Source file src/cmd/cgo/doc.go

     1  // Copyright 2009 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  /*
     6  Cgo enables the creation of Go packages that call C code.
     7  
     8  # Using cgo with the go command
     9  
    10  To use cgo write normal Go code that imports a pseudo-package "C".
    11  The Go code can then refer to types such as C.size_t, variables such
    12  as C.stdout, or functions such as C.putchar.
    13  
    14  If the import of "C" is immediately preceded by a comment, that
    15  comment, called the preamble, is used as a header when compiling
    16  the C parts of the package. For example:
    17  
    18  	// #include <stdio.h>
    19  	// #include <errno.h>
    20  	import "C"
    21  
    22  The preamble may contain any C code, including function and variable
    23  declarations and definitions. These may then be referred to from Go
    24  code as though they were defined in the package "C". All names
    25  declared in the preamble may be used, even if they start with a
    26  lower-case letter. Exception: static variables in the preamble may
    27  not be referenced from Go code; static functions are permitted.
    28  
    29  See $GOROOT/cmd/cgo/internal/teststdio and $GOROOT/misc/cgo/gmp for examples. See
    30  "C? Go? Cgo!" for an introduction to using cgo:
    31  https://golang.org/doc/articles/c_go_cgo.html.
    32  
    33  CFLAGS, CPPFLAGS, CXXFLAGS, FFLAGS and LDFLAGS may be defined with pseudo
    34  #cgo directives within these comments to tweak the behavior of the C, C++
    35  or Fortran compiler. Values defined in multiple directives are concatenated
    36  together. The directive can include a list of build constraints limiting its
    37  effect to systems satisfying one of the constraints
    38  (see https://golang.org/pkg/go/build/#hdr-Build_Constraints for details about the constraint syntax).
    39  For example:
    40  
    41  	// #cgo CFLAGS: -DPNG_DEBUG=1
    42  	// #cgo amd64 386 CFLAGS: -DX86=1
    43  	// #cgo LDFLAGS: -lpng
    44  	// #include <png.h>
    45  	import "C"
    46  
    47  Alternatively, CPPFLAGS and LDFLAGS may be obtained via the pkg-config tool
    48  using a '#cgo pkg-config:' directive followed by the package names.
    49  For example:
    50  
    51  	// #cgo pkg-config: png cairo
    52  	// #include <png.h>
    53  	import "C"
    54  
    55  The default pkg-config tool may be changed by setting the PKG_CONFIG environment variable.
    56  
    57  For security reasons, only a limited set of flags are allowed, notably -D, -U, -I, and -l.
    58  To allow additional flags, set CGO_CFLAGS_ALLOW to a regular expression
    59  matching the new flags. To disallow flags that would otherwise be allowed,
    60  set CGO_CFLAGS_DISALLOW to a regular expression matching arguments
    61  that must be disallowed. In both cases the regular expression must match
    62  a full argument: to allow -mfoo=bar, use CGO_CFLAGS_ALLOW='-mfoo.*',
    63  not just CGO_CFLAGS_ALLOW='-mfoo'. Similarly named variables control
    64  the allowed CPPFLAGS, CXXFLAGS, FFLAGS, and LDFLAGS.
    65  
    66  Also for security reasons, only a limited set of characters are
    67  permitted, notably alphanumeric characters and a few symbols, such as
    68  '.', that will not be interpreted in unexpected ways. Attempts to use
    69  forbidden characters will get a "malformed #cgo argument" error.
    70  
    71  When building, the CGO_CFLAGS, CGO_CPPFLAGS, CGO_CXXFLAGS, CGO_FFLAGS and
    72  CGO_LDFLAGS environment variables are added to the flags derived from
    73  these directives. Package-specific flags should be set using the
    74  directives, not the environment variables, so that builds work in
    75  unmodified environments. Flags obtained from environment variables
    76  are not subject to the security limitations described above.
    77  
    78  All the cgo CPPFLAGS and CFLAGS directives in a package are concatenated and
    79  used to compile C files in that package. All the CPPFLAGS and CXXFLAGS
    80  directives in a package are concatenated and used to compile C++ files in that
    81  package. All the CPPFLAGS and FFLAGS directives in a package are concatenated
    82  and used to compile Fortran files in that package. All the LDFLAGS directives
    83  in any package in the program are concatenated and used at link time. All the
    84  pkg-config directives are concatenated and sent to pkg-config simultaneously
    85  to add to each appropriate set of command-line flags.
    86  
    87  When the cgo directives are parsed, any occurrence of the string ${SRCDIR}
    88  will be replaced by the absolute path to the directory containing the source
    89  file. This allows pre-compiled static libraries to be included in the package
    90  directory and linked properly.
    91  For example if package foo is in the directory /go/src/foo:
    92  
    93  	// #cgo LDFLAGS: -L${SRCDIR}/libs -lfoo
    94  
    95  Will be expanded to:
    96  
    97  	// #cgo LDFLAGS: -L/go/src/foo/libs -lfoo
    98  
    99  When the Go tool sees that one or more Go files use the special import
   100  "C", it will look for other non-Go files in the directory and compile
   101  them as part of the Go package. Any .c, .s, .S or .sx files will be
   102  compiled with the C compiler. Any .cc, .cpp, or .cxx files will be
   103  compiled with the C++ compiler. Any .f, .F, .for or .f90 files will be
   104  compiled with the fortran compiler. Any .h, .hh, .hpp, or .hxx files will
   105  not be compiled separately, but, if these header files are changed,
   106  the package (including its non-Go source files) will be recompiled.
   107  Note that changes to files in other directories do not cause the package
   108  to be recompiled, so all non-Go source code for the package should be
   109  stored in the package directory, not in subdirectories.
   110  The default C and C++ compilers may be changed by the CC and CXX
   111  environment variables, respectively; those environment variables
   112  may include command line options.
   113  
   114  The cgo tool will always invoke the C compiler with the source file's
   115  directory in the include path; i.e. -I${SRCDIR} is always implied. This
   116  means that if a header file foo/bar.h exists both in the source
   117  directory and also in the system include directory (or some other place
   118  specified by a -I flag), then "#include <foo/bar.h>" will always find the
   119  local version in preference to any other version.
   120  
   121  The cgo tool is enabled by default for native builds on systems where
   122  it is expected to work. It is disabled by default when cross-compiling
   123  as well as when the CC environment variable is unset and the default
   124  C compiler (typically gcc or clang) cannot be found on the system PATH.
   125  You can override the default by setting the CGO_ENABLED
   126  environment variable when running the go tool: set it to 1 to enable
   127  the use of cgo, and to 0 to disable it. The go tool will set the
   128  build constraint "cgo" if cgo is enabled. The special import "C"
   129  implies the "cgo" build constraint, as though the file also said
   130  "//go:build cgo".  Therefore, if cgo is disabled, files that import
   131  "C" will not be built by the go tool. (For more about build constraints
   132  see https://golang.org/pkg/go/build/#hdr-Build_Constraints).
   133  
   134  When cross-compiling, you must specify a C cross-compiler for cgo to
   135  use. You can do this by setting the generic CC_FOR_TARGET or the
   136  more specific CC_FOR_${GOOS}_${GOARCH} (for example, CC_FOR_linux_arm)
   137  environment variable when building the toolchain using make.bash,
   138  or you can set the CC environment variable any time you run the go tool.
   139  
   140  The CXX_FOR_TARGET, CXX_FOR_${GOOS}_${GOARCH}, and CXX
   141  environment variables work in a similar way for C++ code.
   142  
   143  # Go references to C
   144  
   145  Within the Go file, C's struct field names that are keywords in Go
   146  can be accessed by prefixing them with an underscore: if x points at a C
   147  struct with a field named "type", x._type accesses the field.
   148  C struct fields that cannot be expressed in Go, such as bit fields
   149  or misaligned data, are omitted in the Go struct, replaced by
   150  appropriate padding to reach the next field or the end of the struct.
   151  
   152  The standard C numeric types are available under the names
   153  C.char, C.schar (signed char), C.uchar (unsigned char),
   154  C.short, C.ushort (unsigned short), C.int, C.uint (unsigned int),
   155  C.long, C.ulong (unsigned long), C.longlong (long long),
   156  C.ulonglong (unsigned long long), C.float, C.double,
   157  C.complexfloat (complex float), and C.complexdouble (complex double).
   158  The C type void* is represented by Go's unsafe.Pointer.
   159  The C types __int128_t and __uint128_t are represented by [16]byte.
   160  
   161  A few special C types which would normally be represented by a pointer
   162  type in Go are instead represented by a uintptr.  See the Special
   163  cases section below.
   164  
   165  To access a struct, union, or enum type directly, prefix it with
   166  struct_, union_, or enum_, as in C.struct_stat. The size of any C type
   167  T is available as C.sizeof_T, as in C.sizeof_struct_stat. These
   168  special prefixes means that there is no way to directly reference a C
   169  identifier that starts with "struct_", "union_", "enum_", or
   170  "sizeof_", such as a function named "struct_function".
   171  A workaround is to use a "#define" in the preamble, as in
   172  "#define c_struct_function struct_function" and then in the
   173  Go code refer to "C.c_struct_function".
   174  
   175  A C function may be declared in the Go file with a parameter type of
   176  the special name _GoString_. This function may be called with an
   177  ordinary Go string value. The string length, and a pointer to the
   178  string contents, may be accessed by calling the C functions
   179  
   180  	size_t _GoStringLen(_GoString_ s);
   181  	const char *_GoStringPtr(_GoString_ s);
   182  
   183  These functions are only available in the preamble, not in other C
   184  files. The C code must not modify the contents of the pointer returned
   185  by _GoStringPtr. Note that the string contents may not have a trailing
   186  NUL byte.
   187  
   188  As Go doesn't have support for C's union type in the general case,
   189  C's union types are represented as a Go byte array with the same length.
   190  
   191  Go structs cannot embed fields with C types.
   192  
   193  Go code cannot refer to zero-sized fields that occur at the end of
   194  non-empty C structs. To get the address of such a field (which is the
   195  only operation you can do with a zero-sized field) you must take the
   196  address of the struct and add the size of the struct.
   197  
   198  Cgo translates C types into equivalent unexported Go types.
   199  Because the translations are unexported, a Go package should not
   200  expose C types in its exported API: a C type used in one Go package
   201  is different from the same C type used in another.
   202  
   203  Any C function (even void functions) may be called in a multiple
   204  assignment context to retrieve both the return value (if any) and the
   205  C errno variable as an error (use _ to skip the result value if the
   206  function returns void). For example:
   207  
   208  	n, err = C.sqrt(-1)
   209  	_, err := C.voidFunc()
   210  	var n, err = C.sqrt(1)
   211  
   212  Note that the C errno value may be non-zero, and thus the err result may be
   213  non-nil, even if the function call is successful. Unlike normal Go conventions,
   214  you should first check whether the call succeeded before checking the error
   215  result. For example:
   216  
   217  	n, err := C.setenv(key, value, 1)
   218  	if n != 0 {
   219  		// we know the call failed, so it is now valid to use err
   220  		return err
   221  	}
   222  
   223  Calling C function pointers is currently not supported, however you can
   224  declare Go variables which hold C function pointers and pass them
   225  back and forth between Go and C. C code may call function pointers
   226  received from Go. For example:
   227  
   228  	package main
   229  
   230  	// typedef int (*intFunc) ();
   231  	//
   232  	// int
   233  	// bridge_int_func(intFunc f)
   234  	// {
   235  	//		return f();
   236  	// }
   237  	//
   238  	// int fortytwo()
   239  	// {
   240  	//	    return 42;
   241  	// }
   242  	import "C"
   243  	import "fmt"
   244  
   245  	func main() {
   246  		f := C.intFunc(C.fortytwo)
   247  		fmt.Println(int(C.bridge_int_func(f)))
   248  		// Output: 42
   249  	}
   250  
   251  In C, a function argument written as a fixed size array
   252  actually requires a pointer to the first element of the array.
   253  C compilers are aware of this calling convention and adjust
   254  the call accordingly, but Go cannot. In Go, you must pass
   255  the pointer to the first element explicitly: C.f(&C.x[0]).
   256  
   257  Calling variadic C functions is not supported. It is possible to
   258  circumvent this by using a C function wrapper. For example:
   259  
   260  	package main
   261  
   262  	// #include <stdio.h>
   263  	// #include <stdlib.h>
   264  	//
   265  	// static void myprint(char* s) {
   266  	//   printf("%s\n", s);
   267  	// }
   268  	import "C"
   269  	import "unsafe"
   270  
   271  	func main() {
   272  		cs := C.CString("Hello from stdio")
   273  		C.myprint(cs)
   274  		C.free(unsafe.Pointer(cs))
   275  	}
   276  
   277  A few special functions convert between Go and C types
   278  by making copies of the data. In pseudo-Go definitions:
   279  
   280  	// Go string to C string
   281  	// The C string is allocated in the C heap using malloc.
   282  	// It is the caller's responsibility to arrange for it to be
   283  	// freed, such as by calling C.free (be sure to include stdlib.h
   284  	// if C.free is needed).
   285  	func C.CString(string) *C.char
   286  
   287  	// Go []byte slice to C array
   288  	// The C array is allocated in the C heap using malloc.
   289  	// It is the caller's responsibility to arrange for it to be
   290  	// freed, such as by calling C.free (be sure to include stdlib.h
   291  	// if C.free is needed).
   292  	func C.CBytes([]byte) unsafe.Pointer
   293  
   294  	// C string to Go string
   295  	func C.GoString(*C.char) string
   296  
   297  	// C data with explicit length to Go string
   298  	func C.GoStringN(*C.char, C.int) string
   299  
   300  	// C data with explicit length to Go []byte
   301  	func C.GoBytes(unsafe.Pointer, C.int) []byte
   302  
   303  As a special case, C.malloc does not call the C library malloc directly
   304  but instead calls a Go helper function that wraps the C library malloc
   305  but guarantees never to return nil. If C's malloc indicates out of memory,
   306  the helper function crashes the program, like when Go itself runs out
   307  of memory. Because C.malloc cannot fail, it has no two-result form
   308  that returns errno.
   309  
   310  # C references to Go
   311  
   312  Go functions can be exported for use by C code in the following way:
   313  
   314  	//export MyFunction
   315  	func MyFunction(arg1, arg2 int, arg3 string) int64 {...}
   316  
   317  	//export MyFunction2
   318  	func MyFunction2(arg1, arg2 int, arg3 string) (int64, *C.char) {...}
   319  
   320  They will be available in the C code as:
   321  
   322  	extern GoInt64 MyFunction(int arg1, int arg2, GoString arg3);
   323  	extern struct MyFunction2_return MyFunction2(int arg1, int arg2, GoString arg3);
   324  
   325  found in the _cgo_export.h generated header, after any preambles
   326  copied from the cgo input files. Functions with multiple
   327  return values are mapped to functions returning a struct.
   328  
   329  Not all Go types can be mapped to C types in a useful way.
   330  Go struct types are not supported; use a C struct type.
   331  Go array types are not supported; use a C pointer.
   332  
   333  Go functions that take arguments of type string may be called with the
   334  C type _GoString_, described above. The _GoString_ type will be
   335  automatically defined in the preamble. Note that there is no way for C
   336  code to create a value of this type; this is only useful for passing
   337  string values from Go to C and back to Go.
   338  
   339  Using //export in a file places a restriction on the preamble:
   340  since it is copied into two different C output files, it must not
   341  contain any definitions, only declarations. If a file contains both
   342  definitions and declarations, then the two output files will produce
   343  duplicate symbols and the linker will fail. To avoid this, definitions
   344  must be placed in preambles in other files, or in C source files.
   345  
   346  # Passing pointers
   347  
   348  Go is a garbage collected language, and the garbage collector needs to
   349  know the location of every pointer to Go memory. Because of this,
   350  there are restrictions on passing pointers between Go and C.
   351  
   352  In this section the term Go pointer means a pointer to memory
   353  allocated by Go (such as by using the & operator or calling the
   354  predefined new function) and the term C pointer means a pointer to
   355  memory allocated by C (such as by a call to C.malloc). Whether a
   356  pointer is a Go pointer or a C pointer is a dynamic property
   357  determined by how the memory was allocated; it has nothing to do with
   358  the type of the pointer.
   359  
   360  Note that values of some Go types, other than the type's zero value,
   361  always include Go pointers. This is true of interface, channel, map,
   362  and function types. A pointer type may hold a Go pointer or a C pointer.
   363  Array, slice, string, and struct types may or may not include Go pointers,
   364  depending on their type and how they are constructed. All the discussion
   365  below about Go pointers applies not just to pointer types,
   366  but also to other types that include Go pointers.
   367  
   368  All Go pointers passed to C must point to pinned Go memory. Go pointers
   369  passed as function arguments to C functions have the memory they point to
   370  implicitly pinned for the duration of the call. Go memory reachable from
   371  these function arguments must be pinned as long as the C code has access
   372  to it. Whether Go memory is pinned is a dynamic property of that memory
   373  region; it has nothing to do with the type of the pointer.
   374  
   375  Go values created by calling new, by taking the address of a composite
   376  literal, or by taking the address of a local variable may also have their
   377  memory pinned using [runtime.Pinner]. This type may be used to manage
   378  the duration of the memory's pinned status, potentially beyond the
   379  duration of a C function call. Memory may be pinned more than once and
   380  must be unpinned exactly the same number of times it has been pinned.
   381  
   382  Go code may pass a Go pointer to C provided the memory to which it
   383  points does not contain any Go pointers to memory that is unpinned. When
   384  passing a pointer to a field in a struct, the Go memory in question is
   385  the memory occupied by the field, not the entire struct. When passing a
   386  pointer to an element in an array or slice, the Go memory in question is
   387  the entire array or the entire backing array of the slice.
   388  
   389  C code may keep a copy of a Go pointer only as long as the memory it
   390  points to is pinned.
   391  
   392  C code may not keep a copy of a Go pointer after the call returns,
   393  unless the memory it points to is pinned with [runtime.Pinner] and the
   394  Pinner is not unpinned while the Go pointer is stored in C memory.
   395  This implies that C code may not keep a copy of a string, slice,
   396  channel, and so forth, because they cannot be pinned with
   397  [runtime.Pinner].
   398  
   399  The _GoString_ type also may not be pinned with [runtime.Pinner].
   400  Because it includes a Go pointer, the memory it points to is only pinned
   401  for the duration of the call; _GoString_ values may not be retained by C
   402  code.
   403  
   404  A Go function called by C code may return a Go pointer to pinned memory
   405  (which implies that it may not return a string, slice, channel, and so
   406  forth). A Go function called by C code may take C pointers as arguments,
   407  and it may store non-pointer data, C pointers, or Go pointers to pinned
   408  memory through those pointers. It may not store a Go pointer to unpinned
   409  memory in memory pointed to by a C pointer (which again, implies that it
   410  may not store a string, slice, channel, and so forth). A Go function
   411  called by C code may take a Go pointer but it must preserve the property
   412  that the Go memory to which it points (and the Go memory to which that
   413  memory points, and so on) is pinned.
   414  
   415  These rules are checked dynamically at runtime. The checking is
   416  controlled by the cgocheck setting of the GODEBUG environment
   417  variable. The default setting is GODEBUG=cgocheck=1, which implements
   418  reasonably cheap dynamic checks. These checks may be disabled
   419  entirely using GODEBUG=cgocheck=0. Complete checking of pointer
   420  handling, at some cost in run time, is available by setting
   421  GOEXPERIMENT=cgocheck2 at build time.
   422  
   423  It is possible to defeat this enforcement by using the unsafe package,
   424  and of course there is nothing stopping the C code from doing anything
   425  it likes. However, programs that break these rules are likely to fail
   426  in unexpected and unpredictable ways.
   427  
   428  The runtime/cgo.Handle type can be used to safely pass Go values
   429  between Go and C. See the runtime/cgo package documentation for details.
   430  
   431  Note: the current implementation has a bug. While Go code is permitted
   432  to write nil or a C pointer (but not a Go pointer) to C memory, the
   433  current implementation may sometimes cause a runtime error if the
   434  contents of the C memory appear to be a Go pointer. Therefore, avoid
   435  passing uninitialized C memory to Go code if the Go code is going to
   436  store pointer values in it. Zero out the memory in C before passing it
   437  to Go.
   438  
   439  # Optimizing calls of C code
   440  
   441  When passing a Go pointer to a C function the compiler normally ensures
   442  that the Go object lives on the heap. If the C function does not keep
   443  a copy of the Go pointer, and never passes the Go pointer back to Go code,
   444  then this is unnecessary. The #cgo noescape directive may be used to tell
   445  the compiler that no Go pointers escape via the named C function.
   446  If the noescape directive is used and the C function does not handle the
   447  pointer safely, the program may crash or see memory corruption.
   448  
   449  For example:
   450  
   451  	// #cgo noescape cFunctionName
   452  
   453  When a Go function calls a C function, it prepares for the C function to
   454  call back to a Go function. The #cgo nocallback directive may be used to
   455  tell the compiler that these preparations are not necessary.
   456  If the nocallback directive is used and the C function does call back into
   457  Go code, the program will panic.
   458  
   459  For example:
   460  
   461  	// #cgo nocallback cFunctionName
   462  
   463  # Special cases
   464  
   465  A few special C types which would normally be represented by a pointer
   466  type in Go are instead represented by a uintptr. Those include:
   467  
   468  1. The *Ref types on Darwin, rooted at CoreFoundation's CFTypeRef type.
   469  
   470  2. The object types from Java's JNI interface:
   471  
   472  	jobject
   473  	jclass
   474  	jthrowable
   475  	jstring
   476  	jarray
   477  	jbooleanArray
   478  	jbyteArray
   479  	jcharArray
   480  	jshortArray
   481  	jintArray
   482  	jlongArray
   483  	jfloatArray
   484  	jdoubleArray
   485  	jobjectArray
   486  	jweak
   487  
   488  3. The EGLDisplay and EGLConfig types from the EGL API.
   489  
   490  These types are uintptr on the Go side because they would otherwise
   491  confuse the Go garbage collector; they are sometimes not really
   492  pointers but data structures encoded in a pointer type. All operations
   493  on these types must happen in C. The proper constant to initialize an
   494  empty such reference is 0, not nil.
   495  
   496  These special cases were introduced in Go 1.10. For auto-updating code
   497  from Go 1.9 and earlier, use the cftype or jni rewrites in the Go fix tool:
   498  
   499  	go tool fix -r cftype <pkg>
   500  	go tool fix -r jni <pkg>
   501  
   502  It will replace nil with 0 in the appropriate places.
   503  
   504  The EGLDisplay case was introduced in Go 1.12. Use the egl rewrite
   505  to auto-update code from Go 1.11 and earlier:
   506  
   507  	go tool fix -r egl <pkg>
   508  
   509  The EGLConfig case was introduced in Go 1.15. Use the eglconf rewrite
   510  to auto-update code from Go 1.14 and earlier:
   511  
   512  	go tool fix -r eglconf <pkg>
   513  
   514  # Using cgo directly
   515  
   516  Usage:
   517  
   518  	go tool cgo [cgo options] [-- compiler options] gofiles...
   519  
   520  Cgo transforms the specified input Go source files into several output
   521  Go and C source files.
   522  
   523  The compiler options are passed through uninterpreted when
   524  invoking the C compiler to compile the C parts of the package.
   525  
   526  The following options are available when running cgo directly:
   527  
   528  	-V
   529  		Print cgo version and exit.
   530  	-debug-define
   531  		Debugging option. Print #defines.
   532  	-debug-gcc
   533  		Debugging option. Trace C compiler execution and output.
   534  	-dynimport file
   535  		Write list of symbols imported by file. Write to
   536  		-dynout argument or to standard output. Used by go
   537  		build when building a cgo package.
   538  	-dynlinker
   539  		Write dynamic linker as part of -dynimport output.
   540  	-dynout file
   541  		Write -dynimport output to file.
   542  	-dynpackage package
   543  		Set Go package for -dynimport output.
   544  	-exportheader file
   545  		If there are any exported functions, write the
   546  		generated export declarations to file.
   547  		C code can #include this to see the declarations.
   548  	-gccgo
   549  		Generate output for the gccgo compiler rather than the
   550  		gc compiler.
   551  	-gccgoprefix prefix
   552  		The -fgo-prefix option to be used with gccgo.
   553  	-gccgopkgpath path
   554  		The -fgo-pkgpath option to be used with gccgo.
   555  	-gccgo_define_cgoincomplete
   556  		Define cgo.Incomplete locally rather than importing it from
   557  		the "runtime/cgo" package. Used for old gccgo versions.
   558  	-godefs
   559  		Write out input file in Go syntax replacing C package
   560  		names with real values. Used to generate files in the
   561  		syscall package when bootstrapping a new target.
   562  	-importpath string
   563  		The import path for the Go package. Optional; used for
   564  		nicer comments in the generated files.
   565  	-import_runtime_cgo
   566  		If set (which it is by default) import runtime/cgo in
   567  		generated output.
   568  	-import_syscall
   569  		If set (which it is by default) import syscall in
   570  		generated output.
   571  	-ldflags flags
   572  		Flags to pass to the C linker. The cmd/go tool uses
   573  		this to pass in the flags in the CGO_LDFLAGS variable.
   574  	-objdir directory
   575  		Put all generated files in directory.
   576  	-srcdir directory
   577  		Find the Go input files, listed on the command line,
   578  		in directory.
   579  	-trimpath rewrites
   580  		Apply trims and rewrites to source file paths.
   581  */
   582  package main
   583  
   584  /*
   585  Implementation details.
   586  
   587  Cgo provides a way for Go programs to call C code linked into the same
   588  address space. This comment explains the operation of cgo.
   589  
   590  Cgo reads a set of Go source files and looks for statements saying
   591  import "C". If the import has a doc comment, that comment is
   592  taken as literal C code to be used as a preamble to any C code
   593  generated by cgo. A typical preamble #includes necessary definitions:
   594  
   595  	// #include <stdio.h>
   596  	import "C"
   597  
   598  For more details about the usage of cgo, see the documentation
   599  comment at the top of this file.
   600  
   601  Understanding C
   602  
   603  Cgo scans the Go source files that import "C" for uses of that
   604  package, such as C.puts. It collects all such identifiers. The next
   605  step is to determine each kind of name. In C.xxx the xxx might refer
   606  to a type, a function, a constant, or a global variable. Cgo must
   607  decide which.
   608  
   609  The obvious thing for cgo to do is to process the preamble, expanding
   610  #includes and processing the corresponding C code. That would require
   611  a full C parser and type checker that was also aware of any extensions
   612  known to the system compiler (for example, all the GNU C extensions) as
   613  well as the system-specific header locations and system-specific
   614  pre-#defined macros. This is certainly possible to do, but it is an
   615  enormous amount of work.
   616  
   617  Cgo takes a different approach. It determines the meaning of C
   618  identifiers not by parsing C code but by feeding carefully constructed
   619  programs into the system C compiler and interpreting the generated
   620  error messages, debug information, and object files. In practice,
   621  parsing these is significantly less work and more robust than parsing
   622  C source.
   623  
   624  Cgo first invokes gcc -E -dM on the preamble, in order to find out
   625  about simple #defines for constants and the like. These are recorded
   626  for later use.
   627  
   628  Next, cgo needs to identify the kinds for each identifier. For the
   629  identifiers C.foo, cgo generates this C program:
   630  
   631  	<preamble>
   632  	#line 1 "not-declared"
   633  	void __cgo_f_1_1(void) { __typeof__(foo) *__cgo_undefined__1; }
   634  	#line 1 "not-type"
   635  	void __cgo_f_1_2(void) { foo *__cgo_undefined__2; }
   636  	#line 1 "not-int-const"
   637  	void __cgo_f_1_3(void) { enum { __cgo_undefined__3 = (foo)*1 }; }
   638  	#line 1 "not-num-const"
   639  	void __cgo_f_1_4(void) { static const double __cgo_undefined__4 = (foo); }
   640  	#line 1 "not-str-lit"
   641  	void __cgo_f_1_5(void) { static const char __cgo_undefined__5[] = (foo); }
   642  
   643  This program will not compile, but cgo can use the presence or absence
   644  of an error message on a given line to deduce the information it
   645  needs. The program is syntactically valid regardless of whether each
   646  name is a type or an ordinary identifier, so there will be no syntax
   647  errors that might stop parsing early.
   648  
   649  An error on not-declared:1 indicates that foo is undeclared.
   650  An error on not-type:1 indicates that foo is not a type (if declared at all, it is an identifier).
   651  An error on not-int-const:1 indicates that foo is not an integer constant.
   652  An error on not-num-const:1 indicates that foo is not a number constant.
   653  An error on not-str-lit:1 indicates that foo is not a string literal.
   654  An error on not-signed-int-const:1 indicates that foo is not a signed integer constant.
   655  
   656  The line number specifies the name involved. In the example, 1 is foo.
   657  
   658  Next, cgo must learn the details of each type, variable, function, or
   659  constant. It can do this by reading object files. If cgo has decided
   660  that t1 is a type, v2 and v3 are variables or functions, and i4, i5
   661  are integer constants, u6 is an unsigned integer constant, and f7 and f8
   662  are float constants, and s9 and s10 are string constants, it generates:
   663  
   664  	<preamble>
   665  	__typeof__(t1) *__cgo__1;
   666  	__typeof__(v2) *__cgo__2;
   667  	__typeof__(v3) *__cgo__3;
   668  	__typeof__(i4) *__cgo__4;
   669  	enum { __cgo_enum__4 = i4 };
   670  	__typeof__(i5) *__cgo__5;
   671  	enum { __cgo_enum__5 = i5 };
   672  	__typeof__(u6) *__cgo__6;
   673  	enum { __cgo_enum__6 = u6 };
   674  	__typeof__(f7) *__cgo__7;
   675  	__typeof__(f8) *__cgo__8;
   676  	__typeof__(s9) *__cgo__9;
   677  	__typeof__(s10) *__cgo__10;
   678  
   679  	long long __cgodebug_ints[] = {
   680  		0, // t1
   681  		0, // v2
   682  		0, // v3
   683  		i4,
   684  		i5,
   685  		u6,
   686  		0, // f7
   687  		0, // f8
   688  		0, // s9
   689  		0, // s10
   690  		1
   691  	};
   692  
   693  	double __cgodebug_floats[] = {
   694  		0, // t1
   695  		0, // v2
   696  		0, // v3
   697  		0, // i4
   698  		0, // i5
   699  		0, // u6
   700  		f7,
   701  		f8,
   702  		0, // s9
   703  		0, // s10
   704  		1
   705  	};
   706  
   707  	const char __cgodebug_str__9[] = s9;
   708  	const unsigned long long __cgodebug_strlen__9 = sizeof(s9)-1;
   709  	const char __cgodebug_str__10[] = s10;
   710  	const unsigned long long __cgodebug_strlen__10 = sizeof(s10)-1;
   711  
   712  and again invokes the system C compiler, to produce an object file
   713  containing debug information. Cgo parses the DWARF debug information
   714  for __cgo__N to learn the type of each identifier. (The types also
   715  distinguish functions from global variables.) Cgo reads the constant
   716  values from the __cgodebug_* from the object file's data segment.
   717  
   718  At this point cgo knows the meaning of each C.xxx well enough to start
   719  the translation process.
   720  
   721  Translating Go
   722  
   723  Given the input Go files x.go and y.go, cgo generates these source
   724  files:
   725  
   726  	x.cgo1.go       # for gc (cmd/compile)
   727  	y.cgo1.go       # for gc
   728  	_cgo_gotypes.go # for gc
   729  	_cgo_import.go  # for gc (if -dynout _cgo_import.go)
   730  	x.cgo2.c        # for gcc
   731  	y.cgo2.c        # for gcc
   732  	_cgo_defun.c    # for gcc (if -gccgo)
   733  	_cgo_export.c   # for gcc
   734  	_cgo_export.h   # for gcc
   735  	_cgo_main.c     # for gcc
   736  	_cgo_flags      # for build tool (if -gccgo)
   737  
   738  The file x.cgo1.go is a copy of x.go with the import "C" removed and
   739  references to C.xxx replaced with names like _Cfunc_xxx or _Ctype_xxx.
   740  The definitions of those identifiers, written as Go functions, types,
   741  or variables, are provided in _cgo_gotypes.go.
   742  
   743  Here is a _cgo_gotypes.go containing definitions for needed C types:
   744  
   745  	type _Ctype_char int8
   746  	type _Ctype_int int32
   747  	type _Ctype_void [0]byte
   748  
   749  The _cgo_gotypes.go file also contains the definitions of the
   750  functions. They all have similar bodies that invoke runtime·cgocall
   751  to make a switch from the Go runtime world to the system C (GCC-based)
   752  world.
   753  
   754  For example, here is the definition of _Cfunc_puts:
   755  
   756  	//go:cgo_import_static _cgo_be59f0f25121_Cfunc_puts
   757  	//go:linkname __cgofn__cgo_be59f0f25121_Cfunc_puts _cgo_be59f0f25121_Cfunc_puts
   758  	var __cgofn__cgo_be59f0f25121_Cfunc_puts byte
   759  	var _cgo_be59f0f25121_Cfunc_puts = unsafe.Pointer(&__cgofn__cgo_be59f0f25121_Cfunc_puts)
   760  
   761  	func _Cfunc_puts(p0 *_Ctype_char) (r1 _Ctype_int) {
   762  		_cgo_runtime_cgocall(_cgo_be59f0f25121_Cfunc_puts, uintptr(unsafe.Pointer(&p0)))
   763  		return
   764  	}
   765  
   766  The hexadecimal number is a hash of cgo's input, chosen to be
   767  deterministic yet unlikely to collide with other uses. The actual
   768  function _cgo_be59f0f25121_Cfunc_puts is implemented in a C source
   769  file compiled by gcc, the file x.cgo2.c:
   770  
   771  	void
   772  	_cgo_be59f0f25121_Cfunc_puts(void *v)
   773  	{
   774  		struct {
   775  			char* p0;
   776  			int r;
   777  			char __pad12[4];
   778  		} __attribute__((__packed__, __gcc_struct__)) *a = v;
   779  		a->r = puts((void*)a->p0);
   780  	}
   781  
   782  It extracts the arguments from the pointer to _Cfunc_puts's argument
   783  frame, invokes the system C function (in this case, puts), stores the
   784  result in the frame, and returns.
   785  
   786  Linking
   787  
   788  Once the _cgo_export.c and *.cgo2.c files have been compiled with gcc,
   789  they need to be linked into the final binary, along with the libraries
   790  they might depend on (in the case of puts, stdio). cmd/link has been
   791  extended to understand basic ELF files, but it does not understand ELF
   792  in the full complexity that modern C libraries embrace, so it cannot
   793  in general generate direct references to the system libraries.
   794  
   795  Instead, the build process generates an object file using dynamic
   796  linkage to the desired libraries. The main function is provided by
   797  _cgo_main.c:
   798  
   799  	int main() { return 0; }
   800  	void crosscall2(void(*fn)(void*), void *a, int c, uintptr_t ctxt) { }
   801  	uintptr_t _cgo_wait_runtime_init_done(void) { return 0; }
   802  	void _cgo_release_context(uintptr_t ctxt) { }
   803  	char* _cgo_topofstack(void) { return (char*)0; }
   804  	void _cgo_allocate(void *a, int c) { }
   805  	void _cgo_panic(void *a, int c) { }
   806  	void _cgo_reginit(void) { }
   807  
   808  The extra functions here are stubs to satisfy the references in the C
   809  code generated for gcc. The build process links this stub, along with
   810  _cgo_export.c and *.cgo2.c, into a dynamic executable and then lets
   811  cgo examine the executable. Cgo records the list of shared library
   812  references and resolved names and writes them into a new file
   813  _cgo_import.go, which looks like:
   814  
   815  	//go:cgo_dynamic_linker "/lib64/ld-linux-x86-64.so.2"
   816  	//go:cgo_import_dynamic puts puts#GLIBC_2.2.5 "libc.so.6"
   817  	//go:cgo_import_dynamic __libc_start_main __libc_start_main#GLIBC_2.2.5 "libc.so.6"
   818  	//go:cgo_import_dynamic stdout stdout#GLIBC_2.2.5 "libc.so.6"
   819  	//go:cgo_import_dynamic fflush fflush#GLIBC_2.2.5 "libc.so.6"
   820  	//go:cgo_import_dynamic _ _ "libpthread.so.0"
   821  	//go:cgo_import_dynamic _ _ "libc.so.6"
   822  
   823  In the end, the compiled Go package, which will eventually be
   824  presented to cmd/link as part of a larger program, contains:
   825  
   826  	_go_.o        # gc-compiled object for _cgo_gotypes.go, _cgo_import.go, *.cgo1.go
   827  	_all.o        # gcc-compiled object for _cgo_export.c, *.cgo2.c
   828  
   829  If there is an error generating the _cgo_import.go file, then, instead
   830  of adding _cgo_import.go to the package, the go tool adds an empty
   831  file named dynimportfail. The _cgo_import.go file is only needed when
   832  using internal linking mode, which is not the default when linking
   833  programs that use cgo (as described below). If the linker sees a file
   834  named dynimportfail it reports an error if it has been told to use
   835  internal linking mode. This approach is taken because generating
   836  _cgo_import.go requires doing a full C link of the package, which can
   837  fail for reasons that are irrelevant when using external linking mode.
   838  
   839  The final program will be a dynamic executable, so that cmd/link can avoid
   840  needing to process arbitrary .o files. It only needs to process the .o
   841  files generated from C files that cgo writes, and those are much more
   842  limited in the ELF or other features that they use.
   843  
   844  In essence, the _cgo_import.o file includes the extra linking
   845  directives that cmd/link is not sophisticated enough to derive from _all.o
   846  on its own. Similarly, the _all.o uses dynamic references to real
   847  system object code because cmd/link is not sophisticated enough to process
   848  the real code.
   849  
   850  The main benefits of this system are that cmd/link remains relatively simple
   851  (it does not need to implement a complete ELF and Mach-O linker) and
   852  that gcc is not needed after the package is compiled. For example,
   853  package net uses cgo for access to name resolution functions provided
   854  by libc. Although gcc is needed to compile package net, gcc is not
   855  needed to link programs that import package net.
   856  
   857  Runtime
   858  
   859  When using cgo, Go must not assume that it owns all details of the
   860  process. In particular it needs to coordinate with C in the use of
   861  threads and thread-local storage. The runtime package declares a few
   862  variables:
   863  
   864  	var (
   865  		iscgo             bool
   866  		_cgo_init         unsafe.Pointer
   867  		_cgo_thread_start unsafe.Pointer
   868  	)
   869  
   870  Any package using cgo imports "runtime/cgo", which provides
   871  initializations for these variables. It sets iscgo to true, _cgo_init
   872  to a gcc-compiled function that can be called early during program
   873  startup, and _cgo_thread_start to a gcc-compiled function that can be
   874  used to create a new thread, in place of the runtime's usual direct
   875  system calls.
   876  
   877  Internal and External Linking
   878  
   879  The text above describes "internal" linking, in which cmd/link parses and
   880  links host object files (ELF, Mach-O, PE, and so on) into the final
   881  executable itself. Keeping cmd/link simple means we cannot possibly
   882  implement the full semantics of the host linker, so the kinds of
   883  objects that can be linked directly into the binary is limited (other
   884  code can only be used as a dynamic library). On the other hand, when
   885  using internal linking, cmd/link can generate Go binaries by itself.
   886  
   887  In order to allow linking arbitrary object files without requiring
   888  dynamic libraries, cgo supports an "external" linking mode too. In
   889  external linking mode, cmd/link does not process any host object files.
   890  Instead, it collects all the Go code and writes a single go.o object
   891  file containing it. Then it invokes the host linker (usually gcc) to
   892  combine the go.o object file and any supporting non-Go code into a
   893  final executable. External linking avoids the dynamic library
   894  requirement but introduces a requirement that the host linker be
   895  present to create such a binary.
   896  
   897  Most builds both compile source code and invoke the linker to create a
   898  binary. When cgo is involved, the compile step already requires gcc, so
   899  it is not problematic for the link step to require gcc too.
   900  
   901  An important exception is builds using a pre-compiled copy of the
   902  standard library. In particular, package net uses cgo on most systems,
   903  and we want to preserve the ability to compile pure Go code that
   904  imports net without requiring gcc to be present at link time. (In this
   905  case, the dynamic library requirement is less significant, because the
   906  only library involved is libc.so, which can usually be assumed
   907  present.)
   908  
   909  This conflict between functionality and the gcc requirement means we
   910  must support both internal and external linking, depending on the
   911  circumstances: if net is the only cgo-using package, then internal
   912  linking is probably fine, but if other packages are involved, so that there
   913  are dependencies on libraries beyond libc, external linking is likely
   914  to work better. The compilation of a package records the relevant
   915  information to support both linking modes, leaving the decision
   916  to be made when linking the final binary.
   917  
   918  Linking Directives
   919  
   920  In either linking mode, package-specific directives must be passed
   921  through to cmd/link. These are communicated by writing //go: directives in a
   922  Go source file compiled by gc. The directives are copied into the .o
   923  object file and then processed by the linker.
   924  
   925  The directives are:
   926  
   927  //go:cgo_import_dynamic <local> [<remote> ["<library>"]]
   928  
   929  	In internal linking mode, allow an unresolved reference to
   930  	<local>, assuming it will be resolved by a dynamic library
   931  	symbol. The optional <remote> specifies the symbol's name and
   932  	possibly version in the dynamic library, and the optional "<library>"
   933  	names the specific library where the symbol should be found.
   934  
   935  	On AIX, the library pattern is slightly different. It must be
   936  	"lib.a/obj.o" with obj.o the member of this library exporting
   937  	this symbol.
   938  
   939  	In the <remote>, # or @ can be used to introduce a symbol version.
   940  
   941  	Examples:
   942  	//go:cgo_import_dynamic puts
   943  	//go:cgo_import_dynamic puts puts#GLIBC_2.2.5
   944  	//go:cgo_import_dynamic puts puts#GLIBC_2.2.5 "libc.so.6"
   945  
   946  	A side effect of the cgo_import_dynamic directive with a
   947  	library is to make the final binary depend on that dynamic
   948  	library. To get the dependency without importing any specific
   949  	symbols, use _ for local and remote.
   950  
   951  	Example:
   952  	//go:cgo_import_dynamic _ _ "libc.so.6"
   953  
   954  	For compatibility with current versions of SWIG,
   955  	#pragma dynimport is an alias for //go:cgo_import_dynamic.
   956  
   957  //go:cgo_dynamic_linker "<path>"
   958  
   959  	In internal linking mode, use "<path>" as the dynamic linker
   960  	in the final binary. This directive is only needed from one
   961  	package when constructing a binary; by convention it is
   962  	supplied by runtime/cgo.
   963  
   964  	Example:
   965  	//go:cgo_dynamic_linker "/lib/ld-linux.so.2"
   966  
   967  //go:cgo_export_dynamic <local> <remote>
   968  
   969  	In internal linking mode, put the Go symbol
   970  	named <local> into the program's exported symbol table as
   971  	<remote>, so that C code can refer to it by that name. This
   972  	mechanism makes it possible for C code to call back into Go or
   973  	to share Go's data.
   974  
   975  	For compatibility with current versions of SWIG,
   976  	#pragma dynexport is an alias for //go:cgo_export_dynamic.
   977  
   978  //go:cgo_import_static <local>
   979  
   980  	In external linking mode, allow unresolved references to
   981  	<local> in the go.o object file prepared for the host linker,
   982  	under the assumption that <local> will be supplied by the
   983  	other object files that will be linked with go.o.
   984  
   985  	Example:
   986  	//go:cgo_import_static puts_wrapper
   987  
   988  //go:cgo_export_static <local> <remote>
   989  
   990  	In external linking mode, put the Go symbol
   991  	named <local> into the program's exported symbol table as
   992  	<remote>, so that C code can refer to it by that name. This
   993  	mechanism makes it possible for C code to call back into Go or
   994  	to share Go's data.
   995  
   996  //go:cgo_ldflag "<arg>"
   997  
   998  	In external linking mode, invoke the host linker (usually gcc)
   999  	with "<arg>" as a command-line argument following the .o files.
  1000  	Note that the arguments are for "gcc", not "ld".
  1001  
  1002  	Example:
  1003  	//go:cgo_ldflag "-lpthread"
  1004  	//go:cgo_ldflag "-L/usr/local/sqlite3/lib"
  1005  
  1006  A package compiled with cgo will include directives for both
  1007  internal and external linking; the linker will select the appropriate
  1008  subset for the chosen linking mode.
  1009  
  1010  Example
  1011  
  1012  As a simple example, consider a package that uses cgo to call C.sin.
  1013  The following code will be generated by cgo:
  1014  
  1015  	// compiled by gc
  1016  
  1017  	//go:cgo_ldflag "-lm"
  1018  
  1019  	type _Ctype_double float64
  1020  
  1021  	//go:cgo_import_static _cgo_gcc_Cfunc_sin
  1022  	//go:linkname __cgo_gcc_Cfunc_sin _cgo_gcc_Cfunc_sin
  1023  	var __cgo_gcc_Cfunc_sin byte
  1024  	var _cgo_gcc_Cfunc_sin = unsafe.Pointer(&__cgo_gcc_Cfunc_sin)
  1025  
  1026  	func _Cfunc_sin(p0 _Ctype_double) (r1 _Ctype_double) {
  1027  		_cgo_runtime_cgocall(_cgo_gcc_Cfunc_sin, uintptr(unsafe.Pointer(&p0)))
  1028  		return
  1029  	}
  1030  
  1031  	// compiled by gcc, into foo.cgo2.o
  1032  
  1033  	void
  1034  	_cgo_gcc_Cfunc_sin(void *v)
  1035  	{
  1036  		struct {
  1037  			double p0;
  1038  			double r;
  1039  		} __attribute__((__packed__)) *a = v;
  1040  		a->r = sin(a->p0);
  1041  	}
  1042  
  1043  What happens at link time depends on whether the final binary is linked
  1044  using the internal or external mode. If other packages are compiled in
  1045  "external only" mode, then the final link will be an external one.
  1046  Otherwise the link will be an internal one.
  1047  
  1048  The linking directives are used according to the kind of final link
  1049  used.
  1050  
  1051  In internal mode, cmd/link itself processes all the host object files, in
  1052  particular foo.cgo2.o. To do so, it uses the cgo_import_dynamic and
  1053  cgo_dynamic_linker directives to learn that the otherwise undefined
  1054  reference to sin in foo.cgo2.o should be rewritten to refer to the
  1055  symbol sin with version GLIBC_2.2.5 from the dynamic library
  1056  "libm.so.6", and the binary should request "/lib/ld-linux.so.2" as its
  1057  runtime dynamic linker.
  1058  
  1059  In external mode, cmd/link does not process any host object files, in
  1060  particular foo.cgo2.o. It links together the gc-generated object
  1061  files, along with any other Go code, into a go.o file. While doing
  1062  that, cmd/link will discover that there is no definition for
  1063  _cgo_gcc_Cfunc_sin, referred to by the gc-compiled source file. This
  1064  is okay, because cmd/link also processes the cgo_import_static directive and
  1065  knows that _cgo_gcc_Cfunc_sin is expected to be supplied by a host
  1066  object file, so cmd/link does not treat the missing symbol as an error when
  1067  creating go.o. Indeed, the definition for _cgo_gcc_Cfunc_sin will be
  1068  provided to the host linker by foo2.cgo.o, which in turn will need the
  1069  symbol 'sin'. cmd/link also processes the cgo_ldflag directives, so that it
  1070  knows that the eventual host link command must include the -lm
  1071  argument, so that the host linker will be able to find 'sin' in the
  1072  math library.
  1073  
  1074  cmd/link Command Line Interface
  1075  
  1076  The go command and any other Go-aware build systems invoke cmd/link
  1077  to link a collection of packages into a single binary. By default, cmd/link will
  1078  present the same interface it does today:
  1079  
  1080  	cmd/link main.a
  1081  
  1082  produces a file named a.out, even if cmd/link does so by invoking the host
  1083  linker in external linking mode.
  1084  
  1085  By default, cmd/link will decide the linking mode as follows: if the only
  1086  packages using cgo are those on a list of known standard library
  1087  packages (net, os/user, runtime/cgo), cmd/link will use internal linking
  1088  mode. Otherwise, there are non-standard cgo packages involved, and cmd/link
  1089  will use external linking mode. The first rule means that a build of
  1090  the godoc binary, which uses net but no other cgo, can run without
  1091  needing gcc available. The second rule means that a build of a
  1092  cgo-wrapped library like sqlite3 can generate a standalone executable
  1093  instead of needing to refer to a dynamic library. The specific choice
  1094  can be overridden using a command line flag: cmd/link -linkmode=internal or
  1095  cmd/link -linkmode=external.
  1096  
  1097  In an external link, cmd/link will create a temporary directory, write any
  1098  host object files found in package archives to that directory (renamed
  1099  to avoid conflicts), write the go.o file to that directory, and invoke
  1100  the host linker. The default value for the host linker is $CC, split
  1101  into fields, or else "gcc". The specific host linker command line can
  1102  be overridden using command line flags: cmd/link -extld=clang
  1103  -extldflags='-ggdb -O3'. If any package in a build includes a .cc or
  1104  other file compiled by the C++ compiler, the go tool will use the
  1105  -extld option to set the host linker to the C++ compiler.
  1106  
  1107  These defaults mean that Go-aware build systems can ignore the linking
  1108  changes and keep running plain 'cmd/link' and get reasonable results, but
  1109  they can also control the linking details if desired.
  1110  
  1111  */
  1112  

View as plain text