.\" .\" This file and its contents are supplied under the terms of the .\" Common Development and Distribution License ("CDDL"), version 1.0. .\" You may only use this file in accordance with the terms of version .\" 1.0 of the CDDL. .\" .\" A full copy of the text of the CDDL should have accompanied this .\" source. A copy of the CDDL is also available via the Internet at .\" http://www.illumos.org/license/CDDL. .\" .\" .\" Copyright (c) 2014 Joyent, Inc. .\" .Dd Sep 26, 2014 .Dt CTF 5 .Os .Sh NAME .Nm ctf .Nd Compact C Type Format .Sh SYNOPSIS .In sys/ctf.h .Sh DESCRIPTION .Nm is designed to be a compact representation of the C programming language's type information focused on serving the needs of dynamic tracing, debuggers, and other in-situ and post-mortem introspection tools. .Nm data is generally included in .Sy ELF objects and is tagged as .Sy SHT_PROGBITS to ensure that the data is accessible in a running process and in subsequent core dumps, if generated. .Lp The .Nm data contained in each file has information about the layout and sizes of C types, including intrinsic types, enumerations, structures, typedefs, and unions, that are used by the corresponding .Sy ELF object. The .Nm data may also include information about the types of global objects and the return type and arguments of functions in the symbol table. .Lp Because a .Nm file is often embedded inside a file, rather than being a standalone file itself, it may also be referred to as a .Nm .Sy container . .Lp On .Fx systems, .Nm data is consumed by .Xr dtrace 1 . Programmatic access to .Nm data can be obtained through libctf. .Lp The .Nm file format is broken down into seven different sections. The first section is the .Sy preamble and .Sy header , which describes the version of the .Nm file, the links it has to other .Nm files, and the sizes of the other sections. The next section is the .Sy label section, which provides a way of identifying similar groups of .Nm data across multiple files. This is followed by the .Sy object information section, which describes the types of global symbols. The subsequent section is the .Sy function information section, which describes the return types and arguments of functions. The next section is the .Sy type information section, which describes the format and layout of the C types themselves, and finally the last section is the .Sy string section, which contains the names of types, enumerations, members, and labels. .Lp While strictly speaking, only the .Sy preamble and .Sy header are required, to be actually useful, both the type and string sections are necessary. .Lp A .Nm file may contain all of the type information that it requires, or it may optionally refer to another .Nm file which holds the remaining types. When a .Nm file refers to another file, it is called the .Sy child and the file it refers to is called the .Sy parent . A given file may only refer to one parent. This process is called .Em uniquification because it ensures each child only has type information that is unique to it. A common example of this is that most kernel modules in illumos are uniquified against the kernel module .Sy genunix and the type information that comes from the .Sy IP module. This means that a module only has types that are unique to itself and the most common types in the kernel are not duplicated. .Sh FILE FORMAT This documents version .Em two of the .Nm file format. All applications and tools on .Fx currently produce and operate on this version. .Lp The file format can be summarized with the following image, the following sections will cover this in more detail. .Bd -literal +-------------+ 0t0 +--------| Preamble | | +-------------+ 0t4 |+-------| Header | || +-------------+ 0t36 + cth_lbloff ||+------| Labels | ||| +-------------+ 0t36 + cth_objtoff |||+-----| Objects | |||| +-------------+ 0t36 + cth_funcoff ||||+----| Functions | ||||| +-------------+ 0t36 + cth_typeoff |||||+---| Types | |||||| +-------------+ 0t36 + cth_stroff ||||||+--| Strings | ||||||| +-------------+ 0t36 + cth_stroff + cth_strlen ||||||| ||||||| ||||||| ||||||| +-- magic - vers flags ||||||| | | | | ||||||| +------+------+------+------+ +---------| 0xcf | 0xf1 | 0x02 | 0x00 | |||||| +------+------+------+------+ |||||| 0 1 2 3 4 |||||| |||||| + parent label + objects |||||| | + parent name | + functions + strings |||||| | | + label | | + types | + strlen |||||| | | | | | | | | |||||| +------+------+------+------+------+-------+-------+-------+ +--------| 0x00 | 0x00 | 0x00 | 0x08 | 0x36 | 0x110 | 0x5f4 | 0x611 | ||||| +------+------+------+------+------+-------+-------+-------+ ||||| 0x04 0x08 0x0c 0x10 0x14 0x18 0x1c 0x20 0x24 ||||| ||||| + Label name ||||| | + Label type ||||| | | + Next label ||||| | | | ||||| +-------+------+-----+ +-----------| 0x01 | 0x42 | ... | |||| +-------+------+-----+ |||| cth_lbloff +0x4 +0x8 cth_objtoff |||| |||| |||| Symidx 0t15 0t43 0t44 |||| +------+------+------+-----+ +----------| 0x00 | 0x42 | 0x36 | ... | ||| +------+------+------+-----+ ||| cth_objtoff +0x2 +0x4 +0x6 cth_funcoff ||| ||| + CTF_TYPE_INFO + CTF_TYPE_INFO ||| | + Return type | ||| | | + arg0 | ||| +--------+------+------+-----+ +---------| 0x2c10 | 0x08 | 0x0c | ... | || +--------+------+------+-----+ || cth_funcff +0x2 +0x4 +0x6 cth_typeoff || || + ctf_stype_t for type 1 || | integer + integer encoding || | | + ctf_stype_t for type 2 || | | | || +--------------------+-----------+-----+ +--------| 0x19 * 0xc01 * 0x0 | 0x1000000 | ... | | +--------------------+-----------+-----+ | cth_typeoff +0x08 +0x0c cth_stroff | | +--- str 0 | | +--- str 1 + str 2 | | | | | v v v | +----+---+---+---+----+---+---+---+---+---+----+ +---| \\0 | i | n | t | \\0 | f | o | o | _ | t | \\0 | +----+---+---+---+----+---+---+---+---+---+----+ 0 1 2 3 4 5 6 7 8 9 10 11 .Ed .Lp Every .Nm file begins with a .Sy preamble , followed by a .Sy header . The .Sy preamble is defined as follows: .Bd -literal typedef struct ctf_preamble { ushort_t ctp_magic; /* magic number (CTF_MAGIC) */ uchar_t ctp_version; /* data format version number (CTF_VERSION) */ uchar_t ctp_flags; /* flags (see below) */ } ctf_preamble_t; .Ed .Pp The .Sy preamble is four bytes long and must be four byte aligned. This .Sy preamble defines the version of the .Nm file which defines the format of the rest of the header. While the header may change in subsequent versions, the preamble will not change across versions, though the interpretation of its flags may change from version to version. The .Em ctp_magic member defines the magic number for the .Nm file format. This must always be .Li 0xcff1 . If another value is encountered, then the file should not be treated as a .Nm file. The .Em ctp_version member defines the version of the .Nm file. The current version is .Li 2 . It is possible to encounter an unsupported version. In that case, software should not try to parse the format, as it may have changed. Finally, the .Em ctp_flags member describes aspects of the file which modify its interpretation. The following flags are currently defined: .Bd -literal #define CTF_F_COMPRESS 0x01 .Ed .Pp The flag .Sy CTF_F_COMPRESS indicates that the body of the .Nm file, all the data following the .Sy header , has been compressed through the .Sy zlib library and its .Sy deflate algorithm. If this flag is not present, then the body has not been compressed and no special action is needed to interpret it. All offsets into the data as described by .Sy header , always refer to the .Sy uncompressed data. .Lp In version two of the .Nm file format, the .Sy header denotes whether or not this .Nm file is the child of another .Nm file and also indicates the size of the remaining sections. The structure for the .Sy header logically contains a copy of the .Sy preamble and the two have a combined size of 36 bytes. .Bd -literal typedef struct ctf_header { ctf_preamble_t cth_preamble; uint_t cth_parlabel; /* ref to name of parent lbl uniq'd against */ uint_t cth_parname; /* ref to basename of parent */ uint_t cth_lbloff; /* offset of label section */ uint_t cth_objtoff; /* offset of object section */ uint_t cth_funcoff; /* offset of function section */ uint_t cth_typeoff; /* offset of type section */ uint_t cth_stroff; /* offset of string section */ uint_t cth_strlen; /* length of string section in bytes */ } ctf_header_t; .Ed .Pp After the .Sy preamble , the next two members .Em cth_parlablel and .Em cth_parname , are used to identify the parent. The value of both members are offsets into the .Sy string section which point to the start of a null-terminated string. For more information on the encoding of strings, see the subsection on .Sx String Identifiers . If the value of either is zero, then there is no entry for that member. If the member .Em cth_parlabel is set, then the .Em ctf_parname member must be set, otherwise it will not be possible to find the parent. If .Em ctf_parname is set, it is not necessary to define .Em cth_parlabel , as the parent may not have a label. For more information on labels and their interpretation, see .Sx The Label Section . .Lp The remaining members (excepting .Em cth_strlen ) describe the beginning of the corresponding sections. These offsets are relative to the end of the .Sy header . Therefore, something with an offset of 0 is at an offset of thirty-six bytes relative to the start of the .Nm file. The difference between members indicates the size of the section itself. Different offsets have different alignment requirements. The start of the .Em cth_objotoff and .Em cth_funcoff must be two byte aligned, while the sections .Em cth_lbloff and .Em cth_typeoff must be four-byte aligned. The section .Em cth_stroff has no alignment requirements. To calculate the size of a given section, excepting the .Sy string section, one should subtract the offset of the section from the following one. For example, the size of the .Sy types section can be calculated by subtracting .Em cth_stroff from .Em cth_typeoff . .Lp Finally, the member .Em cth_strlen describes the length of the string section itself. From it, you can also calculate the size of the entire .Nm file by adding together the size of the .Sy ctf_header_t , the offset of the string section in .Em cth_stroff , and the size of the string section in .Em cth_srlen . .Ss Type Identifiers Through the .Nm ctf data, types are referred to by identifiers. A given .Nm file supports up to 32767 (0x7fff) types. The first valid type identifier is 0x1. When a given .Nm file is a child, indicated by a non-zero entry for the .Sy header Ns 's .Em cth_parname , then the first valid type identifier is 0x8000 and the last is 0xffff. In this case, type identifiers 0x1 through 0x7fff are references to the parent. .Lp The type identifier zero is a sentinel value used to indicate that there is no type information available or it is an unknown type. .Lp Throughout the file format, the identifier is stored in different sized values; however, the minimum size to represent a given identifier is a .Sy uint16_t . Other consumers of .Nm information may use larger or opaque identifiers. .Ss String Identifiers String identifiers are always encoded as four byte unsigned integers which are an offset into a string table. The .Nm format supports two different string tables which have an identifier of zero or one. This identifier is stored in the high-order bit of the unsigned four byte offset. Therefore, the maximum supported offset into one of these tables is 0x7ffffffff. .Lp Table identifier zero, always refers to the .Sy string section in the CTF file itself. String table identifier one refers to an external string table which is the ELF string table for the ELF symbol table associated with the .Nm container. .Ss Type Encoding Every .Nm type begins with metadata encoded into a .Sy uint16_t . This encoded information tells us three different pieces of information: .Bl -bullet -offset indent -compact .It The kind of the type .It Whether this type is a root type or not .It The length of the variable data .El .Lp The 16 bits that make up the encoding are broken down such that you have five bits for the kind, one bit for indicating whether or not it is a root type, and 10 bits for the variable length. This is laid out as follows: .Bd -literal -offset indent +--------------------+ | kind | root | vlen | +--------------------+ 15 11 10 9 0 .Ed .Lp The current version of the file format defines 14 different kinds. The interpretation of these different kinds will be discussed in the section .Sx The Type Section . If a kind is encountered that is not listed below, then it is not a valid .Nm file. The kinds are defined as follows: .Bd -literal -offset indent #define CTF_K_UNKNOWN 0 #define CTF_K_INTEGER 1 #define CTF_K_FLOAT 2 #define CTF_K_POINTER 3 #define CTF_K_ARRAY 4 #define CTF_K_FUNCTION 5 #define CTF_K_STRUCT 6 #define CTF_K_UNION 7 #define CTF_K_ENUM 8 #define CTF_K_FORWARD 9 #define CTF_K_TYPEDEF 10 #define CTF_K_VOLATILE 11 #define CTF_K_CONST 12 #define CTF_K_RESTRICT 13 .Ed .Lp Programs directly reference many types; however, other types are referenced indirectly because they are part of some other structure. These types that are referenced directly and used are called .Sy root types. Other types may be used indirectly, for example, a program may reference a structure directly, but not one of its members which has a type. That type is not considered a .Sy root type. If a type is a .Sy root type, then it will have bit 10 set. .Lp The variable length section is specific to each kind and is discussed in the section .Sx The Type Section . .Lp The following macros are useful for constructing and deconstructing the encoded type information: .Bd -literal -offset indent #define CTF_MAX_VLEN 0x3ff #define CTF_INFO_KIND(info) (((info) & 0xf800) >> 11) #define CTF_INFO_ISROOT(info) (((info) & 0x0400) >> 10) #define CTF_INFO_VLEN(info) (((info) & CTF_MAX_VLEN)) #define CTF_TYPE_INFO(kind, isroot, vlen) \\ (((kind) << 11) | (((isroot) ? 1 : 0) << 10) | ((vlen) & CTF_MAX_VLEN)) .Ed .Ss The Label Section When consuming .Nm data, it is often useful to know whether two different .Nm containers come from the same source base and version. For example, when building illumos, there are many kernel modules that are built against a single collection of source code. A label is encoded into the .Nm files that corresponds with the particular build. This ensures that if files on the system were to become mixed up from multiple releases, that they are not used together by tools, particularly when a child needs to refer to a type in the parent. Because they are linked using the type identifiers, if the wrong parent is used then the wrong type will be encountered. .Lp Each label is encoded in the file format using the following eight byte structure: .Bd -literal typedef struct ctf_lblent { uint_t ctl_label; /* ref to name of label */ uint_t ctl_typeidx; /* last type associated with this label */ } ctf_lblent_t; .Ed .Lp Each label has two different components, a name and a type identifier. The name is encoded in the .Em ctl_label member which is in the format defined in the section .Sx String Identifiers . Generally, the names of all labels are found in the internal string section. .Lp The type identifier encoded in the member .Em ctl_typeidx refers to the last type identifier that a label refers to in the current file. Labels only refer to types in the current file, if the .Nm file is a child, then it will have the same label as its parent; however, its label will only refer to its types, not its parent's. .Lp It is also possible, though rather uncommon, for a .Nm file to have multiple labels. Labels are placed one after another, every eight bytes. When multiple labels are present, types may only belong to a single label. .Ss The Object Section The object section provides a mapping from ELF symbols of type .Sy STT_OBJECT in the symbol table to a type identifier. Every entry in this section is a .Sy uint16_t which contains a type identifier as described in the section .Sx Type Identifiers . If there is no information for an object, then the type identifier 0x0 is stored for that entry. .Lp To walk the object section, you need to have a corresponding .Sy symbol table in the ELF object that contains the .Nm data. Not every object is included in this section. Specifically, when walking the symbol table, an entry is skipped if it matches any of the following conditions: .Lp .Bl -bullet -offset indent -compact .It The symbol type is not .Sy STT_OBJECT .It The symbol's section index is .Sy SHN_UNDEF .It The symbol's name offset is zero .It The symbol's section index is .Sy SHN_ABS and the value of the symbol is zero. .It The symbol's name is .Li _START_ or .Li _END_ . These are skipped because they are used for scoping local symbols in ELF. .El .Lp The following sample code shows an example of iterating the object section and skipping the correct symbols: .Bd -literal #include #include /* * Given the start of the object section in the CTF file, the number of symbols, * and the ELF Data sections for the symbol table and the string table, this * prints the type identifiers that correspond to objects. Note, a more robust * implementation should ensure that they don't walk beyond the end of the CTF * object section. */ static int walk_symbols(uint16_t *objtoff, Elf_Data *symdata, Elf_Data *strdata, long nsyms) { long i; uintptr_t strbase = strdata->d_buf; for (i = 1; i < nsyms; i++, objftoff++) { const char *name; GElf_Sym sym; if (gelf_getsym(symdata, i, &sym) == NULL) return (1); if (GELF_ST_TYPE(sym.st_info) != STT_OBJECT) continue; if (sym.st_shndx == SHN_UNDEF || sym.st_name == 0) continue; if (sym.st_shndx == SHN_ABS && sym.st_value == 0) continue; name = (const char *)(strbase + sym.st_name); if (strcmp(name, "_START_") == 0 || strcmp(name, "_END_") == 0) continue; (void) printf("Symbol %d has type %d\n", i, *objtoff); } return (0); } .Ed .Ss The Function Section The function section of the .Nm file encodes the types of both the function's arguments and the function's return value. Similar to .Sx The Object Section , the function section encodes information for all symbols of type .Sy STT_FUNCTION , excepting those that fit specific criteria. Unlike with objects, because functions have a variable number of arguments, they start with a type encoding as defined in .Sx Type Encoding , which is the size of a .Sy uint16_t . For functions which have no type information available, they are encoded as .Li CTF_TYPE_INFO(CTF_K_UNKNOWN, 0, 0) . Functions with arguments are encoded differently. Here, the variable length is turned into the number of arguments in the function. If a function is a .Sy varargs type function, then the number of arguments is increased by one. Functions with type information are encoded as: .Li CTF_TYPE_INFO(CTF_K_FUNCTION, 0, nargs) . .Lp For functions that have no type information, nothing else is encoded, and the next function is encoded. For functions with type information, the next .Sy uint16_t is encoded with the type identifier of the return type of the function. It is followed by each of the type identifiers of the arguments, if any exist, in the order that they appear in the function. Therefore, argument 0 is the first type identifier and so on. When a function has a final varargs argument, that is encoded with the type identifier of zero. .Lp Like .Sx The Object Section , the function section is encoded in the order of the symbol table. It has similar, but slightly different considerations from objects. While iterating the symbol table, if any of the following conditions are true, then the entry is skipped and no corresponding entry is written: .Lp .Bl -bullet -offset indent -compact .It The symbol type is not .Sy STT_FUNCTION .It The symbol's section index is .Sy SHN_UNDEF .It The symbol's name offset is zero .It The symbol's name is .Li _START_ or .Li _END_ . These are skipped because they are used for scoping local symbols in ELF. .El .Ss The Type Section The type section is the heart of the .Nm data. It encodes all of the information about the types themselves. The base of the type information comes in two forms, a short form and a long form, each of which may be followed by a variable number of arguments. The following definitions describe the short and long forms: .Bd -literal #define CTF_MAX_SIZE 0xfffe /* max size of a type in bytes */ #define CTF_LSIZE_SENT 0xffff /* sentinel for ctt_size */ #define CTF_MAX_LSIZE UINT64_MAX typedef struct ctf_stype { uint_t ctt_name; /* reference to name in string table */ ushort_t ctt_info; /* encoded kind, variant length */ union { ushort_t _size; /* size of entire type in bytes */ ushort_t _type; /* reference to another type */ } _u; } ctf_stype_t; typedef struct ctf_type { uint_t ctt_name; /* reference to name in string table */ ushort_t ctt_info; /* encoded kind, variant length */ union { ushort_t _size; /* always CTF_LSIZE_SENT */ ushort_t _type; /* do not use */ } _u; uint_t ctt_lsizehi; /* high 32 bits of type size in bytes */ uint_t ctt_lsizelo; /* low 32 bits of type size in bytes */ } ctf_type_t; #define ctt_size _u._size /* for fundamental types that have a size */ #define ctt_type _u._type /* for types that reference another type */ .Ed .Pp Type sizes are stored in .Sy bytes . The basic small form uses a .Sy ushort_t to store the number of bytes. If the number of bytes in a structure would exceed 0xfffe, then the alternate form, the .Sy ctf_type_t , is used instead. To indicate that the larger form is being used, the member .Em ctt_size is set to value of .Sy CTF_LSIZE_SENT (0xffff). In general, when going through the type section, consumers use the .Sy ctf_type_t structure, but pay attention to the value of the member .Em ctt_size to determine whether they should increment their scan by the size of the .Sy ctf_stype_t or .Sy ctf_type_t . Not all kinds of types use .Sy ctt_size . Those which do not, will always use the .Sy ctf_stype_t structure. The individual sections for each kind have more information. .Lp Types are written out in order. Therefore the first entry encountered has a type id of 0x1, or 0x8000 if a child. The member .Em ctt_name is encoded as described in the section .Sx String Identifiers . The string that it points to is the name of the type. If the identifier points to an empty string (one that consists solely of a null terminator) then the type does not have a name, this is common with anonymous structures and unions that only have a typedef to name them, as well as pointers and qualifiers. .Lp The next member, the .Em ctt_info , is encoded as described in the section .Sx Type Encoding . The type's kind tells us how to interpret the remaining data in the .Sy ctf_type_t and any variable length data that may exist. The rest of this section will be broken down into the interpretation of the various kinds. .Ss Encoding of Integers Integers, which are of type .Sy CTF_K_INTEGER , have no variable length arguments. Instead, they are followed by a four byte .Sy uint_t which describes their encoding. All integers must be encoded with a variable length of zero. The .Em ctt_size member describes the length of the integer in bytes. In general, integer sizes will be rounded up to the closest power of two. .Lp The integer encoding contains three different pieces of information: .Bl -bullet -offset indent -compact .It The encoding of the integer .It The offset in .Sy bits of the type .It The size in .Sy bits of the type .El .Pp This encoding can be expressed through the following macros: .Bd -literal -offset indent #define CTF_INT_ENCODING(data) (((data) & 0xff000000) >> 24) #define CTF_INT_OFFSET(data) (((data) & 0x00ff0000) >> 16) #define CTF_INT_BITS(data) (((data) & 0x0000ffff)) #define CTF_INT_DATA(encoding, offset, bits) \\ (((encoding) << 24) | ((offset) << 16) | (bits)) .Ed .Pp The following flags are defined for the encoding at this time: .Bd -literal -offset indent #define CTF_INT_SIGNED 0x01 #define CTF_INT_CHAR 0x02 #define CTF_INT_BOOL 0x04 #define CTF_INT_VARARGS 0x08 .Ed .Lp By default, an integer is considered to be unsigned, unless it has the .Sy CTF_INT_SIGNED flag set. If the flag .Sy CTF_INT_CHAR is set, that indicates that the integer is of a type that stores character data, for example the intrinsic C type .Sy char would have the .Sy CTF_INT_CHAR flag set. If the flag .Sy CTF_INT_BOOL is set, that indicates that the integer represents a boolean type. For example, the intrinsic C type .Sy _Bool would have the .Sy CTF_INT_BOOL flag set. Finally, the flag .Sy CTF_INT_VARARGS indicates that the integer is used as part of a variable number of arguments. This encoding is rather uncommon. .Ss Encoding of Floats Floats, which are of type .Sy CTF_K_FLOAT , are similar to their integer counterparts. They have no variable length arguments and are followed by a four byte encoding which describes the kind of float that exists. The .Em ctt_size member is the size, in bytes, of the float. The float encoding has three different pieces of information inside of it: .Lp .Bl -bullet -offset indent -compact .It The specific kind of float that exists .It The offset in .Sy bits of the float .It The size in .Sy bits of the float .El .Lp This encoding can be expressed through the following macros: .Bd -literal -offset indent #define CTF_FP_ENCODING(data) (((data) & 0xff000000) >> 24) #define CTF_FP_OFFSET(data) (((data) & 0x00ff0000) >> 16) #define CTF_FP_BITS(data) (((data) & 0x0000ffff)) #define CTF_FP_DATA(encoding, offset, bits) \\ (((encoding) << 24) | ((offset) << 16) | (bits)) .Ed .Lp Where as the encoding for integers is a series of flags, the encoding for floats maps to a specific kind of float. It is not a flag-based value. The kinds of floats correspond to both their size, and the encoding. This covers all of the basic C intrinsic floating point types. The following are the different kinds of floats represented in the encoding: .Bd -literal -offset indent #define CTF_FP_SINGLE 1 /* IEEE 32-bit float encoding */ #define CTF_FP_DOUBLE 2 /* IEEE 64-bit float encoding */ #define CTF_FP_CPLX 3 /* Complex encoding */ #define CTF_FP_DCPLX 4 /* Double complex encoding */ #define CTF_FP_LDCPLX 5 /* Long double complex encoding */ #define CTF_FP_LDOUBLE 6 /* Long double encoding */ #define CTF_FP_INTRVL 7 /* Interval (2x32-bit) encoding */ #define CTF_FP_DINTRVL 8 /* Double interval (2x64-bit) encoding */ #define CTF_FP_LDINTRVL 9 /* Long double interval (2x128-bit) encoding */ #define CTF_FP_IMAGRY 10 /* Imaginary (32-bit) encoding */ #define CTF_FP_DIMAGRY 11 /* Long imaginary (64-bit) encoding */ #define CTF_FP_LDIMAGRY 12 /* Long double imaginary (128-bit) encoding */ .Ed .Ss Encoding of Arrays Arrays, which are of type .Sy CTF_K_ARRAY , have no variable length arguments. They are followed by a structure which describes the number of elements in the array, the type identifier of the elements in the array, and the type identifier of the index of the array. With arrays, the .Em ctt_size member is set to zero. The structure that follows an array is defined as: .Bd -literal typedef struct ctf_array { ushort_t cta_contents; /* reference to type of array contents */ ushort_t cta_index; /* reference to type of array index */ uint_t cta_nelems; /* number of elements */ } ctf_array_t; .Ed .Lp The .Em cta_contents and .Em cta_index members of the .Sy ctf_array_t are type identifiers which are encoded as per the section .Sx Type Identifiers . The member .Em cta_nelems is a simple four byte unsigned count of the number of elements. This count may be zero when encountering C99's flexible array members. .Ss Encoding of Functions Function types, which are of type .Sy CTF_K_FUNCTION , use the variable length list to be the number of arguments in the function. When the function has a final member which is a varargs, then the argument count is incremented by one to account for the variable argument. Here, the .Em ctt_type member is encoded with the type identifier of the return type of the function. Note that the .Em ctt_size member is not used here. .Lp The variable argument list contains the type identifiers for the arguments of the function, if any. Each one is represented by a .Sy uint16_t and encoded according to the .Sx Type Identifiers section. If the function's last argument is of type varargs, then it is also written out, but the type identifier is zero. This is included in the count of the function's arguments. An extra type identifier may follow the argument and return type identifiers in order to maintain four-byte alignment for the following type definition. Such a type identifier is not included in the argument count and has a value of zero. .Ss Encoding of Structures and Unions Structures and Unions, which are encoded with .Sy CTF_K_STRUCT and .Sy CTF_K_UNION respectively, are very similar constructs in C. The main difference between them is the fact that members of a structure follow one another, where as in a union, all members share the same memory. They are also very similar in terms of their encoding in .Nm . The variable length argument for structures and unions represents the number of members that they have. The value of the member .Em ctt_size is the size of the structure and union. There are two different structures which are used to encode members in the variable list. When the size of a structure or union is greater than or equal to the large member threshold, 8192, then a different structure is used to encode the member, all members are encoded using the same structure. The structure for members is as follows: .Bd -literal typedef struct ctf_member { uint_t ctm_name; /* reference to name in string table */ ushort_t ctm_type; /* reference to type of member */ ushort_t ctm_offset; /* offset of this member in bits */ } ctf_member_t; typedef struct ctf_lmember { uint_t ctlm_name; /* reference to name in string table */ ushort_t ctlm_type; /* reference to type of member */ ushort_t ctlm_pad; /* padding */ uint_t ctlm_offsethi; /* high 32 bits of member offset in bits */ uint_t ctlm_offsetlo; /* low 32 bits of member offset in bits */ } ctf_lmember_t; .Ed .Lp Both the .Em ctm_name and .Em ctlm_name refer to the name of the member. The name is encoded as an offset into the string table as described by the section .Sx String Identifiers . The members .Sy ctm_type and .Sy ctlm_type both refer to the type of the member. They are encoded as per the section .Sx Type Identifiers . .Lp The last piece of information that is present is the offset which describes the offset in memory at which the member begins. For unions, this value will always be zero because each member of a union has an offset of zero. For structures, this is the offset in .Sy bits at which the member begins. Note that a compiler may lay out a type with padding. This means that the difference in offset between two consecutive members may be larger than the size of the member. When the size of the overall structure is strictly less than 8192 bytes, the normal structure, .Sy ctf_member_t , is used and the offset in bits is stored in the member .Em ctm_offset . However, when the size of the structure is greater than or equal to 8192 bytes, then the number of bits is split into two 32-bit quantities. One member, .Em ctlm_offsethi , represents the upper 32 bits of the offset, while the other member, .Em ctlm_offsetlo , represents the lower 32 bits of the offset. These can be joined together to get a 64-bit sized offset in bits by shifting the member .Em ctlm_offsethi to the left by thirty two and then doing a binary or of .Em ctlm_offsetlo . .Ss Encoding of Enumerations Enumerations, noted by the type .Sy CTF_K_ENUM , are similar to structures. Enumerations use the variable list to note the number of values that the enumeration contains, which we'll term enumerators. In C, an enumeration is always equivalent to the intrinsic type .Sy int , thus the value of the member .Em ctt_size is always the size of an integer which is determined based on the current model. For .Fx systems, this will always be 4, as an integer is always defined to be 4 bytes large in both .Sy ILP32 and .Sy LP64 , regardless of the architecture. For further details, see .Xr arch 7 . .Lp The enumerators encoded in an enumeration have the following structure in the variable list: .Bd -literal typedef struct ctf_enum { uint_t cte_name; /* reference to name in string table */ int cte_value; /* value associated with this name */ } ctf_enum_t; .Ed .Pp The member .Em cte_name refers to the name of the enumerator's value, it is encoded according to the rules in the section .Sx String Identifiers . The member .Em cte_value contains the integer value of this enumerator. .Ss Encoding of Forward References Forward references, types of kind .Sy CTF_K_FORWARD , in a .Nm file refer to types which may not have a definition at all, only a name. If the .Nm file is a child, then it may be that the forward is resolved to an actual type in the parent, otherwise the definition may be in another .Nm container or may not be known at all. The only member of the .Sy ctf_type_t that matters for a forward declaration is the .Em ctt_name which points to the name of the forward reference in the string table as described earlier. There is no other information recorded for forward references. .Ss Encoding of Pointers, Typedefs, Volatile, Const, and Restrict Pointers, typedefs, volatile, const, and restrict are all similar in .Nm . They all refer to another type. In the case of typedefs, they provide an alternate name, while volatile, const, and restrict change how the type is interpreted in the C programming language. This covers the .Nm kinds .Sy CTF_K_POINTER , .Sy CTF_K_TYPEDEF , .Sy CTF_K_VOLATILE , .Sy CTF_K_RESTRICT , and .Sy CTF_K_CONST . .Lp These types have no variable list entries and use the member .Em ctt_type to refer to the base type that they modify. .Ss Encoding of Unknown Types Types with the kind .Sy CTF_K_UNKNOWN are used to indicate gaps in the type identifier space. These entries consume an identifier, but do not define anything. Nothing should refer to these gap identifiers. .Ss Dependencies Between Types C types can be imagined as a directed, cyclic, graph. Structures and unions may refer to each other in a way that creates a cyclic dependency. In cases such as these, the entire type section must be read in and processed. Consumers must not assume that every type can be laid out in dependency order; they cannot. .Ss The String Section The last section of the .Nm file is the .Sy string section. This section encodes all of the strings that appear throughout the other sections. It is laid out as a series of characters followed by a null terminator. Generally, all names are written out in ASCII, as most C compilers do not allow any characters to appear in identifiers outside of a subset of ASCII. However, any extended characters sets should be written out as a series of UTF-8 bytes. .Lp The first entry in the section, at offset zero, is a single null terminator to reference the empty string. Following that, each C string should be written out, including the null terminator. Offsets that refer to something in this section should refer to the first byte which begins a string. Beyond the first byte in the section being the null terminator, the order of strings is unimportant. .Ss Data Encoding and ELF Considerations .Nm data is generally included in ELF objects which specify information to identify the architecture and endianness of the file. A .Nm container inside such an object must match the endianness of the ELF object. Aside from the question of the endian encoding of data, there should be no other differences between architectures. While many of the types in this document refer to non-fixed size C integral types, they are equivalent in the models .Sy ILP32 and .Sy LP64 . If any other model is being used with .Nm data that has different sizes, then it must not use the model's sizes for those integral types and instead use the fixed size equivalents based on an .Sy ILP32 environment. .Lp When placing a .Nm container inside of an ELF object, there are certain conventions that are expected for the purposes of tooling being able to find the .Nm data. In particular, a given ELF object should only contain a single .Nm section. Multiple containers should be merged together into a single one. .Lp The .Nm file should be included in its own ELF section. The section's name must be .Ql .SUNW_ctf . The type of the section must be .Sy SHT_PROGBITS . The section should have a link set to the symbol table and its address alignment must be 4. .Sh SEE ALSO .Xr dtrace 1 , .Xr elf 3 , .Xr gelf 3 , .Xr a.out 5 , .Xr elf 5 , .Xr arch 7