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 275The following kernel patch introduced ``kind_flag`` and explained why both 276modes exist: 277 278 https://github.com/torvalds/linux/commit/9d5f9f701b1891466fb3dbb1806ad97716f95cc3#diff-fa650a64fdd3968396883d2fe8215ff3 279 2802.2.6 BTF_KIND_ENUM 281~~~~~~~~~~~~~~~~~~~ 282 283``struct btf_type`` encoding requirement: 284 * ``name_off``: 0 or offset to a valid C identifier 285 * ``info.kind_flag``: 0 for unsigned, 1 for signed 286 * ``info.kind``: BTF_KIND_ENUM 287 * ``info.vlen``: number of enum values 288 * ``size``: 1/2/4/8 289 290``btf_type`` is followed by ``info.vlen`` number of ``struct btf_enum``.:: 291 292 struct btf_enum { 293 __u32 name_off; 294 __s32 val; 295 }; 296 297The ``btf_enum`` encoding: 298 * ``name_off``: offset to a valid C identifier 299 * ``val``: any value 300 301If the original enum value is signed and the size is less than 4, 302that value will be sign extended into 4 bytes. If the size is 8, 303the value will be truncated into 4 bytes. 304 3052.2.7 BTF_KIND_FWD 306~~~~~~~~~~~~~~~~~~ 307 308``struct btf_type`` encoding requirement: 309 * ``name_off``: offset to a valid C identifier 310 * ``info.kind_flag``: 0 for struct, 1 for union 311 * ``info.kind``: BTF_KIND_FWD 312 * ``info.vlen``: 0 313 * ``type``: 0 314 315No additional type data follow ``btf_type``. 316 3172.2.8 BTF_KIND_TYPEDEF 318~~~~~~~~~~~~~~~~~~~~~~ 319 320``struct btf_type`` encoding requirement: 321 * ``name_off``: offset to a valid C identifier 322 * ``info.kind_flag``: 0 323 * ``info.kind``: BTF_KIND_TYPEDEF 324 * ``info.vlen``: 0 325 * ``type``: the type which can be referred by name at ``name_off`` 326 327No additional type data follow ``btf_type``. 328 3292.2.9 BTF_KIND_VOLATILE 330~~~~~~~~~~~~~~~~~~~~~~~ 331 332``struct btf_type`` encoding requirement: 333 * ``name_off``: 0 334 * ``info.kind_flag``: 0 335 * ``info.kind``: BTF_KIND_VOLATILE 336 * ``info.vlen``: 0 337 * ``type``: the type with ``volatile`` qualifier 338 339No additional type data follow ``btf_type``. 340 3412.2.10 BTF_KIND_CONST 342~~~~~~~~~~~~~~~~~~~~~ 343 344``struct btf_type`` encoding requirement: 345 * ``name_off``: 0 346 * ``info.kind_flag``: 0 347 * ``info.kind``: BTF_KIND_CONST 348 * ``info.vlen``: 0 349 * ``type``: the type with ``const`` qualifier 350 351No additional type data follow ``btf_type``. 352 3532.2.11 BTF_KIND_RESTRICT 354~~~~~~~~~~~~~~~~~~~~~~~~ 355 356``struct btf_type`` encoding requirement: 357 * ``name_off``: 0 358 * ``info.kind_flag``: 0 359 * ``info.kind``: BTF_KIND_RESTRICT 360 * ``info.vlen``: 0 361 * ``type``: the type with ``restrict`` qualifier 362 363No additional type data follow ``btf_type``. 364 3652.2.12 BTF_KIND_FUNC 366~~~~~~~~~~~~~~~~~~~~ 367 368``struct btf_type`` encoding requirement: 369 * ``name_off``: offset to a valid C identifier 370 * ``info.kind_flag``: 0 371 * ``info.kind``: BTF_KIND_FUNC 372 * ``info.vlen``: linkage information (BTF_FUNC_STATIC, BTF_FUNC_GLOBAL 373 or BTF_FUNC_EXTERN) 374 * ``type``: a BTF_KIND_FUNC_PROTO type 375 376No additional type data follow ``btf_type``. 377 378A BTF_KIND_FUNC defines not a type, but a subprogram (function) whose 379signature is defined by ``type``. The subprogram is thus an instance of that 380type. The BTF_KIND_FUNC may in turn be referenced by a func_info in the 381:ref:`BTF_Ext_Section` (ELF) or in the arguments to :ref:`BPF_Prog_Load` 382(ABI). 383 384Currently, only linkage values of BTF_FUNC_STATIC and BTF_FUNC_GLOBAL are 385supported in the kernel. 386 3872.2.13 BTF_KIND_FUNC_PROTO 388~~~~~~~~~~~~~~~~~~~~~~~~~~ 389 390``struct btf_type`` encoding requirement: 391 * ``name_off``: 0 392 * ``info.kind_flag``: 0 393 * ``info.kind``: BTF_KIND_FUNC_PROTO 394 * ``info.vlen``: # of parameters 395 * ``type``: the return type 396 397``btf_type`` is followed by ``info.vlen`` number of ``struct btf_param``.:: 398 399 struct btf_param { 400 __u32 name_off; 401 __u32 type; 402 }; 403 404If a BTF_KIND_FUNC_PROTO type is referred by a BTF_KIND_FUNC type, then 405``btf_param.name_off`` must point to a valid C identifier except for the 406possible last argument representing the variable argument. The btf_param.type 407refers to parameter type. 408 409If the function has variable arguments, the last parameter is encoded with 410``name_off = 0`` and ``type = 0``. 411 4122.2.14 BTF_KIND_VAR 413~~~~~~~~~~~~~~~~~~~ 414 415``struct btf_type`` encoding requirement: 416 * ``name_off``: offset to a valid C identifier 417 * ``info.kind_flag``: 0 418 * ``info.kind``: BTF_KIND_VAR 419 * ``info.vlen``: 0 420 * ``type``: the type of the variable 421 422``btf_type`` is followed by a single ``struct btf_variable`` with the 423following data:: 424 425 struct btf_var { 426 __u32 linkage; 427 }; 428 429``struct btf_var`` encoding: 430 * ``linkage``: currently only static variable 0, or globally allocated 431 variable in ELF sections 1 432 433Not all type of global variables are supported by LLVM at this point. 434The following is currently available: 435 436 * static variables with or without section attributes 437 * global variables with section attributes 438 439The latter is for future extraction of map key/value type id's from a 440map definition. 441 4422.2.15 BTF_KIND_DATASEC 443~~~~~~~~~~~~~~~~~~~~~~~ 444 445``struct btf_type`` encoding requirement: 446 * ``name_off``: offset to a valid name associated with a variable or 447 one of .data/.bss/.rodata 448 * ``info.kind_flag``: 0 449 * ``info.kind``: BTF_KIND_DATASEC 450 * ``info.vlen``: # of variables 451 * ``size``: total section size in bytes (0 at compilation time, patched 452 to actual size by BPF loaders such as libbpf) 453 454``btf_type`` is followed by ``info.vlen`` number of ``struct btf_var_secinfo``.:: 455 456 struct btf_var_secinfo { 457 __u32 type; 458 __u32 offset; 459 __u32 size; 460 }; 461 462``struct btf_var_secinfo`` encoding: 463 * ``type``: the type of the BTF_KIND_VAR variable 464 * ``offset``: the in-section offset of the variable 465 * ``size``: the size of the variable in bytes 466 4672.2.16 BTF_KIND_FLOAT 468~~~~~~~~~~~~~~~~~~~~~ 469 470``struct btf_type`` encoding requirement: 471 * ``name_off``: any valid offset 472 * ``info.kind_flag``: 0 473 * ``info.kind``: BTF_KIND_FLOAT 474 * ``info.vlen``: 0 475 * ``size``: the size of the float type in bytes: 2, 4, 8, 12 or 16. 476 477No additional type data follow ``btf_type``. 478 4792.2.17 BTF_KIND_DECL_TAG 480~~~~~~~~~~~~~~~~~~~~~~~~ 481 482``struct btf_type`` encoding requirement: 483 * ``name_off``: offset to a non-empty string 484 * ``info.kind_flag``: 0 485 * ``info.kind``: BTF_KIND_DECL_TAG 486 * ``info.vlen``: 0 487 * ``type``: ``struct``, ``union``, ``func``, ``var`` or ``typedef`` 488 489``btf_type`` is followed by ``struct btf_decl_tag``.:: 490 491 struct btf_decl_tag { 492 __u32 component_idx; 493 }; 494 495The ``name_off`` encodes btf_decl_tag attribute string. 496The ``type`` should be ``struct``, ``union``, ``func``, ``var`` or ``typedef``. 497For ``var`` or ``typedef`` type, ``btf_decl_tag.component_idx`` must be ``-1``. 498For the other three types, if the btf_decl_tag attribute is 499applied to the ``struct``, ``union`` or ``func`` itself, 500``btf_decl_tag.component_idx`` must be ``-1``. Otherwise, 501the attribute is applied to a ``struct``/``union`` member or 502a ``func`` argument, and ``btf_decl_tag.component_idx`` should be a 503valid index (starting from 0) pointing to a member or an argument. 504 5052.2.18 BTF_KIND_TYPE_TAG 506~~~~~~~~~~~~~~~~~~~~~~~~ 507 508``struct btf_type`` encoding requirement: 509 * ``name_off``: offset to a non-empty string 510 * ``info.kind_flag``: 0 511 * ``info.kind``: BTF_KIND_TYPE_TAG 512 * ``info.vlen``: 0 513 * ``type``: the type with ``btf_type_tag`` attribute 514 515Currently, ``BTF_KIND_TYPE_TAG`` is only emitted for pointer types. 516It has the following btf type chain: 517:: 518 519 ptr -> [type_tag]* 520 -> [const | volatile | restrict | typedef]* 521 -> base_type 522 523Basically, a pointer type points to zero or more 524type_tag, then zero or more const/volatile/restrict/typedef 525and finally the base type. The base type is one of 526int, ptr, array, struct, union, enum, func_proto and float types. 527 5282.2.19 BTF_KIND_ENUM64 529~~~~~~~~~~~~~~~~~~~~~~ 530 531``struct btf_type`` encoding requirement: 532 * ``name_off``: 0 or offset to a valid C identifier 533 * ``info.kind_flag``: 0 for unsigned, 1 for signed 534 * ``info.kind``: BTF_KIND_ENUM64 535 * ``info.vlen``: number of enum values 536 * ``size``: 1/2/4/8 537 538``btf_type`` is followed by ``info.vlen`` number of ``struct btf_enum64``.:: 539 540 struct btf_enum64 { 541 __u32 name_off; 542 __u32 val_lo32; 543 __u32 val_hi32; 544 }; 545 546The ``btf_enum64`` encoding: 547 * ``name_off``: offset to a valid C identifier 548 * ``val_lo32``: lower 32-bit value for a 64-bit value 549 * ``val_hi32``: high 32-bit value for a 64-bit value 550 551If the original enum value is signed and the size is less than 8, 552that value will be sign extended into 8 bytes. 553 5543. BTF Kernel API 555================= 556 557The following bpf syscall command involves BTF: 558 * BPF_BTF_LOAD: load a blob of BTF data into kernel 559 * BPF_MAP_CREATE: map creation with btf key and value type info. 560 * BPF_PROG_LOAD: prog load with btf function and line info. 561 * BPF_BTF_GET_FD_BY_ID: get a btf fd 562 * BPF_OBJ_GET_INFO_BY_FD: btf, func_info, line_info 563 and other btf related info are returned. 564 565The workflow typically looks like: 566:: 567 568 Application: 569 BPF_BTF_LOAD 570 | 571 v 572 BPF_MAP_CREATE and BPF_PROG_LOAD 573 | 574 V 575 ...... 576 577 Introspection tool: 578 ...... 579 BPF_{PROG,MAP}_GET_NEXT_ID (get prog/map id's) 580 | 581 V 582 BPF_{PROG,MAP}_GET_FD_BY_ID (get a prog/map fd) 583 | 584 V 585 BPF_OBJ_GET_INFO_BY_FD (get bpf_prog_info/bpf_map_info with btf_id) 586 | | 587 V | 588 BPF_BTF_GET_FD_BY_ID (get btf_fd) | 589 | | 590 V | 591 BPF_OBJ_GET_INFO_BY_FD (get btf) | 592 | | 593 V V 594 pretty print types, dump func signatures and line info, etc. 595 596 5973.1 BPF_BTF_LOAD 598---------------- 599 600Load a blob of BTF data into kernel. A blob of data, described in 601:ref:`BTF_Type_String`, can be directly loaded into the kernel. A ``btf_fd`` 602is returned to a userspace. 603 6043.2 BPF_MAP_CREATE 605------------------ 606 607A map can be created with ``btf_fd`` and specified key/value type id.:: 608 609 __u32 btf_fd; /* fd pointing to a BTF type data */ 610 __u32 btf_key_type_id; /* BTF type_id of the key */ 611 __u32 btf_value_type_id; /* BTF type_id of the value */ 612 613In libbpf, the map can be defined with extra annotation like below: 614:: 615 616 struct { 617 __uint(type, BPF_MAP_TYPE_ARRAY); 618 __type(key, int); 619 __type(value, struct ipv_counts); 620 __uint(max_entries, 4); 621 } btf_map SEC(".maps"); 622 623During ELF parsing, libbpf is able to extract key/value type_id's and assign 624them to BPF_MAP_CREATE attributes automatically. 625 626.. _BPF_Prog_Load: 627 6283.3 BPF_PROG_LOAD 629----------------- 630 631During prog_load, func_info and line_info can be passed to kernel with proper 632values for the following attributes: 633:: 634 635 __u32 insn_cnt; 636 __aligned_u64 insns; 637 ...... 638 __u32 prog_btf_fd; /* fd pointing to BTF type data */ 639 __u32 func_info_rec_size; /* userspace bpf_func_info size */ 640 __aligned_u64 func_info; /* func info */ 641 __u32 func_info_cnt; /* number of bpf_func_info records */ 642 __u32 line_info_rec_size; /* userspace bpf_line_info size */ 643 __aligned_u64 line_info; /* line info */ 644 __u32 line_info_cnt; /* number of bpf_line_info records */ 645 646The func_info and line_info are an array of below, respectively.:: 647 648 struct bpf_func_info { 649 __u32 insn_off; /* [0, insn_cnt - 1] */ 650 __u32 type_id; /* pointing to a BTF_KIND_FUNC type */ 651 }; 652 struct bpf_line_info { 653 __u32 insn_off; /* [0, insn_cnt - 1] */ 654 __u32 file_name_off; /* offset to string table for the filename */ 655 __u32 line_off; /* offset to string table for the source line */ 656 __u32 line_col; /* line number and column number */ 657 }; 658 659func_info_rec_size is the size of each func_info record, and 660line_info_rec_size is the size of each line_info record. Passing the record 661size to kernel make it possible to extend the record itself in the future. 662 663Below are requirements for func_info: 664 * func_info[0].insn_off must be 0. 665 * the func_info insn_off is in strictly increasing order and matches 666 bpf func boundaries. 667 668Below are requirements for line_info: 669 * the first insn in each func must have a line_info record pointing to it. 670 * the line_info insn_off is in strictly increasing order. 671 672For line_info, the line number and column number are defined as below: 673:: 674 675 #define BPF_LINE_INFO_LINE_NUM(line_col) ((line_col) >> 10) 676 #define BPF_LINE_INFO_LINE_COL(line_col) ((line_col) & 0x3ff) 677 6783.4 BPF_{PROG,MAP}_GET_NEXT_ID 679------------------------------ 680 681In kernel, every loaded program, map or btf has a unique id. The id won't 682change during the lifetime of a program, map, or btf. 683 684The bpf syscall command BPF_{PROG,MAP}_GET_NEXT_ID returns all id's, one for 685each command, to user space, for bpf program or maps, respectively, so an 686inspection tool can inspect all programs and maps. 687 6883.5 BPF_{PROG,MAP}_GET_FD_BY_ID 689------------------------------- 690 691An introspection tool cannot use id to get details about program or maps. 692A file descriptor needs to be obtained first for reference-counting purpose. 693 6943.6 BPF_OBJ_GET_INFO_BY_FD 695-------------------------- 696 697Once a program/map fd is acquired, an introspection tool can get the detailed 698information from kernel about this fd, some of which are BTF-related. For 699example, ``bpf_map_info`` returns ``btf_id`` and key/value type ids. 700``bpf_prog_info`` returns ``btf_id``, func_info, and line info for translated 701bpf byte codes, and jited_line_info. 702 7033.7 BPF_BTF_GET_FD_BY_ID 704------------------------ 705 706With ``btf_id`` obtained in ``bpf_map_info`` and ``bpf_prog_info``, bpf 707syscall command BPF_BTF_GET_FD_BY_ID can retrieve a btf fd. Then, with 708command BPF_OBJ_GET_INFO_BY_FD, the btf blob, originally loaded into the 709kernel with BPF_BTF_LOAD, can be retrieved. 710 711With the btf blob, ``bpf_map_info``, and ``bpf_prog_info``, an introspection 712tool has full btf knowledge and is able to pretty print map key/values, dump 713func signatures and line info, along with byte/jit codes. 714 7154. ELF File Format Interface 716============================ 717 7184.1 .BTF section 719---------------- 720 721The .BTF section contains type and string data. The format of this section is 722same as the one describe in :ref:`BTF_Type_String`. 723 724.. _BTF_Ext_Section: 725 7264.2 .BTF.ext section 727-------------------- 728 729The .BTF.ext section encodes func_info and line_info which needs loader 730manipulation before loading into the kernel. 731 732The specification for .BTF.ext section is defined at ``tools/lib/bpf/btf.h`` 733and ``tools/lib/bpf/btf.c``. 734 735The current header of .BTF.ext section:: 736 737 struct btf_ext_header { 738 __u16 magic; 739 __u8 version; 740 __u8 flags; 741 __u32 hdr_len; 742 743 /* All offsets are in bytes relative to the end of this header */ 744 __u32 func_info_off; 745 __u32 func_info_len; 746 __u32 line_info_off; 747 __u32 line_info_len; 748 }; 749 750It is very similar to .BTF section. Instead of type/string section, it 751contains func_info and line_info section. See :ref:`BPF_Prog_Load` for details 752about func_info and line_info record format. 753 754The func_info is organized as below.:: 755 756 func_info_rec_size 757 btf_ext_info_sec for section #1 /* func_info for section #1 */ 758 btf_ext_info_sec for section #2 /* func_info for section #2 */ 759 ... 760 761``func_info_rec_size`` specifies the size of ``bpf_func_info`` structure when 762.BTF.ext is generated. ``btf_ext_info_sec``, defined below, is a collection of 763func_info for each specific ELF section.:: 764 765 struct btf_ext_info_sec { 766 __u32 sec_name_off; /* offset to section name */ 767 __u32 num_info; 768 /* Followed by num_info * record_size number of bytes */ 769 __u8 data[0]; 770 }; 771 772Here, num_info must be greater than 0. 773 774The line_info is organized as below.:: 775 776 line_info_rec_size 777 btf_ext_info_sec for section #1 /* line_info for section #1 */ 778 btf_ext_info_sec for section #2 /* line_info for section #2 */ 779 ... 780 781``line_info_rec_size`` specifies the size of ``bpf_line_info`` structure when 782.BTF.ext is generated. 783 784The interpretation of ``bpf_func_info->insn_off`` and 785``bpf_line_info->insn_off`` is different between kernel API and ELF API. For 786kernel API, the ``insn_off`` is the instruction offset in the unit of ``struct 787bpf_insn``. For ELF API, the ``insn_off`` is the byte offset from the 788beginning of section (``btf_ext_info_sec->sec_name_off``). 789 7904.2 .BTF_ids section 791-------------------- 792 793The .BTF_ids section encodes BTF ID values that are used within the kernel. 794 795This section is created during the kernel compilation with the help of 796macros defined in ``include/linux/btf_ids.h`` header file. Kernel code can 797use them to create lists and sets (sorted lists) of BTF ID values. 798 799The ``BTF_ID_LIST`` and ``BTF_ID`` macros define unsorted list of BTF ID values, 800with following syntax:: 801 802 BTF_ID_LIST(list) 803 BTF_ID(type1, name1) 804 BTF_ID(type2, name2) 805 806resulting in following layout in .BTF_ids section:: 807 808 __BTF_ID__type1__name1__1: 809 .zero 4 810 __BTF_ID__type2__name2__2: 811 .zero 4 812 813The ``u32 list[];`` variable is defined to access the list. 814 815The ``BTF_ID_UNUSED`` macro defines 4 zero bytes. It's used when we 816want to define unused entry in BTF_ID_LIST, like:: 817 818 BTF_ID_LIST(bpf_skb_output_btf_ids) 819 BTF_ID(struct, sk_buff) 820 BTF_ID_UNUSED 821 BTF_ID(struct, task_struct) 822 823The ``BTF_SET_START/END`` macros pair defines sorted list of BTF ID values 824and their count, with following syntax:: 825 826 BTF_SET_START(set) 827 BTF_ID(type1, name1) 828 BTF_ID(type2, name2) 829 BTF_SET_END(set) 830 831resulting in following layout in .BTF_ids section:: 832 833 __BTF_ID__set__set: 834 .zero 4 835 __BTF_ID__type1__name1__3: 836 .zero 4 837 __BTF_ID__type2__name2__4: 838 .zero 4 839 840The ``struct btf_id_set set;`` variable is defined to access the list. 841 842The ``typeX`` name can be one of following:: 843 844 struct, union, typedef, func 845 846and is used as a filter when resolving the BTF ID value. 847 848All the BTF ID lists and sets are compiled in the .BTF_ids section and 849resolved during the linking phase of kernel build by ``resolve_btfids`` tool. 850 8515. Using BTF 852============ 853 8545.1 bpftool map pretty print 855---------------------------- 856 857With BTF, the map key/value can be printed based on fields rather than simply 858raw bytes. This is especially valuable for large structure or if your data 859structure has bitfields. For example, for the following map,:: 860 861 enum A { A1, A2, A3, A4, A5 }; 862 typedef enum A ___A; 863 struct tmp_t { 864 char a1:4; 865 int a2:4; 866 int :4; 867 __u32 a3:4; 868 int b; 869 ___A b1:4; 870 enum A b2:4; 871 }; 872 struct { 873 __uint(type, BPF_MAP_TYPE_ARRAY); 874 __type(key, int); 875 __type(value, struct tmp_t); 876 __uint(max_entries, 1); 877 } tmpmap SEC(".maps"); 878 879bpftool is able to pretty print like below: 880:: 881 882 [{ 883 "key": 0, 884 "value": { 885 "a1": 0x2, 886 "a2": 0x4, 887 "a3": 0x6, 888 "b": 7, 889 "b1": 0x8, 890 "b2": 0xa 891 } 892 } 893 ] 894 8955.2 bpftool prog dump 896--------------------- 897 898The following is an example showing how func_info and line_info can help prog 899dump with better kernel symbol names, function prototypes and line 900information.:: 901 902 $ bpftool prog dump jited pinned /sys/fs/bpf/test_btf_haskv 903 [...] 904 int test_long_fname_2(struct dummy_tracepoint_args * arg): 905 bpf_prog_44a040bf25481309_test_long_fname_2: 906 ; static int test_long_fname_2(struct dummy_tracepoint_args *arg) 907 0: push %rbp 908 1: mov %rsp,%rbp 909 4: sub $0x30,%rsp 910 b: sub $0x28,%rbp 911 f: mov %rbx,0x0(%rbp) 912 13: mov %r13,0x8(%rbp) 913 17: mov %r14,0x10(%rbp) 914 1b: mov %r15,0x18(%rbp) 915 1f: xor %eax,%eax 916 21: mov %rax,0x20(%rbp) 917 25: xor %esi,%esi 918 ; int key = 0; 919 27: mov %esi,-0x4(%rbp) 920 ; if (!arg->sock) 921 2a: mov 0x8(%rdi),%rdi 922 ; if (!arg->sock) 923 2e: cmp $0x0,%rdi 924 32: je 0x0000000000000070 925 34: mov %rbp,%rsi 926 ; counts = bpf_map_lookup_elem(&btf_map, &key); 927 [...] 928 9295.3 Verifier Log 930---------------- 931 932The following is an example of how line_info can help debugging verification 933failure.:: 934 935 /* The code at tools/testing/selftests/bpf/test_xdp_noinline.c 936 * is modified as below. 937 */ 938 data = (void *)(long)xdp->data; 939 data_end = (void *)(long)xdp->data_end; 940 /* 941 if (data + 4 > data_end) 942 return XDP_DROP; 943 */ 944 *(u32 *)data = dst->dst; 945 946 $ bpftool prog load ./test_xdp_noinline.o /sys/fs/bpf/test_xdp_noinline type xdp 947 ; data = (void *)(long)xdp->data; 948 224: (79) r2 = *(u64 *)(r10 -112) 949 225: (61) r2 = *(u32 *)(r2 +0) 950 ; *(u32 *)data = dst->dst; 951 226: (63) *(u32 *)(r2 +0) = r1 952 invalid access to packet, off=0 size=4, R2(id=0,off=0,r=0) 953 R2 offset is outside of the packet 954 9556. BTF Generation 956================= 957 958You need latest pahole 959 960 https://git.kernel.org/pub/scm/devel/pahole/pahole.git/ 961 962or llvm (8.0 or later). The pahole acts as a dwarf2btf converter. It doesn't 963support .BTF.ext and btf BTF_KIND_FUNC type yet. For example,:: 964 965 -bash-4.4$ cat t.c 966 struct t { 967 int a:2; 968 int b:3; 969 int c:2; 970 } g; 971 -bash-4.4$ gcc -c -O2 -g t.c 972 -bash-4.4$ pahole -JV t.o 973 File t.o: 974 [1] STRUCT t kind_flag=1 size=4 vlen=3 975 a type_id=2 bitfield_size=2 bits_offset=0 976 b type_id=2 bitfield_size=3 bits_offset=2 977 c type_id=2 bitfield_size=2 bits_offset=5 978 [2] INT int size=4 bit_offset=0 nr_bits=32 encoding=SIGNED 979 980The llvm is able to generate .BTF and .BTF.ext directly with -g for bpf target 981only. The assembly code (-S) is able to show the BTF encoding in assembly 982format.:: 983 984 -bash-4.4$ cat t2.c 985 typedef int __int32; 986 struct t2 { 987 int a2; 988 int (*f2)(char q1, __int32 q2, ...); 989 int (*f3)(); 990 } g2; 991 int main() { return 0; } 992 int test() { return 0; } 993 -bash-4.4$ clang -c -g -O2 -target bpf t2.c 994 -bash-4.4$ readelf -S t2.o 995 ...... 996 [ 8] .BTF PROGBITS 0000000000000000 00000247 997 000000000000016e 0000000000000000 0 0 1 998 [ 9] .BTF.ext PROGBITS 0000000000000000 000003b5 999 0000000000000060 0000000000000000 0 0 1 1000 [10] .rel.BTF.ext REL 0000000000000000 000007e0 1001 0000000000000040 0000000000000010 16 9 8 1002 ...... 1003 -bash-4.4$ clang -S -g -O2 -target bpf t2.c 1004 -bash-4.4$ cat t2.s 1005 ...... 1006 .section .BTF,"",@progbits 1007 .short 60319 # 0xeb9f 1008 .byte 1 1009 .byte 0 1010 .long 24 1011 .long 0 1012 .long 220 1013 .long 220 1014 .long 122 1015 .long 0 # BTF_KIND_FUNC_PROTO(id = 1) 1016 .long 218103808 # 0xd000000 1017 .long 2 1018 .long 83 # BTF_KIND_INT(id = 2) 1019 .long 16777216 # 0x1000000 1020 .long 4 1021 .long 16777248 # 0x1000020 1022 ...... 1023 .byte 0 # string offset=0 1024 .ascii ".text" # string offset=1 1025 .byte 0 1026 .ascii "/home/yhs/tmp-pahole/t2.c" # string offset=7 1027 .byte 0 1028 .ascii "int main() { return 0; }" # string offset=33 1029 .byte 0 1030 .ascii "int test() { return 0; }" # string offset=58 1031 .byte 0 1032 .ascii "int" # string offset=83 1033 ...... 1034 .section .BTF.ext,"",@progbits 1035 .short 60319 # 0xeb9f 1036 .byte 1 1037 .byte 0 1038 .long 24 1039 .long 0 1040 .long 28 1041 .long 28 1042 .long 44 1043 .long 8 # FuncInfo 1044 .long 1 # FuncInfo section string offset=1 1045 .long 2 1046 .long .Lfunc_begin0 1047 .long 3 1048 .long .Lfunc_begin1 1049 .long 5 1050 .long 16 # LineInfo 1051 .long 1 # LineInfo section string offset=1 1052 .long 2 1053 .long .Ltmp0 1054 .long 7 1055 .long 33 1056 .long 7182 # Line 7 Col 14 1057 .long .Ltmp3 1058 .long 7 1059 .long 58 1060 .long 8206 # Line 8 Col 14 1061 10627. Testing 1063========== 1064 1065Kernel bpf selftest `test_btf.c` provides extensive set of BTF-related tests. 1066