doc/ext_ffi_semantics.html

   1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
   2 <html>
   3 <head>
   4 <title>FFI Semantics</title>
   5 <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
   6 <meta name="Author" content="Mike Pall">
   7 <meta name="Copyright" content="Copyright (C) 2005-2012, Mike Pall">
   8 <meta name="Language" content="en">
   9 <link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
  10 <link rel="stylesheet" type="text/css" href="bluequad-print.css" media="print">
  11 <style type="text/css">
  12 table.convtable { line-height: 1.2; }
  13 tr.convhead td { font-weight: bold; }
  14 td.convin { width: 11em; }
  15 td.convop { font-style: italic; width: 16em; }
  16 </style>
  17 </head>
  18 <body>
  19 <div id="site">
  20 <a href="http://luajit.org"><span>Lua<span id="logo">JIT</span></span></a>
  21 </div>
  22 <div id="head">
  23 <h1>FFI Semantics</h1>
  24 </div>
  25 <div id="nav">
  26 <ul><li>
  27 <a href="luajit.html">LuaJIT</a>
  28 <ul><li>
  29 <a href="install.html">Installation</a>
  30 </li><li>
  31 <a href="running.html">Running</a>
  32 </li></ul>
  33 </li><li>
  34 <a href="extensions.html">Extensions</a>
  35 <ul><li>
  36 <a href="ext_ffi.html">FFI Library</a>
  37 <ul><li>
  38 <a href="ext_ffi_tutorial.html">FFI Tutorial</a>
  39 </li><li>
  40 <a href="ext_ffi_api.html">ffi.* API</a>
  41 </li><li>
  42 <a class="current" href="ext_ffi_semantics.html">FFI Semantics</a>
  43 </li></ul>
  44 </li><li>
  45 <a href="ext_jit.html">jit.* Library</a>
  46 </li><li>
  47 <a href="ext_c_api.html">Lua/C API</a>
  48 </li></ul>
  49 </li><li>
  50 <a href="status.html">Status</a>
  51 <ul><li>
  52 <a href="changes.html">Changes</a>
  53 </li></ul>
  54 </li><li>
  55 <a href="faq.html">FAQ</a>
  56 </li><li>
  57 <a href="http://luajit.org/performance.html">Performance <span class="ext">&raquo;</span></a>
  58 </li><li>
  59 <a href="http://luajit.org/download.html">Download <span class="ext">&raquo;</span></a>
  60 </li><li>
  61 <a href="http://wiki.luajit.org/">Wiki <span class="ext">&raquo;</span></a>
  62 </li><li>
  63 <a href="http://luajit.org/list.html">Mailing List <span class="ext">&raquo;</span></a>
  64 </li></ul>
  65 </div>
  66 <div id="main">
  67 <p>
  68 This page describes the detailed semantics underlying the FFI library
  69 and its interaction with both Lua and C&nbsp;code.
  70 </p>
  71 <p>
  72 Given that the FFI library is designed to interface with C&nbsp;code
  73 and that declarations can be written in plain C&nbsp;syntax, <b>it
  74 closely follows the C&nbsp;language semantics</b>, wherever possible.
  75 Some minor concessions are needed for smoother interoperation with Lua
  76 language semantics.
  77 </p>
  78 <p>
  79 Please don't be overwhelmed by the contents of this page &mdash; this
  80 is a reference and you may need to consult it, if in doubt. It doesn't
  81 hurt to skim this page, but most of the semantics "just work" as you'd
  82 expect them to work. It should be straightforward to write
  83 applications using the LuaJIT FFI for developers with a C or C++
  84 background.
  85 </p>
  86 <p class="indent" style="color: #c00000;">
  87 Please note: this doesn't comprise the final specification for the FFI
  88 semantics, yet. Some semantics may need to be changed, based on your
  89 feedback. Please <a href="contact.html">report</a> any problems you may
  90 encounter or any improvements you'd like to see &mdash; thank you!
  91 </p>
  92
  93 <h2 id="clang">C Language Support</h2>
  94 <p>
  95 The FFI library has a built-in C&nbsp;parser with a minimal memory
  96 footprint. It's used by the <a href="ext_ffi_api.html">ffi.* library
  97 functions</a> to declare C&nbsp;types or external symbols.
  98 </p>
  99 <p>
 100 It's only purpose is to parse C&nbsp;declarations, as found e.g. in
 101 C&nbsp;header files. Although it does evaluate constant expressions,
 102 it's <em>not</em> a C&nbsp;compiler. The body of <tt>inline</tt>
 103 C&nbsp;function definitions is simply ignored.
 104 </p>
 105 <p>
 106 Also, this is <em>not</em> a validating C&nbsp;parser. It expects and
 107 accepts correctly formed C&nbsp;declarations, but it may choose to
 108 ignore bad declarations or show rather generic error messages. If in
 109 doubt, please check the input against your favorite C&nbsp;compiler.
 110 </p>
 111 <p>
 112 The C&nbsp;parser complies to the <b>C99 language standard</b> plus
 113 the following extensions:
 114 </p>
 115 <ul>
 116
 117 <li>The <tt>'\e'</tt> escape in character and string literals.</li>
 118
 119 <li>The C99/C++ boolean type, declared with the keywords <tt>bool</tt>
 120 or <tt>_Bool</tt>.</li>
 121
 122 <li>Complex numbers, declared with the keywords <tt>complex</tt> or
 123 <tt>_Complex</tt>.</li>
 124
 125 <li>Two complex number types: <tt>complex</tt> (aka
 126 <tt>complex&nbsp;double</tt>) and <tt>complex&nbsp;float</tt>.</li>
 127
 128 <li>Vector types, declared with the GCC <tt>mode</tt> or
 129 <tt>vector_size</tt> attribute.</li>
 130
 131 <li>Unnamed ('transparent') <tt>struct</tt>/<tt>union</tt> fields
 132 inside a <tt>struct</tt>/<tt>union</tt>.</li>
 133
 134 <li>Incomplete <tt>enum</tt> declarations, handled like incomplete
 135 <tt>struct</tt> declarations.</li>
 136
 137 <li>Unnamed <tt>enum</tt> fields inside a
 138 <tt>struct</tt>/<tt>union</tt>. This is similar to a scoped C++
 139 <tt>enum</tt>, except that declared constants are visible in the
 140 global namespace, too.</li>
 141
 142 <li>Scoped <tt>static&nbsp;const</tt> declarations inside a
 143 <tt>struct</tt>/<tt>union</tt> (from C++).</li>
 144
 145 <li>Zero-length arrays (<tt>[0]</tt>), empty
 146 <tt>struct</tt>/<tt>union</tt>, variable-length arrays (VLA,
 147 <tt>[?]</tt>) and variable-length structs (VLS, with a trailing
 148 VLA).</li>
 149
 150 <li>C++ reference types (<tt>int&nbsp;&amp;x</tt>).</li>
 151
 152 <li>Alternate GCC keywords with '<tt>__</tt>', e.g.
 153 <tt>__const__</tt>.</li>
 154
 155 <li>GCC <tt>__attribute__</tt> with the following attributes:
 156 <tt>aligned</tt>, <tt>packed</tt>, <tt>mode</tt>,
 157 <tt>vector_size</tt>, <tt>cdecl</tt>, <tt>fastcall</tt>,
 158 <tt>stdcall</tt>.</li>
 159
 160 <li>The GCC <tt>__extension__</tt> keyword and the GCC
 161 <tt>__alignof__</tt> operator.</li>
 162
 163 <li>GCC <tt>__asm__("symname")</tt> symbol name redirection for
 164 function declarations.</li>
 165
 166 <li>MSVC keywords for fixed-length types: <tt>__int8</tt>,
 167 <tt>__int16</tt>, <tt>__int32</tt> and <tt>__int64</tt>.</li>
 168
 169 <li>MSVC <tt>__cdecl</tt>, <tt>__fastcall</tt>, <tt>__stdcall</tt>,
 170 <tt>__ptr32</tt>, <tt>__ptr64</tt>, <tt>__declspec(align(n))</tt>
 171 and <tt>#pragma&nbsp;pack</tt>.</li>
 172
 173 <li>All other GCC/MSVC-specific attributes are ignored.</li>
 174
 175 </ul>
 176 <p>
 177 The following C&nbsp;types are pre-defined by the C&nbsp;parser (like
 178 a <tt>typedef</tt>, except re-declarations will be ignored):
 179 </p>
 180 <ul>
 181
 182 <li>Vararg handling: <tt>va_list</tt>, <tt>__builtin_va_list</tt>,
 183 <tt>__gnuc_va_list</tt>.</li>
 184
 185 <li>From <tt>&lt;stddef.h&gt;</tt>: <tt>ptrdiff_t</tt>,
 186 <tt>size_t</tt>, <tt>wchar_t</tt>.</li>
 187
 188 <li>From <tt>&lt;stdint.h&gt;</tt>: <tt>int8_t</tt>, <tt>int16_t</tt>,
 189 <tt>int32_t</tt>, <tt>int64_t</tt>, <tt>uint8_t</tt>,
 190 <tt>uint16_t</tt>, <tt>uint32_t</tt>, <tt>uint64_t</tt>,
 191 <tt>intptr_t</tt>, <tt>uintptr_t</tt>.</li>
 192
 193 </ul>
 194 <p>
 195 You're encouraged to use these types in preference to the
 196 compiler-specific extensions or the target-dependent standard types.
 197 E.g. <tt>char</tt> differs in signedness and <tt>long</tt> differs in
 198 size, depending on the target architecture and platform ABI.
 199 </p>
 200 <p>
 201 The following C&nbsp;features are <b>not</b> supported:
 202 </p>
 203 <ul>
 204
 205 <li>A declaration must always have a type specifier; it doesn't
 206 default to an <tt>int</tt> type.</li>
 207
 208 <li>Old-style empty function declarations (K&amp;R) are not allowed.
 209 All C&nbsp;functions must have a proper prototype declaration. A
 210 function declared without parameters (<tt>int&nbsp;foo();</tt>) is
 211 treated as a function taking zero arguments, like in C++.</li>
 212
 213 <li>The <tt>long double</tt> C&nbsp;type is parsed correctly, but
 214 there's no support for the related conversions, accesses or arithmetic
 215 operations.</li>
 216
 217 <li>Wide character strings and character literals are not
 218 supported.</li>
 219
 220 <li><a href="#status">See below</a> for features that are currently
 221 not implemented.</li>
 222
 223 </ul>
 224
 225 <h2 id="convert">C Type Conversion Rules</h2>
 226
 227 <h3 id="convert_tolua">Conversions from C&nbsp;types to Lua objects</h3>
 228 <p>
 229 These conversion rules apply for <em>read accesses</em> to
 230 C&nbsp;types: indexing pointers, arrays or
 231 <tt>struct</tt>/<tt>union</tt> types; reading external variables or
 232 constant values; retrieving return values from C&nbsp;calls:
 233 </p>
 234 <table class="convtable">
 235 <tr class="convhead">
 236 <td class="convin">Input</td>
 237 <td class="convop">Conversion</td>
 238 <td class="convout">Output</td>
 239 </tr>
 240 <tr class="odd separate">
 241 <td class="convin"><tt>int8_t</tt>, <tt>int16_t</tt></td><td class="convop">&rarr;<sup>sign-ext</sup> <tt>int32_t</tt> &rarr; <tt>double</tt></td><td class="convout">number</td></tr>
 242 <tr class="even">
 243 <td class="convin"><tt>uint8_t</tt>, <tt>uint16_t</tt></td><td class="convop">&rarr;<sup>zero-ext</sup> <tt>int32_t</tt> &rarr; <tt>double</tt></td><td class="convout">number</td></tr>
 244 <tr class="odd">
 245 <td class="convin"><tt>int32_t</tt>, <tt>uint32_t</tt></td><td class="convop">&rarr; <tt>double</tt></td><td class="convout">number</td></tr>
 246 <tr class="even">
 247 <td class="convin"><tt>int64_t</tt>, <tt>uint64_t</tt></td><td class="convop">boxed value</td><td class="convout">64 bit int cdata</td></tr>
 248 <tr class="odd separate">
 249 <td class="convin"><tt>double</tt>, <tt>float</tt></td><td class="convop">&rarr; <tt>double</tt></td><td class="convout">number</td></tr>
 250 <tr class="even separate">
 251 <td class="convin"><tt>bool</tt></td><td class="convop">0 &rarr; <tt>false</tt>, otherwise <tt>true</tt></td><td class="convout">boolean</td></tr>
 252 <tr class="odd separate">
 253 <td class="convin"><tt>enum</tt></td><td class="convop">boxed value</td><td class="convout">enum cdata</td></tr>
 254 <tr class="even">
 255 <td class="convin">Complex number</td><td class="convop">boxed value</td><td class="convout">complex cdata</td></tr>
 256 <tr class="odd">
 257 <td class="convin">Vector</td><td class="convop">boxed value</td><td class="convout">vector cdata</td></tr>
 258 <tr class="even">
 259 <td class="convin">Pointer</td><td class="convop">boxed value</td><td class="convout">pointer cdata</td></tr>
 260 <tr class="odd separate">
 261 <td class="convin">Array</td><td class="convop">boxed reference</td><td class="convout">reference cdata</td></tr>
 262 <tr class="even">
 263 <td class="convin"><tt>struct</tt>/<tt>union</tt></td><td class="convop">boxed reference</td><td class="convout">reference cdata</td></tr>
 264 </table>
 265 <p>
 266 Bitfields are treated like their underlying type.
 267 </p>
 268 <p>
 269 Reference types are dereferenced <em>before</em> a conversion can take
 270 place &mdash; the conversion is applied to the C&nbsp;type pointed to
 271 by the reference.
 272 </p>
 273
 274 <h3 id="convert_fromlua">Conversions from Lua objects to C&nbsp;types</h3>
 275 <p>
 276 These conversion rules apply for <em>write accesses</em> to
 277 C&nbsp;types: indexing pointers, arrays or
 278 <tt>struct</tt>/<tt>union</tt> types; initializing cdata objects;
 279 casts to C&nbsp;types; writing to external variables; passing
 280 arguments to C&nbsp;calls:
 281 </p>
 282 <table class="convtable">
 283 <tr class="convhead">
 284 <td class="convin">Input</td>
 285 <td class="convop">Conversion</td>
 286 <td class="convout">Output</td>
 287 </tr>
 288 <tr class="odd separate">
 289 <td class="convin">number</td><td class="convop">&rarr;</td><td class="convout"><tt>double</tt></td></tr>
 290 <tr class="even">
 291 <td class="convin">boolean</td><td class="convop"><tt>false</tt> &rarr; 0, <tt>true</tt> &rarr; 1</td><td class="convout"><tt>bool</tt></td></tr>
 292 <tr class="odd separate">
 293 <td class="convin">nil</td><td class="convop"><tt>NULL</tt> &rarr;</td><td class="convout"><tt>(void *)</tt></td></tr>
 294 <tr class="even">
 295 <td class="convin">lightuserdata</td><td class="convop">lightuserdata address &rarr;</td><td class="convout"><tt>(void *)</tt></td></tr>
 296 <tr class="odd">
 297 <td class="convin">userdata</td><td class="convop">userdata payload &rarr;</td><td class="convout"><tt>(void *)</tt></td></tr>
 298 <tr class="even">
 299 <td class="convin">io.* file</td><td class="convop">get FILE * handle &rarr;</td><td class="convout"><tt>(void *)</tt></td></tr>
 300 <tr class="odd separate">
 301 <td class="convin">string</td><td class="convop">match against <tt>enum</tt> constant</td><td class="convout"><tt>enum</tt></td></tr>
 302 <tr class="even">
 303 <td class="convin">string</td><td class="convop">copy string data + zero-byte</td><td class="convout"><tt>int8_t[]</tt>, <tt>uint8_t[]</tt></td></tr>
 304 <tr class="odd">
 305 <td class="convin">string</td><td class="convop">string data &rarr;</td><td class="convout"><tt>const char[]</tt></td></tr>
 306 <tr class="even separate">
 307 <td class="convin">function</td><td class="convop"><a href="#callback">create callback</a> &rarr;</td><td class="convout">C function type</td></tr>
 308 <tr class="odd separate">
 309 <td class="convin">table</td><td class="convop"><a href="#init_table">table initializer</a></td><td class="convout">Array</td></tr>
 310 <tr class="even">
 311 <td class="convin">table</td><td class="convop"><a href="#init_table">table initializer</a></td><td class="convout"><tt>struct</tt>/<tt>union</tt></td></tr>
 312 <tr class="odd separate">
 313 <td class="convin">cdata</td><td class="convop">cdata payload &rarr;</td><td class="convout">C type</td></tr>
 314 </table>
 315 <p>
 316 If the result type of this conversion doesn't match the
 317 C&nbsp;type of the destination, the
 318 <a href="#convert_between">conversion rules between C&nbsp;types</a>
 319 are applied.
 320 </p>
 321 <p>
 322 Reference types are immutable after initialization ("no re-seating of
 323 references"). For initialization purposes or when passing values to
 324 reference parameters, they are treated like pointers. Note that unlike
 325 in C++, there's no way to implement automatic reference generation of
 326 variables under the Lua language semantics. If you want to call a
 327 function with a reference parameter, you need to explicitly pass a
 328 one-element array.
 329 </p>
 330
 331 <h3 id="convert_between">Conversions between C&nbsp;types</h3>
 332 <p>
 333 These conversion rules are more or less the same as the standard
 334 C&nbsp;conversion rules. Some rules only apply to casts, or require
 335 pointer or type compatibility:
 336 </p>
 337 <table class="convtable">
 338 <tr class="convhead">
 339 <td class="convin">Input</td>
 340 <td class="convop">Conversion</td>
 341 <td class="convout">Output</td>
 342 </tr>
 343 <tr class="odd separate">
 344 <td class="convin">Signed integer</td><td class="convop">&rarr;<sup>narrow or sign-extend</sup></td><td class="convout">Integer</td></tr>
 345 <tr class="even">
 346 <td class="convin">Unsigned integer</td><td class="convop">&rarr;<sup>narrow or zero-extend</sup></td><td class="convout">Integer</td></tr>
 347 <tr class="odd">
 348 <td class="convin">Integer</td><td class="convop">&rarr;<sup>round</sup></td><td class="convout"><tt>double</tt>, <tt>float</tt></td></tr>
 349 <tr class="even">
 350 <td class="convin"><tt>double</tt>, <tt>float</tt></td><td class="convop">&rarr;<sup>trunc</sup> <tt>int32_t</tt> &rarr;<sup>narrow</sup></td><td class="convout"><tt>(u)int8_t</tt>, <tt>(u)int16_t</tt></td></tr>
 351 <tr class="odd">
 352 <td class="convin"><tt>double</tt>, <tt>float</tt></td><td class="convop">&rarr;<sup>trunc</sup></td><td class="convout"><tt>(u)int32_t</tt>, <tt>(u)int64_t</tt></td></tr>
 353 <tr class="even">
 354 <td class="convin"><tt>double</tt>, <tt>float</tt></td><td class="convop">&rarr;<sup>round</sup></td><td class="convout"><tt>float</tt>, <tt>double</tt></td></tr>
 355 <tr class="odd separate">
 356 <td class="convin">Number</td><td class="convop">n == 0 &rarr; 0, otherwise 1</td><td class="convout"><tt>bool</tt></td></tr>
 357 <tr class="even">
 358 <td class="convin"><tt>bool</tt></td><td class="convop"><tt>false</tt> &rarr; 0, <tt>true</tt> &rarr; 1</td><td class="convout">Number</td></tr>
 359 <tr class="odd separate">
 360 <td class="convin">Complex number</td><td class="convop">convert real part</td><td class="convout">Number</td></tr>
 361 <tr class="even">
 362 <td class="convin">Number</td><td class="convop">convert real part, imag = 0</td><td class="convout">Complex number</td></tr>
 363 <tr class="odd">
 364 <td class="convin">Complex number</td><td class="convop">convert real and imag part</td><td class="convout">Complex number</td></tr>
 365 <tr class="even separate">
 366 <td class="convin">Number</td><td class="convop">convert scalar and replicate</td><td class="convout">Vector</td></tr>
 367 <tr class="odd">
 368 <td class="convin">Vector</td><td class="convop">copy (same size)</td><td class="convout">Vector</td></tr>
 369 <tr class="even separate">
 370 <td class="convin"><tt>struct</tt>/<tt>union</tt></td><td class="convop">take base address (compat)</td><td class="convout">Pointer</td></tr>
 371 <tr class="odd">
 372 <td class="convin">Array</td><td class="convop">take base address (compat)</td><td class="convout">Pointer</td></tr>
 373 <tr class="even">
 374 <td class="convin">Function</td><td class="convop">take function address</td><td class="convout">Function pointer</td></tr>
 375 <tr class="odd separate">
 376 <td class="convin">Number</td><td class="convop">convert via <tt>uintptr_t</tt> (cast)</td><td class="convout">Pointer</td></tr>
 377 <tr class="even">
 378 <td class="convin">Pointer</td><td class="convop">convert address (compat/cast)</td><td class="convout">Pointer</td></tr>
 379 <tr class="odd">
 380 <td class="convin">Pointer</td><td class="convop">convert address (cast)</td><td class="convout">Integer</td></tr>
 381 <tr class="even">
 382 <td class="convin">Array</td><td class="convop">convert base address (cast)</td><td class="convout">Integer</td></tr>
 383 <tr class="odd separate">
 384 <td class="convin">Array</td><td class="convop">copy (compat)</td><td class="convout">Array</td></tr>
 385 <tr class="even">
 386 <td class="convin"><tt>struct</tt>/<tt>union</tt></td><td class="convop">copy (identical type)</td><td class="convout"><tt>struct</tt>/<tt>union</tt></td></tr>
 387 </table>
 388 <p>
 389 Bitfields or <tt>enum</tt> types are treated like their underlying
 390 type.
 391 </p>
 392 <p>
 393 Conversions not listed above will raise an error. E.g. it's not
 394 possible to convert a pointer to a complex number or vice versa.
 395 </p>
 396
 397 <h3 id="convert_vararg">Conversions for vararg C&nbsp;function arguments</h3>
 398 <p>
 399 The following default conversion rules apply when passing Lua objects
 400 to the variable argument part of vararg C&nbsp;functions:
 401 </p>
 402 <table class="convtable">
 403 <tr class="convhead">
 404 <td class="convin">Input</td>
 405 <td class="convop">Conversion</td>
 406 <td class="convout">Output</td>
 407 </tr>
 408 <tr class="odd separate">
 409 <td class="convin">number</td><td class="convop">&rarr;</td><td class="convout"><tt>double</tt></td></tr>
 410 <tr class="even">
 411 <td class="convin">boolean</td><td class="convop"><tt>false</tt> &rarr; 0, <tt>true</tt> &rarr; 1</td><td class="convout"><tt>bool</tt></td></tr>
 412 <tr class="odd separate">
 413 <td class="convin">nil</td><td class="convop"><tt>NULL</tt> &rarr;</td><td class="convout"><tt>(void *)</tt></td></tr>
 414 <tr class="even">
 415 <td class="convin">userdata</td><td class="convop">userdata payload &rarr;</td><td class="convout"><tt>(void *)</tt></td></tr>
 416 <tr class="odd">
 417 <td class="convin">lightuserdata</td><td class="convop">lightuserdata address &rarr;</td><td class="convout"><tt>(void *)</tt></td></tr>
 418 <tr class="even separate">
 419 <td class="convin">string</td><td class="convop">string data &rarr;</td><td class="convout"><tt>const char *</tt></td></tr>
 420 <tr class="odd separate">
 421 <td class="convin"><tt>float</tt> cdata</td><td class="convop">&rarr;</td><td class="convout"><tt>double</tt></td></tr>
 422 <tr class="even">
 423 <td class="convin">Array cdata</td><td class="convop">take base address</td><td class="convout">Element pointer</td></tr>
 424 <tr class="odd">
 425 <td class="convin"><tt>struct</tt>/<tt>union</tt> cdata</td><td class="convop">take base address</td><td class="convout"><tt>struct</tt>/<tt>union</tt> pointer</td></tr>
 426 <tr class="even">
 427 <td class="convin">Function cdata</td><td class="convop">take function address</td><td class="convout">Function pointer</td></tr>
 428 <tr class="odd">
 429 <td class="convin">Any other cdata</td><td class="convop">no conversion</td><td class="convout">C type</td></tr>
 430 </table>
 431 <p>
 432 To pass a Lua object, other than a cdata object, as a specific type,
 433 you need to override the conversion rules: create a temporary cdata
 434 object with a constructor or a cast and initialize it with the value
 435 to pass:
 436 </p>
 437 <p>
 438 Assuming <tt>x</tt> is a Lua number, here's how to pass it as an
 439 integer to a vararg function:
 440 </p>
 441 <pre class="code">
 442 ffi.cdef[[
 443 int printf(const char *fmt, ...);
 444 ]]
 445 ffi.C.printf("integer value: %d\n", ffi.new("int", x))
 446 </pre>
 447 <p>
 448 If you don't do this, the default Lua number &rarr; <tt>double</tt>
 449 conversion rule applies. A vararg C&nbsp;function expecting an integer
 450 will see a garbled or uninitialized value.
 451 </p>
 452
 453 <h2 id="init">Initializers</h2>
 454 <p>
 455 Creating a cdata object with
 456 <a href="ext_ffi_api.html#ffi_new"><tt>ffi.new()</tt></a> or the
 457 equivalent constructor syntax always initializes its contents, too.
 458 Different rules apply, depending on the number of optional
 459 initializers and the C&nbsp;types involved:
 460 </p>
 461 <ul>
 462 <li>If no initializers are given, the object is filled with zero bytes.</li>
 463
 464 <li>Scalar types (numbers and pointers) accept a single initializer.
 465 The Lua object is <a href="#convert_fromlua">converted to the scalar
 466 C&nbsp;type</a>.</li>
 467
 468 <li>Valarrays (complex numbers and vectors) are treated like scalars
 469 when a single initializer is given. Otherwise they are treated like
 470 regular arrays.</li>
 471
 472 <li>Aggregate types (arrays and structs) accept either a single
 473 <a href="#init_table">table initializer</a> or a flat list of
 474 initializers.</li>
 475
 476 <li>The elements of an array are initialized, starting at index zero.
 477 If a single initializer is given for an array, it's repeated for all
 478 remaining elements. This doesn't happen if two or more initializers
 479 are given: all remaining uninitialized elements are filled with zero
 480 bytes.</li>
 481
 482 <li>Byte arrays may also be initialized with a Lua string. This copies
 483 the whole string plus a terminating zero-byte. The copy stops early only
 484 if the array has a known, fixed size.</li>
 485
 486 <li>The fields of a <tt>struct</tt> are initialized in the order of
 487 their declaration. Uninitialized fields are filled with zero
 488 bytes.</li>
 489
 490 <li>Only the first field of a <tt>union</tt> can be initialized with a
 491 flat initializer.</li>
 492
 493 <li>Elements or fields which are aggregates themselves are initialized
 494 with a <em>single</em> initializer, but this may be a table
 495 initializer or a compatible aggregate.</li>
 496
 497 <li>Excess initializers cause an error.</li>
 498
 499 </ul>
 500
 501 <h2 id="init_table">Table Initializers</h2>
 502 <p>
 503 The following rules apply if a Lua table is used to initialize an
 504 Array or a <tt>struct</tt>/<tt>union</tt>:
 505 </p>
 506 <ul>
 507
 508 <li>If the table index <tt>[0]</tt> is non-<tt>nil</tt>, then the
 509 table is assumed to be zero-based. Otherwise it's assumed to be
 510 one-based.</li>
 511
 512 <li>Array elements, starting at index zero, are initialized one-by-one
 513 with the consecutive table elements, starting at either index
 514 <tt>[0]</tt> or <tt>[1]</tt>. This process stops at the first
 515 <tt>nil</tt> table element.</li>
 516
 517 <li>If exactly one array element was initialized, it's repeated for
 518 all the remaining elements. Otherwise all remaining uninitialized
 519 elements are filled with zero bytes.</li>
 520
 521 <li>The above logic only applies to arrays with a known fixed size.
 522 A VLA is only initialized with the element(s) given in the table.
 523 Depending on the use case, you may need to explicitly add a
 524 <tt>NULL</tt> or <tt>0</tt> terminator to a VLA.</li>
 525
 526 <li>If the table has a non-empty hash part, a
 527 <tt>struct</tt>/<tt>union</tt> is initialized by looking up each field
 528 name (as a string key) in the table. Each non-<tt>nil</tt> value is
 529 used to initialize the corresponding field.</li>
 530
 531 <li>Otherwise a <tt>struct</tt>/<tt>union</tt> is initialized in the
 532 order of the declaration of its fields. Each field is initialized with
 533 the consecutive table elements, starting at either index <tt>[0]</tt>
 534 or <tt>[1]</tt>. This process stops at the first <tt>nil</tt> table
 535 element.</li>
 536
 537 <li>Uninitialized fields of a <tt>struct</tt> are filled with zero
 538 bytes, except for the trailing VLA of a VLS.</li>
 539
 540 <li>Initialization of a <tt>union</tt> stops after one field has been
 541 initialized. If no field has been initialized, the <tt>union</tt> is
 542 filled with zero bytes.</li>
 543
 544 <li>Elements or fields which are aggregates themselves are initialized
 545 with a <em>single</em> initializer, but this may be a nested table
 546 initializer (or a compatible aggregate).</li>
 547
 548 <li>Excess initializers for an array cause an error. Excess
 549 initializers for a <tt>struct</tt>/<tt>union</tt> are ignored.
 550 Unrelated table entries are ignored, too.</li>
 551
 552 </ul>
 553 <p>
 554 Example:
 555 </p>
 556 <pre class="code">
 557 local ffi = require("ffi")
 558
 559 ffi.cdef[[
 560 struct foo { int a, b; };
 561 union bar { int i; double d; };
 562 struct nested { int x; struct foo y; };
 563 ]]
 564
 565 ffi.new("int[3]", {})            --> 0, 0, 0
 566 ffi.new("int[3]", {1})           --> 1, 1, 1
 567 ffi.new("int[3]", {1,2})         --> 1, 2, 0
 568 ffi.new("int[3]", {1,2,3})       --> 1, 2, 3
 569 ffi.new("int[3]", {[0]=1})       --> 1, 1, 1
 570 ffi.new("int[3]", {[0]=1,2})     --> 1, 2, 0
 571 ffi.new("int[3]", {[0]=1,2,3})   --> 1, 2, 3
 572 ffi.new("int[3]", {[0]=1,2,3,4}) --> error: too many initializers
 573
 574 ffi.new("struct foo", {})            --> a = 0, b = 0
 575 ffi.new("struct foo", {1})           --> a = 1, b = 0
 576 ffi.new("struct foo", {1,2})         --> a = 1, b = 2
 577 ffi.new("struct foo", {[0]=1,2})     --> a = 1, b = 2
 578 ffi.new("struct foo", {b=2})         --> a = 0, b = 2
 579 ffi.new("struct foo", {a=1,b=2,c=3}) --> a = 1, b = 2  'c' is ignored
 580
 581 ffi.new("union bar", {})        --> i = 0, d = 0.0
 582 ffi.new("union bar", {1})       --> i = 1, d = ?
 583 ffi.new("union bar", {[0]=1,2}) --> i = 1, d = ?    '2' is ignored
 584 ffi.new("union bar", {d=2})     --> i = ?, d = 2.0
 585
 586 ffi.new("struct nested", {1,{2,3}})     --> x = 1, y.a = 2, y.b = 3
 587 ffi.new("struct nested", {x=1,y={2,3}}) --> x = 1, y.a = 2, y.b = 3
 588 </pre>
 589
 590 <h2 id="cdata_ops">Operations on cdata Objects</h2>
 591 <p>
 592 All of the standard Lua operators can be applied to cdata objects or a
 593 mix of a cdata object and another Lua object. The following list shows
 594 the pre-defined operations.
 595 </p>
 596 <p>
 597 Reference types are dereferenced <em>before</em> performing each of
 598 the operations below &mdash; the operation is applied to the
 599 C&nbsp;type pointed to by the reference.
 600 </p>
 601 <p>
 602 The pre-defined operations are always tried first before deferring to a
 603 metamethod or index table (if any) for the corresponding ctype (except
 604 for <tt>__new</tt>). An error is raised if the metamethod lookup or
 605 index table lookup fails.
 606 </p>
 607
 608 <h3 id="cdata_array">Indexing a cdata object</h3>
 609 <ul>
 610
 611 <li><b>Indexing a pointer/array</b>: a cdata pointer/array can be
 612 indexed by a cdata number or a Lua number. The element address is
 613 computed as the base address plus the number value multiplied by the
 614 element size in bytes. A read access loads the element value and
 615 <a href="#convert_tolua">converts it to a Lua object</a>. A write
 616 access <a href="#convert_fromlua">converts a Lua object to the element
 617 type</a> and stores the converted value to the element. An error is
 618 raised if the element size is undefined or a write access to a
 619 constant element is attempted.</li>
 620
 621 <li><b>Dereferencing a <tt>struct</tt>/<tt>union</tt> field</b>: a
 622 cdata <tt>struct</tt>/<tt>union</tt> or a pointer to a
 623 <tt>struct</tt>/<tt>union</tt> can be dereferenced by a string key,
 624 giving the field name. The field address is computed as the base
 625 address plus the relative offset of the field. A read access loads the
 626 field value and <a href="#convert_tolua">converts it to a Lua
 627 object</a>. A write access <a href="#convert_fromlua">converts a Lua
 628 object to the field type</a> and stores the converted value to the
 629 field. An error is raised if a write access to a constant
 630 <tt>struct</tt>/<tt>union</tt> or a constant field is attempted.
 631 Scoped enum constants or static constants are treated like a constant
 632 field.</li>
 633
 634 <li><b>Indexing a complex number</b>: a complex number can be indexed
 635 either by a cdata number or a Lua number with the values 0 or 1, or by
 636 the strings <tt>"re"</tt> or <tt>"im"</tt>. A read access loads the
 637 real part (<tt>[0]</tt>, <tt>.re</tt>) or the imaginary part
 638 (<tt>[1]</tt>, <tt>.im</tt>) part of a complex number and
 639 <a href="#convert_tolua">converts it to a Lua number</a>. The
 640 sub-parts of a complex number are immutable &mdash; assigning to an
 641 index of a complex number raises an error. Accessing out-of-bound
 642 indexes returns unspecified results, but is guaranteed not to trigger
 643 memory access violations.</li>
 644
 645 <li><b>Indexing a vector</b>: a vector is treated like an array for
 646 indexing purposes, except the vector elements are immutable &mdash;
 647 assigning to an index of a vector raises an error.</li>
 648
 649 </ul>
 650 <p>
 651 A ctype object can be indexed with a string key, too. The only
 652 pre-defined operation is reading scoped constants of
 653 <tt>struct</tt>/<tt>union</tt> types. All other accesses defer
 654 to the corresponding metamethods or index tables (if any).
 655 </p>
 656 <p>
 657 Note: since there's (deliberately) no address-of operator, a cdata
 658 object holding a value type is effectively immutable after
 659 initialization. The JIT compiler benefits from this fact when applying
 660 certain optimizations.
 661 </p>
 662 <p>
 663 As a consequence of this, the <em>elements</em> of complex numbers and
 664 vectors are immutable. But the elements of an aggregate holding these
 665 types <em>may</em> be modified of course. I.e. you cannot assign to
 666 <tt>foo.c.im</tt>, but you can assign a (newly created) complex number
 667 to <tt>foo.c</tt>.
 668 </p>
 669
 670 <h3 id="cdata_call">Calling a cdata object</h3>
 671 <ul>
 672
 673 <li><b>Constructor</b>: a ctype object can be called and used as a
 674 <a href="ext_ffi_api.html#ffi_new">constructor</a>. This is equivalent
 675 to <tt>ffi.new(ct, ...)</tt>, unless a <tt>__new</tt> metamethod is
 676 defined. The <tt>__new</tt> metamethod is called with the ctype object
 677 plus any other arguments passed to the contructor. Note that you have to
 678 use <tt>ffi.new</tt> inside of it, since calling <tt>ct(...)</tt> would
 679 cause infinite recursion.</li>
 680
 681 <li><b>C&nbsp;function call</b>: a cdata function or cdata function
 682 pointer can be called. The passed arguments are
 683 <a href="#convert_fromlua">converted to the C&nbsp;types</a> of the
 684 parameters given by the function declaration. Arguments passed to the
 685 variable argument part of vararg C&nbsp;function use
 686 <a href="#convert_vararg">special conversion rules</a>. This
 687 C&nbsp;function is called and the return value (if any) is
 688 <a href="#convert_tolua">converted to a Lua object</a>.<br>
 689 On Windows/x86 systems, <tt>__stdcall</tt> functions are automatically
 690 detected and a function declared as <tt>__cdecl</tt> (the default) is
 691 silently fixed up after the first call.</li>
 692
 693 </ul>
 694
 695 <h3 id="cdata_arith">Arithmetic on cdata objects</h3>
 696 <ul>
 697
 698 <li><b>Pointer arithmetic</b>: a cdata pointer/array and a cdata
 699 number or a Lua number can be added or subtracted. The number must be
 700 on the right hand side for a subtraction. The result is a pointer of
 701 the same type with an address plus or minus the number value
 702 multiplied by the element size in bytes. An error is raised if the
 703 element size is undefined.</li>
 704
 705 <li><b>Pointer difference</b>: two compatible cdata pointers/arrays
 706 can be subtracted. The result is the difference between their
 707 addresses, divided by the element size in bytes. An error is raised if
 708 the element size is undefined or zero.</li>
 709
 710 <li><b>64&nbsp;bit integer arithmetic</b>: the standard arithmetic
 711 operators (<tt>+&nbsp;-&nbsp;*&nbsp;/&nbsp;%&nbsp;^</tt> and unary
 712 minus) can be applied to two cdata numbers, or a cdata number and a
 713 Lua number. If one of them is an <tt>uint64_t</tt>, the other side is
 714 converted to an <tt>uint64_t</tt> and an unsigned arithmetic operation
 715 is performed. Otherwise both sides are converted to an
 716 <tt>int64_t</tt> and a signed arithmetic operation is performed. The
 717 result is a boxed 64&nbsp;bit cdata object.<br>
 718
 719 If one of the operands is an <tt>enum</tt> and the other operand is a
 720 string, the string is converted to the value of a matching <tt>enum</tt>
 721 constant before the above conversion.<br>
 722
 723 These rules ensure that 64&nbsp;bit integers are "sticky". Any
 724 expression involving at least one 64&nbsp;bit integer operand results
 725 in another one. The undefined cases for the division, modulo and power
 726 operators return <tt>2LL&nbsp;^&nbsp;63</tt> or
 727 <tt>2ULL&nbsp;^&nbsp;63</tt>.<br>
 728
 729 You'll have to explicitly convert a 64&nbsp;bit integer to a Lua
 730 number (e.g. for regular floating-point calculations) with
 731 <tt>tonumber()</tt>. But note this may incur a precision loss.</li>
 732
 733 </ul>
 734
 735 <h3 id="cdata_comp">Comparisons of cdata objects</h3>
 736 <ul>
 737
 738 <li><b>Pointer comparison</b>: two compatible cdata pointers/arrays
 739 can be compared. The result is the same as an unsigned comparison of
 740 their addresses. <tt>nil</tt> is treated like a <tt>NULL</tt> pointer,
 741 which is compatible with any other pointer type.</li>
 742
 743 <li><b>64&nbsp;bit integer comparison</b>: two cdata numbers, or a
 744 cdata number and a Lua number can be compared with each other. If one
 745 of them is an <tt>uint64_t</tt>, the other side is converted to an
 746 <tt>uint64_t</tt> and an unsigned comparison is performed. Otherwise
 747 both sides are converted to an <tt>int64_t</tt> and a signed
 748 comparison is performed.<br>
 749
 750 If one of the operands is an <tt>enum</tt> and the other operand is a
 751 string, the string is converted to the value of a matching <tt>enum</tt>
 752 constant before the above conversion.<br>
 753
 754 <li><b>Comparisons for equality/inequality</b> never raise an error.
 755 Even incompatible pointers can be compared for equality by address. Any
 756 other incompatible comparison (also with non-cdata objects) treats the
 757 two sides as unequal.</li>
 758
 759 </ul>
 760
 761 <h3 id="cdata_key">cdata objects as table keys</h3>
 762 <p>
 763 Lua tables may be indexed by cdata objects, but this doesn't provide
 764 any useful semantics &mdash; <b>cdata objects are unsuitable as table
 765 keys!</b>
 766 </p>
 767 <p>
 768 A cdata object is treated like any other garbage-collected object and
 769 is hashed and compared by its address for table indexing. Since
 770 there's no interning for cdata value types, the same value may be
 771 boxed in different cdata objects with different addresses. Thus
 772 <tt>t[1LL+1LL]</tt> and <tt>t[2LL]</tt> usually <b>do not</b> point to
 773 the same hash slot and they certainly <b>do not</b> point to the same
 774 hash slot as <tt>t[2]</tt>.
 775 </p>
 776 <p>
 777 It would seriously drive up implementation complexity and slow down
 778 the common case, if one were to add extra handling for by-value
 779 hashing and comparisons to Lua tables. Given the ubiquity of their use
 780 inside the VM, this is not acceptable.
 781 </p>
 782 <p>
 783 There are three viable alternatives, if you really need to use cdata
 784 objects as keys:
 785 </p>
 786 <ul>
 787
 788 <li>If you can get by with the precision of Lua numbers
 789 (52&nbsp;bits), then use <tt>tonumber()</tt> on a cdata number or
 790 combine multiple fields of a cdata aggregate to a Lua number. Then use
 791 the resulting Lua number as a key when indexing tables.<br>
 792 One obvious benefit: <tt>t[tonumber(2LL)]</tt> <b>does</b> point to
 793 the same slot as <tt>t[2]</tt>.</li>
 794
 795 <li>Otherwise use either <tt>tostring()</tt> on 64&nbsp;bit integers
 796 or complex numbers or combine multiple fields of a cdata aggregate to
 797 a Lua string (e.g. with
 798 <a href="ext_ffi_api.html#ffi_string"><tt>ffi.string()</tt></a>). Then
 799 use the resulting Lua string as a key when indexing tables.</li>
 800
 801 <li>Create your own specialized hash table implementation using the
 802 C&nbsp;types provided by the FFI library, just like you would in
 803 C&nbsp;code. Ultimately this may give much better performance than the
 804 other alternatives or what a generic by-value hash table could
 805 possibly provide.</li>
 806
 807 </ul>
 808
 809 <h2 id="gc">Garbage Collection of cdata Objects</h2>
 810 <p>
 811 All explicitly (<tt>ffi.new()</tt>, <tt>ffi.cast()</tt> etc.) or
 812 implicitly (accessors) created cdata objects are garbage collected.
 813 You need to ensure to retain valid references to cdata objects
 814 somewhere on a Lua stack, an upvalue or in a Lua table while they are
 815 still in use. Once the last reference to a cdata object is gone, the
 816 garbage collector will automatically free the memory used by it (at
 817 the end of the next GC cycle).
 818 </p>
 819 <p>
 820 Please note that pointers themselves are cdata objects, however they
 821 are <b>not</b> followed by the garbage collector. So e.g. if you
 822 assign a cdata array to a pointer, you must keep the cdata object
 823 holding the array alive as long as the pointer is still in use:
 824 </p>
 825 <pre class="code">
 826 ffi.cdef[[
 827 typedef struct { int *a; } foo_t;
 828 ]]
 829
 830 local s = ffi.new("foo_t", ffi.new("int[10]")) -- <span style="color:#c00000;">WRONG!</span>
 831
 832 local a = ffi.new("int[10]") -- <span style="color:#00a000;">OK</span>
 833 local s = ffi.new("foo_t", a)
 834 -- Now do something with 's', but keep 'a' alive until you're done.
 835 </pre>
 836 <p>
 837 Similar rules apply for Lua strings which are implicitly converted to
 838 <tt>"const&nbsp;char&nbsp;*"</tt>: the string object itself must be
 839 referenced somewhere or it'll be garbage collected eventually. The
 840 pointer will then point to stale data, which may have already been
 841 overwritten. Note that <em>string literals</em> are automatically kept
 842 alive as long as the function containing it (actually its prototype)
 843 is not garbage collected.
 844 </p>
 845 <p>
 846 Objects which are passed as an argument to an external C&nbsp;function
 847 are kept alive until the call returns. So it's generally safe to
 848 create temporary cdata objects in argument lists. This is a common
 849 idiom for <a href="#convert_vararg">passing specific C&nbsp;types to
 850 vararg functions</a>.
 851 </p>
 852 <p>
 853 Memory areas returned by C functions (e.g. from <tt>malloc()</tt>)
 854 must be manually managed, of course (or use
 855 <a href="ext_ffi_api.html#ffi_gc"><tt>ffi.gc()</tt></a>). Pointers to
 856 cdata objects are indistinguishable from pointers returned by C
 857 functions (which is one of the reasons why the GC cannot follow them).
 858 </p>
 859
 860 <h2 id="callback">Callbacks</h2>
 861 <p>
 862 The LuaJIT FFI automatically generates special callback functions
 863 whenever a Lua function is converted to a C&nbsp;function pointer. This
 864 associates the generated callback function pointer with the C&nbsp;type
 865 of the function pointer and the Lua function object (closure).
 866 </p>
 867 <p>
 868 This can happen implicitly due to the usual conversions, e.g. when
 869 passing a Lua function to a function pointer argument. Or you can use
 870 <tt>ffi.cast()</tt> to explicitly cast a Lua function to a
 871 C&nbsp;function pointer.
 872 </p>
 873 <p>
 874 Currently only certain C&nbsp;function types can be used as callback
 875 functions. Neither C&nbsp;vararg functions nor functions with
 876 pass-by-value aggregate argument or result types are supported. There
 877 are no restrictions for the kind of Lua functions that can be called
 878 from the callback &mdash; no checks for the proper number of arguments
 879 are made. The return value of the Lua function will be converted to the
 880 result type and an error will be thrown for invalid conversions.
 881 </p>
 882 <p>
 883 It's allowed to throw errors across a callback invocation, but it's not
 884 advisable in general. Do this only if you know the C&nbsp;function, that
 885 called the callback, copes with the forced stack unwinding and doesn't
 886 leak resources.
 887 </p>
 888
 889 <h3 id="callback_resources">Callback resource handling</h3>
 890 <p>
 891 Callbacks take up resources &mdash; you can only have a limited number
 892 of them at the same time (500&nbsp;-&nbsp;1000, depending on the
 893 architecture). The associated Lua functions are anchored to prevent
 894 garbage collection, too.
 895 </p>
 896 <p>
 897 <b>Callbacks due to implicit conversions are permanent!</b> There is no
 898 way to guess their lifetime, since the C&nbsp;side might store the
 899 function pointer for later use (typical for GUI toolkits). The associated
 900 resources cannot be reclaimed until termination:
 901 </p>
 902 <pre class="code">
 903 ffi.cdef[[
 904 typedef int (__stdcall *WNDENUMPROC)(void *hwnd, intptr_t l);
 905 int EnumWindows(WNDENUMPROC func, intptr_t l);
 906 ]]
 907
 908 -- Implicit conversion to a callback via function pointer argument.
 909 local count = 0
 910 ffi.C.EnumWindows(function(hwnd, l)
 911   count = count + 1
 912   return true
 913 end, 0)
 914 -- The callback is permanent and its resources cannot be reclaimed!
 915 -- Ok, so this may not be a problem, if you do this only once.
 916 </pre>
 917 <p>
 918 Note: this example shows that you <em>must</em> properly declare
 919 <tt>__stdcall</tt> callbacks on Windows/x86 systems. The calling
 920 convention cannot be automatically detected, unlike for
 921 <tt>__stdcall</tt> calls <em>to</em> Windows functions.
 922 </p>
 923 <p>
 924 For some use cases it's necessary to free up the resources or to
 925 dynamically redirect callbacks. Use an explicit cast to a
 926 C&nbsp;function pointer and keep the resulting cdata object. Then use
 927 the <a href="ext_ffi_api.html#callback_free"><tt>cb:free()</tt></a>
 928 or <a href="ext_ffi_api.html#callback_set"><tt>cb:set()</tt></a> methods
 929 on the cdata object:
 930 </p>
 931 <pre class="code">
 932 -- Explicitly convert to a callback via cast.
 933 local count = 0
 934 local cb = ffi.cast("WNDENUMPROC", function(hwnd, l)
 935   count = count + 1
 936   return true
 937 end)
 938
 939 -- Pass it to a C function.
 940 ffi.C.EnumWindows(cb, 0)
 941 -- EnumWindows doesn't need the callback after it returns, so free it.
 942
 943 cb:free()
 944 -- The callback function pointer is no longer valid and its resources
 945 -- will be reclaimed. The created Lua closure will be garbage collected.
 946 </pre>
 947
 948 <h3 id="callback_performance">Callback performance</h3>
 949 <p>
 950 <b>Callbacks are slow!</b> First, the C&nbsp;to Lua transition itself
 951 has an unavoidable cost, similar to a <tt>lua_call()</tt> or
 952 <tt>lua_pcall()</tt>. Argument and result marshalling add to that cost.
 953 And finally, neither the C&nbsp;compiler nor LuaJIT can inline or
 954 optimize across the language barrier and hoist repeated computations out
 955 of a callback function.
 956 </p>
 957 <p>
 958 Do not use callbacks for performance-sensitive work: e.g. consider a
 959 numerical integration routine which takes a user-defined function to
 960 integrate over. It's a bad idea to call a user-defined Lua function from
 961 C&nbsp;code millions of times. The callback overhead will be absolutely
 962 detrimental for performance.
 963 </p>
 964 <p>
 965 It's considerably faster to write the numerical integration routine
 966 itself in Lua &mdash; the JIT compiler will be able to inline the
 967 user-defined function and optimize it together with its calling context,
 968 with very competitive performance.
 969 </p>
 970 <p>
 971 As a general guideline: <b>use callbacks only when you must</b>, because
 972 of existing C&nbsp;APIs. E.g. callback performance is irrelevant for a
 973 GUI application, which waits for user input most of the time, anyway.
 974 </p>
 975 <p>
 976 For new designs <b>avoid push-style APIs</b> (C&nbsp;function repeatedly
 977 calling a callback for each result). Instead <b>use pull-style APIs</b>
 978 (call a C&nbsp;function repeatedly to get a new result). Calls from Lua
 979 to C via the FFI are much faster than the other way round. Most well-designed
 980 libraries already use pull-style APIs (read/write, get/put).
 981 </p>
 982
 983 <h2 id="clib">C Library Namespaces</h2>
 984 <p>
 985 A C&nbsp;library namespace is a special kind of object which allows
 986 access to the symbols contained in shared libraries or the default
 987 symbol namespace. The default
 988 <a href="ext_ffi_api.html#ffi_C"><tt>ffi.C</tt></a> namespace is
 989 automatically created when the FFI library is loaded. C&nbsp;library
 990 namespaces for specific shared libraries may be created with the
 991 <a href="ext_ffi_api.html#ffi_load"><tt>ffi.load()</tt></a> API
 992 function.
 993 </p>
 994 <p>
 995 Indexing a C&nbsp;library namespace object with a symbol name (a Lua
 996 string) automatically binds it to the library. First the symbol type
 997 is resolved &mdash; it must have been declared with
 998 <a href="ext_ffi_api.html#ffi_cdef"><tt>ffi.cdef</tt></a>. Then the
 999 symbol address is resolved by searching for the symbol name in the
1000 associated shared libraries or the default symbol namespace. Finally,
1001 the resulting binding between the symbol name, the symbol type and its
1002 address is cached. Missing symbol declarations or nonexistent symbol
1003 names cause an error.
1004 </p>
1005 <p>
1006 This is what happens on a <b>read access</b> for the different kinds of
1007 symbols:
1008 </p>
1009 <ul>
1010
1011 <li>External functions: a cdata object with the type of the function
1012 and its address is returned.</li>
1013
1014 <li>External variables: the symbol address is dereferenced and the
1015 loaded value is <a href="#convert_tolua">converted to a Lua object</a>
1016 and returned.</li>
1017
1018 <li>Constant values (<tt>static&nbsp;const</tt> or <tt>enum</tt>
1019 constants): the constant is <a href="#convert_tolua">converted to a
1020 Lua object</a> and returned.</li>
1021
1022 </ul>
1023 <p>
1024 This is what happens on a <b>write access</b>:
1025 </p>
1026 <ul>
1027
1028 <li>External variables: the value to be written is
1029 <a href="#convert_fromlua">converted to the C&nbsp;type</a> of the
1030 variable and then stored at the symbol address.</li>
1031
1032 <li>Writing to constant variables or to any other symbol type causes
1033 an error, like any other attempted write to a constant location.</li>
1034
1035 </ul>
1036 <p>
1037 C&nbsp;library namespaces themselves are garbage collected objects. If
1038 the last reference to the namespace object is gone, the garbage
1039 collector will eventually release the shared library reference and
1040 remove all memory associated with the namespace. Since this may
1041 trigger the removal of the shared library from the memory of the
1042 running process, it's generally <em>not safe</em> to use function
1043 cdata objects obtained from a library if the namespace object may be
1044 unreferenced.
1045 </p>
1046 <p>
1047 Performance notice: the JIT compiler specializes to the identity of
1048 namespace objects and to the strings used to index it. This
1049 effectively turns function cdata objects into constants. It's not
1050 useful and actually counter-productive to explicitly cache these
1051 function objects, e.g. <tt>local strlen = ffi.C.strlen</tt>. OTOH it
1052 <em>is</em> useful to cache the namespace itself, e.g. <tt>local C =
1053 ffi.C</tt>.
1054 </p>
1055
1056 <h2 id="policy">No Hand-holding!</h2>
1057 <p>
1058 The FFI library has been designed as <b>a low-level library</b>. The
1059 goal is to interface with C&nbsp;code and C&nbsp;data types with a
1060 minimum of overhead. This means <b>you can do anything you can do
1061 from&nbsp;C</b>: access all memory, overwrite anything in memory, call
1062 machine code at any memory address and so on.
1063 </p>
1064 <p>
1065 The FFI library provides <b>no memory safety</b>, unlike regular Lua
1066 code. It will happily allow you to dereference a <tt>NULL</tt>
1067 pointer, to access arrays out of bounds or to misdeclare
1068 C&nbsp;functions. If you make a mistake, your application might crash,
1069 just like equivalent C&nbsp;code would.
1070 </p>
1071 <p>
1072 This behavior is inevitable, since the goal is to provide full
1073 interoperability with C&nbsp;code. Adding extra safety measures, like
1074 bounds checks, would be futile. There's no way to detect
1075 misdeclarations of C&nbsp;functions, since shared libraries only
1076 provide symbol names, but no type information. Likewise there's no way
1077 to infer the valid range of indexes for a returned pointer.
1078 </p>
1079 <p>
1080 Again: the FFI library is a low-level library. This implies it needs
1081 to be used with care, but it's flexibility and performance often
1082 outweigh this concern. If you're a C or C++ developer, it'll be easy
1083 to apply your existing knowledge. OTOH writing code for the FFI
1084 library is not for the faint of heart and probably shouldn't be the
1085 first exercise for someone with little experience in Lua, C or C++.
1086 </p>
1087 <p>
1088 As a corollary of the above, the FFI library is <b>not safe for use by
1089 untrusted Lua code</b>. If you're sandboxing untrusted Lua code, you
1090 definitely don't want to give this code access to the FFI library or
1091 to <em>any</em> cdata object (except 64&nbsp;bit integers or complex
1092 numbers). Any properly engineered Lua sandbox needs to provide safety
1093 wrappers for many of the standard Lua library functions &mdash;
1094 similar wrappers need to be written for high-level operations on FFI
1095 data types, too.
1096 </p>
1097
1098 <h2 id="status">Current Status</h2>
1099 <p>
1100 The initial release of the FFI library has some limitations and is
1101 missing some features. Most of these will be fixed in future releases.
1102 </p>
1103 <p>
1104 <a href="#clang">C language support</a> is
1105 currently incomplete:
1106 </p>
1107 <ul>
1108 <li>C&nbsp;declarations are not passed through a C&nbsp;pre-processor,
1109 yet.</li>
1110 <li>The C&nbsp;parser is able to evaluate most constant expressions
1111 commonly found in C&nbsp;header files. However it doesn't handle the
1112 full range of C&nbsp;expression semantics and may fail for some
1113 obscure constructs.</li>
1114 <li><tt>static const</tt> declarations only work for integer types
1115 up to 32&nbsp;bits. Neither declaring string constants nor
1116 floating-point constants is supported.</li>
1117 <li>Packed <tt>struct</tt> bitfields that cross container boundaries
1118 are not implemented.</li>
1119 <li>Native vector types may be defined with the GCC <tt>mode</tt> or
1120 <tt>vector_size</tt> attribute. But no operations other than loading,
1121 storing and initializing them are supported, yet.</li>
1122 <li>The <tt>volatile</tt> type qualifier is currently ignored by
1123 compiled code.</li>
1124 <li><a href="ext_ffi_api.html#ffi_cdef"><tt>ffi.cdef</tt></a> silently
1125 ignores all re-declarations.</li>
1126 </ul>
1127 <p>
1128 The JIT compiler already handles a large subset of all FFI operations.
1129 It automatically falls back to the interpreter for unimplemented
1130 operations (you can check for this with the
1131 <a href="running.html#opt_j"><tt>-jv</tt></a> command line option).
1132 The following operations are currently not compiled and may exhibit
1133 suboptimal performance, especially when used in inner loops:
1134 </p>
1135 <ul>
1136 <li>Array/<tt>struct</tt> copies and bulk initializations.</li>
1137 <li>Bitfield accesses and initializations.</li>
1138 <li>Vector operations.</li>
1139 <li>Table initializers.</li>
1140 <li>Initialization of nested <tt>struct</tt>/<tt>union</tt> types.</li>
1141 <li>Allocations of variable-length arrays or structs.</li>
1142 <li>Allocations of C&nbsp;types with a size &gt; 128&nbsp;bytes or an
1143 alignment &gt; 8&nbsp;bytes.</li>
1144 <li>Conversions from lightuserdata to <tt>void&nbsp;*</tt>.</li>
1145 <li>Pointer differences for element sizes that are not a power of
1146 two.</li>
1147 <li>Calls to C&nbsp;functions with aggregates passed or returned by
1148 value.</li>
1149 <li>Calls to ctype metamethods which are not plain functions.</li>
1150 <li>ctype <tt>__newindex</tt> tables and non-string lookups in ctype
1151 <tt>__index</tt> tables.</li>
1152 <li><tt>tostring()</tt> for cdata types.</li>
1153 <li>Calls to the following <a href="ext_ffi_api.html">ffi.* API</a>
1154 functions: <tt>cdef</tt>, <tt>load</tt>, <tt>typeof</tt>,
1155 <tt>metatype</tt>, <tt>gc</tt>, <tt>sizeof</tt>, <tt>alignof</tt>,
1156 <tt>offsetof</tt>.</li>
1157 </ul>
1158 <p>
1159 Other missing features:
1160 </p>
1161 <ul>
1162 <li>Bit operations for 64&nbsp;bit types.</li>
1163 <li>Arithmetic for <tt>complex</tt> numbers.</li>
1164 <li>Passing structs by value to vararg C&nbsp;functions.</li>
1165 <li><a href="extensions.html#exceptions">C++ exception interoperability</a>
1166 does not extend to C&nbsp;functions called via the FFI, if the call is
1167 compiled.</li>
1168 </ul>
1169 <br class="flush">
1170 </div>
1171 <div id="foot">
1172 <hr class="hide">
1173 Copyright &copy; 2005-2012 Mike Pall
1174 <span class="noprint">
1175 &middot;
1176 <a href="contact.html">Contact</a>
1177 </span>
1178 </div>
1179 </body>
1180 </html>