1119 lines · plain
1=====================2BPF Type Format (BTF)3=====================4 51. Introduction6===============7 8BTF (BPF Type Format) is the metadata format which encodes the debug info9related to BPF program/map. The name BTF was used initially to describe data10types. The BTF was later extended to include function info for defined11subroutines, and line info for source/line information.12 13The debug info is used for map pretty print, function signature, etc. The14function signature enables better bpf program/function kernel symbol. The line15info helps generate source annotated translated byte code, jited code and16verifier log.17 18The BTF specification contains two parts,19 * BTF kernel API20 * BTF ELF file format21 22The kernel API is the contract between user space and kernel. The kernel23verifies the BTF info before using it. The ELF file format is a user space24contract between ELF file and libbpf loader.25 26The type and string sections are part of the BTF kernel API, describing the27debug info (mostly types related) referenced by the bpf program. These two28sections are discussed in details in :ref:`BTF_Type_String`.29 30.. _BTF_Type_String:31 322. BTF Type and String Encoding33===============================34 35The file ``include/uapi/linux/btf.h`` provides high-level definition of how36types/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 little54endian systems, and can be used to test whether BTF is generated for big- or55little-endian target. The ``btf_header`` is designed to be extensible with56``hdr_len`` equal to ``sizeof(struct btf_header)`` when a data blob is57generated.58 592.1 String Encoding60-------------------61 62The first string in the string section must be a null string. The rest of63string table is a concatenation of other null-terminated strings.64 652.2 Type Encoding66-----------------67 68The type id ``0`` is reserved for ``void`` type. The type section is parsed69sequentially and type id is assigned to each recognized type starting from id70``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 arrangement100 * bits 0-15: vlen (e.g. # of struct's members)101 * bits 16-23: unused102 * bits 24-28: kind (e.g. int, ptr, array...etc)103 * bits 29-30: unused104 * bit 31: kind_flag, currently used by105 * 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. The122``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_INT126~~~~~~~~~~~~~~~~~~127 128``struct btf_type`` encoding requirement:129 * ``name_off``: any valid offset130 * ``info.kind_flag``: 0131 * ``info.kind``: BTF_KIND_INT132 * ``info.vlen``: 0133 * ``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, or148bool, for the int type. The char and bool encoding are mostly useful for149pretty 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 int152type. 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 values157for 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 starting164from bits ``100 + 2 = 102``.165 166Alternatively, the bitfield struct member can be the following to access the167same 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 of174bitfield encoding. Currently, both llvm and pahole generate175``BTF_INT_OFFSET() = 0`` for all int types.176 1772.2.2 BTF_KIND_PTR178~~~~~~~~~~~~~~~~~~179 180``struct btf_type`` encoding requirement:181 * ``name_off``: 0182 * ``info.kind_flag``: 0183 * ``info.kind``: BTF_KIND_PTR184 * ``info.vlen``: 0185 * ``type``: the pointee type of the pointer186 187No additional type data follow ``btf_type``.188 1892.2.3 BTF_KIND_ARRAY190~~~~~~~~~~~~~~~~~~~~191 192``struct btf_type`` encoding requirement:193 * ``name_off``: 0194 * ``info.kind_flag``: 0195 * ``info.kind``: BTF_KIND_ARRAY196 * ``info.vlen``: 0197 * ``size/type``: 0, not used198 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 type209 * ``index_type``: the index type210 * ``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 including214``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 represent218multidimensional arrays. For example, for ``int a[5][6]``, the following type219information illustrates the chaining:220 221 * [1]: int222 * [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 into226one-dimensional array, e.g., for ``a[5][6]``, the ``btf_array.nelems`` is227equal to ``30``. This is because the original use case is map pretty print228where the whole array is dumped out so one-dimensional array is enough. As229more BTF usage is explored, pahole and llvm can be changed to generate proper230chained representation for multidimensional arrays.231 2322.2.4 BTF_KIND_STRUCT233~~~~~~~~~~~~~~~~~~~~~2342.2.5 BTF_KIND_UNION235~~~~~~~~~~~~~~~~~~~~236 237``struct btf_type`` encoding requirement:238 * ``name_off``: 0 or offset to a valid C identifier239 * ``info.kind_flag``: 0 or 1240 * ``info.kind``: BTF_KIND_STRUCT or BTF_KIND_UNION241 * ``info.vlen``: the number of struct/union members242 * ``info.size``: the size of the struct/union in bytes243 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 identifier254 * ``type``: the member type255 * ``offset``: <see below>256 257If the type info ``kind_flag`` is not set, the offset contains only bit offset258of the member. Note that the base type of the bitfield can only be int or enum259type. If the bitfield size is 32, the base type can be either int or enum260type. If the bitfield size is not 32, the base type must be int, and int type261``BTF_INT_BITS()`` encodes the bitfield size.262 263If the ``kind_flag`` is set, the ``btf_member.offset`` contains both member264bitfield size and bit offset. The bitfield size and bit offset are calculated265as 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 modes276exist.277 2782.2.6 BTF_KIND_ENUM279~~~~~~~~~~~~~~~~~~~280 281``struct btf_type`` encoding requirement:282 * ``name_off``: 0 or offset to a valid C identifier283 * ``info.kind_flag``: 0 for unsigned, 1 for signed284 * ``info.kind``: BTF_KIND_ENUM285 * ``info.vlen``: number of enum values286 * ``size``: 1/2/4/8287 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 identifier297 * ``val``: any value298 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_FWD304~~~~~~~~~~~~~~~~~~305 306``struct btf_type`` encoding requirement:307 * ``name_off``: offset to a valid C identifier308 * ``info.kind_flag``: 0 for struct, 1 for union309 * ``info.kind``: BTF_KIND_FWD310 * ``info.vlen``: 0311 * ``type``: 0312 313No additional type data follow ``btf_type``.314 3152.2.8 BTF_KIND_TYPEDEF316~~~~~~~~~~~~~~~~~~~~~~317 318``struct btf_type`` encoding requirement:319 * ``name_off``: offset to a valid C identifier320 * ``info.kind_flag``: 0321 * ``info.kind``: BTF_KIND_TYPEDEF322 * ``info.vlen``: 0323 * ``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_VOLATILE328~~~~~~~~~~~~~~~~~~~~~~~329 330``struct btf_type`` encoding requirement:331 * ``name_off``: 0332 * ``info.kind_flag``: 0333 * ``info.kind``: BTF_KIND_VOLATILE334 * ``info.vlen``: 0335 * ``type``: the type with ``volatile`` qualifier336 337No additional type data follow ``btf_type``.338 3392.2.10 BTF_KIND_CONST340~~~~~~~~~~~~~~~~~~~~~341 342``struct btf_type`` encoding requirement:343 * ``name_off``: 0344 * ``info.kind_flag``: 0345 * ``info.kind``: BTF_KIND_CONST346 * ``info.vlen``: 0347 * ``type``: the type with ``const`` qualifier348 349No additional type data follow ``btf_type``.350 3512.2.11 BTF_KIND_RESTRICT352~~~~~~~~~~~~~~~~~~~~~~~~353 354``struct btf_type`` encoding requirement:355 * ``name_off``: 0356 * ``info.kind_flag``: 0357 * ``info.kind``: BTF_KIND_RESTRICT358 * ``info.vlen``: 0359 * ``type``: the type with ``restrict`` qualifier360 361No additional type data follow ``btf_type``.362 3632.2.12 BTF_KIND_FUNC364~~~~~~~~~~~~~~~~~~~~365 366``struct btf_type`` encoding requirement:367 * ``name_off``: offset to a valid C identifier368 * ``info.kind_flag``: 0369 * ``info.kind``: BTF_KIND_FUNC370 * ``info.vlen``: linkage information (BTF_FUNC_STATIC, BTF_FUNC_GLOBAL371 or BTF_FUNC_EXTERN - see :ref:`BTF_Function_Linkage_Constants`)372 * ``type``: a BTF_KIND_FUNC_PROTO type373 374No additional type data follow ``btf_type``.375 376A BTF_KIND_FUNC defines not a type, but a subprogram (function) whose377signature is defined by ``type``. The subprogram is thus an instance of that378type. The BTF_KIND_FUNC may in turn be referenced by a func_info in the379: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 are383supported in the kernel.384 3852.2.13 BTF_KIND_FUNC_PROTO386~~~~~~~~~~~~~~~~~~~~~~~~~~387 388``struct btf_type`` encoding requirement:389 * ``name_off``: 0390 * ``info.kind_flag``: 0391 * ``info.kind``: BTF_KIND_FUNC_PROTO392 * ``info.vlen``: # of parameters393 * ``type``: the return type394 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, then403``btf_param.name_off`` must point to a valid C identifier except for the404possible last argument representing the variable argument. The btf_param.type405refers to parameter type.406 407If the function has variable arguments, the last parameter is encoded with408``name_off = 0`` and ``type = 0``.409 4102.2.14 BTF_KIND_VAR411~~~~~~~~~~~~~~~~~~~412 413``struct btf_type`` encoding requirement:414 * ``name_off``: offset to a valid C identifier415 * ``info.kind_flag``: 0416 * ``info.kind``: BTF_KIND_VAR417 * ``info.vlen``: 0418 * ``type``: the type of the variable419 420``btf_type`` is followed by a single ``struct btf_variable`` with the421following data::422 423 struct btf_var {424 __u32 linkage;425 };426 427``btf_var.linkage`` may take the values: BTF_VAR_STATIC, BTF_VAR_GLOBAL_ALLOCATED or BTF_VAR_GLOBAL_EXTERN -428see :ref:`BTF_Var_Linkage_Constants`.429 430Not all type of global variables are supported by LLVM at this point.431The following is currently available:432 433 * static variables with or without section attributes434 * global variables with section attributes435 436The latter is for future extraction of map key/value type id's from a437map definition.438 4392.2.15 BTF_KIND_DATASEC440~~~~~~~~~~~~~~~~~~~~~~~441 442``struct btf_type`` encoding requirement:443 * ``name_off``: offset to a valid name associated with a variable or444 one of .data/.bss/.rodata445 * ``info.kind_flag``: 0446 * ``info.kind``: BTF_KIND_DATASEC447 * ``info.vlen``: # of variables448 * ``size``: total section size in bytes (0 at compilation time, patched449 to actual size by BPF loaders such as libbpf)450 451``btf_type`` is followed by ``info.vlen`` number of ``struct btf_var_secinfo``.::452 453 struct btf_var_secinfo {454 __u32 type;455 __u32 offset;456 __u32 size;457 };458 459``struct btf_var_secinfo`` encoding:460 * ``type``: the type of the BTF_KIND_VAR variable461 * ``offset``: the in-section offset of the variable462 * ``size``: the size of the variable in bytes463 4642.2.16 BTF_KIND_FLOAT465~~~~~~~~~~~~~~~~~~~~~466 467``struct btf_type`` encoding requirement:468 * ``name_off``: any valid offset469 * ``info.kind_flag``: 0470 * ``info.kind``: BTF_KIND_FLOAT471 * ``info.vlen``: 0472 * ``size``: the size of the float type in bytes: 2, 4, 8, 12 or 16.473 474No additional type data follow ``btf_type``.475 4762.2.17 BTF_KIND_DECL_TAG477~~~~~~~~~~~~~~~~~~~~~~~~478 479``struct btf_type`` encoding requirement:480 * ``name_off``: offset to a non-empty string481 * ``info.kind_flag``: 0482 * ``info.kind``: BTF_KIND_DECL_TAG483 * ``info.vlen``: 0484 * ``type``: ``struct``, ``union``, ``func``, ``var`` or ``typedef``485 486``btf_type`` is followed by ``struct btf_decl_tag``.::487 488 struct btf_decl_tag {489 __u32 component_idx;490 };491 492The ``name_off`` encodes btf_decl_tag attribute string.493The ``type`` should be ``struct``, ``union``, ``func``, ``var`` or ``typedef``.494For ``var`` or ``typedef`` type, ``btf_decl_tag.component_idx`` must be ``-1``.495For the other three types, if the btf_decl_tag attribute is496applied to the ``struct``, ``union`` or ``func`` itself,497``btf_decl_tag.component_idx`` must be ``-1``. Otherwise,498the attribute is applied to a ``struct``/``union`` member or499a ``func`` argument, and ``btf_decl_tag.component_idx`` should be a500valid index (starting from 0) pointing to a member or an argument.501 5022.2.18 BTF_KIND_TYPE_TAG503~~~~~~~~~~~~~~~~~~~~~~~~504 505``struct btf_type`` encoding requirement:506 * ``name_off``: offset to a non-empty string507 * ``info.kind_flag``: 0508 * ``info.kind``: BTF_KIND_TYPE_TAG509 * ``info.vlen``: 0510 * ``type``: the type with ``btf_type_tag`` attribute511 512Currently, ``BTF_KIND_TYPE_TAG`` is only emitted for pointer types.513It has the following btf type chain:514::515 516 ptr -> [type_tag]*517 -> [const | volatile | restrict | typedef]*518 -> base_type519 520Basically, a pointer type points to zero or more521type_tag, then zero or more const/volatile/restrict/typedef522and finally the base type. The base type is one of523int, ptr, array, struct, union, enum, func_proto and float types.524 5252.2.19 BTF_KIND_ENUM64526~~~~~~~~~~~~~~~~~~~~~~527 528``struct btf_type`` encoding requirement:529 * ``name_off``: 0 or offset to a valid C identifier530 * ``info.kind_flag``: 0 for unsigned, 1 for signed531 * ``info.kind``: BTF_KIND_ENUM64532 * ``info.vlen``: number of enum values533 * ``size``: 1/2/4/8534 535``btf_type`` is followed by ``info.vlen`` number of ``struct btf_enum64``.::536 537 struct btf_enum64 {538 __u32 name_off;539 __u32 val_lo32;540 __u32 val_hi32;541 };542 543The ``btf_enum64`` encoding:544 * ``name_off``: offset to a valid C identifier545 * ``val_lo32``: lower 32-bit value for a 64-bit value546 * ``val_hi32``: high 32-bit value for a 64-bit value547 548If the original enum value is signed and the size is less than 8,549that value will be sign extended into 8 bytes.550 5512.3 Constant Values552-------------------553 554.. _BTF_Function_Linkage_Constants:555 5562.3.1 Function Linkage Constant Values557~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~558.. table:: Function Linkage Values and Meanings559 560 =================== ===== ===========561 kind value description562 =================== ===== ===========563 ``BTF_FUNC_STATIC`` 0x0 definition of subprogram not visible outside containing compilation unit564 ``BTF_FUNC_GLOBAL`` 0x1 definition of subprogram visible outside containing compilation unit565 ``BTF_FUNC_EXTERN`` 0x2 declaration of a subprogram whose definition is outside the containing compilation unit566 =================== ===== ===========567 568 569.. _BTF_Var_Linkage_Constants:570 5712.3.2 Variable Linkage Constant Values572~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~573.. table:: Variable Linkage Values and Meanings574 575 ============================ ===== ===========576 kind value description577 ============================ ===== ===========578 ``BTF_VAR_STATIC`` 0x0 definition of global variable not visible outside containing compilation unit579 ``BTF_VAR_GLOBAL_ALLOCATED`` 0x1 definition of global variable visible outside containing compilation unit580 ``BTF_VAR_GLOBAL_EXTERN`` 0x2 declaration of global variable whose definition is outside the containing compilation unit581 ============================ ===== ===========582 5833. BTF Kernel API584=================585 586The following bpf syscall command involves BTF:587 * BPF_BTF_LOAD: load a blob of BTF data into kernel588 * BPF_MAP_CREATE: map creation with btf key and value type info.589 * BPF_PROG_LOAD: prog load with btf function and line info.590 * BPF_BTF_GET_FD_BY_ID: get a btf fd591 * BPF_OBJ_GET_INFO_BY_FD: btf, func_info, line_info592 and other btf related info are returned.593 594The workflow typically looks like:595::596 597 Application:598 BPF_BTF_LOAD599 |600 v601 BPF_MAP_CREATE and BPF_PROG_LOAD602 |603 V604 ......605 606 Introspection tool:607 ......608 BPF_{PROG,MAP}_GET_NEXT_ID (get prog/map id's)609 |610 V611 BPF_{PROG,MAP}_GET_FD_BY_ID (get a prog/map fd)612 |613 V614 BPF_OBJ_GET_INFO_BY_FD (get bpf_prog_info/bpf_map_info with btf_id)615 | |616 V |617 BPF_BTF_GET_FD_BY_ID (get btf_fd) |618 | |619 V |620 BPF_OBJ_GET_INFO_BY_FD (get btf) |621 | |622 V V623 pretty print types, dump func signatures and line info, etc.624 625 6263.1 BPF_BTF_LOAD627----------------628 629Load a blob of BTF data into kernel. A blob of data, described in630:ref:`BTF_Type_String`, can be directly loaded into the kernel. A ``btf_fd``631is returned to a userspace.632 6333.2 BPF_MAP_CREATE634------------------635 636A map can be created with ``btf_fd`` and specified key/value type id.::637 638 __u32 btf_fd; /* fd pointing to a BTF type data */639 __u32 btf_key_type_id; /* BTF type_id of the key */640 __u32 btf_value_type_id; /* BTF type_id of the value */641 642In libbpf, the map can be defined with extra annotation like below:643::644 645 struct {646 __uint(type, BPF_MAP_TYPE_ARRAY);647 __type(key, int);648 __type(value, struct ipv_counts);649 __uint(max_entries, 4);650 } btf_map SEC(".maps");651 652During ELF parsing, libbpf is able to extract key/value type_id's and assign653them to BPF_MAP_CREATE attributes automatically.654 655.. _BPF_Prog_Load:656 6573.3 BPF_PROG_LOAD658-----------------659 660During prog_load, func_info and line_info can be passed to kernel with proper661values for the following attributes:662::663 664 __u32 insn_cnt;665 __aligned_u64 insns;666 ......667 __u32 prog_btf_fd; /* fd pointing to BTF type data */668 __u32 func_info_rec_size; /* userspace bpf_func_info size */669 __aligned_u64 func_info; /* func info */670 __u32 func_info_cnt; /* number of bpf_func_info records */671 __u32 line_info_rec_size; /* userspace bpf_line_info size */672 __aligned_u64 line_info; /* line info */673 __u32 line_info_cnt; /* number of bpf_line_info records */674 675The func_info and line_info are an array of below, respectively.::676 677 struct bpf_func_info {678 __u32 insn_off; /* [0, insn_cnt - 1] */679 __u32 type_id; /* pointing to a BTF_KIND_FUNC type */680 };681 struct bpf_line_info {682 __u32 insn_off; /* [0, insn_cnt - 1] */683 __u32 file_name_off; /* offset to string table for the filename */684 __u32 line_off; /* offset to string table for the source line */685 __u32 line_col; /* line number and column number */686 };687 688func_info_rec_size is the size of each func_info record, and689line_info_rec_size is the size of each line_info record. Passing the record690size to kernel make it possible to extend the record itself in the future.691 692Below are requirements for func_info:693 * func_info[0].insn_off must be 0.694 * the func_info insn_off is in strictly increasing order and matches695 bpf func boundaries.696 697Below are requirements for line_info:698 * the first insn in each func must have a line_info record pointing to it.699 * the line_info insn_off is in strictly increasing order.700 701For line_info, the line number and column number are defined as below:702::703 704 #define BPF_LINE_INFO_LINE_NUM(line_col) ((line_col) >> 10)705 #define BPF_LINE_INFO_LINE_COL(line_col) ((line_col) & 0x3ff)706 7073.4 BPF_{PROG,MAP}_GET_NEXT_ID708------------------------------709 710In kernel, every loaded program, map or btf has a unique id. The id won't711change during the lifetime of a program, map, or btf.712 713The bpf syscall command BPF_{PROG,MAP}_GET_NEXT_ID returns all id's, one for714each command, to user space, for bpf program or maps, respectively, so an715inspection tool can inspect all programs and maps.716 7173.5 BPF_{PROG,MAP}_GET_FD_BY_ID718-------------------------------719 720An introspection tool cannot use id to get details about program or maps.721A file descriptor needs to be obtained first for reference-counting purpose.722 7233.6 BPF_OBJ_GET_INFO_BY_FD724--------------------------725 726Once a program/map fd is acquired, an introspection tool can get the detailed727information from kernel about this fd, some of which are BTF-related. For728example, ``bpf_map_info`` returns ``btf_id`` and key/value type ids.729``bpf_prog_info`` returns ``btf_id``, func_info, and line info for translated730bpf byte codes, and jited_line_info.731 7323.7 BPF_BTF_GET_FD_BY_ID733------------------------734 735With ``btf_id`` obtained in ``bpf_map_info`` and ``bpf_prog_info``, bpf736syscall command BPF_BTF_GET_FD_BY_ID can retrieve a btf fd. Then, with737command BPF_OBJ_GET_INFO_BY_FD, the btf blob, originally loaded into the738kernel with BPF_BTF_LOAD, can be retrieved.739 740With the btf blob, ``bpf_map_info``, and ``bpf_prog_info``, an introspection741tool has full btf knowledge and is able to pretty print map key/values, dump742func signatures and line info, along with byte/jit codes.743 7444. ELF File Format Interface745============================746 7474.1 .BTF section748----------------749 750The .BTF section contains type and string data. The format of this section is751same as the one describe in :ref:`BTF_Type_String`.752 753.. _BTF_Ext_Section:754 7554.2 .BTF.ext section756--------------------757 758The .BTF.ext section encodes func_info, line_info and CO-RE relocations759which needs loader manipulation before loading into the kernel.760 761The specification for .BTF.ext section is defined at ``tools/lib/bpf/btf.h``762and ``tools/lib/bpf/btf.c``.763 764The current header of .BTF.ext section::765 766 struct btf_ext_header {767 __u16 magic;768 __u8 version;769 __u8 flags;770 __u32 hdr_len;771 772 /* All offsets are in bytes relative to the end of this header */773 __u32 func_info_off;774 __u32 func_info_len;775 __u32 line_info_off;776 __u32 line_info_len;777 778 /* optional part of .BTF.ext header */779 __u32 core_relo_off;780 __u32 core_relo_len;781 };782 783It is very similar to .BTF section. Instead of type/string section, it784contains func_info, line_info and core_relo sub-sections.785See :ref:`BPF_Prog_Load` for details about func_info and line_info786record format.787 788The func_info is organized as below.::789 790 func_info_rec_size /* __u32 value */791 btf_ext_info_sec for section #1 /* func_info for section #1 */792 btf_ext_info_sec for section #2 /* func_info for section #2 */793 ...794 795``func_info_rec_size`` specifies the size of ``bpf_func_info`` structure when796.BTF.ext is generated. ``btf_ext_info_sec``, defined below, is a collection of797func_info for each specific ELF section.::798 799 struct btf_ext_info_sec {800 __u32 sec_name_off; /* offset to section name */801 __u32 num_info;802 /* Followed by num_info * record_size number of bytes */803 __u8 data[0];804 };805 806Here, num_info must be greater than 0.807 808The line_info is organized as below.::809 810 line_info_rec_size /* __u32 value */811 btf_ext_info_sec for section #1 /* line_info for section #1 */812 btf_ext_info_sec for section #2 /* line_info for section #2 */813 ...814 815``line_info_rec_size`` specifies the size of ``bpf_line_info`` structure when816.BTF.ext is generated.817 818The interpretation of ``bpf_func_info->insn_off`` and819``bpf_line_info->insn_off`` is different between kernel API and ELF API. For820kernel API, the ``insn_off`` is the instruction offset in the unit of ``struct821bpf_insn``. For ELF API, the ``insn_off`` is the byte offset from the822beginning of section (``btf_ext_info_sec->sec_name_off``).823 824The core_relo is organized as below.::825 826 core_relo_rec_size /* __u32 value */827 btf_ext_info_sec for section #1 /* core_relo for section #1 */828 btf_ext_info_sec for section #2 /* core_relo for section #2 */829 830``core_relo_rec_size`` specifies the size of ``bpf_core_relo``831structure when .BTF.ext is generated. All ``bpf_core_relo`` structures832within a single ``btf_ext_info_sec`` describe relocations applied to833section named by ``btf_ext_info_sec->sec_name_off``.834 835See :ref:`Documentation/bpf/llvm_reloc.rst <btf-co-re-relocations>`836for more information on CO-RE relocations.837 8384.2 .BTF_ids section839--------------------840 841The .BTF_ids section encodes BTF ID values that are used within the kernel.842 843This section is created during the kernel compilation with the help of844macros defined in ``include/linux/btf_ids.h`` header file. Kernel code can845use them to create lists and sets (sorted lists) of BTF ID values.846 847The ``BTF_ID_LIST`` and ``BTF_ID`` macros define unsorted list of BTF ID values,848with following syntax::849 850 BTF_ID_LIST(list)851 BTF_ID(type1, name1)852 BTF_ID(type2, name2)853 854resulting in following layout in .BTF_ids section::855 856 __BTF_ID__type1__name1__1:857 .zero 4858 __BTF_ID__type2__name2__2:859 .zero 4860 861The ``u32 list[];`` variable is defined to access the list.862 863The ``BTF_ID_UNUSED`` macro defines 4 zero bytes. It's used when we864want to define unused entry in BTF_ID_LIST, like::865 866 BTF_ID_LIST(bpf_skb_output_btf_ids)867 BTF_ID(struct, sk_buff)868 BTF_ID_UNUSED869 BTF_ID(struct, task_struct)870 871The ``BTF_SET_START/END`` macros pair defines sorted list of BTF ID values872and their count, with following syntax::873 874 BTF_SET_START(set)875 BTF_ID(type1, name1)876 BTF_ID(type2, name2)877 BTF_SET_END(set)878 879resulting in following layout in .BTF_ids section::880 881 __BTF_ID__set__set:882 .zero 4883 __BTF_ID__type1__name1__3:884 .zero 4885 __BTF_ID__type2__name2__4:886 .zero 4887 888The ``struct btf_id_set set;`` variable is defined to access the list.889 890The ``typeX`` name can be one of following::891 892 struct, union, typedef, func893 894and is used as a filter when resolving the BTF ID value.895 896All the BTF ID lists and sets are compiled in the .BTF_ids section and897resolved during the linking phase of kernel build by ``resolve_btfids`` tool.898 8995. Using BTF900============901 9025.1 bpftool map pretty print903----------------------------904 905With BTF, the map key/value can be printed based on fields rather than simply906raw bytes. This is especially valuable for large structure or if your data907structure has bitfields. For example, for the following map,::908 909 enum A { A1, A2, A3, A4, A5 };910 typedef enum A ___A;911 struct tmp_t {912 char a1:4;913 int a2:4;914 int :4;915 __u32 a3:4;916 int b;917 ___A b1:4;918 enum A b2:4;919 };920 struct {921 __uint(type, BPF_MAP_TYPE_ARRAY);922 __type(key, int);923 __type(value, struct tmp_t);924 __uint(max_entries, 1);925 } tmpmap SEC(".maps");926 927bpftool is able to pretty print like below:928::929 930 [{931 "key": 0,932 "value": {933 "a1": 0x2,934 "a2": 0x4,935 "a3": 0x6,936 "b": 7,937 "b1": 0x8,938 "b2": 0xa939 }940 }941 ]942 9435.2 bpftool prog dump944---------------------945 946The following is an example showing how func_info and line_info can help prog947dump with better kernel symbol names, function prototypes and line948information.::949 950 $ bpftool prog dump jited pinned /sys/fs/bpf/test_btf_haskv951 [...]952 int test_long_fname_2(struct dummy_tracepoint_args * arg):953 bpf_prog_44a040bf25481309_test_long_fname_2:954 ; static int test_long_fname_2(struct dummy_tracepoint_args *arg)955 0: push %rbp956 1: mov %rsp,%rbp957 4: sub $0x30,%rsp958 b: sub $0x28,%rbp959 f: mov %rbx,0x0(%rbp)960 13: mov %r13,0x8(%rbp)961 17: mov %r14,0x10(%rbp)962 1b: mov %r15,0x18(%rbp)963 1f: xor %eax,%eax964 21: mov %rax,0x20(%rbp)965 25: xor %esi,%esi966 ; int key = 0;967 27: mov %esi,-0x4(%rbp)968 ; if (!arg->sock)969 2a: mov 0x8(%rdi),%rdi970 ; if (!arg->sock)971 2e: cmp $0x0,%rdi972 32: je 0x0000000000000070973 34: mov %rbp,%rsi974 ; counts = bpf_map_lookup_elem(&btf_map, &key);975 [...]976 9775.3 Verifier Log978----------------979 980The following is an example of how line_info can help debugging verification981failure.::982 983 /* The code at tools/testing/selftests/bpf/test_xdp_noinline.c984 * is modified as below.985 */986 data = (void *)(long)xdp->data;987 data_end = (void *)(long)xdp->data_end;988 /*989 if (data + 4 > data_end)990 return XDP_DROP;991 */992 *(u32 *)data = dst->dst;993 994 $ bpftool prog load ./test_xdp_noinline.o /sys/fs/bpf/test_xdp_noinline type xdp995 ; data = (void *)(long)xdp->data;996 224: (79) r2 = *(u64 *)(r10 -112)997 225: (61) r2 = *(u32 *)(r2 +0)998 ; *(u32 *)data = dst->dst;999 226: (63) *(u32 *)(r2 +0) = r11000 invalid access to packet, off=0 size=4, R2(id=0,off=0,r=0)1001 R2 offset is outside of the packet1002 10036. BTF Generation1004=================1005 1006You need latest pahole1007 1008 https://git.kernel.org/pub/scm/devel/pahole/pahole.git/1009 1010or llvm (8.0 or later). The pahole acts as a dwarf2btf converter. It doesn't1011support .BTF.ext and btf BTF_KIND_FUNC type yet. For example,::1012 1013 -bash-4.4$ cat t.c1014 struct t {1015 int a:2;1016 int b:3;1017 int c:2;1018 } g;1019 -bash-4.4$ gcc -c -O2 -g t.c1020 -bash-4.4$ pahole -JV t.o1021 File t.o:1022 [1] STRUCT t kind_flag=1 size=4 vlen=31023 a type_id=2 bitfield_size=2 bits_offset=01024 b type_id=2 bitfield_size=3 bits_offset=21025 c type_id=2 bitfield_size=2 bits_offset=51026 [2] INT int size=4 bit_offset=0 nr_bits=32 encoding=SIGNED1027 1028The llvm is able to generate .BTF and .BTF.ext directly with -g for bpf target1029only. The assembly code (-S) is able to show the BTF encoding in assembly1030format.::1031 1032 -bash-4.4$ cat t2.c1033 typedef int __int32;1034 struct t2 {1035 int a2;1036 int (*f2)(char q1, __int32 q2, ...);1037 int (*f3)();1038 } g2;1039 int main() { return 0; }1040 int test() { return 0; }1041 -bash-4.4$ clang -c -g -O2 --target=bpf t2.c1042 -bash-4.4$ readelf -S t2.o1043 ......1044 [ 8] .BTF PROGBITS 0000000000000000 000002471045 000000000000016e 0000000000000000 0 0 11046 [ 9] .BTF.ext PROGBITS 0000000000000000 000003b51047 0000000000000060 0000000000000000 0 0 11048 [10] .rel.BTF.ext REL 0000000000000000 000007e01049 0000000000000040 0000000000000010 16 9 81050 ......1051 -bash-4.4$ clang -S -g -O2 --target=bpf t2.c1052 -bash-4.4$ cat t2.s1053 ......1054 .section .BTF,"",@progbits1055 .short 60319 # 0xeb9f1056 .byte 11057 .byte 01058 .long 241059 .long 01060 .long 2201061 .long 2201062 .long 1221063 .long 0 # BTF_KIND_FUNC_PROTO(id = 1)1064 .long 218103808 # 0xd0000001065 .long 21066 .long 83 # BTF_KIND_INT(id = 2)1067 .long 16777216 # 0x10000001068 .long 41069 .long 16777248 # 0x10000201070 ......1071 .byte 0 # string offset=01072 .ascii ".text" # string offset=11073 .byte 01074 .ascii "/home/yhs/tmp-pahole/t2.c" # string offset=71075 .byte 01076 .ascii "int main() { return 0; }" # string offset=331077 .byte 01078 .ascii "int test() { return 0; }" # string offset=581079 .byte 01080 .ascii "int" # string offset=831081 ......1082 .section .BTF.ext,"",@progbits1083 .short 60319 # 0xeb9f1084 .byte 11085 .byte 01086 .long 241087 .long 01088 .long 281089 .long 281090 .long 441091 .long 8 # FuncInfo1092 .long 1 # FuncInfo section string offset=11093 .long 21094 .long .Lfunc_begin01095 .long 31096 .long .Lfunc_begin11097 .long 51098 .long 16 # LineInfo1099 .long 1 # LineInfo section string offset=11100 .long 21101 .long .Ltmp01102 .long 71103 .long 331104 .long 7182 # Line 7 Col 141105 .long .Ltmp31106 .long 71107 .long 581108 .long 8206 # Line 8 Col 141109 11107. Testing1111==========1112 1113The kernel BPF selftest `tools/testing/selftests/bpf/prog_tests/btf.c`_1114provides an extensive set of BTF-related tests.1115 1116.. Links1117.. _tools/testing/selftests/bpf/prog_tests/btf.c:1118 https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/tools/testing/selftests/bpf/prog_tests/btf.c1119