doc/changes.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>LuaJIT Change History</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 div.major { max-width: 600px; padding: 1em; margin: 1em 0 1em 0; }
  13 </style>
  14 </head>
  15 <body>
  16 <div id="site">
  17 <a href="http://luajit.org"><span>Lua<span id="logo">JIT</span></span></a>
  18 </div>
  19 <div id="head">
  20 <h1>LuaJIT Change History</h1>
  21 </div>
  22 <div id="nav">
  23 <ul><li>
  24 <a href="luajit.html">LuaJIT</a>
  25 <ul><li>
  26 <a href="install.html">Installation</a>
  27 </li><li>
  28 <a href="running.html">Running</a>
  29 </li></ul>
  30 </li><li>
  31 <a href="extensions.html">Extensions</a>
  32 <ul><li>
  33 <a href="ext_ffi.html">FFI Library</a>
  34 <ul><li>
  35 <a href="ext_ffi_tutorial.html">FFI Tutorial</a>
  36 </li><li>
  37 <a href="ext_ffi_api.html">ffi.* API</a>
  38 </li><li>
  39 <a href="ext_ffi_int64.html">64 bit Integers</a>
  40 </li><li>
  41 <a href="ext_ffi_semantics.html">FFI Semantics</a>
  42 </li></ul>
  43 </li><li>
  44 <a href="ext_jit.html">jit.* Library</a>
  45 </li><li>
  46 <a href="ext_c_api.html">Lua/C API</a>
  47 </li></ul>
  48 </li><li>
  49 <a href="status.html">Status</a>
  50 <ul><li>
  51 <a class="current" href="changes.html">Changes</a>
  52 </li></ul>
  53 </li><li>
  54 <a href="faq.html">FAQ</a>
  55 </li><li>
  56 <a href="http://luajit.org/performance.html">Performance <span class="ext">&raquo;</span></a>
  57 </li><li>
  58 <a href="http://luajit.org/download.html">Download <span class="ext">&raquo;</span></a>
  59 </li></ul>
  60 </div>
  61 <div id="main">
  62 <p>
  63 This is a list of changes between the released versions of LuaJIT.<br>
  64 The current <span style="color: #c00000;">development version</span> is <strong>LuaJIT&nbsp;2.0.0-beta5</strong>.<br>
  65 The current <span style="color: #0000c0;">stable version</span> is <strong>LuaJIT&nbsp;1.1.6</strong>.
  66 </p>
  67 <p>
  68 Please check the
  69 <a href="http://luajit.org/changes.html"><span class="ext">&raquo;</span>&nbsp;Online Change History</a>
  70 to see whether newer versions are available.
  71 </p>
  72
  73 <div class="major" style="background: #ffd0d0;">
  74 <h2 id="LuaJIT-2.0.0-beta5">LuaJIT 2.0.0-beta5 &mdash; 2010-08-24</h2>
  75 <ul>
  76 <li>Correctness and completeness:
  77 <ul>
  78 <li>Fix trace exit dispatch to function headers.</li>
  79 <li>Fix Windows and OSX builds with LUAJIT_DISABLE_JIT.</li>
  80 <li>Reorganize and fix placement of generated machine code on x64.</li>
  81 <li>Fix TNEW in x64 interpreter.</li>
  82 <li>Do not eliminate PHIs for values only referenced from side exits.</li>
  83 <li>OS-independent canonicalization of strings for non-finite numbers.</li>
  84 <li>Fix <tt>string.char()</tt> range check on x64.</li>
  85 <li>Fix <tt>tostring()</tt> resolving within <tt>print()</tt>.</li>
  86 <li>Fix error handling for <tt>next()</tt>.</li>
  87 <li>Fix passing of constant arguments to external calls on x64.</li>
  88 <li>Fix interpreter argument check for two-argument SSE math functions.</li>
  89 <li>Fix C frame chain corruption caused by <tt>lua_cpcall()</tt>.</li>
  90 <li>Fix return from <tt>pcall()</tt> within active hook.</li>
  91 </ul></li>
  92 <li>Structural and performance enhancements:
  93 <ul>
  94 <li>Replace on-trace GC frame syncing with interpreter exit.</li>
  95 <li>Improve hash lookup specialization by not removing dead keys during GC.</li>
  96 <li>Turn traces into true GC objects.</li>
  97 <li>Avoid starting a GC cycle immediately after library init.</li>
  98 <li>Add weak guards to improve dead-code elimination.</li>
  99 <li>Speed up string interning.</li>
 100 </ul></li>
 101 </ul>
 102
 103 <h2 id="LuaJIT-2.0.0-beta4">LuaJIT 2.0.0-beta4 &mdash; 2010-03-28</h2>
 104 <ul>
 105 <li>Correctness and completeness:
 106 <ul>
 107 <li>Fix precondition for on-trace creation of table keys.</li>
 108 <li>Fix <tt>{f()}</tt> on x64 when table is resized.</li>
 109 <li>Fix folding of ordered comparisons with same references.</li>
 110 <li>Fix snapshot restores for multi-result bytecodes.</li>
 111 <li>Fix potential hang when recording bytecode with nested closures.</li>
 112 <li>Fix recording of <tt>getmetatable()</tt>, <tt>tonumber()</tt> and bad argument types.</li>
 113 <li>Fix SLOAD fusion across returns to lower frames.</li>
 114 </ul></li>
 115 <li>Structural and performance enhancements:
 116 <ul>
 117 <li>Add array bounds check elimination. <tt>-Oabc</tt> is enabled by default.</li>
 118 <li>More tuning for x64, e.g. smaller table objects.</li>
 119 </ul></li>
 120 </ul>
 121
 122 <h2 id="LuaJIT-2.0.0-beta3">LuaJIT 2.0.0-beta3 &mdash; 2010-03-07</h2>
 123 <ul>
 124 <li>LuaJIT x64 port:</li>
 125 <ul>
 126 <li>Port integrated memory allocator to Linux/x64, Windows/x64 and OSX/x64.</li>
 127 <li>Port interpreter and JIT compiler to x64.</li>
 128 <li>Port DynASM to x64.</li>
 129 <li>Many 32/64 bit cleanups in the VM.</li>
 130 <li>Allow building the interpreter with either x87 or SSE2 arithmetics.</li>
 131 <li>Add external unwinding and C++ exception interop (default on x64).</li>
 132 </ul></li>
 133 <li>Correctness and completeness:
 134 <ul>
 135 <li>Fix constructor bytecode generation for certain conditional values.</li>
 136 <li>Fix some cases of ordered string comparisons.</li>
 137 <li>Fix <tt>lua_tocfunction()</tt>.</li>
 138 <li>Fix cutoff register in JMP bytecode for some conditional expressions.</li>
 139 <li>Fix PHI marking algorithm for references from variant slots.</li>
 140 <li>Fix <tt>package.cpath</tt> for non-default PREFIX.</li>
 141 <li>Fix DWARF2 frame unwind information for interpreter on OSX.</li>
 142 <li>Drive the GC forward on string allocations in the parser.</li>
 143 <li>Implement call/return hooks (zero-cost if disabled).</li>
 144 <li>Implement yield from C hooks.</li>
 145 <li>Disable JIT compiler on older non-SSE2 CPUs instead of aborting.</li>
 146 </ul></li>
 147 <li>Structural and performance enhancements:
 148 <ul>
 149 <li>Compile recursive code (tail-, up- and down-recursion).</li>
 150 <li>Improve heuristics for bytecode penalties and blacklisting.</li>
 151 <li>Split CALL/FUNC recording and clean up fast function call semantics.</li>
 152 <li>Major redesign of internal function call handling.</li>
 153 <li>Improve FOR loop const specialization and integerness checks.</li>
 154 <li>Switch to pre-initialized stacks. Avoid frame-clearing.</li>
 155 <li>Colocation of prototypes and related data: bytecode, constants, debug info.</li>
 156 <li>Cleanup parser and streamline bytecode generation.</li>
 157 <li>Add support for weak IR references to register allocator.</li>
 158 <li>Switch to compressed, extensible snapshots.</li>
 159 <li>Compile returns to frames below the start frame.</li>
 160 <li>Improve alias analysis of upvalues using a disambiguation hash value.</li>
 161 <li>Compile floor/ceil/trunc to SSE2 helper calls or SSE4.1 instructions.</li>
 162 <li>Add generic C call handling to IR and backend.</li>
 163 <li>Improve KNUM fuse vs. load heuristics.</li>
 164 <li>Compile various <tt>io.*()</tt> functions.</li>
 165 <li>Compile <tt>math.sinh()</tt>, <tt>math.cosh()</tt>, <tt>math.tanh()</tt>
 166 and <tt>math.random()</tt>.</li>
 167 </ul></li>
 168 </ul>
 169
 170 <h2 id="LuaJIT-2.0.0-beta2">LuaJIT 2.0.0-beta2 &mdash; 2009-11-09</h2>
 171 <ul>
 172 <li>Reorganize build system. Build static+shared library on POSIX.</li>
 173 <li>Allow C++ exception conversion on all platforms
 174 using a wrapper function.</li>
 175 <li>Automatically catch C++ exceptions and rethrow Lua error
 176 (DWARF2 only).</li>
 177 <li>Check for the correct x87 FPU precision at strategic points.</li>
 178 <li>Always use wrappers for libm functions.</li>
 179 <li>Resurrect metamethod name strings before copying them.</li>
 180 <li>Mark current trace, even if compiler is idle.</li>
 181 <li>Ensure FILE metatable is created only once.</li>
 182 <li>Fix type comparisons when different integer types are involved.</li>
 183 <li>Fix <tt>getmetatable()</tt> recording.</li>
 184 <li>Fix TDUP with dead keys in template table.</li>
 185 <li><tt>jit.flush(tr)</tt> returns status.
 186 Prevent manual flush of a trace that's still linked.</li>
 187 <li>Improve register allocation heuristics for invariant references.</li>
 188 <li>Compile the push/pop variants of <tt>table.insert()</tt> and
 189 <tt>table.remove()</tt>.</li>
 190 <li>Compatibility with MSVC <tt>link&nbsp/debug</tt>.</li>
 191 <li>Fix <tt>lua_iscfunction()</tt>.</li>
 192 <li>Fix <tt>math.random()</tt> when compiled with <tt>-fpic</tt> (OSX).</li>
 193 <li>Fix <tt>table.maxn()</tt>.</li>
 194 <li>Bump <tt>MACOSX_DEPLOYMENT_TARGET</tt> to <tt>10.4</tt></li>
 195 <li><tt>luaL_check*()</tt> and <tt>luaL_opt*()</tt> now support
 196 negative arguments, too.<br>
 197 This matches the behavior of Lua 5.1, but not the specification.</li>
 198 </ul>
 199
 200 <h2 id="LuaJIT-2.0.0-beta1">LuaJIT 2.0.0-beta1 &mdash; 2009-10-31</h2>
 201 <ul>
 202 <li>This is the first public release of LuaJIT 2.0.</li>
 203 <li>The whole VM has been rewritten from the ground up, so there's
 204 no point in listing differences over earlier versions.</li>
 205 </ul>
 206 </div>
 207
 208 <div class="major" style="background: #d0d0ff;">
 209 <h2 id="LuaJIT-1.1.6">LuaJIT 1.1.6 &mdash; 2010-03-28</h2>
 210 <ul>
 211 <li>Added fixes for the
 212 <a href="http://www.lua.org/bugs.html#5.1.4"><span class="ext">&raquo;</span>&nbsp;currently known bugs in Lua 5.1.4</a>.</li>
 213 <li>Removed wrong GC check in <tt>jit_createstate()</tt>.
 214 Thanks to Tim Mensch.</li>
 215 <li>Fixed bad assertions while compiling <tt>table.insert()</tt> and
 216 <tt>table.remove()</tt>.</li>
 217 </ul>
 218
 219 <h2 id="LuaJIT-1.1.5">LuaJIT 1.1.5 &mdash; 2008-10-25</h2>
 220 <ul>
 221 <li>Merged with Lua 5.1.4. Fixes all
 222 <a href="http://www.lua.org/bugs.html#5.1.3"><span class="ext">&raquo;</span>&nbsp;known bugs in Lua 5.1.3</a>.</li>
 223 </ul>
 224
 225 <h2 id="LuaJIT-1.1.4">LuaJIT 1.1.4 &mdash; 2008-02-05</h2>
 226 <ul>
 227 <li>Merged with Lua 5.1.3. Fixes all
 228 <a href="http://www.lua.org/bugs.html#5.1.2"><span class="ext">&raquo;</span>&nbsp;known bugs in Lua 5.1.2</a>.</li>
 229 <li>Fixed possible (but unlikely) stack corruption while compiling
 230 <tt>k^x</tt> expressions.</li>
 231 <li>Fixed DynASM template for cmpss instruction.</li>
 232 </ul>
 233
 234 <h2 id="LuaJIT-1.1.3">LuaJIT 1.1.3 &mdash; 2007-05-24</h2>
 235 <ul>
 236 <li>Merged with Lua 5.1.2. Fixes all
 237 <a href="http://www.lua.org/bugs.html#5.1.1"><span class="ext">&raquo;</span>&nbsp;known bugs in Lua 5.1.1</a>.</li>
 238 <li>Merged pending Lua 5.1.x fixes: "return -nil" bug, spurious count hook call.</li>
 239 <li>Remove a (sometimes) wrong assertion in <tt>luaJIT_findpc()</tt>.</li>
 240 <li>DynASM now allows labels for displacements and <tt>.aword</tt>.</li>
 241 <li>Fix some compiler warnings for DynASM glue (internal API change).</li>
 242 <li>Correct naming for SSSE3 (temporarily known as SSE4) in DynASM and x86 disassembler.</li>
 243 <li>The loadable debug modules now handle redirection to stdout
 244 (e.g. <tt>-j&nbsp;trace=-</tt>).</li>
 245 </ul>
 246
 247 <h2 id="LuaJIT-1.1.2">LuaJIT 1.1.2 &mdash; 2006-06-24</h2>
 248 <ul>
 249 <li>Fix MSVC inline assembly: use only local variables with
 250 <tt>lua_number2int()</tt>.</li>
 251 <li>Fix "attempt to call a thread value" bug on Mac OS X:
 252 make values of consts used as lightuserdata keys unique
 253 to avoid joining by the compiler/linker.</li>
 254 </ul>
 255
 256 <h2 id="LuaJIT-1.1.1">LuaJIT 1.1.1 &mdash; 2006-06-20</h2>
 257 <ul>
 258 <li>Merged with Lua 5.1.1. Fixes all
 259 <a href="http://www.lua.org/bugs.html#5.1"><span class="ext">&raquo;</span>&nbsp;known bugs in Lua 5.1</a>.</li>
 260 <li>Enforce (dynamic) linker error for EXE/DLL version mismatches.</li>
 261 <li>Minor changes to DynASM: faster preprocessing, smaller encoding
 262 for some immediates.</li>
 263 </ul>
 264 <p>
 265 This release is in sync with Coco 1.1.1 (see the
 266 <a href="http://coco.luajit.org/changes.html"><span class="ext">&raquo;</span>&nbsp;Coco Change History</a>).
 267 </p>
 268
 269 <h2 id="LuaJIT-1.1.0">LuaJIT 1.1.0 &mdash; 2006-03-13</h2>
 270 <ul>
 271 <li>Merged with Lua 5.1 (final).</li>
 272
 273 <li>New JIT call frame setup:
 274 <ul>
 275 <li>The C stack is kept 16 byte aligned (faster).
 276 Mandatory for Mac OS X on Intel, too.</li>
 277 <li>Faster calling conventions for internal C helper functions.</li>
 278 <li>Better instruction scheduling for function prologue, OP_CALL and
 279 OP_RETURN.</li>
 280 </ul></li>
 281
 282 <li>Miscellaneous optimizations:
 283 <ul>
 284 <li>Faster loads of FP constants. Remove narrow-to-wide store-to-load
 285 forwarding stalls.</li>
 286 <li>Use (scalar) SSE2 ops (if the CPU supports it) to speed up slot moves
 287 and FP to integer conversions.</li>
 288 <li>Optimized the two-argument form of <tt>OP_CONCAT</tt> (<tt>a..b</tt>).</li>
 289 <li>Inlined <tt>OP_MOD</tt> (<tt>a%b</tt>).
 290 With better accuracy than the C variant, too.</li>
 291 <li>Inlined <tt>OP_POW</tt> (<tt>a^b</tt>). Unroll <tt>x^k</tt> or
 292 use <tt>k^x = 2^(log2(k)*x)</tt> or call <tt>pow()</tt>.</li>
 293 </ul></li>
 294
 295 <li>Changes in the optimizer:
 296 <ul>
 297 <li>Improved hinting for table keys derived from table values
 298 (<tt>t1[t2[x]]</tt>).</li>
 299 <li>Lookup hinting now works with arbitrary object types and
 300 supports index chains, too.</li>
 301 <li>Generate type hints for arithmetic and comparison operators,
 302 OP_LEN, OP_CONCAT and OP_FORPREP.</li>
 303 <li>Remove several hint definitions in favour of a generic COMBINE hint.</li>
 304 <li>Complete rewrite of <tt>jit.opt_inline</tt> module
 305 (ex <tt>jit.opt_lib</tt>).</li>
 306 </ul></li>
 307
 308 <li>Use adaptive deoptimization:
 309 <ul>
 310 <li>If runtime verification of a contract fails, the affected
 311 instruction is recompiled and patched on-the-fly.
 312 Regular programs will trigger deoptimization only occasionally.</li>
 313 <li>This avoids generating code for uncommon fallback cases
 314 most of the time. Generated code is up to 30% smaller compared to
 315 LuaJIT&nbsp;1.0.3.</li>
 316 <li>Deoptimization is used for many opcodes and contracts:
 317 <ul>
 318 <li>OP_CALL, OP_TAILCALL: type mismatch for callable.</li>
 319 <li>Inlined calls: closure mismatch, parameter number and type mismatches.</li>
 320 <li>OP_GETTABLE, OP_SETTABLE: table or key type and range mismatches.</li>
 321 <li>All arithmetic and comparison operators, OP_LEN, OP_CONCAT,
 322 OP_FORPREP: operand type and range mismatches.</li>
 323 </ul></li>
 324 <li>Complete redesign of the debug and traceback info
 325 (bytecode &harr; mcode) to support deoptimization.
 326 Much more flexible and needs only 50% of the space.</li>
 327 <li>The modules <tt>jit.trace</tt>, <tt>jit.dumphints</tt> and
 328 <tt>jit.dump</tt> handle deoptimization.</li>
 329 </ul></li>
 330
 331 <li>Inlined many popular library functions
 332 (for commonly used arguments only):
 333 <ul>
 334 <li>Most <tt>math.*</tt> functions (the 18 most used ones)
 335 [2x-10x faster].</li>
 336 <li><tt>string.len</tt>, <tt>string.sub</tt> and <tt>string.char</tt>
 337 [2x-10x faster].</li>
 338 <li><tt>table.insert</tt>, <tt>table.remove</tt> and <tt>table.getn</tt>
 339 [3x-5x faster].</li>
 340 <li><tt>coroutine.yield</tt> and <tt>coroutine.resume</tt>
 341 [3x-5x faster].</li>
 342 <li><tt>pairs</tt>, <tt>ipairs</tt> and the corresponding iterators
 343 [8x-15x faster].</li>
 344 </ul></li>
 345
 346 <li>Changes in the core and loadable modules and the stand-alone executable:
 347 <ul>
 348 <li>Added <tt>jit.version</tt>, <tt>jit.version_num</tt>
 349 and <tt>jit.arch</tt>.</li>
 350 <li>Reorganized some internal API functions (<tt>jit.util.*mcode*</tt>).</li>
 351 <li>The <tt>-j dump</tt> output now shows JSUB names, too.</li>
 352 <li>New x86 disassembler module written in pure Lua. No dependency
 353 on ndisasm anymore. Flexible API, very compact (500 lines)
 354 and complete (x87, MMX, SSE, SSE2, SSE3, SSSE3, privileged instructions).</li>
 355 <li><tt>luajit -v</tt> prints the LuaJIT version and copyright
 356 on a separate line.</li>
 357 </ul></li>
 358
 359 <li>Added SSE, SSE2, SSE3 and SSSE3 support to DynASM.</li>
 360 <li>Miscellaneous doc changes. Added a section about
 361 <a href="install.html#embedding">embedding LuaJIT</a>.</li>
 362 </ul>
 363 <p>
 364 This release is in sync with Coco 1.1.0 (see the
 365 <a href="http://coco.luajit.org/changes.html"><span class="ext">&raquo;</span>&nbsp;Coco Change History</a>).
 366 </p>
 367 </div>
 368
 369 <div class="major" style="background: #ffffd0;">
 370 <h2 id="LuaJIT-1.0.3">LuaJIT 1.0.3 &mdash; 2005-09-08</h2>
 371 <ul>
 372 <li>Even more docs.</li>
 373 <li>Unified closure checks in <tt>jit.*</tt>.</li>
 374 <li>Fixed some range checks in <tt>jit.util.*</tt>.</li>
 375 <li>Fixed __newindex call originating from <tt>jit_settable_str()</tt>.</li>
 376 <li>Merged with Lua 5.1 alpha (including early bug fixes).</li>
 377 </ul>
 378 <p>
 379 This is the first public release of LuaJIT.
 380 </p>
 381
 382 <h2 id="LuaJIT-1.0.2">LuaJIT 1.0.2 &mdash; 2005-09-02</h2>
 383 <ul>
 384 <li>Add support for flushing the Valgrind translation cache <br>
 385 (<tt>MYCFLAGS= -DUSE_VALGRIND</tt>).</li>
 386 <li>Add support for freeing executable mcode memory to the <tt>mmap()</tt>-based
 387 variant for POSIX systems.</li>
 388 <li>Reorganized the C&nbsp;function signature handling in
 389 <tt>jit.opt_lib</tt>.</li>
 390 <li>Changed to index-based hints for inlining C&nbsp;functions.
 391 Still no support in the backend for inlining.</li>
 392 <li>Hardcode <tt>HEAP_CREATE_ENABLE_EXECUTE</tt> value if undefined.</li>
 393 <li>Misc. changes to the <tt>jit.*</tt> modules.</li>
 394 <li>Misc. changes to the Makefiles.</li>
 395 <li>Lots of new docs.</li>
 396 <li>Complete doc reorg.</li>
 397 </ul>
 398 <p>
 399 Not released because Lua 5.1 alpha came out today.
 400 </p>
 401
 402 <h2 id="LuaJIT-1.0.1">LuaJIT 1.0.1 &mdash; 2005-08-31</h2>
 403 <ul>
 404 <li>Missing GC step in <tt>OP_CONCAT</tt>.</li>
 405 <li>Fix result handling for C &ndash;> JIT calls.</li>
 406 <li>Detect CPU feature bits.</li>
 407 <li>Encode conditional moves (<tt>fucomip</tt>) only when supported.</li>
 408 <li>Add fallback instructions for FP compares.</li>
 409 <li>Add support for <tt>LUA_COMPAT_VARARG</tt>. Still disabled by default.</li>
 410 <li>MSVC needs a specific place for the <tt>CALLBACK</tt> attribute
 411 (David Burgess).</li>
 412 <li>Misc. doc updates.</li>
 413 </ul>
 414 <p>
 415 Interim non-public release.
 416 Special thanks to Adam D. Moss for reporting most of the bugs.
 417 </p>
 418
 419 <h2 id="LuaJIT-1.0.0">LuaJIT 1.0.0 &mdash; 2005-08-29</h2>
 420 <p>
 421 This is the initial non-public release of LuaJIT.
 422 </p>
 423 </div>
 424 <br class="flush">
 425 </div>
 426 <div id="foot">
 427 <hr class="hide">
 428 Copyright &copy; 2005-2011 Mike Pall
 429 <span class="noprint">
 430 &middot;
 431 <a href="contact.html">Contact</a>
 432 </span>
 433 </div>
 434 </body>
 435 </html>