1===================== 2BPF Type Format (BTF) 3===================== 4 51. Introduction 6=============== 7 8BTF (BPF Type Format) is the metadata format which encodes the debug info 9related to BPF program/map. The name BTF was used initially to describe data 10types. The BTF was later extended to include function info for defined 11subroutines, and line info for source/line information. 12 13The debug info is used for map pretty print, function signature, etc. The 14function signature enables better bpf program/function kernel symbol. The line 15info helps generate source annotated translated byte code, jited code and 16verifier log. 17 18The BTF specification contains two parts, 19 * BTF kernel API 20 * BTF ELF file format 21 22The kernel API is the contract between user space and kernel. The kernel 23verifies the BTF info before using it. The ELF file format is a user space 24contract between ELF file and libbpf loader. 25 26The type and string sections are part of the BTF kernel API, describing the 27debug info (mostly types related) referenced by the bpf program. These two 28sections are discussed in details in :ref:`BTF_Type_String`. 29 30.. _BTF_Type_String: 31 322. BTF Type and String Encoding 33=============================== 34 35The file ``include/uapi/linux/btf.h`` provides high-level definition of how 36types/strings are encoded. 37 38The beginning of data blob must be:: 39 40 struct btf_header { 41 __u16 magic; 42 __u8 version; 43 __u8 flags; 44 __u32 hdr_len; 45 46 /* All offsets are in bytes relative to the end of this header */ 47 __u32 type_off; /* offset of type section */ 48 __u32 type_len; /* length of type section */ 49 __u32 str_off; /* offset of string section */ 50 __u32 str_len; /* length of string section */ 51 }; 52 53The magic is ``0xeB9F``, which has different encoding for big and little 54endian systems, and can be used to test whether BTF is generated for big- or 55little-endian target. The ``btf_header`` is designed to be extensible with 56``hdr_len`` equal to ``sizeof(struct btf_header)`` when a data blob is 57generated. 58 592.1 String Encoding 60------------------- 61 62The first string in the string section must be a null string. The rest of 63string table is a concatenation of other null-terminated strings. 64 652.2 Type Encoding 66----------------- 67 68The type id ``0`` is reserved for ``void`` type. The type section is parsed 69sequentially and type id is assigned to each recognized type starting from id 70``1``. Currently, the following types are supported:: 71 72 #define BTF_KIND_INT 1 /* Integer */ 73 #define BTF_KIND_PTR 2 /* Pointer */ 74 #define BTF_KIND_ARRAY 3 /* Array */ 75 #define BTF_KIND_STRUCT 4 /* Struct */ 76 #define BTF_KIND_UNION 5 /* Union */ 77 #define BTF_KIND_ENUM 6 /* Enumeration up to 32-bit values */ 78 #define BTF_KIND_FWD 7 /* Forward */ 79 #define BTF_KIND_TYPEDEF 8 /* Typedef */ 80 #define BTF_KIND_VOLATILE 9 /* Volatile */ 81 #define BTF_KIND_CONST 10 /* Const */ 82 #define BTF_KIND_RESTRICT 11 /* Restrict */ 83 #define BTF_KIND_FUNC 12 /* Function */ 84 #define BTF_KIND_FUNC_PROTO 13 /* Function Proto */ 85 #define BTF_KIND_VAR 14 /* Variable */ 86 #define BTF_KIND_DATASEC 15 /* Section */ 87 #define BTF_KIND_FLOAT 16 /* Floating point */ 88 #define BTF_KIND_DECL_TAG 17 /* Decl Tag */ 89 #define BTF_KIND_TYPE_TAG 18 /* Type Tag */ 90 #define BTF_KIND_ENUM64 19 /* Enumeration up to 64-bit values */ 91 92Note that the type section encodes debug info, not just pure types. 93``BTF_KIND_FUNC`` is not a type, and it represents a defined subprogram. 94 95Each type contains the following common data:: 96 97 struct btf_type { 98 __u32 name_off; 99 /* "info" bits arrangement 100 * bits 0-15: vlen (e.g. # of struct's members) 101 * bits 16-23: unused 102 * bits 24-28: kind (e.g. int, ptr, array...etc) 103 * bits 29-30: unused 104 * bit 31: kind_flag, currently used by 105 * struct, union, fwd, enum and enum64. 106 */ 107 __u32 info; 108 /* "size" is used by INT, ENUM, STRUCT, UNION and ENUM64. 109 * "size" tells the size of the type it is describing. 110 * 111 * "type" is used by PTR, TYPEDEF, VOLATILE, CONST, RESTRICT, 112 * FUNC, FUNC_PROTO, DECL_TAG and TYPE_TAG. 113 * "type" is a type_id referring to another type. 114 */ 115 union { 116 __u32 size; 117 __u32 type; 118 }; 119 }; 120 121For certain kinds, the common data are followed by kind-specific data. The 122``name_off`` in ``struct btf_type`` specifies the offset in the string table. 123The following sections detail encoding of each kind. 124 1252.2.1 BTF_KIND_INT 126~~~~~~~~~~~~~~~~~~ 127 128``struct btf_type`` encoding requirement: 129 * ``name_off``: any valid offset 130 * ``info.kind_flag``: 0 131 * ``info.kind``: BTF_KIND_INT 132 * ``info.vlen``: 0 133 * ``size``: the size of the int type in bytes. 134 135``btf_type`` is followed by a ``u32`` with the following bits arrangement:: 136 137 #define BTF_INT_ENCODING(VAL) (((VAL) & 0x0f000000) >> 24) 138 #define BTF_INT_OFFSET(VAL) (((VAL) & 0x00ff0000) >> 16) 139 #define BTF_INT_BITS(VAL) ((VAL) & 0x000000ff) 140 141The ``BTF_INT_ENCODING`` has the following attributes:: 142 143 #define BTF_INT_SIGNED (1 << 0) 144 #define BTF_INT_CHAR (1 << 1) 145 #define BTF_INT_BOOL (1 << 2) 146 147The ``BTF_INT_ENCODING()`` provides extra information: signedness, char, or 148bool, for the int type. The char and bool encoding are mostly useful for 149pretty print. At most one encoding can be specified for the int type. 150 151The ``BTF_INT_BITS()`` specifies the number of actual bits held by this int 152type. For example, a 4-bit bitfield encodes ``BTF_INT_BITS()`` equals to 4. 153The ``btf_type.size * 8`` must be equal to or greater than ``BTF_INT_BITS()`` 154for the type. The maximum value of ``BTF_INT_BITS()`` is 128. 155 156The ``BTF_INT_OFFSET()`` specifies the starting bit offset to calculate values 157for this int. For example, a bitfield struct member has: 158 159 * btf member bit offset 100 from the start of the structure, 160 * btf member pointing to an int type, 161 * the int type has ``BTF_INT_OFFSET() = 2`` and ``BTF_INT_BITS() = 4`` 162 163Then in the struct memory layout, this member will occupy ``4`` bits starting 164from bits ``100 + 2 = 102``. 165 166Alternatively, the bitfield struct member can be the following to access the 167same bits as the above: 168 169 * btf member bit offset 102, 170 * btf member pointing to an int type, 171 * the int type has ``BTF_INT_OFFSET() = 0`` and ``BTF_INT_BITS() = 4`` 172 173The original intention of ``BTF_INT_OFFSET()`` is to provide flexibility of 174bitfield encoding. Currently, both llvm and pahole generate 175``BTF_INT_OFFSET() = 0`` for all int types. 176 1772.2.2 BTF_KIND_PTR 178~~~~~~~~~~~~~~~~~~ 179 180``struct btf_type`` encoding requirement: 181 * ``name_off``: 0 182 * ``info.kind_flag``: 0 183 * ``info.kind``: BTF_KIND_PTR 184 * ``info.vlen``: 0 185 * ``type``: the pointee type of the pointer 186 187No additional type data follow ``btf_type``. 188 1892.2.3 BTF_KIND_ARRAY 190~~~~~~~~~~~~~~~~~~~~ 191 192``struct btf_type`` encoding requirement: 193 * ``name_off``: 0 194 * ``info.kind_flag``: 0 195 * ``info.kind``: BTF_KIND_ARRAY 196 * ``info.vlen``: 0 197 * ``size/type``: 0, not used 198 199``btf_type`` is followed by one ``struct btf_array``:: 200 201 struct btf_array { 202 __u32 type; 203 __u32 index_type; 204 __u32 nelems; 205 }; 206 207The ``struct btf_array`` encoding: 208 * ``type``: the element type 209 * ``index_type``: the index type 210 * ``nelems``: the number of elements for this array (``0`` is also allowed). 211 212The ``index_type`` can be any regular int type (``u8``, ``u16``, ``u32``, 213``u64``, ``unsigned __int128``). The original design of including 214``index_type`` follows DWARF, which has an ``index_type`` for its array type. 215Currently in BTF, beyond type verification, the ``index_type`` is not used. 216 217The ``struct btf_array`` allows chaining through element type to represent 218multidimensional arrays. For example, for ``int a[5][6]``, the following type 219information illustrates the chaining: 220 221 * [1]: int 222 * [2]: array, ``btf_array.type = [1]``, ``btf_array.nelems = 6`` 223 * [3]: array, ``btf_array.type = [2]``, ``btf_array.nelems = 5`` 224 225Currently, both pahole and llvm collapse multidimensional array into 226one-dimensional array, e.g., for ``a[5][6]``, the ``btf_array.nelems`` is 227equal to ``30``. This is because the original use case is map pretty print 228where the whole array is dumped out so one-dimensional array is enough. As 229more BTF usage is explored, pahole and llvm can be changed to generate proper 230chained representation for multidimensional arrays. 231 2322.2.4 BTF_KIND_STRUCT 233~~~~~~~~~~~~~~~~~~~~~ 2342.2.5 BTF_KIND_UNION 235~~~~~~~~~~~~~~~~~~~~ 236 237``struct btf_type`` encoding requirement: 238 * ``name_off``: 0 or offset to a valid C identifier 239 * ``info.kind_flag``: 0 or 1 240 * ``info.kind``: BTF_KIND_STRUCT or BTF_KIND_UNION 241 * ``info.vlen``: the number of struct/union members 242 * ``info.size``: the size of the struct/union in bytes 243 244``btf_type`` is followed by ``info.vlen`` number of ``struct btf_member``.:: 245 246 struct btf_member { 247 __u32 name_off; 248 __u32 type; 249 __u32 offset; 250 }; 251 252``struct btf_member`` encoding: 253 * ``name_off``: offset to a valid C identifier 254 * ``type``: the member type 255 * ``offset``: <see below> 256 257If the type info ``kind_flag`` is not set, the offset contains only bit offset 258of the member. Note that the base type of the bitfield can only be int or enum 259type. If the bitfield size is 32, the base type can be either int or enum 260type. If the bitfield size is not 32, the base type must be int, and int type 261``BTF_INT_BITS()`` encodes the bitfield size. 262 263If the ``kind_flag`` is set, the ``btf_member.offset`` contains both member 264bitfield size and bit offset. The bitfield size and bit offset are calculated 265as below.:: 266 267 #define BTF_MEMBER_BITFIELD_SIZE(val) ((val) >> 24) 268 #define BTF_MEMBER_BIT_OFFSET(val) ((val) & 0xffffff) 269 270In this case, if the base type is an int type, it must be a regular int type: 271 272 * ``BTF_INT_OFFSET()`` must be 0. 273 * ``BTF_INT_BITS()`` must be equal to ``{1,2,4,8,16} * 8``. 274 275Commit 9d5f9f701b18 introduced ``kind_flag`` and explains why both modes 276exist. 277 2782.2.6 BTF_KIND_ENUM 279~~~~~~~~~~~~~~~~~~~ 280 281``struct btf_type`` encoding requirement: 282 * ``name_off``: 0 or offset to a valid C identifier 283 * ``info.kind_flag``: 0 for unsigned, 1 for signed 284 * ``info.kind``: BTF_KIND_ENUM 285 * ``info.vlen``: number of enum values 286 * ``size``: 1/2/4/8 287 288``btf_type`` is followed by ``info.vlen`` number of ``struct btf_enum``.:: 289 290 struct btf_enum { 291 __u32 name_off; 292 __s32 val; 293 }; 294 295The ``btf_enum`` encoding: 296 * ``name_off``: offset to a valid C identifier 297 * ``val``: any value 298 299If the original enum value is signed and the size is less than 4, 300that value will be sign extended into 4 bytes. If the size is 8, 301the value will be truncated into 4 bytes. 302 3032.2.7 BTF_KIND_FWD 304~~~~~~~~~~~~~~~~~~ 305 306``struct btf_type`` encoding requirement: 307 * ``name_off``: offset to a valid C identifier 308 * ``info.kind_flag``: 0 for struct, 1 for union 309 * ``info.kind``: BTF_KIND_FWD 310 * ``info.vlen``: 0 311 * ``type``: 0 312 313No additional type data follow ``btf_type``. 314 3152.2.8 BTF_KIND_TYPEDEF 316~~~~~~~~~~~~~~~~~~~~~~ 317 318``struct btf_type`` encoding requirement: 319 * ``name_off``: offset to a valid C identifier 320 * ``info.kind_flag``: 0 321 * ``info.kind``: BTF_KIND_TYPEDEF 322 * ``info.vlen``: 0 323 * ``type``: the type which can be referred by name at ``name_off`` 324 325No additional type data follow ``btf_type``. 326 3272.2.9 BTF_KIND_VOLATILE 328~~~~~~~~~~~~~~~~~~~~~~~ 329 330``struct btf_type`` encoding requirement: 331 * ``name_off``: 0 332 * ``info.kind_flag``: 0 333 * ``info.kind``: BTF_KIND_VOLATILE 334 * ``info.vlen``: 0 335 * ``type``: the type with ``volatile`` qualifier 336 337No additional type data follow ``btf_type``. 338 3392.2.10 BTF_KIND_CONST 340~~~~~~~~~~~~~~~~~~~~~ 341 342``struct btf_type`` encoding requirement: 343 * ``name_off``: 0 344 * ``info.kind_flag``: 0 345 * ``info.kind``: BTF_KIND_CONST 346 * ``info.vlen``: 0 347 * ``type``: the type with ``const`` qualifier 348 349No additional type data follow ``btf_type``. 350 3512.2.11 BTF_KIND_RESTRICT 352~~~~~~~~~~~~~~~~~~~~~~~~ 353 354``struct btf_type`` encoding requirement: 355 * ``name_off``: 0 356 * ``info.kind_flag``: 0 357 * ``info.kind``: BTF_KIND_RESTRICT 358 * ``info.vlen``: 0 359 * ``type``: the type with ``restrict`` qualifier 360 361No additional type data follow ``btf_type``. 362 3632.2.12 BTF_KIND_FUNC 364~~~~~~~~~~~~~~~~~~~~ 365 366``struct btf_type`` encoding requirement: 367 * ``name_off``: offset to a valid C identifier 368 * ``info.kind_flag``: 0 369 * ``info.kind``: BTF_KIND_FUNC 370 * ``info.vlen``: linkage information (BTF_FUNC_STATIC, BTF_FUNC_GLOBAL 371 or BTF_FUNC_EXTERN) 372 * ``type``: a BTF_KIND_FUNC_PROTO type 373 374No additional type data follow ``btf_type``. 375 376A BTF_KIND_FUNC defines not a type, but a subprogram (function) whose 377signature is defined by ``type``. The subprogram is thus an instance of that 378type. The BTF_KIND_FUNC may in turn be referenced by a func_info in the 379:ref:`BTF_Ext_Section` (ELF) or in the arguments to :ref:`BPF_Prog_Load` 380(ABI). 381 382Currently, only linkage values of BTF_FUNC_STATIC and BTF_FUNC_GLOBAL are 383supported in the kernel. 384 3852.2.13 BTF_KIND_FUNC_PROTO 386~~~~~~~~~~~~~~~~~~~~~~~~~~ 387 388``struct btf_type`` encoding requirement: 389 * ``name_off``: 0 390 * ``info.kind_flag``: 0 391 * ``info.kind``: BTF_KIND_FUNC_PROTO 392 * ``info.vlen``: # of parameters 393 * ``type``: the return type 394 395``btf_type`` is followed by ``info.vlen`` number of ``struct btf_param``.:: 396 397 struct btf_param { 398 __u32 name_off; 399 __u32 type; 400 }; 401 402If a BTF_KIND_FUNC_PROTO type is referred by a BTF_KIND_FUNC type, then 403``btf_param.name_off`` must point to a valid C identifier except for the 404possible last argument representing the variable argument. The btf_param.type 405refers to parameter type. 406 407If the function has variable arguments, the last parameter is encoded with 408``name_off = 0`` and ``type = 0``. 409 4102.2.14 BTF_KIND_VAR 411~~~~~~~~~~~~~~~~~~~ 412 413``struct btf_type`` encoding requirement: 414 * ``name_off``: offset to a valid C identifier 415 * ``info.kind_flag``: 0 416 * ``info.kind``: BTF_KIND_VAR 417 * ``info.vlen``: 0 418 * ``type``: the type of the variable 419 420``btf_type`` is followed by a single ``struct btf_variable`` with the 421following data:: 422 423 struct btf_var { 424 __u32 linkage; 425 }; 426 427``struct btf_var`` encoding: 428 * ``linkage``: currently only static variable 0, or globally allocated 429 variable in ELF sections 1 430 431Not all type of global variables are supported by LLVM at this point. 432The following is currently available: 433 434 * static variables with or without section attributes 435 * global variables with section attributes 436 437The latter is for future extraction of map key/value type id's from a 438map definition. 439 4402.2.15 BTF_KIND_DATASEC 441~~~~~~~~~~~~~~~~~~~~~~~ 442 443``struct btf_type`` encoding requirement: 444 * ``name_off``: offset to a valid name associated with a variable or 445 one of .data/.bss/.rodata 446 * ``info.kind_flag``: 0 447 * ``info.kind``: BTF_KIND_DATASEC 448 * ``info.vlen``: # of variables 449 * ``size``: total section size in bytes (0 at compilation time, patched 450 to actual size by BPF loaders such as libbpf) 451 452``btf_type`` is followed by ``info.vlen`` number of ``struct btf_var_secinfo``.:: 453 454 struct btf_var_secinfo { 455 __u32 type; 456 __u32 offset; 457 __u32 size; 458 }; 459 460``struct btf_var_secinfo`` encoding: 461 * ``type``: the type of the BTF_KIND_VAR variable 462 * ``offset``: the in-section offset of the variable 463 * ``size``: the size of the variable in bytes 464 4652.2.16 BTF_KIND_FLOAT 466~~~~~~~~~~~~~~~~~~~~~ 467 468``struct btf_type`` encoding requirement: 469 * ``name_off``: any valid offset 470 * ``info.kind_flag``: 0 471 * ``info.kind``: BTF_KIND_FLOAT 472 * ``info.vlen``: 0 473 * ``size``: the size of the float type in bytes: 2, 4, 8, 12 or 16. 474 475No additional type data follow ``btf_type``. 476 4772.2.17 BTF_KIND_DECL_TAG 478~~~~~~~~~~~~~~~~~~~~~~~~ 479 480``struct btf_type`` encoding requirement: 481 * ``name_off``: offset to a non-empty string 482 * ``info.kind_flag``: 0 483 * ``info.kind``: BTF_KIND_DECL_TAG 484 * ``info.vlen``: 0 485 * ``type``: ``struct``, ``union``, ``func``, ``var`` or ``typedef`` 486 487``btf_type`` is followed by ``struct btf_decl_tag``.:: 488 489 struct btf_decl_tag { 490 __u32 component_idx; 491 }; 492 493The ``name_off`` encodes btf_decl_tag attribute string. 494The ``type`` should be ``struct``, ``union``, ``func``, ``var`` or ``typedef``. 495For ``var`` or ``typedef`` type, ``btf_decl_tag.component_idx`` must be ``-1``. 496For the other three types, if the btf_decl_tag attribute is 497applied to the ``struct``, ``union`` or ``func`` itself, 498``btf_decl_tag.component_idx`` must be ``-1``. Otherwise, 499the attribute is applied to a ``struct``/``union`` member or 500a ``func`` argument, and ``btf_decl_tag.component_idx`` should be a 501valid index (starting from 0) pointing to a member or an argument. 502 5032.2.18 BTF_KIND_TYPE_TAG 504~~~~~~~~~~~~~~~~~~~~~~~~ 505 506``struct btf_type`` encoding requirement: 507 * ``name_off``: offset to a non-empty string 508 * ``info.kind_flag``: 0 509 * ``info.kind``: BTF_KIND_TYPE_TAG 510 * ``info.vlen``: 0 511 * ``type``: the type with ``btf_type_tag`` attribute 512 513Currently, ``BTF_KIND_TYPE_TAG`` is only emitted for pointer types. 514It has the following btf type chain: 515:: 516 517 ptr -> [type_tag]* 518 -> [const | volatile | restrict | typedef]* 519 -> base_type 520 521Basically, a pointer type points to zero or more 522type_tag, then zero or more const/volatile/restrict/typedef 523and finally the base type. The base type is one of 524int, ptr, array, struct, union, enum, func_proto and float types. 525 5262.2.19 BTF_KIND_ENUM64 527~~~~~~~~~~~~~~~~~~~~~~ 528 529``struct btf_type`` encoding requirement: 530 * ``name_off``: 0 or offset to a valid C identifier 531 * ``info.kind_flag``: 0 for unsigned, 1 for signed 532 * ``info.kind``: BTF_KIND_ENUM64 533 * ``info.vlen``: number of enum values 534 * ``size``: 1/2/4/8 535 536``btf_type`` is followed by ``info.vlen`` number of ``struct btf_enum64``.:: 537 538 struct btf_enum64 { 539 __u32 name_off; 540 __u32 val_lo32; 541 __u32 val_hi32; 542 }; 543 544The ``btf_enum64`` encoding: 545 * ``name_off``: offset to a valid C identifier 546 * ``val_lo32``: lower 32-bit value for a 64-bit value 547 * ``val_hi32``: high 32-bit value for a 64-bit value 548 549If the original enum value is signed and the size is less than 8, 550that value will be sign extended into 8 bytes. 551 5523. BTF Kernel API 553================= 554 555The following bpf syscall command involves BTF: 556 * BPF_BTF_LOAD: load a blob of BTF data into kernel 557 * BPF_MAP_CREATE: map creation with btf key and value type info. 558 * BPF_PROG_LOAD: prog load with btf function and line info. 559 * BPF_BTF_GET_FD_BY_ID: get a btf fd 560 * BPF_OBJ_GET_INFO_BY_FD: btf, func_info, line_info 561 and other btf related info are returned. 562 563The workflow typically looks like: 564:: 565 566 Application: 567 BPF_BTF_LOAD 568 | 569 v 570 BPF_MAP_CREATE and BPF_PROG_LOAD 571 | 572 V 573 ...... 574 575 Introspection tool: 576 ...... 577 BPF_{PROG,MAP}_GET_NEXT_ID (get prog/map id's) 578 | 579 V 580 BPF_{PROG,MAP}_GET_FD_BY_ID (get a prog/map fd) 581 | 582 V 583 BPF_OBJ_GET_INFO_BY_FD (get bpf_prog_info/bpf_map_info with btf_id) 584 | | 585 V | 586 BPF_BTF_GET_FD_BY_ID (get btf_fd) | 587 | | 588 V | 589 BPF_OBJ_GET_INFO_BY_FD (get btf) | 590 | | 591 V V 592 pretty print types, dump func signatures and line info, etc. 593 594 5953.1 BPF_BTF_LOAD 596---------------- 597 598Load a blob of BTF data into kernel. A blob of data, described in 599:ref:`BTF_Type_String`, can be directly loaded into the kernel. A ``btf_fd`` 600is returned to a userspace. 601 6023.2 BPF_MAP_CREATE 603------------------ 604 605A map can be created with ``btf_fd`` and specified key/value type id.:: 606 607 __u32 btf_fd; /* fd pointing to a BTF type data */ 608 __u32 btf_key_type_id; /* BTF type_id of the key */ 609 __u32 btf_value_type_id; /* BTF type_id of the value */ 610 611In libbpf, the map can be defined with extra annotation like below: 612:: 613 614 struct { 615 __uint(type, BPF_MAP_TYPE_ARRAY); 616 __type(key, int); 617 __type(value, struct ipv_counts); 618 __uint(max_entries, 4); 619 } btf_map SEC(".maps"); 620 621During ELF parsing, libbpf is able to extract key/value type_id's and assign 622them to BPF_MAP_CREATE attributes automatically. 623 624.. _BPF_Prog_Load: 625 6263.3 BPF_PROG_LOAD 627----------------- 628 629During prog_load, func_info and line_info can be passed to kernel with proper 630values for the following attributes: 631:: 632 633 __u32 insn_cnt; 634 __aligned_u64 insns; 635 ...... 636 __u32 prog_btf_fd; /* fd pointing to BTF type data */ 637 __u32 func_info_rec_size; /* userspace bpf_func_info size */ 638 __aligned_u64 func_info; /* func info */ 639 __u32 func_info_cnt; /* number of bpf_func_info records */ 640 __u32 line_info_rec_size; /* userspace bpf_line_info size */ 641 __aligned_u64 line_info; /* line info */ 642 __u32 line_info_cnt; /* number of bpf_line_info records */ 643 644The func_info and line_info are an array of below, respectively.:: 645 646 struct bpf_func_info { 647 __u32 insn_off; /* [0, insn_cnt - 1] */ 648 __u32 type_id; /* pointing to a BTF_KIND_FUNC type */ 649 }; 650 struct bpf_line_info { 651 __u32 insn_off; /* [0, insn_cnt - 1] */ 652 __u32 file_name_off; /* offset to string table for the filename */ 653 __u32 line_off; /* offset to string table for the source line */ 654 __u32 line_col; /* line number and column number */ 655 }; 656 657func_info_rec_size is the size of each func_info record, and 658line_info_rec_size is the size of each line_info record. Passing the record 659size to kernel make it possible to extend the record itself in the future. 660 661Below are requirements for func_info: 662 * func_info[0].insn_off must be 0. 663 * the func_info insn_off is in strictly increasing order and matches 664 bpf func boundaries. 665 666Below are requirements for line_info: 667 * the first insn in each func must have a line_info record pointing to it. 668 * the line_info insn_off is in strictly increasing order. 669 670For line_info, the line number and column number are defined as below: 671:: 672 673 #define BPF_LINE_INFO_LINE_NUM(line_col) ((line_col) >> 10) 674 #define BPF_LINE_INFO_LINE_COL(line_col) ((line_col) & 0x3ff) 675 6763.4 BPF_{PROG,MAP}_GET_NEXT_ID 677------------------------------ 678 679In kernel, every loaded program, map or btf has a unique id. The id won't 680change during the lifetime of a program, map, or btf. 681 682The bpf syscall command BPF_{PROG,MAP}_GET_NEXT_ID returns all id's, one for 683each command, to user space, for bpf program or maps, respectively, so an 684inspection tool can inspect all programs and maps. 685 6863.5 BPF_{PROG,MAP}_GET_FD_BY_ID 687------------------------------- 688 689An introspection tool cannot use id to get details about program or maps. 690A file descriptor needs to be obtained first for reference-counting purpose. 691 6923.6 BPF_OBJ_GET_INFO_BY_FD 693-------------------------- 694 695Once a program/map fd is acquired, an introspection tool can get the detailed 696information from kernel about this fd, some of which are BTF-related. For 697example, ``bpf_map_info`` returns ``btf_id`` and key/value type ids. 698``bpf_prog_info`` returns ``btf_id``, func_info, and line info for translated 699bpf byte codes, and jited_line_info. 700 7013.7 BPF_BTF_GET_FD_BY_ID 702------------------------ 703 704With ``btf_id`` obtained in ``bpf_map_info`` and ``bpf_prog_info``, bpf 705syscall command BPF_BTF_GET_FD_BY_ID can retrieve a btf fd. Then, with 706command BPF_OBJ_GET_INFO_BY_FD, the btf blob, originally loaded into the 707kernel with BPF_BTF_LOAD, can be retrieved. 708 709With the btf blob, ``bpf_map_info``, and ``bpf_prog_info``, an introspection 710tool has full btf knowledge and is able to pretty print map key/values, dump 711func signatures and line info, along with byte/jit codes. 712 7134. ELF File Format Interface 714============================ 715 7164.1 .BTF section 717---------------- 718 719The .BTF section contains type and string data. The format of this section is 720same as the one describe in :ref:`BTF_Type_String`. 721 722.. _BTF_Ext_Section: 723 7244.2 .BTF.ext section 725-------------------- 726 727The .BTF.ext section encodes func_info, line_info and CO-RE relocations 728which needs loader manipulation before loading into the kernel. 729 730The specification for .BTF.ext section is defined at ``tools/lib/bpf/btf.h`` 731and ``tools/lib/bpf/btf.c``. 732 733The current header of .BTF.ext section:: 734 735 struct btf_ext_header { 736 __u16 magic; 737 __u8 version; 738 __u8 flags; 739 __u32 hdr_len; 740 741 /* All offsets are in bytes relative to the end of this header */ 742 __u32 func_info_off; 743 __u32 func_info_len; 744 __u32 line_info_off; 745 __u32 line_info_len; 746 747 /* optional part of .BTF.ext header */ 748 __u32 core_relo_off; 749 __u32 core_relo_len; 750 }; 751 752It is very similar to .BTF section. Instead of type/string section, it 753contains func_info, line_info and core_relo sub-sections. 754See :ref:`BPF_Prog_Load` for details about func_info and line_info 755record format. 756 757The func_info is organized as below.:: 758 759 func_info_rec_size /* __u32 value */ 760 btf_ext_info_sec for section #1 /* func_info for section #1 */ 761 btf_ext_info_sec for section #2 /* func_info for section #2 */ 762 ... 763 764``func_info_rec_size`` specifies the size of ``bpf_func_info`` structure when 765.BTF.ext is generated. ``btf_ext_info_sec``, defined below, is a collection of 766func_info for each specific ELF section.:: 767 768 struct btf_ext_info_sec { 769 __u32 sec_name_off; /* offset to section name */ 770 __u32 num_info; 771 /* Followed by num_info * record_size number of bytes */ 772 __u8 data[0]; 773 }; 774 775Here, num_info must be greater than 0. 776 777The line_info is organized as below.:: 778 779 line_info_rec_size /* __u32 value */ 780 btf_ext_info_sec for section #1 /* line_info for section #1 */ 781 btf_ext_info_sec for section #2 /* line_info for section #2 */ 782 ... 783 784``line_info_rec_size`` specifies the size of ``bpf_line_info`` structure when 785.BTF.ext is generated. 786 787The interpretation of ``bpf_func_info->insn_off`` and 788``bpf_line_info->insn_off`` is different between kernel API and ELF API. For 789kernel API, the ``insn_off`` is the instruction offset in the unit of ``struct 790bpf_insn``. For ELF API, the ``insn_off`` is the byte offset from the 791beginning of section (``btf_ext_info_sec->sec_name_off``). 792 793The core_relo is organized as below.:: 794 795 core_relo_rec_size /* __u32 value */ 796 btf_ext_info_sec for section #1 /* core_relo for section #1 */ 797 btf_ext_info_sec for section #2 /* core_relo for section #2 */ 798 799``core_relo_rec_size`` specifies the size of ``bpf_core_relo`` 800structure when .BTF.ext is generated. All ``bpf_core_relo`` structures 801within a single ``btf_ext_info_sec`` describe relocations applied to 802section named by ``btf_ext_info_sec->sec_name_off``. 803 804See :ref:`Documentation/bpf/llvm_reloc.rst <btf-co-re-relocations>` 805for more information on CO-RE relocations. 806 8074.2 .BTF_ids section 808-------------------- 809 810The .BTF_ids section encodes BTF ID values that are used within the kernel. 811 812This section is created during the kernel compilation with the help of 813macros defined in ``include/linux/btf_ids.h`` header file. Kernel code can 814use them to create lists and sets (sorted lists) of BTF ID values. 815 816The ``BTF_ID_LIST`` and ``BTF_ID`` macros define unsorted list of BTF ID values, 817with following syntax:: 818 819 BTF_ID_LIST(list) 820 BTF_ID(type1, name1) 821 BTF_ID(type2, name2) 822 823resulting in following layout in .BTF_ids section:: 824 825 __BTF_ID__type1__name1__1: 826 .zero 4 827 __BTF_ID__type2__name2__2: 828 .zero 4 829 830The ``u32 list[];`` variable is defined to access the list. 831 832The ``BTF_ID_UNUSED`` macro defines 4 zero bytes. It's used when we 833want to define unused entry in BTF_ID_LIST, like:: 834 835 BTF_ID_LIST(bpf_skb_output_btf_ids) 836 BTF_ID(struct, sk_buff) 837 BTF_ID_UNUSED 838 BTF_ID(struct, task_struct) 839 840The ``BTF_SET_START/END`` macros pair defines sorted list of BTF ID values 841and their count, with following syntax:: 842 843 BTF_SET_START(set) 844 BTF_ID(type1, name1) 845 BTF_ID(type2, name2) 846 BTF_SET_END(set) 847 848resulting in following layout in .BTF_ids section:: 849 850 __BTF_ID__set__set: 851 .zero 4 852 __BTF_ID__type1__name1__3: 853 .zero 4 854 __BTF_ID__type2__name2__4: 855 .zero 4 856 857The ``struct btf_id_set set;`` variable is defined to access the list. 858 859The ``typeX`` name can be one of following:: 860 861 struct, union, typedef, func 862 863and is used as a filter when resolving the BTF ID value. 864 865All the BTF ID lists and sets are compiled in the .BTF_ids section and 866resolved during the linking phase of kernel build by ``resolve_btfids`` tool. 867 8685. Using BTF 869============ 870 8715.1 bpftool map pretty print 872---------------------------- 873 874With BTF, the map key/value can be printed based on fields rather than simply 875raw bytes. This is especially valuable for large structure or if your data 876structure has bitfields. For example, for the following map,:: 877 878 enum A { A1, A2, A3, A4, A5 }; 879 typedef enum A ___A; 880 struct tmp_t { 881 char a1:4; 882 int a2:4; 883 int :4; 884 __u32 a3:4; 885 int b; 886 ___A b1:4; 887 enum A b2:4; 888 }; 889 struct { 890 __uint(type, BPF_MAP_TYPE_ARRAY); 891 __type(key, int); 892 __type(value, struct tmp_t); 893 __uint(max_entries, 1); 894 } tmpmap SEC(".maps"); 895 896bpftool is able to pretty print like below: 897:: 898 899 [{ 900 "key": 0, 901 "value": { 902 "a1": 0x2, 903 "a2": 0x4, 904 "a3": 0x6, 905 "b": 7, 906 "b1": 0x8, 907 "b2": 0xa 908 } 909 } 910 ] 911 9125.2 bpftool prog dump 913--------------------- 914 915The following is an example showing how func_info and line_info can help prog 916dump with better kernel symbol names, function prototypes and line 917information.:: 918 919 $ bpftool prog dump jited pinned /sys/fs/bpf/test_btf_haskv 920 [...] 921 int test_long_fname_2(struct dummy_tracepoint_args * arg): 922 bpf_prog_44a040bf25481309_test_long_fname_2: 923 ; static int test_long_fname_2(struct dummy_tracepoint_args *arg) 924 0: push %rbp 925 1: mov %rsp,%rbp 926 4: sub $0x30,%rsp 927 b: sub $0x28,%rbp 928 f: mov %rbx,0x0(%rbp) 929 13: mov %r13,0x8(%rbp) 930 17: mov %r14,0x10(%rbp) 931 1b: mov %r15,0x18(%rbp) 932 1f: xor %eax,%eax 933 21: mov %rax,0x20(%rbp) 934 25: xor %esi,%esi 935 ; int key = 0; 936 27: mov %esi,-0x4(%rbp) 937 ; if (!arg->sock) 938 2a: mov 0x8(%rdi),%rdi 939 ; if (!arg->sock) 940 2e: cmp $0x0,%rdi 941 32: je 0x0000000000000070 942 34: mov %rbp,%rsi 943 ; counts = bpf_map_lookup_elem(&btf_map, &key); 944 [...] 945 9465.3 Verifier Log 947---------------- 948 949The following is an example of how line_info can help debugging verification 950failure.:: 951 952 /* The code at tools/testing/selftests/bpf/test_xdp_noinline.c 953 * is modified as below. 954 */ 955 data = (void *)(long)xdp->data; 956 data_end = (void *)(long)xdp->data_end; 957 /* 958 if (data + 4 > data_end) 959 return XDP_DROP; 960 */ 961 *(u32 *)data = dst->dst; 962 963 $ bpftool prog load ./test_xdp_noinline.o /sys/fs/bpf/test_xdp_noinline type xdp 964 ; data = (void *)(long)xdp->data; 965 224: (79) r2 = *(u64 *)(r10 -112) 966 225: (61) r2 = *(u32 *)(r2 +0) 967 ; *(u32 *)data = dst->dst; 968 226: (63) *(u32 *)(r2 +0) = r1 969 invalid access to packet, off=0 size=4, R2(id=0,off=0,r=0) 970 R2 offset is outside of the packet 971 9726. BTF Generation 973================= 974 975You need latest pahole 976 977 https://git.kernel.org/pub/scm/devel/pahole/pahole.git/ 978 979or llvm (8.0 or later). The pahole acts as a dwarf2btf converter. It doesn't 980support .BTF.ext and btf BTF_KIND_FUNC type yet. For example,:: 981 982 -bash-4.4$ cat t.c 983 struct t { 984 int a:2; 985 int b:3; 986 int c:2; 987 } g; 988 -bash-4.4$ gcc -c -O2 -g t.c 989 -bash-4.4$ pahole -JV t.o 990 File t.o: 991 [1] STRUCT t kind_flag=1 size=4 vlen=3 992 a type_id=2 bitfield_size=2 bits_offset=0 993 b type_id=2 bitfield_size=3 bits_offset=2 994 c type_id=2 bitfield_size=2 bits_offset=5 995 [2] INT int size=4 bit_offset=0 nr_bits=32 encoding=SIGNED 996 997The llvm is able to generate .BTF and .BTF.ext directly with -g for bpf target 998only. The assembly code (-S) is able to show the BTF encoding in assembly 999format.:: 1000 1001 -bash-4.4$ cat t2.c 1002 typedef int __int32; 1003 struct t2 { 1004 int a2; 1005 int (*f2)(char q1, __int32 q2, ...); 1006 int (*f3)(); 1007 } g2; 1008 int main() { return 0; } 1009 int test() { return 0; } 1010 -bash-4.4$ clang -c -g -O2 --target=bpf t2.c 1011 -bash-4.4$ readelf -S t2.o 1012 ...... 1013 [ 8] .BTF PROGBITS 0000000000000000 00000247 1014 000000000000016e 0000000000000000 0 0 1 1015 [ 9] .BTF.ext PROGBITS 0000000000000000 000003b5 1016 0000000000000060 0000000000000000 0 0 1 1017 [10] .rel.BTF.ext REL 0000000000000000 000007e0 1018 0000000000000040 0000000000000010 16 9 8 1019 ...... 1020 -bash-4.4$ clang -S -g -O2 --target=bpf t2.c 1021 -bash-4.4$ cat t2.s 1022 ...... 1023 .section .BTF,"",@progbits 1024 .short 60319 # 0xeb9f 1025 .byte 1 1026 .byte 0 1027 .long 24 1028 .long 0 1029 .long 220 1030 .long 220 1031 .long 122 1032 .long 0 # BTF_KIND_FUNC_PROTO(id = 1) 1033 .long 218103808 # 0xd000000 1034 .long 2 1035 .long 83 # BTF_KIND_INT(id = 2) 1036 .long 16777216 # 0x1000000 1037 .long 4 1038 .long 16777248 # 0x1000020 1039 ...... 1040 .byte 0 # string offset=0 1041 .ascii ".text" # string offset=1 1042 .byte 0 1043 .ascii "/home/yhs/tmp-pahole/t2.c" # string offset=7 1044 .byte 0 1045 .ascii "int main() { return 0; }" # string offset=33 1046 .byte 0 1047 .ascii "int test() { return 0; }" # string offset=58 1048 .byte 0 1049 .ascii "int" # string offset=83 1050 ...... 1051 .section .BTF.ext,"",@progbits 1052 .short 60319 # 0xeb9f 1053 .byte 1 1054 .byte 0 1055 .long 24 1056 .long 0 1057 .long 28 1058 .long 28 1059 .long 44 1060 .long 8 # FuncInfo 1061 .long 1 # FuncInfo section string offset=1 1062 .long 2 1063 .long .Lfunc_begin0 1064 .long 3 1065 .long .Lfunc_begin1 1066 .long 5 1067 .long 16 # LineInfo 1068 .long 1 # LineInfo section string offset=1 1069 .long 2 1070 .long .Ltmp0 1071 .long 7 1072 .long 33 1073 .long 7182 # Line 7 Col 14 1074 .long .Ltmp3 1075 .long 7 1076 .long 58 1077 .long 8206 # Line 8 Col 14 1078 10797. Testing 1080========== 1081 1082The kernel BPF selftest `tools/testing/selftests/bpf/prog_tests/btf.c`_ 1083provides an extensive set of BTF-related tests. 1084 1085.. Links 1086.. _tools/testing/selftests/bpf/prog_tests/btf.c: 1087 https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/tools/testing/selftests/bpf/prog_tests/btf.c 1088