mark mark                       Set mark
//   -c, --secmark secmark                 Set selinux secmark
//   -e, --event-mask eventmask            Event mask, eg. NEW,DESTROY
//   -z, --zero                            Zero counters while listing
//   -o, --output type[,...]               Output format, eg. xml
//   -l, --label label[,...]               conntrack labels

// Common parameters and options:
//   -s, --src, --orig-src ip              Source address from original direction
//   -d, --dst, --orig-dst ip              Destination address from original direction
//   -r, --reply-src ip            Source address from reply direction
//   -q, --reply-dst ip            Destination address from reply direction
//   -p, --protonum proto          Layer 4 Protocol, eg. 'tcp'
//   -f, --family proto            Layer 3 Protocol, eg. 'ipv6'
//   -t, --timeout timeout         Set timeout
//   -u, --status status           Set status, eg. ASSURED
//   -w, --zone value              Set conntrack zone
//   --orig-zone value             Set zone for original direction
//   --reply-zone value            Set zone for reply direction
//   -b, --buffer-size             Netlink socket buffer size
//   --mask-src ip                 Source mask address
//   --mask-dst ip                 Destination mask address

// Filter types
type ConntrackFilterType uint8

const (
	ConntrackOrigSrcIP  = iota                // -orig-src ip    Source address from original direction
	ConntrackOrigDstIP                        // -orig-dst ip    Destination address from original direction
	ConntrackReplySrcIP                       // --reply-src ip  Reply Source IP
	ConntrackReplyDstIP                       // --reply-dst ip  Reply Destination IP
	ConntrackReplyAnyIP                       // Match source or destination reply IP
	ConntrackNatSrcIP   = ConntrackReplySrcIP // deprecated use instead ConntrackReplySrcIP
	ConntrackNatDstIP   = ConntrackReplyDstIP // deprecated use instead ConntrackReplyDstIP
	ConntrackNatAnyIP   = ConntrackReplyAnyIP // deprecated use instaed ConntrackReplyAnyIP
)

type CustomConntrackFilter interface {
	// MatchConntrackFlow applies the filter to the flow and returns true if the flow matches
	// the filter or false otherwise
	MatchConntrackFlow(flow *ConntrackFlow) bool
}

type ConntrackFilter struct {
	ipFilter map[ConntrackFilterType]net.IP
}

// AddIP adds an IP to the conntrack filter
func (f *ConntrackFilter) AddIP(tp ConntrackFilterType, ip net.IP) error {
	if f.ipFilter == nil {
		f.ipFilter = make(map[ConntrackFilterType]net.IP)
	}
	if _, ok := f.ipFilter[tp]; ok {
		return errors.New("Filter attribute already present")
	}
	f.ipFilter[tp] = ip
	return nil
}

// MatchConntrackFlow applies the filter to the flow and returns true if the flow matches the filter
// false otherwise
func (f *ConntrackFilter) MatchConntrackFlow(flow *ConntrackFlow) bool {
	if len(f.ipFilter) == 0 {
		// empty filter always not match
		return false
	}

	match := true
	// -orig-src ip   Source address from original direction
	if elem, found := f.ipFilter[ConntrackOrigSrcIP]; found {
		match = match && elem.Equal(flow.Forward.SrcIP)
	}

	// -orig-dst ip   Destination address from original direction
	if elem, found := f.ipFilter[ConntrackOrigDstIP]; match && found {
		match = match && elem.Equal(flow.Forward.DstIP)
	}

	// -src-nat ip    Source NAT ip
	if elem, found := f.ipFilter[ConntrackReplySrcIP]; match && found {
		match = match && elem.Equal(flow.Reverse.SrcIP)
	}

	// -dst-nat ip    Destination NAT ip
	if elem, found := f.ipFilter[ConntrackReplyDstIP]; match && found {
		match = match && elem.Equal(flow.Reverse.DstIP)
	}

	// Match source or destination reply IP
	if elem, found := f.ipFilter[ConntrackReplyAnyIP]; match && found {
		match = match && (elem.Equal(flow.Reverse.SrcIP) || elem.Equal(flow.Reverse.DstIP))
	}

	return match
}

var _ CustomConntrackFilter = (*ConntrackFilter)(nil)