Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / dec8cfb43c4dd4db7110414507528439c6200b40 / . / docs / OpenProjects.html
blob: a6ab3446ebf0c13b8ec565db942730d4795c7a48 [file] [log] [blame]
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>Open LLVM Projects</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
7</head>
8<body>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]9
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]10<div class="doc_title">
11 Open LLVM Projects
12</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]13
14<ul>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]15 <li><a href="#what">What is this?</a></li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]16 <li><a href="#improving">Improving the current system</a>
17 <ol>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]18 <li><a href="#glibc">Port glibc to LLVM</a></li>
19 <li><a href="#NightlyTest">Improving the Nightly Tester</a></li>
20 <li><a href="#programs">Compile programs with the LLVM Compiler</a></li>
21 <li><a href="#llvm_ir">Extend the LLVM intermediate representation</a></li>
22 <li><a href="#misc_imp">Miscellaneous Improvements</a></li>
23 </ol></li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]24
25 <li><a href="#new">Adding new capabilities to LLVM</a>
26 <ol>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]27 <li><a href="#pointeranalysis">Pointer and Alias Analysis</a></li>
28 <li><a href="#profileguided">Profile Guided Optimization</a></li>
29 <li><a href="#xforms">New Transformations and Analyses</a></li>
30 <li><a href="#x86be">X86 Back-end Improvements</a></li>
31 <li><a href="#misc_new">Miscellaneous Additions</a></li>
32 </ol></li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]33</ul>
34
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]35<!-- *********************************************************************** -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]36<div class="doc_section">
37 <a name="what">What is this?</a>
38</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]39<!-- *********************************************************************** -->
40
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]41<div class="doc_text">
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]42
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]43<p>This document is meant to be a sort of "big TODO list" for LLVM. Each
44project in this document is something that would be useful for LLVM to have, and
45would also be a great way to get familiar with the system. Some of these
46projects are small and self-contained, which may be implemented in a couple of
47days, others are larger. Several of these projects may lead to interesting
48research projects in their own right. In any case, we welcome all
49contributions.</p>
50
51<p>If you are thinking about tackling one of these projects, please send a mail
52to the <a href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]53Developer's</a> mailing list, so that we know the project is being worked on.
54Additionally this is a good way to get more information about a specific project
Alkis Evlogimenos45ff0272004-03-11 02:50:35 +0000[diff] [blame]55or to suggest other projects to add to this page.
56</p>
57
58<p>The projects in this page are open ended. More specific projects are
59filed as unassigned enhancements in our <a href="http://llvm.cs.uiuc.edu/bugs/">
60LLVM bug tracker</a>. Here is the current list:
61</p>
62
63<iframe src="http://llvm.cs.uiuc.edu/bugs/buglist.cgi?keywords_type=allwords&keywords=&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&bug_severity=enhancement&emailassigned_to1=1&emailtype1=substring&email1=unassigned" frameborder="1" align="center" width="100%" height="400">
64</iframe>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]65
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]66</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]67
68<!-- *********************************************************************** -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]69<div class="doc_section">
70 <a name="improving">Improving the current system</a>
71</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]72<!-- *********************************************************************** -->
73
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]74<div class="doc_text">
75
76<p>Improvements to the current infrastructure are always very welcome and tend
77to be fairly straight-forward to implement. Here are some of the key areas that
78can use improvement...</p>
79
80</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]81
82<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]83<div class="doc_subsection">
84 <a name="glibc">Port glibc to LLVM</a>
85</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]86
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]87<div class="doc_text">
88
89<p>It would be very useful to <a
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]90href="http://www.gnu.org/software/libc/porting.html">port</a> <a
91href="http://www.gnu.org/software/glibc/">glibc</a> to LLVM. This would allow a
92variety of interprocedural algorithms to be much more effective in the face of
93library calls. The most important pieces to port are things like the string
94library and the <tt>stdio</tt> related functions... low-level system calls like
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]95'<tt>read</tt>' should stay unimplemented in LLVM.</p>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]96
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]97</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]98
99<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]100<div class="doc_subsection">
101 <a name="NightlyTest">Improving the Nightly Tester</a>
102</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]103
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]104<div class="doc_text">
105
106<p>The <a href="/testresults/">Nightly Tester</a> is a simple perl script
107(located in <tt>utils/NightlyTest.pl</tt>) which runs every night to generate a
108daily report. It could use the following improvements:</p>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]109
110<ol>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]111<li>Regression tests - We should run the regression tests in addition to the
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]112 program tests...</li>
113</ol>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]114
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]115</div>
Chris Lattner1b360042003-02-07 19:25:35 +0000[diff] [blame]116
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]117<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]118<div class="doc_subsection">
119 <a name="programs">Compile programs with the LLVM Compiler</a>
120</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]121
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]122<div class="doc_text">
123
124<p>We are always looking for new testcases and benchmarks for use with LLVM. In
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]125particular, it is useful to try compiling your favorite C source code with LLVM.
126If it doesn't compile, try to figure out why or report it to the <a
127href="http://mail.cs.uiuc.edu/pipermail/llvmbugs/">llvm-bugs</a> list. If you
128get the program to compile, it would be extremely useful to convert the build
129system to be compatible with the LLVM Programs testsuite so that we can check it
Chris Lattner1b360042003-02-07 19:25:35 +0000[diff] [blame]130into CVS and the automated tester can use it to track progress of the
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]131compiler.</p>
Chris Lattner1b360042003-02-07 19:25:35 +0000[diff] [blame]132
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]133<p>When testing a code, try running it with a variety of optimizations, and with
134all the back-ends: CBE, llc, and lli.</p>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]135
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]136</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]137
138<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]139<div class="doc_subsection">
140 <a name="llvm_ir">Extend the LLVM intermediate representation</a>
141</div>
142
143<div class="doc_text">
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]144
145<ol>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]146<li>Add support for platform-independent prefetch support. The GCC <a
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]147 href="http://gcc.gnu.org/projects/prefetch.html">prefetch project</a> page
148 has a good survey of the prefetching capabilities of a variety of modern
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]149 processors.</li>
150
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]151</ol>
152
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]153</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]154
155<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]156<div class="doc_subsection">
157 <a name="misc_imp">Miscellaneous Improvements</a>
158</div>
159
160<div class="doc_text">
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]161
162<ol>
Chris Lattner8bde96f2003-05-06 20:47:11 +0000[diff] [blame]163<li>Someone needs to look into getting the <tt>ranlib</tt> tool to index LLVM
164 bytecode files, so that linking in .a files is not hideously slow. They
165 would also then have to implement the reader for this index in
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]166 <tt>gccld</tt>.</li>
Chris Lattner8bde96f2003-05-06 20:47:11 +0000[diff] [blame]167
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]168<li>Rework the PassManager to be more flexible</li>
169
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]170<li>Some transformations and analyses only work on reducible flow graphs. It
171would be nice to have a transformation which could be "required" by these passes
172which makes irreducible graphs reducible. This can easily be accomplished
173through code duplication. See <a
174href="http://citeseer.nj.nec.com/janssen97making.html">Making Graphs Reducible
175with Controlled Node Splitting</a> and perhaps <a
176href="http://doi.acm.org/10.1145/262004.262005">Nesting of Reducible and
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]177Irreducible Loops</a>.</li>
178
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]179</ol>
180
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]181</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]182
183<!-- *********************************************************************** -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]184<div class="doc_section">
185 <a name="new">Adding new capabilities to LLVM</a>
186</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]187<!-- *********************************************************************** -->
188
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]189<div class="doc_text">
190
Brian Gaeke2d261352004-04-07 15:31:23 +0000[diff] [blame]191<p>Sometimes creating new things is more fun than improving existing things.
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]192These projects tend to be more involved and perhaps require more work, but can
193also be very rewarding.</p>
194
195</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]196
197<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]198<div class="doc_subsection">
199 <a name="pointeranalysis">Pointer and Alias Analysis</a>
200</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]201
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]202<div class="doc_text">
203
204<p>We have a <a href="AliasAnalysis.html">strong base for development</a> of
205both pointer analysis based optimizations as well as pointer analyses
206themselves. It seems natural to want to take advantage of this...</p>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]207
208<ol>
209<li>Implement a flow-sensitive context-sensitive alias analysis algorithm<br>
210 - Pick one of the somewhat efficient algorithms, but strive for maximum
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]211 precision</li>
212
Chris Lattnerf56703b2003-06-25 14:31:06 +0000[diff] [blame]213<li>Implement a flow-sensitive context-insensitive alias analysis algorithm<br>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]214 - Just an efficient local algorithm perhaps?</li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]215
216<li>Implement an interface to update analyses in response to common code motion
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]217 transformations</li>
218
219<li>Implement alias-analysis-based optimizations:
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]220 <ul>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]221 <li>Dead store elimination</li>
222 <li>...</li>
223 </ul></li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]224</ol>
225
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]226</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]227
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]228<!-- ======================================================================= -->
229<div class="doc_subsection">
230 <a name="profileguided">Profile Guided Optimization</a>
231</div>
232
233<div class="doc_text">
234
Chris Lattnera5422e42004-03-08 22:29:35 +0000[diff] [blame]235<p>We now have a unified infrastructure for writing profile-guided
236transformations, which will work either at offline-compile-time or in the JIT,
237but we don't have many transformations. We would welcome new profile-guided
238transformations as well as improvements to the current profiling system.
239</p>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]240
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]241<p>Ideas for profile guided transformations:</p>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]242
243<ol>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]244<li>Superblock formation (with many optimizations)</li>
245<li>Loop unrolling/peeling</li>
246<li>Profile directed inlining</li>
247<li>Code layout</li>
248<li>...</li>
249</ol>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]250
Chris Lattnera5422e42004-03-08 22:29:35 +0000[diff] [blame]251<p>Improvements to the existing support:</p>
252
253<ol>
254<li>The current block and edge profiling code that gets inserted is very simple
255and inefficient. Through the use of control-dependence information, many fewer
256counters could be inserted into the code. Also, if the execution count of a
257loop is known to be a compile-time or runtime constant, all of the counters in
258the loop could be avoided.</li>
259
260<li>You could implement one of the "static profiling" algorithms which analyze a
261piece of code an make educated guesses about the relative execution frequencies
262of various parts of the code.</li>
263
264<li>You could add path profiling support, or adapt the existing LLVM path
265profiling code to work with the generic profiling interfaces.</li>
266</ol>
267
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]268</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]269
270<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]271<div class="doc_subsection">
272 <a name="xforms">New Transformations and Analyses</a>
273</div>
274
275<div class="doc_text">
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]276
277<ol>
278<li>Implement a Dependence Analysis Infrastructure<br>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]279 - Design some way to represent and query dep analysis</li>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]280<li>Implement a strength reduction pass</li>
281<li>Value range propagation pass</li>
282<li>Implement an unswitching pass</li>
283<li>Write a loop unroller, with a simple heuristic for when to unroll</li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]284</ol>
285
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]286</div>
287
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]288<!-- ======================================================================= -->
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]289<div class="doc_section">
290 <a name="x86be">X86 Back-end Improvements</a>
291</div>
292
293<div class="doc_text">
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]294
295<ol>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]296<li>Implement a better instruction selector</li>
Chris Lattner99df25f2003-04-28 03:42:31 +0000[diff] [blame]297<li>Implement support for the "switch" instruction without requiring the
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]298 lower-switches pass.</li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]299</ol>
300
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]301</div>
302
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]303<!-- ======================================================================= -->
Misha Brukman3497ae92004-03-11 21:26:29 +0000[diff] [blame]304<div class="doc_section">
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]305 <a name="misc_new">Miscellaneous Additions</a>
306</div>
307
308<div class="doc_text">
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]309
310<ol>
Brian Gaeke2d261352004-04-07 15:31:23 +0000[diff] [blame]311<li>Port the <A HREF="http://www-sop.inria.fr/mimosa/fp/Bigloo/">Bigloo</A>
312Scheme compiler, from Manuel Serrano at INRIA Sophia-Antipolis, to
313output LLVM bytecode. It seems that it can already output .NET
314bytecode, JVM bytecode, and C, so LLVM would ostensibly be another good
315candidate.</li>
316<li>Write a new frontend for some other language (Java? OCaml? Forth?)</li>
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]317<li>Write a new backend for a target (IA64? MIPS? MMIX?)</li>
Misha Brukmandec8cfb2004-04-16 16:55:30 +0000[diff] [blame^]318<li>Write a disassembler for machine code that would use TableGen to output
319<tt>MachineInstr</tt>s for transformations, optimizations, etc.</li>
Brian Gaekea2245a32004-03-10 19:15:50 +0000[diff] [blame]320<li>Random test vector generator: Use a C grammar to generate random C code;
321run it through llvm-gcc, then run a random set of passes on it using opt.
322Try to crash opt. When opt crashes, use bugpoint to reduce the test case and
323mail the result to yourself. Repeat ad infinitum.</li>
Misha Brukman1a49bc82004-03-10 19:22:29 +0000[diff] [blame]324<li>Design a simple, recognizable logo.</li>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]325</ol>
326
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]327</div>
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]328
329<!-- *********************************************************************** -->
Chris Lattnerb7131092003-02-06 19:39:21 +0000[diff] [blame]330
Misha Brukman076438b2003-11-14 17:34:38 +0000[diff] [blame]331<hr>
332<div class="doc_footer">
333 <address><a href="mailto:sabre@nondot.org">Chris Lattner</a></address>
334 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a>
335 <br>
336 Last modified: $Date$
337</div>
338
339</body>
340</html>