distinguish functions from global variables.) Cgo reads the constant
values from the __cgodebug_* from the object file's data segment.

At this point cgo knows the meaning of each C.xxx well enough to start
the translation process.

Translating Go

Given the input Go files x.go and y.go, cgo generates these source
files:

	x.cgo1.go       # for gc (cmd/compile)
	y.cgo1.go       # for gc
	_cgo_gotypes.go # for gc

	// right place when we start editing the AST behind its back,
	// so we use ast1 to look for the doc comments on import "C"
	// and on exported functions, and we use ast2 for translating
	// and reprinting.
	// In cgo mode, we ignore ast2 and just apply edits directly
	// the text behind ast1. In godefs mode we modify and print ast2.
	ast1 := parse(abspath, src, parser.SkipObjectResolution|parser.ParseComments)

Cgo's most interesting translation is for functions.  If xxx is a C
function, then cgo rewrites C.xxx into a new function _C_xxx that
calls the C xxx in a standard pthread.  The new function translates
its arguments, calls xxx, and translates the return value.

Translation of parameters and the return value follows the type
translation above except that arrays passed as parameters translate

			// This was an extension in the old C version of 'go tool addr2line'
			// and is probably not used by anyone, but recognize the syntax.
			// We don't have an implementation.
			fmt.Fprintf(stdout, "!reverse translation not implemented\n")
			continue
		}
		pc, _ := strconv.ParseUint(strings.TrimPrefix(p, "0x"), 16, 64)
		file, line, fn := tab.PCToLine(pc)
		name := "?"
		if fn != nil {
			name = fn.Name
		} else {

			t.Go = c.complex64
		case 16:
			t.Go = c.complex128
		}
		if t.Align = t.Size / 2; t.Align >= c.ptrSize {
			t.Align = c.ptrSize
		}

	case *dwarf.FuncType:
		// No attempt at translation: would enable calls
		// directly between worlds, but we need to moderate those.
		t.Go = c.uintptr
		t.Align = c.ptrSize

	case *dwarf.IntType:
		if dt.BitSize > 0 {

provided it can be proved that <code>*shared</code> will not fault on access,
because the potential added read will not affect any existing concurrent reads or writes.
On the other hand, the rewrite would not be valid in a source-to-source translator.
</p>

<h2 id="conclusion">Conclusion</h2>

<p>
Go programmers writing data-race-free programs can rely on
sequentially consistent execution of those programs,

    The <a href="https://go.dev/wiki/LoopvarExperiment#my-test-fails-with-the-change-how-can-i-debug-it">transition support tooling</a>
    described in the proposal continues to work in the same way it did in Go 1.21.
  </li>
  <li>
    "For" loops may now range over integers.