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 33 /* jmp encodings */ 34 #define BPF_JNE 0x50 /* jump != */ 35 #define BPF_JLT 0xa0 /* LT is unsigned, '<' */ 36 #define BPF_JLE 0xb0 /* LE is unsigned, '<=' */ 37 #define BPF_JSGT 0x60 /* SGT is signed '>', GT in x86 */ 38 #define BPF_JSGE 0x70 /* SGE is signed '>=', GE in x86 */ 39 #define BPF_JSLT 0xc0 /* SLT is signed, '<' */ 40 #define BPF_JSLE 0xd0 /* SLE is signed, '<=' */ 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 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 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, 87 BPF_PROG_ATTACH, 88 BPF_PROG_DETACH, 89 BPF_PROG_TEST_RUN, 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, 95 BPF_PROG_QUERY, 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, 107 BPF_MAP_TYPE_CGROUP_ARRAY, 108 BPF_MAP_TYPE_LRU_HASH, 109 BPF_MAP_TYPE_LRU_PERCPU_HASH, 110 BPF_MAP_TYPE_LPM_TRIE, 111 BPF_MAP_TYPE_ARRAY_OF_MAPS, 112 BPF_MAP_TYPE_HASH_OF_MAPS, 113 BPF_MAP_TYPE_DEVMAP, 114 BPF_MAP_TYPE_SOCKMAP, 115 BPF_MAP_TYPE_CPUMAP, 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, 125 BPF_PROG_TYPE_XDP, 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, 132 BPF_PROG_TYPE_SOCK_OPS, 133 BPF_PROG_TYPE_SK_SKB, 134 BPF_PROG_TYPE_CGROUP_DEVICE, 135 }; 136 137 enum bpf_attach_type { 138 BPF_CGROUP_INET_INGRESS, 139 BPF_CGROUP_INET_EGRESS, 140 BPF_CGROUP_INET_SOCK_CREATE, 141 BPF_CGROUP_SOCK_OPS, 142 BPF_SK_SKB_STREAM_PARSER, 143 BPF_SK_SKB_STREAM_VERDICT, 144 BPF_CGROUP_DEVICE, 145 __MAX_BPF_ATTACH_TYPE 146 }; 147 148 #define MAX_BPF_ATTACH_TYPE __MAX_BPF_ATTACH_TYPE 149 150 /* cgroup-bpf attach flags used in BPF_PROG_ATTACH command 151 * 152 * NONE(default): No further bpf programs allowed in the subtree. 153 * 154 * BPF_F_ALLOW_OVERRIDE: If a sub-cgroup installs some bpf program, 155 * the program in this cgroup yields to sub-cgroup program. 156 * 157 * BPF_F_ALLOW_MULTI: If a sub-cgroup installs some bpf program, 158 * that cgroup program gets run in addition to the program in this cgroup. 159 * 160 * Only one program is allowed to be attached to a cgroup with 161 * NONE or BPF_F_ALLOW_OVERRIDE flag. 162 * Attaching another program on top of NONE or BPF_F_ALLOW_OVERRIDE will 163 * release old program and attach the new one. Attach flags has to match. 164 * 165 * Multiple programs are allowed to be attached to a cgroup with 166 * BPF_F_ALLOW_MULTI flag. They are executed in FIFO order 167 * (those that were attached first, run first) 168 * The programs of sub-cgroup are executed first, then programs of 169 * this cgroup and then programs of parent cgroup. 170 * When children program makes decision (like picking TCP CA or sock bind) 171 * parent program has a chance to override it. 172 * 173 * A cgroup with MULTI or OVERRIDE flag allows any attach flags in sub-cgroups. 174 * A cgroup with NONE doesn't allow any programs in sub-cgroups. 175 * Ex1: 176 * cgrp1 (MULTI progs A, B) -> 177 * cgrp2 (OVERRIDE prog C) -> 178 * cgrp3 (MULTI prog D) -> 179 * cgrp4 (OVERRIDE prog E) -> 180 * cgrp5 (NONE prog F) 181 * the event in cgrp5 triggers execution of F,D,A,B in that order. 182 * if prog F is detached, the execution is E,D,A,B 183 * if prog F and D are detached, the execution is E,A,B 184 * if prog F, E and D are detached, the execution is C,A,B 185 * 186 * All eligible programs are executed regardless of return code from 187 * earlier programs. 188 */ 189 #define BPF_F_ALLOW_OVERRIDE (1U << 0) 190 #define BPF_F_ALLOW_MULTI (1U << 1) 191 192 /* If BPF_F_STRICT_ALIGNMENT is used in BPF_PROG_LOAD command, the 193 * verifier will perform strict alignment checking as if the kernel 194 * has been built with CONFIG_EFFICIENT_UNALIGNED_ACCESS not set, 195 * and NET_IP_ALIGN defined to 2. 196 */ 197 #define BPF_F_STRICT_ALIGNMENT (1U << 0) 198 199 #define BPF_PSEUDO_MAP_FD 1 200 201 /* flags for BPF_MAP_UPDATE_ELEM command */ 202 #define BPF_ANY 0 /* create new element or update existing */ 203 #define BPF_NOEXIST 1 /* create new element if it didn't exist */ 204 #define BPF_EXIST 2 /* update existing element */ 205 206 /* flags for BPF_MAP_CREATE command */ 207 #define BPF_F_NO_PREALLOC (1U << 0) 208 /* Instead of having one common LRU list in the 209 * BPF_MAP_TYPE_LRU_[PERCPU_]HASH map, use a percpu LRU list 210 * which can scale and perform better. 211 * Note, the LRU nodes (including free nodes) cannot be moved 212 * across different LRU lists. 213 */ 214 #define BPF_F_NO_COMMON_LRU (1U << 1) 215 /* Specify numa node during map creation */ 216 #define BPF_F_NUMA_NODE (1U << 2) 217 218 /* flags for BPF_PROG_QUERY */ 219 #define BPF_F_QUERY_EFFECTIVE (1U << 0) 220 221 #define BPF_OBJ_NAME_LEN 16U 222 223 /* Flags for accessing BPF object */ 224 #define BPF_F_RDONLY (1U << 3) 225 #define BPF_F_WRONLY (1U << 4) 226 227 union bpf_attr { 228 struct { /* anonymous struct used by BPF_MAP_CREATE command */ 229 __u32 map_type; /* one of enum bpf_map_type */ 230 __u32 key_size; /* size of key in bytes */ 231 __u32 value_size; /* size of value in bytes */ 232 __u32 max_entries; /* max number of entries in a map */ 233 __u32 map_flags; /* BPF_MAP_CREATE related 234 * flags defined above. 235 */ 236 __u32 inner_map_fd; /* fd pointing to the inner map */ 237 __u32 numa_node; /* numa node (effective only if 238 * BPF_F_NUMA_NODE is set). 239 */ 240 char map_name[BPF_OBJ_NAME_LEN]; 241 }; 242 243 struct { /* anonymous struct used by BPF_MAP_*_ELEM commands */ 244 __u32 map_fd; 245 __aligned_u64 key; 246 union { 247 __aligned_u64 value; 248 __aligned_u64 next_key; 249 }; 250 __u64 flags; 251 }; 252 253 struct { /* anonymous struct used by BPF_PROG_LOAD command */ 254 __u32 prog_type; /* one of enum bpf_prog_type */ 255 __u32 insn_cnt; 256 __aligned_u64 insns; 257 __aligned_u64 license; 258 __u32 log_level; /* verbosity level of verifier */ 259 __u32 log_size; /* size of user buffer */ 260 __aligned_u64 log_buf; /* user supplied buffer */ 261 __u32 kern_version; /* checked when prog_type=kprobe */ 262 __u32 prog_flags; 263 char prog_name[BPF_OBJ_NAME_LEN]; 264 __u32 prog_target_ifindex; /* ifindex of netdev to prep for */ 265 }; 266 267 struct { /* anonymous struct used by BPF_OBJ_* commands */ 268 __aligned_u64 pathname; 269 __u32 bpf_fd; 270 __u32 file_flags; 271 }; 272 273 struct { /* anonymous struct used by BPF_PROG_ATTACH/DETACH commands */ 274 __u32 target_fd; /* container object to attach to */ 275 __u32 attach_bpf_fd; /* eBPF program to attach */ 276 __u32 attach_type; 277 __u32 attach_flags; 278 }; 279 280 struct { /* anonymous struct used by BPF_PROG_TEST_RUN command */ 281 __u32 prog_fd; 282 __u32 retval; 283 __u32 data_size_in; 284 __u32 data_size_out; 285 __aligned_u64 data_in; 286 __aligned_u64 data_out; 287 __u32 repeat; 288 __u32 duration; 289 } test; 290 291 struct { /* anonymous struct used by BPF_*_GET_*_ID */ 292 union { 293 __u32 start_id; 294 __u32 prog_id; 295 __u32 map_id; 296 }; 297 __u32 next_id; 298 __u32 open_flags; 299 }; 300 301 struct { /* anonymous struct used by BPF_OBJ_GET_INFO_BY_FD */ 302 __u32 bpf_fd; 303 __u32 info_len; 304 __aligned_u64 info; 305 } info; 306 307 struct { /* anonymous struct used by BPF_PROG_QUERY command */ 308 __u32 target_fd; /* container object to query */ 309 __u32 attach_type; 310 __u32 query_flags; 311 __u32 attach_flags; 312 __aligned_u64 prog_ids; 313 __u32 prog_cnt; 314 } query; 315 } __attribute__((aligned(8))); 316 317 /* BPF helper function descriptions: 318 * 319 * void *bpf_map_lookup_elem(&map, &key) 320 * Return: Map value or NULL 321 * 322 * int bpf_map_update_elem(&map, &key, &value, flags) 323 * Return: 0 on success or negative error 324 * 325 * int bpf_map_delete_elem(&map, &key) 326 * Return: 0 on success or negative error 327 * 328 * int bpf_probe_read(void *dst, int size, void *src) 329 * Return: 0 on success or negative error 330 * 331 * u64 bpf_ktime_get_ns(void) 332 * Return: current ktime 333 * 334 * int bpf_trace_printk(const char *fmt, int fmt_size, ...) 335 * Return: length of buffer written or negative error 336 * 337 * u32 bpf_prandom_u32(void) 338 * Return: random value 339 * 340 * u32 bpf_raw_smp_processor_id(void) 341 * Return: SMP processor ID 342 * 343 * int bpf_skb_store_bytes(skb, offset, from, len, flags) 344 * store bytes into packet 345 * @skb: pointer to skb 346 * @offset: offset within packet from skb->mac_header 347 * @from: pointer where to copy bytes from 348 * @len: number of bytes to store into packet 349 * @flags: bit 0 - if true, recompute skb->csum 350 * other bits - reserved 351 * Return: 0 on success or negative error 352 * 353 * int bpf_l3_csum_replace(skb, offset, from, to, flags) 354 * recompute IP checksum 355 * @skb: pointer to skb 356 * @offset: offset within packet where IP checksum is located 357 * @from: old value of header field 358 * @to: new value of header field 359 * @flags: bits 0-3 - size of header field 360 * other bits - reserved 361 * Return: 0 on success or negative error 362 * 363 * int bpf_l4_csum_replace(skb, offset, from, to, flags) 364 * recompute TCP/UDP checksum 365 * @skb: pointer to skb 366 * @offset: offset within packet where TCP/UDP checksum is located 367 * @from: old value of header field 368 * @to: new value of header field 369 * @flags: bits 0-3 - size of header field 370 * bit 4 - is pseudo header 371 * other bits - reserved 372 * Return: 0 on success or negative error 373 * 374 * int bpf_tail_call(ctx, prog_array_map, index) 375 * jump into another BPF program 376 * @ctx: context pointer passed to next program 377 * @prog_array_map: pointer to map which type is BPF_MAP_TYPE_PROG_ARRAY 378 * @index: 32-bit index inside array that selects specific program to run 379 * Return: 0 on success or negative error 380 * 381 * int bpf_clone_redirect(skb, ifindex, flags) 382 * redirect to another netdev 383 * @skb: pointer to skb 384 * @ifindex: ifindex of the net device 385 * @flags: bit 0 - if set, redirect to ingress instead of egress 386 * other bits - reserved 387 * Return: 0 on success or negative error 388 * 389 * u64 bpf_get_current_pid_tgid(void) 390 * Return: current->tgid << 32 | current->pid 391 * 392 * u64 bpf_get_current_uid_gid(void) 393 * Return: current_gid << 32 | current_uid 394 * 395 * int bpf_get_current_comm(char *buf, int size_of_buf) 396 * stores current->comm into buf 397 * Return: 0 on success or negative error 398 * 399 * u32 bpf_get_cgroup_classid(skb) 400 * retrieve a proc's classid 401 * @skb: pointer to skb 402 * Return: classid if != 0 403 * 404 * int bpf_skb_vlan_push(skb, vlan_proto, vlan_tci) 405 * Return: 0 on success or negative error 406 * 407 * int bpf_skb_vlan_pop(skb) 408 * Return: 0 on success or negative error 409 * 410 * int bpf_skb_get_tunnel_key(skb, key, size, flags) 411 * int bpf_skb_set_tunnel_key(skb, key, size, flags) 412 * retrieve or populate tunnel metadata 413 * @skb: pointer to skb 414 * @key: pointer to 'struct bpf_tunnel_key' 415 * @size: size of 'struct bpf_tunnel_key' 416 * @flags: room for future extensions 417 * Return: 0 on success or negative error 418 * 419 * u64 bpf_perf_event_read(map, flags) 420 * read perf event counter value 421 * @map: pointer to perf_event_array map 422 * @flags: index of event in the map or bitmask flags 423 * Return: value of perf event counter read or error code 424 * 425 * int bpf_redirect(ifindex, flags) 426 * redirect to another netdev 427 * @ifindex: ifindex of the net device 428 * @flags: 429 * cls_bpf: 430 * bit 0 - if set, redirect to ingress instead of egress 431 * other bits - reserved 432 * xdp_bpf: 433 * all bits - reserved 434 * Return: cls_bpf: TC_ACT_REDIRECT on success or TC_ACT_SHOT on error 435 * xdp_bfp: XDP_REDIRECT on success or XDP_ABORT on error 436 * int bpf_redirect_map(map, key, flags) 437 * redirect to endpoint in map 438 * @map: pointer to dev map 439 * @key: index in map to lookup 440 * @flags: -- 441 * Return: XDP_REDIRECT on success or XDP_ABORT on error 442 * 443 * u32 bpf_get_route_realm(skb) 444 * retrieve a dst's tclassid 445 * @skb: pointer to skb 446 * Return: realm if != 0 447 * 448 * int bpf_perf_event_output(ctx, map, flags, data, size) 449 * output perf raw sample 450 * @ctx: struct pt_regs* 451 * @map: pointer to perf_event_array map 452 * @flags: index of event in the map or bitmask flags 453 * @data: data on stack to be output as raw data 454 * @size: size of data 455 * Return: 0 on success or negative error 456 * 457 * int bpf_get_stackid(ctx, map, flags) 458 * walk user or kernel stack and return id 459 * @ctx: struct pt_regs* 460 * @map: pointer to stack_trace map 461 * @flags: bits 0-7 - numer of stack frames to skip 462 * bit 8 - collect user stack instead of kernel 463 * bit 9 - compare stacks by hash only 464 * bit 10 - if two different stacks hash into the same stackid 465 * discard old 466 * other bits - reserved 467 * Return: >= 0 stackid on success or negative error 468 * 469 * s64 bpf_csum_diff(from, from_size, to, to_size, seed) 470 * calculate csum diff 471 * @from: raw from buffer 472 * @from_size: length of from buffer 473 * @to: raw to buffer 474 * @to_size: length of to buffer 475 * @seed: optional seed 476 * Return: csum result or negative error code 477 * 478 * int bpf_skb_get_tunnel_opt(skb, opt, size) 479 * retrieve tunnel options metadata 480 * @skb: pointer to skb 481 * @opt: pointer to raw tunnel option data 482 * @size: size of @opt 483 * Return: option size 484 * 485 * int bpf_skb_set_tunnel_opt(skb, opt, size) 486 * populate tunnel options metadata 487 * @skb: pointer to skb 488 * @opt: pointer to raw tunnel option data 489 * @size: size of @opt 490 * Return: 0 on success or negative error 491 * 492 * int bpf_skb_change_proto(skb, proto, flags) 493 * Change protocol of the skb. Currently supported is v4 -> v6, 494 * v6 -> v4 transitions. The helper will also resize the skb. eBPF 495 * program is expected to fill the new headers via skb_store_bytes 496 * and lX_csum_replace. 497 * @skb: pointer to skb 498 * @proto: new skb->protocol type 499 * @flags: reserved 500 * Return: 0 on success or negative error 501 * 502 * int bpf_skb_change_type(skb, type) 503 * Change packet type of skb. 504 * @skb: pointer to skb 505 * @type: new skb->pkt_type type 506 * Return: 0 on success or negative error 507 * 508 * int bpf_skb_under_cgroup(skb, map, index) 509 * Check cgroup2 membership of skb 510 * @skb: pointer to skb 511 * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type 512 * @index: index of the cgroup in the bpf_map 513 * Return: 514 * == 0 skb failed the cgroup2 descendant test 515 * == 1 skb succeeded the cgroup2 descendant test 516 * < 0 error 517 * 518 * u32 bpf_get_hash_recalc(skb) 519 * Retrieve and possibly recalculate skb->hash. 520 * @skb: pointer to skb 521 * Return: hash 522 * 523 * u64 bpf_get_current_task(void) 524 * Returns current task_struct 525 * Return: current 526 * 527 * int bpf_probe_write_user(void *dst, void *src, int len) 528 * safely attempt to write to a location 529 * @dst: destination address in userspace 530 * @src: source address on stack 531 * @len: number of bytes to copy 532 * Return: 0 on success or negative error 533 * 534 * int bpf_current_task_under_cgroup(map, index) 535 * Check cgroup2 membership of current task 536 * @map: pointer to bpf_map in BPF_MAP_TYPE_CGROUP_ARRAY type 537 * @index: index of the cgroup in the bpf_map 538 * Return: 539 * == 0 current failed the cgroup2 descendant test 540 * == 1 current succeeded the cgroup2 descendant test 541 * < 0 error 542 * 543 * int bpf_skb_change_tail(skb, len, flags) 544 * The helper will resize the skb to the given new size, to be used f.e. 545 * with control messages. 546 * @skb: pointer to skb 547 * @len: new skb length 548 * @flags: reserved 549 * Return: 0 on success or negative error 550 * 551 * int bpf_skb_pull_data(skb, len) 552 * The helper will pull in non-linear data in case the skb is non-linear 553 * and not all of len are part of the linear section. Only needed for 554 * read/write with direct packet access. 555 * @skb: pointer to skb 556 * @len: len to make read/writeable 557 * Return: 0 on success or negative error 558 * 559 * s64 bpf_csum_update(skb, csum) 560 * Adds csum into skb->csum in case of CHECKSUM_COMPLETE. 561 * @skb: pointer to skb 562 * @csum: csum to add 563 * Return: csum on success or negative error 564 * 565 * void bpf_set_hash_invalid(skb) 566 * Invalidate current skb->hash. 567 * @skb: pointer to skb 568 * 569 * int bpf_get_numa_node_id() 570 * Return: Id of current NUMA node. 571 * 572 * int bpf_skb_change_head() 573 * Grows headroom of skb and adjusts MAC header offset accordingly. 574 * Will extends/reallocae as required automatically. 575 * May change skb data pointer and will thus invalidate any check 576 * performed for direct packet access. 577 * @skb: pointer to skb 578 * @len: length of header to be pushed in front 579 * @flags: Flags (unused for now) 580 * Return: 0 on success or negative error 581 * 582 * int bpf_xdp_adjust_head(xdp_md, delta) 583 * Adjust the xdp_md.data by delta 584 * @xdp_md: pointer to xdp_md 585 * @delta: An positive/negative integer to be added to xdp_md.data 586 * Return: 0 on success or negative on error 587 * 588 * int bpf_probe_read_str(void *dst, int size, const void *unsafe_ptr) 589 * Copy a NUL terminated string from unsafe address. In case the string 590 * length is smaller than size, the target is not padded with further NUL 591 * bytes. In case the string length is larger than size, just count-1 592 * bytes are copied and the last byte is set to NUL. 593 * @dst: destination address 594 * @size: maximum number of bytes to copy, including the trailing NUL 595 * @unsafe_ptr: unsafe address 596 * Return: 597 * > 0 length of the string including the trailing NUL on success 598 * < 0 error 599 * 600 * u64 bpf_get_socket_cookie(skb) 601 * Get the cookie for the socket stored inside sk_buff. 602 * @skb: pointer to skb 603 * Return: 8 Bytes non-decreasing number on success or 0 if the socket 604 * field is missing inside sk_buff 605 * 606 * u32 bpf_get_socket_uid(skb) 607 * Get the owner uid of the socket stored inside sk_buff. 608 * @skb: pointer to skb 609 * Return: uid of the socket owner on success or overflowuid if failed. 610 * 611 * u32 bpf_set_hash(skb, hash) 612 * Set full skb->hash. 613 * @skb: pointer to skb 614 * @hash: hash to set 615 * 616 * int bpf_setsockopt(bpf_socket, level, optname, optval, optlen) 617 * Calls setsockopt. Not all opts are available, only those with 618 * integer optvals plus TCP_CONGESTION. 619 * Supported levels: SOL_SOCKET and IPPROTO_TCP 620 * @bpf_socket: pointer to bpf_socket 621 * @level: SOL_SOCKET or IPPROTO_TCP 622 * @optname: option name 623 * @optval: pointer to option value 624 * @optlen: length of optval in bytes 625 * Return: 0 or negative error 626 * 627 * int bpf_getsockopt(bpf_socket, level, optname, optval, optlen) 628 * Calls getsockopt. Not all opts are available. 629 * Supported levels: IPPROTO_TCP 630 * @bpf_socket: pointer to bpf_socket 631 * @level: IPPROTO_TCP 632 * @optname: option name 633 * @optval: pointer to option value 634 * @optlen: length of optval in bytes 635 * Return: 0 or negative error 636 * 637 * int bpf_skb_adjust_room(skb, len_diff, mode, flags) 638 * Grow or shrink room in sk_buff. 639 * @skb: pointer to skb 640 * @len_diff: (signed) amount of room to grow/shrink 641 * @mode: operation mode (enum bpf_adj_room_mode) 642 * @flags: reserved for future use 643 * Return: 0 on success or negative error code 644 * 645 * int bpf_sk_redirect_map(map, key, flags) 646 * Redirect skb to a sock in map using key as a lookup key for the 647 * sock in map. 648 * @map: pointer to sockmap 649 * @key: key to lookup sock in map 650 * @flags: reserved for future use 651 * Return: SK_PASS 652 * 653 * int bpf_sock_map_update(skops, map, key, flags) 654 * @skops: pointer to bpf_sock_ops 655 * @map: pointer to sockmap to update 656 * @key: key to insert/update sock in map 657 * @flags: same flags as map update elem 658 * 659 * int bpf_xdp_adjust_meta(xdp_md, delta) 660 * Adjust the xdp_md.data_meta by delta 661 * @xdp_md: pointer to xdp_md 662 * @delta: An positive/negative integer to be added to xdp_md.data_meta 663 * Return: 0 on success or negative on error 664 * 665 * int bpf_perf_event_read_value(map, flags, buf, buf_size) 666 * read perf event counter value and perf event enabled/running time 667 * @map: pointer to perf_event_array map 668 * @flags: index of event in the map or bitmask flags 669 * @buf: buf to fill 670 * @buf_size: size of the buf 671 * Return: 0 on success or negative error code 672 * 673 * int bpf_perf_prog_read_value(ctx, buf, buf_size) 674 * read perf prog attached perf event counter and enabled/running time 675 * @ctx: pointer to ctx 676 * @buf: buf to fill 677 * @buf_size: size of the buf 678 * Return : 0 on success or negative error code 679 */ 680 #define __BPF_FUNC_MAPPER(FN) \ 681 FN(unspec), \ 682 FN(map_lookup_elem), \ 683 FN(map_update_elem), \ 684 FN(map_delete_elem), \ 685 FN(probe_read), \ 686 FN(ktime_get_ns), \ 687 FN(trace_printk), \ 688 FN(get_prandom_u32), \ 689 FN(get_smp_processor_id), \ 690 FN(skb_store_bytes), \ 691 FN(l3_csum_replace), \ 692 FN(l4_csum_replace), \ 693 FN(tail_call), \ 694 FN(clone_redirect), \ 695 FN(get_current_pid_tgid), \ 696 FN(get_current_uid_gid), \ 697 FN(get_current_comm), \ 698 FN(get_cgroup_classid), \ 699 FN(skb_vlan_push), \ 700 FN(skb_vlan_pop), \ 701 FN(skb_get_tunnel_key), \ 702 FN(skb_set_tunnel_key), \ 703 FN(perf_event_read), \ 704 FN(redirect), \ 705 FN(get_route_realm), \ 706 FN(perf_event_output), \ 707 FN(skb_load_bytes), \ 708 FN(get_stackid), \ 709 FN(csum_diff), \ 710 FN(skb_get_tunnel_opt), \ 711 FN(skb_set_tunnel_opt), \ 712 FN(skb_change_proto), \ 713 FN(skb_change_type), \ 714 FN(skb_under_cgroup), \ 715 FN(get_hash_recalc), \ 716 FN(get_current_task), \ 717 FN(probe_write_user), \ 718 FN(current_task_under_cgroup), \ 719 FN(skb_change_tail), \ 720 FN(skb_pull_data), \ 721 FN(csum_update), \ 722 FN(set_hash_invalid), \ 723 FN(get_numa_node_id), \ 724 FN(skb_change_head), \ 725 FN(xdp_adjust_head), \ 726 FN(probe_read_str), \ 727 FN(get_socket_cookie), \ 728 FN(get_socket_uid), \ 729 FN(set_hash), \ 730 FN(setsockopt), \ 731 FN(skb_adjust_room), \ 732 FN(redirect_map), \ 733 FN(sk_redirect_map), \ 734 FN(sock_map_update), \ 735 FN(xdp_adjust_meta), \ 736 FN(perf_event_read_value), \ 737 FN(perf_prog_read_value), \ 738 FN(getsockopt), 739 740 /* integer value in 'imm' field of BPF_CALL instruction selects which helper 741 * function eBPF program intends to call 742 */ 743 #define __BPF_ENUM_FN(x) BPF_FUNC_ ## x 744 enum bpf_func_id { 745 __BPF_FUNC_MAPPER(__BPF_ENUM_FN) 746 __BPF_FUNC_MAX_ID, 747 }; 748 #undef __BPF_ENUM_FN 749 750 /* All flags used by eBPF helper functions, placed here. */ 751 752 /* BPF_FUNC_skb_store_bytes flags. */ 753 #define BPF_F_RECOMPUTE_CSUM (1ULL << 0) 754 #define BPF_F_INVALIDATE_HASH (1ULL << 1) 755 756 /* BPF_FUNC_l3_csum_replace and BPF_FUNC_l4_csum_replace flags. 757 * First 4 bits are for passing the header field size. 758 */ 759 #define BPF_F_HDR_FIELD_MASK 0xfULL 760 761 /* BPF_FUNC_l4_csum_replace flags. */ 762 #define BPF_F_PSEUDO_HDR (1ULL << 4) 763 #define BPF_F_MARK_MANGLED_0 (1ULL << 5) 764 #define BPF_F_MARK_ENFORCE (1ULL << 6) 765 766 /* BPF_FUNC_clone_redirect and BPF_FUNC_redirect flags. */ 767 #define BPF_F_INGRESS (1ULL << 0) 768 769 /* BPF_FUNC_skb_set_tunnel_key and BPF_FUNC_skb_get_tunnel_key flags. */ 770 #define BPF_F_TUNINFO_IPV6 (1ULL << 0) 771 772 /* BPF_FUNC_get_stackid flags. */ 773 #define BPF_F_SKIP_FIELD_MASK 0xffULL 774 #define BPF_F_USER_STACK (1ULL << 8) 775 #define BPF_F_FAST_STACK_CMP (1ULL << 9) 776 #define BPF_F_REUSE_STACKID (1ULL << 10) 777 778 /* BPF_FUNC_skb_set_tunnel_key flags. */ 779 #define BPF_F_ZERO_CSUM_TX (1ULL << 1) 780 #define BPF_F_DONT_FRAGMENT (1ULL << 2) 781 782 /* BPF_FUNC_perf_event_output, BPF_FUNC_perf_event_read and 783 * BPF_FUNC_perf_event_read_value flags. 784 */ 785 #define BPF_F_INDEX_MASK 0xffffffffULL 786 #define BPF_F_CURRENT_CPU BPF_F_INDEX_MASK 787 /* BPF_FUNC_perf_event_output for sk_buff input context. */ 788 #define BPF_F_CTXLEN_MASK (0xfffffULL << 32) 789 790 /* Mode for BPF_FUNC_skb_adjust_room helper. */ 791 enum bpf_adj_room_mode { 792 BPF_ADJ_ROOM_NET, 793 }; 794 795 /* user accessible mirror of in-kernel sk_buff. 796 * new fields can only be added to the end of this structure 797 */ 798 struct __sk_buff { 799 __u32 len; 800 __u32 pkt_type; 801 __u32 mark; 802 __u32 queue_mapping; 803 __u32 protocol; 804 __u32 vlan_present; 805 __u32 vlan_tci; 806 __u32 vlan_proto; 807 __u32 priority; 808 __u32 ingress_ifindex; 809 __u32 ifindex; 810 __u32 tc_index; 811 __u32 cb[5]; 812 __u32 hash; 813 __u32 tc_classid; 814 __u32 data; 815 __u32 data_end; 816 __u32 napi_id; 817 818 /* Accessed by BPF_PROG_TYPE_sk_skb types from here to ... */ 819 __u32 family; 820 __u32 remote_ip4; /* Stored in network byte order */ 821 __u32 local_ip4; /* Stored in network byte order */ 822 __u32 remote_ip6[4]; /* Stored in network byte order */ 823 __u32 local_ip6[4]; /* Stored in network byte order */ 824 __u32 remote_port; /* Stored in network byte order */ 825 __u32 local_port; /* stored in host byte order */ 826 /* ... here. */ 827 828 __u32 data_meta; 829 }; 830 831 struct bpf_tunnel_key { 832 __u32 tunnel_id; 833 union { 834 __u32 remote_ipv4; 835 __u32 remote_ipv6[4]; 836 }; 837 __u8 tunnel_tos; 838 __u8 tunnel_ttl; 839 __u16 tunnel_ext; 840 __u32 tunnel_label; 841 }; 842 843 /* Generic BPF return codes which all BPF program types may support. 844 * The values are binary compatible with their TC_ACT_* counter-part to 845 * provide backwards compatibility with existing SCHED_CLS and SCHED_ACT 846 * programs. 847 * 848 * XDP is handled seprately, see XDP_*. 849 */ 850 enum bpf_ret_code { 851 BPF_OK = 0, 852 /* 1 reserved */ 853 BPF_DROP = 2, 854 /* 3-6 reserved */ 855 BPF_REDIRECT = 7, 856 /* >127 are reserved for prog type specific return codes */ 857 }; 858 859 struct bpf_sock { 860 __u32 bound_dev_if; 861 __u32 family; 862 __u32 type; 863 __u32 protocol; 864 __u32 mark; 865 __u32 priority; 866 }; 867 868 #define XDP_PACKET_HEADROOM 256 869 870 /* User return codes for XDP prog type. 871 * A valid XDP program must return one of these defined values. All other 872 * return codes are reserved for future use. Unknown return codes will 873 * result in packet drops and a warning via bpf_warn_invalid_xdp_action(). 874 */ 875 enum xdp_action { 876 XDP_ABORTED = 0, 877 XDP_DROP, 878 XDP_PASS, 879 XDP_TX, 880 XDP_REDIRECT, 881 }; 882 883 /* user accessible metadata for XDP packet hook 884 * new fields must be added to the end of this structure 885 */ 886 struct xdp_md { 887 __u32 data; 888 __u32 data_end; 889 __u32 data_meta; 890 }; 891 892 enum sk_action { 893 SK_DROP = 0, 894 SK_PASS, 895 }; 896 897 #define BPF_TAG_SIZE 8 898 899 enum bpf_prog_status { 900 BPF_PROG_STATUS_DEV_BOUND = (1 << 0), 901 }; 902 903 struct bpf_prog_info { 904 __u32 type; 905 __u32 id; 906 __u8 tag[BPF_TAG_SIZE]; 907 __u32 jited_prog_len; 908 __u32 xlated_prog_len; 909 __aligned_u64 jited_prog_insns; 910 __aligned_u64 xlated_prog_insns; 911 __u64 load_time; /* ns since boottime */ 912 __u32 created_by_uid; 913 __u32 nr_map_ids; 914 __aligned_u64 map_ids; 915 char name[BPF_OBJ_NAME_LEN]; 916 __u32 ifindex; 917 __u32 status; 918 } __attribute__((aligned(8))); 919 920 struct bpf_map_info { 921 __u32 type; 922 __u32 id; 923 __u32 key_size; 924 __u32 value_size; 925 __u32 max_entries; 926 __u32 map_flags; 927 char name[BPF_OBJ_NAME_LEN]; 928 } __attribute__((aligned(8))); 929 930 /* User bpf_sock_ops struct to access socket values and specify request ops 931 * and their replies. 932 * Some of this fields are in network (bigendian) byte order and may need 933 * to be converted before use (bpf_ntohl() defined in samples/bpf/bpf_endian.h). 934 * New fields can only be added at the end of this structure 935 */ 936 struct bpf_sock_ops { 937 __u32 op; 938 union { 939 __u32 reply; 940 __u32 replylong[4]; 941 }; 942 __u32 family; 943 __u32 remote_ip4; /* Stored in network byte order */ 944 __u32 local_ip4; /* Stored in network byte order */ 945 __u32 remote_ip6[4]; /* Stored in network byte order */ 946 __u32 local_ip6[4]; /* Stored in network byte order */ 947 __u32 remote_port; /* Stored in network byte order */ 948 __u32 local_port; /* stored in host byte order */ 949 }; 950 951 /* List of known BPF sock_ops operators. 952 * New entries can only be added at the end 953 */ 954 enum { 955 BPF_SOCK_OPS_VOID, 956 BPF_SOCK_OPS_TIMEOUT_INIT, /* Should return SYN-RTO value to use or 957 * -1 if default value should be used 958 */ 959 BPF_SOCK_OPS_RWND_INIT, /* Should return initial advertized 960 * window (in packets) or -1 if default 961 * value should be used 962 */ 963 BPF_SOCK_OPS_TCP_CONNECT_CB, /* Calls BPF program right before an 964 * active connection is initialized 965 */ 966 BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB, /* Calls BPF program when an 967 * active connection is 968 * established 969 */ 970 BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB, /* Calls BPF program when a 971 * passive connection is 972 * established 973 */ 974 BPF_SOCK_OPS_NEEDS_ECN, /* If connection's congestion control 975 * needs ECN 976 */ 977 BPF_SOCK_OPS_BASE_RTT, /* Get base RTT. The correct value is 978 * based on the path and may be 979 * dependent on the congestion control 980 * algorithm. In general it indicates 981 * a congestion threshold. RTTs above 982 * this indicate congestion 983 */ 984 }; 985 986 #define TCP_BPF_IW 1001 /* Set TCP initial congestion window */ 987 #define TCP_BPF_SNDCWND_CLAMP 1002 /* Set sndcwnd_clamp */ 988 989 struct bpf_perf_event_value { 990 __u64 counter; 991 __u64 enabled; 992 __u64 running; 993 }; 994 995 #define BPF_DEVCG_ACC_MKNOD (1ULL << 0) 996 #define BPF_DEVCG_ACC_READ (1ULL << 1) 997 #define BPF_DEVCG_ACC_WRITE (1ULL << 2) 998 999 #define BPF_DEVCG_DEV_BLOCK (1ULL << 0) 1000 #define BPF_DEVCG_DEV_CHAR (1ULL << 1) 1001 1002 struct bpf_cgroup_dev_ctx { 1003 __u32 access_type; /* (access << 16) | type */ 1004 __u32 major; 1005 __u32 minor; 1006 }; 1007 1008 #endif /* _UAPI__LINUX_BPF_H__ */ 1009