Commit | Line | Data |
---|---|---|
971e827b ACM |
1 | /* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com |
2 | * | |
3 | * This program is free software; you can redistribute it and/or | |
4 | * modify it under the terms of version 2 of the GNU General Public | |
5 | * License as published by the Free Software Foundation. | |
6 | */ | |
7 | #ifndef _UAPI__LINUX_BPF_H__ | |
8 | #define _UAPI__LINUX_BPF_H__ | |
9 | ||
10 | #include <linux/types.h> | |
11 | #include <linux/bpf_common.h> | |
12 | ||
13 | /* Extended instruction set based on top of classic BPF */ | |
14 | ||
15 | /* instruction classes */ | |
16 | #define BPF_ALU64 0x07 /* alu mode in double word width */ | |
17 | ||
18 | /* ld/ldx fields */ | |
19 | #define BPF_DW 0x18 /* double word */ | |
20 | #define BPF_XADD 0xc0 /* exclusive add */ | |
21 | ||
22 | /* alu/jmp fields */ | |
23 | #define BPF_MOV 0xb0 /* mov reg to reg */ | |
24 | #define BPF_ARSH 0xc0 /* sign extending arithmetic shift right */ | |
25 | ||
26 | /* change endianness of a register */ | |
27 | #define BPF_END 0xd0 /* flags for endianness conversion: */ | |
28 | #define BPF_TO_LE 0x00 /* convert to little-endian */ | |
29 | #define BPF_TO_BE 0x08 /* convert to big-endian */ | |
30 | #define BPF_FROM_LE BPF_TO_LE | |
31 | #define BPF_FROM_BE BPF_TO_BE | |
32 | ||
92b31a9a | 33 | /* jmp encodings */ |
971e827b | 34 | #define BPF_JNE 0x50 /* jump != */ |
92b31a9a DB |
35 | #define BPF_JLT 0xa0 /* LT is unsigned, '<' */ |
36 | #define BPF_JLE 0xb0 /* LE is unsigned, '<=' */ | |
971e827b ACM |
37 | #define BPF_JSGT 0x60 /* SGT is signed '>', GT in x86 */ |
38 | #define BPF_JSGE 0x70 /* SGE is signed '>=', GE in x86 */ | |
92b31a9a DB |
39 | #define BPF_JSLT 0xc0 /* SLT is signed, '<' */ |
40 | #define BPF_JSLE 0xd0 /* SLE is signed, '<=' */ | |
971e827b ACM |
41 | #define BPF_CALL 0x80 /* function call */ |
42 | #define BPF_EXIT 0x90 /* function return */ | |
43 | ||
44 | /* Register numbers */ | |
45 | enum { | |
46 | BPF_REG_0 = 0, | |
47 | BPF_REG_1, | |
48 | BPF_REG_2, | |
49 | BPF_REG_3, | |
50 | BPF_REG_4, | |
51 | BPF_REG_5, | |
52 | BPF_REG_6, | |
53 | BPF_REG_7, | |
54 | BPF_REG_8, | |
55 | BPF_REG_9, | |
56 | BPF_REG_10, | |
57 | __MAX_BPF_REG, | |
58 | }; | |
59 | ||
60 | /* BPF has 10 general purpose 64-bit registers and stack frame. */ | |
61 | #define MAX_BPF_REG __MAX_BPF_REG | |
62 | ||
63 | struct bpf_insn { | |
64 | __u8 code; /* opcode */ | |
65 | __u8 dst_reg:4; /* dest register */ | |
66 | __u8 src_reg:4; /* source register */ | |
67 | __s16 off; /* signed offset */ | |
68 | __s32 imm; /* signed immediate constant */ | |
69 | }; | |
70 | ||
9a738266 MS |
71 | /* Key of an a BPF_MAP_TYPE_LPM_TRIE entry */ |
72 | struct bpf_lpm_trie_key { | |
73 | __u32 prefixlen; /* up to 32 for AF_INET, 128 for AF_INET6 */ | |
74 | __u8 data[0]; /* Arbitrary size */ | |
75 | }; | |
76 | ||
971e827b ACM |
77 | /* BPF syscall commands, see bpf(2) man-page for details. */ |
78 | enum bpf_cmd { | |
79 | BPF_MAP_CREATE, | |
80 | BPF_MAP_LOOKUP_ELEM, | |
81 | BPF_MAP_UPDATE_ELEM, | |
82 | BPF_MAP_DELETE_ELEM, | |
83 | BPF_MAP_GET_NEXT_KEY, | |
84 | BPF_PROG_LOAD, | |
85 | BPF_OBJ_PIN, | |
86 | BPF_OBJ_GET, | |
0cb34dc2 JS |
87 | BPF_PROG_ATTACH, |
88 | BPF_PROG_DETACH, | |
30848873 | 89 | BPF_PROG_TEST_RUN, |
95b9afd3 MKL |
90 | BPF_PROG_GET_NEXT_ID, |
91 | BPF_MAP_GET_NEXT_ID, | |
92 | BPF_PROG_GET_FD_BY_ID, | |
93 | BPF_MAP_GET_FD_BY_ID, | |
94 | BPF_OBJ_GET_INFO_BY_FD, | |
defd9c47 | 95 | BPF_PROG_QUERY, |
971e827b ACM |
96 | }; |
97 | ||
98 | enum bpf_map_type { | |
99 | BPF_MAP_TYPE_UNSPEC, | |
100 | BPF_MAP_TYPE_HASH, | |
101 | BPF_MAP_TYPE_ARRAY, | |
102 | BPF_MAP_TYPE_PROG_ARRAY, | |
103 | BPF_MAP_TYPE_PERF_EVENT_ARRAY, | |
104 | BPF_MAP_TYPE_PERCPU_HASH, | |
105 | BPF_MAP_TYPE_PERCPU_ARRAY, | |
106 | BPF_MAP_TYPE_STACK_TRACE, | |
791cceb8 | 107 | BPF_MAP_TYPE_CGROUP_ARRAY, |
0cb34dc2 JS |
108 | BPF_MAP_TYPE_LRU_HASH, |
109 | BPF_MAP_TYPE_LRU_PERCPU_HASH, | |
9a738266 | 110 | BPF_MAP_TYPE_LPM_TRIE, |
fb30d4b7 MKL |
111 | BPF_MAP_TYPE_ARRAY_OF_MAPS, |
112 | BPF_MAP_TYPE_HASH_OF_MAPS, | |
81f6bf81 | 113 | BPF_MAP_TYPE_DEVMAP, |
69e8cc13 | 114 | BPF_MAP_TYPE_SOCKMAP, |
6710e112 | 115 | BPF_MAP_TYPE_CPUMAP, |
971e827b ACM |
116 | }; |
117 | ||
118 | enum bpf_prog_type { | |
119 | BPF_PROG_TYPE_UNSPEC, | |
120 | BPF_PROG_TYPE_SOCKET_FILTER, | |
121 | BPF_PROG_TYPE_KPROBE, | |
122 | BPF_PROG_TYPE_SCHED_CLS, | |
123 | BPF_PROG_TYPE_SCHED_ACT, | |
124 | BPF_PROG_TYPE_TRACEPOINT, | |
791cceb8 | 125 | BPF_PROG_TYPE_XDP, |
0cb34dc2 JS |
126 | BPF_PROG_TYPE_PERF_EVENT, |
127 | BPF_PROG_TYPE_CGROUP_SKB, | |
128 | BPF_PROG_TYPE_CGROUP_SOCK, | |
129 | BPF_PROG_TYPE_LWT_IN, | |
130 | BPF_PROG_TYPE_LWT_OUT, | |
131 | BPF_PROG_TYPE_LWT_XMIT, | |
04df41e3 | 132 | BPF_PROG_TYPE_SOCK_OPS, |
69e8cc13 | 133 | BPF_PROG_TYPE_SK_SKB, |
971e827b ACM |
134 | }; |
135 | ||
0cb34dc2 JS |
136 | enum bpf_attach_type { |
137 | BPF_CGROUP_INET_INGRESS, | |
138 | BPF_CGROUP_INET_EGRESS, | |
139 | BPF_CGROUP_INET_SOCK_CREATE, | |
04df41e3 | 140 | BPF_CGROUP_SOCK_OPS, |
464bc0fd JF |
141 | BPF_SK_SKB_STREAM_PARSER, |
142 | BPF_SK_SKB_STREAM_VERDICT, | |
0cb34dc2 JS |
143 | __MAX_BPF_ATTACH_TYPE |
144 | }; | |
145 | ||
146 | #define MAX_BPF_ATTACH_TYPE __MAX_BPF_ATTACH_TYPE | |
147 | ||
defd9c47 AS |
148 | /* cgroup-bpf attach flags used in BPF_PROG_ATTACH command |
149 | * | |
150 | * NONE(default): No further bpf programs allowed in the subtree. | |
151 | * | |
152 | * BPF_F_ALLOW_OVERRIDE: If a sub-cgroup installs some bpf program, | |
153 | * the program in this cgroup yields to sub-cgroup program. | |
154 | * | |
155 | * BPF_F_ALLOW_MULTI: If a sub-cgroup installs some bpf program, | |
156 | * that cgroup program gets run in addition to the program in this cgroup. | |
157 | * | |
158 | * Only one program is allowed to be attached to a cgroup with | |
159 | * NONE or BPF_F_ALLOW_OVERRIDE flag. | |
160 | * Attaching another program on top of NONE or BPF_F_ALLOW_OVERRIDE will | |
161 | * release old program and attach the new one. Attach flags has to match. | |
162 | * | |
163 | * Multiple programs are allowed to be attached to a cgroup with | |
164 | * BPF_F_ALLOW_MULTI flag. They are executed in FIFO order | |
165 | * (those that were attached first, run first) | |
166 | * The programs of sub-cgroup are executed first, then programs of | |
167 | * this cgroup and then programs of parent cgroup. | |
168 | * When children program makes decision (like picking TCP CA or sock bind) | |
169 | * parent program has a chance to override it. | |
170 | * | |
171 | * A cgroup with MULTI or OVERRIDE flag allows any attach flags in sub-cgroups. | |
172 | * A cgroup with NONE doesn't allow any programs in sub-cgroups. | |
173 | * Ex1: | |
174 | * cgrp1 (MULTI progs A, B) -> | |
175 | * cgrp2 (OVERRIDE prog C) -> | |
176 | * cgrp3 (MULTI prog D) -> | |
177 | * cgrp4 (OVERRIDE prog E) -> | |
178 | * cgrp5 (NONE prog F) | |
179 | * the event in cgrp5 triggers execution of F,D,A,B in that order. | |
180 | * if prog F is detached, the execution is E,D,A,B | |
181 | * if prog F and D are detached, the execution is E,A,B | |
182 | * if prog F, E and D are detached, the execution is C,A,B | |
183 | * | |
184 | * All eligible programs are executed regardless of return code from | |
185 | * earlier programs. | |
5463b3d0 SR |
186 | */ |
187 | #define BPF_F_ALLOW_OVERRIDE (1U << 0) | |
defd9c47 | 188 | #define BPF_F_ALLOW_MULTI (1U << 1) |
5463b3d0 | 189 | |
e07b98d9 DM |
190 | /* If BPF_F_STRICT_ALIGNMENT is used in BPF_PROG_LOAD command, the |
191 | * verifier will perform strict alignment checking as if the kernel | |
192 | * has been built with CONFIG_EFFICIENT_UNALIGNED_ACCESS not set, | |
193 | * and NET_IP_ALIGN defined to 2. | |
194 | */ | |
195 | #define BPF_F_STRICT_ALIGNMENT (1U << 0) | |
196 | ||
971e827b ACM |
197 | #define BPF_PSEUDO_MAP_FD 1 |
198 | ||
199 | /* flags for BPF_MAP_UPDATE_ELEM command */ | |
200 | #define BPF_ANY 0 /* create new element or update existing */ | |
201 | #define BPF_NOEXIST 1 /* create new element if it didn't exist */ | |
202 | #define BPF_EXIST 2 /* update existing element */ | |
203 | ||
ad17d0e6 | 204 | /* flags for BPF_MAP_CREATE command */ |
971e827b | 205 | #define BPF_F_NO_PREALLOC (1U << 0) |
0cb34dc2 JS |
206 | /* Instead of having one common LRU list in the |
207 | * BPF_MAP_TYPE_LRU_[PERCPU_]HASH map, use a percpu LRU list | |
208 | * which can scale and perform better. | |
209 | * Note, the LRU nodes (including free nodes) cannot be moved | |
210 | * across different LRU lists. | |
211 | */ | |
212 | #define BPF_F_NO_COMMON_LRU (1U << 1) | |
ad17d0e6 MKL |
213 | /* Specify numa node during map creation */ |
214 | #define BPF_F_NUMA_NODE (1U << 2) | |
971e827b | 215 | |
defd9c47 AS |
216 | /* flags for BPF_PROG_QUERY */ |
217 | #define BPF_F_QUERY_EFFECTIVE (1U << 0) | |
218 | ||
88cda1c9 MKL |
219 | #define BPF_OBJ_NAME_LEN 16U |
220 | ||
971e827b ACM |
221 | union bpf_attr { |
222 | struct { /* anonymous struct used by BPF_MAP_CREATE command */ | |
223 | __u32 map_type; /* one of enum bpf_map_type */ | |
224 | __u32 key_size; /* size of key in bytes */ | |
225 | __u32 value_size; /* size of value in bytes */ | |
226 | __u32 max_entries; /* max number of entries in a map */ | |
ad17d0e6 MKL |
227 | __u32 map_flags; /* BPF_MAP_CREATE related |
228 | * flags defined above. | |
229 | */ | |
fb30d4b7 | 230 | __u32 inner_map_fd; /* fd pointing to the inner map */ |
ad17d0e6 MKL |
231 | __u32 numa_node; /* numa node (effective only if |
232 | * BPF_F_NUMA_NODE is set). | |
233 | */ | |
067cae47 | 234 | char map_name[BPF_OBJ_NAME_LEN]; |
971e827b ACM |
235 | }; |
236 | ||
237 | struct { /* anonymous struct used by BPF_MAP_*_ELEM commands */ | |
238 | __u32 map_fd; | |
239 | __aligned_u64 key; | |
240 | union { | |
241 | __aligned_u64 value; | |
242 | __aligned_u64 next_key; | |
243 | }; | |
244 | __u64 flags; | |
245 | }; | |
246 | ||
247 | struct { /* anonymous struct used by BPF_PROG_LOAD command */ | |
248 | __u32 prog_type; /* one of enum bpf_prog_type */ | |
249 | __u32 insn_cnt; | |
250 | __aligned_u64 insns; | |
251 | __aligned_u64 license; | |
252 | __u32 log_level; /* verbosity level of verifier */ | |
253 | __u32 log_size; /* size of user buffer */ | |
254 | __aligned_u64 log_buf; /* user supplied buffer */ | |
255 | __u32 kern_version; /* checked when prog_type=kprobe */ | |
e07b98d9 | 256 | __u32 prog_flags; |
067cae47 | 257 | char prog_name[BPF_OBJ_NAME_LEN]; |
971e827b ACM |
258 | }; |
259 | ||
260 | struct { /* anonymous struct used by BPF_OBJ_* commands */ | |
261 | __aligned_u64 pathname; | |
262 | __u32 bpf_fd; | |
263 | }; | |
0cb34dc2 JS |
264 | |
265 | struct { /* anonymous struct used by BPF_PROG_ATTACH/DETACH commands */ | |
266 | __u32 target_fd; /* container object to attach to */ | |
267 | __u32 attach_bpf_fd; /* eBPF program to attach */ | |
268 | __u32 attach_type; | |
5463b3d0 | 269 | __u32 attach_flags; |
0cb34dc2 | 270 | }; |
30848873 AS |
271 | |
272 | struct { /* anonymous struct used by BPF_PROG_TEST_RUN command */ | |
273 | __u32 prog_fd; | |
274 | __u32 retval; | |
275 | __u32 data_size_in; | |
276 | __u32 data_size_out; | |
277 | __aligned_u64 data_in; | |
278 | __aligned_u64 data_out; | |
279 | __u32 repeat; | |
280 | __u32 duration; | |
281 | } test; | |
95b9afd3 MKL |
282 | |
283 | struct { /* anonymous struct used by BPF_*_GET_*_ID */ | |
284 | union { | |
285 | __u32 start_id; | |
286 | __u32 prog_id; | |
287 | __u32 map_id; | |
288 | }; | |
289 | __u32 next_id; | |
290 | }; | |
291 | ||
292 | struct { /* anonymous struct used by BPF_OBJ_GET_INFO_BY_FD */ | |
293 | __u32 bpf_fd; | |
294 | __u32 info_len; | |
295 | __aligned_u64 info; | |
296 | } info; | |
defd9c47 AS |
297 | |
298 | struct { /* anonymous struct used by BPF_PROG_QUERY command */ | |
299 | __u32 target_fd; /* container object to query */ | |
300 | __u32 attach_type; | |
301 | __u32 query_flags; | |
302 | __u32 attach_flags; | |
303 | __aligned_u64 prog_ids; | |
304 | __u32 prog_cnt; | |
305 | } query; | |
971e827b ACM |
306 | } __attribute__((aligned(8))); |
307 | ||
0cb34dc2 JS |
308 | /* BPF helper function descriptions: |
309 | * | |
310 | * void *bpf_map_lookup_elem(&map, &key) | |
311 | * Return: Map value or NULL | |
312 | * | |
313 | * int bpf_map_update_elem(&map, &key, &value, flags) | |
314 | * Return: 0 on success or negative error | |
315 | * | |
316 | * int bpf_map_delete_elem(&map, &key) | |
317 | * Return: 0 on success or negative error | |
318 | * | |
319 | * int bpf_probe_read(void *dst, int size, void *src) | |
320 | * Return: 0 on success or negative error | |
321 | * | |
322 | * u64 bpf_ktime_get_ns(void) | |
323 | * Return: current ktime | |
324 | * | |
325 | * int bpf_trace_printk(const char *fmt, int fmt_size, ...) | |
326 | * Return: length of buffer written or negative error | |
327 | * | |
328 | * u32 bpf_prandom_u32(void) | |
329 | * Return: random value | |
330 | * | |
331 | * u32 bpf_raw_smp_processor_id(void) | |
332 | * Return: SMP processor ID | |
333 | * | |
334 | * int bpf_skb_store_bytes(skb, offset, from, len, flags) | |
335 | * store bytes into packet | |
336 | * @skb: pointer to skb | |
337 | * @offset: offset within packet from skb->mac_header | |
338 | * @from: pointer where to copy bytes from | |
339 | * @len: number of bytes to store into packet | |
340 | * @flags: bit 0 - if true, recompute skb->csum | |
341 | * other bits - reserved | |
342 | * Return: 0 on success or negative error | |
343 | * | |
344 | * int bpf_l3_csum_replace(skb, offset, from, to, flags) | |
345 | * recompute IP checksum | |
346 | * @skb: pointer to skb | |
347 | * @offset: offset within packet where IP checksum is located | |
348 | * @from: old value of header field | |
349 | * @to: new value of header field | |
350 | * @flags: bits 0-3 - size of header field | |
351 | * other bits - reserved | |
352 | * Return: 0 on success or negative error | |
353 | * | |
354 | * int bpf_l4_csum_replace(skb, offset, from, to, flags) | |
355 | * recompute TCP/UDP checksum | |
356 | * @skb: pointer to skb | |
357 | * @offset: offset within packet where TCP/UDP checksum is located | |
358 | * @from: old value of header field | |
359 | * @to: new value of header field | |
360 | * @flags: bits 0-3 - size of header field | |
361 | * bit 4 - is pseudo header | |
362 | * other bits - reserved | |
363 | * Return: 0 on success or negative error | |
364 | * | |
365 | * int bpf_tail_call(ctx, prog_array_map, index) | |
366 | * jump into another BPF program | |
367 | * @ctx: context pointer passed to next program | |
368 | * @prog_array_map: pointer to map which type is BPF_MAP_TYPE_PROG_ARRAY | |
aa7b4e02 | 369 | * @index: 32-bit index inside array that selects specific program to run |
0cb34dc2 JS |
370 | * Return: 0 on success or negative error |
371 | * | |
372 | * int bpf_clone_redirect(skb, ifindex, flags) | |
373 | * redirect to another netdev | |
374 | * @skb: pointer to skb | |
375 | * @ifindex: ifindex of the net device | |
376 | * @flags: bit 0 - if set, redirect to ingress instead of egress | |
377 | * other bits - reserved | |
378 | * Return: 0 on success or negative error | |
379 | * | |
380 | * u64 bpf_get_current_pid_tgid(void) | |
381 | * Return: current->tgid << 32 | current->pid | |
382 | * | |
383 | * u64 bpf_get_current_uid_gid(void) | |
384 | * Return: current_gid << 32 | current_uid | |
385 | * | |
386 | * int bpf_get_current_comm(char *buf, int size_of_buf) | |
387 | * stores current->comm into buf | |
388 | * Return: 0 on success or negative error | |
389 | * | |
390 | * u32 bpf_get_cgroup_classid(skb) | |
391 | * retrieve a proc's classid | |
392 | * @skb: pointer to skb | |
393 | * Return: classid if != 0 | |
394 | * | |
395 | * int bpf_skb_vlan_push(skb, vlan_proto, vlan_tci) | |
396 | * Return: 0 on success or negative error | |
397 | * | |
398 | * int bpf_skb_vlan_pop(skb) | |
399 | * Return: 0 on success or negative error | |
400 | * | |
401 | * int bpf_skb_get_tunnel_key(skb, key, size, flags) | |
402 | * int bpf_skb_set_tunnel_key(skb, key, size, flags) | |
403 | * retrieve or populate tunnel metadata | |
404 | * @skb: pointer to skb | |
405 | * @key: pointer to 'struct bpf_tunnel_key' | |
406 | * @size: size of 'struct bpf_tunnel_key' | |
407 | * @flags: room for future extensions | |
408 | * Return: 0 on success or negative error | |
409 | * | |
b7d3ed5b TQ |
410 | * u64 bpf_perf_event_read(map, flags) |
411 | * read perf event counter value | |
412 | * @map: pointer to perf_event_array map | |
413 | * @flags: index of event in the map or bitmask flags | |
414 | * Return: value of perf event counter read or error code | |
0cb34dc2 JS |
415 | * |
416 | * int bpf_redirect(ifindex, flags) | |
417 | * redirect to another netdev | |
418 | * @ifindex: ifindex of the net device | |
ac29991b DB |
419 | * @flags: |
420 | * cls_bpf: | |
421 | * bit 0 - if set, redirect to ingress instead of egress | |
422 | * other bits - reserved | |
423 | * xdp_bpf: | |
424 | * all bits - reserved | |
425 | * Return: cls_bpf: TC_ACT_REDIRECT on success or TC_ACT_SHOT on error | |
426 | * xdp_bfp: XDP_REDIRECT on success or XDP_ABORT on error | |
427 | * int bpf_redirect_map(map, key, flags) | |
428 | * redirect to endpoint in map | |
429 | * @map: pointer to dev map | |
430 | * @key: index in map to lookup | |
431 | * @flags: -- | |
432 | * Return: XDP_REDIRECT on success or XDP_ABORT on error | |
0cb34dc2 JS |
433 | * |
434 | * u32 bpf_get_route_realm(skb) | |
435 | * retrieve a dst's tclassid | |
436 | * @skb: pointer to skb | |
437 | * Return: realm if != 0 | |
438 | * | |
b7d3ed5b | 439 | * int bpf_perf_event_output(ctx, map, flags, data, size) |
0cb34dc2 JS |
440 | * output perf raw sample |
441 | * @ctx: struct pt_regs* | |
442 | * @map: pointer to perf_event_array map | |
b7d3ed5b | 443 | * @flags: index of event in the map or bitmask flags |
0cb34dc2 JS |
444 | * @data: data on stack to be output as raw data |
445 | * @size: size of data | |
446 | * Return: 0 on success or negative error | |
447 | * | |
448 | * int bpf_get_stackid(ctx, map, flags) | |
449 | * walk user or kernel stack and return id | |
450 | * @ctx: struct pt_regs* | |
451 | * @map: pointer to stack_trace map | |
452 | * @flags: bits 0-7 - numer of stack frames to skip | |
453 | * bit 8 - collect user stack instead of kernel | |
454 | * bit 9 - compare stacks by hash only | |
455 | * bit 10 - if two different stacks hash into the same stackid | |
456 | * discard old | |
457 | * other bits - reserved | |
458 | * Return: >= 0 stackid on success or negative error | |
459 | * | |
460 | * s64 bpf_csum_diff(from, from_size, to, to_size, seed) | |
461 | * calculate csum diff | |
462 | * @from: raw from buffer | |
463 | * @from_size: length of from buffer | |
464 | * @to: raw to buffer | |
465 | * @to_size: length of to buffer | |
466 | * @seed: optional seed | |
467 | * Return: csum result or negative error code | |
468 | * | |
469 | * int bpf_skb_get_tunnel_opt(skb, opt, size) | |
470 | * retrieve tunnel options metadata | |
471 | * @skb: pointer to skb | |
472 | * @opt: pointer to raw tunnel option data | |
473 | * @size: size of @opt | |
474 | * Return: option size | |
475 | * | |
476 | * int bpf_skb_set_tunnel_opt(skb, opt, size) | |
477 | * populate tunnel options metadata | |
478 | * @skb: pointer to skb | |
479 | * @opt: pointer to raw tunnel option data | |
480 | * @size: size of @opt | |
481 | * Return: 0 on success or negative error | |
482 | * | |
483 | * int bpf_skb_change_proto(skb, proto, flags) | |
484 | * Change protocol of the skb. Currently supported is v4 -> v6, | |
485 | * v6 -> v4 transitions. The helper will also resize the skb. eBPF | |
486 | * program is expected to fill the new headers via skb_store_bytes | |
487 | * and lX_csum_replace. | |
488 | * @skb: pointer to skb | |
489 | * @proto: new skb->protocol type | |
490 | * @flags: reserved | |
491 | * Return: 0 on success or negative error | |
492 | * | |
493 | * int bpf_skb_change_type(skb, type) | |
494 | * Change packet type of skb. | |
495 | * @skb: pointer to skb | |
496 | * @type: new skb->pkt_type type | |
497 | * Return: 0 on success or negative error | |
498 | * | |
499 | * int bpf_skb_under_cgroup(skb, map, index) | |
500 | * Check cgroup2 membership of skb | |
501 | * @skb: pointer to skb | |
502 | * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type | |
503 | * @index: index of the cgroup in the bpf_map | |
504 | * Return: | |
505 | * == 0 skb failed the cgroup2 descendant test | |
506 | * == 1 skb succeeded the cgroup2 descendant test | |
507 | * < 0 error | |
508 | * | |
509 | * u32 bpf_get_hash_recalc(skb) | |
510 | * Retrieve and possibly recalculate skb->hash. | |
511 | * @skb: pointer to skb | |
512 | * Return: hash | |
513 | * | |
514 | * u64 bpf_get_current_task(void) | |
515 | * Returns current task_struct | |
516 | * Return: current | |
517 | * | |
518 | * int bpf_probe_write_user(void *dst, void *src, int len) | |
519 | * safely attempt to write to a location | |
520 | * @dst: destination address in userspace | |
521 | * @src: source address on stack | |
522 | * @len: number of bytes to copy | |
523 | * Return: 0 on success or negative error | |
524 | * | |
525 | * int bpf_current_task_under_cgroup(map, index) | |
526 | * Check cgroup2 membership of current task | |
527 | * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type | |
528 | * @index: index of the cgroup in the bpf_map | |
529 | * Return: | |
530 | * == 0 current failed the cgroup2 descendant test | |
531 | * == 1 current succeeded the cgroup2 descendant test | |
532 | * < 0 error | |
533 | * | |
534 | * int bpf_skb_change_tail(skb, len, flags) | |
535 | * The helper will resize the skb to the given new size, to be used f.e. | |
536 | * with control messages. | |
537 | * @skb: pointer to skb | |
538 | * @len: new skb length | |
539 | * @flags: reserved | |
540 | * Return: 0 on success or negative error | |
541 | * | |
542 | * int bpf_skb_pull_data(skb, len) | |
543 | * The helper will pull in non-linear data in case the skb is non-linear | |
544 | * and not all of len are part of the linear section. Only needed for | |
545 | * read/write with direct packet access. | |
546 | * @skb: pointer to skb | |
547 | * @len: len to make read/writeable | |
548 | * Return: 0 on success or negative error | |
549 | * | |
550 | * s64 bpf_csum_update(skb, csum) | |
551 | * Adds csum into skb->csum in case of CHECKSUM_COMPLETE. | |
552 | * @skb: pointer to skb | |
553 | * @csum: csum to add | |
554 | * Return: csum on success or negative error | |
555 | * | |
556 | * void bpf_set_hash_invalid(skb) | |
557 | * Invalidate current skb->hash. | |
558 | * @skb: pointer to skb | |
559 | * | |
560 | * int bpf_get_numa_node_id() | |
561 | * Return: Id of current NUMA node. | |
562 | * | |
563 | * int bpf_skb_change_head() | |
564 | * Grows headroom of skb and adjusts MAC header offset accordingly. | |
565 | * Will extends/reallocae as required automatically. | |
566 | * May change skb data pointer and will thus invalidate any check | |
567 | * performed for direct packet access. | |
568 | * @skb: pointer to skb | |
569 | * @len: length of header to be pushed in front | |
570 | * @flags: Flags (unused for now) | |
571 | * Return: 0 on success or negative error | |
572 | * | |
573 | * int bpf_xdp_adjust_head(xdp_md, delta) | |
574 | * Adjust the xdp_md.data by delta | |
575 | * @xdp_md: pointer to xdp_md | |
576 | * @delta: An positive/negative integer to be added to xdp_md.data | |
577 | * Return: 0 on success or negative on error | |
9a738266 MS |
578 | * |
579 | * int bpf_probe_read_str(void *dst, int size, const void *unsafe_ptr) | |
580 | * Copy a NUL terminated string from unsafe address. In case the string | |
581 | * length is smaller than size, the target is not padded with further NUL | |
582 | * bytes. In case the string length is larger than size, just count-1 | |
583 | * bytes are copied and the last byte is set to NUL. | |
584 | * @dst: destination address | |
585 | * @size: maximum number of bytes to copy, including the trailing NUL | |
586 | * @unsafe_ptr: unsafe address | |
587 | * Return: | |
588 | * > 0 length of the string including the trailing NUL on success | |
589 | * < 0 error | |
30848873 | 590 | * |
3c60a531 | 591 | * u64 bpf_get_socket_cookie(skb) |
30848873 AS |
592 | * Get the cookie for the socket stored inside sk_buff. |
593 | * @skb: pointer to skb | |
594 | * Return: 8 Bytes non-decreasing number on success or 0 if the socket | |
595 | * field is missing inside sk_buff | |
596 | * | |
597 | * u32 bpf_get_socket_uid(skb) | |
598 | * Get the owner uid of the socket stored inside sk_buff. | |
599 | * @skb: pointer to skb | |
e07b98d9 | 600 | * Return: uid of the socket owner on success or overflowuid if failed. |
ded092cd DB |
601 | * |
602 | * u32 bpf_set_hash(skb, hash) | |
603 | * Set full skb->hash. | |
604 | * @skb: pointer to skb | |
605 | * @hash: hash to set | |
04df41e3 LB |
606 | * |
607 | * int bpf_setsockopt(bpf_socket, level, optname, optval, optlen) | |
608 | * Calls setsockopt. Not all opts are available, only those with | |
609 | * integer optvals plus TCP_CONGESTION. | |
610 | * Supported levels: SOL_SOCKET and IPROTO_TCP | |
611 | * @bpf_socket: pointer to bpf_socket | |
612 | * @level: SOL_SOCKET or IPROTO_TCP | |
613 | * @optname: option name | |
614 | * @optval: pointer to option value | |
615 | * @optlen: length of optval in byes | |
616 | * Return: 0 or negative error | |
2be7e212 DB |
617 | * |
618 | * int bpf_skb_adjust_room(skb, len_diff, mode, flags) | |
619 | * Grow or shrink room in sk_buff. | |
620 | * @skb: pointer to skb | |
621 | * @len_diff: (signed) amount of room to grow/shrink | |
622 | * @mode: operation mode (enum bpf_adj_room_mode) | |
623 | * @flags: reserved for future use | |
624 | * Return: 0 on success or negative error code | |
69e8cc13 | 625 | * |
34f79502 | 626 | * int bpf_sk_redirect_map(skb, map, key, flags) |
69e8cc13 JF |
627 | * Redirect skb to a sock in map using key as a lookup key for the |
628 | * sock in map. | |
34f79502 | 629 | * @skb: pointer to skb |
69e8cc13 JF |
630 | * @map: pointer to sockmap |
631 | * @key: key to lookup sock in map | |
632 | * @flags: reserved for future use | |
633 | * Return: SK_REDIRECT | |
634 | * | |
464bc0fd | 635 | * int bpf_sock_map_update(skops, map, key, flags) |
69e8cc13 JF |
636 | * @skops: pointer to bpf_sock_ops |
637 | * @map: pointer to sockmap to update | |
638 | * @key: key to insert/update sock in map | |
639 | * @flags: same flags as map update elem | |
ac29991b DB |
640 | * |
641 | * int bpf_xdp_adjust_meta(xdp_md, delta) | |
642 | * Adjust the xdp_md.data_meta by delta | |
643 | * @xdp_md: pointer to xdp_md | |
644 | * @delta: An positive/negative integer to be added to xdp_md.data_meta | |
645 | * Return: 0 on success or negative on error | |
0cb34dc2 JS |
646 | */ |
647 | #define __BPF_FUNC_MAPPER(FN) \ | |
648 | FN(unspec), \ | |
649 | FN(map_lookup_elem), \ | |
650 | FN(map_update_elem), \ | |
651 | FN(map_delete_elem), \ | |
652 | FN(probe_read), \ | |
653 | FN(ktime_get_ns), \ | |
654 | FN(trace_printk), \ | |
655 | FN(get_prandom_u32), \ | |
656 | FN(get_smp_processor_id), \ | |
657 | FN(skb_store_bytes), \ | |
658 | FN(l3_csum_replace), \ | |
659 | FN(l4_csum_replace), \ | |
660 | FN(tail_call), \ | |
661 | FN(clone_redirect), \ | |
662 | FN(get_current_pid_tgid), \ | |
663 | FN(get_current_uid_gid), \ | |
664 | FN(get_current_comm), \ | |
665 | FN(get_cgroup_classid), \ | |
666 | FN(skb_vlan_push), \ | |
667 | FN(skb_vlan_pop), \ | |
668 | FN(skb_get_tunnel_key), \ | |
669 | FN(skb_set_tunnel_key), \ | |
670 | FN(perf_event_read), \ | |
671 | FN(redirect), \ | |
672 | FN(get_route_realm), \ | |
673 | FN(perf_event_output), \ | |
674 | FN(skb_load_bytes), \ | |
675 | FN(get_stackid), \ | |
676 | FN(csum_diff), \ | |
677 | FN(skb_get_tunnel_opt), \ | |
678 | FN(skb_set_tunnel_opt), \ | |
679 | FN(skb_change_proto), \ | |
680 | FN(skb_change_type), \ | |
681 | FN(skb_under_cgroup), \ | |
682 | FN(get_hash_recalc), \ | |
683 | FN(get_current_task), \ | |
684 | FN(probe_write_user), \ | |
685 | FN(current_task_under_cgroup), \ | |
686 | FN(skb_change_tail), \ | |
687 | FN(skb_pull_data), \ | |
688 | FN(csum_update), \ | |
689 | FN(set_hash_invalid), \ | |
690 | FN(get_numa_node_id), \ | |
691 | FN(skb_change_head), \ | |
9a738266 | 692 | FN(xdp_adjust_head), \ |
91b8270f | 693 | FN(probe_read_str), \ |
6acc5c29 | 694 | FN(get_socket_cookie), \ |
ded092cd | 695 | FN(get_socket_uid), \ |
04df41e3 | 696 | FN(set_hash), \ |
2be7e212 | 697 | FN(setsockopt), \ |
996139e8 | 698 | FN(skb_adjust_room), \ |
69e8cc13 JF |
699 | FN(redirect_map), \ |
700 | FN(sk_redirect_map), \ | |
ac29991b | 701 | FN(sock_map_update), \ |
020a32d9 | 702 | FN(xdp_adjust_meta), \ |
81b9cf80 YS |
703 | FN(perf_event_read_value), \ |
704 | FN(perf_prog_read_value), | |
0cb34dc2 | 705 | |
971e827b ACM |
706 | /* integer value in 'imm' field of BPF_CALL instruction selects which helper |
707 | * function eBPF program intends to call | |
708 | */ | |
0cb34dc2 | 709 | #define __BPF_ENUM_FN(x) BPF_FUNC_ ## x |
971e827b | 710 | enum bpf_func_id { |
0cb34dc2 | 711 | __BPF_FUNC_MAPPER(__BPF_ENUM_FN) |
971e827b ACM |
712 | __BPF_FUNC_MAX_ID, |
713 | }; | |
0cb34dc2 | 714 | #undef __BPF_ENUM_FN |
971e827b ACM |
715 | |
716 | /* All flags used by eBPF helper functions, placed here. */ | |
717 | ||
718 | /* BPF_FUNC_skb_store_bytes flags. */ | |
719 | #define BPF_F_RECOMPUTE_CSUM (1ULL << 0) | |
720 | #define BPF_F_INVALIDATE_HASH (1ULL << 1) | |
721 | ||
722 | /* BPF_FUNC_l3_csum_replace and BPF_FUNC_l4_csum_replace flags. | |
723 | * First 4 bits are for passing the header field size. | |
724 | */ | |
725 | #define BPF_F_HDR_FIELD_MASK 0xfULL | |
726 | ||
727 | /* BPF_FUNC_l4_csum_replace flags. */ | |
728 | #define BPF_F_PSEUDO_HDR (1ULL << 4) | |
729 | #define BPF_F_MARK_MANGLED_0 (1ULL << 5) | |
9a738266 | 730 | #define BPF_F_MARK_ENFORCE (1ULL << 6) |
971e827b ACM |
731 | |
732 | /* BPF_FUNC_clone_redirect and BPF_FUNC_redirect flags. */ | |
733 | #define BPF_F_INGRESS (1ULL << 0) | |
734 | ||
735 | /* BPF_FUNC_skb_set_tunnel_key and BPF_FUNC_skb_get_tunnel_key flags. */ | |
736 | #define BPF_F_TUNINFO_IPV6 (1ULL << 0) | |
737 | ||
738 | /* BPF_FUNC_get_stackid flags. */ | |
739 | #define BPF_F_SKIP_FIELD_MASK 0xffULL | |
740 | #define BPF_F_USER_STACK (1ULL << 8) | |
741 | #define BPF_F_FAST_STACK_CMP (1ULL << 9) | |
742 | #define BPF_F_REUSE_STACKID (1ULL << 10) | |
743 | ||
744 | /* BPF_FUNC_skb_set_tunnel_key flags. */ | |
745 | #define BPF_F_ZERO_CSUM_TX (1ULL << 1) | |
746 | #define BPF_F_DONT_FRAGMENT (1ULL << 2) | |
747 | ||
791cceb8 | 748 | /* BPF_FUNC_perf_event_output and BPF_FUNC_perf_event_read flags. */ |
971e827b ACM |
749 | #define BPF_F_INDEX_MASK 0xffffffffULL |
750 | #define BPF_F_CURRENT_CPU BPF_F_INDEX_MASK | |
791cceb8 ACM |
751 | /* BPF_FUNC_perf_event_output for sk_buff input context. */ |
752 | #define BPF_F_CTXLEN_MASK (0xfffffULL << 32) | |
971e827b | 753 | |
2be7e212 DB |
754 | /* Mode for BPF_FUNC_skb_adjust_room helper. */ |
755 | enum bpf_adj_room_mode { | |
d62c1d72 | 756 | BPF_ADJ_ROOM_NET, |
2be7e212 DB |
757 | }; |
758 | ||
971e827b ACM |
759 | /* user accessible mirror of in-kernel sk_buff. |
760 | * new fields can only be added to the end of this structure | |
761 | */ | |
762 | struct __sk_buff { | |
763 | __u32 len; | |
764 | __u32 pkt_type; | |
765 | __u32 mark; | |
766 | __u32 queue_mapping; | |
767 | __u32 protocol; | |
768 | __u32 vlan_present; | |
769 | __u32 vlan_tci; | |
770 | __u32 vlan_proto; | |
771 | __u32 priority; | |
772 | __u32 ingress_ifindex; | |
773 | __u32 ifindex; | |
774 | __u32 tc_index; | |
775 | __u32 cb[5]; | |
776 | __u32 hash; | |
777 | __u32 tc_classid; | |
778 | __u32 data; | |
779 | __u32 data_end; | |
b1d9fc41 | 780 | __u32 napi_id; |
69e8cc13 | 781 | |
ac29991b | 782 | /* Accessed by BPF_PROG_TYPE_sk_skb types from here to ... */ |
69e8cc13 JF |
783 | __u32 family; |
784 | __u32 remote_ip4; /* Stored in network byte order */ | |
785 | __u32 local_ip4; /* Stored in network byte order */ | |
786 | __u32 remote_ip6[4]; /* Stored in network byte order */ | |
787 | __u32 local_ip6[4]; /* Stored in network byte order */ | |
788 | __u32 remote_port; /* Stored in network byte order */ | |
789 | __u32 local_port; /* stored in host byte order */ | |
ac29991b DB |
790 | /* ... here. */ |
791 | ||
792 | __u32 data_meta; | |
971e827b ACM |
793 | }; |
794 | ||
795 | struct bpf_tunnel_key { | |
796 | __u32 tunnel_id; | |
797 | union { | |
798 | __u32 remote_ipv4; | |
799 | __u32 remote_ipv6[4]; | |
800 | }; | |
801 | __u8 tunnel_tos; | |
802 | __u8 tunnel_ttl; | |
803 | __u16 tunnel_ext; | |
804 | __u32 tunnel_label; | |
805 | }; | |
806 | ||
0cb34dc2 JS |
807 | /* Generic BPF return codes which all BPF program types may support. |
808 | * The values are binary compatible with their TC_ACT_* counter-part to | |
809 | * provide backwards compatibility with existing SCHED_CLS and SCHED_ACT | |
810 | * programs. | |
811 | * | |
812 | * XDP is handled seprately, see XDP_*. | |
813 | */ | |
814 | enum bpf_ret_code { | |
815 | BPF_OK = 0, | |
816 | /* 1 reserved */ | |
817 | BPF_DROP = 2, | |
818 | /* 3-6 reserved */ | |
819 | BPF_REDIRECT = 7, | |
820 | /* >127 are reserved for prog type specific return codes */ | |
821 | }; | |
822 | ||
823 | struct bpf_sock { | |
824 | __u32 bound_dev_if; | |
825 | __u32 family; | |
826 | __u32 type; | |
827 | __u32 protocol; | |
ac29991b DB |
828 | __u32 mark; |
829 | __u32 priority; | |
0cb34dc2 JS |
830 | }; |
831 | ||
832 | #define XDP_PACKET_HEADROOM 256 | |
833 | ||
791cceb8 ACM |
834 | /* User return codes for XDP prog type. |
835 | * A valid XDP program must return one of these defined values. All other | |
ac29991b DB |
836 | * return codes are reserved for future use. Unknown return codes will |
837 | * result in packet drops and a warning via bpf_warn_invalid_xdp_action(). | |
791cceb8 ACM |
838 | */ |
839 | enum xdp_action { | |
840 | XDP_ABORTED = 0, | |
841 | XDP_DROP, | |
842 | XDP_PASS, | |
843 | XDP_TX, | |
ac29991b | 844 | XDP_REDIRECT, |
791cceb8 ACM |
845 | }; |
846 | ||
847 | /* user accessible metadata for XDP packet hook | |
848 | * new fields must be added to the end of this structure | |
849 | */ | |
850 | struct xdp_md { | |
851 | __u32 data; | |
852 | __u32 data_end; | |
ac29991b | 853 | __u32 data_meta; |
791cceb8 ACM |
854 | }; |
855 | ||
69e8cc13 JF |
856 | enum sk_action { |
857 | SK_ABORTED = 0, | |
858 | SK_DROP, | |
859 | SK_REDIRECT, | |
860 | }; | |
861 | ||
95b9afd3 MKL |
862 | #define BPF_TAG_SIZE 8 |
863 | ||
864 | struct bpf_prog_info { | |
865 | __u32 type; | |
866 | __u32 id; | |
867 | __u8 tag[BPF_TAG_SIZE]; | |
868 | __u32 jited_prog_len; | |
869 | __u32 xlated_prog_len; | |
870 | __aligned_u64 jited_prog_insns; | |
871 | __aligned_u64 xlated_prog_insns; | |
88cda1c9 MKL |
872 | __u64 load_time; /* ns since boottime */ |
873 | __u32 created_by_uid; | |
874 | __u32 nr_map_ids; | |
875 | __aligned_u64 map_ids; | |
067cae47 | 876 | char name[BPF_OBJ_NAME_LEN]; |
95b9afd3 MKL |
877 | } __attribute__((aligned(8))); |
878 | ||
879 | struct bpf_map_info { | |
880 | __u32 type; | |
881 | __u32 id; | |
882 | __u32 key_size; | |
883 | __u32 value_size; | |
884 | __u32 max_entries; | |
885 | __u32 map_flags; | |
067cae47 | 886 | char name[BPF_OBJ_NAME_LEN]; |
95b9afd3 MKL |
887 | } __attribute__((aligned(8))); |
888 | ||
04df41e3 LB |
889 | /* User bpf_sock_ops struct to access socket values and specify request ops |
890 | * and their replies. | |
f1d6cb2d ACM |
891 | * Some of this fields are in network (bigendian) byte order and may need |
892 | * to be converted before use (bpf_ntohl() defined in samples/bpf/bpf_endian.h). | |
04df41e3 LB |
893 | * New fields can only be added at the end of this structure |
894 | */ | |
895 | struct bpf_sock_ops { | |
896 | __u32 op; | |
897 | union { | |
898 | __u32 reply; | |
899 | __u32 replylong[4]; | |
900 | }; | |
901 | __u32 family; | |
f1d6cb2d ACM |
902 | __u32 remote_ip4; /* Stored in network byte order */ |
903 | __u32 local_ip4; /* Stored in network byte order */ | |
904 | __u32 remote_ip6[4]; /* Stored in network byte order */ | |
905 | __u32 local_ip6[4]; /* Stored in network byte order */ | |
906 | __u32 remote_port; /* Stored in network byte order */ | |
907 | __u32 local_port; /* stored in host byte order */ | |
04df41e3 LB |
908 | }; |
909 | ||
910 | /* List of known BPF sock_ops operators. | |
911 | * New entries can only be added at the end | |
912 | */ | |
913 | enum { | |
914 | BPF_SOCK_OPS_VOID, | |
915 | BPF_SOCK_OPS_TIMEOUT_INIT, /* Should return SYN-RTO value to use or | |
916 | * -1 if default value should be used | |
917 | */ | |
918 | BPF_SOCK_OPS_RWND_INIT, /* Should return initial advertized | |
919 | * window (in packets) or -1 if default | |
920 | * value should be used | |
921 | */ | |
922 | BPF_SOCK_OPS_TCP_CONNECT_CB, /* Calls BPF program right before an | |
923 | * active connection is initialized | |
924 | */ | |
925 | BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB, /* Calls BPF program when an | |
926 | * active connection is | |
927 | * established | |
928 | */ | |
929 | BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB, /* Calls BPF program when a | |
930 | * passive connection is | |
931 | * established | |
932 | */ | |
933 | BPF_SOCK_OPS_NEEDS_ECN, /* If connection's congestion control | |
934 | * needs ECN | |
935 | */ | |
936 | }; | |
937 | ||
938 | #define TCP_BPF_IW 1001 /* Set TCP initial congestion window */ | |
939 | #define TCP_BPF_SNDCWND_CLAMP 1002 /* Set sndcwnd_clamp */ | |
940 | ||
971e827b | 941 | #endif /* _UAPI__LINUX_BPF_H__ */ |