FFI: Fix unfused loads of float fields.
[luajit-2.0.git] / doc / running.html
blob109df5dfd85671142be9c21adffe5ac1af960748
1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
2 <html>
3 <head>
4 <title>Running LuaJIT</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) 2005-2011, 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.opt {
13 line-height: 1.2;
15 tr.opthead td {
16 font-weight: bold;
18 td.flag_name {
19 width: 4em;
21 td.flag_level {
22 width: 2em;
23 text-align: center;
25 td.param_name {
26 width: 6em;
28 td.param_default {
29 width: 4em;
30 text-align: right;
32 </style>
33 </head>
34 <body>
35 <div id="site">
36 <a href="http://luajit.org"><span>Lua<span id="logo">JIT</span></span></a>
37 </div>
38 <div id="head">
39 <h1>Running LuaJIT</h1>
40 </div>
41 <div id="nav">
42 <ul><li>
43 <a href="luajit.html">LuaJIT</a>
44 <ul><li>
45 <a href="install.html">Installation</a>
46 </li><li>
47 <a class="current" href="running.html">Running</a>
48 </li></ul>
49 </li><li>
50 <a href="extensions.html">Extensions</a>
51 <ul><li>
52 <a href="ext_ffi.html">FFI Library</a>
53 <ul><li>
54 <a href="ext_ffi_tutorial.html">FFI Tutorial</a>
55 </li><li>
56 <a href="ext_ffi_api.html">ffi.* API</a>
57 </li><li>
58 <a href="ext_ffi_semantics.html">FFI Semantics</a>
59 </li></ul>
60 </li><li>
61 <a href="ext_jit.html">jit.* Library</a>
62 </li><li>
63 <a href="ext_c_api.html">Lua/C API</a>
64 </li></ul>
65 </li><li>
66 <a href="status.html">Status</a>
67 <ul><li>
68 <a href="changes.html">Changes</a>
69 </li></ul>
70 </li><li>
71 <a href="faq.html">FAQ</a>
72 </li><li>
73 <a href="http://luajit.org/performance.html">Performance <span class="ext">&raquo;</span></a>
74 </li><li>
75 <a href="http://luajit.org/download.html">Download <span class="ext">&raquo;</span></a>
76 </li></ul>
77 </div>
78 <div id="main">
79 <p>
80 LuaJIT has only a single stand-alone executable, called <tt>luajit</tt> on
81 POSIX systems or <tt>luajit.exe</tt> on Windows. It can be used to run simple
82 Lua statements or whole Lua applications from the command line. It has an
83 interactive mode, too.
84 </p>
85 <p class="indent" style="color: #c00000;">
86 Note: the beta test releases only install under the versioned name on
87 POSIX systems (to avoid overwriting a previous version). You either need
88 to type <tt>luajit-2.0.0-beta8</tt> to start it or create a symlink
89 with a command like this:
90 </p>
91 <pre class="code" style="color: #c00000;">
92 sudo ln -sf luajit-2.0.0-beta8&nbsp;/usr/local/bin/luajit
93 </pre>
94 <p>
95 Unlike previous versions <b>optimization is turned on by default</b> in
96 LuaJIT 2.0!<br>It's no longer necessary to use <tt>luajit&nbsp;-O</tt>.
97 </p>
99 <h2 id="options">Command Line Options</h2>
101 The <tt>luajit</tt> stand-alone executable is just a slightly modified
102 version of the regular <tt>lua</tt> stand-alone executable.
103 It supports the same basic options, too. <tt>luajit&nbsp;-h</tt>
104 prints a short list of the available options. Please have a look at the
105 <a href="http://www.lua.org/manual/5.1/manual.html#6"><span class="ext">&raquo;</span>&nbsp;Lua manual</a>
106 for details.
107 </p>
109 LuaJIT has some additional options:
110 </p>
112 <h3 id="opt_b"><tt>-b[options] input output</tt></h3>
114 This option saves or lists bytecode. The following additional options
115 are accepted:
116 </p>
117 <ul>
118 <li><tt>-l</tt> &mdash; Only list bytecode.</li>
119 <li><tt>-s</tt> &mdash; Strip debug info (this is the default).</li>
120 <li><tt>-g</tt> &mdash; Keep debug info.</li>
121 <li><tt>-n name</tt> &mdash; Set module name (default: auto-detect from input name)</li>
122 <li><tt>-t type</tt> &mdash; Set output file type (default: auto-detect from output name).</li>
123 <li><tt>-a arch</tt> &mdash; Override architecture for object files (default: native).</li>
124 <li><tt>-o os</tt> &mdash; Override OS for object files (default: native).</li>
125 <li><tt>-e chunk</tt> &mdash; Use chunk string as input.</li>
126 <li><tt>-</tt> (a single minus sign) &mdash; Use stdin as input and/or stdout as output.</li>
127 </ul>
129 The output file type is auto-detected from the extension of the output
130 file name:
131 </p>
132 <ul>
133 <li><tt>c</tt> &mdash; C source file, exported bytecode data.</li>
134 <li><tt>h</tt> &mdash; C header file, static bytecode data.</li>
135 <li><tt>obj</tt> or <tt>o</tt> &mdash; Object file, exported bytecode data
136 (OS- and architecture-specific).</li>
137 <li><tt>raw</tt> or any other extension &mdash; Raw bytecode file (portable).
138 </ul>
140 Notes:
141 </p>
142 <ul>
143 <li>See also <a href="extensions.html#string_dump">string.dump()</a>
144 for information on bytecode portability and compatibility.</li>
145 <li>A file in raw bytecode format is auto-detected and can be loaded like
146 any Lua source file. E.g. directly from the command line or with
147 <tt>loadfile()</tt>, <tt>dofile()</tt> etc.</li>
148 <li>To statically embed the bytecode of a module in your application,
149 generate an object file and just link it with your application.</li>
150 <li>On most ELF-based systems (e.g. Linux) you need to explicitly export the
151 global symbols when linking your application, e.g. with: <tt>-Wl,-E</tt></li>
152 <li><tt>require()</tt> tries to load embedded bytecode data from exported
153 symbols (in <tt>*.exe</tt> or <tt>lua51.dll</tt> on Windows) and from
154 shared libraries in <tt>package.cpath</tt>.</li>
155 </ul>
157 Typical usage examples:
158 </p>
159 <pre class="code">
160 luajit -b test.lua test.out # Save bytecode to test.out
161 luajit -bg test.lua test.out # Keep debug info
162 luajit -be "print('hello world')" test.out # Save cmdline script
164 luajit -bl test.lua # List to stdout
165 luajit -bl test.lua test.txt # List to test.txt
166 luajit -ble "print('hello world')" # List cmdline script
168 luajit -b test.lua test.obj # Generate object file
169 # Link test.obj with your application and load it with require("test")
170 </pre>
172 <h3 id="opt_j"><tt>-j cmd[=arg[,arg...]]</tt></h3>
174 This option performs a LuaJIT control command or activates one of the
175 loadable extension modules. The command is first looked up in the
176 <tt>jit.*</tt> library. If no matching function is found, a module
177 named <tt>jit.&lt;cmd&gt;</tt> is loaded and the <tt>start()</tt>
178 function of the module is called with the specified arguments (if
179 any). The space between <tt>-j</tt> and <tt>cmd</tt> is optional.
180 </p>
182 Here are the available LuaJIT control commands:
183 </p>
184 <ul>
185 <li id="j_on"><tt>-jon</tt> &mdash; Turns the JIT compiler on (default).</li>
186 <li id="j_off"><tt>-joff</tt> &mdash; Turns the JIT compiler off (only use the interpreter).</li>
187 <li id="j_flush"><tt>-jflush</tt> &mdash; Flushes the whole cache of compiled code.</li>
188 <li id="j_v"><tt>-jv</tt> &mdash; Shows verbose information about the progress of the JIT compiler.</li>
189 <li id="j_dump"><tt>-jdump</tt> &mdash; Dumps the code and structures used in various compiler stages.</li>
190 </ul>
192 The <tt>-jv</tt> and <tt>-jdump</tt> commands are extension modules
193 written in Lua. They are mainly used for debugging the JIT compiler
194 itself. For a description of their options and output format, please
195 read the comment block at the start of their source.
196 They can be found in the <tt>lib</tt> directory of the source
197 distribution or installed under the <tt>jit</tt> directory. By default
198 this is <tt>/usr/local/share/luajit-2.0.0-beta8/jit</tt> on POSIX
199 systems.
200 </p>
202 <h3 id="opt_O"><tt>-O[level]</tt><br>
203 <tt>-O[+]flag</tt>&nbsp;&nbsp;&nbsp;<tt>-O-flag</tt><br>
204 <tt>-Oparam=value</tt></h3>
206 This options allows fine-tuned control of the optimizations used by
207 the JIT compiler. This is mainly intended for debugging LuaJIT itself.
208 Please note that the JIT compiler is extremely fast (we are talking
209 about the microsecond to millisecond range). Disabling optimizations
210 doesn't have any visible impact on its overhead, but usually generates
211 code that runs slower.
212 </p>
214 The first form sets an optimization level &mdash; this enables a
215 specific mix of optimization flags. <tt>-O0</tt> turns off all
216 optimizations and higher numbers enable more optimizations. Omitting
217 the level (i.e. just <tt>-O</tt>) sets the default optimization level,
218 which is <tt>-O3</tt> in the current version.
219 </p>
221 The second form adds or removes individual optimization flags.
222 The third form sets a parameter for the VM or the JIT compiler
223 to a specific value.
224 </p>
226 You can either use this option multiple times (like <tt>-Ocse
227 -O-dce -Ohotloop=10</tt>) or separate several settings with a comma
228 (like <tt>-O+cse,-dce,hotloop=10</tt>). The settings are applied from
229 left to right and later settings override earlier ones. You can freely
230 mix the three forms, but note that setting an optimization level
231 overrides all earlier flags.
232 </p>
234 Here are the available flags and at what optimization levels they
235 are enabled:
236 </p>
237 <table class="opt">
238 <tr class="opthead">
239 <td class="flag_name">Flag</td>
240 <td class="flag_level">-O1</td>
241 <td class="flag_level">-O2</td>
242 <td class="flag_level">-O3</td>
243 <td class="flag_desc">&nbsp;</td>
244 </tr>
245 <tr class="odd separate">
246 <td class="flag_name">fold</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_desc">Constant Folding, Simplifications and Reassociation</td></tr>
247 <tr class="even">
248 <td class="flag_name">cse</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_desc">Common-Subexpression Elimination</td></tr>
249 <tr class="odd">
250 <td class="flag_name">dce</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_desc">Dead-Code Elimination</td></tr>
251 <tr class="even">
252 <td class="flag_name">narrow</td><td class="flag_level">&nbsp;</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_desc">Narrowing of numbers to integers</td></tr>
253 <tr class="odd">
254 <td class="flag_name">loop</td><td class="flag_level">&nbsp;</td><td class="flag_level">&bull;</td><td class="flag_level">&bull;</td><td class="flag_desc">Loop Optimizations (code hoisting)</td></tr>
255 <tr class="even">
256 <td class="flag_name">fwd</td><td class="flag_level">&nbsp;</td><td class="flag_level">&nbsp;</td><td class="flag_level">&bull;</td><td class="flag_desc">Load Forwarding (L2L) and Store Forwarding (S2L)</td></tr>
257 <tr class="odd">
258 <td class="flag_name">dse</td><td class="flag_level">&nbsp;</td><td class="flag_level">&nbsp;</td><td class="flag_level">&bull;</td><td class="flag_desc">Dead-Store Elimination</td></tr>
259 <tr class="even">
260 <td class="flag_name">abc</td><td class="flag_level">&nbsp;</td><td class="flag_level">&nbsp;</td><td class="flag_level">&bull;</td><td class="flag_desc">Array Bounds Check Elimination</td></tr>
261 <tr class="odd">
262 <td class="flag_name">fuse</td><td class="flag_level">&nbsp;</td><td class="flag_level">&nbsp;</td><td class="flag_level">&bull;</td><td class="flag_desc">Fusion of operands into instructions</td></tr>
263 </table>
265 Here are the parameters and their default settings:
266 </p>
267 <table class="opt">
268 <tr class="opthead">
269 <td class="param_name">Parameter</td>
270 <td class="param_default">Default</td>
271 <td class="param_desc">&nbsp;</td>
272 </tr>
273 <tr class="odd separate">
274 <td class="param_name">maxtrace</td><td class="param_default">1000</td><td class="param_desc">Max. number of traces in the cache</td></tr>
275 <tr class="even">
276 <td class="param_name">maxrecord</td><td class="param_default">4000</td><td class="param_desc">Max. number of recorded IR instructions</td></tr>
277 <tr class="odd">
278 <td class="param_name">maxirconst</td><td class="param_default">500</td><td class="param_desc">Max. number of IR constants of a trace</td></tr>
279 <tr class="even">
280 <td class="param_name">maxside</td><td class="param_default">100</td><td class="param_desc">Max. number of side traces of a root trace</td></tr>
281 <tr class="odd">
282 <td class="param_name">maxsnap</td><td class="param_default">500</td><td class="param_desc">Max. number of snapshots for a trace</td></tr>
283 <tr class="even separate">
284 <td class="param_name">hotloop</td><td class="param_default">56</td><td class="param_desc">Number of iterations to detect a hot loop or hot call</td></tr>
285 <tr class="odd">
286 <td class="param_name">hotexit</td><td class="param_default">10</td><td class="param_desc">Number of taken exits to start a side trace</td></tr>
287 <tr class="even">
288 <td class="param_name">tryside</td><td class="param_default">4</td><td class="param_desc">Number of attempts to compile a side trace</td></tr>
289 <tr class="odd separate">
290 <td class="param_name">instunroll</td><td class="param_default">4</td><td class="param_desc">Max. unroll factor for instable loops</td></tr>
291 <tr class="even">
292 <td class="param_name">loopunroll</td><td class="param_default">15</td><td class="param_desc">Max. unroll factor for loop ops in side traces</td></tr>
293 <tr class="odd">
294 <td class="param_name">callunroll</td><td class="param_default">3</td><td class="param_desc">Max. unroll factor for pseudo-recursive calls</td></tr>
295 <tr class="even">
296 <td class="param_name">recunroll</td><td class="param_default">2</td><td class="param_desc">Min. unroll factor for true recursion</td></tr>
297 <tr class="odd separate">
298 <td class="param_name">sizemcode</td><td class="param_default">32</td><td class="param_desc">Size of each machine code area in KBytes (Windows: 64K)</td></tr>
299 <tr class="even">
300 <td class="param_name">maxmcode</td><td class="param_default">512</td><td class="param_desc">Max. total size of all machine code areas in KBytes</td></tr>
301 </table>
302 <br class="flush">
303 </div>
304 <div id="foot">
305 <hr class="hide">
306 Copyright &copy; 2005-2011 Mike Pall
307 <span class="noprint">
308 &middot;
309 <a href="contact.html">Contact</a>
310 </span>
311 </div>
312 </body>
313 </html>