]>
Commit | Line | Data |
---|---|---|
e2be04c7 | 1 | /* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */ |
daedfb22 AS |
2 | /* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com |
3 | * | |
4 | * This program is free software; you can redistribute it and/or | |
5 | * modify it under the terms of version 2 of the GNU General Public | |
6 | * License as published by the Free Software Foundation. | |
7 | */ | |
8 | #ifndef _UAPI__LINUX_BPF_H__ | |
9 | #define _UAPI__LINUX_BPF_H__ | |
10 | ||
11 | #include <linux/types.h> | |
c15952dc | 12 | #include <linux/bpf_common.h> |
daedfb22 AS |
13 | |
14 | /* Extended instruction set based on top of classic BPF */ | |
15 | ||
16 | /* instruction classes */ | |
17 | #define BPF_ALU64 0x07 /* alu mode in double word width */ | |
18 | ||
19 | /* ld/ldx fields */ | |
cb5f7334 | 20 | #define BPF_DW 0x18 /* double word (64-bit) */ |
daedfb22 AS |
21 | #define BPF_XADD 0xc0 /* exclusive add */ |
22 | ||
23 | /* alu/jmp fields */ | |
24 | #define BPF_MOV 0xb0 /* mov reg to reg */ | |
25 | #define BPF_ARSH 0xc0 /* sign extending arithmetic shift right */ | |
26 | ||
27 | /* change endianness of a register */ | |
28 | #define BPF_END 0xd0 /* flags for endianness conversion: */ | |
29 | #define BPF_TO_LE 0x00 /* convert to little-endian */ | |
30 | #define BPF_TO_BE 0x08 /* convert to big-endian */ | |
31 | #define BPF_FROM_LE BPF_TO_LE | |
32 | #define BPF_FROM_BE BPF_TO_BE | |
33 | ||
92b31a9a | 34 | /* jmp encodings */ |
daedfb22 | 35 | #define BPF_JNE 0x50 /* jump != */ |
92b31a9a DB |
36 | #define BPF_JLT 0xa0 /* LT is unsigned, '<' */ |
37 | #define BPF_JLE 0xb0 /* LE is unsigned, '<=' */ | |
daedfb22 AS |
38 | #define BPF_JSGT 0x60 /* SGT is signed '>', GT in x86 */ |
39 | #define BPF_JSGE 0x70 /* SGE is signed '>=', GE in x86 */ | |
92b31a9a DB |
40 | #define BPF_JSLT 0xc0 /* SLT is signed, '<' */ |
41 | #define BPF_JSLE 0xd0 /* SLE is signed, '<=' */ | |
daedfb22 AS |
42 | #define BPF_CALL 0x80 /* function call */ |
43 | #define BPF_EXIT 0x90 /* function return */ | |
44 | ||
45 | /* Register numbers */ | |
46 | enum { | |
47 | BPF_REG_0 = 0, | |
48 | BPF_REG_1, | |
49 | BPF_REG_2, | |
50 | BPF_REG_3, | |
51 | BPF_REG_4, | |
52 | BPF_REG_5, | |
53 | BPF_REG_6, | |
54 | BPF_REG_7, | |
55 | BPF_REG_8, | |
56 | BPF_REG_9, | |
57 | BPF_REG_10, | |
58 | __MAX_BPF_REG, | |
59 | }; | |
60 | ||
61 | /* BPF has 10 general purpose 64-bit registers and stack frame. */ | |
62 | #define MAX_BPF_REG __MAX_BPF_REG | |
63 | ||
64 | struct bpf_insn { | |
65 | __u8 code; /* opcode */ | |
66 | __u8 dst_reg:4; /* dest register */ | |
67 | __u8 src_reg:4; /* source register */ | |
68 | __s16 off; /* signed offset */ | |
69 | __s32 imm; /* signed immediate constant */ | |
70 | }; | |
71 | ||
b95a5c4d DM |
72 | /* Key of an a BPF_MAP_TYPE_LPM_TRIE entry */ |
73 | struct bpf_lpm_trie_key { | |
74 | __u32 prefixlen; /* up to 32 for AF_INET, 128 for AF_INET6 */ | |
75 | __u8 data[0]; /* Arbitrary size */ | |
76 | }; | |
77 | ||
b2197755 | 78 | /* BPF syscall commands, see bpf(2) man-page for details. */ |
99c55f7d | 79 | enum bpf_cmd { |
99c55f7d | 80 | BPF_MAP_CREATE, |
db20fd2b | 81 | BPF_MAP_LOOKUP_ELEM, |
db20fd2b | 82 | BPF_MAP_UPDATE_ELEM, |
db20fd2b | 83 | BPF_MAP_DELETE_ELEM, |
db20fd2b | 84 | BPF_MAP_GET_NEXT_KEY, |
09756af4 | 85 | BPF_PROG_LOAD, |
b2197755 DB |
86 | BPF_OBJ_PIN, |
87 | BPF_OBJ_GET, | |
f4324551 DM |
88 | BPF_PROG_ATTACH, |
89 | BPF_PROG_DETACH, | |
1cf1cae9 | 90 | BPF_PROG_TEST_RUN, |
34ad5580 MKL |
91 | BPF_PROG_GET_NEXT_ID, |
92 | BPF_MAP_GET_NEXT_ID, | |
b16d9aa4 | 93 | BPF_PROG_GET_FD_BY_ID, |
bd5f5f4e | 94 | BPF_MAP_GET_FD_BY_ID, |
1e270976 | 95 | BPF_OBJ_GET_INFO_BY_FD, |
468e2f64 | 96 | BPF_PROG_QUERY, |
c4f6699d | 97 | BPF_RAW_TRACEPOINT_OPEN, |
99c55f7d AS |
98 | }; |
99 | ||
100 | enum bpf_map_type { | |
101 | BPF_MAP_TYPE_UNSPEC, | |
0f8e4bd8 | 102 | BPF_MAP_TYPE_HASH, |
28fbcfa0 | 103 | BPF_MAP_TYPE_ARRAY, |
04fd61ab | 104 | BPF_MAP_TYPE_PROG_ARRAY, |
ea317b26 | 105 | BPF_MAP_TYPE_PERF_EVENT_ARRAY, |
824bd0ce | 106 | BPF_MAP_TYPE_PERCPU_HASH, |
a10423b8 | 107 | BPF_MAP_TYPE_PERCPU_ARRAY, |
d5a3b1f6 | 108 | BPF_MAP_TYPE_STACK_TRACE, |
4ed8ec52 | 109 | BPF_MAP_TYPE_CGROUP_ARRAY, |
29ba732a | 110 | BPF_MAP_TYPE_LRU_HASH, |
8f844938 | 111 | BPF_MAP_TYPE_LRU_PERCPU_HASH, |
b95a5c4d | 112 | BPF_MAP_TYPE_LPM_TRIE, |
56f668df | 113 | BPF_MAP_TYPE_ARRAY_OF_MAPS, |
bcc6b1b7 | 114 | BPF_MAP_TYPE_HASH_OF_MAPS, |
546ac1ff | 115 | BPF_MAP_TYPE_DEVMAP, |
174a79ff | 116 | BPF_MAP_TYPE_SOCKMAP, |
6710e112 | 117 | BPF_MAP_TYPE_CPUMAP, |
99c55f7d AS |
118 | }; |
119 | ||
09756af4 AS |
120 | enum bpf_prog_type { |
121 | BPF_PROG_TYPE_UNSPEC, | |
ddd872bc | 122 | BPF_PROG_TYPE_SOCKET_FILTER, |
2541517c | 123 | BPF_PROG_TYPE_KPROBE, |
96be4325 | 124 | BPF_PROG_TYPE_SCHED_CLS, |
94caee8c | 125 | BPF_PROG_TYPE_SCHED_ACT, |
98b5c2c6 | 126 | BPF_PROG_TYPE_TRACEPOINT, |
6a773a15 | 127 | BPF_PROG_TYPE_XDP, |
0515e599 | 128 | BPF_PROG_TYPE_PERF_EVENT, |
0e33661d | 129 | BPF_PROG_TYPE_CGROUP_SKB, |
61023658 | 130 | BPF_PROG_TYPE_CGROUP_SOCK, |
3a0af8fd TG |
131 | BPF_PROG_TYPE_LWT_IN, |
132 | BPF_PROG_TYPE_LWT_OUT, | |
133 | BPF_PROG_TYPE_LWT_XMIT, | |
40304b2a | 134 | BPF_PROG_TYPE_SOCK_OPS, |
b005fd18 | 135 | BPF_PROG_TYPE_SK_SKB, |
ebc614f6 | 136 | BPF_PROG_TYPE_CGROUP_DEVICE, |
4f738adb | 137 | BPF_PROG_TYPE_SK_MSG, |
c4f6699d | 138 | BPF_PROG_TYPE_RAW_TRACEPOINT, |
4fbac77d | 139 | BPF_PROG_TYPE_CGROUP_SOCK_ADDR, |
09756af4 AS |
140 | }; |
141 | ||
0e33661d DM |
142 | enum bpf_attach_type { |
143 | BPF_CGROUP_INET_INGRESS, | |
144 | BPF_CGROUP_INET_EGRESS, | |
61023658 | 145 | BPF_CGROUP_INET_SOCK_CREATE, |
40304b2a | 146 | BPF_CGROUP_SOCK_OPS, |
464bc0fd JF |
147 | BPF_SK_SKB_STREAM_PARSER, |
148 | BPF_SK_SKB_STREAM_VERDICT, | |
ebc614f6 | 149 | BPF_CGROUP_DEVICE, |
4f738adb | 150 | BPF_SK_MSG_VERDICT, |
4fbac77d AI |
151 | BPF_CGROUP_INET4_BIND, |
152 | BPF_CGROUP_INET6_BIND, | |
d74bad4e AI |
153 | BPF_CGROUP_INET4_CONNECT, |
154 | BPF_CGROUP_INET6_CONNECT, | |
aac3fc32 AI |
155 | BPF_CGROUP_INET4_POST_BIND, |
156 | BPF_CGROUP_INET6_POST_BIND, | |
0e33661d DM |
157 | __MAX_BPF_ATTACH_TYPE |
158 | }; | |
159 | ||
160 | #define MAX_BPF_ATTACH_TYPE __MAX_BPF_ATTACH_TYPE | |
161 | ||
324bda9e AS |
162 | /* cgroup-bpf attach flags used in BPF_PROG_ATTACH command |
163 | * | |
164 | * NONE(default): No further bpf programs allowed in the subtree. | |
165 | * | |
166 | * BPF_F_ALLOW_OVERRIDE: If a sub-cgroup installs some bpf program, | |
167 | * the program in this cgroup yields to sub-cgroup program. | |
168 | * | |
169 | * BPF_F_ALLOW_MULTI: If a sub-cgroup installs some bpf program, | |
170 | * that cgroup program gets run in addition to the program in this cgroup. | |
171 | * | |
172 | * Only one program is allowed to be attached to a cgroup with | |
173 | * NONE or BPF_F_ALLOW_OVERRIDE flag. | |
174 | * Attaching another program on top of NONE or BPF_F_ALLOW_OVERRIDE will | |
175 | * release old program and attach the new one. Attach flags has to match. | |
176 | * | |
177 | * Multiple programs are allowed to be attached to a cgroup with | |
178 | * BPF_F_ALLOW_MULTI flag. They are executed in FIFO order | |
179 | * (those that were attached first, run first) | |
180 | * The programs of sub-cgroup are executed first, then programs of | |
181 | * this cgroup and then programs of parent cgroup. | |
182 | * When children program makes decision (like picking TCP CA or sock bind) | |
183 | * parent program has a chance to override it. | |
184 | * | |
185 | * A cgroup with MULTI or OVERRIDE flag allows any attach flags in sub-cgroups. | |
186 | * A cgroup with NONE doesn't allow any programs in sub-cgroups. | |
187 | * Ex1: | |
188 | * cgrp1 (MULTI progs A, B) -> | |
189 | * cgrp2 (OVERRIDE prog C) -> | |
190 | * cgrp3 (MULTI prog D) -> | |
191 | * cgrp4 (OVERRIDE prog E) -> | |
192 | * cgrp5 (NONE prog F) | |
193 | * the event in cgrp5 triggers execution of F,D,A,B in that order. | |
194 | * if prog F is detached, the execution is E,D,A,B | |
195 | * if prog F and D are detached, the execution is E,A,B | |
196 | * if prog F, E and D are detached, the execution is C,A,B | |
197 | * | |
198 | * All eligible programs are executed regardless of return code from | |
199 | * earlier programs. | |
7f677633 AS |
200 | */ |
201 | #define BPF_F_ALLOW_OVERRIDE (1U << 0) | |
324bda9e | 202 | #define BPF_F_ALLOW_MULTI (1U << 1) |
7f677633 | 203 | |
e07b98d9 DM |
204 | /* If BPF_F_STRICT_ALIGNMENT is used in BPF_PROG_LOAD command, the |
205 | * verifier will perform strict alignment checking as if the kernel | |
206 | * has been built with CONFIG_EFFICIENT_UNALIGNED_ACCESS not set, | |
207 | * and NET_IP_ALIGN defined to 2. | |
208 | */ | |
209 | #define BPF_F_STRICT_ALIGNMENT (1U << 0) | |
210 | ||
cc8b0b92 | 211 | /* when bpf_ldimm64->src_reg == BPF_PSEUDO_MAP_FD, bpf_ldimm64->imm == fd */ |
f1a66f85 DB |
212 | #define BPF_PSEUDO_MAP_FD 1 |
213 | ||
cc8b0b92 AS |
214 | /* when bpf_call->src_reg == BPF_PSEUDO_CALL, bpf_call->imm == pc-relative |
215 | * offset to another bpf function | |
216 | */ | |
217 | #define BPF_PSEUDO_CALL 1 | |
218 | ||
3274f520 AS |
219 | /* flags for BPF_MAP_UPDATE_ELEM command */ |
220 | #define BPF_ANY 0 /* create new element or update existing */ | |
221 | #define BPF_NOEXIST 1 /* create new element if it didn't exist */ | |
222 | #define BPF_EXIST 2 /* update existing element */ | |
223 | ||
96eabe7a | 224 | /* flags for BPF_MAP_CREATE command */ |
6c905981 | 225 | #define BPF_F_NO_PREALLOC (1U << 0) |
29ba732a | 226 | /* Instead of having one common LRU list in the |
8f844938 | 227 | * BPF_MAP_TYPE_LRU_[PERCPU_]HASH map, use a percpu LRU list |
29ba732a MKL |
228 | * which can scale and perform better. |
229 | * Note, the LRU nodes (including free nodes) cannot be moved | |
230 | * across different LRU lists. | |
231 | */ | |
232 | #define BPF_F_NO_COMMON_LRU (1U << 1) | |
96eabe7a MKL |
233 | /* Specify numa node during map creation */ |
234 | #define BPF_F_NUMA_NODE (1U << 2) | |
6c905981 | 235 | |
468e2f64 AS |
236 | /* flags for BPF_PROG_QUERY */ |
237 | #define BPF_F_QUERY_EFFECTIVE (1U << 0) | |
238 | ||
cb4d2b3f MKL |
239 | #define BPF_OBJ_NAME_LEN 16U |
240 | ||
6e71b04a CF |
241 | /* Flags for accessing BPF object */ |
242 | #define BPF_F_RDONLY (1U << 3) | |
243 | #define BPF_F_WRONLY (1U << 4) | |
244 | ||
615755a7 SL |
245 | /* Flag for stack_map, store build_id+offset instead of pointer */ |
246 | #define BPF_F_STACK_BUILD_ID (1U << 5) | |
247 | ||
248 | enum bpf_stack_build_id_status { | |
249 | /* user space need an empty entry to identify end of a trace */ | |
250 | BPF_STACK_BUILD_ID_EMPTY = 0, | |
251 | /* with valid build_id and offset */ | |
252 | BPF_STACK_BUILD_ID_VALID = 1, | |
253 | /* couldn't get build_id, fallback to ip */ | |
254 | BPF_STACK_BUILD_ID_IP = 2, | |
255 | }; | |
256 | ||
257 | #define BPF_BUILD_ID_SIZE 20 | |
258 | struct bpf_stack_build_id { | |
259 | __s32 status; | |
260 | unsigned char build_id[BPF_BUILD_ID_SIZE]; | |
261 | union { | |
262 | __u64 offset; | |
263 | __u64 ip; | |
264 | }; | |
265 | }; | |
266 | ||
99c55f7d AS |
267 | union bpf_attr { |
268 | struct { /* anonymous struct used by BPF_MAP_CREATE command */ | |
269 | __u32 map_type; /* one of enum bpf_map_type */ | |
270 | __u32 key_size; /* size of key in bytes */ | |
271 | __u32 value_size; /* size of value in bytes */ | |
272 | __u32 max_entries; /* max number of entries in a map */ | |
96eabe7a MKL |
273 | __u32 map_flags; /* BPF_MAP_CREATE related |
274 | * flags defined above. | |
275 | */ | |
56f668df | 276 | __u32 inner_map_fd; /* fd pointing to the inner map */ |
96eabe7a MKL |
277 | __u32 numa_node; /* numa node (effective only if |
278 | * BPF_F_NUMA_NODE is set). | |
279 | */ | |
067cae47 | 280 | char map_name[BPF_OBJ_NAME_LEN]; |
a3884572 | 281 | __u32 map_ifindex; /* ifindex of netdev to create on */ |
99c55f7d | 282 | }; |
db20fd2b AS |
283 | |
284 | struct { /* anonymous struct used by BPF_MAP_*_ELEM commands */ | |
285 | __u32 map_fd; | |
286 | __aligned_u64 key; | |
287 | union { | |
288 | __aligned_u64 value; | |
289 | __aligned_u64 next_key; | |
290 | }; | |
3274f520 | 291 | __u64 flags; |
db20fd2b | 292 | }; |
09756af4 AS |
293 | |
294 | struct { /* anonymous struct used by BPF_PROG_LOAD command */ | |
295 | __u32 prog_type; /* one of enum bpf_prog_type */ | |
296 | __u32 insn_cnt; | |
297 | __aligned_u64 insns; | |
298 | __aligned_u64 license; | |
cbd35700 AS |
299 | __u32 log_level; /* verbosity level of verifier */ |
300 | __u32 log_size; /* size of user buffer */ | |
301 | __aligned_u64 log_buf; /* user supplied buffer */ | |
2541517c | 302 | __u32 kern_version; /* checked when prog_type=kprobe */ |
e07b98d9 | 303 | __u32 prog_flags; |
067cae47 | 304 | char prog_name[BPF_OBJ_NAME_LEN]; |
1f6f4cb7 | 305 | __u32 prog_ifindex; /* ifindex of netdev to prep for */ |
5e43f899 AI |
306 | /* For some prog types expected attach type must be known at |
307 | * load time to verify attach type specific parts of prog | |
308 | * (context accesses, allowed helpers, etc). | |
309 | */ | |
310 | __u32 expected_attach_type; | |
09756af4 | 311 | }; |
b2197755 DB |
312 | |
313 | struct { /* anonymous struct used by BPF_OBJ_* commands */ | |
314 | __aligned_u64 pathname; | |
315 | __u32 bpf_fd; | |
6e71b04a | 316 | __u32 file_flags; |
b2197755 | 317 | }; |
f4324551 DM |
318 | |
319 | struct { /* anonymous struct used by BPF_PROG_ATTACH/DETACH commands */ | |
320 | __u32 target_fd; /* container object to attach to */ | |
321 | __u32 attach_bpf_fd; /* eBPF program to attach */ | |
322 | __u32 attach_type; | |
7f677633 | 323 | __u32 attach_flags; |
f4324551 | 324 | }; |
1cf1cae9 AS |
325 | |
326 | struct { /* anonymous struct used by BPF_PROG_TEST_RUN command */ | |
327 | __u32 prog_fd; | |
328 | __u32 retval; | |
329 | __u32 data_size_in; | |
330 | __u32 data_size_out; | |
331 | __aligned_u64 data_in; | |
332 | __aligned_u64 data_out; | |
333 | __u32 repeat; | |
334 | __u32 duration; | |
335 | } test; | |
34ad5580 | 336 | |
b16d9aa4 MKL |
337 | struct { /* anonymous struct used by BPF_*_GET_*_ID */ |
338 | union { | |
339 | __u32 start_id; | |
340 | __u32 prog_id; | |
bd5f5f4e | 341 | __u32 map_id; |
b16d9aa4 | 342 | }; |
34ad5580 | 343 | __u32 next_id; |
6e71b04a | 344 | __u32 open_flags; |
34ad5580 | 345 | }; |
1e270976 MKL |
346 | |
347 | struct { /* anonymous struct used by BPF_OBJ_GET_INFO_BY_FD */ | |
348 | __u32 bpf_fd; | |
349 | __u32 info_len; | |
350 | __aligned_u64 info; | |
351 | } info; | |
468e2f64 AS |
352 | |
353 | struct { /* anonymous struct used by BPF_PROG_QUERY command */ | |
354 | __u32 target_fd; /* container object to query */ | |
355 | __u32 attach_type; | |
356 | __u32 query_flags; | |
357 | __u32 attach_flags; | |
358 | __aligned_u64 prog_ids; | |
359 | __u32 prog_cnt; | |
360 | } query; | |
c4f6699d AS |
361 | |
362 | struct { | |
363 | __u64 name; | |
364 | __u32 prog_fd; | |
365 | } raw_tracepoint; | |
99c55f7d AS |
366 | } __attribute__((aligned(8))); |
367 | ||
ebb676da TG |
368 | /* BPF helper function descriptions: |
369 | * | |
370 | * void *bpf_map_lookup_elem(&map, &key) | |
371 | * Return: Map value or NULL | |
372 | * | |
373 | * int bpf_map_update_elem(&map, &key, &value, flags) | |
374 | * Return: 0 on success or negative error | |
375 | * | |
376 | * int bpf_map_delete_elem(&map, &key) | |
377 | * Return: 0 on success or negative error | |
378 | * | |
379 | * int bpf_probe_read(void *dst, int size, void *src) | |
380 | * Return: 0 on success or negative error | |
381 | * | |
382 | * u64 bpf_ktime_get_ns(void) | |
383 | * Return: current ktime | |
384 | * | |
385 | * int bpf_trace_printk(const char *fmt, int fmt_size, ...) | |
386 | * Return: length of buffer written or negative error | |
387 | * | |
388 | * u32 bpf_prandom_u32(void) | |
389 | * Return: random value | |
390 | * | |
391 | * u32 bpf_raw_smp_processor_id(void) | |
392 | * Return: SMP processor ID | |
393 | * | |
394 | * int bpf_skb_store_bytes(skb, offset, from, len, flags) | |
395 | * store bytes into packet | |
396 | * @skb: pointer to skb | |
397 | * @offset: offset within packet from skb->mac_header | |
398 | * @from: pointer where to copy bytes from | |
399 | * @len: number of bytes to store into packet | |
400 | * @flags: bit 0 - if true, recompute skb->csum | |
401 | * other bits - reserved | |
402 | * Return: 0 on success or negative error | |
403 | * | |
404 | * int bpf_l3_csum_replace(skb, offset, from, to, flags) | |
405 | * recompute IP checksum | |
406 | * @skb: pointer to skb | |
407 | * @offset: offset within packet where IP checksum is located | |
408 | * @from: old value of header field | |
409 | * @to: new value of header field | |
410 | * @flags: bits 0-3 - size of header field | |
411 | * other bits - reserved | |
412 | * Return: 0 on success or negative error | |
413 | * | |
414 | * int bpf_l4_csum_replace(skb, offset, from, to, flags) | |
415 | * recompute TCP/UDP checksum | |
416 | * @skb: pointer to skb | |
417 | * @offset: offset within packet where TCP/UDP checksum is located | |
418 | * @from: old value of header field | |
419 | * @to: new value of header field | |
420 | * @flags: bits 0-3 - size of header field | |
421 | * bit 4 - is pseudo header | |
422 | * other bits - reserved | |
423 | * Return: 0 on success or negative error | |
424 | * | |
425 | * int bpf_tail_call(ctx, prog_array_map, index) | |
426 | * jump into another BPF program | |
427 | * @ctx: context pointer passed to next program | |
428 | * @prog_array_map: pointer to map which type is BPF_MAP_TYPE_PROG_ARRAY | |
90caccdd | 429 | * @index: 32-bit index inside array that selects specific program to run |
ebb676da TG |
430 | * Return: 0 on success or negative error |
431 | * | |
432 | * int bpf_clone_redirect(skb, ifindex, flags) | |
433 | * redirect to another netdev | |
434 | * @skb: pointer to skb | |
435 | * @ifindex: ifindex of the net device | |
436 | * @flags: bit 0 - if set, redirect to ingress instead of egress | |
437 | * other bits - reserved | |
438 | * Return: 0 on success or negative error | |
439 | * | |
440 | * u64 bpf_get_current_pid_tgid(void) | |
441 | * Return: current->tgid << 32 | current->pid | |
442 | * | |
443 | * u64 bpf_get_current_uid_gid(void) | |
444 | * Return: current_gid << 32 | current_uid | |
445 | * | |
446 | * int bpf_get_current_comm(char *buf, int size_of_buf) | |
447 | * stores current->comm into buf | |
448 | * Return: 0 on success or negative error | |
449 | * | |
450 | * u32 bpf_get_cgroup_classid(skb) | |
451 | * retrieve a proc's classid | |
452 | * @skb: pointer to skb | |
453 | * Return: classid if != 0 | |
454 | * | |
455 | * int bpf_skb_vlan_push(skb, vlan_proto, vlan_tci) | |
456 | * Return: 0 on success or negative error | |
457 | * | |
458 | * int bpf_skb_vlan_pop(skb) | |
459 | * Return: 0 on success or negative error | |
460 | * | |
461 | * int bpf_skb_get_tunnel_key(skb, key, size, flags) | |
462 | * int bpf_skb_set_tunnel_key(skb, key, size, flags) | |
463 | * retrieve or populate tunnel metadata | |
464 | * @skb: pointer to skb | |
465 | * @key: pointer to 'struct bpf_tunnel_key' | |
466 | * @size: size of 'struct bpf_tunnel_key' | |
467 | * @flags: room for future extensions | |
468 | * Return: 0 on success or negative error | |
469 | * | |
b7d3ed5b TQ |
470 | * u64 bpf_perf_event_read(map, flags) |
471 | * read perf event counter value | |
472 | * @map: pointer to perf_event_array map | |
473 | * @flags: index of event in the map or bitmask flags | |
474 | * Return: value of perf event counter read or error code | |
ebb676da TG |
475 | * |
476 | * int bpf_redirect(ifindex, flags) | |
477 | * redirect to another netdev | |
478 | * @ifindex: ifindex of the net device | |
56ce097c JF |
479 | * @flags: |
480 | * cls_bpf: | |
481 | * bit 0 - if set, redirect to ingress instead of egress | |
482 | * other bits - reserved | |
483 | * xdp_bpf: | |
484 | * all bits - reserved | |
485 | * Return: cls_bpf: TC_ACT_REDIRECT on success or TC_ACT_SHOT on error | |
486 | * xdp_bfp: XDP_REDIRECT on success or XDP_ABORT on error | |
487 | * int bpf_redirect_map(map, key, flags) | |
97f91a7c | 488 | * redirect to endpoint in map |
56ce097c | 489 | * @map: pointer to dev map |
97f91a7c | 490 | * @key: index in map to lookup |
97f91a7c | 491 | * @flags: -- |
56ce097c | 492 | * Return: XDP_REDIRECT on success or XDP_ABORT on error |
ebb676da TG |
493 | * |
494 | * u32 bpf_get_route_realm(skb) | |
495 | * retrieve a dst's tclassid | |
496 | * @skb: pointer to skb | |
497 | * Return: realm if != 0 | |
498 | * | |
b7d3ed5b | 499 | * int bpf_perf_event_output(ctx, map, flags, data, size) |
ebb676da TG |
500 | * output perf raw sample |
501 | * @ctx: struct pt_regs* | |
502 | * @map: pointer to perf_event_array map | |
b7d3ed5b | 503 | * @flags: index of event in the map or bitmask flags |
ebb676da TG |
504 | * @data: data on stack to be output as raw data |
505 | * @size: size of data | |
506 | * Return: 0 on success or negative error | |
507 | * | |
508 | * int bpf_get_stackid(ctx, map, flags) | |
509 | * walk user or kernel stack and return id | |
510 | * @ctx: struct pt_regs* | |
511 | * @map: pointer to stack_trace map | |
512 | * @flags: bits 0-7 - numer of stack frames to skip | |
513 | * bit 8 - collect user stack instead of kernel | |
514 | * bit 9 - compare stacks by hash only | |
515 | * bit 10 - if two different stacks hash into the same stackid | |
516 | * discard old | |
517 | * other bits - reserved | |
518 | * Return: >= 0 stackid on success or negative error | |
519 | * | |
520 | * s64 bpf_csum_diff(from, from_size, to, to_size, seed) | |
521 | * calculate csum diff | |
522 | * @from: raw from buffer | |
523 | * @from_size: length of from buffer | |
524 | * @to: raw to buffer | |
525 | * @to_size: length of to buffer | |
526 | * @seed: optional seed | |
527 | * Return: csum result or negative error code | |
528 | * | |
529 | * int bpf_skb_get_tunnel_opt(skb, opt, size) | |
530 | * retrieve tunnel options metadata | |
531 | * @skb: pointer to skb | |
532 | * @opt: pointer to raw tunnel option data | |
533 | * @size: size of @opt | |
534 | * Return: option size | |
535 | * | |
536 | * int bpf_skb_set_tunnel_opt(skb, opt, size) | |
537 | * populate tunnel options metadata | |
538 | * @skb: pointer to skb | |
539 | * @opt: pointer to raw tunnel option data | |
540 | * @size: size of @opt | |
541 | * Return: 0 on success or negative error | |
542 | * | |
543 | * int bpf_skb_change_proto(skb, proto, flags) | |
544 | * Change protocol of the skb. Currently supported is v4 -> v6, | |
545 | * v6 -> v4 transitions. The helper will also resize the skb. eBPF | |
546 | * program is expected to fill the new headers via skb_store_bytes | |
547 | * and lX_csum_replace. | |
548 | * @skb: pointer to skb | |
549 | * @proto: new skb->protocol type | |
550 | * @flags: reserved | |
551 | * Return: 0 on success or negative error | |
552 | * | |
553 | * int bpf_skb_change_type(skb, type) | |
554 | * Change packet type of skb. | |
555 | * @skb: pointer to skb | |
556 | * @type: new skb->pkt_type type | |
557 | * Return: 0 on success or negative error | |
558 | * | |
559 | * int bpf_skb_under_cgroup(skb, map, index) | |
560 | * Check cgroup2 membership of skb | |
561 | * @skb: pointer to skb | |
562 | * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type | |
563 | * @index: index of the cgroup in the bpf_map | |
564 | * Return: | |
565 | * == 0 skb failed the cgroup2 descendant test | |
566 | * == 1 skb succeeded the cgroup2 descendant test | |
567 | * < 0 error | |
568 | * | |
569 | * u32 bpf_get_hash_recalc(skb) | |
570 | * Retrieve and possibly recalculate skb->hash. | |
571 | * @skb: pointer to skb | |
572 | * Return: hash | |
573 | * | |
574 | * u64 bpf_get_current_task(void) | |
575 | * Returns current task_struct | |
576 | * Return: current | |
577 | * | |
578 | * int bpf_probe_write_user(void *dst, void *src, int len) | |
579 | * safely attempt to write to a location | |
580 | * @dst: destination address in userspace | |
581 | * @src: source address on stack | |
582 | * @len: number of bytes to copy | |
583 | * Return: 0 on success or negative error | |
584 | * | |
585 | * int bpf_current_task_under_cgroup(map, index) | |
586 | * Check cgroup2 membership of current task | |
587 | * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type | |
588 | * @index: index of the cgroup in the bpf_map | |
589 | * Return: | |
590 | * == 0 current failed the cgroup2 descendant test | |
591 | * == 1 current succeeded the cgroup2 descendant test | |
592 | * < 0 error | |
593 | * | |
594 | * int bpf_skb_change_tail(skb, len, flags) | |
595 | * The helper will resize the skb to the given new size, to be used f.e. | |
596 | * with control messages. | |
597 | * @skb: pointer to skb | |
598 | * @len: new skb length | |
599 | * @flags: reserved | |
600 | * Return: 0 on success or negative error | |
601 | * | |
602 | * int bpf_skb_pull_data(skb, len) | |
603 | * The helper will pull in non-linear data in case the skb is non-linear | |
604 | * and not all of len are part of the linear section. Only needed for | |
605 | * read/write with direct packet access. | |
606 | * @skb: pointer to skb | |
607 | * @len: len to make read/writeable | |
608 | * Return: 0 on success or negative error | |
609 | * | |
610 | * s64 bpf_csum_update(skb, csum) | |
611 | * Adds csum into skb->csum in case of CHECKSUM_COMPLETE. | |
612 | * @skb: pointer to skb | |
613 | * @csum: csum to add | |
614 | * Return: csum on success or negative error | |
615 | * | |
616 | * void bpf_set_hash_invalid(skb) | |
617 | * Invalidate current skb->hash. | |
618 | * @skb: pointer to skb | |
619 | * | |
620 | * int bpf_get_numa_node_id() | |
621 | * Return: Id of current NUMA node. | |
3a0af8fd TG |
622 | * |
623 | * int bpf_skb_change_head() | |
624 | * Grows headroom of skb and adjusts MAC header offset accordingly. | |
625 | * Will extends/reallocae as required automatically. | |
626 | * May change skb data pointer and will thus invalidate any check | |
627 | * performed for direct packet access. | |
628 | * @skb: pointer to skb | |
629 | * @len: length of header to be pushed in front | |
630 | * @flags: Flags (unused for now) | |
631 | * Return: 0 on success or negative error | |
17bedab2 MKL |
632 | * |
633 | * int bpf_xdp_adjust_head(xdp_md, delta) | |
634 | * Adjust the xdp_md.data by delta | |
635 | * @xdp_md: pointer to xdp_md | |
636 | * @delta: An positive/negative integer to be added to xdp_md.data | |
637 | * Return: 0 on success or negative on error | |
a5e8c070 GB |
638 | * |
639 | * int bpf_probe_read_str(void *dst, int size, const void *unsafe_ptr) | |
640 | * Copy a NUL terminated string from unsafe address. In case the string | |
641 | * length is smaller than size, the target is not padded with further NUL | |
642 | * bytes. In case the string length is larger than size, just count-1 | |
643 | * bytes are copied and the last byte is set to NUL. | |
644 | * @dst: destination address | |
645 | * @size: maximum number of bytes to copy, including the trailing NUL | |
646 | * @unsafe_ptr: unsafe address | |
647 | * Return: | |
648 | * > 0 length of the string including the trailing NUL on success | |
649 | * < 0 error | |
91b8270f | 650 | * |
3c60a531 | 651 | * u64 bpf_get_socket_cookie(skb) |
91b8270f CF |
652 | * Get the cookie for the socket stored inside sk_buff. |
653 | * @skb: pointer to skb | |
654 | * Return: 8 Bytes non-decreasing number on success or 0 if the socket | |
655 | * field is missing inside sk_buff | |
6acc5c29 CF |
656 | * |
657 | * u32 bpf_get_socket_uid(skb) | |
658 | * Get the owner uid of the socket stored inside sk_buff. | |
659 | * @skb: pointer to skb | |
5d4e3443 | 660 | * Return: uid of the socket owner on success or overflowuid if failed. |
ded092cd DB |
661 | * |
662 | * u32 bpf_set_hash(skb, hash) | |
663 | * Set full skb->hash. | |
664 | * @skb: pointer to skb | |
665 | * @hash: hash to set | |
8c4b4c7e LB |
666 | * |
667 | * int bpf_setsockopt(bpf_socket, level, optname, optval, optlen) | |
668 | * Calls setsockopt. Not all opts are available, only those with | |
669 | * integer optvals plus TCP_CONGESTION. | |
cd86d1fd | 670 | * Supported levels: SOL_SOCKET and IPPROTO_TCP |
8c4b4c7e | 671 | * @bpf_socket: pointer to bpf_socket |
cd86d1fd | 672 | * @level: SOL_SOCKET or IPPROTO_TCP |
8c4b4c7e LB |
673 | * @optname: option name |
674 | * @optval: pointer to option value | |
cd86d1fd LB |
675 | * @optlen: length of optval in bytes |
676 | * Return: 0 or negative error | |
677 | * | |
678 | * int bpf_getsockopt(bpf_socket, level, optname, optval, optlen) | |
679 | * Calls getsockopt. Not all opts are available. | |
680 | * Supported levels: IPPROTO_TCP | |
681 | * @bpf_socket: pointer to bpf_socket | |
682 | * @level: IPPROTO_TCP | |
683 | * @optname: option name | |
684 | * @optval: pointer to option value | |
685 | * @optlen: length of optval in bytes | |
8c4b4c7e | 686 | * Return: 0 or negative error |
2be7e212 | 687 | * |
b13d8807 LB |
688 | * int bpf_sock_ops_cb_flags_set(bpf_sock_ops, flags) |
689 | * Set callback flags for sock_ops | |
690 | * @bpf_sock_ops: pointer to bpf_sock_ops_kern struct | |
691 | * @flags: flags value | |
692 | * Return: 0 for no error | |
693 | * -EINVAL if there is no full tcp socket | |
694 | * bits in flags that are not supported by current kernel | |
695 | * | |
2be7e212 DB |
696 | * int bpf_skb_adjust_room(skb, len_diff, mode, flags) |
697 | * Grow or shrink room in sk_buff. | |
698 | * @skb: pointer to skb | |
699 | * @len_diff: (signed) amount of room to grow/shrink | |
700 | * @mode: operation mode (enum bpf_adj_room_mode) | |
701 | * @flags: reserved for future use | |
702 | * Return: 0 on success or negative error code | |
174a79ff JF |
703 | * |
704 | * int bpf_sk_redirect_map(map, key, flags) | |
705 | * Redirect skb to a sock in map using key as a lookup key for the | |
706 | * sock in map. | |
707 | * @map: pointer to sockmap | |
708 | * @key: key to lookup sock in map | |
709 | * @flags: reserved for future use | |
bfa64075 | 710 | * Return: SK_PASS |
174a79ff | 711 | * |
464bc0fd | 712 | * int bpf_sock_map_update(skops, map, key, flags) |
174a79ff JF |
713 | * @skops: pointer to bpf_sock_ops |
714 | * @map: pointer to sockmap to update | |
715 | * @key: key to insert/update sock in map | |
716 | * @flags: same flags as map update elem | |
de8f3a83 DB |
717 | * |
718 | * int bpf_xdp_adjust_meta(xdp_md, delta) | |
719 | * Adjust the xdp_md.data_meta by delta | |
720 | * @xdp_md: pointer to xdp_md | |
721 | * @delta: An positive/negative integer to be added to xdp_md.data_meta | |
722 | * Return: 0 on success or negative on error | |
908432ca YS |
723 | * |
724 | * int bpf_perf_event_read_value(map, flags, buf, buf_size) | |
725 | * read perf event counter value and perf event enabled/running time | |
726 | * @map: pointer to perf_event_array map | |
727 | * @flags: index of event in the map or bitmask flags | |
728 | * @buf: buf to fill | |
729 | * @buf_size: size of the buf | |
730 | * Return: 0 on success or negative error code | |
4bebdc7a YS |
731 | * |
732 | * int bpf_perf_prog_read_value(ctx, buf, buf_size) | |
733 | * read perf prog attached perf event counter and enabled/running time | |
734 | * @ctx: pointer to ctx | |
735 | * @buf: buf to fill | |
736 | * @buf_size: size of the buf | |
737 | * Return : 0 on success or negative error code | |
9802d865 JB |
738 | * |
739 | * int bpf_override_return(pt_regs, rc) | |
740 | * @pt_regs: pointer to struct pt_regs | |
741 | * @rc: the return value to set | |
4f738adb JF |
742 | * |
743 | * int bpf_msg_redirect_map(map, key, flags) | |
744 | * Redirect msg to a sock in map using key as a lookup key for the | |
745 | * sock in map. | |
746 | * @map: pointer to sockmap | |
747 | * @key: key to lookup sock in map | |
748 | * @flags: reserved for future use | |
749 | * Return: SK_PASS | |
750 | * | |
d74bad4e AI |
751 | * int bpf_bind(ctx, addr, addr_len) |
752 | * Bind socket to address. Only binding to IP is supported, no port can be | |
753 | * set in addr. | |
754 | * @ctx: pointer to context of type bpf_sock_addr | |
755 | * @addr: pointer to struct sockaddr to bind socket to | |
756 | * @addr_len: length of sockaddr structure | |
757 | * Return: 0 on success or negative error code | |
ebb676da TG |
758 | */ |
759 | #define __BPF_FUNC_MAPPER(FN) \ | |
760 | FN(unspec), \ | |
761 | FN(map_lookup_elem), \ | |
762 | FN(map_update_elem), \ | |
763 | FN(map_delete_elem), \ | |
764 | FN(probe_read), \ | |
765 | FN(ktime_get_ns), \ | |
766 | FN(trace_printk), \ | |
767 | FN(get_prandom_u32), \ | |
768 | FN(get_smp_processor_id), \ | |
769 | FN(skb_store_bytes), \ | |
770 | FN(l3_csum_replace), \ | |
771 | FN(l4_csum_replace), \ | |
772 | FN(tail_call), \ | |
773 | FN(clone_redirect), \ | |
774 | FN(get_current_pid_tgid), \ | |
775 | FN(get_current_uid_gid), \ | |
776 | FN(get_current_comm), \ | |
777 | FN(get_cgroup_classid), \ | |
778 | FN(skb_vlan_push), \ | |
779 | FN(skb_vlan_pop), \ | |
780 | FN(skb_get_tunnel_key), \ | |
781 | FN(skb_set_tunnel_key), \ | |
782 | FN(perf_event_read), \ | |
783 | FN(redirect), \ | |
784 | FN(get_route_realm), \ | |
785 | FN(perf_event_output), \ | |
786 | FN(skb_load_bytes), \ | |
787 | FN(get_stackid), \ | |
788 | FN(csum_diff), \ | |
789 | FN(skb_get_tunnel_opt), \ | |
790 | FN(skb_set_tunnel_opt), \ | |
791 | FN(skb_change_proto), \ | |
792 | FN(skb_change_type), \ | |
793 | FN(skb_under_cgroup), \ | |
794 | FN(get_hash_recalc), \ | |
795 | FN(get_current_task), \ | |
796 | FN(probe_write_user), \ | |
797 | FN(current_task_under_cgroup), \ | |
798 | FN(skb_change_tail), \ | |
799 | FN(skb_pull_data), \ | |
800 | FN(csum_update), \ | |
801 | FN(set_hash_invalid), \ | |
3a0af8fd | 802 | FN(get_numa_node_id), \ |
17bedab2 | 803 | FN(skb_change_head), \ |
a5e8c070 | 804 | FN(xdp_adjust_head), \ |
91b8270f | 805 | FN(probe_read_str), \ |
6acc5c29 | 806 | FN(get_socket_cookie), \ |
ded092cd | 807 | FN(get_socket_uid), \ |
8c4b4c7e | 808 | FN(set_hash), \ |
2be7e212 | 809 | FN(setsockopt), \ |
97f91a7c | 810 | FN(skb_adjust_room), \ |
174a79ff JF |
811 | FN(redirect_map), \ |
812 | FN(sk_redirect_map), \ | |
813 | FN(sock_map_update), \ | |
908432ca | 814 | FN(xdp_adjust_meta), \ |
4bebdc7a | 815 | FN(perf_event_read_value), \ |
cd86d1fd | 816 | FN(perf_prog_read_value), \ |
9802d865 | 817 | FN(getsockopt), \ |
b13d8807 | 818 | FN(override_return), \ |
4f738adb | 819 | FN(sock_ops_cb_flags_set), \ |
2a100317 | 820 | FN(msg_redirect_map), \ |
91843d54 | 821 | FN(msg_apply_bytes), \ |
015632bb | 822 | FN(msg_cork_bytes), \ |
d74bad4e AI |
823 | FN(msg_pull_data), \ |
824 | FN(bind), | |
ebb676da | 825 | |
09756af4 AS |
826 | /* integer value in 'imm' field of BPF_CALL instruction selects which helper |
827 | * function eBPF program intends to call | |
828 | */ | |
ebb676da | 829 | #define __BPF_ENUM_FN(x) BPF_FUNC_ ## x |
09756af4 | 830 | enum bpf_func_id { |
ebb676da | 831 | __BPF_FUNC_MAPPER(__BPF_ENUM_FN) |
09756af4 AS |
832 | __BPF_FUNC_MAX_ID, |
833 | }; | |
ebb676da | 834 | #undef __BPF_ENUM_FN |
09756af4 | 835 | |
781c53bc DB |
836 | /* All flags used by eBPF helper functions, placed here. */ |
837 | ||
838 | /* BPF_FUNC_skb_store_bytes flags. */ | |
839 | #define BPF_F_RECOMPUTE_CSUM (1ULL << 0) | |
8afd54c8 | 840 | #define BPF_F_INVALIDATE_HASH (1ULL << 1) |
781c53bc DB |
841 | |
842 | /* BPF_FUNC_l3_csum_replace and BPF_FUNC_l4_csum_replace flags. | |
843 | * First 4 bits are for passing the header field size. | |
844 | */ | |
845 | #define BPF_F_HDR_FIELD_MASK 0xfULL | |
846 | ||
847 | /* BPF_FUNC_l4_csum_replace flags. */ | |
848 | #define BPF_F_PSEUDO_HDR (1ULL << 4) | |
2f72959a | 849 | #define BPF_F_MARK_MANGLED_0 (1ULL << 5) |
d1b662ad | 850 | #define BPF_F_MARK_ENFORCE (1ULL << 6) |
781c53bc DB |
851 | |
852 | /* BPF_FUNC_clone_redirect and BPF_FUNC_redirect flags. */ | |
853 | #define BPF_F_INGRESS (1ULL << 0) | |
854 | ||
c6c33454 DB |
855 | /* BPF_FUNC_skb_set_tunnel_key and BPF_FUNC_skb_get_tunnel_key flags. */ |
856 | #define BPF_F_TUNINFO_IPV6 (1ULL << 0) | |
857 | ||
d5a3b1f6 AS |
858 | /* BPF_FUNC_get_stackid flags. */ |
859 | #define BPF_F_SKIP_FIELD_MASK 0xffULL | |
860 | #define BPF_F_USER_STACK (1ULL << 8) | |
861 | #define BPF_F_FAST_STACK_CMP (1ULL << 9) | |
862 | #define BPF_F_REUSE_STACKID (1ULL << 10) | |
863 | ||
2da897e5 DB |
864 | /* BPF_FUNC_skb_set_tunnel_key flags. */ |
865 | #define BPF_F_ZERO_CSUM_TX (1ULL << 1) | |
22080870 | 866 | #define BPF_F_DONT_FRAGMENT (1ULL << 2) |
77a5196a | 867 | #define BPF_F_SEQ_NUMBER (1ULL << 3) |
2da897e5 | 868 | |
908432ca YS |
869 | /* BPF_FUNC_perf_event_output, BPF_FUNC_perf_event_read and |
870 | * BPF_FUNC_perf_event_read_value flags. | |
871 | */ | |
1e33759c DB |
872 | #define BPF_F_INDEX_MASK 0xffffffffULL |
873 | #define BPF_F_CURRENT_CPU BPF_F_INDEX_MASK | |
555c8a86 DB |
874 | /* BPF_FUNC_perf_event_output for sk_buff input context. */ |
875 | #define BPF_F_CTXLEN_MASK (0xfffffULL << 32) | |
1e33759c | 876 | |
2be7e212 DB |
877 | /* Mode for BPF_FUNC_skb_adjust_room helper. */ |
878 | enum bpf_adj_room_mode { | |
879 | BPF_ADJ_ROOM_NET, | |
880 | }; | |
881 | ||
9bac3d6d AS |
882 | /* user accessible mirror of in-kernel sk_buff. |
883 | * new fields can only be added to the end of this structure | |
884 | */ | |
885 | struct __sk_buff { | |
886 | __u32 len; | |
887 | __u32 pkt_type; | |
888 | __u32 mark; | |
889 | __u32 queue_mapping; | |
c2497395 AS |
890 | __u32 protocol; |
891 | __u32 vlan_present; | |
892 | __u32 vlan_tci; | |
27cd5452 | 893 | __u32 vlan_proto; |
bcad5718 | 894 | __u32 priority; |
37e82c2f AS |
895 | __u32 ingress_ifindex; |
896 | __u32 ifindex; | |
d691f9e8 AS |
897 | __u32 tc_index; |
898 | __u32 cb[5]; | |
ba7591d8 | 899 | __u32 hash; |
045efa82 | 900 | __u32 tc_classid; |
969bf05e AS |
901 | __u32 data; |
902 | __u32 data_end; | |
b1d9fc41 | 903 | __u32 napi_id; |
8a31db56 | 904 | |
de8f3a83 | 905 | /* Accessed by BPF_PROG_TYPE_sk_skb types from here to ... */ |
8a31db56 JF |
906 | __u32 family; |
907 | __u32 remote_ip4; /* Stored in network byte order */ | |
908 | __u32 local_ip4; /* Stored in network byte order */ | |
909 | __u32 remote_ip6[4]; /* Stored in network byte order */ | |
910 | __u32 local_ip6[4]; /* Stored in network byte order */ | |
911 | __u32 remote_port; /* Stored in network byte order */ | |
912 | __u32 local_port; /* stored in host byte order */ | |
de8f3a83 DB |
913 | /* ... here. */ |
914 | ||
915 | __u32 data_meta; | |
9bac3d6d AS |
916 | }; |
917 | ||
d3aa45ce AS |
918 | struct bpf_tunnel_key { |
919 | __u32 tunnel_id; | |
c6c33454 DB |
920 | union { |
921 | __u32 remote_ipv4; | |
922 | __u32 remote_ipv6[4]; | |
923 | }; | |
924 | __u8 tunnel_tos; | |
925 | __u8 tunnel_ttl; | |
c0e760c9 | 926 | __u16 tunnel_ext; |
4018ab18 | 927 | __u32 tunnel_label; |
d3aa45ce AS |
928 | }; |
929 | ||
3a0af8fd TG |
930 | /* Generic BPF return codes which all BPF program types may support. |
931 | * The values are binary compatible with their TC_ACT_* counter-part to | |
932 | * provide backwards compatibility with existing SCHED_CLS and SCHED_ACT | |
933 | * programs. | |
934 | * | |
935 | * XDP is handled seprately, see XDP_*. | |
936 | */ | |
937 | enum bpf_ret_code { | |
938 | BPF_OK = 0, | |
939 | /* 1 reserved */ | |
940 | BPF_DROP = 2, | |
941 | /* 3-6 reserved */ | |
942 | BPF_REDIRECT = 7, | |
943 | /* >127 are reserved for prog type specific return codes */ | |
944 | }; | |
945 | ||
61023658 DA |
946 | struct bpf_sock { |
947 | __u32 bound_dev_if; | |
aa4c1037 DA |
948 | __u32 family; |
949 | __u32 type; | |
950 | __u32 protocol; | |
482dca93 DA |
951 | __u32 mark; |
952 | __u32 priority; | |
aac3fc32 AI |
953 | __u32 src_ip4; /* Allows 1,2,4-byte read. |
954 | * Stored in network byte order. | |
955 | */ | |
956 | __u32 src_ip6[4]; /* Allows 1,2,4-byte read. | |
957 | * Stored in network byte order. | |
958 | */ | |
959 | __u32 src_port; /* Allows 4-byte read. | |
960 | * Stored in host byte order | |
961 | */ | |
61023658 DA |
962 | }; |
963 | ||
17bedab2 MKL |
964 | #define XDP_PACKET_HEADROOM 256 |
965 | ||
6a773a15 BB |
966 | /* User return codes for XDP prog type. |
967 | * A valid XDP program must return one of these defined values. All other | |
9beb8bed DB |
968 | * return codes are reserved for future use. Unknown return codes will |
969 | * result in packet drops and a warning via bpf_warn_invalid_xdp_action(). | |
6a773a15 BB |
970 | */ |
971 | enum xdp_action { | |
972 | XDP_ABORTED = 0, | |
973 | XDP_DROP, | |
974 | XDP_PASS, | |
6ce96ca3 | 975 | XDP_TX, |
814abfab | 976 | XDP_REDIRECT, |
6a773a15 BB |
977 | }; |
978 | ||
979 | /* user accessible metadata for XDP packet hook | |
980 | * new fields must be added to the end of this structure | |
981 | */ | |
982 | struct xdp_md { | |
983 | __u32 data; | |
984 | __u32 data_end; | |
de8f3a83 | 985 | __u32 data_meta; |
daaf24c6 | 986 | /* Below access go through struct xdp_rxq_info */ |
02dd3291 JDB |
987 | __u32 ingress_ifindex; /* rxq->dev->ifindex */ |
988 | __u32 rx_queue_index; /* rxq->queue_index */ | |
6a773a15 BB |
989 | }; |
990 | ||
174a79ff | 991 | enum sk_action { |
bfa64075 JF |
992 | SK_DROP = 0, |
993 | SK_PASS, | |
174a79ff JF |
994 | }; |
995 | ||
4f738adb JF |
996 | /* user accessible metadata for SK_MSG packet hook, new fields must |
997 | * be added to the end of this structure | |
998 | */ | |
999 | struct sk_msg_md { | |
1000 | void *data; | |
1001 | void *data_end; | |
1002 | }; | |
1003 | ||
1e270976 MKL |
1004 | #define BPF_TAG_SIZE 8 |
1005 | ||
1006 | struct bpf_prog_info { | |
1007 | __u32 type; | |
1008 | __u32 id; | |
1009 | __u8 tag[BPF_TAG_SIZE]; | |
1010 | __u32 jited_prog_len; | |
1011 | __u32 xlated_prog_len; | |
1012 | __aligned_u64 jited_prog_insns; | |
1013 | __aligned_u64 xlated_prog_insns; | |
cb4d2b3f MKL |
1014 | __u64 load_time; /* ns since boottime */ |
1015 | __u32 created_by_uid; | |
1016 | __u32 nr_map_ids; | |
1017 | __aligned_u64 map_ids; | |
067cae47 | 1018 | char name[BPF_OBJ_NAME_LEN]; |
675fc275 JK |
1019 | __u32 ifindex; |
1020 | __u64 netns_dev; | |
1021 | __u64 netns_ino; | |
1e270976 MKL |
1022 | } __attribute__((aligned(8))); |
1023 | ||
1024 | struct bpf_map_info { | |
1025 | __u32 type; | |
1026 | __u32 id; | |
1027 | __u32 key_size; | |
1028 | __u32 value_size; | |
1029 | __u32 max_entries; | |
1030 | __u32 map_flags; | |
067cae47 | 1031 | char name[BPF_OBJ_NAME_LEN]; |
52775b33 JK |
1032 | __u32 ifindex; |
1033 | __u64 netns_dev; | |
1034 | __u64 netns_ino; | |
1e270976 MKL |
1035 | } __attribute__((aligned(8))); |
1036 | ||
4fbac77d AI |
1037 | /* User bpf_sock_addr struct to access socket fields and sockaddr struct passed |
1038 | * by user and intended to be used by socket (e.g. to bind to, depends on | |
1039 | * attach attach type). | |
1040 | */ | |
1041 | struct bpf_sock_addr { | |
1042 | __u32 user_family; /* Allows 4-byte read, but no write. */ | |
1043 | __u32 user_ip4; /* Allows 1,2,4-byte read and 4-byte write. | |
1044 | * Stored in network byte order. | |
1045 | */ | |
1046 | __u32 user_ip6[4]; /* Allows 1,2,4-byte read an 4-byte write. | |
1047 | * Stored in network byte order. | |
1048 | */ | |
1049 | __u32 user_port; /* Allows 4-byte read and write. | |
1050 | * Stored in network byte order | |
1051 | */ | |
1052 | __u32 family; /* Allows 4-byte read, but no write */ | |
1053 | __u32 type; /* Allows 4-byte read, but no write */ | |
1054 | __u32 protocol; /* Allows 4-byte read, but no write */ | |
1055 | }; | |
1056 | ||
40304b2a LB |
1057 | /* User bpf_sock_ops struct to access socket values and specify request ops |
1058 | * and their replies. | |
1059 | * Some of this fields are in network (bigendian) byte order and may need | |
1060 | * to be converted before use (bpf_ntohl() defined in samples/bpf/bpf_endian.h). | |
1061 | * New fields can only be added at the end of this structure | |
1062 | */ | |
1063 | struct bpf_sock_ops { | |
1064 | __u32 op; | |
1065 | union { | |
de525be2 LB |
1066 | __u32 args[4]; /* Optionally passed to bpf program */ |
1067 | __u32 reply; /* Returned by bpf program */ | |
1068 | __u32 replylong[4]; /* Optionally returned by bpf prog */ | |
40304b2a LB |
1069 | }; |
1070 | __u32 family; | |
1071 | __u32 remote_ip4; /* Stored in network byte order */ | |
1072 | __u32 local_ip4; /* Stored in network byte order */ | |
1073 | __u32 remote_ip6[4]; /* Stored in network byte order */ | |
1074 | __u32 local_ip6[4]; /* Stored in network byte order */ | |
1075 | __u32 remote_port; /* Stored in network byte order */ | |
1076 | __u32 local_port; /* stored in host byte order */ | |
f19397a5 LB |
1077 | __u32 is_fullsock; /* Some TCP fields are only valid if |
1078 | * there is a full socket. If not, the | |
1079 | * fields read as zero. | |
1080 | */ | |
1081 | __u32 snd_cwnd; | |
1082 | __u32 srtt_us; /* Averaged RTT << 3 in usecs */ | |
b13d8807 | 1083 | __u32 bpf_sock_ops_cb_flags; /* flags defined in uapi/linux/tcp.h */ |
44f0e430 LB |
1084 | __u32 state; |
1085 | __u32 rtt_min; | |
1086 | __u32 snd_ssthresh; | |
1087 | __u32 rcv_nxt; | |
1088 | __u32 snd_nxt; | |
1089 | __u32 snd_una; | |
1090 | __u32 mss_cache; | |
1091 | __u32 ecn_flags; | |
1092 | __u32 rate_delivered; | |
1093 | __u32 rate_interval_us; | |
1094 | __u32 packets_out; | |
1095 | __u32 retrans_out; | |
1096 | __u32 total_retrans; | |
1097 | __u32 segs_in; | |
1098 | __u32 data_segs_in; | |
1099 | __u32 segs_out; | |
1100 | __u32 data_segs_out; | |
1101 | __u32 lost_out; | |
1102 | __u32 sacked_out; | |
1103 | __u32 sk_txhash; | |
1104 | __u64 bytes_received; | |
1105 | __u64 bytes_acked; | |
40304b2a LB |
1106 | }; |
1107 | ||
b13d8807 | 1108 | /* Definitions for bpf_sock_ops_cb_flags */ |
f89013f6 | 1109 | #define BPF_SOCK_OPS_RTO_CB_FLAG (1<<0) |
a31ad29e | 1110 | #define BPF_SOCK_OPS_RETRANS_CB_FLAG (1<<1) |
d4487491 LB |
1111 | #define BPF_SOCK_OPS_STATE_CB_FLAG (1<<2) |
1112 | #define BPF_SOCK_OPS_ALL_CB_FLAGS 0x7 /* Mask of all currently | |
b13d8807 LB |
1113 | * supported cb flags |
1114 | */ | |
1115 | ||
40304b2a LB |
1116 | /* List of known BPF sock_ops operators. |
1117 | * New entries can only be added at the end | |
1118 | */ | |
1119 | enum { | |
1120 | BPF_SOCK_OPS_VOID, | |
8550f328 LB |
1121 | BPF_SOCK_OPS_TIMEOUT_INIT, /* Should return SYN-RTO value to use or |
1122 | * -1 if default value should be used | |
1123 | */ | |
13d3b1eb LB |
1124 | BPF_SOCK_OPS_RWND_INIT, /* Should return initial advertized |
1125 | * window (in packets) or -1 if default | |
1126 | * value should be used | |
1127 | */ | |
9872a4bd LB |
1128 | BPF_SOCK_OPS_TCP_CONNECT_CB, /* Calls BPF program right before an |
1129 | * active connection is initialized | |
1130 | */ | |
1131 | BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB, /* Calls BPF program when an | |
1132 | * active connection is | |
1133 | * established | |
1134 | */ | |
1135 | BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB, /* Calls BPF program when a | |
1136 | * passive connection is | |
1137 | * established | |
1138 | */ | |
91b5b21c LB |
1139 | BPF_SOCK_OPS_NEEDS_ECN, /* If connection's congestion control |
1140 | * needs ECN | |
1141 | */ | |
e6546ef6 LB |
1142 | BPF_SOCK_OPS_BASE_RTT, /* Get base RTT. The correct value is |
1143 | * based on the path and may be | |
1144 | * dependent on the congestion control | |
1145 | * algorithm. In general it indicates | |
1146 | * a congestion threshold. RTTs above | |
1147 | * this indicate congestion | |
1148 | */ | |
f89013f6 LB |
1149 | BPF_SOCK_OPS_RTO_CB, /* Called when an RTO has triggered. |
1150 | * Arg1: value of icsk_retransmits | |
1151 | * Arg2: value of icsk_rto | |
1152 | * Arg3: whether RTO has expired | |
1153 | */ | |
a31ad29e LB |
1154 | BPF_SOCK_OPS_RETRANS_CB, /* Called when skb is retransmitted. |
1155 | * Arg1: sequence number of 1st byte | |
1156 | * Arg2: # segments | |
1157 | * Arg3: return value of | |
1158 | * tcp_transmit_skb (0 => success) | |
1159 | */ | |
d4487491 LB |
1160 | BPF_SOCK_OPS_STATE_CB, /* Called when TCP changes state. |
1161 | * Arg1: old_state | |
1162 | * Arg2: new_state | |
1163 | */ | |
1164 | }; | |
1165 | ||
1166 | /* List of TCP states. There is a build check in net/ipv4/tcp.c to detect | |
1167 | * changes between the TCP and BPF versions. Ideally this should never happen. | |
1168 | * If it does, we need to add code to convert them before calling | |
1169 | * the BPF sock_ops function. | |
1170 | */ | |
1171 | enum { | |
1172 | BPF_TCP_ESTABLISHED = 1, | |
1173 | BPF_TCP_SYN_SENT, | |
1174 | BPF_TCP_SYN_RECV, | |
1175 | BPF_TCP_FIN_WAIT1, | |
1176 | BPF_TCP_FIN_WAIT2, | |
1177 | BPF_TCP_TIME_WAIT, | |
1178 | BPF_TCP_CLOSE, | |
1179 | BPF_TCP_CLOSE_WAIT, | |
1180 | BPF_TCP_LAST_ACK, | |
1181 | BPF_TCP_LISTEN, | |
1182 | BPF_TCP_CLOSING, /* Now a valid state */ | |
1183 | BPF_TCP_NEW_SYN_RECV, | |
1184 | ||
1185 | BPF_TCP_MAX_STATES /* Leave at the end! */ | |
40304b2a LB |
1186 | }; |
1187 | ||
fc747810 | 1188 | #define TCP_BPF_IW 1001 /* Set TCP initial congestion window */ |
13bf9641 | 1189 | #define TCP_BPF_SNDCWND_CLAMP 1002 /* Set sndcwnd_clamp */ |
fc747810 | 1190 | |
908432ca YS |
1191 | struct bpf_perf_event_value { |
1192 | __u64 counter; | |
1193 | __u64 enabled; | |
1194 | __u64 running; | |
1195 | }; | |
1196 | ||
ebc614f6 RG |
1197 | #define BPF_DEVCG_ACC_MKNOD (1ULL << 0) |
1198 | #define BPF_DEVCG_ACC_READ (1ULL << 1) | |
1199 | #define BPF_DEVCG_ACC_WRITE (1ULL << 2) | |
1200 | ||
1201 | #define BPF_DEVCG_DEV_BLOCK (1ULL << 0) | |
1202 | #define BPF_DEVCG_DEV_CHAR (1ULL << 1) | |
1203 | ||
1204 | struct bpf_cgroup_dev_ctx { | |
06ef0ccb YS |
1205 | /* access_type encoded as (BPF_DEVCG_ACC_* << 16) | BPF_DEVCG_DEV_* */ |
1206 | __u32 access_type; | |
ebc614f6 RG |
1207 | __u32 major; |
1208 | __u32 minor; | |
1209 | }; | |
1210 | ||
c4f6699d AS |
1211 | struct bpf_raw_tracepoint_args { |
1212 | __u64 args[0]; | |
1213 | }; | |
1214 | ||
daedfb22 | 1215 | #endif /* _UAPI__LINUX_BPF_H__ */ |