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 3.0 Release Notes</title>
11 <h1>LLVM 3.0 Release Notes</h1>
13 <img align=right src="http://llvm.org/img/DragonSmall.png"
14 width="136" height="136" alt="LLVM Dragon Logo">
17 <li><a href="#intro">Introduction</a></li>
18 <li><a href="#subproj">Sub-project Status Update</a></li>
19 <li><a href="#externalproj">External Projects Using LLVM 3.0</a></li>
20 <li><a href="#whatsnew">What's New in LLVM 3.0?</a></li>
21 <li><a href="GettingStarted.html">Installation Instructions</a></li>
22 <li><a href="#knownproblems">Known Problems</a></li>
23 <li><a href="#additionalinfo">Additional Information</a></li>
26 <div class="doc_author">
27 <p>Written by the <a href="http://llvm.org/">LLVM Team</a></p>
31 <h1 style="color:red">These are in-progress notes for the upcoming LLVM 3.0
34 <a href="http://llvm.org/releases/2.9/docs/ReleaseNotes.html">LLVM 2.9
35 Release Notes</a>.</h1>
38 <!-- *********************************************************************** -->
40 <a name="intro">Introduction</a>
42 <!-- *********************************************************************** -->
46 <p>This document contains the release notes for the LLVM Compiler
47 Infrastructure, release 3.0. Here we describe the status of LLVM, including
48 major improvements from the previous release, improvements in various
49 subprojects of LLVM, and some of the current users of the code.
50 All LLVM releases may be downloaded from
51 the <a href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
53 <p>For more information about LLVM, including information about the latest
54 release, please check out the <a href="http://llvm.org/">main LLVM web
55 site</a>. If you have questions or comments,
56 the <a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM
57 Developer's Mailing List</a> is a good place to send them.</p>
59 <p>Note that if you are reading this file from a Subversion checkout or the main
60 LLVM web page, this document applies to the <i>next</i> release, not the
61 current one. To see the release notes for a specific release, please see the
62 <a href="http://llvm.org/releases/">releases page</a>.</p>
67 <!-- *********************************************************************** -->
69 <a name="subproj">Sub-project Status Update</a>
71 <!-- *********************************************************************** -->
75 <p>The LLVM 3.0 distribution currently consists of code from the core LLVM
76 repository (which roughly includes the LLVM optimizers, code generators and
77 supporting tools), and the Clang repository. In
78 addition to this code, the LLVM Project includes other sub-projects that are
79 in development. Here we include updates on these subprojects.</p>
81 <!--=========================================================================-->
83 <a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
88 <p><a href="http://clang.llvm.org/">Clang</a> is an LLVM front end for the C,
89 C++, and Objective-C languages. Clang aims to provide a better user
90 experience through expressive diagnostics, a high level of conformance to
91 language standards, fast compilation, and low memory use. Like LLVM, Clang
92 provides a modular, library-based architecture that makes it suitable for
93 creating or integrating with other development tools. Clang is considered a
94 production-quality compiler for C, Objective-C, C++ and Objective-C++ on x86
95 (32- and 64-bit), and for Darwin/ARM targets.</p>
97 <p>In the LLVM 3.0 time-frame, the Clang team has made many improvements:</p>
100 <li>Greatly improved support for building C++ applications, with greater
101 stability and better diagnostics.</li>
103 <li><a href="http://clang.llvm.org/cxx_status.html">Improved support</a> for
104 the <a href="http://www.iso.org/iso/iso_catalogue/catalogue_tc/catalogue_detail.htm?csnumber=50372">C++
105 2011</a> standard (aka "C++'0x"), including implementations of non-static data member
106 initializers, alias templates, delegating constructors, range-based
107 for loops, and implicitly-generated move constructors and move assignment
108 operators, among others.</li>
110 <li>Implemented support for some features of the upcoming C1x standard,
111 including static assertions and generic selections.</li>
113 <li>Better detection of include and linking paths for system headers and
114 libraries, especially for Linux distributions.</li>
116 <li>Several improvements to Objective-C support, including:
119 <li><a href="http://clang.llvm.org/docs/AutomaticReferenceCounting.html">
120 Automatic Reference Counting</a> (ARC) and an improved memory model
121 cleanly separating object and C memory.</li>
123 <li>A migration tool for moving manual retain/release code to ARC</li>
125 <li>Better support for data hiding, allowing instance variables to be
126 declared in implementation contexts or class extensions</li>
127 <li>Weak linking support for Objective-C classes</li>
128 <li>Improved static type checking by inferring the return type of methods
129 such as +alloc and -init.</li>
132 Some new Objective-C features require either the Mac OS X 10.7 / iOS 5
133 Objective-C runtime, or version 1.6 or later of the GNUstep Objective-C
134 runtime version.</li>
136 <li>Implemented a number of optimizations in <tt>libclang</tt>, the Clang C
137 interface, to improve the performance of code completion and the mapping
138 from source locations to abstract syntax tree nodes.</li>
142 <p>If Clang rejects your code but another compiler accepts it, please take a
143 look at the <a href="http://clang.llvm.org/compatibility.html">language
144 compatibility</a> guide to make sure this is not intentional or a known
149 <!--=========================================================================-->
151 <a name="dragonegg">DragonEgg: GCC front-ends, LLVM back-end</a>
155 <p><a href="http://dragonegg.llvm.org/">DragonEgg</a> is a
156 <a href="http://gcc.gnu.org/wiki/plugins">gcc plugin</a> that replaces GCC's
157 optimizers and code generators with LLVM's. It works with gcc-4.5 or gcc-4.6,
158 targets the x86-32 and x86-64 processor families, and has been successfully
159 used on the Darwin, FreeBSD, KFreeBSD, Linux and OpenBSD platforms. It fully
160 supports Ada, C, C++ and Fortran. It has partial support for Go, Java, Obj-C
163 <p>The 3.0 release has the following notable changes:</p>
166 <li>GCC version 4.6 is now fully supported.</li>
168 <li>Patching and building GCC is no longer required: the plugin should work
169 with your system GCC (version 4.5 or 4.6; on Debian/Ubuntu systems the
170 gcc-4.5-plugin-dev or gcc-4.6-plugin-dev package is also needed).</li>
172 <li>The <tt>-fplugin-arg-dragonegg-enable-gcc-optzns</tt> option, which runs
173 GCC's optimizers as well as LLVM's, now works much better. This is the
174 option to use if you want ultimate performance! It is still experimental
175 though: it may cause the plugin to crash.</li>
177 <li>The type and constant conversion logic has been almost entirely rewritten,
178 fixing a multitude of obscure bugs.</li>
184 <!--=========================================================================-->
186 <a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
191 <p>The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
192 is a simple library that provides an implementation of the low-level
193 target-specific hooks required by code generation and other runtime
194 components. For example, when compiling for a 32-bit target, converting a
195 double to a 64-bit unsigned integer is compiled into a runtime call to the
196 "__fixunsdfdi" function. The compiler-rt library provides highly optimized
197 implementations of this and other low-level routines (some are 3x faster than
198 the equivalent libgcc routines).</p>
200 <p>In the LLVM 3.0 timeframe, the target specific ARM code has converted to
201 "unified" assembly syntax, and several new functions have been added to the
206 <!--=========================================================================-->
208 <a name="lldb">LLDB: Low Level Debugger</a>
213 <p>LLDB is a ground-up implementation of a command line debugger, as well as a
214 debugger API that can be used from other applications. LLDB makes use of the
215 Clang parser to provide high-fidelity expression parsing (particularly for
216 C++) and uses the LLVM JIT for target support.</p>
218 <p>LLDB has advanced by leaps and bounds in the 3.0 timeframe. It is
219 dramatically more stable and useful, and includes both a
220 new <a href="http://lldb.llvm.org/tutorial.html">tutorial</a> and
221 a <a href="http://lldb.llvm.org/lldb-gdb.html">side-by-side comparison with
226 <!--=========================================================================-->
228 <a name="libc++">libc++: C++ Standard Library</a>
233 <p>Like compiler_rt, libc++ is now <a href="DeveloperPolicy.html#license">dual
234 licensed</a> under the MIT and UIUC license, allowing it to be used more
237 <p>Libc++ has been ported to FreeBSD and imported into the base system. It is
238 planned to be the default STL implementation for FreeBSD 10.</p>
242 <!--=========================================================================-->
244 <a name="vmkit">VMKit</a>
249 <p>The <a href="http://vmkit.llvm.org/">VMKit project</a> is an
250 implementation of a Java Virtual Machine (Java VM or JVM) that uses LLVM for
251 static and just-in-time compilation.
253 <p>In the LLVM 3.0 time-frame, VMKit has had significant improvements on both
254 runtime and startup performance:</p>
257 <li>Precompilation: by compiling ahead of time a small subset of Java's core
258 library, the startup performance have been highly optimized to the point that
259 running a 'Hello World' program takes less than 30 milliseconds.</li>
261 <li>Customization: by customizing virtual methods for individual classes,
262 the VM can statically determine the target of a virtual call, and decide to
265 <li>Inlining: the VM does more inlining than it did before, by allowing more
266 bytecode instructions to be inlined, and thanks to customization. It also
267 inlines GC barriers, and object allocations.</li>
269 <li>New exception model: the generated code for a method that does not do
270 any try/catch is not penalized anymore by the eventuality of calling a
271 method that throws an exception. Instead, the method that throws the
272 exception jumps directly to the method that could catch it.</li>
278 <!--=========================================================================-->
280 <a name="LLBrowse">LLBrowse: IR Browser</a>
285 <p><a href="http://llvm.org/svn/llvm-project/llbrowse/trunk/doc/LLBrowse.html">
286 LLBrowse</a> is an interactive viewer for LLVM modules. It can load any LLVM
287 module and displays its contents as an expandable tree view, facilitating an
288 easy way to inspect types, functions, global variables, or metadata nodes. It
289 is fully cross-platform, being based on the popular wxWidgets GUI
295 <!--=========================================================================-->
298 <a name="klee">KLEE: A Symbolic Execution Virtual Machine</a>
303 <a href="http://klee.llvm.org/">KLEE</a> is a symbolic execution framework for
304 programs in LLVM bitcode form. KLEE tries to symbolically evaluate "all" paths
305 through the application and records state transitions that lead to fault
306 states. This allows it to construct testcases that lead to faults and can even
307 be used to verify some algorithms.
315 <!-- *********************************************************************** -->
317 <a name="externalproj">External Open Source Projects Using LLVM 3.0</a>
319 <!-- *********************************************************************** -->
323 <p>An exciting aspect of LLVM is that it is used as an enabling technology for
324 a lot of other language and tools projects. This section lists some of the
325 projects that have already been updated to work with LLVM 3.0.</p>
327 <!--=========================================================================-->
328 <h3>AddressSanitizer</h3>
332 <p><a href="http://code.google.com/p/address-sanitizer/">AddressSanitizer</a>
333 uses compiler instrumentation and a specialized malloc library to find C/C++
334 bugs such as use-after-free and out-of-bound accesses to heap, stack, and
335 globals. The key feature of the tool is speed: the average slowdown
336 introduced by AddressSanitizer is less than 2x.</p>
340 <!--=========================================================================-->
345 <p><a href="http://www.clamav.net">Clam AntiVirus</a> is an open source (GPL)
346 anti-virus toolkit for UNIX, designed especially for e-mail scanning on mail
349 <p>Since version 0.96 it
350 has <a href="http://vrt-sourcefire.blogspot.com/2010/09/introduction-to-clamavs-low-level.html">bytecode
351 signatures</a> that allow writing detections for complex malware.
352 It uses LLVM's JIT to speed up the execution of bytecode on X86, X86-64,
353 PPC32/64, falling back to its own interpreter otherwise. The git version was
354 updated to work with LLVM 3.0.</p>
358 <!--=========================================================================-->
359 <h3>clang_complete for VIM</h3>
363 <p><a href="https://github.com/Rip-Rip/clang_complete">clang_complete</a> is a
364 VIM plugin, that provides accurate C/C++ autocompletion using the clang front
365 end. The development version of clang complete, can directly use libclang
366 which can maintain a cache to speed up auto completion.</p>
370 <!--=========================================================================-->
375 <p><a href="https://bitbucket.org/dwilliamson/clreflect">clReflect</a> is a C++
376 parser that uses clang/LLVM to derive a light-weight reflection database
377 suitable for use in game development. It comes with a very simple runtime
378 library for loading and querying the database, requiring no external
379 dependencies (including CRT), and an additional utility library for object
380 management and serialisation.</p>
384 <!--=========================================================================-->
385 <h3>Cling C++ Interpreter</h3>
389 <p><a href="http://cern.ch/cling">Cling</a> is an interactive compiler interface
390 (aka C++ interpreter). It supports C++ and C, and uses LLVM's JIT and the
391 Clang parser. It has a prompt interface, runs source files, calls into shared
392 libraries, prints the value of expressions, even does runtime lookup of
393 identifiers (dynamic scopes). And it just behaves like one would expect from
398 <!--=========================================================================-->
399 <h3>Crack Programming Language</h3>
403 <p><a href="http://code.google.com/p/crack-language/">Crack</a> aims to provide
404 the ease of development of a scripting language with the performance of a
405 compiled language. The language derives concepts from C++, Java and Python,
406 incorporating object-oriented programming, operator overloading and strong
411 <!--=========================================================================-->
416 <p><a href="http://eerolanguage.org/">Eero</a> is a fully
417 header-and-binary-compatible dialect of Objective-C 2.0, implemented with a
418 patched version of the Clang/LLVM compiler. It features a streamlined syntax,
419 Python-like indentation, and new operators, for improved readability and
420 reduced code clutter. It also has new features such as limited forms of
421 operator overloading and namespaces, and strict (type-and-operator-safe)
422 enumerations. It is inspired by languages such as Smalltalk, Python, and
427 <!--=========================================================================-->
428 <h3>FAUST Real-Time Audio Signal Processing Language</h3>
432 <p><a href="http://faust.grame.fr/">FAUST</a> is a compiled language for
433 real-time audio signal processing. The name FAUST stands for Functional
434 AUdio STream. Its programming model combines two approaches: functional
435 programming and block diagram composition. In addition with the C, C++, Java
436 output formats, the Faust compiler can now generate LLVM bitcode, and works
442 <!--=========================================================================-->
443 <h3>Glasgow Haskell Compiler (GHC)</h3>
447 <p>GHC is an open source, state-of-the-art programming suite for Haskell, a
448 standard lazy functional programming language. It includes an optimizing
449 static compiler generating good code for a variety of platforms, together
450 with an interactive system for convenient, quick development.</p>
452 <p>GHC 7.0 and onwards include an LLVM code generator, supporting LLVM 2.8 and
453 later. Since LLVM 2.9, GHC now includes experimental support for the ARM
454 platform with LLVM 3.0.</p>
458 <!--=========================================================================-->
463 <p><a href="http://botwars.tk/gwscript/">gwXscript</a> is an object oriented,
464 aspect oriented programming language which can create both executables (ELF,
465 EXE) and shared libraries (DLL, SO, DYNLIB). The compiler is implemented in
466 its own language and translates scripts into LLVM-IR which can be optimized
467 and translated into native code by the LLVM framework. Source code in
468 gwScript contains definitions that expand the namespaces. So you can build
469 your project and simply 'plug out' features by removing a file. The remaining
470 project does not leave scars since you directly separate concerns by the
471 'template' feature of gwX. It is also possible to add new features to a
472 project by just adding files and without editing the original project. This
473 language is used for example to create games or content management systems
474 that should be extendable.</p>
476 <p>gwXscript is strongly typed and offers comfort with its native types string,
477 hash and array. You can easily write new libraries in gwXscript or native
478 code. gwXscript is type safe and users should not be able to crash your
479 program or execute malicious code except code that is eating CPU time.</p>
483 <!--=========================================================================-->
484 <h3>include-what-you-use</h3>
488 <p><a href="http://code.google.com/p/include-what-you-use">include-what-you-use</a>
489 is a tool to ensure that a file directly <code>#include</code>s
490 all <code>.h</code> files that provide a symbol that the file uses. It also
491 removes superfluous <code>#include</code>s from source files.</p>
495 <!--=========================================================================-->
496 <h3>ispc: The Intel SPMD Program Compiler</h3>
500 <p><a href="http://ispc.github.com">ispc</a> is a compiler for "single program,
501 multiple data" (SPMD) programs. It compiles a C-based SPMD programming
502 language to run on the SIMD units of CPUs; it often delivers 5-6x speedups on
503 a single core of a CPU with an 8-wide SIMD unit compared to serial code,
504 while still providing a clean and easy-to-understand programming model. For
505 an introduction to the language and its performance,
506 see <a href="http://ispc.github.com/example.html">the walkthrough</a> of a short
507 example program. ispc is licensed under the BSD license.</p>
511 <!--=========================================================================-->
512 <h3>The Julia Programming Language</h3>
516 <p><a href="http://github.com/JuliaLang/julia">Julia</a> is a high-level,
517 high-performance dynamic language for technical
518 computing. It provides a sophisticated compiler, distributed parallel
519 execution, numerical accuracy, and an extensive mathematical function
520 library. The compiler uses type inference to generate fast code
521 without any type declarations, and uses LLVM's optimization passes and
522 JIT compiler. The language is designed around multiple dispatch,
523 giving programs a large degree of flexibility. It is ready for use on many
524 kinds of problems.</p>
527 <!--=========================================================================-->
528 <h3>LanguageKit and Pragmatic Smalltalk</h3>
532 <p><a href="http://etoileos.com/etoile/features/languagekit/">LanguageKit</a> is
533 a framework for implementing dynamic languages sharing an object model with
534 Objective-C. It provides static and JIT compilation using LLVM along with
535 its own interpreter. Pragmatic Smalltalk is a dialect of Smalltalk, built on
536 top of LanguageKit, that interfaces directly with Objective-C, sharing the
537 same object representation and message sending behaviour. These projects are
538 developed as part of the Étoilé desktop environment.</p>
542 <!--=========================================================================-->
547 <p><a href="http://lua-av.mat.ucsb.edu/blog/">LuaAV</a> is a real-time
548 audiovisual scripting environment based around the Lua language and a
549 collection of libraries for sound, graphics, and other media protocols. LuaAV
550 uses LLVM and Clang to JIT compile efficient user-defined audio synthesis
551 routines specified in a declarative syntax.</p>
555 <!--=========================================================================-->
560 <p>An open source, cross-platform implementation of C# and the CLR that is
561 binary compatible with Microsoft.NET. Has an optional, dynamically-loaded
562 LLVM code generation backend in Mini, the JIT compiler.</p>
564 <p>Note that we use a Git mirror of LLVM <a
565 href="https://github.com/mono/llvm">with some patches</a>.</p>
569 <!--=========================================================================-->
574 <p><a href="http://polly.grosser.es">Polly</a> is an advanced data-locality
575 optimizer and automatic parallelizer. It uses an advanced, mathematical
576 model to calculate detailed data dependency information which it uses to
577 optimize the loop structure of a program. Polly can speed up sequential code
578 by improving memory locality and consequently the cache use. Furthermore,
579 Polly is able to expose different kind of parallelism which it exploits by
580 introducing (basic) OpenMP and SIMD code. A mid-term goal of Polly is to
581 automatically create optimized GPU code.</p>
585 <!--=========================================================================-->
586 <h3>Portable OpenCL (pocl)</h3>
590 <p>Portable OpenCL is an open source implementation of the OpenCL standard which
591 can be easily adapted for new targets. One of the goals of the project is
592 improving performance portability of OpenCL programs, avoiding the need for
593 target-dependent manual optimizations. A "native" target is included, which
594 allows running OpenCL kernels on the host (CPU).</p>
598 <!--=========================================================================-->
602 <p><a href="http://pure-lang.googlecode.com/">Pure</a> is an
603 algebraic/functional programming language based on term rewriting. Programs
604 are collections of equations which are used to evaluate expressions in a
605 symbolic fashion. The interpreter uses LLVM as a backend to JIT-compile Pure
606 programs to fast native code. Pure offers dynamic typing, eager and lazy
607 evaluation, lexical closures, a hygienic macro system (also based on term
608 rewriting), built-in list and matrix support (including list and matrix
609 comprehensions) and an easy-to-use interface to C and other programming
610 languages (including the ability to load LLVM bitcode modules, and inline C,
611 C++, Fortran and Faust code in Pure programs if the corresponding LLVM-enabled
612 compilers are installed).</p>
614 <p>Pure version 0.48 has been tested and is known to work with LLVM 3.0
615 (and continues to work with older LLVM releases >= 2.5).</p>
619 <!--=========================================================================-->
620 <h3>Renderscript</h3>
624 <p><a href="http://developer.android.com/guide/topics/renderscript/index.html">Renderscript</a>
625 is Android's advanced 3D graphics rendering and compute API. It provides a
626 portable C99-based language with extensions to facilitate common use cases
627 for enhancing graphics and thread level parallelism. The Renderscript
628 compiler frontend is based on Clang/LLVM. It emits a portable bitcode format
629 for the actual compiled script code, as well as reflects a Java interface for
630 developers to control the execution of the compiled bitcode. Executable
631 machine code is then generated from this bitcode by an LLVM backend on the
632 device. Renderscript is thus able to provide a mechanism by which Android
633 developers can improve performance of their applications while retaining
638 <!--=========================================================================-->
643 <p><a href="http://safecode.cs.illinois.edu">SAFECode</a> is a memory safe C/C++
644 compiler built using LLVM. It takes standard, unannotated C/C++ code,
645 analyzes the code to ensure that memory accesses and array indexing
646 operations are safe, and instruments the code with run-time checks when
647 safety cannot be proven statically. SAFECode can be used as a debugging aid
648 (like Valgrind) to find and repair memory safety bugs. It can also be used
649 to protect code from security attacks at run-time.</p>
653 <!--=========================================================================-->
654 <h3>The Stupid D Compiler (SDC)</h3>
658 <p><a href="https://github.com/bhelyer/SDC">The Stupid D Compiler</a> is a
659 project seeking to write a self-hosting compiler for the D programming
660 language without using the frontend of the reference compiler (DMD).</p>
664 <!--=========================================================================-->
665 <h3>TTA-based Co-design Environment (TCE)</h3>
669 <p>TCE is a toolset for designing application-specific processors (ASP) based on
670 the Transport triggered architecture (TTA). The toolset provides a complete
671 co-design flow from C/C++ programs down to synthesizable VHDL and parallel
672 program binaries. Processor customization points include the register files,
673 function units, supported operations, and the interconnection network.</p>
675 <p>TCE uses Clang and LLVM for C/C++ language support, target independent
676 optimizations and also for parts of code generation. It generates new
677 LLVM-based code generators "on the fly" for the designed TTA processors and
678 loads them in to the compiler backend as runtime libraries to avoid
679 per-target recompilation of larger parts of the compiler chain.</p>
683 <!--=========================================================================-->
684 <h3>Tart Programming Language</h3>
688 <p><a href="http://code.google.com/p/tart/">Tart</a> is a general-purpose,
689 strongly typed programming language designed for application
690 developers. Strongly inspired by Python and C#, Tart focuses on practical
691 solutions for the professional software developer, while avoiding the clutter
692 and boilerplate of legacy languages like Java and C++. Although Tart is still
693 in development, the current implementation supports many features expected of
694 a modern programming language, such as garbage collection, powerful
695 bidirectional type inference, a greatly simplified syntax for template
696 metaprogramming, closures and function literals, reflection, operator
697 overloading, explicit mutability and immutability, and much more. Tart is
698 flexible enough to accommodate a broad range of programming styles and
699 philosophies, while maintaining a strong commitment to simplicity, minimalism
700 and elegance in design.</p>
704 <!--=========================================================================-->
705 <h3>ThreadSanitizer</h3>
709 <p><a href="http://code.google.com/p/data-race-test/">ThreadSanitizer</a> is a
710 data race detector for (mostly) C and C++ code, available for Linux, Mac OS
711 and Windows. On different systems, we use binary instrumentation frameworks
712 (Valgrind and Pin) as frontends that generate the program events for the race
713 detection algorithm. On Linux, there's an option of using LLVM-based
714 compile-time instrumentation.</p>
720 <!-- *********************************************************************** -->
722 <a name="whatsnew">What's New in LLVM 3.0?</a>
724 <!-- *********************************************************************** -->
728 <p>This release includes a huge number of bug fixes, performance tweaks and
729 minor improvements. Some of the major improvements and new features are
730 listed in this section.</p>
732 <!--=========================================================================-->
734 <a name="majorfeatures">Major New Features</a>
739 <!-- Features that need text if they're finished for 3.1:
743 loop dependence analysis
744 CorrelatedValuePropagation
745 lib/Transforms/IPO/MergeFunctions.cpp => consider for 3.1.
746 Integrated assembler on by default for arm/thumb?
751 Analysis/RegionInfo.h + Dom Frontiers
752 SparseBitVector: used in LiveVar.
753 llvm/lib/Archive - replace with lib object?
756 <p>LLVM 3.0 includes several major changes and big features:</p>
759 <li>llvm-gcc is no longer supported, and not included in the release. We
760 recommend switching to <a
761 href="http://clang.llvm.org/">Clang</a> or <a
762 href="http://dragonegg.llvm.org/">DragonEgg</a>.</li>
764 <li>The linear scan register allocator has been replaced with a new "greedy"
765 register allocator, enabling live range splitting and many other
766 optimizations that lead to better code quality. Please see its <a
767 href="http://blog.llvm.org/2011/09/greedy-register-allocation-in-llvm-30.html">blog post</a> or its talk at the <a
768 href="http://llvm.org/devmtg/2011-11/">Developer Meeting</a>
769 for more information.</li>
770 <li>LLVM IR now includes full support for <a href="Atomics.html">atomics
771 memory operations</a> intended to support the C++'11 and C'1x memory models.
772 This includes <a href="LangRef.html#memoryops">atomic load and store,
773 compare and exchange, and read/modify/write instructions</a> as well as a
774 full set of <a href="LangRef.html#ordering">memory ordering constraints</a>.
775 Please see the <a href="Atomics.html">Atomics Guide</a> for more
778 <li>The LLVM IR exception handling representation has been redesigned and
779 reimplemented, making it more elegant, fixing a huge number of bugs, and
780 enabling inlining and other optimizations. Please see its blog post (XXX
781 not yet) and the <a href="ExceptionHandling.html">Exception Handling
782 documentation</a> for more information.</li>
783 <li>The LLVM IR Type system has been redesigned and reimplemented, making it
784 faster and solving some long-standing problems.
786 href="http://blog.llvm.org/2011/11/llvm-30-type-system-rewrite.html">blog
787 post</a> for more information.</li>
789 <li>The MIPS backend has made major leaps in this release, going from an
790 experimental target to being virtually production quality and supporting a
791 wide variety of MIPS subtargets. See the <a href="#MIPS">MIPS section</a>
792 below for more information.</li>
794 <li>The optimizer and code generator now supports gprof and gcov-style coverage
795 and profiling information, and includes a new llvm-cov tool (but also works
796 with gcov). Clang exposes coverage and profiling through GCC-compatible
797 command line options.</li>
803 <!--=========================================================================-->
805 <a name="coreimprovements">LLVM IR and Core Improvements</a>
810 <p>LLVM IR has several new features for better support of new targets and that
811 expose new optimization opportunities:</p>
814 <li><a href="Atomics.html">Atomic memory accesses and memory ordering</a> are
815 now directly expressible in the IR.</li>
816 <li>A new <a href="LangRef.html#int_fma">llvm.fma intrinsic</a> directly
817 represents floating point multiply accumulate operations without an
818 intermediate rounding stage.</li>
819 <li>A new llvm.expect intrinsic (XXX not documented in langref) allows a
820 frontend to express expected control flow (and the __builtin_expect builtin
822 <li>The <a href="LangRef.html#int_prefetch">llvm.prefetch intrinsic</a> now
823 takes a 4th argument that specifies whether the prefetch happens from the
824 icache or dcache.</li>
825 <li>The new <a href="LangRef.html#uwtable">uwtable function attribute</a>
826 allows a frontend to control emission of unwind tables.</li>
827 <li>The new <a href="LangRef.html#fnattrs">nonlazybind function
828 attribute</a> allow optimization of Global Offset Table (GOT) accesses.</li>
829 <li>The new <a href="LangRef.html#returns_twice">returns_twice attribute</a>
830 allows better modeling of functions like setjmp.</li>
831 <li>The <a href="LangRef.html#datalayout">target datalayout</a> string can now
832 encode the natural alignment of the target's stack for better optimization.
837 <!--=========================================================================-->
839 <a name="optimizer">Optimizer Improvements</a>
844 <p>In addition to many minor performance tweaks and bug fixes, this
845 release includes a few major enhancements and additions to the
849 <li>The pass manager now has an extension API that allows front-ends and plugins
850 to insert their own optimizations in the well-known places in the standard
851 pass optimization pipeline.</li>
853 <li>Information about <a href="BranchWeightMetadata.html">branch probability</a>
854 and basic block frequency is now available within LLVM, based on a
855 combination of static branch prediction heuristics and
856 <code>__builtin_expect</code> calls. That information is currently used for
857 register spill placement and if-conversion, with additional optimizations
858 planned for future releases. The same framework is intended for eventual
859 use with profile-guided optimization.</li>
861 <li>The "-indvars" induction variable simplification pass only modifies
862 induction variables when profitable. Sign and zero extension
863 elimination, linear function test replacement, loop unrolling, and
864 other simplifications that require induction variable analysis have
865 been generalized so they no longer require loops to be rewritten into
866 canonical form prior to optimization. This new design
867 preserves more IR level information, avoids undoing earlier loop
868 optimizations (particularly hand-optimized loops), and no longer
869 requires the code generator to reconstruct loops into an optimal form -
870 an intractable problem.</li>
872 <li>LLVM now includes a pass to optimize retain/release calls for the
873 <a href="http://clang.llvm.org/docs/AutomaticReferenceCounting.html">Automatic
874 Reference Counting</a> (ARC) Objective-C language feature (in
875 lib/Transforms/Scalar/ObjCARC.cpp). It is a decent example of implementing
876 a source-language-specific optimization in LLVM.</li>
882 <!--=========================================================================-->
884 <a name="mc">MC Level Improvements</a>
889 <p>The LLVM Machine Code (aka MC) subsystem was created to solve a number of
890 problems in the realm of assembly, disassembly, object file format handling,
891 and a number of other related areas that CPU instruction-set level tools work
892 in. For more information, please see
893 the <a href="http://blog.llvm.org/2010/04/intro-to-llvm-mc-project.html">Intro
894 to the LLVM MC Project Blog Post</a>.</p>
897 <li>The MC layer has undergone significant refactoring to eliminate layering
898 violations that caused it to pull in the LLVM compiler backend code.</li>
899 <li>The ELF object file writers are much more full featured.</li>
900 <li>The integrated assembler now supports #line directives.</li>
901 <li>An early implementation of a JIT built on top of the MC framework (known
902 as MC-JIT) has been implemented and will eventually replace the old JIT.
903 It emits object files direct to memory and uses a runtime dynamic linker to
904 resolve references and drive lazy compilation. The MC-JIT enables much
905 greater code reuse between the JIT and the static compiler and provides
906 better integration with the platform ABI as a result.
908 <li>The assembly printer now makes uses of assemblers instruction aliases
909 (InstAliases) to print simplified mneumonics when possible.</li>
910 <li>TableGen can now autogenerate MC expansion logic for pseudo
911 instructions that expand to multiple MC instructions (through the
912 PseudoInstExpansion class).</li>
913 <li>A new llvm-objdump and llvm-dwarfdump tools provide a start of a drop-in
914 replacement for the corresponding tools that use LLVM libraries. As part of
915 this, LLVM has the beginnings of a dwarf parsing library.</li>
916 <li>XXX: object file parsing stuff and llvm-size (mspencer). Status?</li>
921 <!--=========================================================================-->
923 <a name="codegen">Target Independent Code Generator Improvements</a>
928 <p>We have put a significant amount of work into the code generator
929 infrastructure, which allows us to implement more aggressive algorithms and
930 make it run faster:</p>
933 <li>XXX: Segmented stacks.</li>
934 <li>LLVM generates substantially better code for indirect gotos due to a new
935 tail duplication pass, which can be a substantial performance win for
936 interpreter loops that use them.</li>
937 <li>Exception handling and debug information is now emitted with CFI directives,
938 yielding <a href="http://blog.mozilla.com/respindola/2011/05/12/cfi-directives/">much smaller executables</a> for some C++ applications.
941 <li>The code generator now supports vector "select" operations on vector
942 comparisons, turning them into various optimized code sequences (e.g.
943 using the SSE4/AVX "blend" instructions).</li>
944 <li>The SSE execution domain fix pass and the ARM NEON move fix pass have been
945 merged to a target independent execution dependency fix pass. Targets can
946 override the <code>getExecutionDomain</code> and
947 <code>setExecutionDomain</code> hooks to use it.</li>
951 <!--=========================================================================-->
953 <a name="x86">X86-32 and X86-64 Target Improvements</a>
958 <p>New features and major changes in the X86 target include:</p>
961 <li>The X86 backend, assembler and disassembler now have full support for AVX 1.
962 To enable it pass <code>-mavx</code> to the compiler. AVX2 implementation is
963 underway on mainline.</li>
964 <li>The integrated assembler and disassembler now support a broad range of new
965 instructions including Atom, Ivy Bridge, <a
966 href="http://en.wikipedia.org/wiki/SSE4a">SSE4a/BMI</a> instructions, <a
967 href="http://en.wikipedia.org/wiki/RdRand">rdrand</a> and many others.</li>
968 <li>The X86 backend now fully supports the <a href="http://llvm.org/PR879">X87
969 floating point stack inline assembly constraints</a>.</li>
970 <li>The integrated assembler now supports the <tt>.code32</tt> and
971 <tt>.code64</tt> directives to switch between 32-bit and 64-bit
973 <li>The X86 backend now synthesizes horizontal add/sub instructions from generic
974 vector code when the appropriate instructions are enabled.</li>
975 <li>The X86-64 backend generates smaller and faster code at -O0 due to
976 improvements in fast instruction selection.</li>
977 <li><a href="http://code.google.com/p/nativeclient/">Native Client</a>
978 subtarget support has been added.</li>
980 <li>The CRC32 intrinsics have been renamed. The intrinsics were previously
981 <code>@llvm.x86.sse42.crc32.[8|16|32]</code>
982 and <code>@llvm.x86.sse42.crc64.[8|64]</code>. They have been renamed to
983 <code>@llvm.x86.sse42.crc32.32.[8|16|32]</code> and
984 <code>@llvm.x86.sse42.crc32.64.[8|64]</code>.</li>
989 <!--=========================================================================-->
991 <a name="ARM">ARM Target Improvements</a>
996 <p>New features of the ARM target include:</p>
999 <li>The ARM backend generates much faster code for Cortex-A9 chips.</li>
1000 <li>The ARM backend has improved support for Cortex-M series processors.</li>
1001 <li>The ARM inline assembly constraints have been implemented and are now fully
1003 <li>NEON code produced by Clang often runs much faster due to improvements in
1004 the Scalar Replacement of Aggregates pass.</li>
1005 <li>The old ARM disassembler is replaced with a new one based on autogenerated
1006 encoding information from ARM .td files.</li>
1007 <li>The integrated assembler has made major leaps forward, but is still beta quality in LLVM 3.0.</li>
1012 <!--=========================================================================-->
1014 <a name="MIPS">MIPS Target Improvements</a>
1019 <p>This release has seen major new work on just about every aspect of the MIPS
1020 backend. Some of the major new features include:</p>
1023 <li>Most MIPS32r1 and r2 instructions are now supported.</li>
1024 <li>LE/BE MIPS32r1/r2 has been tested extensively.</li>
1025 <li>O32 ABI has been fully tested.</li>
1026 <li>MIPS backend has migrated to using the MC infrastructure for assembly printing. Initial support for direct object code emission has been implemented too.</li>
1027 <li>Delay slot filler has been updated. Now it tries to fill delay slots with useful instructions instead of always filling them with NOPs.</li>
1028 <li>Support for old-style JIT is complete.</li>
1029 <li>Support for old architectures (MIPS1 and MIPS2) has been removed.</li>
1030 <li>Initial support for MIPS64 has been added.</li>
1034 <!--=========================================================================-->
1036 <a name="PTX">PTX Target Improvements</a>
1042 The PTX back-end is still experimental, but is fairly usable for compute kernels
1043 in LLVM 3.0. Most scalar arithmetic is implemented, as well as intrinsics to
1044 access the special PTX registers and sync instructions. The major missing
1045 pieces are texture/sampler support and some vector operations.</p>
1047 <p>That said, the backend is already being used for domain-specific languages
1048 and works well with the <a href="http://www.pcc.me.uk/~peter/libclc/">libclc
1049 library</a> to supply OpenCL built-ins. With it, you can use Clang to compile
1050 OpenCL code into PTX and execute it by loading the resulting PTX as a binary
1051 blob using the nVidia OpenCL library. It has been tested with several OpenCL
1052 programs, including some from the nVidia GPU Computing SDK, and the performance
1053 is on par with the nVidia compiler.</p>
1057 <!--=========================================================================-->
1059 <a name="OtherTS">Other Target Specific Improvements</a>
1065 <li>Many PowerPC improvements have been implemented for ELF targets, including
1066 support for varargs and initial support for direct .o file emission.</li>
1068 <li>MicroBlaze scheduling itineraries were added that model the
1069 3-stage and the 5-stage pipeline architectures. The 3-stage
1070 pipeline model can be selected with <code>-mcpu=mblaze3</code>
1071 and the 5-stage pipeline model can be selected with
1072 <code>-mcpu=mblaze5</code>.</li>
1078 <!--=========================================================================-->
1080 <a name="changes">Major Changes and Removed Features</a>
1085 <p>If you're already an LLVM user or developer with out-of-tree changes based on
1086 LLVM 2.9, this section lists some "gotchas" that you may run into upgrading
1087 from the previous release.</p>
1090 <li>LLVM 3.0 removes support for reading LLVM 2.8 and earlier files, and LLVM
1091 3.1 will eliminate support for reading LLVM 2.9 files. Going forward, we
1092 aim for all future versions of LLVM to read bitcode files and .ll files
1093 produced by LLVM 3.0.</li>
1094 <li>Tablegen has been split into a library, allowing the clang tblgen pieces
1095 now live in the clang tree. The llvm version has been renamed to
1096 llvm-tblgen instead of tblgen.</li>
1097 <li>The <code>LLVMC</code> meta compiler driver was removed.</li>
1098 <li>The unused PostOrder Dominator Frontiers and LowerSetJmp passes were removed.</li>
1101 <li>The old <code>TailDup</code> pass was not used in the standard pipeline
1102 and was unable to update ssa form, so it has been removed.
1103 <li>The syntax of volatile loads and stores in IR has been changed to
1104 "<code>load volatile</code>"/"<code>store volatile</code>". The old
1105 syntax ("<code>volatile load</code>"/"<code>volatile store</code>")
1106 is still accepted, but is now considered deprecated and will be removed in
1108 <li>llvm-gcc's frontend tests have been removed from llvm/test/Frontend*, sunk
1109 into the clang and dragonegg testsuites.</li>
1110 <li>The old atomic intrinsics (<code>llvm.memory.barrier</code> and
1111 <code>llvm.atomic.*</code>) are now gone. Please use the new atomic
1112 instructions, described in the <a href="Atomics.html">atomics guide</a>.
1113 <li>LLVM's configure script doesn't depend on llvm-gcc anymore, eliminating a
1114 strange circular dependence between projects.</li>
1117 <h4>Windows (32-bit)</h4>
1121 <li>On Win32(MinGW32 and MSVC), Windows 2000 will not be supported.
1122 Windows XP or higher is required.</li>
1129 <!--=========================================================================-->
1131 <a name="api_changes">Internal API Changes</a>
1136 <p>In addition, many APIs have changed in this release. Some of the major
1137 LLVM API changes are:</p>
1140 <li>The biggest and most pervasive change is that llvm::Types are no longer
1141 returned or accepted as 'const' values. Instead, just pass around
1142 non-const Types.</li>
1144 <li><code>PHINode::reserveOperandSpace</code> has been removed. Instead, you
1145 must specify how many operands to reserve space for when you create the
1146 PHINode, by passing an extra argument
1147 into <code>PHINode::Create</code>.</li>
1149 <li>PHINodes no longer store their incoming BasicBlocks as operands. Instead,
1150 the list of incoming BasicBlocks is stored separately, and can be accessed
1151 with new functions <code>PHINode::block_begin</code>
1152 and <code>PHINode::block_end</code>.</li>
1154 <li>Various functions now take an <code>ArrayRef</code> instead of either a
1155 pair of pointers (or iterators) to the beginning and end of a range, or a
1156 pointer and a length. Others now return an <code>ArrayRef</code> instead
1157 of a reference to a <code>SmallVector</code>
1158 or <code>std::vector</code>. These include:
1160 <!-- Please keep this list sorted. -->
1161 <li><code>CallInst::Create</code></li>
1162 <li><code>ComputeLinearIndex</code> (in <code>llvm/CodeGen/Analysis.h</code>)</li>
1163 <li><code>ConstantArray::get</code></li>
1164 <li><code>ConstantExpr::getExtractElement</code></li>
1165 <li><code>ConstantExpr::getGetElementPtr</code></li>
1166 <li><code>ConstantExpr::getInBoundsGetElementPtr</code></li>
1167 <li><code>ConstantExpr::getIndices</code></li>
1168 <li><code>ConstantExpr::getInsertElement</code></li>
1169 <li><code>ConstantExpr::getWithOperands</code></li>
1170 <li><code>ConstantFoldCall</code> (in <code>llvm/Analysis/ConstantFolding.h</code>)</li>
1171 <li><code>ConstantFoldInstOperands</code> (in <code>llvm/Analysis/ConstantFolding.h</code>)</li>
1172 <li><code>ConstantVector::get</code></li>
1173 <li><code>DIBuilder::createComplexVariable</code></li>
1174 <li><code>DIBuilder::getOrCreateArray</code></li>
1175 <li><code>ExtractValueInst::Create</code></li>
1176 <li><code>ExtractValueInst::getIndexedType</code></li>
1177 <li><code>ExtractValueInst::getIndices</code></li>
1178 <li><code>FindInsertedValue</code> (in <code>llvm/Analysis/ValueTracking.h</code>)</li>
1179 <li><code>gep_type_begin</code> (in <code>llvm/Support/GetElementPtrTypeIterator.h</code>)</li>
1180 <li><code>gep_type_end</code> (in <code>llvm/Support/GetElementPtrTypeIterator.h</code>)</li>
1181 <li><code>GetElementPtrInst::Create</code></li>
1182 <li><code>GetElementPtrInst::CreateInBounds</code></li>
1183 <li><code>GetElementPtrInst::getIndexedType</code></li>
1184 <li><code>InsertValueInst::Create</code></li>
1185 <li><code>InsertValueInst::getIndices</code></li>
1186 <li><code>InvokeInst::Create</code></li>
1187 <li><code>IRBuilder::CreateCall</code></li>
1188 <li><code>IRBuilder::CreateExtractValue</code></li>
1189 <li><code>IRBuilder::CreateGEP</code></li>
1190 <li><code>IRBuilder::CreateInBoundsGEP</code></li>
1191 <li><code>IRBuilder::CreateInsertValue</code></li>
1192 <li><code>IRBuilder::CreateInvoke</code></li>
1193 <li><code>MDNode::get</code></li>
1194 <li><code>MDNode::getIfExists</code></li>
1195 <li><code>MDNode::getTemporary</code></li>
1196 <li><code>MDNode::getWhenValsUnresolved</code></li>
1197 <li><code>SimplifyGEPInst</code> (in <code>llvm/Analysis/InstructionSimplify.h</code>)</li>
1198 <li><code>TargetData::getIndexedOffset</code></li>
1201 <li>All forms of <code>StringMap::getOrCreateValue</code> have been remove
1202 except for the one which takes a <code>StringRef</code>.</li>
1204 <li>The <code>LLVMBuildUnwind</code> function from the C API was removed. The
1205 LLVM <code>unwind</code> instruction has been deprecated for a long time
1206 and isn't used by the current front-ends. So this was removed during the
1207 exception handling rewrite.</li>
1209 <li>The <code>LLVMAddLowerSetJmpPass</code> function from the C API was
1210 removed because the <code>LowerSetJmp</code> pass was removed.</li>
1212 <li>The <code>DIBuilder</code> interface used by front ends to encode
1213 debugging information in the LLVM IR now expects clients to
1214 use <code>DIBuilder::finalize()</code> at the end of translation unit to
1215 complete debugging information encoding.</li>
1217 <li>The way the type system works has been
1218 rewritten: <code>PATypeHolder</code> and <code>OpaqueType</code> are gone,
1219 and all APIs deal with <code>Type*</code> instead of <code>const
1220 Type*</code>. If you need to create recursive structures, then create a
1221 named structure, and use <code>setBody()</code> when all its elements are
1222 built. Type merging and refining is gone too: named structures are not
1223 merged with other structures, even if their layout is identical. (of
1224 course anonymous structures are still uniqued by layout).</li>
1226 <li>TargetSelect.h moved to Support/ from Target/</li>
1228 <li>UpgradeIntrinsicCall no longer upgrades pre-2.9 intrinsic calls (for
1229 example <code>llvm.memset.i32</code>).</li>
1231 <li>It is mandatory to initialize all out-of-tree passes too and their dependencies now with
1232 <code>INITIALIZE_PASS{BEGIN,END,}</code>
1233 and <code>INITIALIZE_{PASS,AG}_DEPENDENCY</code>.</li>
1235 <li>The interface for MemDepResult in MemoryDependenceAnalysis has been
1236 enhanced with new return types Unknown and NonFuncLocal, in addition to
1237 the existing types Clobber, Def, and NonLocal.</li>
1244 <!-- *********************************************************************** -->
1246 <a name="knownproblems">Known Problems</a>
1248 <!-- *********************************************************************** -->
1252 <p>LLVM is generally a production quality compiler, and is used by a broad range
1253 of applications and shipping in many products. That said, not every
1254 subsystem is as mature as the aggregate, particularly the more obscure
1255 targets. If you run into a problem, please check the <a
1256 href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
1257 there isn't already one or ask on the <a
1258 href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev
1261 <p>Known problem areas include:</p>
1264 <li>The Alpha, Blackfin, CellSPU, MSP430, PTX, SystemZ and
1265 XCore backends are experimental, and the Alpha, Blackfin and SystemZ
1266 targets have already been removed from mainline.</li>
1268 <li>The integrated assembler, disassembler, and JIT is not supported by
1269 several targets. If an integrated assembler is not supported, then a
1270 system assembler is required. For more details, see the <a
1271 href="CodeGenerator.html#targetfeatures">Target Features Matrix</a>.
1274 <li>The C backend has numerous problems and is not being actively maintained.
1275 Depending on it for anything serious is not advised.</li>
1282 <!-- *********************************************************************** -->
1284 <a name="additionalinfo">Additional Information</a>
1286 <!-- *********************************************************************** -->
1290 <p>A wide variety of additional information is available on
1291 the <a href="http://llvm.org/">LLVM web page</a>, in particular in
1292 the <a href="http://llvm.org/docs/">documentation</a> section. The web page
1293 also contains versions of the API documentation which is up-to-date with the
1294 Subversion version of the source code. You can access versions of these
1295 documents specific to this release by going into the "<tt>llvm/doc/</tt>"
1296 directory in the LLVM tree.</p>
1298 <p>If you have any questions or comments about LLVM, please feel free to contact
1299 us via the <a href="http://llvm.org/docs/#maillist"> mailing lists</a>.</p>
1303 <!--=========================================================================-->
1305 <!-- EH details: to be moved to a blog post:
1310 <p>One of the biggest changes is that 3.0 has a new exception handling
1311 system. The old system used LLVM intrinsics to convey the exception handling
1312 information to the code generator. It worked in most cases, but not
1313 all. Inlining was especially difficult to get right. Also, the intrinsics
1314 could be moved away from the <code>invoke</code> instruction, making it hard
1315 to recover that information.</p>
1317 <p>The new EH system makes exception handling a first-class member of the IR. It
1318 adds two new instructions:</p>
1321 <li><a href="LangRef.html#i_landingpad"><code>landingpad</code></a> —
1322 this instruction defines a landing pad basic block. It contains all of the
1323 information that's needed by the code generator. It's also required to be
1324 the first non-PHI instruction in the landing pad. In addition, a landing
1325 pad may be jumped to only by the unwind edge of an <code>invoke</code>
1328 <li><a href="LangRef.html#i_resume"><code>resume</code></a> — this
1329 instruction causes the current exception to resume traveling up the
1330 stack. It replaces the <code>@llvm.eh.resume</code> intrinsic.</li>
1333 <p>Converting from the old EH API to the new EH API is rather simple, because a
1334 lot of complexity has been removed. The two intrinsics,
1335 <code>@llvm.eh.exception</code> and <code>@llvm.eh.selector</code> have been
1336 superseded by the <code>landingpad</code> instruction. Instead of generating
1337 a call to <code>@llvm.eh.exception</code> and <code>@llvm.eh.selector</code>:
1339 <div class="doc_code">
1341 Function *ExcIntr = Intrinsic::getDeclaration(TheModule,
1342 Intrinsic::eh_exception);
1343 Function *SlctrIntr = Intrinsic::getDeclaration(TheModule,
1344 Intrinsic::eh_selector);
1346 // The exception pointer.
1347 Value *ExnPtr = Builder.CreateCall(ExcIntr, "exc_ptr");
1349 std::vector<Value*> Args;
1350 Args.push_back(ExnPtr);
1351 Args.push_back(Builder.CreateBitCast(Personality,
1352 Type::getInt8PtrTy(Context)));
1354 <i>// Add selector clauses to Args.</i>
1356 // The selector call.
1357 Builder.CreateCall(SlctrIntr, Args, "exc_sel");
1361 <p>You should instead generate a <code>landingpad</code> instruction, that
1362 returns an exception object and selector value:</p>
1364 <div class="doc_code">
1366 LandingPadInst *LPadInst =
1367 Builder.CreateLandingPad(StructType::get(Int8PtrTy, Int32Ty, NULL),
1370 Value *LPadExn = Builder.CreateExtractValue(LPadInst, 0);
1371 Builder.CreateStore(LPadExn, getExceptionSlot());
1373 Value *LPadSel = Builder.CreateExtractValue(LPadInst, 1);
1374 Builder.CreateStore(LPadSel, getEHSelectorSlot());
1378 <p>It's now trivial to add the individual clauses to the <code>landingpad</code>
1381 <div class="doc_code">
1383 <i><b>// Adding a catch clause</b></i>
1384 Constant *TypeInfo = getTypeInfo();
1385 LPadInst->addClause(TypeInfo);
1387 <i><b>// Adding a C++ catch-all</b></i>
1388 LPadInst->addClause(Constant::getNullValue(Builder.getInt8PtrTy()));
1390 <i><b>// Adding a cleanup</b></i>
1391 LPadInst->setCleanup(true);
1393 <i><b>// Adding a filter clause</b></i>
1394 std::vector<Constant*> TypeInfos;
1395 Constant *TypeInfo = getFilterTypeInfo();
1396 TypeInfos.push_back(Builder.CreateBitCast(TypeInfo, Builder.getInt8PtrTy()));
1398 ArrayType *FilterTy = ArrayType::get(Int8PtrTy, TypeInfos.size());
1399 LPadInst->addClause(ConstantArray::get(FilterTy, TypeInfos));
1403 <p>Converting from using the <code>@llvm.eh.resume</code> intrinsic to
1404 the <code>resume</code> instruction is trivial. It takes the exception
1405 pointer and exception selector values returned by
1406 the <code>landingpad</code> instruction:</p>
1408 <div class="doc_code">
1410 Type *UnwindDataTy = StructType::get(Builder.getInt8PtrTy(),
1411 Builder.getInt32Ty(), NULL);
1412 Value *UnwindData = UndefValue::get(UnwindDataTy);
1413 Value *ExcPtr = Builder.CreateLoad(getExceptionObjSlot());
1414 Value *ExcSel = Builder.CreateLoad(getExceptionSelSlot());
1415 UnwindData = Builder.CreateInsertValue(UnwindData, ExcPtr, 0, "exc_ptr");
1416 UnwindData = Builder.CreateInsertValue(UnwindData, ExcSel, 1, "exc_sel");
1417 Builder.CreateResume(UnwindData);
1427 <!-- *********************************************************************** -->
1431 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1432 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
1433 <a href="http://validator.w3.org/check/referer"><img
1434 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
1436 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
1437 Last modified: $Date$