1 /* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */ 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> 12 #include <linux/bpf_common.h> 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 */ 20 #define BPF_DW 0x18 /* double word */ 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 34 /* jmp encodings */ 35 #define BPF_JNE 0x50 /* jump != */ 36 #define BPF_JLT 0xa0 /* LT is unsigned, '<' */ 37 #define BPF_JLE 0xb0 /* LE is unsigned, '<=' */ 38 #define BPF_JSGT 0x60 /* SGT is signed '>', GT in x86 */ 39 #define BPF_JSGE 0x70 /* SGE is signed '>=', GE in x86 */ 40 #define BPF_JSLT 0xc0 /* SLT is signed, '<' */ 41 #define BPF_JSLE 0xd0 /* SLE is signed, '<=' */ 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 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 78 /* BPF syscall commands, see bpf(2) man-page for details. */ 79 enum bpf_cmd { 80 BPF_MAP_CREATE, 81 BPF_MAP_LOOKUP_ELEM, 82 BPF_MAP_UPDATE_ELEM, 83 BPF_MAP_DELETE_ELEM, 84 BPF_MAP_GET_NEXT_KEY, 85 BPF_PROG_LOAD, 86 BPF_OBJ_PIN, 87 BPF_OBJ_GET, 88 BPF_PROG_ATTACH, 89 BPF_PROG_DETACH, 90 BPF_PROG_TEST_RUN, 91 BPF_PROG_GET_NEXT_ID, 92 BPF_MAP_GET_NEXT_ID, 93 BPF_PROG_GET_FD_BY_ID, 94 BPF_MAP_GET_FD_BY_ID, 95 BPF_OBJ_GET_INFO_BY_FD, 96 BPF_PROG_QUERY, 97 }; 98 99 enum bpf_map_type { 100 BPF_MAP_TYPE_UNSPEC, 101 BPF_MAP_TYPE_HASH, 102 BPF_MAP_TYPE_ARRAY, 103 BPF_MAP_TYPE_PROG_ARRAY, 104 BPF_MAP_TYPE_PERF_EVENT_ARRAY, 105 BPF_MAP_TYPE_PERCPU_HASH, 106 BPF_MAP_TYPE_PERCPU_ARRAY, 107 BPF_MAP_TYPE_STACK_TRACE, 108 BPF_MAP_TYPE_CGROUP_ARRAY, 109 BPF_MAP_TYPE_LRU_HASH, 110 BPF_MAP_TYPE_LRU_PERCPU_HASH, 111 BPF_MAP_TYPE_LPM_TRIE, 112 BPF_MAP_TYPE_ARRAY_OF_MAPS, 113 BPF_MAP_TYPE_HASH_OF_MAPS, 114 BPF_MAP_TYPE_DEVMAP, 115 BPF_MAP_TYPE_SOCKMAP, 116 BPF_MAP_TYPE_CPUMAP, 117 }; 118 119 enum bpf_prog_type { 120 BPF_PROG_TYPE_UNSPEC, 121 BPF_PROG_TYPE_SOCKET_FILTER, 122 BPF_PROG_TYPE_KPROBE, 123 BPF_PROG_TYPE_SCHED_CLS, 124 BPF_PROG_TYPE_SCHED_ACT, 125 BPF_PROG_TYPE_TRACEPOINT, 126 BPF_PROG_TYPE_XDP, 127 BPF_PROG_TYPE_PERF_EVENT, 128 BPF_PROG_TYPE_CGROUP_SKB, 129 BPF_PROG_TYPE_CGROUP_SOCK, 130 BPF_PROG_TYPE_LWT_IN, 131 BPF_PROG_TYPE_LWT_OUT, 132 BPF_PROG_TYPE_LWT_XMIT, 133 BPF_PROG_TYPE_SOCK_OPS, 134 BPF_PROG_TYPE_SK_SKB, 135 BPF_PROG_TYPE_CGROUP_DEVICE, 136 }; 137 138 enum bpf_attach_type { 139 BPF_CGROUP_INET_INGRESS, 140 BPF_CGROUP_INET_EGRESS, 141 BPF_CGROUP_INET_SOCK_CREATE, 142 BPF_CGROUP_SOCK_OPS, 143 BPF_SK_SKB_STREAM_PARSER, 144 BPF_SK_SKB_STREAM_VERDICT, 145 BPF_CGROUP_DEVICE, 146 __MAX_BPF_ATTACH_TYPE 147 }; 148 149 #define MAX_BPF_ATTACH_TYPE __MAX_BPF_ATTACH_TYPE 150 151 /* cgroup-bpf attach flags used in BPF_PROG_ATTACH command 152 * 153 * NONE(default): No further bpf programs allowed in the subtree. 154 * 155 * BPF_F_ALLOW_OVERRIDE: If a sub-cgroup installs some bpf program, 156 * the program in this cgroup yields to sub-cgroup program. 157 * 158 * BPF_F_ALLOW_MULTI: If a sub-cgroup installs some bpf program, 159 * that cgroup program gets run in addition to the program in this cgroup. 160 * 161 * Only one program is allowed to be attached to a cgroup with 162 * NONE or BPF_F_ALLOW_OVERRIDE flag. 163 * Attaching another program on top of NONE or BPF_F_ALLOW_OVERRIDE will 164 * release old program and attach the new one. Attach flags has to match. 165 * 166 * Multiple programs are allowed to be attached to a cgroup with 167 * BPF_F_ALLOW_MULTI flag. They are executed in FIFO order 168 * (those that were attached first, run first) 169 * The programs of sub-cgroup are executed first, then programs of 170 * this cgroup and then programs of parent cgroup. 171 * When children program makes decision (like picking TCP CA or sock bind) 172 * parent program has a chance to override it. 173 * 174 * A cgroup with MULTI or OVERRIDE flag allows any attach flags in sub-cgroups. 175 * A cgroup with NONE doesn't allow any programs in sub-cgroups. 176 * Ex1: 177 * cgrp1 (MULTI progs A, B) -> 178 * cgrp2 (OVERRIDE prog C) -> 179 * cgrp3 (MULTI prog D) -> 180 * cgrp4 (OVERRIDE prog E) -> 181 * cgrp5 (NONE prog F) 182 * the event in cgrp5 triggers execution of F,D,A,B in that order. 183 * if prog F is detached, the execution is E,D,A,B 184 * if prog F and D are detached, the execution is E,A,B 185 * if prog F, E and D are detached, the execution is C,A,B 186 * 187 * All eligible programs are executed regardless of return code from 188 * earlier programs. 189 */ 190 #define BPF_F_ALLOW_OVERRIDE (1U << 0) 191 #define BPF_F_ALLOW_MULTI (1U << 1) 192 193 /* If BPF_F_STRICT_ALIGNMENT is used in BPF_PROG_LOAD command, the 194 * verifier will perform strict alignment checking as if the kernel 195 * has been built with CONFIG_EFFICIENT_UNALIGNED_ACCESS not set, 196 * and NET_IP_ALIGN defined to 2. 197 */ 198 #define BPF_F_STRICT_ALIGNMENT (1U << 0) 199 200 #define BPF_PSEUDO_MAP_FD 1 201 202 /* flags for BPF_MAP_UPDATE_ELEM command */ 203 #define BPF_ANY 0 /* create new element or update existing */ 204 #define BPF_NOEXIST 1 /* create new element if it didn't exist */ 205 #define BPF_EXIST 2 /* update existing element */ 206 207 /* flags for BPF_MAP_CREATE command */ 208 #define BPF_F_NO_PREALLOC (1U << 0) 209 /* Instead of having one common LRU list in the 210 * BPF_MAP_TYPE_LRU_[PERCPU_]HASH map, use a percpu LRU list 211 * which can scale and perform better. 212 * Note, the LRU nodes (including free nodes) cannot be moved 213 * across different LRU lists. 214 */ 215 #define BPF_F_NO_COMMON_LRU (1U << 1) 216 /* Specify numa node during map creation */ 217 #define BPF_F_NUMA_NODE (1U << 2) 218 219 /* flags for BPF_PROG_QUERY */ 220 #define BPF_F_QUERY_EFFECTIVE (1U << 0) 221 222 #define BPF_OBJ_NAME_LEN 16U 223 224 /* Flags for accessing BPF object */ 225 #define BPF_F_RDONLY (1U << 3) 226 #define BPF_F_WRONLY (1U << 4) 227 228 union bpf_attr { 229 struct { /* anonymous struct used by BPF_MAP_CREATE command */ 230 __u32 map_type; /* one of enum bpf_map_type */ 231 __u32 key_size; /* size of key in bytes */ 232 __u32 value_size; /* size of value in bytes */ 233 __u32 max_entries; /* max number of entries in a map */ 234 __u32 map_flags; /* BPF_MAP_CREATE related 235 * flags defined above. 236 */ 237 __u32 inner_map_fd; /* fd pointing to the inner map */ 238 __u32 numa_node; /* numa node (effective only if 239 * BPF_F_NUMA_NODE is set). 240 */ 241 char map_name[BPF_OBJ_NAME_LEN]; 242 }; 243 244 struct { /* anonymous struct used by BPF_MAP_*_ELEM commands */ 245 __u32 map_fd; 246 __aligned_u64 key; 247 union { 248 __aligned_u64 value; 249 __aligned_u64 next_key; 250 }; 251 __u64 flags; 252 }; 253 254 struct { /* anonymous struct used by BPF_PROG_LOAD command */ 255 __u32 prog_type; /* one of enum bpf_prog_type */ 256 __u32 insn_cnt; 257 __aligned_u64 insns; 258 __aligned_u64 license; 259 __u32 log_level; /* verbosity level of verifier */ 260 __u32 log_size; /* size of user buffer */ 261 __aligned_u64 log_buf; /* user supplied buffer */ 262 __u32 kern_version; /* checked when prog_type=kprobe */ 263 __u32 prog_flags; 264 char prog_name[BPF_OBJ_NAME_LEN]; 265 __u32 prog_ifindex; /* ifindex of netdev to prep for */ 266 }; 267 268 struct { /* anonymous struct used by BPF_OBJ_* commands */ 269 __aligned_u64 pathname; 270 __u32 bpf_fd; 271 __u32 file_flags; 272 }; 273 274 struct { /* anonymous struct used by BPF_PROG_ATTACH/DETACH commands */ 275 __u32 target_fd; /* container object to attach to */ 276 __u32 attach_bpf_fd; /* eBPF program to attach */ 277 __u32 attach_type; 278 __u32 attach_flags; 279 }; 280 281 struct { /* anonymous struct used by BPF_PROG_TEST_RUN command */ 282 __u32 prog_fd; 283 __u32 retval; 284 __u32 data_size_in; 285 __u32 data_size_out; 286 __aligned_u64 data_in; 287 __aligned_u64 data_out; 288 __u32 repeat; 289 __u32 duration; 290 } test; 291 292 struct { /* anonymous struct used by BPF_*_GET_*_ID */ 293 union { 294 __u32 start_id; 295 __u32 prog_id; 296 __u32 map_id; 297 }; 298 __u32 next_id; 299 __u32 open_flags; 300 }; 301 302 struct { /* anonymous struct used by BPF_OBJ_GET_INFO_BY_FD */ 303 __u32 bpf_fd; 304 __u32 info_len; 305 __aligned_u64 info; 306 } info; 307 308 struct { /* anonymous struct used by BPF_PROG_QUERY command */ 309 __u32 target_fd; /* container object to query */ 310 __u32 attach_type; 311 __u32 query_flags; 312 __u32 attach_flags; 313 __aligned_u64 prog_ids; 314 __u32 prog_cnt; 315 } query; 316 } __attribute__((aligned(8))); 317 318 /* BPF helper function descriptions: 319 * 320 * void *bpf_map_lookup_elem(&map, &key) 321 * Return: Map value or NULL 322 * 323 * int bpf_map_update_elem(&map, &key, &value, flags) 324 * Return: 0 on success or negative error 325 * 326 * int bpf_map_delete_elem(&map, &key) 327 * Return: 0 on success or negative error 328 * 329 * int bpf_probe_read(void *dst, int size, void *src) 330 * Return: 0 on success or negative error 331 * 332 * u64 bpf_ktime_get_ns(void) 333 * Return: current ktime 334 * 335 * int bpf_trace_printk(const char *fmt, int fmt_size, ...) 336 * Return: length of buffer written or negative error 337 * 338 * u32 bpf_prandom_u32(void) 339 * Return: random value 340 * 341 * u32 bpf_raw_smp_processor_id(void) 342 * Return: SMP processor ID 343 * 344 * int bpf_skb_store_bytes(skb, offset, from, len, flags) 345 * store bytes into packet 346 * @skb: pointer to skb 347 * @offset: offset within packet from skb->mac_header 348 * @from: pointer where to copy bytes from 349 * @len: number of bytes to store into packet 350 * @flags: bit 0 - if true, recompute skb->csum 351 * other bits - reserved 352 * Return: 0 on success or negative error 353 * 354 * int bpf_l3_csum_replace(skb, offset, from, to, flags) 355 * recompute IP checksum 356 * @skb: pointer to skb 357 * @offset: offset within packet where IP 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 * other bits - reserved 362 * Return: 0 on success or negative error 363 * 364 * int bpf_l4_csum_replace(skb, offset, from, to, flags) 365 * recompute TCP/UDP checksum 366 * @skb: pointer to skb 367 * @offset: offset within packet where TCP/UDP checksum is located 368 * @from: old value of header field 369 * @to: new value of header field 370 * @flags: bits 0-3 - size of header field 371 * bit 4 - is pseudo header 372 * other bits - reserved 373 * Return: 0 on success or negative error 374 * 375 * int bpf_tail_call(ctx, prog_array_map, index) 376 * jump into another BPF program 377 * @ctx: context pointer passed to next program 378 * @prog_array_map: pointer to map which type is BPF_MAP_TYPE_PROG_ARRAY 379 * @index: 32-bit index inside array that selects specific program to run 380 * Return: 0 on success or negative error 381 * 382 * int bpf_clone_redirect(skb, ifindex, flags) 383 * redirect to another netdev 384 * @skb: pointer to skb 385 * @ifindex: ifindex of the net device 386 * @flags: bit 0 - if set, redirect to ingress instead of egress 387 * other bits - reserved 388 * Return: 0 on success or negative error 389 * 390 * u64 bpf_get_current_pid_tgid(void) 391 * Return: current->tgid << 32 | current->pid 392 * 393 * u64 bpf_get_current_uid_gid(void) 394 * Return: current_gid << 32 | current_uid 395 * 396 * int bpf_get_current_comm(char *buf, int size_of_buf) 397 * stores current->comm into buf 398 * Return: 0 on success or negative error 399 * 400 * u32 bpf_get_cgroup_classid(skb) 401 * retrieve a proc's classid 402 * @skb: pointer to skb 403 * Return: classid if != 0 404 * 405 * int bpf_skb_vlan_push(skb, vlan_proto, vlan_tci) 406 * Return: 0 on success or negative error 407 * 408 * int bpf_skb_vlan_pop(skb) 409 * Return: 0 on success or negative error 410 * 411 * int bpf_skb_get_tunnel_key(skb, key, size, flags) 412 * int bpf_skb_set_tunnel_key(skb, key, size, flags) 413 * retrieve or populate tunnel metadata 414 * @skb: pointer to skb 415 * @key: pointer to 'struct bpf_tunnel_key' 416 * @size: size of 'struct bpf_tunnel_key' 417 * @flags: room for future extensions 418 * Return: 0 on success or negative error 419 * 420 * u64 bpf_perf_event_read(map, flags) 421 * read perf event counter value 422 * @map: pointer to perf_event_array map 423 * @flags: index of event in the map or bitmask flags 424 * Return: value of perf event counter read or error code 425 * 426 * int bpf_redirect(ifindex, flags) 427 * redirect to another netdev 428 * @ifindex: ifindex of the net device 429 * @flags: 430 * cls_bpf: 431 * bit 0 - if set, redirect to ingress instead of egress 432 * other bits - reserved 433 * xdp_bpf: 434 * all bits - reserved 435 * Return: cls_bpf: TC_ACT_REDIRECT on success or TC_ACT_SHOT on error 436 * xdp_bfp: XDP_REDIRECT on success or XDP_ABORT on error 437 * int bpf_redirect_map(map, key, flags) 438 * redirect to endpoint in map 439 * @map: pointer to dev map 440 * @key: index in map to lookup 441 * @flags: -- 442 * Return: XDP_REDIRECT on success or XDP_ABORT on error 443 * 444 * u32 bpf_get_route_realm(skb) 445 * retrieve a dst's tclassid 446 * @skb: pointer to skb 447 * Return: realm if != 0 448 * 449 * int bpf_perf_event_output(ctx, map, flags, data, size) 450 * output perf raw sample 451 * @ctx: struct pt_regs* 452 * @map: pointer to perf_event_array map 453 * @flags: index of event in the map or bitmask flags 454 * @data: data on stack to be output as raw data 455 * @size: size of data 456 * Return: 0 on success or negative error 457 * 458 * int bpf_get_stackid(ctx, map, flags) 459 * walk user or kernel stack and return id 460 * @ctx: struct pt_regs* 461 * @map: pointer to stack_trace map 462 * @flags: bits 0-7 - numer of stack frames to skip 463 * bit 8 - collect user stack instead of kernel 464 * bit 9 - compare stacks by hash only 465 * bit 10 - if two different stacks hash into the same stackid 466 * discard old 467 * other bits - reserved 468 * Return: >= 0 stackid on success or negative error 469 * 470 * s64 bpf_csum_diff(from, from_size, to, to_size, seed) 471 * calculate csum diff 472 * @from: raw from buffer 473 * @from_size: length of from buffer 474 * @to: raw to buffer 475 * @to_size: length of to buffer 476 * @seed: optional seed 477 * Return: csum result or negative error code 478 * 479 * int bpf_skb_get_tunnel_opt(skb, opt, size) 480 * retrieve tunnel options metadata 481 * @skb: pointer to skb 482 * @opt: pointer to raw tunnel option data 483 * @size: size of @opt 484 * Return: option size 485 * 486 * int bpf_skb_set_tunnel_opt(skb, opt, size) 487 * populate tunnel options metadata 488 * @skb: pointer to skb 489 * @opt: pointer to raw tunnel option data 490 * @size: size of @opt 491 * Return: 0 on success or negative error 492 * 493 * int bpf_skb_change_proto(skb, proto, flags) 494 * Change protocol of the skb. Currently supported is v4 -> v6, 495 * v6 -> v4 transitions. The helper will also resize the skb. eBPF 496 * program is expected to fill the new headers via skb_store_bytes 497 * and lX_csum_replace. 498 * @skb: pointer to skb 499 * @proto: new skb->protocol type 500 * @flags: reserved 501 * Return: 0 on success or negative error 502 * 503 * int bpf_skb_change_type(skb, type) 504 * Change packet type of skb. 505 * @skb: pointer to skb 506 * @type: new skb->pkt_type type 507 * Return: 0 on success or negative error 508 * 509 * int bpf_skb_under_cgroup(skb, map, index) 510 * Check cgroup2 membership of skb 511 * @skb: pointer to skb 512 * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type 513 * @index: index of the cgroup in the bpf_map 514 * Return: 515 * == 0 skb failed the cgroup2 descendant test 516 * == 1 skb succeeded the cgroup2 descendant test 517 * < 0 error 518 * 519 * u32 bpf_get_hash_recalc(skb) 520 * Retrieve and possibly recalculate skb->hash. 521 * @skb: pointer to skb 522 * Return: hash 523 * 524 * u64 bpf_get_current_task(void) 525 * Returns current task_struct 526 * Return: current 527 * 528 * int bpf_probe_write_user(void *dst, void *src, int len) 529 * safely attempt to write to a location 530 * @dst: destination address in userspace 531 * @src: source address on stack 532 * @len: number of bytes to copy 533 * Return: 0 on success or negative error 534 * 535 * int bpf_current_task_under_cgroup(map, index) 536 * Check cgroup2 membership of current task 537 * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type 538 * @index: index of the cgroup in the bpf_map 539 * Return: 540 * == 0 current failed the cgroup2 descendant test 541 * == 1 current succeeded the cgroup2 descendant test 542 * < 0 error 543 * 544 * int bpf_skb_change_tail(skb, len, flags) 545 * The helper will resize the skb to the given new size, to be used f.e. 546 * with control messages. 547 * @skb: pointer to skb 548 * @len: new skb length 549 * @flags: reserved 550 * Return: 0 on success or negative error 551 * 552 * int bpf_skb_pull_data(skb, len) 553 * The helper will pull in non-linear data in case the skb is non-linear 554 * and not all of len are part of the linear section. Only needed for 555 * read/write with direct packet access. 556 * @skb: pointer to skb 557 * @len: len to make read/writeable 558 * Return: 0 on success or negative error 559 * 560 * s64 bpf_csum_update(skb, csum) 561 * Adds csum into skb->csum in case of CHECKSUM_COMPLETE. 562 * @skb: pointer to skb 563 * @csum: csum to add 564 * Return: csum on success or negative error 565 * 566 * void bpf_set_hash_invalid(skb) 567 * Invalidate current skb->hash. 568 * @skb: pointer to skb 569 * 570 * int bpf_get_numa_node_id() 571 * Return: Id of current NUMA node. 572 * 573 * int bpf_skb_change_head() 574 * Grows headroom of skb and adjusts MAC header offset accordingly. 575 * Will extends/reallocae as required automatically. 576 * May change skb data pointer and will thus invalidate any check 577 * performed for direct packet access. 578 * @skb: pointer to skb 579 * @len: length of header to be pushed in front 580 * @flags: Flags (unused for now) 581 * Return: 0 on success or negative error 582 * 583 * int bpf_xdp_adjust_head(xdp_md, delta) 584 * Adjust the xdp_md.data by delta 585 * @xdp_md: pointer to xdp_md 586 * @delta: An positive/negative integer to be added to xdp_md.data 587 * Return: 0 on success or negative on error 588 * 589 * int bpf_probe_read_str(void *dst, int size, const void *unsafe_ptr) 590 * Copy a NUL terminated string from unsafe address. In case the string 591 * length is smaller than size, the target is not padded with further NUL 592 * bytes. In case the string length is larger than size, just count-1 593 * bytes are copied and the last byte is set to NUL. 594 * @dst: destination address 595 * @size: maximum number of bytes to copy, including the trailing NUL 596 * @unsafe_ptr: unsafe address 597 * Return: 598 * > 0 length of the string including the trailing NUL on success 599 * < 0 error 600 * 601 * u64 bpf_get_socket_cookie(skb) 602 * Get the cookie for the socket stored inside sk_buff. 603 * @skb: pointer to skb 604 * Return: 8 Bytes non-decreasing number on success or 0 if the socket 605 * field is missing inside sk_buff 606 * 607 * u32 bpf_get_socket_uid(skb) 608 * Get the owner uid of the socket stored inside sk_buff. 609 * @skb: pointer to skb 610 * Return: uid of the socket owner on success or overflowuid if failed. 611 * 612 * u32 bpf_set_hash(skb, hash) 613 * Set full skb->hash. 614 * @skb: pointer to skb 615 * @hash: hash to set 616 * 617 * int bpf_setsockopt(bpf_socket, level, optname, optval, optlen) 618 * Calls setsockopt. Not all opts are available, only those with 619 * integer optvals plus TCP_CONGESTION. 620 * Supported levels: SOL_SOCKET and IPPROTO_TCP 621 * @bpf_socket: pointer to bpf_socket 622 * @level: SOL_SOCKET or IPPROTO_TCP 623 * @optname: option name 624 * @optval: pointer to option value 625 * @optlen: length of optval in bytes 626 * Return: 0 or negative error 627 * 628 * int bpf_getsockopt(bpf_socket, level, optname, optval, optlen) 629 * Calls getsockopt. Not all opts are available. 630 * Supported levels: IPPROTO_TCP 631 * @bpf_socket: pointer to bpf_socket 632 * @level: IPPROTO_TCP 633 * @optname: option name 634 * @optval: pointer to option value 635 * @optlen: length of optval in bytes 636 * Return: 0 or negative error 637 * 638 * int bpf_skb_adjust_room(skb, len_diff, mode, flags) 639 * Grow or shrink room in sk_buff. 640 * @skb: pointer to skb 641 * @len_diff: (signed) amount of room to grow/shrink 642 * @mode: operation mode (enum bpf_adj_room_mode) 643 * @flags: reserved for future use 644 * Return: 0 on success or negative error code 645 * 646 * int bpf_sk_redirect_map(map, key, flags) 647 * Redirect skb to a sock in map using key as a lookup key for the 648 * sock in map. 649 * @map: pointer to sockmap 650 * @key: key to lookup sock in map 651 * @flags: reserved for future use 652 * Return: SK_PASS 653 * 654 * int bpf_sock_map_update(skops, map, key, flags) 655 * @skops: pointer to bpf_sock_ops 656 * @map: pointer to sockmap to update 657 * @key: key to insert/update sock in map 658 * @flags: same flags as map update elem 659 * 660 * int bpf_xdp_adjust_meta(xdp_md, delta) 661 * Adjust the xdp_md.data_meta by delta 662 * @xdp_md: pointer to xdp_md 663 * @delta: An positive/negative integer to be added to xdp_md.data_meta 664 * Return: 0 on success or negative on error 665 * 666 * int bpf_perf_event_read_value(map, flags, buf, buf_size) 667 * read perf event counter value and perf event enabled/running time 668 * @map: pointer to perf_event_array map 669 * @flags: index of event in the map or bitmask flags 670 * @buf: buf to fill 671 * @buf_size: size of the buf 672 * Return: 0 on success or negative error code 673 * 674 * int bpf_perf_prog_read_value(ctx, buf, buf_size) 675 * read perf prog attached perf event counter and enabled/running time 676 * @ctx: pointer to ctx 677 * @buf: buf to fill 678 * @buf_size: size of the buf 679 * Return : 0 on success or negative error code 680 */ 681 #define __BPF_FUNC_MAPPER(FN) \ 682 FN(unspec), \ 683 FN(map_lookup_elem), \ 684 FN(map_update_elem), \ 685 FN(map_delete_elem), \ 686 FN(probe_read), \ 687 FN(ktime_get_ns), \ 688 FN(trace_printk), \ 689 FN(get_prandom_u32), \ 690 FN(get_smp_processor_id), \ 691 FN(skb_store_bytes), \ 692 FN(l3_csum_replace), \ 693 FN(l4_csum_replace), \ 694 FN(tail_call), \ 695 FN(clone_redirect), \ 696 FN(get_current_pid_tgid), \ 697 FN(get_current_uid_gid), \ 698 FN(get_current_comm), \ 699 FN(get_cgroup_classid), \ 700 FN(skb_vlan_push), \ 701 FN(skb_vlan_pop), \ 702 FN(skb_get_tunnel_key), \ 703 FN(skb_set_tunnel_key), \ 704 FN(perf_event_read), \ 705 FN(redirect), \ 706 FN(get_route_realm), \ 707 FN(perf_event_output), \ 708 FN(skb_load_bytes), \ 709 FN(get_stackid), \ 710 FN(csum_diff), \ 711 FN(skb_get_tunnel_opt), \ 712 FN(skb_set_tunnel_opt), \ 713 FN(skb_change_proto), \ 714 FN(skb_change_type), \ 715 FN(skb_under_cgroup), \ 716 FN(get_hash_recalc), \ 717 FN(get_current_task), \ 718 FN(probe_write_user), \ 719 FN(current_task_under_cgroup), \ 720 FN(skb_change_tail), \ 721 FN(skb_pull_data), \ 722 FN(csum_update), \ 723 FN(set_hash_invalid), \ 724 FN(get_numa_node_id), \ 725 FN(skb_change_head), \ 726 FN(xdp_adjust_head), \ 727 FN(probe_read_str), \ 728 FN(get_socket_cookie), \ 729 FN(get_socket_uid), \ 730 FN(set_hash), \ 731 FN(setsockopt), \ 732 FN(skb_adjust_room), \ 733 FN(redirect_map), \ 734 FN(sk_redirect_map), \ 735 FN(sock_map_update), \ 736 FN(xdp_adjust_meta), \ 737 FN(perf_event_read_value), \ 738 FN(perf_prog_read_value), \ 739 FN(getsockopt), 740 741 /* integer value in 'imm' field of BPF_CALL instruction selects which helper 742 * function eBPF program intends to call 743 */ 744 #define __BPF_ENUM_FN(x) BPF_FUNC_ ## x 745 enum bpf_func_id { 746 __BPF_FUNC_MAPPER(__BPF_ENUM_FN) 747 __BPF_FUNC_MAX_ID, 748 }; 749 #undef __BPF_ENUM_FN 750 751 /* All flags used by eBPF helper functions, placed here. */ 752 753 /* BPF_FUNC_skb_store_bytes flags. */ 754 #define BPF_F_RECOMPUTE_CSUM (1ULL << 0) 755 #define BPF_F_INVALIDATE_HASH (1ULL << 1) 756 757 /* BPF_FUNC_l3_csum_replace and BPF_FUNC_l4_csum_replace flags. 758 * First 4 bits are for passing the header field size. 759 */ 760 #define BPF_F_HDR_FIELD_MASK 0xfULL 761 762 /* BPF_FUNC_l4_csum_replace flags. */ 763 #define BPF_F_PSEUDO_HDR (1ULL << 4) 764 #define BPF_F_MARK_MANGLED_0 (1ULL << 5) 765 #define BPF_F_MARK_ENFORCE (1ULL << 6) 766 767 /* BPF_FUNC_clone_redirect and BPF_FUNC_redirect flags. */ 768 #define BPF_F_INGRESS (1ULL << 0) 769 770 /* BPF_FUNC_skb_set_tunnel_key and BPF_FUNC_skb_get_tunnel_key flags. */ 771 #define BPF_F_TUNINFO_IPV6 (1ULL << 0) 772 773 /* BPF_FUNC_get_stackid flags. */ 774 #define BPF_F_SKIP_FIELD_MASK 0xffULL 775 #define BPF_F_USER_STACK (1ULL << 8) 776 #define BPF_F_FAST_STACK_CMP (1ULL << 9) 777 #define BPF_F_REUSE_STACKID (1ULL << 10) 778 779 /* BPF_FUNC_skb_set_tunnel_key flags. */ 780 #define BPF_F_ZERO_CSUM_TX (1ULL << 1) 781 #define BPF_F_DONT_FRAGMENT (1ULL << 2) 782 783 /* BPF_FUNC_perf_event_output, BPF_FUNC_perf_event_read and 784 * BPF_FUNC_perf_event_read_value flags. 785 */ 786 #define BPF_F_INDEX_MASK 0xffffffffULL 787 #define BPF_F_CURRENT_CPU BPF_F_INDEX_MASK 788 /* BPF_FUNC_perf_event_output for sk_buff input context. */ 789 #define BPF_F_CTXLEN_MASK (0xfffffULL << 32) 790 791 /* Mode for BPF_FUNC_skb_adjust_room helper. */ 792 enum bpf_adj_room_mode { 793 BPF_ADJ_ROOM_NET, 794 }; 795 796 /* user accessible mirror of in-kernel sk_buff. 797 * new fields can only be added to the end of this structure 798 */ 799 struct __sk_buff { 800 __u32 len; 801 __u32 pkt_type; 802 __u32 mark; 803 __u32 queue_mapping; 804 __u32 protocol; 805 __u32 vlan_present; 806 __u32 vlan_tci; 807 __u32 vlan_proto; 808 __u32 priority; 809 __u32 ingress_ifindex; 810 __u32 ifindex; 811 __u32 tc_index; 812 __u32 cb[5]; 813 __u32 hash; 814 __u32 tc_classid; 815 __u32 data; 816 __u32 data_end; 817 __u32 napi_id; 818 819 /* Accessed by BPF_PROG_TYPE_sk_skb types from here to ... */ 820 __u32 family; 821 __u32 remote_ip4; /* Stored in network byte order */ 822 __u32 local_ip4; /* Stored in network byte order */ 823 __u32 remote_ip6[4]; /* Stored in network byte order */ 824 __u32 local_ip6[4]; /* Stored in network byte order */ 825 __u32 remote_port; /* Stored in network byte order */ 826 __u32 local_port; /* stored in host byte order */ 827 /* ... here. */ 828 829 __u32 data_meta; 830 }; 831 832 struct bpf_tunnel_key { 833 __u32 tunnel_id; 834 union { 835 __u32 remote_ipv4; 836 __u32 remote_ipv6[4]; 837 }; 838 __u8 tunnel_tos; 839 __u8 tunnel_ttl; 840 __u16 tunnel_ext; 841 __u32 tunnel_label; 842 }; 843 844 /* Generic BPF return codes which all BPF program types may support. 845 * The values are binary compatible with their TC_ACT_* counter-part to 846 * provide backwards compatibility with existing SCHED_CLS and SCHED_ACT 847 * programs. 848 * 849 * XDP is handled seprately, see XDP_*. 850 */ 851 enum bpf_ret_code { 852 BPF_OK = 0, 853 /* 1 reserved */ 854 BPF_DROP = 2, 855 /* 3-6 reserved */ 856 BPF_REDIRECT = 7, 857 /* >127 are reserved for prog type specific return codes */ 858 }; 859 860 struct bpf_sock { 861 __u32 bound_dev_if; 862 __u32 family; 863 __u32 type; 864 __u32 protocol; 865 __u32 mark; 866 __u32 priority; 867 }; 868 869 #define XDP_PACKET_HEADROOM 256 870 871 /* User return codes for XDP prog type. 872 * A valid XDP program must return one of these defined values. All other 873 * return codes are reserved for future use. Unknown return codes will 874 * result in packet drops and a warning via bpf_warn_invalid_xdp_action(). 875 */ 876 enum xdp_action { 877 XDP_ABORTED = 0, 878 XDP_DROP, 879 XDP_PASS, 880 XDP_TX, 881 XDP_REDIRECT, 882 }; 883 884 /* user accessible metadata for XDP packet hook 885 * new fields must be added to the end of this structure 886 */ 887 struct xdp_md { 888 __u32 data; 889 __u32 data_end; 890 __u32 data_meta; 891 }; 892 893 enum sk_action { 894 SK_DROP = 0, 895 SK_PASS, 896 }; 897 898 #define BPF_TAG_SIZE 8 899 900 struct bpf_prog_info { 901 __u32 type; 902 __u32 id; 903 __u8 tag[BPF_TAG_SIZE]; 904 __u32 jited_prog_len; 905 __u32 xlated_prog_len; 906 __aligned_u64 jited_prog_insns; 907 __aligned_u64 xlated_prog_insns; 908 __u64 load_time; /* ns since boottime */ 909 __u32 created_by_uid; 910 __u32 nr_map_ids; 911 __aligned_u64 map_ids; 912 char name[BPF_OBJ_NAME_LEN]; 913 } __attribute__((aligned(8))); 914 915 struct bpf_map_info { 916 __u32 type; 917 __u32 id; 918 __u32 key_size; 919 __u32 value_size; 920 __u32 max_entries; 921 __u32 map_flags; 922 char name[BPF_OBJ_NAME_LEN]; 923 } __attribute__((aligned(8))); 924 925 /* User bpf_sock_ops struct to access socket values and specify request ops 926 * and their replies. 927 * Some of this fields are in network (bigendian) byte order and may need 928 * to be converted before use (bpf_ntohl() defined in samples/bpf/bpf_endian.h). 929 * New fields can only be added at the end of this structure 930 */ 931 struct bpf_sock_ops { 932 __u32 op; 933 union { 934 __u32 reply; 935 __u32 replylong[4]; 936 }; 937 __u32 family; 938 __u32 remote_ip4; /* Stored in network byte order */ 939 __u32 local_ip4; /* Stored in network byte order */ 940 __u32 remote_ip6[4]; /* Stored in network byte order */ 941 __u32 local_ip6[4]; /* Stored in network byte order */ 942 __u32 remote_port; /* Stored in network byte order */ 943 __u32 local_port; /* stored in host byte order */ 944 }; 945 946 /* List of known BPF sock_ops operators. 947 * New entries can only be added at the end 948 */ 949 enum { 950 BPF_SOCK_OPS_VOID, 951 BPF_SOCK_OPS_TIMEOUT_INIT, /* Should return SYN-RTO value to use or 952 * -1 if default value should be used 953 */ 954 BPF_SOCK_OPS_RWND_INIT, /* Should return initial advertized 955 * window (in packets) or -1 if default 956 * value should be used 957 */ 958 BPF_SOCK_OPS_TCP_CONNECT_CB, /* Calls BPF program right before an 959 * active connection is initialized 960 */ 961 BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB, /* Calls BPF program when an 962 * active connection is 963 * established 964 */ 965 BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB, /* Calls BPF program when a 966 * passive connection is 967 * established 968 */ 969 BPF_SOCK_OPS_NEEDS_ECN, /* If connection's congestion control 970 * needs ECN 971 */ 972 BPF_SOCK_OPS_BASE_RTT, /* Get base RTT. The correct value is 973 * based on the path and may be 974 * dependent on the congestion control 975 * algorithm. In general it indicates 976 * a congestion threshold. RTTs above 977 * this indicate congestion 978 */ 979 }; 980 981 #define TCP_BPF_IW 1001 /* Set TCP initial congestion window */ 982 #define TCP_BPF_SNDCWND_CLAMP 1002 /* Set sndcwnd_clamp */ 983 984 struct bpf_perf_event_value { 985 __u64 counter; 986 __u64 enabled; 987 __u64 running; 988 }; 989 990 #define BPF_DEVCG_ACC_MKNOD (1ULL << 0) 991 #define BPF_DEVCG_ACC_READ (1ULL << 1) 992 #define BPF_DEVCG_ACC_WRITE (1ULL << 2) 993 994 #define BPF_DEVCG_DEV_BLOCK (1ULL << 0) 995 #define BPF_DEVCG_DEV_CHAR (1ULL << 1) 996 997 struct bpf_cgroup_dev_ctx { 998 __u32 access_type; /* (access << 16) | type */ 999 __u32 major; 1000 __u32 minor; 1001 }; 1002 1003 #endif /* _UAPI__LINUX_BPF_H__ */ 1004