1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>LLVM Testing Infrastructure Guide</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
10 <div class="doc_title">
11 LLVM Testing Infrastructure Guide
15 <li><a href="#overview">Overview</a></li>
16 <li><a href="#requirements">Requirements</a></li>
17 <li><a href="#org">LLVM testing infrastructure organization</a>
19 <li><a href="#dejagnu">DejaGNU tests</a></li>
20 <li><a href="#testsuite">Test suite</a></li>
23 <li><a href="#quick">Quick start</a>
25 <li><a href="#quickdejagnu">DejaGNU tests</a></li>
26 <li><a href="#quicktestsuite">Test suite</a></li>
29 <li><a href="#dgstructure">DejaGNU structure</a>
31 <li><a href="#dgcustom">Writing new DejaGNU tests</a></li>
32 <li><a href="#dgvars">Variables and substitutions</a></li>
33 <li><a href="#dgfeatures">Other features</a></li>
36 <li><a href="#testsuitestructure">Test suite structure</a></li>
37 <li><a href="#testsuiterun">Running the test suite</a>
39 <li><a href="#testsuiteexternal">Configuring external tests</a></li>
40 <li><a href="#testsuitetests">Running different tests</a></li>
41 <li><a href="#testsuiteoutput">Generating test output</a></li>
42 <li><a href="#testsuitecustom">Writing custom tests for llvm-test</a></li>
45 <li><a href="#nightly">Running the nightly tester</a></li>
48 <div class="doc_author">
49 <p>Written by John T. Criswell, <a
50 href="http://llvm.x10sys.com/rspencer">Reid Spencer</a>, and Tanya Lattner</p>
53 <!--=========================================================================-->
54 <div class="doc_section"><a name="overview">Overview</a></div>
55 <!--=========================================================================-->
57 <div class="doc_text">
59 <p>This document is the reference manual for the LLVM testing infrastructure. It documents
60 the structure of the LLVM testing infrastructure, the tools needed to use it,
61 and how to add and run tests.</p>
65 <!--=========================================================================-->
66 <div class="doc_section"><a name="requirements">Requirements</a></div>
67 <!--=========================================================================-->
69 <div class="doc_text">
71 <p>In order to use the LLVM testing infrastructure, you will need all of the software
72 required to build LLVM, plus the following:</p>
75 <dt><a href="http://www.gnu.org/software/dejagnu/">DejaGNU</a></dt>
76 <dd>The Feature and Regressions tests are organized and run by DejaGNU.</dd>
77 <dt><a href="http://expect.nist.gov/">Expect</a></dt>
78 <dd>Expect is required by DejaGNU.</dd>
79 <dt><a href="http://www.tcl.tk/software/tcltk/">tcl</a></dt>
80 <dd>Tcl is required by DejaGNU. </dd>
82 <dt><a href="http://www.netlib.org/f2c">F2C</a></dt>
83 <dd>For now, LLVM does not have a Fortran front-end, but using F2C, we can run
84 Fortran benchmarks. F2C support must be enabled via <tt>configure</tt> if not
85 installed in a standard place. F2C requires three items: the <tt>f2c</tt>
86 executable, <tt>f2c.h</tt> to compile the generated code, and <tt>libf2c.a</tt>
87 to link generated code. By default, given an F2C directory <tt>$DIR</tt>, the
88 configure script will search <tt>$DIR/bin</tt> for <tt>f2c</tt>,
89 <tt>$DIR/include</tt> for <tt>f2c.h</tt>, and <tt>$DIR/lib</tt> for
90 <tt>libf2c.a</tt>. The default <tt>$DIR</tt> values are: <tt>/usr</tt>,
91 <tt>/usr/local</tt>, <tt>/sw</tt>, and <tt>/opt</tt>. If you installed F2C in a
92 different location, you must tell <tt>configure</tt>:
95 <li><tt>./configure --with-f2c=$DIR</tt><br>
96 This will specify a new <tt>$DIR</tt> for the above-described search
97 process. This will only work if the binary, header, and library are in their
98 respective subdirectories of <tt>$DIR</tt>.</li>
100 <li><tt>./configure --with-f2c-bin=/binary/path --with-f2c-inc=/include/path
101 --with-f2c-lib=/lib/path</tt><br>
102 This allows you to specify the F2C components separately. Note: if you choose
103 this route, you MUST specify all three components, and you need to only specify
104 <em>directories</em> where the files are located; do NOT include the
105 filenames themselves on the <tt>configure</tt> line.</li>
109 <p>Darwin (Mac OS X) developers can simplify the installation of Expect and tcl
110 by using fink. <tt>fink install expect</tt> will install both. Alternatively,
111 Darwinports users can use <tt>sudo port install expect</tt> to install Expect
116 <!--=========================================================================-->
117 <div class="doc_section"><a name="org">LLVM testing infrastructure organization</a></div>
118 <!--=========================================================================-->
120 <div class="doc_text">
122 <p>The LLVM testing infrastructure contains two major categories of tests: code
123 fragments and whole programs. Code fragments are referred to as the "DejaGNU
124 tests" and are in the <tt>llvm</tt> module in subversion under the
125 <tt>llvm/test</tt> directory. The whole programs tests are referred to as the
126 "Test suite" and are in the <tt>test-suite</tt> module in subversion.
131 <!-- _______________________________________________________________________ -->
132 <div class="doc_subsection"><a name="dejagnu">DejaGNU tests</a></div>
133 <!-- _______________________________________________________________________ -->
135 <div class="doc_text">
137 <p>Code fragments are small pieces of code that test a specific feature of LLVM
138 or trigger a specific bug in LLVM. They are usually written in LLVM assembly
139 language, but can be written in other languages if the test targets a particular
140 language front end. These tests are driven by the DejaGNU testing framework,
141 which is hidden behind a few simple makefiles.</p>
143 <p>These code fragments are not complete programs. The code generated from them is
144 never executed to determine correct behavior.</p>
146 <p>These code fragment tests are located in the <tt>llvm/test</tt>
149 <p>Typically when a bug is found in LLVM, a regression test containing
150 just enough code to reproduce the problem should be written and placed
151 somewhere underneath this directory. In most cases, this will be a small
152 piece of LLVM assembly language code, often distilled from an actual
153 application or benchmark.</p>
157 <!-- _______________________________________________________________________ -->
158 <div class="doc_subsection"><a name="testsuite">Test suite</a></div>
159 <!-- _______________________________________________________________________ -->
161 <div class="doc_text">
163 <p>The test suite contains whole programs, which are pieces of
164 code which can be compiled and linked into a stand-alone program that can be
165 executed. These programs are generally written in high level languages such as
166 C or C++, but sometimes they are written straight in LLVM assembly.</p>
168 <p>These programs are compiled and then executed using several different
169 methods (native compiler, LLVM C backend, LLVM JIT, LLVM native code generation,
170 etc). The output of these programs is compared to ensure that LLVM is compiling
171 the program correctly.</p>
173 <p>In addition to compiling and executing programs, whole program tests serve as
174 a way of benchmarking LLVM performance, both in terms of the efficiency of the
175 programs generated as well as the speed with which LLVM compiles, optimizes, and
178 <p>The test-suite is located in the <tt>test-suite</tt> Subversion module.</p>
182 <!--=========================================================================-->
183 <div class="doc_section"><a name="quick">Quick start</a></div>
184 <!--=========================================================================-->
186 <div class="doc_text">
188 <p>The tests are located in two separate Subversion modules. The
189 DejaGNU tests are in the main "llvm" module under the directory
190 <tt>llvm/test</tt> (so you get these tests for free with the main llvm tree).
191 The more comprehensive test suite that includes whole
192 programs in C and C++ is in the <tt>test-suite</tt> module. This module should
193 be checked out to the <tt>llvm/projects</tt> directory (don't use another name
194 then the default "test-suite", for then the test suite will be run every time
195 you run <tt>make</tt> in the main <tt>llvm</tt> directory).
196 When you <tt>configure</tt> the <tt>llvm</tt> module,
197 the <tt>test-suite</tt> directory will be automatically configured.
198 Alternatively, you can configure the <tt>test-suite</tt> module manually.</p>
200 <!-- _______________________________________________________________________ -->
201 <div class="doc_subsection"><a name="quickdejagnu">DejaGNU tests</a></div>
202 <!-- _______________________________________________________________________ -->
203 <p>To run all of the simple tests in LLVM using DejaGNU, use the master Makefile
204 in the <tt>llvm/test</tt> directory:</p>
206 <div class="doc_code">
214 <div class="doc_code">
220 <p>To run only a subdirectory of tests in <tt>llvm/test</tt> using DejaGNU (ie.
221 Transforms), just set the TESTSUITE variable to the path of the
222 subdirectory (relative to <tt>llvm/test</tt>):</p>
224 <div class="doc_code">
226 % gmake TESTSUITE=Transforms check
230 <p><b>Note: If you are running the tests with <tt>objdir != subdir</tt>, you
231 must have run the complete testsuite before you can specify a
232 subdirectory.</b></p>
234 <p>To run only a single test, set <tt>TESTONE</tt> to its path (relative to
235 <tt>llvm/test</tt>) and make the <tt>check-one</tt> target:</p>
237 <div class="doc_code">
239 % gmake TESTONE=Feature/basictest.ll check-one
243 <!-- _______________________________________________________________________ -->
244 <div class="doc_subsection"><a name="quicktestsuite">Test suite</a></div>
245 <!-- _______________________________________________________________________ -->
247 <p>To run the comprehensive test suite (tests that compile and execute whole
248 programs), first checkout and setup the <tt>test-suite</tt> module:</p>
250 <div class="doc_code">
253 % svn co http://llvm.org/svn/llvm-project/test-suite/trunk test-suite
255 % ./configure --with-llvmgccdir=$LLVM_GCC_DIR
257 <p>where <tt>$LLVM_GCC_DIR</tt> is the directory where you <em>installed</em>
258 llvm-gcc, not it's src or obj dir.</p>
261 <p>Then, run the entire test suite by running make in the <tt>test-suite</tt>
264 <div class="doc_code">
266 % cd projects/test-suite
271 <p>Usually, running the "nightly" set of tests is a good idea, and you can also
272 let it generate a report by running:</p>
274 <div class="doc_code">
276 % cd projects/test-suite
277 % gmake TEST=nightly report report.html
281 <p>Any of the above commands can also be run in a subdirectory of
282 <tt>projects/test-suite</tt> to run the specified test only on the programs in
283 that subdirectory.</p>
287 <!--=========================================================================-->
288 <div class="doc_section"><a name="dgstructure">DejaGNU structure</a></div>
289 <!--=========================================================================-->
290 <div class="doc_text">
291 <p>The LLVM DejaGNU tests are driven by DejaGNU together with GNU Make and are
292 located in the <tt>llvm/test</tt> directory.
294 <p>This directory contains a large array of small tests
295 that exercise various features of LLVM and to ensure that regressions do not
296 occur. The directory is broken into several sub-directories, each focused on
297 a particular area of LLVM. A few of the important ones are:</p>
300 <li><tt>Analysis</tt>: checks Analysis passes.</li>
301 <li><tt>Archive</tt>: checks the Archive library.</li>
302 <li><tt>Assembler</tt>: checks Assembly reader/writer functionality.</li>
303 <li><tt>Bitcode</tt>: checks Bitcode reader/writer functionality.</li>
304 <li><tt>CodeGen</tt>: checks code generation and each target.</li>
305 <li><tt>Features</tt>: checks various features of the LLVM language.</li>
306 <li><tt>Linker</tt>: tests bitcode linking.</li>
307 <li><tt>Transforms</tt>: tests each of the scalar, IPO, and utility
308 transforms to ensure they make the right transformations.</li>
309 <li><tt>Verifier</tt>: tests the IR verifier.</li>
314 <!-- _______________________________________________________________________ -->
315 <div class="doc_subsection"><a name="dgcustom">Writing new DejaGNU tests</a></div>
316 <!-- _______________________________________________________________________ -->
317 <div class="doc_text">
318 <p>The DejaGNU structure is very simple, but does require some information to
319 be set. This information is gathered via <tt>configure</tt> and is written
320 to a file, <tt>site.exp</tt> in <tt>llvm/test</tt>. The <tt>llvm/test</tt>
321 Makefile does this work for you.</p>
323 <p>In order for DejaGNU to work, each directory of tests must have a
324 <tt>dg.exp</tt> file. DejaGNU looks for this file to determine how to run the
325 tests. This file is just a Tcl script and it can do anything you want, but
326 we've standardized it for the LLVM regression tests. If you're adding a
327 directory of tests, just copy <tt>dg.exp</tt> from another directory to get
328 running. The standard <tt>dg.exp</tt> simply loads a Tcl
329 library (<tt>test/lib/llvm.exp</tt>) and calls the <tt>llvm_runtests</tt>
330 function defined in that library with a list of file names to run. The names
331 are obtained by using Tcl's glob command. Any directory that contains only
332 directories does not need the <tt>dg.exp</tt> file.</p>
334 <p>The <tt>llvm-runtests</tt> function lookas at each file that is passed to
335 it and gathers any lines together that match "RUN:". This are the "RUN" lines
336 that specify how the test is to be run. So, each test script must contain
337 RUN lines if it is to do anything. If there are no RUN lines, the
338 <tt>llvm-runtests</tt> function will issue an error and the test will
341 <p>RUN lines are specified in the comments of the test program using the
342 keyword <tt>RUN</tt> followed by a colon, and lastly the command (pipeline)
343 to execute. Together, these lines form the "script" that
344 <tt>llvm-runtests</tt> executes to run the test case. The syntax of the
345 RUN lines is similar to a shell's syntax for pipelines including I/O
346 redirection and variable substitution. However, even though these lines
347 may <i>look</i> like a shell script, they are not. RUN lines are interpreted
348 directly by the Tcl <tt>exec</tt> command. They are never executed by a
349 shell. Consequently the syntax differs from normal shell script syntax in a
350 few ways. You can specify as many RUN lines as needed.</p>
352 <p>Each RUN line is executed on its own, distinct from other lines unless
353 its last character is <tt>\</tt>. This continuation character causes the RUN
354 line to be concatenated with the next one. In this way you can build up long
355 pipelines of commands without making huge line lengths. The lines ending in
356 <tt>\</tt> are concatenated until a RUN line that doesn't end in <tt>\</tt> is
357 found. This concatenated set of RUN lines then constitutes one execution.
358 Tcl will substitute variables and arrange for the pipeline to be executed. If
359 any process in the pipeline fails, the entire line (and test case) fails too.
362 <p> Below is an example of legal RUN lines in a <tt>.ll</tt> file:</p>
364 <div class="doc_code">
366 ; RUN: llvm-as < %s | llvm-dis > %t1
367 ; RUN: llvm-dis < %s.bc-13 > %t2
372 <p>As with a Unix shell, the RUN: lines permit pipelines and I/O redirection
373 to be used. However, the usage is slightly different than for Bash. To check
374 what's legal, see the documentation for the
375 <a href="http://www.tcl.tk/man/tcl8.5/TclCmd/exec.htm#M2">Tcl exec</a>
377 <a href="http://www.tcl.tk/man/tcl8.5/tutorial/Tcl26.html">tutorial</a>.
378 The major differences are:</p>
380 <li>You can't do <tt>2>&1</tt>. That will cause Tcl to write to a
381 file named <tt>&1</tt>. Usually this is done to get stderr to go through
382 a pipe. You can do that in tcl with <tt>|&</tt> so replace this idiom:
383 <tt>... 2>&1 | grep</tt> with <tt>... |& grep</tt></li>
384 <li>You can only redirect to a file, not to another descriptor and not from
385 a here document.</li>
386 <li>tcl supports redirecting to open files with the @ syntax but you
387 shouldn't use that here.</li>
390 <p>There are some quoting rules that you must pay attention to when writing
391 your RUN lines. In general nothing needs to be quoted. Tcl won't strip off any
392 ' or " so they will get passed to the invoked program. For example:</p>
394 <div class="doc_code">
396 ... | grep 'find this string'
400 <p>This will fail because the ' characters are passed to grep. This would
401 instruction grep to look for <tt>'find</tt> in the files <tt>this</tt> and
402 <tt>string'</tt>. To avoid this use curly braces to tell Tcl that it should
403 treat everything enclosed as one value. So our example would become:</p>
405 <div class="doc_code">
407 ... | grep {find this string}
411 <p>Additionally, the characters <tt>[</tt> and <tt>]</tt> are treated
412 specially by Tcl. They tell Tcl to interpret the content as a command to
413 execute. Since these characters are often used in regular expressions this can
414 have disastrous results and cause the entire test run in a directory to fail.
415 For example, a common idiom is to look for some basicblock number:</p>
417 <div class="doc_code">
423 <p>This, however, will cause Tcl to fail because its going to try to execute
424 a program named "2-8". Instead, what you want is this:</p>
426 <div class="doc_code">
428 ... | grep {bb\[2-8\]}
432 <p>Finally, if you need to pass the <tt>\</tt> character down to a program,
433 then it must be doubled. This is another Tcl special character. So, suppose
436 <div class="doc_code">
442 <p>This will fail to match what you want (a pointer to i32). First, the
443 <tt>'</tt> do not get stripped off. Second, the <tt>\</tt> gets stripped off
444 by Tcl so what grep sees is: <tt>'i32*'</tt>. That's not likely to match
445 anything. To resolve this you must use <tt>\\</tt> and the <tt>{}</tt>, like
448 <div class="doc_code">
456 <!-- _______________________________________________________________________ -->
457 <div class="doc_subsection"><a name="dgvars">Variables and substitutions</a></div>
458 <!-- _______________________________________________________________________ -->
459 <div class="doc_text">
460 <p>With a RUN line there are a number of substitutions that are permitted. In
461 general, any Tcl variable that is available in the <tt>substitute</tt>
462 function (in <tt>test/lib/llvm.exp</tt>) can be substituted into a RUN line.
463 To make a substitution just write the variable's name preceded by a $.
464 Additionally, for compatibility reasons with previous versions of the test
465 library, certain names can be accessed with an alternate syntax: a % prefix.
466 These alternates are deprecated and may go away in a future version.
468 <p>Here are the available variable names. The alternate syntax is listed in
471 <dl style="margin-left: 25px">
472 <dt><b>$test</b> (%s)</dt>
473 <dd>The full path to the test case's source. This is suitable for passing
474 on the command line as the input to an llvm tool.</dd>
476 <dt><b>$srcdir</b></dt>
477 <dd>The source directory from where the "<tt>make check</tt>" was run.</dd>
479 <dt><b>objdir</b></dt>
480 <dd>The object directory that corresponds to the <tt>$srcdir</tt>.</dd>
482 <dt><b>subdir</b></dt>
483 <dd>A partial path from the <tt>test</tt> directory that contains the
484 sub-directory that contains the test source being executed.</dd>
486 <dt><b>srcroot</b></dt>
487 <dd>The root directory of the LLVM src tree.</dd>
489 <dt><b>objroot</b></dt>
490 <dd>The root directory of the LLVM object tree. This could be the same
494 <dd>The path to the directory that contains the test case source. This is
495 for locating any supporting files that are not generated by the test, but
496 used by the test.</dd>
499 <dd>The path to a temporary file name that could be used for this test case.
500 The file name won't conflict with other test cases. You can append to it if
501 you need multiple temporaries. This is useful as the destination of some
502 redirected output.</dd>
504 <dt><b>llvmlibsdir</b> (%llvmlibsdir)</dt>
505 <dd>The directory where the LLVM libraries are located.</dd>
507 <dt><b>target_triplet</b> (%target_triplet)</dt>
508 <dd>The target triplet that corresponds to the current host machine (the one
509 running the test cases). This should probably be called "host".<dd>
511 <dt><b>prcontext</b> (%prcontext)</dt>
512 <dd>Path to the prcontext tcl script that prints some context around a
513 line that matches a pattern. This isn't strictly necessary as the test suite
514 is run with its PATH altered to include the test/Scripts directory where
515 the prcontext script is located. Note that this script is similar to
516 <tt>grep -C</tt> but you should use the <tt>prcontext</tt> script because
517 not all platforms support <tt>grep -C</tt>.</dd>
519 <dt><b>llvmgcc</b> (%llvmgcc)</dt>
520 <dd>The full path to the <tt>llvm-gcc</tt> executable as specified in the
521 configured LLVM environment</dd>
523 <dt><b>llvmgxx</b> (%llvmgxx)</dt>
524 <dd>The full path to the <tt>llvm-gxx</tt> executable as specified in the
525 configured LLVM environment</dd>
527 <dt><b>llvmgcc_version</b> (%llvmgcc_version)</dt>
528 <dd>The full version number of the <tt>llvm-gcc</tt> executable.</dd>
530 <dt><b>llvmgccmajvers</b> (%llvmgccmajvers)</dt>
531 <dd>The major version number of the <tt>llvm-gcc</tt> executable.</dd>
533 <dt><b>gccpath</b></dt>
534 <dd>The full path to the C compiler used to <i>build </i> LLVM. Note that
535 this might not be gcc.</dd>
537 <dt><b>gxxpath</b></dt>
538 <dd>The full path to the C++ compiler used to <i>build </i> LLVM. Note that
539 this might not be g++.</dd>
541 <dt><b>compile_c</b> (%compile_c)</dt>
542 <dd>The full command line used to compile LLVM C source code. This has all
543 the configured -I, -D and optimization options.</dd>
545 <dt><b>compile_cxx</b> (%compile_cxx)</dt>
546 <dd>The full command used to compile LLVM C++ source code. This has
547 all the configured -I, -D and optimization options.</dd>
549 <dt><b>link</b> (%link)</dt>
550 <dd>This full link command used to link LLVM executables. This has all the
551 configured -I, -L and -l options.</dd>
553 <dt><b>shlibext</b> (%shlibext)</dt>
554 <dd>The suffix for the host platforms share library (dll) files. This
555 includes the period as the first character.</dd>
557 <p>To add more variables, two things need to be changed. First, add a line in
558 the <tt>test/Makefile</tt> that creates the <tt>site.exp</tt> file. This will
559 "set" the variable as a global in the site.exp file. Second, in the
560 <tt>test/lib/llvm.exp</tt> file, in the substitute proc, add the variable name
561 to the list of "global" declarations at the beginning of the proc. That's it,
562 the variable can then be used in test scripts.</p>
565 <!-- _______________________________________________________________________ -->
566 <div class="doc_subsection"><a name="dgfeatures">Other Features</a></div>
567 <!-- _______________________________________________________________________ -->
568 <div class="doc_text">
569 <p>To make RUN line writing easier, there are several shell scripts located
570 in the <tt>llvm/test/Scripts</tt> directory. This directory is in the PATH
571 when running tests, so you can just call these scripts using their name. For
574 <dt><b>ignore</b></dt>
575 <dd>This script runs its arguments and then always returns 0. This is useful
576 in cases where the test needs to cause a tool to generate an error (e.g. to
577 check the error output). However, any program in a pipeline that returns a
578 non-zero result will cause the test to fail. This script overcomes that
579 issue and nicely documents that the test case is purposefully ignoring the
580 result code of the tool</dd>
583 <dd>This script runs its arguments and then inverts the result code from
584 it. Zero result codes become 1. Non-zero result codes become 0. This is
585 useful to invert the result of a grep. For example "not grep X" means
586 succeed only if you don't find X in the input.</dd>
589 <p>Sometimes it is necessary to mark a test case as "expected fail" or XFAIL.
590 You can easily mark a test as XFAIL just by including <tt>XFAIL: </tt> on a
591 line near the top of the file. This signals that the test case should succeed
592 if the test fails. Such test cases are counted separately by DejaGnu. To
593 specify an expected fail, use the XFAIL keyword in the comments of the test
594 program followed by a colon and one or more regular expressions (separated by
595 a comma). The regular expressions allow you to XFAIL the test conditionally
596 by host platform. The regular expressions following the : are matched against
597 the target triplet or llvmgcc version number for the host machine. If there is
598 a match, the test is expected to fail. If not, the test is expected to
599 succeed. To XFAIL everywhere just specify <tt>XFAIL: *</tt>. When matching
600 the llvm-gcc version, you can specify the major (e.g. 3) or full version
601 (i.e. 3.4) number. Here is an example of an <tt>XFAIL</tt> line:</p>
603 <div class="doc_code">
605 ; XFAIL: darwin,sun,llvmgcc4
609 <p>To make the output more useful, the <tt>llvm_runtest</tt> function wil
610 scan the lines of the test case for ones that contain a pattern that matches
611 PR[0-9]+. This is the syntax for specifying a PR (Problem Report) number that
612 is related to the test case. The number after "PR" specifies the LLVM bugzilla
613 number. When a PR number is specified, it will be used in the pass/fail
614 reporting. This is useful to quickly get some context when a test fails.</p>
616 <p>Finally, any line that contains "END." will cause the special
617 interpretation of lines to terminate. This is generally done right after the
618 last RUN: line. This has two side effects: (a) it prevents special
619 interpretation of lines that are part of the test program, not the
620 instructions to the test case, and (b) it speeds things up for really big test
621 cases by avoiding interpretation of the remainder of the file.</p>
625 <!--=========================================================================-->
626 <div class="doc_section"><a name="testsuitestructure">Test suite
628 <!--=========================================================================-->
630 <div class="doc_text">
632 <p>The <tt>test-suite</tt> module contains a number of programs that can be compiled
633 with LLVM and executed. These programs are compiled using the native compiler
634 and various LLVM backends. The output from the program compiled with the
635 native compiler is assumed correct; the results from the other programs are
636 compared to the native program output and pass if they match.</p>
638 <p>When executing tests, it is usually a good idea to start out with a subset of
639 the available tests or programs. This makes test run times smaller at first and
640 later on this is useful to investigate individual test failures. To run some
641 test only on a subset of programs, simply change directory to the programs you
642 want tested and run <tt>gmake</tt> there. Alternatively, you can run a different
643 test using the <tt>TEST</tt> variable to change what tests or run on the
644 selected programs (see below for more info).</p>
646 <p>In addition for testing correctness, the <tt>llvm-test</tt> directory also
647 performs timing tests of various LLVM optimizations. It also records
648 compilation times for the compilers and the JIT. This information can be
649 used to compare the effectiveness of LLVM's optimizations and code
652 <p><tt>llvm-test</tt> tests are divided into three types of tests: MultiSource,
653 SingleSource, and External.</p>
656 <li><tt>llvm-test/SingleSource</tt>
657 <p>The SingleSource directory contains test programs that are only a single
658 source file in size. These are usually small benchmark programs or small
659 programs that calculate a particular value. Several such programs are grouped
660 together in each directory.</p></li>
662 <li><tt>llvm-test/MultiSource</tt>
663 <p>The MultiSource directory contains subdirectories which contain entire
664 programs with multiple source files. Large benchmarks and whole applications
667 <li><tt>llvm-test/External</tt>
668 <p>The External directory contains Makefiles for building code that is external
669 to (i.e., not distributed with) LLVM. The most prominent members of this
670 directory are the SPEC 95 and SPEC 2000 benchmark suites. The <tt>External</tt>
671 directory does not contain these actual tests,but only the Makefiles that know
672 how to properly compile these programs from somewhere else. The presence and
673 location of these external programs is configured by the llvm-test
674 <tt>configure</tt> script.</p></li>
677 <p>Each tree is then subdivided into several categories, including applications,
678 benchmarks, regression tests, code that is strange grammatically, etc. These
679 organizations should be relatively self explanatory.</p>
681 <p>Some tests are known to fail. Some are bugs that we have not fixed yet;
682 others are features that we haven't added yet (or may never add). In DejaGNU,
683 the result for such tests will be XFAIL (eXpected FAILure). In this way, you
684 can tell the difference between an expected and unexpected failure.</p>
686 <p>The tests in the test suite have no such feature at this time. If the
687 test passes, only warnings and other miscellaneous output will be generated. If
688 a test fails, a large <program> FAILED message will be displayed. This
689 will help you separate benign warnings from actual test failures.</p>
693 <!--=========================================================================-->
694 <div class="doc_section"><a name="testsuiterun">Running the test suite</a></div>
695 <!--=========================================================================-->
697 <div class="doc_text">
699 <p>First, all tests are executed within the LLVM object directory tree. They
700 <i>are not</i> executed inside of the LLVM source tree. This is because the
701 test suite creates temporary files during execution.</p>
703 <p>To run the test suite, you need to use the following steps:</p>
706 <li><tt>cd</tt> into the <tt>llvm/projects</tt> directory</li>
708 <li><p>Check out the <tt>test-suite</tt> module with:</p>
710 <div class="doc_code">
712 % svn co http://llvm.org/svn/llvm-project/test-suite/trunk test-suite
716 <p>This will get the test suite into <tt>llvm/projects/llvm-test</tt></p>
718 <li><p>Configure the test suite using llvm configure. This will automatically
719 configure test-suite. You must do it from the top level otherwise llvm-gcc
720 will not be set which is required to run llvm-test:</p>
721 <div class="doc_code">
723 % cd $LLVM_OBJ_ROOT ; $LLVM_SRC_ROOT/configure --with-llvmgccdir=$LLVM_GCC_DIR
726 <p>Note that that <tt>$LLVM_GCC_DIR</tt> is the directory where you
727 <em>installed</em> llvm-gcc, not its src or obj directory.</p>
730 <li><p>Change back to the <tt>llvm/projects/test-suite</tt> directory you created before
731 and run <tt>gmake</tt> (or just "<tt>make</tt>" on systems where GNU make is
732 the default, such as linux.</p></li>
734 <p>Note that the second and third steps only need to be done once. After you
735 have the suite checked out and configured, you don't need to do it again (unless
736 the test code or configure script changes).</p>
738 <!-- _______________________________________________________________________ -->
739 <div class="doc_subsection">
740 <a name="testsuiteexternal">Configuring external tests</a></div>
741 <!-- _______________________________________________________________________ -->
743 <div class="doc_text">
744 <p>Note, when configuring the <tt>test-suite</tt> module, you might want to
745 specify the following configuration options:</p>
747 <dt><i>--enable-spec2000</i>
748 <dt><i>--enable-spec2000=<<tt>directory</tt>></i>
750 Enable the use of SPEC2000 when testing LLVM. This is disabled by default
751 (unless <tt>configure</tt> finds SPEC2000 installed). By specifying
752 <tt>directory</tt>, you can tell configure where to find the SPEC2000
753 benchmarks. If <tt>directory</tt> is left unspecified, <tt>configure</tt>
754 uses the default value
755 <tt>/home/vadve/shared/benchmarks/speccpu2000/benchspec</tt>.
758 <dt><i>--enable-spec95</i>
759 <dt><i>--enable-spec95=<<tt>directory</tt>></i>
761 Enable the use of SPEC95 when testing LLVM. It is similar to the
762 <i>--enable-spec2000</i> option.
765 <dt><i>--enable-povray</i>
766 <dt><i>--enable-povray=<<tt>directory</tt>></i>
768 Enable the use of Povray as an external test. Versions of Povray written
769 in C should work. This option is similar to the <i>--enable-spec2000</i>
774 <!-- _______________________________________________________________________ -->
775 <div class="doc_subsection">
776 <a name="testsuitetests">Running different tests</a></div>
777 <!-- _______________________________________________________________________ -->
778 <div class="doc_text">
779 <p>In addition to the regular "whole program" tests, the <tt>test-suite</tt>
780 module also provides a mechanism for compiling the programs in different ways.
781 If the variable TEST is defined on the <tt>gmake</tt> command line, the test system will
782 include a Makefile named <tt>TEST.<value of TEST variable>.Makefile</tt>.
783 This Makefile can modify build rules to yield different results.</p>
785 <p>For example, the LLVM nightly tester uses <tt>TEST.nightly.Makefile</tt> to
786 create the nightly test reports. To run the nightly tests, run <tt>gmake
787 TEST=nightly</tt>.</p>
789 <p>There are several TEST Makefiles available in the tree. Some of them are
790 designed for internal LLVM research and will not work outside of the LLVM
791 research group. They may still be valuable, however, as a guide to writing your
792 own TEST Makefile for any optimization or analysis passes that you develop with
797 <!-- _______________________________________________________________________ -->
798 <div class="doc_subsection">
799 <a name="testsuiteoutput">Generating test output</a></div>
800 <!-- _______________________________________________________________________ -->
801 <div class="doc_text">
802 <p>There are a number of ways to run the tests and generate output. The most
803 simple one is simply running <tt>gmake</tt> with no arguments. This will
804 compile and run all programs in the tree using a number of different methods
805 and compare results. Any failures are reported in the output, but are likely
806 drowned in the other output. Passes are not reported explicitely.</p>
808 <p>Somewhat better is running <tt>gmake TEST=sometest test</tt>, which runs
809 the specified test and usually adds per-program summaries to the output
810 (depending on which sometest you use). For example, the <tt>nightly</tt> test
811 explicitely outputs TEST-PASS or TEST-FAIL for every test after each program.
812 Though these lines are still drowned in the output, it's easy to grep the
813 output logs in the Output directories.</p>
815 <p>Even better are the <tt>report</tt> and <tt>report.format</tt> targets
816 (where <tt>format</tt> is one of <tt>html</tt>, <tt>csv</tt>, <tt>text</tt> or
817 <tt>graphs</tt>). The exact contents of the report are dependent on which
818 <tt>TEST</tt> you are running, but the text results are always shown at the
819 end of the run and the results are always stored in the
820 <tt>report.<type>.format</tt> file (when running with
821 <tt>TEST=<type></tt>).
823 The <tt>report</tt> also generate a file called
824 <tt>report.<type>.raw.out</tt> containing the output of the entire test
828 <!-- _______________________________________________________________________ -->
829 <div class="doc_subsection">
830 <a name="testsuitecustom">Writing custom tests for the test suite</a></div>
831 <!-- _______________________________________________________________________ -->
833 <div class="doc_text">
835 <p>Assuming you can run the test suite, (e.g. "<tt>gmake TEST=nightly report</tt>"
836 should work), it is really easy to run optimizations or code generator
837 components against every program in the tree, collecting statistics or running
838 custom checks for correctness. At base, this is how the nightly tester works,
839 it's just one example of a general framework.</p>
841 <p>Lets say that you have an LLVM optimization pass, and you want to see how
842 many times it triggers. First thing you should do is add an LLVM
843 <a href="ProgrammersManual.html#Statistic">statistic</a> to your pass, which
844 will tally counts of things you care about.</p>
846 <p>Following this, you can set up a test and a report that collects these and
847 formats them for easy viewing. This consists of two files, an
848 "<tt>test-suite/TEST.XXX.Makefile</tt>" fragment (where XXX is the name of your
849 test) and an "<tt>llvm-test/TEST.XXX.report</tt>" file that indicates how to
850 format the output into a table. There are many example reports of various
851 levels of sophistication included with the test suite, and the framework is very
854 <p>If you are interested in testing an optimization pass, check out the
855 "libcalls" test as an example. It can be run like this:<p>
857 <div class="doc_code">
859 % cd llvm/projects/test-suite/MultiSource/Benchmarks # or some other level
860 % make TEST=libcalls report
864 <p>This will do a bunch of stuff, then eventually print a table like this:</p>
866 <div class="doc_code">
868 Name | total | #exit |
870 FreeBench/analyzer/analyzer | 51 | 6 |
871 FreeBench/fourinarow/fourinarow | 1 | 1 |
872 FreeBench/neural/neural | 19 | 9 |
873 FreeBench/pifft/pifft | 5 | 3 |
874 MallocBench/cfrac/cfrac | 1 | * |
875 MallocBench/espresso/espresso | 52 | 12 |
876 MallocBench/gs/gs | 4 | * |
877 Prolangs-C/TimberWolfMC/timberwolfmc | 302 | * |
878 Prolangs-C/agrep/agrep | 33 | 12 |
879 Prolangs-C/allroots/allroots | * | * |
880 Prolangs-C/assembler/assembler | 47 | * |
881 Prolangs-C/bison/mybison | 74 | * |
886 <p>This basically is grepping the -stats output and displaying it in a table.
887 You can also use the "TEST=libcalls report.html" target to get the table in HTML
888 form, similarly for report.csv and report.tex.</p>
890 <p>The source for this is in test-suite/TEST.libcalls.*. The format is pretty
891 simple: the Makefile indicates how to run the test (in this case,
892 "<tt>opt -simplify-libcalls -stats</tt>"), and the report contains one line for
893 each column of the output. The first value is the header for the column and the
894 second is the regex to grep the output of the command for. There are lots of
895 example reports that can do fancy stuff.</p>
900 <!--=========================================================================-->
901 <div class="doc_section"><a name="nightly">Running the nightly tester</a></div>
902 <!--=========================================================================-->
904 <div class="doc_text">
907 The <a href="http://llvm.org/nightlytest/">LLVM Nightly Testers</a>
908 automatically check out an LLVM tree, build it, run the "nightly"
909 program test (described above), run all of the DejaGNU tests,
910 delete the checked out tree, and then submit the results to
911 <a href="http://llvm.org/nightlytest/">http://llvm.org/nightlytest/</a>.
912 After test results are submitted to
913 <a href="http://llvm.org/nightlytest/">http://llvm.org/nightlytest/</a>,
914 they are processed and displayed on the tests page. An email to
915 <a href="http://lists.cs.uiuc.edu/pipermail/llvm-testresults/">
916 llvm-testresults@cs.uiuc.edu</a> summarizing the results is also generated.
917 This testing scheme is designed to ensure that programs don't break as well
918 as keep track of LLVM's progress over time.</p>
920 <p>If you'd like to set up an instance of the nightly tester to run on your
921 machine, take a look at the comments at the top of the
922 <tt>utils/NewNightlyTest.pl</tt> file. If you decide to set up a nightly tester
923 please choose a unique nickname and invoke <tt>utils/NewNightlyTest.pl</tt>
924 with the "-nickname [yournickname]" command line option.
926 <p>You can create a shell script to encapsulate the running of the script.
927 The optimized x86 Linux nightly test is run from just such a script:</p>
929 <div class="doc_code">
932 BASE=/proj/work/llvm/nightlytest
933 export BUILDDIR=$BASE/build
934 export WEBDIR=$BASE/testresults
935 export LLVMGCCDIR=/proj/work/llvm/cfrontend/install
936 export PATH=/proj/install/bin:$LLVMGCCDIR/bin:$PATH
937 export LD_LIBRARY_PATH=/proj/install/lib
939 cp /proj/work/llvm/llvm/utils/NewNightlyTest.pl .
940 nice ./NewNightlyTest.pl -nice -release -verbose -parallel -enable-linscan \
941 -nickname NightlyTester -noexternals > output.log 2>&1
945 <p>It is also possible to specify the the location your nightly test results
946 are submitted. You can do this by passing the command line option
947 "-submit-server [server_address]" and "-submit-script [script_on_server]" to
948 <tt>utils/NewNightlyTest.pl</tt>. For example, to submit to the llvm.org
949 nightly test results page, you would invoke the nightly test script with
950 "-submit-server llvm.org -submit-script /nightlytest/NightlyTestAccept.cgi".
951 If these options are not specified, the nightly test script sends the results
952 to the llvm.org nightly test results page.</p>
954 <p>Take a look at the <tt>NewNightlyTest.pl</tt> file to see what all of the
955 flags and strings do. If you start running the nightly tests, please let us
960 <!-- *********************************************************************** -->
964 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
965 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
966 <a href="http://validator.w3.org/check/referer"><img
967 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
969 John T. Criswell, Reid Spencer, and Tanya Lattner<br>
970 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
971 Last modified: $Date$