1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
6 <link rel="stylesheet" href="llvm.css" type="text/css">
7 <title>LLVM 2.6 Release Notes</title>
11 <div class="doc_title">LLVM 2.6 Release Notes</div>
14 <li><a href="#intro">Introduction</a></li>
15 <li><a href="#subproj">Sub-project Status Update</a></li>
16 <li><a href="#externalproj">External Projects Using LLVM 2.6</a></li>
17 <li><a href="#whatsnew">What's New in LLVM 2.6?</a></li>
18 <li><a href="GettingStarted.html">Installation Instructions</a></li>
19 <li><a href="#portability">Portability and Supported Platforms</a></li>
20 <li><a href="#knownproblems">Known Problems</a></li>
21 <li><a href="#additionalinfo">Additional Information</a></li>
24 <div class="doc_author">
25 <p>Written by the <a href="http://llvm.org">LLVM Team</a></p>
28 <!-- *********************************************************************** -->
29 <div class="doc_section">
30 <a name="intro">Introduction</a>
32 <!-- *********************************************************************** -->
34 <div class="doc_text">
36 <p>This document contains the release notes for the LLVM Compiler
37 Infrastructure, release 2.6. Here we describe the status of LLVM, including
38 major improvements from the previous release and significant known problems.
39 All LLVM releases may be downloaded from the <a
40 href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
42 <p>For more information about LLVM, including information about the latest
43 release, please check out the <a href="http://llvm.org/">main LLVM
44 web site</a>. If you have questions or comments, the <a
45 href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM Developer's Mailing
46 List</a> is a good place to send them.</p>
48 <p>Note that if you are reading this file from a Subversion checkout or the
49 main LLVM web page, this document applies to the <i>next</i> release, not the
50 current one. To see the release notes for a specific release, please see the
51 <a href="http://llvm.org/releases/">releases page</a>.</p>
58 include/llvm/Analysis/LiveValues.h => Dan
59 lib/Transforms/IPO/MergeFunctions.cpp => consider for 2.8.
60 llvm/Analysis/PointerTracking.h => Edwin wants this, consider for 2.8.
64 <!-- Unfinished features in 2.6:
67 variable debug info for optimized code
68 postalloc scheduler: anti dependence breaking, hazard recognizer?
70 loop dependence analysis
71 ELF Writer? How stable?
72 <li>PostRA scheduler improvements, ARM adoption (David Goodwin).</li>
73 2.7 supports the GDB 7.0 jit interfaces for debug info.
76 <!-- for announcement email:
80 klee web page at klee.llvm.org
81 Many new papers added to /pubs/
86 <!-- *********************************************************************** -->
87 <div class="doc_section">
88 <a name="subproj">Sub-project Status Update</a>
90 <!-- *********************************************************************** -->
92 <div class="doc_text">
94 The LLVM 2.6 distribution currently consists of code from the core LLVM
95 repository —which roughly includes the LLVM optimizers, code generators
96 and supporting tools — and the llvm-gcc repository. In addition to this
97 code, the LLVM Project includes other sub-projects that are in development. The
98 two which are the most actively developed are the <a href="#clang">Clang
99 Project</a> and the <a href="#vmkit">VMKit Project</a>.
105 <!--=========================================================================-->
106 <div class="doc_subsection">
107 <a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
110 <div class="doc_text">
112 <p>The <a href="http://clang.llvm.org/">Clang project</a> is an effort to build
113 a set of new 'LLVM native' front-end technologies for the C family of languages.
114 LLVM 2.6 is the first release to officially include Clang, and it provides a
115 production quality C and Objective-C compiler. If you are interested in fast
116 compiles and good diagnostics, we encourage you to try it out.</p>
118 <p>In addition to supporting these languages, C++ support is also <a
119 href="http://clang.llvm.org/cxx_status.html">well under way</a>, and mainline
120 Clang is able to parse the libstdc++ 4.2 headers and even codegen simple apps.
121 If you are interested in Clang C++ support or any other Clang feature, we
122 strongly encourage you to get involved on the <a
123 href="http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev">Clang front-end mailing
126 <p>In the LLVM 2.6 time-frame, the Clang team has made many improvements:</p>
129 <li>C and Objective-C support are now considered production quality.</li>
130 <li>AuroraUX / FreeBSD & OpenBSD Toolchain support.</li>
131 <li>Most of Objective-C 2.0 is now supported with the GNU runtime.</li>
132 <li>Many many bugs are fixed and many features have been added.</li>
136 <!--=========================================================================-->
137 <div class="doc_subsection">
138 <a name="clangsa">Clang Static Analyzer</a>
141 <div class="doc_text">
143 <p><b>UPDATE!</b> Previously announced in the 2.4 and 2.5 LLVM releases, the Clang project also
144 includes an early stage static source code analysis tool for <a
145 href="http://clang.llvm.org/StaticAnalysis.html">automatically finding bugs</a>
146 in C and Objective-C programs. The tool performs a growing set of checks to find
147 bugs that occur on a specific path within a program.</p>
149 <p>In the LLVM 2.6 time-frame there have been many significant improvements to
152 <p>The set of checks performed by the static analyzer continues to expand, and
153 future plans for the tool include full source-level inter-procedural analysis
154 and deeper checks such as buffer overrun detection. There are many opportunities
155 to extend and enhance the static analyzer, and anyone interested in working on
156 this project is encouraged to get involved!</p>
160 <!--=========================================================================-->
161 <div class="doc_subsection">
162 <a name="vmkit">VMKit: JVM/CLI Virtual Machine Implementation</a>
165 <div class="doc_text">
167 The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation of
168 a JVM and a CLI Virtual Machine (Microsoft .NET is an
169 implementation of the CLI) using LLVM for static and just-in-time
173 VMKit version 0.26 builds with LLVM 2.6 and you can find it on its
174 <a href="http://vmkit.llvm.org/releases/">webpage</a>. The release includes
175 bug fixes, cleanup and new features. The major changes are:</p>
179 <li>A new llcj tool to generate shared libraries or executables of Java
181 <li>Cooperative garbage collection. </li>
182 <li>Fast subtype checking (paper from Click et al [JGI'02]). </li>
183 <li>Implementation of a two-word header for Java objects instead of the orginal
184 three-word header. </li>
185 <li>Better Java specification-compliance: division by zero checks, stack
186 overflow checks, finalization and references support. </li>
192 <!--=========================================================================-->
193 <div class="doc_subsection">
194 <a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
197 <div class="doc_text">
199 The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
200 is a simple library that provides an implementation of the low-level
201 target-specific hooks required by code generation and other runtime components.
202 For example, when compiling for a 32-bit target, converting a double to a 64-bit
203 unsigned integer is compiling into a runtime call to the "__fixunsdfdi"
204 function. The compiler-rt library provides optimized implementations of this and
205 other low-level routines.</p>
208 All of the code in the compiler-rt project is available under the standard LLVM
209 License, a "BSD-style" license.</p>
213 <!--=========================================================================-->
214 <div class="doc_subsection">
215 <a name="klee">klee: Symbolic Execution and Automatic Test Case Generator</a>
218 <div class="doc_text">
220 The new LLVM <a href="http://klee.llvm.org/">klee project</a> is a symbolic
221 execution framework for programs in LLVM bitcode form. Klee tries to
222 symbolically evaluate "all" paths through the application and records state
223 transitions that lead to fault states. This allows it to construct testcases
224 that lead to faults and can even be used to verify algorithms. For more
225 details, please see the <a
226 href="http://llvm.org/pubs/2008-12-OSDI-KLEE.html">OSDI 2008 paper</a> about
231 <!--=========================================================================-->
232 <div class="doc_subsection">
233 <a name="dragonegg">Dragon Egg: An LLVM backend plugin for GCC</a>
236 <div class="doc_text">
238 <b>Duncan needs to write me</b>.
244 <!--=========================================================================-->
245 <div class="doc_subsection">
246 <a name="mc">llvm-mc: Machine Code Toolkit</a>
249 <div class="doc_text">
251 The LLVM Machine Code (MC) Toolkit project is a (very early) effort to build
252 better tools for dealing with machine code, object file formats, etc. The idea
253 is to be able to generate most of the target specific details of assemblers and
254 disassemblers from existing LLVM target .td files (with suitable enhancements),
255 and to build infrastructure for reading and writing common object file formats.
256 One of the first deliverables is to build a full assembler and integrate it into
257 the compiler, which is predicted to substantially reduce compile time in some
261 <p>In the LLVM 2.6 timeframe, the MC framework has grown to the point where it
262 can reliably parse and pretty print (with some encoding information) a
263 darwin/x86 .s file successfully, and has the very early phases of a Mach-O
264 assembler in progress. Beyond the MC framework itself, major refactoring of the
265 LLVM code generator has started. The idea is to make the code generator reason
266 about the code it is producing in a much more semantic way, rather than a
267 textual way. For example, the code generator now uses MCSection objects to
268 represent section assignments, instead of text strings that print to .section
271 <p>MC is an early and ongoing project that will hopefully continue to lead to
272 many improvements in the code generator and build infrastructure useful for many
279 <!-- *********************************************************************** -->
280 <div class="doc_section">
281 <a name="externalproj">External Projects Using LLVM 2.6</a>
283 <!-- *********************************************************************** -->
285 <!--=========================================================================-->
286 <div class="doc_subsection">
287 <a name="Rubinius">Rubinius</a>
290 <div class="doc_text">
291 <p><a href="http://github.com/evanphx/rubinius">Rubinius</a> is an environment
292 for running Ruby code which strives to write as much of the core class
293 implementation in Ruby as possible. Combined with a bytecode interpreting VM, it
294 uses LLVM to optimize and compile ruby code down to machine code. Techniques
295 such as type feedback, method inlining, and uncommon traps are all used to
296 remove dynamism from ruby execution and increase performance.</p>
298 <p>Since LLVM 2.5, Rubinius has made several major leaps forward, implementing
299 a counter based JIT, type feedback, and speculative method inlining.
304 <!--=========================================================================-->
305 <div class="doc_subsection">
306 <a name="macruby">MacRuby</a>
309 <div class="doc_text">
312 <a href="http://macruby.org">MacRuby</a> is an implementation of Ruby on top of
313 core Mac OS X technologies, such as the Objective-C common runtime and garbage
314 collector, and the CoreFoundation framework. It is principally developed by
315 Apple and aims at enabling the creation of full-fledged Mac OS X applications.
319 MacRuby uses LLVM for optimization passes, JIT and AOT compilation of Ruby
320 expressions. It also uses zero-cost DWARF exceptions to implement Ruby exception
326 <!--=========================================================================-->
327 <div class="doc_subsection">
328 <a name="pure">Pure</a>
331 <div class="doc_text">
333 <a href="http://pure-lang.googlecode.com/">Pure</a>
334 is an algebraic/functional programming language based on term rewriting.
335 Programs are collections of equations which are used to evaluate expressions in
336 a symbolic fashion. Pure offers dynamic typing, eager and lazy evaluation,
337 lexical closures, a hygienic macro system (also based on term rewriting),
338 built-in list and matrix support (including list and matrix comprehensions) and
339 an easy-to-use C interface. The interpreter uses LLVM as a backend to
340 JIT-compile Pure programs to fast native code.</p>
342 <p>Pure versions 0.31 and later have been tested and are known to work with
343 LLVM 2.6 (and continue to work with older LLVM releases >= 2.3 as well).
348 <!--=========================================================================-->
349 <div class="doc_subsection">
350 <a name="ldc">LLVM D Compiler</a>
353 <div class="doc_text">
355 <a href="http://www.dsource.org/projects/ldc">LDC</a> is an implementation of
356 the D Programming Language using the LLVM optimizer and code generator.
357 The LDC project works great with the LLVM 2.6 release. General improvements in
359 cycle have included new inline asm constraint handling, better debug info
360 support, general bugfixes, and better x86-64 support. This has allowed
361 some major improvements in LDC, getting us much closer to being as
362 fully featured as the original DMD compiler from DigitalMars.
366 <!--=========================================================================-->
367 <div class="doc_subsection">
368 <a name="RoadsendPHP">Roadsend PHP</a>
371 <div class="doc_text">
373 <a href="http://code.roadsend.com/rphp">Roadsend PHP</a> (rphp) is an open
374 source implementation of the PHP programming
375 language that uses LLVM for its optimizer, JIT, and static compiler. This is a
376 reimplementation of an earlier project that is now based on LLVM.</p>
379 <!--=========================================================================-->
380 <div class="doc_subsection">
381 <a name="UnladenSwallow">Unladen Swallow</a>
384 <div class="doc_text">
386 <a href="http://code.google.com/p/unladen-swallow/">Unladen Swallow</a> is a
387 branch of <a href="http://python.org/">Python</a> intended to be fully
388 compatible and significantly faster. It uses LLVM's optimization passes and JIT
392 <!--=========================================================================-->
393 <div class="doc_subsection">
394 <a name="llvm-lua">llvm-lua</a>
397 <div class="doc_text">
399 <a href="http://code.google.com/p/llvm-lua/">LLVM-Lua</a> uses LLVM to add JIT
400 & static compiling support to the Lua VM. Lua bytecode is analyzed to
401 remove type checks, then LLVM is used to compile those bytecodes down to machine
407 <!-- *********************************************************************** -->
408 <div class="doc_section">
409 <a name="whatsnew">What's New in LLVM 2.6?</a>
411 <!-- *********************************************************************** -->
413 <div class="doc_text">
415 <p>This release includes a huge number of bug fixes, performance tweaks, and
416 minor improvements. Some of the major improvements and new features are listed
422 <!--=========================================================================-->
423 <div class="doc_subsection">
424 <a name="majorfeatures">Major New Features</a>
427 <div class="doc_text">
429 <p>LLVM 2.6 includes several major new capabilities:</p>
432 <li>New <a href="#compiler-rt">compiler-rt</a>, <A href="#klee">klee</a>,
433 and <a href="#mc">machine code toolkit</a> sub-projects.</li>
434 <li>Debug information now includes line numbers when optimizations are enabled.
435 This allows statistical sampling tools like oprofile and Shark to map
436 samples back to source lines.</li>
437 <li>LLVM now includes new experimental backends to support the MSP430, SystemZ,
438 and BlackFin architectures.</li>
439 <li>LLVM supports a new <a href="GoldPlugin.html">Gold Linker Plugin</a> which
440 enables support for <a href="LinkTimeOptimization.html">transparent
441 link-time optimization</a> on ELF targets when used with the Gold binutils
443 <li>LLVM now supports doing optimization and code generation on multiple threads
444 by allowing multiple "LLVMContext" objects to exist. Please see the <a
445 href="ProgrammersManual.html#threading">threading entry in the Programmer's
446 Manual</a> for more information.</li>
447 <li>LLVM now has experimental support for <a
448 href="http://nondot.org/~sabre/LLVMNotes/EmbeddedMetadata.txt">embedded
449 metadata</a> in LLVM IR, though the implementation is not guaranteed to be
450 final and the .bc file format may change in future releases. Debug info
451 does not yet use this format in LLVM 2.6.</p>
457 <!--=========================================================================-->
458 <div class="doc_subsection">
459 <a name="coreimprovements">LLVM IR and Core Improvements</a>
462 <div class="doc_text">
463 <p>LLVM IR has several new features that are used by our existing front-ends and
464 can be useful if you are writing a front-end for LLVM:</p>
467 <li>Getelementpr instruction now allows any integer type for array/pointer indexes.</li>
468 <li>Inbounds for GEP</li>
469 <li>NSW/NUW/exact div</li>
470 <li>LSR promotes int induction variables to 64-bit on 64-bit targets, major perf boost for numerical code.</li>
471 <li>LSR now analyzes pointer expressions (e.g. getelementptrs), not just integers.</li>
472 <li>new linkage types linkonce_odr, weak_odr, linker_private, and available_externally.</li>
473 <li>New fadd, fsub, fmul instructions and classes. </li>
474 <li>Target intrinsics can now return multiple results.</li>
479 <!--=========================================================================-->
480 <div class="doc_subsection">
481 <a name="optimizer">Optimizer Improvements</a>
484 <div class="doc_text">
486 <p>In addition to a large array of bug fixes and minor performance tweaks, this
487 release includes a few major enhancements and additions to the optimizers:</p>
491 <li>SRoA improvements for vector unions, memset, arbitrary weird bitfield accesses etc. It now produces "strange" sized integers.</li>
492 <li>Inliner reuse stack space when inlining arrays?</li>
493 <li>Enabled GVN Load PRE.</li>
494 <li>New Static Single Information (SSI) construction pass (not used by anything yet, experimental).</li>
501 <!--=========================================================================-->
502 <div class="doc_subsection">
503 <a name="codegen">Target Independent Code Generator Improvements</a>
506 <div class="doc_text">
508 <p>We have put a significant amount of work into the code generator
509 infrastructure, which allows us to implement more aggressive algorithms and make
514 <li> -asm-verbose now prints location info (with -g) and loop nest info.</li>
515 <li>Tblgen now supports multiclass inheritance and a number of new string and
516 list operations like !(subst), !(foreach), !car, !cdr, !null, !if, !cast.
517 These make the .td files more expressive and allow more aggressive factoring
518 of duplication across instruction patterns.</li>
519 <li>New MachineVerifier pass.</li>
520 <li>Machine LICM, hoists things like constant pool loads, loads from readonly stubs, vector constant synthesization code, etc.</li>
521 <li>Machine Sinking</li>
522 <li>target-specific intrinsics (r63765)</li>
523 <li>Regalloc improvements for commuting, various spiller peephole optimizations, cross-class coalescing.</li>
524 <li><tt>llc -enable-value-prop</tt>, propagation of value info (sign/zero ext info) from one MBB to another</li>
525 <li>Regalloc hints for allocation stuff: Evan r73381/r73671. Finished/enabled?</li>
526 <li>Stack slot coloring for register spills (denser stack frames)</li>
527 <li>SelectionDAGS: New BuildVectorSDNode (r65296), and ISD::VECTOR_SHUFFLE (r69952 / PR2957)</li>
528 <li>Experimental support for shrink wrapping support in PEI.</li>
529 <li>Experimental support for writing ELF .o files directly from the compiler,
530 it works well for many simple C testcases, but doesn't support exception
531 handling, debug info, inline assembly, etc.</li>
532 <li>Targets can now specify register allocation hints through
533 MachineRegisterInfo:: setRegAllocationHint. A regalloc hint consists 1) hint
534 type, 2) physical register number. A hint type of zero specifies a register
535 allocation preference. Other hint type values are target specific which are
536 resolved by TargetRegisterInfo::ResolveRegAllocHint. An example of which is
537 the ARM target can uses register hint to request that the register allocator
538 provide an even / odd register pair to two virtual registers. It is
539 important to note the register allocation hints are just hints. There is no
540 guarantee the register allocators will be able to satisfy the hints.</li>
545 <!--=========================================================================-->
546 <div class="doc_subsection">
547 <a name="x86">X86-32 and X86-64 Target Improvements</a>
550 <div class="doc_text">
551 <p>New features of the X86 target include:
556 <li>Preliminary support for addrspace 256 -> GS, 257 -> FS, known problems: CodeGenerator.html#x86_memory</li>
557 <li>Support for softfloat modes, typically used by OS kernels.</li>
559 <li>X86-64: better modeling of implicit zero extensions, eliminates a lot of redundant zexts</li>
560 <li>X86-64 TLS support for local exec and initial exec.</li>
561 <li>Better modeling of H registerts as subregs.</li>
562 <li>Vector icmp/fcmp now work with SSE codegen.</li>
563 <li>SSE 4.2 support.</li>
564 <li>all global variable reference logic is now in ClassifyGlobalReference.</li>
570 <!--=========================================================================-->
571 <div class="doc_subsection">
572 <a name="pic16">PIC16 Target Improvements</a>
575 <div class="doc_text">
576 <p>New features of the PIC16 target include:
580 <li>Support for floating-point, indirect function calls, and
581 passing/returning aggregate types to functions.
582 <li>The code generator is able to generate debug info into output COFF files.
583 <li>Support for placing an object into a specific section or at a specific
584 address in memory.</li>
587 <p>Things not yet supported:</p>
590 <li>Variable arguments.</li>
591 <li>Interrupts/programs.</li>
596 <!--=========================================================================-->
597 <div class="doc_subsection">
598 <a name="ARM">ARM Target Improvements</a>
601 <div class="doc_text">
602 <p>New features of the ARM target include:
607 <li>Preliminary support for processors, such as the Cortex-A8 and Cortex-A9,
608 that implement version v7-A of the ARM architecture. The ARM backend now
609 supports both the Thumb2 and Advanced SIMD (Neon) instruction sets. The
610 AAPCS-VFP "hard float" calling conventions are also supported with the
611 <tt>-float-abi=hard</tt> flag. These features are still somewhat experimental
612 and subject to change. The Neon intrinsics, in particular, may change in future
616 ARM AAPCS-VFP hard float ABI is supported.
617 ARM calling convention code is now tblgen generated instead of manual.
618 ARM: NEON support. neonfp for doing single precision fp with neon instead of VFP.
624 <!--=========================================================================-->
625 <div class="doc_subsection">
626 <a name="OtherTarget">Other Target Specific Improvements</a>
629 <div class="doc_text">
630 <p>New features of other targets include:
634 <li>Mips now supports O32 Calling Convention.</li>
635 <li>Many improvements to the 32-bit PowerPC SVR4 ABI (used on powerpc-linux)
636 support, lots of bugs fixed.</li>
637 <li>Added support for the 64-bit PowerPC SVR4 ABI (used on powerpc64-linux).
638 Needs more testing.</li>
643 <!--=========================================================================-->
644 <div class="doc_subsection">
645 <a name="executionengine">Interpreter and JIT Improvements</a>
648 <div class="doc_text">
651 <li>The JIT now supports generating more than 16M of code.</li>
652 <li>When configured with --with-oprofile, the JIT can now inform oprofile about
653 JIT'd code, allowing oprofile to get line number and function name
654 information for JIT'd functions.</li>
655 <li>When "libffi" is available, the LLVM interpreter now uses it, which supports
656 calling almost arbitrary external (natively compiled) functions.</li>
657 <li>Clients of the JIT can now register a 'JITEventListener' object to receive
658 callbacks when the JIT emits or frees machine code. The OProfile support
659 uses this mechanism.</li>
665 <!--=========================================================================-->
666 <div class="doc_subsection">
667 <a name="newapis">New Useful APIs</a>
670 <div class="doc_text">
673 <li>New EngineBuilder class for creating JITs: r76276</li>
674 New PrettyStackTrace, crashes of llvm tools should give some indication of what the compiler was doing at the time of the crash (e.g. running a pass), and print out command line arguments.
675 StringRef class, Twine class.
676 New WeakVH and AssertingVH and CallbackVH classes.
677 New llvm/ADT/Triple class.
678 llvm_report_error() error handling API (llvm/Support/ErrorHandling.h)
679 New llvm/System/Atomic.h, llvm/System/RWMutex.h for portable atomic ops, rw locks.
680 New SourceMgr, SMLoc classes for simple parsers with caret diagnostics and #include support, (used by
681 tablegen, llvm-mc, the .ll parser, FileCheck, etc)
688 <!--=========================================================================-->
689 <div class="doc_subsection">
690 <a name="otherimprovements">Other Improvements and New Features</a>
693 <div class="doc_text">
694 <p>Other miscellaneous features include:</p>
697 <li>LLVM now includes a new internal '<a
698 href="http://llvm.org/cmds/FileCheck.html">FileCheck</a>' tool which allows
699 writing much more accurate regression tests that run faster. Please see the
700 <a href="TestingGuide.html#FileCheck">FileCheck section of the Testing
701 Guide</a> for more information.</li>
702 <li>LLVM profile information support has been significantly improved to produce
703 correct use counts, and has support for edge profiling with reduced runtime
704 overhead. Combined, the generated profile information is both more correct and
705 imposes about half as much overhead (2.6. from 12% to 6% overhead on SPEC
707 <li>Many extensions to the C APIs.</li>
708 <li>LLVM 2.6 includes a brand new experimental LLVM bindings to the Ada2005
709 programming language.</li>
713 * Dynamic plugins now work on Windows.
714 * New option property: init. Makes possible to provide default values for
715 options defined in plugins (interface to cl::init).
716 * New example: Skeleton, shows how to create a standalone LLVMC-based driver.
717 * New example: mcc16, a driver for the PIC16 toolchain.</li>
724 <!--=========================================================================-->
725 <div class="doc_subsection">
726 <a name="changes">Major Changes and Removed Features</a>
729 <div class="doc_text">
731 <p>If you're already an LLVM user or developer with out-of-tree changes based
732 on LLVM 2.5, this section lists some "gotchas" that you may run into upgrading
733 from the previous release.</p>
737 <li>The Itanium (IA64) backend has been removed. It was not supported and
739 <li>The BigBlock register allocator has been removed, it also bitrotted.</li>
740 <li>The C Backend (-march=c) is no longer considered part of the LLVM release
741 criteria. We still want it to work, but no one is maintaining it and it lacks
742 support for arbitrary precision integers and other important IR features.</li>
745 LLVM build now builds all libraries as .a files instead of some
746 libraries as relinked .o files. This requires some APIs like
747 InitializeAllTargets.h. TargetRegistry!
752 <p>In addition, many APIs have changed in this release. Some of the major LLVM
757 no use of hash_set/hash_map, no more llvm::OStream
758 Use raw_ostream for everything, killed off llvm/Streams.h and DOUT
761 <li>LLVM's global uniquing tables for <tt>Type</tt>s and <tt>Constant</tt>s have
762 been privatized into members of an <tt>LLVMContext</tt>. A number of APIs
763 now take an <tt>LLVMContext</tt> as a parameter. To smooth the transition
764 for clients that will only ever use a single context, the new
765 <tt>getGlobalContext()</tt> API can be used to access a default global
766 context which can be passed in any and all cases where a context is
768 <li>The <tt>getABITypeSize</tt> methods are now called <tt>getAllocSize</tt>.</li>
769 <li>The <tt>Add</tt>, <tt>Sub</tt>, and <tt>Mul</tt> operators are no longer
770 overloaded for floating-point types. Floating-point addition, subtraction,
771 and multiplication are now represented with new operators <tt>FAdd</tt>,
772 <tt>FSub</tt>, and <tt>FMul</tt>. In the <tt>IRBuilder</tt> API,
773 <tt>CreateAdd</tt>, <tt>CreateSub</tt>, <tt>CreateMul</tt>, and
774 <tt>CreateNeg</tt> should only be used for integer arithmetic now;
775 <tt>CreateFAdd</tt>, <tt>CreateFSub</tt>, <tt>CreateFMul</tt>, and
776 <tt>CreateFNeg</tt> should now be used for floating-point arithmetic.</li>
777 <li>The DynamicLibrary class can no longer be constructed, its functionality has
778 moved to static member functions.</li>
779 <li><tt>raw_fd_ostream</tt>'s constructor for opening a given filename now
780 takes an extra <tt>Force</tt> argument. If <tt>Force</tt> is set to
781 <tt>false</tt>, an error will be reported if a file with the given name
782 already exists. If <tt>Force</tt> is set to <tt>true</tt>, the file will
783 be silently truncated (which is the behavior before this flag was
785 <li><tt>SCEVHandle</tt> no longer exists, because reference counting is no
786 longer done for <tt>SCEV*</tt> objects, instead <tt>const SCEV*</tt> should be
789 <li>Many APIs, notably <tt>llvm::Value</tt>, now use the <tt>StringRef</tt>
790 and <tt>Twine</tt> classes instead of passing <tt>const char*</tt>
791 or <tt>std::string</tt>, as described in
792 the <a href="ProgrammersManual.html#string_apis">Programmer's Manual</a>. Most
793 clients should be unaffected by this transition, unless they are used to <tt>Value::getName()</tt> returning a string. Here are some tips on updating to 2.6:
795 <li><tt>getNameStr()</tt> is still available, and matches the old
796 behavior. Replacing <tt>getName()</tt> calls with this is an safe option,
797 although more efficient alternatives are now possible.</li>
799 <li>If you were just relying on <tt>getName()</tt> being able to be sent to
800 a <tt>std::ostream</tt>, consider migrating
801 to <tt>llvm::raw_ostream</tt>.</li>
803 <li>If you were using <tt>getName().c_str()</tt> to get a <tt>const
804 char*</tt> pointer to the name, you can use <tt>getName().data()</tt>.
805 Note that this string (as before), may not be the entire name if the
806 name containts embedded null characters.</li>
808 <li>If you were using operator plus on the result of <tt>getName()</tt> and
809 treating the result as an <tt>std::string</tt>, you can either
810 uses <tt>Twine::str</tt> to get the result as an <tt>std::string</tt>, or
811 could move to a <tt>Twine</tt> based design.</li>
813 <li><tt>isName()</tt> should be replaced with comparison
814 against <tt>getName()</tt> (this is now efficient).
818 <li>The registration interfaces for backend Targets has changed (what was
819 previously TargetMachineRegistry). For backend authors, see the <a href="WritingAnLLVMBackend.html#TargetRegistration">Writing An LLVM Backend</a> guide. For clients, the notable API changes are:
821 <li><tt>TargetMachineRegistry</tt> has been renamed
822 to <tt>TargetRegistry</tt>.</li>
824 <li>Clients should move to using the <tt>TargetRegistry::lookupTarget()</tt>
825 function to find targets.</li>
829 <li>llvm-dis now fails if output file exists, instead of dumping to stdout.
830 FIXME: describe any other tool changes due to the raw_fd_ostream change. FIXME:
831 This is not an API change, maybe there should be a tool changes section?</li>
832 <li>temporarely due to Context API change passes should call doInitialization()
833 method of the pass they inherit from, otherwise Context is NULL.
834 FIXME: remove this entry when this is no longer needed.<li>
841 <!-- *********************************************************************** -->
842 <div class="doc_section">
843 <a name="portability">Portability and Supported Platforms</a>
845 <!-- *********************************************************************** -->
847 <div class="doc_text">
849 <p>LLVM is known to work on the following platforms:</p>
852 <li>Intel and AMD machines (IA32, X86-64, AMD64, EMT-64) running Red Hat
853 Linux, Fedora Core, FreeBSD and AuroraUX (and probably other unix-like systems).</li>
854 <li>PowerPC and X86-based Mac OS X systems, running 10.3 and above in 32-bit
855 and 64-bit modes.</li>
856 <li>Intel and AMD machines running on Win32 using MinGW libraries (native).</li>
857 <li>Intel and AMD machines running on Win32 with the Cygwin libraries (limited
858 support is available for native builds with Visual C++).</li>
859 <li>Sun UltraSPARC workstations running Solaris 10.</li>
860 <li>Alpha-based machines running Debian GNU/Linux.</li>
863 <p>The core LLVM infrastructure uses GNU autoconf to adapt itself
864 to the machine and operating system on which it is built. However, minor
865 porting may be required to get LLVM to work on new platforms. We welcome your
866 portability patches and reports of successful builds or error messages.</p>
870 <!-- *********************************************************************** -->
871 <div class="doc_section">
872 <a name="knownproblems">Known Problems</a>
874 <!-- *********************************************************************** -->
876 <div class="doc_text">
878 <p>This section contains significant known problems with the LLVM system,
879 listed by component. If you run into a problem, please check the <a
880 href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
881 there isn't already one.</p>
884 <li>LLVM will not correctly compile on Solaris and/or OpenSolaris
885 using the stock GCC 3.x.x series 'out the box',
886 See: <a href="#brokengcc">Broken versions of GCC and other tools</a>.
887 However, A <a href="http://pkg.auroraux.org/GCC">Modern GCC Build</a>
888 for x86/x64 has been made available from the third party AuroraUX Project
889 that has been meticulously tested for bootstrapping LLVM & Clang.</li>
894 <!-- ======================================================================= -->
895 <div class="doc_subsection">
896 <a name="experimental">Experimental features included with this release</a>
899 <div class="doc_text">
901 <p>The following components of this LLVM release are either untested, known to
902 be broken or unreliable, or are in early development. These components should
903 not be relied on, and bugs should not be filed against them, but they may be
904 useful to some people. In particular, if you would like to work on one of these
905 components, please contact us on the <a
906 href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
909 <li>The MSIL, Alpha, SPU, MIPS, and PIC16 backends are experimental.</li>
910 <li>The <tt>llc</tt> "<tt>-filetype=asm</tt>" (the default) is the only
911 supported value for this option.</li>
916 <!-- ======================================================================= -->
917 <div class="doc_subsection">
918 <a name="x86-be">Known problems with the X86 back-end</a>
921 <div class="doc_text">
924 <li>The X86 backend does not yet support
925 all <a href="http://llvm.org/PR879">inline assembly that uses the X86
926 floating point stack</a>. It supports the 'f' and 't' constraints, but not
928 <li>The X86 backend generates inefficient floating point code when configured
929 to generate code for systems that don't have SSE2.</li>
930 <li>Win64 code generation wasn't widely tested. Everything should work, but we
931 expect small issues to happen. Also, llvm-gcc cannot build the mingw64
932 runtime currently due
933 to <a href="http://llvm.org/PR2255">several</a>
934 <a href="http://llvm.org/PR2257">bugs</a> and due to lack of support for
936 'u' inline assembly constraint and for X87 floating point inline assembly.</li>
937 <li>The X86-64 backend does not yet support the LLVM IR instruction
938 <tt>va_arg</tt>. Currently, the llvm-gcc and front-ends support variadic
939 argument constructs on X86-64 by lowering them manually.</li>
944 <!-- ======================================================================= -->
945 <div class="doc_subsection">
946 <a name="ppc-be">Known problems with the PowerPC back-end</a>
949 <div class="doc_text">
952 <li>The Linux PPC32/ABI support needs testing for the interpreter and static
953 compilation, and lacks support for debug information.</li>
958 <!-- ======================================================================= -->
959 <div class="doc_subsection">
960 <a name="arm-be">Known problems with the ARM back-end</a>
963 <div class="doc_text">
966 <li>Support for the Advanced SIMD (Neon) instruction set is still incomplete
967 and not well tested. Some features may not work at all, and the code quality
968 may be poor in some cases.</li>
969 <li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
970 processors, thumb programs can crash or produce wrong
971 results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
972 <li>Compilation for ARM Linux OABI (old ABI) is supported but not fully tested.
978 <!-- ======================================================================= -->
979 <div class="doc_subsection">
980 <a name="sparc-be">Known problems with the SPARC back-end</a>
983 <div class="doc_text">
986 <li>The SPARC backend only supports the 32-bit SPARC ABI (-m32); it does not
987 support the 64-bit SPARC ABI (-m64).</li>
992 <!-- ======================================================================= -->
993 <div class="doc_subsection">
994 <a name="mips-be">Known problems with the MIPS back-end</a>
997 <div class="doc_text">
1000 <li>The O32 ABI is not fully supported.</li>
1001 <li>64-bit MIPS targets are not supported yet.</li>
1006 <!-- ======================================================================= -->
1007 <div class="doc_subsection">
1008 <a name="alpha-be">Known problems with the Alpha back-end</a>
1011 <div class="doc_text">
1015 <li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
1016 appropriate nops inserted to ensure restartability.</li>
1021 <!-- ======================================================================= -->
1022 <div class="doc_subsection">
1023 <a name="c-be">Known problems with the C back-end</a>
1026 <div class="doc_text">
1029 <li><a href="http://llvm.org/PR802">The C backend has only basic support for
1030 inline assembly code</a>.</li>
1031 <li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common
1032 C++ programs</a>, preventing intermixing between C++ compiled by the CBE and
1033 C++ code compiled with <tt>llc</tt> or native compilers.</li>
1034 <li>The C backend does not support all exception handling constructs.</li>
1035 <li>The C backend does not support arbitrary precision integers.</li>
1041 <!-- ======================================================================= -->
1042 <div class="doc_subsection">
1043 <a name="c-fe">Known problems with the llvm-gcc C front-end</a>
1046 <div class="doc_text">
1048 <p>llvm-gcc does not currently support <a href="http://llvm.org/PR869">Link-Time
1049 Optimization</a> on most platforms "out-of-the-box". Please inquire on the
1050 LLVMdev mailing list if you are interested.</p>
1052 <p>The only major language feature of GCC not supported by llvm-gcc is
1053 the <tt>__builtin_apply</tt> family of builtins. However, some extensions
1054 are only supported on some targets. For example, trampolines are only
1055 supported on some targets (these are used when you take the address of a
1056 nested function).</p>
1058 <p>If you run into GCC extensions which are not supported, please let us know.
1063 <!-- ======================================================================= -->
1064 <div class="doc_subsection">
1065 <a name="c++-fe">Known problems with the llvm-gcc C++ front-end</a>
1068 <div class="doc_text">
1070 <p>The C++ front-end is considered to be fully
1071 tested and works for a number of non-trivial programs, including LLVM
1072 itself, Qt, Mozilla, etc.</p>
1075 <li>Exception handling works well on the X86 and PowerPC targets. Currently
1076 only Linux and Darwin targets are supported (both 32 and 64 bit).</li>
1081 <!-- ======================================================================= -->
1082 <div class="doc_subsection">
1083 <a name="fortran-fe">Known problems with the llvm-gcc Fortran front-end</a>
1086 <div class="doc_text">
1088 <li>Fortran support generally works, but there are still several unresolved bugs
1089 in Bugzilla. Please see the tools/gfortran component for details.</li>
1093 <!-- ======================================================================= -->
1094 <div class="doc_subsection">
1095 <a name="ada-fe">Known problems with the llvm-gcc Ada front-end</a>
1098 <div class="doc_text">
1099 The llvm-gcc 4.2 Ada compiler works fairly well; however, this is not a mature
1100 technology, and problems should be expected.
1102 <li>The Ada front-end currently only builds on X86-32. This is mainly due
1103 to lack of trampoline support (pointers to nested functions) on other platforms.
1104 However, it <a href="http://llvm.org/PR2006">also fails to build on X86-64</a>
1105 which does support trampolines.</li>
1106 <li>The Ada front-end <a href="http://llvm.org/PR2007">fails to bootstrap</a>.
1107 This is due to lack of LLVM support for <tt>setjmp</tt>/<tt>longjmp</tt> style
1108 exception handling, which is used internally by the compiler.
1109 Workaround: configure with --disable-bootstrap.</li>
1110 <li>The c380004, <a href="http://llvm.org/PR2010">c393010</a>
1111 and <a href="http://llvm.org/PR2421">cxg2021</a> ACATS tests fail
1112 (c380004 also fails with gcc-4.2 mainline).
1113 If the compiler is built with checks disabled then <a href="http://llvm.org/PR2010">c393010</a>
1114 causes the compiler to go into an infinite loop, using up all system memory.</li>
1115 <li>Some GCC specific Ada tests continue to crash the compiler.</li>
1116 <li>The -E binder option (exception backtraces)
1117 <a href="http://llvm.org/PR1982">does not work</a> and will result in programs
1118 crashing if an exception is raised. Workaround: do not use -E.</li>
1119 <li>Only discrete types <a href="http://llvm.org/PR1981">are allowed to start
1120 or finish at a non-byte offset</a> in a record. Workaround: do not pack records
1121 or use representation clauses that result in a field of a non-discrete type
1122 starting or finishing in the middle of a byte.</li>
1123 <li>The <tt>lli</tt> interpreter <a href="http://llvm.org/PR2009">considers
1124 'main' as generated by the Ada binder to be invalid</a>.
1125 Workaround: hand edit the file to use pointers for <tt>argv</tt> and
1126 <tt>envp</tt> rather than integers.</li>
1127 <li>The <tt>-fstack-check</tt> option <a href="http://llvm.org/PR2008">is
1132 <!-- ======================================================================= -->
1133 <div class="doc_subsection">
1134 <a name="ocaml-bindingse">Known problems with the O'Caml bindings</a>
1137 <div class="doc_text">
1139 <p>The Llvm.Linkage module is broken, and has incorrect values. Only
1140 Llvm.Linkage.External, Llvm.Linkage.Available_externally, and
1141 Llvm.Linkage.Link_once will be correct. If you need any of the other linkage
1142 modes, you'll have to write an external C library in order to expose the
1143 functionality. This has been fixed in the trunk.</p>
1146 <!-- *********************************************************************** -->
1147 <div class="doc_section">
1148 <a name="additionalinfo">Additional Information</a>
1150 <!-- *********************************************************************** -->
1152 <div class="doc_text">
1154 <p>A wide variety of additional information is available on the <a
1155 href="http://llvm.org">LLVM web page</a>, in particular in the <a
1156 href="http://llvm.org/docs/">documentation</a> section. The web page also
1157 contains versions of the API documentation which is up-to-date with the
1158 Subversion version of the source code.
1159 You can access versions of these documents specific to this release by going
1160 into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
1162 <p>If you have any questions or comments about LLVM, please feel free to contact
1163 us via the <a href="http://llvm.org/docs/#maillist"> mailing
1168 <!-- *********************************************************************** -->
1172 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1173 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
1174 <a href="http://validator.w3.org/check/referer"><img
1175 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
1177 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
1178 Last modified: $Date$