RPCS3/llvm
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm.git synced 2024-11-29 22:50:55 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

New Release Notes for LLVM 1.6.

Browse Source 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@22124 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
John Criswell 2005-05-18 20:28:46 +00:00
parent 0e9dfcb10c
commit 98391cedd5
1 changed files with 11 additions and 265 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

276
docs/ReleaseNotes.html
View File

@ -4,11 +4,11 @@
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<link rel="stylesheet" href="llvm.css" type="text/css">
<title>LLVM 1.5 Release Notes</title>
<title>LLVM 1.6 Release Notes</title>
</head>
<body>
<div class="doc_title">LLVM 1.5 Release Notes</div>
<div class="doc_title">LLVM 1.6 Release Notes</div>
<ol>
<li><a href="#intro">Introduction</a></li>
@ -32,10 +32,10 @@
<div class="doc_text">
<p>This document contains the release notes for the LLVM compiler
infrastructure, release 1.5. Here we describe the status of LLVM, including any
infrastructure, release 1.6. Here we describe the status of LLVM, including any
known problems and major improvements from the previous release. The most
up-to-date version of this document can be found on the <a
href="http://llvm.cs.uiuc.edu/releases/1.5/">LLVM 1.5 web site</a>. If you are
href="http://llvm.cs.uiuc.edu/releases/1.6/">LLVM 1.6 web site</a>. If you are
not reading this on the LLVM web pages, you should probably go there because
this document may be updated after the release.</p>
@ -60,286 +60,32 @@ href="http://llvm.cs.uiuc.edu/releases/">releases page</a>.</p>
<div class="doc_text">
<p>This is the sixth public release of the LLVM Compiler Infrastructure.</p>
<p>This is the seventh public release of the LLVM Compiler Infrastructure.</p>
<p>LLVM 1.5 is known to correctly compile a wide range of C and C++ programs,
includes bug fixes for those problems found since the 1.4 release, and includes
<p>LLVM 1.6 is known to correctly compile a wide range of C and C++ programs,
includes bug fixes for those problems found since the 1.5 release, and includes
a large number of new features and enhancements, described below.</p>
</div>
<!--=========================================================================-->
<div class="doc_subsection">
<a name="newfeatures">New Features in LLVM 1.5</a>
</div>
<!--_________________________________________________________________________-->
<div class="doc_subsubsection"><a name="newcg">New Native Code
Generators</a></div>
<div class="doc_text">
<p>
This release includes new native code generators for <a
href="#alpha-be">Alpha</a>, <a href="#ia64-be">IA-64</a>, and <a
href="#sparcv8">SPARC-V8</a> (32-bit SPARC). These code generators are still
beta quality, but are progressing rapidly. The Alpha backend is implemented
with an eye towards being compatible with the widely used SimpleScalar
simulator.
</p>
</div>
<!--_________________________________________________________________________-->
<div class="doc_subsubsection"><a name="selectiondag">New Instruction
Selector Framework</a></div>
<div class="doc_text">
<p>This release includes a <a href="CodeGenerator.html#instselect">new framework
for building instruction selectors</a>, which has long been the hardest part of
building a new LLVM target. This framework handles a lot of the mundane (but
easy to get wrong) details of writing the instruction selector, such as
generating efficient code for <a
href="LangRef.html#i_getelementptr">getelementptr</a> instructions, promoting
small integer types to larger types (e.g. for RISC targets with one size of
integer registers), expanding 64-bit integer operations for 32-bit targets, etc.
Currently, the X86, PowerPC, Alpha, and IA-64 backends use this framework. The
SPARC backends will be migrated when time permits.
</p>
</div>
<!--_________________________________________________________________________-->
<div class="doc_subsubsection"><a name="customccs">New Support for Per-Function
Calling Conventions</a></div>
<div class="doc_text">
<p>LLVM 1.5 adds supports for <a href="LangRef.html#callingconv">per-function
calling conventions</a>. Traditionally, the LLVM code generators match the
native C calling conventions for a target. This is important for compatibility,
but is not very flexible. This release allows custom calling conventions to be
established for functions, and defines three target-independent conventions (<a
href="LangRef.html#callingconv">C call, fast call, and cold call</a>) which may
be supported by code generators. When possible, the LLVM optimizer promotes C
functions to use the "fastcc" convention, allowing the use of more efficient
calling sequences (e.g., parameters are passed in registers in the X86 target).
</p>
<p>Targets may now also define target-specific calling conventions, allowing
LLVM to fully support calling convention altering options (e.g. GCC's
<tt>-mregparm</tt> flag) and well-defined target conventions (e.g. stdcall and
fastcall on X86).</p>
</div>
<!--_________________________________________________________________________-->
<div class="doc_subsubsection"><a name="tailcalls">New Support for
Proper Tail Calls</a></div>
<div class="doc_text">
<p>The release now includes support for <a
href="http://doi.acm.org/10.1145/277650.277719">proper tail calls</a>, as
required to implement languages like Scheme. Tail calls make use of two
features: custom calling conventions (described above), which allow the code
generator to use a convention where the caller deallocates its stack before it
returns. The second feature is a flag on the <a href="LangRef.html#i_call">call
instruction</a>, which indicates that the callee does not access the caller's
stack frame (indicating that it is acceptable to deallocate the caller stack
before invoking the callee). LLVM proper tail calls run on the system stack (as
do normal calls), supports indirect tail calls, tail calls with arbitrary
numbers of arguments, tail calls where the callee requires more argument space
than the caller, etc. The only case not supported are varargs calls, but that
could be added if desired.
</p>
<p>To ensure a call is interpreted as a tail call, a front-end must mark
functions as "fastcc", mark calls with the 'tail' marker, and follow the call
with a return of the called value (or void). The optimizer and code generator
attempt to handle more general cases, but the simple case will always work if
the code generator supports tail calls. Here is an example:</p>
<pre>
fastcc int %bar(int %X, int(double, int)* %FP) { ;<i> fastcc</i>
%Y = tail call fastcc int %FP(double 0.0, int %X) ;<i> tail, fastcc</i>
ret int %Y
}
</pre>
<p>In LLVM 1.5, the X86 code generator is the only target that has been enhanced
to support proper tail calls (other targets will be enhanced in future).
Further, because this support was added very close to the release, it is
disabled by default. Pass <tt>-enable-x86-fastcc</tt> to llc to enable it (this
will be enabled by default in the next release). The example above compiles to:
</p>
<pre>
bar:
sub ESP, 8 # Callee uses more space than the caller
mov ECX, DWORD PTR [ESP + 8] # Get the old return address
mov DWORD PTR [ESP + 4], 0 # First half of 0.0
mov DWORD PTR [ESP + 8], 0 # Second half of 0.0
mov DWORD PTR [ESP], ECX # Put the return address where it belongs
jmp EDX # Tail call "FP"
</pre>
<p>
With fastcc on X86, the first two integer arguments are passed in EAX/EDX, the
callee pops its arguments off the stack, and the argument area is always a
multiple of 8 bytes in size.
</p>
</div>
<!--_________________________________________________________________________-->
<div class="doc_subsubsection">Other New Features</div>
<div class="doc_text">
<ol>
<li>LLVM now includes an <a href="http://llvm.cs.uiuc.edu/PR415">
Interprocedural Sparse Conditional Constant Propagation</a> pass, named
-ipsccp, which is run by default at link-time.</li>
<li>LLVM 1.5 is now about 15% faster than LLVM 1.4 and its core data
structures use about 30% less memory.</li>
<li>Support for Microsoft Visual Studio is improved, and <a
href="GettingStartedVS.html">now documented</a>. Most LLVM tools build
natively with Visual C++ now.</li>
<li><a href="GettingStarted.html#config">Configuring LLVM to build a subset
of the available targets</a> is now implemented, via the
<tt>--enable-targets=</tt> option.</li>
<li>LLVM can now create native shared libraries with '<tt>llvm-gcc ...
-shared -Wl,-native</tt>' (or with <tt>-Wl,-native-cbe</tt>).</li>
<li>LLVM now supports a new "<a href="LangRef.html#i_prefetch">llvm.prefetch
</a>" intrinsic, and llvm-gcc now supports __builtin_prefetch.
<li>LLVM now supports intrinsics for <a href="LangRef.html#int_count">bit
counting</a> and llvm-gcc now implements the GCC
<tt>__builtin_popcount</tt>, <tt>__builtin_ctz</tt>, and
<tt>__builtin_clz</tt> builtins.</li>
<li>LLVM now mostly builds on HP-UX with the HP aCC Compiler.</li>
<li>The LLVM X86 backend can now emit Cygwin-compatible .s files.</li>
<li>LLVM now includes workarounds in the code generator generator which
reduces the likelyhood of <a href="http://llvm.cs.uiuc.edu/PR448">GCC
hitting swap during optimized builds</a>.</li>
<li>The <a href="http://llvm.cs.uiuc.edu/ProjectsWithLLVM/#llvmtv">LLVM
Transformation Visualizer</a> (llvm-tv) project has been updated to
work with LLVM 1.5.</li>
<li>Nightly tester output is now archived on the <a
href="http://mail.cs.uiuc.edu/pipermail/llvm-testresults/">
llvm-testresults</a> mailing list.</li>
</ol>
<a name="newfeatures">New Features in LLVM 1.6</a>
</div>
<!--=========================================================================-->
<div class="doc_subsection">
<a name="codequality">Code Quality Improvements in LLVM 1.5</a>
<a name="codequality">Code Quality Improvements in LLVM 1.6</a>
</div>
<div class="doc_text">
<ol>
<li>The new -simplify-libcalls pass improves code generated for well-known
library calls. The pass optimizes calls to many of the string, memory, and
standard I/O functions (e.g. replace the calls with simpler/faster calls) when
possible, given information known statically about the arguments to the call.
</li>
<li>The -globalopt pass now promotes non-address-taken static globals that are
only accessed in main to SSA registers.</li>
<li>Loops with trip counts based on array pointer comparisons (e.g. "<tt>for (i
= 0; &amp;A[i] != &amp;A[n]; ++i) ...</tt>") are optimized better than before,
which primarily helps iterator-intensive C++ code.</li>
<li>The optimizer now eliminates simple cases where redundant conditions exist
between neighboring blocks.</li>
<li>The reassociation pass (which turns (1+X+3) into (X+1+3) among other
things), is more aggressive and intelligent.</li>
<li>The -prune-eh pass now detects no-return functions in addition to the
no-unwind functions it did before.</li>
<li>The -globalsmodref alias analysis generates more precise results in some
cases.</li>
</ol>
</div>
<!--=========================================================================-->
<div class="doc_subsection">
<a name="codequality">Code Generator Improvements in LLVM 1.5</a>
<a name="codequality">Code Generator Improvements in LLVM 1.6</a>
</div>
<div class="doc_text">
<ol>
<li>The code generator now can provide and use information about commutative
two-address instructions when performing register allocation.</li>
<li>The code generator now tracks function live-in registers explicitly,
instead of requiring the target to generate 'implicit defs' at the
entry to a function.</li>
<li>The code generator can lower integer division by a constant to
multiplication by a magic constant and multiplication by a constant into
shift/add sequences.</li>
<li>The code generator compiles fabs/fneg/sin/cos/sqrt to assembly instructions
when possible.</li>
<li>The PowerPC backend generates better code in many cases, making use of
FMA instructions and the recording ("dot") forms of various PowerPC
instructions.</li>
</ol>
</div>
<!--=========================================================================-->
<div class="doc_subsection">
<a name="bugfix">Significant Bugs Fixed in LLVM 1.5</a>
</div>
<div class="doc_text">
<p>Bugs fixed in the LLVM Core:</p>
<ol>
<li><a href="http://llvm.cs.uiuc.edu/PR491">[dse] DSE deletes stores that
are partially overwritten by smaller stores</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR548">[instcombine] miscompilation of
setcc or setcc in one case</a></li>
<li>Transition code for LLVM 1.0 style varargs was removed from the .ll file
parser. LLVM 1.0 bytecode files are still supported. </li>
</ol>
<p>Code Generator Bugs:</p>
<ol>
<li><a href="http://llvm.cs.uiuc.edu/PR490">[cbackend] Logical constant
expressions (and/or/xor) not implemented</a>.</li>
<li><a href="http://llvm.cs.uiuc.edu/PR511">[cbackend] C backend does not
respect 'volatile'</a>.</li>
<li>The JIT sometimes miscompiled globals and constant pool entries for
64-bit integer constants on 32-bit hosts.</li>
<li>The C backend should no longer produce code that crashes ICC 8.1.</li>
</ol>
<p>Bugs in the C/C++ front-end:</p>
<ol>
<li><a href="http://llvm.cs.uiuc.edu/PR487">[llvmgcc] llvm-gcc incorrectly
rejects some constant initializers involving the addresses of array
elements</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR501">[llvm-g++] Crash compiling
anonymous union</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR509">[llvm-g++] Do not use dynamic
initialization where static init will do</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR510">[llvmgcc] Field offset
miscalculated for some structure fields following bit fields</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR513">[llvm-g++] Temporary lifetimes
incorrect for short circuit logical operations</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR517">[llvm-gcc] Crash compiling
bitfield &lt;-&gt; aggregate assignment</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR520">[llvm-g++] Error compiling
virtual function thunk with an unnamed argument</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR522">[llvm-gcc] Crash on certain
C99 complex number routines</a></li>
<li><a href="http://llvm.cs.uiuc.edu/PR529">[llvm-g++] Crash using placement
new on an array type</a></li>
</ol>
<a name="bugfix">Significant Bugs Fixed in LLVM 1.6</a>
</div>
<!-- *********************************************************************** -->