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 2.8 Release Notes</title>
11 <div class="doc_title">LLVM 2.8 Release Notes</div>
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 2.8</a></li>
20 <li><a href="#whatsnew">What's New in LLVM 2.8?</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 2.8
34 <a href="http://llvm.org/releases/2.7/docs/ReleaseNotes.html">LLVM 2.7
35 Release Notes</a>.</h1>
38 <!-- *********************************************************************** -->
39 <div class="doc_section">
40 <a name="intro">Introduction</a>
42 <!-- *********************************************************************** -->
44 <div class="doc_text">
46 <p>This document contains the release notes for the LLVM Compiler
47 Infrastructure, release 2.8. Here we describe the status of LLVM, including
48 major improvements from the previous release and significant known problems.
49 All LLVM releases may be downloaded from the <a
50 href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
52 <p>For more information about LLVM, including information about the latest
53 release, please check out the <a href="http://llvm.org/">main LLVM
54 web site</a>. If you have questions or comments, the <a
55 href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM Developer's
56 Mailing List</a> is a good place to send them.</p>
58 <p>Note that if you are reading this file from a Subversion checkout or the
59 main LLVM web page, this document applies to the <i>next</i> release, not the
60 current one. To see the release notes for a specific release, please see the
61 <a href="http://llvm.org/releases/">releases page</a>.</p>
68 include/llvm/Analysis/LiveValues.h => Dan
69 lib/Transforms/IPO/MergeFunctions.cpp => consider for 2.8.
70 llvm/Analysis/PointerTracking.h => Edwin wants this, consider for 2.8.
75 <!-- Features that need text if they're finished for 2.9:
78 loop dependence analysis
80 CorrelatedValuePropagation
83 <!-- Announcement, lldb, libc++ -->
86 MachineCSE tuned and on by default.
87 llvm.dbg.value: variable debug info for optimized code
88 MC Assembler backend is now real, does relaxation and is bitwise identical
89 with darwin assembler in huge majority of all cases.
90 new GHC calling convention
91 New half float intrinsics LangRef.html#int_fp16
92 Rewrote tblgen's type inference for backends to be more consistent and
93 diagnose more target bugs. This also allows limited support for writing
94 patterns for instructions that return multiple results, e.g. a virtual
95 register and a flag result. Stuff that used 'parallel' before should use
97 New ARM/Thumb disassembler support in MC.
98 New SSEDomainFix pass:
99 On Nehalem and newer CPUs there is a 2 cycle latency penalty on using a
100 register in a different domain than where it was defined. Some instructions
101 have equvivalents for different domains, like por/orps/orpd. The
102 SSEDomainFix pass tries to minimize the number of domain crossings by
103 changing between equvivalent opcodes where possible.
104 Support for the Intel AES instructions in the assembler.
105 memcpy, memmove, and memset now take address space qualified pointers + volatile.
110 <!-- *********************************************************************** -->
111 <div class="doc_section">
112 <a name="subproj">Sub-project Status Update</a>
114 <!-- *********************************************************************** -->
116 <div class="doc_text">
118 The LLVM 2.8 distribution currently consists of code from the core LLVM
119 repository (which roughly includes the LLVM optimizers, code generators
120 and supporting tools), the Clang repository and the llvm-gcc repository. In
121 addition to this code, the LLVM Project includes other sub-projects that are in
122 development. Here we include updates on these subprojects.
128 <!--=========================================================================-->
129 <div class="doc_subsection">
130 <a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
133 <div class="doc_text">
135 <p><a href="http://clang.llvm.org/">Clang</a> is an LLVM front end for the C,
136 C++, and Objective-C languages. Clang aims to provide a better user experience
137 through expressive diagnostics, a high level of conformance to language
138 standards, fast compilation, and low memory use. Like LLVM, Clang provides a
139 modular, library-based architecture that makes it suitable for creating or
140 integrating with other development tools. Clang is considered a
141 production-quality compiler for C, Objective-C, C++ and Objective-C++ on x86
142 (32- and 64-bit), and for darwin-arm targets.</p>
144 <p>In the LLVM 2.8 time-frame, the Clang team has made many improvements:</p>
147 <li>Surely these guys have done something</li>
151 <!--=========================================================================-->
152 <div class="doc_subsection">
153 <a name="clangsa">Clang Static Analyzer</a>
156 <div class="doc_text">
158 <p>The <a href="http://clang-analyzer.llvm.org/">Clang Static Analyzer</a>
159 project is an effort to use static source code analysis techniques to
160 automatically find bugs in C and Objective-C programs (and hopefully <a
161 href="http://clang-analyzer.llvm.org/dev_cxx.html">C++ in the
162 future</a>!). The tool is very good at finding bugs that occur on specific
163 paths through code, such as on error conditions.</p>
165 <p>The LLVM 2.8 release fixes a number of bugs and slightly improves precision
166 over 2.7, but there are no major new features in the release.
171 <!--=========================================================================-->
172 <div class="doc_subsection">
173 <a name="vmkit">VMKit: JVM/CLI Virtual Machine Implementation</a>
176 <div class="doc_text">
178 The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation of
179 a JVM and a CLI Virtual Machine (Microsoft .NET is an
180 implementation of the CLI) using LLVM for static and just-in-time
183 <p>With the release of LLVM 2.8, ...</p>
188 <!--=========================================================================-->
189 <div class="doc_subsection">
190 <a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
193 <div class="doc_text">
195 The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
196 is a simple library that provides an implementation of the low-level
197 target-specific hooks required by code generation and other runtime components.
198 For example, when compiling for a 32-bit target, converting a double to a 64-bit
199 unsigned integer is compiled into a runtime call to the "__fixunsdfdi"
200 function. The compiler-rt library provides highly optimized implementations of
201 this and other low-level routines (some are 3x faster than the equivalent
202 libgcc routines).</p>
205 All of the code in the compiler-rt project is available under the standard LLVM
206 License, a "BSD-style" license. New in LLVM 2.8:
213 <!--=========================================================================-->
214 <div class="doc_subsection">
215 <a name="dragonegg">DragonEgg: llvm-gcc ported to gcc-4.5</a>
218 <div class="doc_text">
220 <a href="http://dragonegg.llvm.org/">DragonEgg</a> is a port of llvm-gcc to
221 gcc-4.5. Unlike llvm-gcc, which makes many intrusive changes to the underlying
222 gcc-4.2 code, dragonegg in theory does not require any gcc-4.5 modifications
223 whatsoever (currently one small patch is needed). This is thanks to the new
224 <a href="http://gcc.gnu.org/wiki/plugins">gcc plugin architecture</a>, which
225 makes it possible to modify the behaviour of gcc at runtime by loading a plugin,
226 which is nothing more than a dynamic library which conforms to the gcc plugin
227 interface. DragonEgg is a gcc plugin that causes the LLVM optimizers to be run
228 instead of the gcc optimizers, and the LLVM code generators instead of the gcc
229 code generators, just like llvm-gcc. To use it, you add
230 "-fplugin=path/dragonegg.so" to the gcc-4.5 command line, and gcc-4.5 magically
231 becomes llvm-gcc-4.5!
235 DragonEgg is still a work in progress. Currently C works very well, while C++,
236 Ada and Fortran work fairly well. All other languages either don't work at all,
237 or only work poorly. For the moment only the x86-32 and x86-64 targets are
238 supported, and only on linux and darwin (darwin needs an additional gcc patch).
248 <!--=========================================================================-->
249 <div class="doc_subsection">
250 <a name="mc">llvm-mc: Machine Code Toolkit</a>
253 <div class="doc_text">
255 The LLVM Machine Code (aka MC) sub-project of LLVM was created to solve a number
256 of problems in the realm of assembly, disassembly, object file format handling,
257 and a number of other related areas that CPU instruction-set level tools work
258 in. It is a sub-project of LLVM which provides it with a number of advantages
259 over other compilers that do not have tightly integrated assembly-level tools.
260 For a gentle introduction, please see the <a
261 href="http://blog.llvm.org/2010/04/intro-to-llvm-mc-project.html">Intro to the
262 LLVM MC Project Blog Post</a>.
265 <p>2.8 status here. Basic correctness, some obscure missing instructions on
266 mainline, on by default in clang.
267 Entire compiler backend converted to use mcstreamer.
271 <!--=========================================================================-->
272 <div class="doc_subsection">
273 <a name="lldb">LLDB: Low Level Debugger</a>
276 <div class="doc_text">
278 <a href="http://lldb.llvm.org/">LLDB</a> is</p>
289 <!--=========================================================================-->
290 <div class="doc_subsection">
291 <a name="libc++">libc++: C++ Standard Library</a>
294 <div class="doc_text">
296 <a href="http://libc++.llvm.org/">libc++</a> is</p>
308 <!-- *********************************************************************** -->
309 <div class="doc_section">
310 <a name="externalproj">External Open Source Projects Using LLVM 2.8</a>
312 <!-- *********************************************************************** -->
314 <div class="doc_text">
316 <p>An exciting aspect of LLVM is that it is used as an enabling technology for
317 a lot of other language and tools projects. This section lists some of the
318 projects that have already been updated to work with LLVM 2.8.</p>
321 <!--=========================================================================-->
322 <div class="doc_subsection">
323 <a name="tce">TTA-based Codesign Environment (TCE)</a>
326 <div class="doc_text">
328 <a href="http://tce.cs.tut.fi/">TCE</a> is a toolset for designing
329 application-specific processors (ASP) based on the Transport triggered
330 architecture (TTA). The toolset provides a complete co-design flow from C/C++
331 programs down to synthesizable VHDL and parallel program binaries. Processor
332 customization points include the register files, function units, supported
333 operations, and the interconnection network.</p>
335 <p>TCE uses llvm-gcc/Clang and LLVM for C/C++ language support, target
336 independent optimizations and also for parts of code generation. It generates
337 new LLVM-based code generators "on the fly" for the designed TTA processors and
338 loads them in to the compiler backend as runtime libraries to avoid per-target
339 recompilation of larger parts of the compiler chain.</p>
343 <!--=========================================================================-->
344 <div class="doc_subsection">
345 <a name="Horizon">Horizon Bytecode Compiler</a>
348 <div class="doc_text">
350 <a href="http://www.quokforge.org/projects/horizon">Horizon</a> is a bytecode
351 language and compiler written on top of LLVM, intended for producing
352 single-address-space managed code operating systems that
353 run faster than the equivalent multiple-address-space C systems.
354 More in-depth blurb is available on <a
355 href="http://www.quokforge.org/projects/horizon/wiki/Wiki">the wiki</a>.</p>
359 <!--=========================================================================-->
360 <div class="doc_subsection">
361 <a name="clamav">Clam AntiVirus</a>
364 <div class="doc_text">
366 <a href=http://www.clamav.net>Clam AntiVirus</a> is an open source (GPL)
367 anti-virus toolkit for UNIX, designed especially for e-mail scanning on mail
368 gateways. Since version 0.96 it has <a
369 href="http://vrt-sourcefire.blogspot.com/2010/09/introduction-to-clamavs-low-level.html">bytecode
370 signatures</a> that allow writing detections for complex malware. It
371 uses LLVM's JIT to speed up the execution of bytecode on
372 X86,X86-64,PPC32/64, falling back to its own interpreter otherwise.
373 The git version was updated to work with LLVM 2.8
377 href="http://git.clamav.net/gitweb?p=clamav-bytecode-compiler.git;a=blob_plain;f=docs/user/clambc-user.pdf">
378 ClamAV bytecode compiler</a> uses Clang and LLVM to compile a C-like
379 language, insert runtime checks, and generate ClamAV bytecode.</p>
383 <!--=========================================================================-->
384 <div class="doc_subsection">
385 <a name="pure">Pure</a>
388 <div class="doc_text">
390 <a href="http://pure-lang.googlecode.com/">Pure</a>
391 is an algebraic/functional
392 programming language based on term rewriting. Programs are collections
393 of equations which are used to evaluate expressions in a symbolic
394 fashion. Pure offers dynamic typing, eager and lazy evaluation, lexical
395 closures, a hygienic macro system (also based on term rewriting),
396 built-in list and matrix support (including list and matrix
397 comprehensions) and an easy-to-use C interface. The interpreter uses
398 LLVM as a backend to JIT-compile Pure programs to fast native code.</p>
400 <p>Pure versions 0.44 and later have been tested and are known to work with
401 LLVM 2.8 (and continue to work with older LLVM releases >= 2.5).</p>
405 <!--=========================================================================-->
406 <div class="doc_subsection">
407 <a name="GHC">Glasgow Haskell Compiler (GHC)</a>
410 <div class="doc_text">
412 <a href="http://www.haskell.org/ghc/">GHC</a> is an open source,
413 state-of-the-art programming suite for
414 Haskell, a standard lazy functional programming language. It includes
415 an optimizing static compiler generating good code for a variety of
416 platforms, together with an interactive system for convenient, quick
419 <p>In addition to the existing C and native code generators, GHC 7.0 now
421 href="http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM">LLVM
422 code generator</a>. GHC supports LLVM 2.7 and later.</p>
426 <!--=========================================================================-->
427 <div class="doc_subsection">
428 <a name="Clay">Clay Programming Language</a>
431 <div class="doc_text">
433 <a href="http://tachyon.in/clay/">Clay</a> is a new systems programming
434 language that is specifically designed for generic programming. It makes
435 generic programming very concise thanks to whole program type propagation. It
436 uses LLVM as its backend.</p>
440 <!-- *********************************************************************** -->
441 <div class="doc_section">
442 <a name="whatsnew">What's New in LLVM 2.8?</a>
444 <!-- *********************************************************************** -->
446 <div class="doc_text">
448 <p>This release includes a huge number of bug fixes, performance tweaks and
449 minor improvements. Some of the major improvements and new features are listed
455 <!--=========================================================================-->
456 <div class="doc_subsection">
457 <a name="orgchanges">LLVM Community Changes</a>
460 <div class="doc_text">
462 <p>In addition to changes to the code, between LLVM 2.7 and 2.8, a number of
463 organization changes have happened:
467 <li>libc++ and lldb are new</li>
468 <li>Debugging optimized code support.</li>
472 <!--=========================================================================-->
473 <div class="doc_subsection">
474 <a name="majorfeatures">Major New Features</a>
477 <div class="doc_text">
479 <p>LLVM 2.8 includes several major new capabilities:</p>
482 <li>atomic lowering pass.</li>
483 <li>RegionInfo pass: opt -regions analyze" or "opt -view-regions".
484 <!-- Tobias Grosser --></li>
485 <li>ARMGlobalMerge: <!-- Anton --> </li>
491 <!--=========================================================================-->
492 <div class="doc_subsection">
493 <a name="coreimprovements">LLVM IR and Core Improvements</a>
496 <div class="doc_text">
497 <p>LLVM IR has several new features for better support of new targets and that
498 expose new optimization opportunities:</p>
502 <li>LLVM 2.8 changes the internal order of operands in <a
503 href="http://llvm.org/doxygen/classllvm_1_1InvokeInst.html"><tt>InvokeInst</tt></a>
504 and <a href="http://llvm.org/doxygen/classllvm_1_1CallInst.html"><tt>CallInst</tt></a>.
505 To be portable across releases, resort to <tt>CallSite</tt> and the
506 high-level accessors, such as <tt>getCalledValue</tt> and <tt>setUnwindDest</tt>.
509 You can no longer pass use_iterators directly to cast<> (and similar), because
510 these routines tend to perform costly dereference operations more than once. You
511 have to dereference the iterators yourself and pass them in.
514 llvm.memcpy.*, llvm.memset.*, llvm.memmove.* (and possibly other?) intrinsics
515 take an extra parameter now (i1 isVolatile), totaling 5 parameters.
516 If you were creating these intrinsic calls and prototypes yourself (as opposed
517 to using Intrinsic::getDeclaration), you can use UpgradeIntrinsicFunction/UpgradeIntrinsicCall
518 to be portable accross releases.
519 Note that you cannot use Intrinsic::getDeclaration() in a backwards compatible
520 way (needs 2/3 types now, in 2.7 it needed just 1).
523 SetCurrentDebugLocation takes a DebugLoc now instead of a MDNode.
524 Change your code to use
525 SetCurrentDebugLocation(DebugLoc::getFromDILocation(...)).
528 VISIBILITY_HIDDEN is gone.
531 The <tt>RegisterPass</tt> and <tt>RegisterAnalysisGroup</tt> templates are
532 considered deprecated, but continue to function in LLVM 2.8. Clients are
533 strongly advised to use the upcoming <tt>INITIALIZE_PASS()</tt> and
534 <tt>INITIALIZE_AG_PASS()</tt> macros instead.
536 SMDiagnostic takes different parameters now. //FIXME: how to upgrade?
539 The constructor for the Triple class no longer tries to understand odd triple
540 specifications. Frontends should ensure that they only pass valid triples to
541 LLVM. The Triple::normalize utility method has been added to help front-ends
542 deal with funky triples.
544 Some APIs got renamed:
546 <li>llvm_report_error -> report_fatal_error</li>
547 <li>llvm_install_error_handler -> install_fatal_error_handler</li>
548 <li>llvm::DwarfExceptionHandling -> llvm::JITExceptionHandling</li>
555 <!--=========================================================================-->
556 <div class="doc_subsection">
557 <a name="optimizer">Optimizer Improvements</a>
560 <div class="doc_text">
562 <p>In addition to a large array of minor performance tweaks and bug fixes, this
563 release includes a few major enhancements and additions to the optimizers:</p>
574 <!--=========================================================================-->
575 <div class="doc_subsection">
576 <a name="executionengine">Interpreter and JIT Improvements</a>
579 <div class="doc_text">
588 <!--=========================================================================-->
589 <div class="doc_subsection">
590 <a name="codegen">Target Independent Code Generator Improvements</a>
593 <div class="doc_text">
595 <p>We have put a significant amount of work into the code generator
596 infrastructure, which allows us to implement more aggressive algorithms and make
600 <li>MachO writer works.</li>
604 <!--=========================================================================-->
605 <div class="doc_subsection">
606 <a name="x86">X86-32 and X86-64 Target Improvements</a>
609 <div class="doc_text">
610 <p>New features of the X86 target include:
614 <li>The X86 backend now supports holding X87 floating point stack values
615 in registers across basic blocks, dramatically improving performance of code
616 that uses long double, and when targetting CPUs that don't support SSE.</li>
622 <!--=========================================================================-->
623 <div class="doc_subsection">
624 <a name="ARM">ARM Target Improvements</a>
627 <div class="doc_text">
628 <p>New features of the ARM target include:
634 All of the NEON load and store intrinsics (llvm.arm.neon.vld* and
635 llvm.arm.neon.vst*) take an extra parameter to specify the alignment in bytes
636 of the memory being accessed.
639 The llvm.arm.neon.vaba intrinsic (vector absolute difference and
640 accumulate) has been removed. This operation is now represented using
641 the llvm.arm.neon.vabd intrinsic (vector absolute difference) followed by a
645 The llvm.arm.neon.vabdl and llvm.arm.neon.vabal intrinsics (lengthening
646 vector absolute difference with and without accumlation) have been removed.
647 They are represented using the llvm.arm.neon.vabd intrinsic (vector absolute
648 difference) followed by a vector zero-extend operation, and for vabal,
652 The llvm.arm.neon.vmovn intrinsic has been removed. Calls of this intrinsic
653 are now replaced by vector truncate operations.
656 The llvm.arm.neon.vmovls and llvm.arm.neon.vmovlu intrinsics have been
657 removed. They are now represented as vector sign-extend (vmovls) and
658 zero-extend (vmovlu) operations.
661 The llvm.arm.neon.vaddl*, llvm.arm.neon.vaddw*, llvm.arm.neon.vsubl*, and
662 llvm.arm.neon.vsubw* intrinsics (lengthening vector add and subtract) have
663 been removed. They are replaced by vector add and vector subtract operations
664 where one (vaddw, vsubw) or both (vaddl, vsubl) of the operands are either
665 sign-extended or zero-extended.
668 The llvm.arm.neon.vmulls, llvm.arm.neon.vmullu, llvm.arm.neon.vmlal*, and
669 llvm.arm.neon.vmlsl* intrinsics (lengthening vector multiply with and without
670 accumulation and subtraction) have been removed. These operations are now
671 represented as vector multiplications where the operands are either
672 sign-extended or zero-extended, followed by a vector add for vmlal or a
673 vector subtract for vmlsl. Note that the polynomial vector multiply
674 intrinsic, llvm.arm.neon.vmullp, remains unchanged.
680 <!--=========================================================================-->
681 <div class="doc_subsection">
682 <a name="newapis">New Useful APIs</a>
685 <div class="doc_text">
687 <p>This release includes a number of new APIs that are used internally, which
688 may also be useful for external clients.
698 <!--=========================================================================-->
699 <div class="doc_subsection">
700 <a name="otherimprovements">Other Improvements and New Features</a>
703 <div class="doc_text">
704 <p>Other miscellaneous features include:</p>
713 <!--=========================================================================-->
714 <div class="doc_subsection">
715 <a name="changes">Major Changes and Removed Features</a>
718 <div class="doc_text">
720 <p>If you're already an LLVM user or developer with out-of-tree changes based
721 on LLVM 2.7, this section lists some "gotchas" that you may run into upgrading
722 from the previous release.</p>
725 <li>.ll file doesn't produce #uses comments anymore, to get them, run a .bc file
726 through "llvm-dis --show-annotations".</li>
727 <li>MSIL Backend removed.</li>
728 <li>ABCD and SSI passes removed.</li>
729 <li>'Union' LLVM IR feature removed.</li>
732 <p>In addition, many APIs have changed in this release. Some of the major LLVM
741 <!-- *********************************************************************** -->
742 <div class="doc_section">
743 <a name="knownproblems">Known Problems</a>
745 <!-- *********************************************************************** -->
747 <div class="doc_text">
749 <p>This section contains significant known problems with the LLVM system,
750 listed by component. If you run into a problem, please check the <a
751 href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
752 there isn't already one.</p>
756 <!-- ======================================================================= -->
757 <div class="doc_subsection">
758 <a name="experimental">Experimental features included with this release</a>
761 <div class="doc_text">
763 <p>The following components of this LLVM release are either untested, known to
764 be broken or unreliable, or are in early development. These components should
765 not be relied on, and bugs should not be filed against them, but they may be
766 useful to some people. In particular, if you would like to work on one of these
767 components, please contact us on the <a
768 href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
771 <li>The Alpha, SPU, MIPS, PIC16, Blackfin, MSP430, SystemZ and MicroBlaze
772 backends are experimental.</li>
773 <li><tt>llc</tt> "<tt>-filetype=asm</tt>" (the default) is the only
774 supported value for this option. XXX Update me</li>
779 <!-- ======================================================================= -->
780 <div class="doc_subsection">
781 <a name="x86-be">Known problems with the X86 back-end</a>
784 <div class="doc_text">
787 <li>The X86 backend does not yet support
788 all <a href="http://llvm.org/PR879">inline assembly that uses the X86
789 floating point stack</a>. It supports the 'f' and 't' constraints, but not
791 <li>Win64 code generation wasn't widely tested. Everything should work, but we
792 expect small issues to happen. Also, llvm-gcc cannot build the mingw64
793 runtime currently due to lack of support for the 'u' inline assembly
794 constraint and for X87 floating point inline assembly.</li>
795 <li>The X86-64 backend does not yet support the LLVM IR instruction
796 <tt>va_arg</tt>. Currently, front-ends support variadic
797 argument constructs on X86-64 by lowering them manually.</li>
802 <!-- ======================================================================= -->
803 <div class="doc_subsection">
804 <a name="ppc-be">Known problems with the PowerPC back-end</a>
807 <div class="doc_text">
810 <li>The Linux PPC32/ABI support needs testing for the interpreter and static
811 compilation, and lacks support for debug information.</li>
816 <!-- ======================================================================= -->
817 <div class="doc_subsection">
818 <a name="arm-be">Known problems with the ARM back-end</a>
821 <div class="doc_text">
824 <li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
825 processors, thumb programs can crash or produce wrong
826 results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
827 <li>Compilation for ARM Linux OABI (old ABI) is supported but not fully tested.
833 <!-- ======================================================================= -->
834 <div class="doc_subsection">
835 <a name="sparc-be">Known problems with the SPARC back-end</a>
838 <div class="doc_text">
841 <li>The SPARC backend only supports the 32-bit SPARC ABI (-m32); it does not
842 support the 64-bit SPARC ABI (-m64).</li>
847 <!-- ======================================================================= -->
848 <div class="doc_subsection">
849 <a name="mips-be">Known problems with the MIPS back-end</a>
852 <div class="doc_text">
855 <li>64-bit MIPS targets are not supported yet.</li>
860 <!-- ======================================================================= -->
861 <div class="doc_subsection">
862 <a name="alpha-be">Known problems with the Alpha back-end</a>
865 <div class="doc_text">
869 <li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
870 appropriate nops inserted to ensure restartability.</li>
875 <!-- ======================================================================= -->
876 <div class="doc_subsection">
877 <a name="c-be">Known problems with the C back-end</a>
880 <div class="doc_text">
882 <p>The C backend has numerous problems and is not being actively maintained.
883 Depending on it for anything serious is not advised.</p>
886 <li><a href="http://llvm.org/PR802">The C backend has only basic support for
887 inline assembly code</a>.</li>
888 <li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common
889 C++ programs</a>, preventing intermixing between C++ compiled by the CBE and
890 C++ code compiled with <tt>llc</tt> or native compilers.</li>
891 <li>The C backend does not support all exception handling constructs.</li>
892 <li>The C backend does not support arbitrary precision integers.</li>
898 <!-- ======================================================================= -->
899 <div class="doc_subsection">
900 <a name="llvm-gcc">Known problems with the llvm-gcc front-end</a>
903 <div class="doc_text">
905 <p>llvm-gcc is generally very stable for the C family of languages. The only
906 major language feature of GCC not supported by llvm-gcc is the
907 <tt>__builtin_apply</tt> family of builtins. However, some extensions
908 are only supported on some targets. For example, trampolines are only
909 supported on some targets (these are used when you take the address of a
910 nested function).</p>
912 <p>Fortran support generally works, but there are still several unresolved bugs
913 in <a href="http://llvm.org/bugs/">Bugzilla</a>. Please see the
914 tools/gfortran component for details. Note that llvm-gcc is missing major
915 Fortran performance work in the frontend and library that went into GCC after
916 4.2. If you are interested in Fortran, we recommend that you consider using
917 <a href="#dragonegg">dragonegg</a> instead.</p>
919 <p>The llvm-gcc 4.2 Ada compiler has basic functionality. However, this is not a
920 mature technology, and problems should be expected. For example:</p>
922 <li>The Ada front-end currently only builds on X86-32. This is mainly due
923 to lack of trampoline support (pointers to nested functions) on other platforms.
924 However, it <a href="http://llvm.org/PR2006">also fails to build on X86-64</a>
925 which does support trampolines.</li>
926 <li>The Ada front-end <a href="http://llvm.org/PR2007">fails to bootstrap</a>.
927 This is due to lack of LLVM support for <tt>setjmp</tt>/<tt>longjmp</tt> style
928 exception handling, which is used internally by the compiler.
929 Workaround: configure with <tt>--disable-bootstrap</tt>.</li>
930 <li>The c380004, <a href="http://llvm.org/PR2010">c393010</a>
931 and <a href="http://llvm.org/PR2421">cxg2021</a> ACATS tests fail
932 (c380004 also fails with gcc-4.2 mainline).
933 If the compiler is built with checks disabled then <a href="http://llvm.org/PR2010">c393010</a>
934 causes the compiler to go into an infinite loop, using up all system memory.</li>
935 <li>Some GCC specific Ada tests continue to crash the compiler.</li>
936 <li>The <tt>-E</tt> binder option (exception backtraces)
937 <a href="http://llvm.org/PR1982">does not work</a> and will result in programs
938 crashing if an exception is raised. Workaround: do not use <tt>-E</tt>.</li>
939 <li>Only discrete types <a href="http://llvm.org/PR1981">are allowed to start
940 or finish at a non-byte offset</a> in a record. Workaround: do not pack records
941 or use representation clauses that result in a field of a non-discrete type
942 starting or finishing in the middle of a byte.</li>
943 <li>The <tt>lli</tt> interpreter <a href="http://llvm.org/PR2009">considers
944 'main' as generated by the Ada binder to be invalid</a>.
945 Workaround: hand edit the file to use pointers for <tt>argv</tt> and
946 <tt>envp</tt> rather than integers.</li>
947 <li>The <tt>-fstack-check</tt> option <a href="http://llvm.org/PR2008">is
952 <!-- *********************************************************************** -->
953 <div class="doc_section">
954 <a name="additionalinfo">Additional Information</a>
956 <!-- *********************************************************************** -->
958 <div class="doc_text">
960 <p>A wide variety of additional information is available on the <a
961 href="http://llvm.org">LLVM web page</a>, in particular in the <a
962 href="http://llvm.org/docs/">documentation</a> section. The web page also
963 contains versions of the API documentation which is up-to-date with the
964 Subversion version of the source code.
965 You can access versions of these documents specific to this release by going
966 into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
968 <p>If you have any questions or comments about LLVM, please feel free to contact
969 us via the <a href="http://llvm.org/docs/#maillist"> mailing
974 <!-- *********************************************************************** -->
978 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
979 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
980 <a href="http://validator.w3.org/check/referer"><img
981 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
983 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
984 Last modified: $Date$