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_semantics.html">FFI Semantics</a>
  40 </li></ul>
  41 </li><li>
  42 <a href="ext_jit.html">jit.* Library</a>
  43 </li><li>
  44 <a href="ext_c_api.html">Lua/C API</a>
  45 </li></ul>
  46 </li><li>
  47 <a href="status.html">Status</a>
  48 <ul><li>
  49 <a class="current" href="changes.html">Changes</a>
  50 </li></ul>
  51 </li><li>
  52 <a href="faq.html">FAQ</a>
  53 </li><li>
  54 <a href="http://luajit.org/performance.html">Performance <span class="ext">&raquo;</span></a>
  55 </li><li>
  56 <a href="http://luajit.org/download.html">Download <span class="ext">&raquo;</span></a>
  57 </li></ul>
  58 </div>
  59 <div id="main">
  60 <p>
  61 This is a list of changes between the released versions of LuaJIT.<br>
  62 The current <span style="color: #c00000;">development version</span> is <strong>LuaJIT&nbsp;2.0.0-beta9</strong>.<br>
  63 The current <span style="color: #0000c0;">stable version</span> is <strong>LuaJIT&nbsp;1.1.7</strong>.
  64 </p>
  65 <p>
  66 Please check the
  67 <a href="http://luajit.org/changes.html"><span class="ext">&raquo;</span>&nbsp;Online Change History</a>
  68 to see whether newer versions are available.
  69 </p>
  70
  71 <div class="major" style="background: #ffd0d0;">
  72 <h2 id="LuaJIT-2.0.0-beta9">LuaJIT 2.0.0-beta9 &mdash; 2011-12-14</h2>
  73 <ul>
  74 <li>New features:
  75 <ul>
  76 <li>PPC port of LuaJIT is complete. Default is the dual-number port
  77 (usually faster). Single-number port selectable via <tt>src/Makefile</tt>
  78 at build time.</li>
  79 <li>Add FFI callback support.</li>
  80 <li>Extend <tt>-b</tt> to generate <tt>.c</tt>, <tt>.h</tt> or <tt>.obj/.o</tt>
  81 files with embedded bytecode.</li>
  82 <li>Allow loading embedded bytecode with <tt>require()</tt>.</li>
  83 <li>From Lua 5.2: Change to <tt>'\z'</tt> escape. Reject undefined escape
  84 sequences.</li>
  85 </ul></li>
  86 <li>Correctness and completeness:
  87 <ul>
  88 <li>Fix OSX 10.7 build. Fix <tt>install_name</tt> and versioning on OSX.</li>
  89 <li>Fix iOS build.</li>
  90 <li>Install <tt>dis_arm.lua</tt>, too.</li>
  91 <li>Mark installed shared library as executable.</li>
  92 <li>Add debug option to <tt>msvcbuild.bat</tt> and improve error handling.</li>
  93 <li>Fix data-flow analysis for iterators.</li>
  94 <li>Fix forced unwinding triggered by external unwinder.</li>
  95 <li>Record missing <tt>for</tt> loop slot loads (return to lower frame).</li>
  96 <li>Always use ANSI variants of Windows system functions.</li>
  97 <li>Fix GC barrier for multi-result table constructor (<tt>TSETM</tt>).</li>
  98 <li>Fix/add various FOLD rules.</li>
  99 <li>Add potential PHI for number conversions due to type instability.</li>
 100 <li>Do not eliminate PHIs only referenced from other PHIs.</li>
 101 <li>Correctly anchor implicit number to string conversions in Lua/C API.</li>
 102 <li>Fix various stack limit checks.</li>
 103 <li>x64: Use thread-safe exceptions for external unwinding (GCC platforms).</li>
 104 <li>x64: Fix result type of cdata index conversions.</li>
 105 <li>x64: Fix <tt>math.random()</tt> and <tt>bit.bswap()</tt> code generation.</li>
 106 <li>x64: Fix <tt>lightuserdata</tt> comparisons.</li>
 107 <li>x64: Always extend stack-passed arguments to pointer size.</li>
 108 <li>ARM: Many fixes to code generation backend.</li>
 109 <li>PPC/e500: Fix dispatch for binop metamethods.</li>
 110 <li>PPC/e500: Save/restore condition registers when entering/leaving the VM.</li>
 111 <li>PPC/e500: Fix write barrier in stores of strings to upvalues.</li>
 112 </ul></li>
 113 <li>FFI library:
 114 <ul>
 115 <li>Fix C comment parsing.</li>
 116 <li>Fix snapshot optimization for cdata comparisons.</li>
 117 <li>Fix recording of const/enum lookups in namespaces.</li>
 118 <li>Fix call argument and return handling for <tt>I8/U8/I16/U16</tt> types.</li>
 119 <li>Fix unfused loads of float fields.</li>
 120 <li>Fix <tt>ffi.string()</tt> recording.</li>
 121 <li>Save <tt>GetLastError()</tt> around <tt>ffi.load()</tt> and symbol
 122 resolving, too.</li>
 123 <li>Improve ld script detection in <tt>ffi.load()</tt>.</li>
 124 <li>Record loads/stores to external variables in namespaces.</li>
 125 <li>Compile calls to stdcall, fastcall and vararg functions.</li>
 126 <li>Treat function ctypes like pointers in comparisons.</li>
 127 <li>Resolve <tt>__call</tt> metamethod for pointers, too.</li>
 128 <li>Record C function calls with bool return values.</li>
 129 <li>Record <tt>ffi.errno()</tt>.</li>
 130 <li>x86: Fix number to <tt>uint32_t</tt> conversion rounding.</li>
 131 <li>x86: Fix 64 bit arithmetic in assembler backend.</li>
 132 <li>x64: Fix struct-by-value calling conventions.</li>
 133 <li>ARM: Ensure invocation of SPLIT pass for float conversions.</li>
 134 </ul></li>
 135 <li>Structural and performance enhancements:
 136 <ul>
 137 <li>Display trace types with <tt>-jv</tt> and <tt>-jdump</tt>.</li>
 138 <li>Record isolated calls. But prefer recording loops over calls.</li>
 139 <li>Specialize to prototype for non-monomorphic functions. Solves the
 140 trace-explosion problem for closure-heavy programming styles.</li>
 141 <li>Always generate a portable <tt>vmdef.lua</tt>. Easier for distros.</li>
 142 </ul></li>
 143 </ul>
 144
 145 <h2 id="LuaJIT-2.0.0-beta8">LuaJIT 2.0.0-beta8 &mdash; 2011-06-23</h2>
 146 <ul>
 147 <li>New features:
 148 <ul>
 149 <li>Soft-float ARM port of LuaJIT is complete.</li>
 150 <li>Add support for bytecode loading/saving and <tt>-b</tt> command line
 151 option.</li>
 152 <li>From Lua 5.2: <tt>__len</tt> metamethod for tables
 153 (disabled by default).</li>
 154 </ul></li>
 155 <li>Correctness and completeness:
 156 <ul>
 157 <li>ARM: Misc. fixes for interpreter.</li>
 158 <li>x86/x64: Fix <tt>bit.*</tt> argument checking in interpreter.</li>
 159 <li>Catch early out-of-memory in memory allocator initialization.</li>
 160 <li>Fix data-flow analysis for paths leading to an upvalue close.</li>
 161 <li>Fix check for missing arguments in <tt>string.format()</tt>.</li>
 162 <li>Fix Solaris/x86 build (note: not a supported target).</li>
 163 <li>Fix recording of loops with instable directions in side traces.</li>
 164 <li>x86/x64: Fix fusion of comparisons with <tt>u8</tt>/<tt>u16</tt>
 165 <tt>XLOAD</tt>.</li>
 166 <li>x86/x64: Fix register allocation for variable shifts.</li>
 167 </ul></li>
 168 <li>FFI library:
 169 <ul>
 170 <li>Add <tt>ffi.errno()</tt>. Save <tt>errno</tt>/<tt>GetLastError()</tt>
 171 around allocations etc.</li>
 172 <li>Fix <tt>__gc</tt> for VLA/VLS cdata objects.</li>
 173 <li>Fix recording of casts from 32 bit cdata pointers to integers.</li>
 174 <li><tt>tonumber(cdata)</tt> returns <tt>nil</tt> for non-numbers.</li>
 175 <li>Show address pointed to for <tt>tostring(pointer)</tt>.</li>
 176 <li>Print <tt>NULL</tt> pointers as <tt>"cdata&lt;... *&gt;: NULL"</tt>.</li>
 177 <li>Support <tt>__tostring</tt> metamethod for pointers to structs, too.</li>
 178 </ul></li>
 179 <li>Structural and performance enhancements:
 180 <ul>
 181 <li>More tuning for loop unrolling heuristics.</li>
 182 <li>Flatten and compress in-memory debug info (saves ~70%).</li>
 183 </ul></li>
 184 </ul>
 185
 186 <h2 id="LuaJIT-2.0.0-beta7">LuaJIT 2.0.0-beta7 &mdash; 2011-05-05</h2>
 187 <ul>
 188 <li>New features:
 189 <ul>
 190 <li>ARM port of the LuaJIT interpreter is complete.</li>
 191 <li>FFI library: Add <tt>ffi.gc()</tt>, <tt>ffi.metatype()</tt>,
 192 <tt>ffi.istype()</tt>.</li>
 193 <li>FFI library: Resolve ld script redirection in <tt>ffi.load()</tt>.</li>
 194 <li>From Lua 5.2: <tt>package.searchpath()</tt>, <tt>fp:read("*L")</tt>,
 195 <tt>load(string)</tt>.</li>
 196 <li>From Lua 5.2, disabled by default: empty statement,
 197 <tt>table.unpack()</tt>, modified <tt>coroutine.running()</tt>.</li>
 198 </ul></li>
 199 <li>Correctness and completeness:
 200 <ul>
 201 <li>FFI library: numerous fixes.</li>
 202 <li>Fix type mismatches in store-to-load forwarding.</li>
 203 <li>Fix error handling within metamethods.</li>
 204 <li>Fix <tt>table.maxn()</tt>.</li>
 205 <li>Improve accuracy of <tt>x^-k</tt> on x64.</li>
 206 <li>Fix code generation for Intel Atom in x64 mode.</li>
 207 <li>Fix narrowing of POW.</li>
 208 <li>Fix recording of retried fast functions.</li>
 209 <li>Fix code generation for <tt>bit.bnot()</tt> and multiplies.</li>
 210 <li>Fix error location within cpcall frames.</li>
 211 <li>Add workaround for old libgcc unwind bug.</li>
 212 <li>Fix <tt>lua_yield()</tt> and <tt>getmetatable(lightuserdata)</tt> on x64.</li>
 213 <li>Misc. fixes for PPC/e500 interpreter.</li>
 214 <li>Fix stack slot updates for down-recursion.</li>
 215 </ul></li>
 216 <li>Structural and performance enhancements:
 217 <ul>
 218 <li>Add dual-number mode (int/double) for the VM. Enabled for ARM.</li>
 219 <li>Improve narrowing of arithmetic operators and <tt>for</tt> loops.</li>
 220 <li>Tune loop unrolling heuristics and increase trace recorder limits.</li>
 221 <li>Eliminate dead slots in snapshots using bytecode data-flow analysis.</li>
 222 <li>Avoid phantom stores to proxy tables.</li>
 223 <li>Optimize lookups in empty proxy tables.</li>
 224 <li>Improve bytecode optimization of <tt>and</tt>/<tt>or</tt> operators.</li>
 225 </ul></li>
 226 </ul>
 227
 228 <h2 id="LuaJIT-2.0.0-beta6">LuaJIT 2.0.0-beta6 &mdash; 2011-02-11</h2>
 229 <ul>
 230 <li>New features:
 231 <ul>
 232 <li>PowerPC/e500v2 port of the LuaJIT interpreter is complete.</li>
 233 <li>Various minor features from Lua 5.2: Hex escapes in literals,
 234 <tt>'\*'</tt> escape, reversible <tt>string.format("%q",s)</tt>,
 235 <tt>"%g"</tt> pattern, <tt>table.sort</tt> checks callbacks,
 236 <tt>os.exit(status|true|false[,close])</tt>.</li>
 237 <li>Lua 5.2 <tt>__pairs</tt> and <tt>__ipairs</tt> metamethods
 238 (disabled by default).</li>
 239 <li>Initial release of the FFI library.</li>
 240 </ul></li>
 241 <li>Correctness and completeness:
 242 <ul>
 243 <li>Fix <tt>string.format()</tt> for non-finite numbers.</li>
 244 <li>Fix memory leak when compiled to use the built-in allocator.</li>
 245 <li>x86/x64: Fix unnecessary resize in <tt>TSETM</tt> bytecode.</li>
 246 <li>Fix various GC issues with traces and <tt>jit.flush()</tt>.</li>
 247 <li>x64: Fix fusion of indexes for array references.</li>
 248 <li>x86/x64: Fix stack overflow handling for coroutine results.</li>
 249 <li>Enable low-2GB memory allocation on FreeBSD/x64.</li>
 250 <li>Fix <tt>collectgarbage("count")</tt> result if more than 2GB is in use.</li>
 251 <li>Fix parsing of hex floats.</li>
 252 <li>x86/x64: Fix loop branch inversion with trailing
 253 <tt>HREF+NE/EQ</tt>.</li>
 254 <li>Add <tt>jit.os</tt> string.</li>
 255 <li><tt>coroutine.create()</tt> permits running C functions, too.</li>
 256 <li>Fix OSX build to work with newer ld64 versions.</li>
 257 <li>Fix bytecode optimization of <tt>and</tt>/<tt>or</tt> operators.</li>
 258 </ul></li>
 259 <li>Structural and performance enhancements:
 260 <ul>
 261 <li>Emit specialized bytecode for <tt>pairs()</tt>/<tt>next()</tt>.</li>
 262 <li>Improve bytecode coalescing of <tt>nil</tt> constants.</li>
 263 <li>Compile calls to vararg functions.</li>
 264 <li>Compile <tt>select()</tt>.</li>
 265 <li>Improve alias analysis, esp. for loads from allocations.</li>
 266 <li>Tuning of various compiler heuristics.</li>
 267 <li>Refactor and extend IR conversion instructions.</li>
 268 <li>x86/x64: Various backend enhancements related to the FFI.</li>
 269 <li>Add SPLIT pass to split 64 bit IR instructions for 32 bit CPUs.</li>
 270 </ul></li>
 271 </ul>
 272
 273 <h2 id="LuaJIT-2.0.0-beta5">LuaJIT 2.0.0-beta5 &mdash; 2010-08-24</h2>
 274 <ul>
 275 <li>Correctness and completeness:
 276 <ul>
 277 <li>Fix trace exit dispatch to function headers.</li>
 278 <li>Fix Windows and OSX builds with LUAJIT_DISABLE_JIT.</li>
 279 <li>Reorganize and fix placement of generated machine code on x64.</li>
 280 <li>Fix TNEW in x64 interpreter.</li>
 281 <li>Do not eliminate PHIs for values only referenced from side exits.</li>
 282 <li>OS-independent canonicalization of strings for non-finite numbers.</li>
 283 <li>Fix <tt>string.char()</tt> range check on x64.</li>
 284 <li>Fix <tt>tostring()</tt> resolving within <tt>print()</tt>.</li>
 285 <li>Fix error handling for <tt>next()</tt>.</li>
 286 <li>Fix passing of constant arguments to external calls on x64.</li>
 287 <li>Fix interpreter argument check for two-argument SSE math functions.</li>
 288 <li>Fix C frame chain corruption caused by <tt>lua_cpcall()</tt>.</li>
 289 <li>Fix return from <tt>pcall()</tt> within active hook.</li>
 290 </ul></li>
 291 <li>Structural and performance enhancements:
 292 <ul>
 293 <li>Replace on-trace GC frame syncing with interpreter exit.</li>
 294 <li>Improve hash lookup specialization by not removing dead keys during GC.</li>
 295 <li>Turn traces into true GC objects.</li>
 296 <li>Avoid starting a GC cycle immediately after library init.</li>
 297 <li>Add weak guards to improve dead-code elimination.</li>
 298 <li>Speed up string interning.</li>
 299 </ul></li>
 300 </ul>
 301
 302 <h2 id="LuaJIT-2.0.0-beta4">LuaJIT 2.0.0-beta4 &mdash; 2010-03-28</h2>
 303 <ul>
 304 <li>Correctness and completeness:
 305 <ul>
 306 <li>Fix precondition for on-trace creation of table keys.</li>
 307 <li>Fix <tt>{f()}</tt> on x64 when table is resized.</li>
 308 <li>Fix folding of ordered comparisons with same references.</li>
 309 <li>Fix snapshot restores for multi-result bytecodes.</li>
 310 <li>Fix potential hang when recording bytecode with nested closures.</li>
 311 <li>Fix recording of <tt>getmetatable()</tt>, <tt>tonumber()</tt> and bad argument types.</li>
 312 <li>Fix SLOAD fusion across returns to lower frames.</li>
 313 </ul></li>
 314 <li>Structural and performance enhancements:
 315 <ul>
 316 <li>Add array bounds check elimination. <tt>-Oabc</tt> is enabled by default.</li>
 317 <li>More tuning for x64, e.g. smaller table objects.</li>
 318 </ul></li>
 319 </ul>
 320
 321 <h2 id="LuaJIT-2.0.0-beta3">LuaJIT 2.0.0-beta3 &mdash; 2010-03-07</h2>
 322 <ul>
 323 <li>LuaJIT x64 port:
 324 <ul>
 325 <li>Port integrated memory allocator to Linux/x64, Windows/x64 and OSX/x64.</li>
 326 <li>Port interpreter and JIT compiler to x64.</li>
 327 <li>Port DynASM to x64.</li>
 328 <li>Many 32/64 bit cleanups in the VM.</li>
 329 <li>Allow building the interpreter with either x87 or SSE2 arithmetics.</li>
 330 <li>Add external unwinding and C++ exception interop (default on x64).</li>
 331 </ul></li>
 332 <li>Correctness and completeness:
 333 <ul>
 334 <li>Fix constructor bytecode generation for certain conditional values.</li>
 335 <li>Fix some cases of ordered string comparisons.</li>
 336 <li>Fix <tt>lua_tocfunction()</tt>.</li>
 337 <li>Fix cutoff register in JMP bytecode for some conditional expressions.</li>
 338 <li>Fix PHI marking algorithm for references from variant slots.</li>
 339 <li>Fix <tt>package.cpath</tt> for non-default PREFIX.</li>
 340 <li>Fix DWARF2 frame unwind information for interpreter on OSX.</li>
 341 <li>Drive the GC forward on string allocations in the parser.</li>
 342 <li>Implement call/return hooks (zero-cost if disabled).</li>
 343 <li>Implement yield from C hooks.</li>
 344 <li>Disable JIT compiler on older non-SSE2 CPUs instead of aborting.</li>
 345 </ul></li>
 346 <li>Structural and performance enhancements:
 347 <ul>
 348 <li>Compile recursive code (tail-, up- and down-recursion).</li>
 349 <li>Improve heuristics for bytecode penalties and blacklisting.</li>
 350 <li>Split CALL/FUNC recording and clean up fast function call semantics.</li>
 351 <li>Major redesign of internal function call handling.</li>
 352 <li>Improve FOR loop const specialization and integerness checks.</li>
 353 <li>Switch to pre-initialized stacks. Avoid frame-clearing.</li>
 354 <li>Colocation of prototypes and related data: bytecode, constants, debug info.</li>
 355 <li>Cleanup parser and streamline bytecode generation.</li>
 356 <li>Add support for weak IR references to register allocator.</li>
 357 <li>Switch to compressed, extensible snapshots.</li>
 358 <li>Compile returns to frames below the start frame.</li>
 359 <li>Improve alias analysis of upvalues using a disambiguation hash value.</li>
 360 <li>Compile floor/ceil/trunc to SSE2 helper calls or SSE4.1 instructions.</li>
 361 <li>Add generic C call handling to IR and backend.</li>
 362 <li>Improve KNUM fuse vs. load heuristics.</li>
 363 <li>Compile various <tt>io.*()</tt> functions.</li>
 364 <li>Compile <tt>math.sinh()</tt>, <tt>math.cosh()</tt>, <tt>math.tanh()</tt>
 365 and <tt>math.random()</tt>.</li>
 366 </ul></li>
 367 </ul>
 368
 369 <h2 id="LuaJIT-2.0.0-beta2">LuaJIT 2.0.0-beta2 &mdash; 2009-11-09</h2>
 370 <ul>
 371 <li>Reorganize build system. Build static+shared library on POSIX.</li>
 372 <li>Allow C++ exception conversion on all platforms
 373 using a wrapper function.</li>
 374 <li>Automatically catch C++ exceptions and rethrow Lua error
 375 (DWARF2 only).</li>
 376 <li>Check for the correct x87 FPU precision at strategic points.</li>
 377 <li>Always use wrappers for libm functions.</li>
 378 <li>Resurrect metamethod name strings before copying them.</li>
 379 <li>Mark current trace, even if compiler is idle.</li>
 380 <li>Ensure FILE metatable is created only once.</li>
 381 <li>Fix type comparisons when different integer types are involved.</li>
 382 <li>Fix <tt>getmetatable()</tt> recording.</li>
 383 <li>Fix TDUP with dead keys in template table.</li>
 384 <li><tt>jit.flush(tr)</tt> returns status.
 385 Prevent manual flush of a trace that's still linked.</li>
 386 <li>Improve register allocation heuristics for invariant references.</li>
 387 <li>Compile the push/pop variants of <tt>table.insert()</tt> and
 388 <tt>table.remove()</tt>.</li>
 389 <li>Compatibility with MSVC <tt>link&nbsp/debug</tt>.</li>
 390 <li>Fix <tt>lua_iscfunction()</tt>.</li>
 391 <li>Fix <tt>math.random()</tt> when compiled with <tt>-fpic</tt> (OSX).</li>
 392 <li>Fix <tt>table.maxn()</tt>.</li>
 393 <li>Bump <tt>MACOSX_DEPLOYMENT_TARGET</tt> to <tt>10.4</tt></li>
 394 <li><tt>luaL_check*()</tt> and <tt>luaL_opt*()</tt> now support
 395 negative arguments, too.<br>
 396 This matches the behavior of Lua 5.1, but not the specification.</li>
 397 </ul>
 398
 399 <h2 id="LuaJIT-2.0.0-beta1">LuaJIT 2.0.0-beta1 &mdash; 2009-10-31</h2>
 400 <ul>
 401 <li>This is the first public release of LuaJIT 2.0.</li>
 402 <li>The whole VM has been rewritten from the ground up, so there's
 403 no point in listing differences over earlier versions.</li>
 404 </ul>
 405 </div>
 406
 407 <div class="major" style="background: #d0d0ff;">
 408 <h2 id="LuaJIT-1.1.7">LuaJIT 1.1.7 &mdash; 2011-05-05</h2>
 409 <ul>
 410 <li>Added fixes for the
 411 <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>
 412 </ul>
 413
 414 <h2 id="LuaJIT-1.1.6">LuaJIT 1.1.6 &mdash; 2010-03-28</h2>
 415 <ul>
 416 <li>Added fixes for the
 417 <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>
 418 <li>Removed wrong GC check in <tt>jit_createstate()</tt>.
 419 Thanks to Tim Mensch.</li>
 420 <li>Fixed bad assertions while compiling <tt>table.insert()</tt> and
 421 <tt>table.remove()</tt>.</li>
 422 </ul>
 423
 424 <h2 id="LuaJIT-1.1.5">LuaJIT 1.1.5 &mdash; 2008-10-25</h2>
 425 <ul>
 426 <li>Merged with Lua 5.1.4. Fixes all
 427 <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>
 428 </ul>
 429
 430 <h2 id="LuaJIT-1.1.4">LuaJIT 1.1.4 &mdash; 2008-02-05</h2>
 431 <ul>
 432 <li>Merged with Lua 5.1.3. Fixes all
 433 <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>
 434 <li>Fixed possible (but unlikely) stack corruption while compiling
 435 <tt>k^x</tt> expressions.</li>
 436 <li>Fixed DynASM template for cmpss instruction.</li>
 437 </ul>
 438
 439 <h2 id="LuaJIT-1.1.3">LuaJIT 1.1.3 &mdash; 2007-05-24</h2>
 440 <ul>
 441 <li>Merged with Lua 5.1.2. Fixes all
 442 <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>
 443 <li>Merged pending Lua 5.1.x fixes: "return -nil" bug, spurious count hook call.</li>
 444 <li>Remove a (sometimes) wrong assertion in <tt>luaJIT_findpc()</tt>.</li>
 445 <li>DynASM now allows labels for displacements and <tt>.aword</tt>.</li>
 446 <li>Fix some compiler warnings for DynASM glue (internal API change).</li>
 447 <li>Correct naming for SSSE3 (temporarily known as SSE4) in DynASM and x86 disassembler.</li>
 448 <li>The loadable debug modules now handle redirection to stdout
 449 (e.g. <tt>-j&nbsp;trace=-</tt>).</li>
 450 </ul>
 451
 452 <h2 id="LuaJIT-1.1.2">LuaJIT 1.1.2 &mdash; 2006-06-24</h2>
 453 <ul>
 454 <li>Fix MSVC inline assembly: use only local variables with
 455 <tt>lua_number2int()</tt>.</li>
 456 <li>Fix "attempt to call a thread value" bug on Mac OS X:
 457 make values of consts used as lightuserdata keys unique
 458 to avoid joining by the compiler/linker.</li>
 459 </ul>
 460
 461 <h2 id="LuaJIT-1.1.1">LuaJIT 1.1.1 &mdash; 2006-06-20</h2>
 462 <ul>
 463 <li>Merged with Lua 5.1.1. Fixes all
 464 <a href="http://www.lua.org/bugs.html#5.1"><span class="ext">&raquo;</span>&nbsp;known bugs in Lua 5.1</a>.</li>
 465 <li>Enforce (dynamic) linker error for EXE/DLL version mismatches.</li>
 466 <li>Minor changes to DynASM: faster pre-processing, smaller encoding
 467 for some immediates.</li>
 468 </ul>
 469 <p>
 470 This release is in sync with Coco 1.1.1 (see the
 471 <a href="http://coco.luajit.org/changes.html"><span class="ext">&raquo;</span>&nbsp;Coco Change History</a>).
 472 </p>
 473
 474 <h2 id="LuaJIT-1.1.0">LuaJIT 1.1.0 &mdash; 2006-03-13</h2>
 475 <ul>
 476 <li>Merged with Lua 5.1 (final).</li>
 477
 478 <li>New JIT call frame setup:
 479 <ul>
 480 <li>The C stack is kept 16 byte aligned (faster).
 481 Mandatory for Mac OS X on Intel, too.</li>
 482 <li>Faster calling conventions for internal C helper functions.</li>
 483 <li>Better instruction scheduling for function prologue, OP_CALL and
 484 OP_RETURN.</li>
 485 </ul></li>
 486
 487 <li>Miscellaneous optimizations:
 488 <ul>
 489 <li>Faster loads of FP constants. Remove narrow-to-wide store-to-load
 490 forwarding stalls.</li>
 491 <li>Use (scalar) SSE2 ops (if the CPU supports it) to speed up slot moves
 492 and FP to integer conversions.</li>
 493 <li>Optimized the two-argument form of <tt>OP_CONCAT</tt> (<tt>a..b</tt>).</li>
 494 <li>Inlined <tt>OP_MOD</tt> (<tt>a%b</tt>).
 495 With better accuracy than the C variant, too.</li>
 496 <li>Inlined <tt>OP_POW</tt> (<tt>a^b</tt>). Unroll <tt>x^k</tt> or
 497 use <tt>k^x = 2^(log2(k)*x)</tt> or call <tt>pow()</tt>.</li>
 498 </ul></li>
 499
 500 <li>Changes in the optimizer:
 501 <ul>
 502 <li>Improved hinting for table keys derived from table values
 503 (<tt>t1[t2[x]]</tt>).</li>
 504 <li>Lookup hinting now works with arbitrary object types and
 505 supports index chains, too.</li>
 506 <li>Generate type hints for arithmetic and comparison operators,
 507 OP_LEN, OP_CONCAT and OP_FORPREP.</li>
 508 <li>Remove several hint definitions in favour of a generic COMBINE hint.</li>
 509 <li>Complete rewrite of <tt>jit.opt_inline</tt> module
 510 (ex <tt>jit.opt_lib</tt>).</li>
 511 </ul></li>
 512
 513 <li>Use adaptive deoptimization:
 514 <ul>
 515 <li>If runtime verification of a contract fails, the affected
 516 instruction is recompiled and patched on-the-fly.
 517 Regular programs will trigger deoptimization only occasionally.</li>
 518 <li>This avoids generating code for uncommon fallback cases
 519 most of the time. Generated code is up to 30% smaller compared to
 520 LuaJIT&nbsp;1.0.3.</li>
 521 <li>Deoptimization is used for many opcodes and contracts:
 522 <ul>
 523 <li>OP_CALL, OP_TAILCALL: type mismatch for callable.</li>
 524 <li>Inlined calls: closure mismatch, parameter number and type mismatches.</li>
 525 <li>OP_GETTABLE, OP_SETTABLE: table or key type and range mismatches.</li>
 526 <li>All arithmetic and comparison operators, OP_LEN, OP_CONCAT,
 527 OP_FORPREP: operand type and range mismatches.</li>
 528 </ul></li>
 529 <li>Complete redesign of the debug and traceback info
 530 (bytecode &harr; mcode) to support deoptimization.
 531 Much more flexible and needs only 50% of the space.</li>
 532 <li>The modules <tt>jit.trace</tt>, <tt>jit.dumphints</tt> and
 533 <tt>jit.dump</tt> handle deoptimization.</li>
 534 </ul></li>
 535
 536 <li>Inlined many popular library functions
 537 (for commonly used arguments only):
 538 <ul>
 539 <li>Most <tt>math.*</tt> functions (the 18 most used ones)
 540 [2x-10x faster].</li>
 541 <li><tt>string.len</tt>, <tt>string.sub</tt> and <tt>string.char</tt>
 542 [2x-10x faster].</li>
 543 <li><tt>table.insert</tt>, <tt>table.remove</tt> and <tt>table.getn</tt>
 544 [3x-5x faster].</li>
 545 <li><tt>coroutine.yield</tt> and <tt>coroutine.resume</tt>
 546 [3x-5x faster].</li>
 547 <li><tt>pairs</tt>, <tt>ipairs</tt> and the corresponding iterators
 548 [8x-15x faster].</li>
 549 </ul></li>
 550
 551 <li>Changes in the core and loadable modules and the stand-alone executable:
 552 <ul>
 553 <li>Added <tt>jit.version</tt>, <tt>jit.version_num</tt>
 554 and <tt>jit.arch</tt>.</li>
 555 <li>Reorganized some internal API functions (<tt>jit.util.*mcode*</tt>).</li>
 556 <li>The <tt>-j dump</tt> output now shows JSUB names, too.</li>
 557 <li>New x86 disassembler module written in pure Lua. No dependency
 558 on ndisasm anymore. Flexible API, very compact (500 lines)
 559 and complete (x87, MMX, SSE, SSE2, SSE3, SSSE3, privileged instructions).</li>
 560 <li><tt>luajit -v</tt> prints the LuaJIT version and copyright
 561 on a separate line.</li>
 562 </ul></li>
 563
 564 <li>Added SSE, SSE2, SSE3 and SSSE3 support to DynASM.</li>
 565 <li>Miscellaneous doc changes. Added a section about
 566 <a href="install.html#embedding">embedding LuaJIT</a>.</li>
 567 </ul>
 568 <p>
 569 This release is in sync with Coco 1.1.0 (see the
 570 <a href="http://coco.luajit.org/changes.html"><span class="ext">&raquo;</span>&nbsp;Coco Change History</a>).
 571 </p>
 572 </div>
 573
 574 <div class="major" style="background: #ffffd0;">
 575 <h2 id="LuaJIT-1.0.3">LuaJIT 1.0.3 &mdash; 2005-09-08</h2>
 576 <ul>
 577 <li>Even more docs.</li>
 578 <li>Unified closure checks in <tt>jit.*</tt>.</li>
 579 <li>Fixed some range checks in <tt>jit.util.*</tt>.</li>
 580 <li>Fixed __newindex call originating from <tt>jit_settable_str()</tt>.</li>
 581 <li>Merged with Lua 5.1 alpha (including early bug fixes).</li>
 582 </ul>
 583 <p>
 584 This is the first public release of LuaJIT.
 585 </p>
 586
 587 <h2 id="LuaJIT-1.0.2">LuaJIT 1.0.2 &mdash; 2005-09-02</h2>
 588 <ul>
 589 <li>Add support for flushing the Valgrind translation cache <br>
 590 (<tt>MYCFLAGS= -DUSE_VALGRIND</tt>).</li>
 591 <li>Add support for freeing executable mcode memory to the <tt>mmap()</tt>-based
 592 variant for POSIX systems.</li>
 593 <li>Reorganized the C&nbsp;function signature handling in
 594 <tt>jit.opt_lib</tt>.</li>
 595 <li>Changed to index-based hints for inlining C&nbsp;functions.
 596 Still no support in the backend for inlining.</li>
 597 <li>Hardcode <tt>HEAP_CREATE_ENABLE_EXECUTE</tt> value if undefined.</li>
 598 <li>Misc. changes to the <tt>jit.*</tt> modules.</li>
 599 <li>Misc. changes to the Makefiles.</li>
 600 <li>Lots of new docs.</li>
 601 <li>Complete doc reorg.</li>
 602 </ul>
 603 <p>
 604 Not released because Lua 5.1 alpha came out today.
 605 </p>
 606
 607 <h2 id="LuaJIT-1.0.1">LuaJIT 1.0.1 &mdash; 2005-08-31</h2>
 608 <ul>
 609 <li>Missing GC step in <tt>OP_CONCAT</tt>.</li>
 610 <li>Fix result handling for C &ndash;> JIT calls.</li>
 611 <li>Detect CPU feature bits.</li>
 612 <li>Encode conditional moves (<tt>fucomip</tt>) only when supported.</li>
 613 <li>Add fallback instructions for FP compares.</li>
 614 <li>Add support for <tt>LUA_COMPAT_VARARG</tt>. Still disabled by default.</li>
 615 <li>MSVC needs a specific place for the <tt>CALLBACK</tt> attribute
 616 (David Burgess).</li>
 617 <li>Misc. doc updates.</li>
 618 </ul>
 619 <p>
 620 Interim non-public release.
 621 Special thanks to Adam D. Moss for reporting most of the bugs.
 622 </p>
 623
 624 <h2 id="LuaJIT-1.0.0">LuaJIT 1.0.0 &mdash; 2005-08-29</h2>
 625 <p>
 626 This is the initial non-public release of LuaJIT.
 627 </p>
 628 </div>
 629 <br class="flush">
 630 </div>
 631 <div id="foot">
 632 <hr class="hide">
 633 Copyright &copy; 2005-2011 Mike Pall
 634 <span class="noprint">
 635 &middot;
 636 <a href="contact.html">Contact</a>
 637 </span>
 638 </div>
 639 </body>
 640 </html>