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="#regressiontests">Regression tests</a></li>
20 <li><a href="#testsuite">Test suite</a></li>
21 <li><a href="#debuginfotests">Debugging Information tests</a></li>
24 <li><a href="#quick">Quick start</a>
26 <li><a href="#quickregressiontests">Regression tests</a></li>
27 <li><a href="#quicktestsuite">Test suite</a></li>
28 <li><a href="#quickdebuginfotests">Debugging Information tests</a></li>
31 <li><a href="#rtstructure">Regression test structure</a>
33 <li><a href="#rtcustom">Writing new regression tests</a></li>
34 <li><a href="#FileCheck">The FileCheck utility</a></li>
35 <li><a href="#rtvars">Variables and substitutions</a></li>
36 <li><a href="#rtfeatures">Other features</a></li>
39 <li><a href="#testsuitestructure">Test suite structure</a></li>
40 <li><a href="#testsuiterun">Running the test suite</a>
42 <li><a href="#testsuiteexternal">Configuring External Tests</a></li>
43 <li><a href="#testsuitetests">Running different tests</a></li>
44 <li><a href="#testsuiteoutput">Generating test output</a></li>
45 <li><a href="#testsuitecustom">Writing custom tests for test-suite</a></li>
50 <div class="doc_author">
51 <p>Written by John T. Criswell, Daniel Dunbar, Reid Spencer, and Tanya Lattner</p>
54 <!--=========================================================================-->
55 <div class="doc_section"><a name="overview">Overview</a></div>
56 <!--=========================================================================-->
58 <div class="doc_text">
60 <p>This document is the reference manual for the LLVM testing infrastructure. It
61 documents the structure of the LLVM testing infrastructure, the tools needed to
62 use it, and how to add and run tests.</p>
66 <!--=========================================================================-->
67 <div class="doc_section"><a name="requirements">Requirements</a></div>
68 <!--=========================================================================-->
70 <div class="doc_text">
72 <p>In order to use the LLVM testing infrastructure, you will need all of the
73 software required to build LLVM, as well
74 as <a href="http://python.org">Python</a> 2.4 or later.</p>
78 <!--=========================================================================-->
79 <div class="doc_section"><a name="org">LLVM testing infrastructure organization</a></div>
80 <!--=========================================================================-->
82 <div class="doc_text">
84 <p>The LLVM testing infrastructure contains two major categories of tests:
85 regression tests and whole programs. The regression tests are contained inside
86 the LLVM repository itself under <tt>llvm/test</tt> and are expected to always
87 pass -- they should be run before every commit. The whole programs tests are
88 referred to as the "LLVM test suite" and are in the <tt>test-suite</tt> module
94 <!-- _______________________________________________________________________ -->
95 <div class="doc_subsection"><a name="regressiontests">Regression tests</a></div>
96 <!-- _______________________________________________________________________ -->
98 <div class="doc_text">
100 <p>The regression tests are small pieces of code that test a specific feature of
101 LLVM or trigger a specific bug in LLVM. They are usually written in LLVM
102 assembly language, but can be written in other languages if the test targets a
103 particular language front end (and the appropriate <tt>--with-llvmgcc</tt>
104 options were used at <tt>configure</tt> time of the <tt>llvm</tt> module). These
105 tests are driven by the 'lit' testing tool, which is part of LLVM.</p>
107 <p>These code fragments are not complete programs. The code generated
108 from them is never executed to determine correct behavior.</p>
110 <p>These code fragment tests are located in the <tt>llvm/test</tt>
113 <p>Typically when a bug is found in LLVM, a regression test containing
114 just enough code to reproduce the problem should be written and placed
115 somewhere underneath this directory. In most cases, this will be a small
116 piece of LLVM assembly language code, often distilled from an actual
117 application or benchmark.</p>
121 <!-- _______________________________________________________________________ -->
122 <div class="doc_subsection"><a name="testsuite">Test suite</a></div>
123 <!-- _______________________________________________________________________ -->
125 <div class="doc_text">
127 <p>The test suite contains whole programs, which are pieces of
128 code which can be compiled and linked into a stand-alone program that can be
129 executed. These programs are generally written in high level languages such as
130 C or C++, but sometimes they are written straight in LLVM assembly.</p>
132 <p>These programs are compiled and then executed using several different
133 methods (native compiler, LLVM C backend, LLVM JIT, LLVM native code generation,
134 etc). The output of these programs is compared to ensure that LLVM is compiling
135 the program correctly.</p>
137 <p>In addition to compiling and executing programs, whole program tests serve as
138 a way of benchmarking LLVM performance, both in terms of the efficiency of the
139 programs generated as well as the speed with which LLVM compiles, optimizes, and
142 <p>The test-suite is located in the <tt>test-suite</tt> Subversion module.</p>
146 <!-- _______________________________________________________________________ -->
147 <div class="doc_subsection"><a name="debuginfotests">Debugging Information
149 <!-- _______________________________________________________________________ -->
151 <div class="doc_text">
153 <p>The test suite contains tests to check quality of debugging information.
154 The test are written in C based languages or in LLVM assembly language. </p>
156 <p>These tests are compiled and run under a debugger. The debugger output
157 is checked to validate of debugging information. See README.txt in the
158 test suite for more information . This test suite is located in the
159 <tt>debuginfo-tests</tt> Subversion module. </p>
163 <!--=========================================================================-->
164 <div class="doc_section"><a name="quick">Quick start</a></div>
165 <!--=========================================================================-->
167 <div class="doc_text">
169 <p>The tests are located in two separate Subversion modules. The regressions
170 tests are in the main "llvm" module under the directory
171 <tt>llvm/test</tt> (so you get these tests for free with the main llvm tree).
172 The more comprehensive test suite that includes whole
173 programs in C and C++ is in the <tt>test-suite</tt> module. This module should
174 be checked out to the <tt>llvm/projects</tt> directory (don't use another name
175 than the default "test-suite", for then the test suite will be run every time
176 you run <tt>make</tt> in the main <tt>llvm</tt> directory).
177 When you <tt>configure</tt> the <tt>llvm</tt> module,
178 the <tt>test-suite</tt> directory will be automatically configured.
179 Alternatively, you can configure the <tt>test-suite</tt> module manually.</p>
181 <!-- _______________________________________________________________________ -->
182 <div class="doc_subsection"><a name="quickregressiontests">Regression tests</a></div>
183 <!-- _______________________________________________________________________ -->
184 <p>To run all of the LLVM regression tests, use master Makefile in
185 the <tt>llvm/test</tt> directory:</p>
187 <div class="doc_code">
195 <div class="doc_code">
201 <p>If you have <a href="http://clang.llvm.org">Clang</a> checked out and built,
202 you can run the LLVM and Clang tests simultaneously using:</p>
206 <div class="doc_code">
212 <p>To run the tests with Valgrind (Memcheck by default), just append
213 <tt>VG=1</tt> to the commands above, e.g.:</p>
215 <div class="doc_code">
221 <p>To run individual tests or subsets of tests, you can use the 'llvm-lit'
222 script which is built as part of LLVM. For example, to run the
223 'Integer/BitCast.ll' test by itself you can run:</p>
225 <div class="doc_code">
227 % llvm-lit ~/llvm/test/Integer/BitCast.ll
231 <p>or to run all of the ARM CodeGen tests:</p>
233 <div class="doc_code">
235 % llvm-lit ~/llvm/test/CodeGen/ARM
239 <p>For more information on using the 'lit' tool, see 'llvm-lit --help' or the
242 <!-- _______________________________________________________________________ -->
243 <div class="doc_subsection"><a name="quicktestsuite">Test suite</a></div>
244 <!-- _______________________________________________________________________ -->
246 <p>To run the comprehensive test suite (tests that compile and execute whole
247 programs), first checkout and setup the <tt>test-suite</tt> module:</p>
249 <div class="doc_code">
252 % svn co http://llvm.org/svn/llvm-project/test-suite/trunk test-suite
254 % ./configure --with-llvmgccdir=$LLVM_GCC_DIR
258 <p>where <tt>$LLVM_GCC_DIR</tt> is the directory where
259 you <em>installed</em> llvm-gcc, not its src or obj
260 dir. The <tt>--with-llvmgccdir</tt> option assumes that
261 the <tt>llvm-gcc-4.2</tt> module was configured with
262 <tt>--program-prefix=llvm-</tt>, and therefore that the C and C++
263 compiler drivers are called <tt>llvm-gcc</tt> and <tt>llvm-g++</tt>
264 respectively. If this is not the case,
265 use <tt>--with-llvmgcc</tt>/<tt>--with-llvmgxx</tt> to specify each
266 executable's location.</p>
268 <p>Then, run the entire test suite by running make in the <tt>test-suite</tt>
271 <div class="doc_code">
273 % cd projects/test-suite
278 <p>Usually, running the "nightly" set of tests is a good idea, and you can also
279 let it generate a report by running:</p>
281 <div class="doc_code">
283 % cd projects/test-suite
284 % gmake TEST=nightly report report.html
288 <p>Any of the above commands can also be run in a subdirectory of
289 <tt>projects/test-suite</tt> to run the specified test only on the programs in
290 that subdirectory.</p>
294 <!-- _______________________________________________________________________ -->
295 <div class="doc_subsection"><a name="quickdebuginfotests">Debugging Information
297 <!-- _______________________________________________________________________ -->
299 <p> To run debugging information tests simply checkout the tests inside
300 clang/test directory. </p>
302 <div class="doc_code">
305 % svn co http://llvm.org/svn/llvm-project/debuginfo-tests/trunk debuginfo-tests
309 <p> These tests are already set up to run as part of clang regression tests.</p>
313 <!--=========================================================================-->
314 <div class="doc_section"><a name="rtstructure">Regression test structure</a></div>
315 <!--=========================================================================-->
316 <div class="doc_text">
317 <p>The LLVM regression tests are driven by 'lit' and are located in
318 the <tt>llvm/test</tt> directory.
320 <p>This directory contains a large array of small tests
321 that exercise various features of LLVM and to ensure that regressions do not
322 occur. The directory is broken into several sub-directories, each focused on
323 a particular area of LLVM. A few of the important ones are:</p>
326 <li><tt>Analysis</tt>: checks Analysis passes.</li>
327 <li><tt>Archive</tt>: checks the Archive library.</li>
328 <li><tt>Assembler</tt>: checks Assembly reader/writer functionality.</li>
329 <li><tt>Bitcode</tt>: checks Bitcode reader/writer functionality.</li>
330 <li><tt>CodeGen</tt>: checks code generation and each target.</li>
331 <li><tt>Features</tt>: checks various features of the LLVM language.</li>
332 <li><tt>Linker</tt>: tests bitcode linking.</li>
333 <li><tt>Transforms</tt>: tests each of the scalar, IPO, and utility
334 transforms to ensure they make the right transformations.</li>
335 <li><tt>Verifier</tt>: tests the IR verifier.</li>
340 <!-- _______________________________________________________________________ -->
341 <div class="doc_subsection"><a name="rtcustom">Writing new regression tests</a></div>
342 <!-- _______________________________________________________________________ -->
343 <div class="doc_text">
344 <p>The regression test structure is very simple, but does require some
345 information to be set. This information is gathered via <tt>configure</tt> and
346 is written to a file, <tt>lit.site.cfg</tt>
347 in <tt>llvm/test</tt>. The <tt>llvm/test</tt> Makefile does this work for
350 <p>In order for the regression tests to work, each directory of tests must
351 have a <tt>dg.exp</tt> file. Lit looks for this file to determine how to
352 run the tests. This file is just a Tcl script and it can do anything you want,
353 but we've standardized it for the LLVM regression tests. If you're adding a
354 directory of tests, just copy <tt>dg.exp</tt> from another directory to get
355 running. The standard <tt>dg.exp</tt> simply loads a Tcl library
356 (<tt>test/lib/llvm.exp</tt>) and calls the <tt>llvm_runtests</tt> function
357 defined in that library with a list of file names to run. The names are
358 obtained by using Tcl's glob command. Any directory that contains only
359 directories does not need the <tt>dg.exp</tt> file.</p>
361 <p>The <tt>llvm-runtests</tt> function lookas at each file that is passed to
362 it and gathers any lines together that match "RUN:". This are the "RUN" lines
363 that specify how the test is to be run. So, each test script must contain
364 RUN lines if it is to do anything. If there are no RUN lines, the
365 <tt>llvm-runtests</tt> function will issue an error and the test will
368 <p>RUN lines are specified in the comments of the test program using the
369 keyword <tt>RUN</tt> followed by a colon, and lastly the command (pipeline)
370 to execute. Together, these lines form the "script" that
371 <tt>llvm-runtests</tt> executes to run the test case. The syntax of the
372 RUN lines is similar to a shell's syntax for pipelines including I/O
373 redirection and variable substitution. However, even though these lines
374 may <i>look</i> like a shell script, they are not. RUN lines are interpreted
375 directly by the Tcl <tt>exec</tt> command. They are never executed by a
376 shell. Consequently the syntax differs from normal shell script syntax in a
377 few ways. You can specify as many RUN lines as needed.</p>
379 <p>Each RUN line is executed on its own, distinct from other lines unless
380 its last character is <tt>\</tt>. This continuation character causes the RUN
381 line to be concatenated with the next one. In this way you can build up long
382 pipelines of commands without making huge line lengths. The lines ending in
383 <tt>\</tt> are concatenated until a RUN line that doesn't end in <tt>\</tt> is
384 found. This concatenated set of RUN lines then constitutes one execution.
385 Tcl will substitute variables and arrange for the pipeline to be executed. If
386 any process in the pipeline fails, the entire line (and test case) fails too.
389 <p> Below is an example of legal RUN lines in a <tt>.ll</tt> file:</p>
391 <div class="doc_code">
393 ; RUN: llvm-as < %s | llvm-dis > %t1
394 ; RUN: llvm-dis < %s.bc-13 > %t2
399 <p>As with a Unix shell, the RUN: lines permit pipelines and I/O redirection
400 to be used. However, the usage is slightly different than for Bash. To check
401 what's legal, see the documentation for the
402 <a href="http://www.tcl.tk/man/tcl8.5/TclCmd/exec.htm#M2">Tcl exec</a>
404 <a href="http://www.tcl.tk/man/tcl8.5/tutorial/Tcl26.html">tutorial</a>.
405 The major differences are:</p>
407 <li>You can't do <tt>2>&1</tt>. That will cause Tcl to write to a
408 file named <tt>&1</tt>. Usually this is done to get stderr to go through
409 a pipe. You can do that in tcl with <tt>|&</tt> so replace this idiom:
410 <tt>... 2>&1 | grep</tt> with <tt>... |& grep</tt></li>
411 <li>You can only redirect to a file, not to another descriptor and not from
412 a here document.</li>
413 <li>tcl supports redirecting to open files with the @ syntax but you
414 shouldn't use that here.</li>
417 <p>There are some quoting rules that you must pay attention to when writing
418 your RUN lines. In general nothing needs to be quoted. Tcl won't strip off any
419 quote characters so they will get passed to the invoked program. For
422 <div class="doc_code">
424 ... | grep 'find this string'
428 <p>This will fail because the ' characters are passed to grep. This would
429 instruction grep to look for <tt>'find</tt> in the files <tt>this</tt> and
430 <tt>string'</tt>. To avoid this use curly braces to tell Tcl that it should
431 treat everything enclosed as one value. So our example would become:</p>
433 <div class="doc_code">
435 ... | grep {find this string}
439 <p>Additionally, the characters <tt>[</tt> and <tt>]</tt> are treated
440 specially by Tcl. They tell Tcl to interpret the content as a command to
441 execute. Since these characters are often used in regular expressions this can
442 have disastrous results and cause the entire test run in a directory to fail.
443 For example, a common idiom is to look for some basicblock number:</p>
445 <div class="doc_code">
451 <p>This, however, will cause Tcl to fail because its going to try to execute
452 a program named "2-8". Instead, what you want is this:</p>
454 <div class="doc_code">
456 ... | grep {bb\[2-8\]}
460 <p>Finally, if you need to pass the <tt>\</tt> character down to a program,
461 then it must be doubled. This is another Tcl special character. So, suppose
464 <div class="doc_code">
470 <p>This will fail to match what you want (a pointer to i32). First, the
471 <tt>'</tt> do not get stripped off. Second, the <tt>\</tt> gets stripped off
472 by Tcl so what grep sees is: <tt>'i32*'</tt>. That's not likely to match
473 anything. To resolve this you must use <tt>\\</tt> and the <tt>{}</tt>, like
476 <div class="doc_code">
482 <p>If your system includes GNU <tt>grep</tt>, make sure
483 that <tt>GREP_OPTIONS</tt> is not set in your environment. Otherwise,
484 you may get invalid results (both false positives and false
489 <!-- _______________________________________________________________________ -->
490 <div class="doc_subsection"><a name="FileCheck">The FileCheck utility</a></div>
491 <!-- _______________________________________________________________________ -->
493 <div class="doc_text">
495 <p>A powerful feature of the RUN: lines is that it allows any arbitrary commands
496 to be executed as part of the test harness. While standard (portable) unix
497 tools like 'grep' work fine on run lines, as you see above, there are a lot
498 of caveats due to interaction with Tcl syntax, and we want to make sure the
499 run lines are portable to a wide range of systems. Another major problem is
500 that grep is not very good at checking to verify that the output of a tools
501 contains a series of different output in a specific order. The FileCheck
502 tool was designed to help with these problems.</p>
504 <p>FileCheck (whose basic command line arguments are described in <a
505 href="http://llvm.org/cmds/FileCheck.html">the FileCheck man page</a> is
506 designed to read a file to check from standard input, and the set of things
507 to verify from a file specified as a command line argument. A simple example
508 of using FileCheck from a RUN line looks like this:</p>
510 <div class="doc_code">
512 ; RUN: llvm-as < %s | llc -march=x86-64 | <b>FileCheck %s</b>
516 <p>This syntax says to pipe the current file ("%s") into llvm-as, pipe that into
517 llc, then pipe the output of llc into FileCheck. This means that FileCheck will
518 be verifying its standard input (the llc output) against the filename argument
519 specified (the original .ll file specified by "%s"). To see how this works,
520 lets look at the rest of the .ll file (after the RUN line):</p>
522 <div class="doc_code">
524 define void @sub1(i32* %p, i32 %v) {
526 ; <b>CHECK: sub1:</b>
528 %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v)
532 define void @inc4(i64* %p) {
534 ; <b>CHECK: inc4:</b>
536 %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1)
542 <p>Here you can see some "CHECK:" lines specified in comments. Now you can see
543 how the file is piped into llvm-as, then llc, and the machine code output is
544 what we are verifying. FileCheck checks the machine code output to verify that
545 it matches what the "CHECK:" lines specify.</p>
547 <p>The syntax of the CHECK: lines is very simple: they are fixed strings that
548 must occur in order. FileCheck defaults to ignoring horizontal whitespace
549 differences (e.g. a space is allowed to match a tab) but otherwise, the contents
550 of the CHECK: line is required to match some thing in the test file exactly.</p>
552 <p>One nice thing about FileCheck (compared to grep) is that it allows merging
553 test cases together into logical groups. For example, because the test above
554 is checking for the "sub1:" and "inc4:" labels, it will not match unless there
555 is a "subl" in between those labels. If it existed somewhere else in the file,
556 that would not count: "grep subl" matches if subl exists anywhere in the
561 <!-- _______________________________________________________________________ -->
562 <div class="doc_subsubsection"><a
563 name="FileCheck-check-prefix">The FileCheck -check-prefix option</a></div>
565 <div class="doc_text">
567 <p>The FileCheck -check-prefix option allows multiple test configurations to be
568 driven from one .ll file. This is useful in many circumstances, for example,
569 testing different architectural variants with llc. Here's a simple example:</p>
571 <div class="doc_code">
573 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \
574 ; RUN: | <b>FileCheck %s -check-prefix=X32</b>
575 ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \
576 ; RUN: | <b>FileCheck %s -check-prefix=X64</b>
578 define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind {
579 %tmp1 = insertelement <4 x i32> %tmp, i32 %s, i32 1
580 ret <4 x i32> %tmp1
581 ; <b>X32:</b> pinsrd_1:
582 ; <b>X32:</b> pinsrd $1, 4(%esp), %xmm0
584 ; <b>X64:</b> pinsrd_1:
585 ; <b>X64:</b> pinsrd $1, %edi, %xmm0
590 <p>In this case, we're testing that we get the expected code generation with
591 both 32-bit and 64-bit code generation.</p>
595 <!-- _______________________________________________________________________ -->
596 <div class="doc_subsubsection"><a
597 name="FileCheck-CHECK-NEXT">The "CHECK-NEXT:" directive</a></div>
599 <div class="doc_text">
601 <p>Sometimes you want to match lines and would like to verify that matches
602 happen on exactly consequtive lines with no other lines in between them. In
603 this case, you can use CHECK: and CHECK-NEXT: directives to specify this. If
604 you specified a custom check prefix, just use "<PREFIX>-NEXT:". For
605 example, something like this works as you'd expect:</p>
607 <div class="doc_code">
609 define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) {
610 %tmp3 = load <2 x double>* %A, align 16
611 %tmp7 = insertelement <2 x double> undef, double %B, i32 0
612 %tmp9 = shufflevector <2 x double> %tmp3,
613 <2 x double> %tmp7,
614 <2 x i32> < i32 0, i32 2 >
615 store <2 x double> %tmp9, <2 x double>* %r, align 16
619 ; <b>CHECK:</b> movl 8(%esp), %eax
620 ; <b>CHECK-NEXT:</b> movapd (%eax), %xmm0
621 ; <b>CHECK-NEXT:</b> movhpd 12(%esp), %xmm0
622 ; <b>CHECK-NEXT:</b> movl 4(%esp), %eax
623 ; <b>CHECK-NEXT:</b> movapd %xmm0, (%eax)
624 ; <b>CHECK-NEXT:</b> ret
629 <p>CHECK-NEXT: directives reject the input unless there is exactly one newline
630 between it an the previous directive. A CHECK-NEXT cannot be the first
631 directive in a file.</p>
635 <!-- _______________________________________________________________________ -->
636 <div class="doc_subsubsection"><a
637 name="FileCheck-CHECK-NOT">The "CHECK-NOT:" directive</a></div>
639 <div class="doc_text">
641 <p>The CHECK-NOT: directive is used to verify that a string doesn't occur
642 between two matches (or the first match and the beginning of the file). For
643 example, to verify that a load is removed by a transformation, a test like this
646 <div class="doc_code">
648 define i8 @coerce_offset0(i32 %V, i32* %P) {
649 store i32 %V, i32* %P
651 %P2 = bitcast i32* %P to i8*
652 %P3 = getelementptr i8* %P2, i32 2
656 ; <b>CHECK:</b> @coerce_offset0
657 ; <b>CHECK-NOT:</b> load
658 ; <b>CHECK:</b> ret i8
665 <!-- _______________________________________________________________________ -->
666 <div class="doc_subsubsection"><a
667 name="FileCheck-Matching">FileCheck Pattern Matching Syntax</a></div>
669 <div class="doc_text">
671 <p>The CHECK: and CHECK-NOT: directives both take a pattern to match. For most
672 uses of FileCheck, fixed string matching is perfectly sufficient. For some
673 things, a more flexible form of matching is desired. To support this, FileCheck
674 allows you to specify regular expressions in matching strings, surrounded by
675 double braces: <b>{{yourregex}}</b>. Because we want to use fixed string
676 matching for a majority of what we do, FileCheck has been designed to support
677 mixing and matching fixed string matching with regular expressions. This allows
678 you to write things like this:</p>
680 <div class="doc_code">
682 ; CHECK: movhpd <b>{{[0-9]+}}</b>(%esp), <b>{{%xmm[0-7]}}</b>
686 <p>In this case, any offset from the ESP register will be allowed, and any xmm
687 register will be allowed.</p>
689 <p>Because regular expressions are enclosed with double braces, they are
690 visually distinct, and you don't need to use escape characters within the double
691 braces like you would in C. In the rare case that you want to match double
692 braces explicitly from the input, you can use something ugly like
693 <b>{{[{][{]}}</b> as your pattern.</p>
697 <!-- _______________________________________________________________________ -->
698 <div class="doc_subsubsection"><a
699 name="FileCheck-Variables">FileCheck Variables</a></div>
701 <div class="doc_text">
703 <p>It is often useful to match a pattern and then verify that it occurs again
704 later in the file. For codegen tests, this can be useful to allow any register,
705 but verify that that register is used consistently later. To do this, FileCheck
706 allows named variables to be defined and substituted into patterns. Here is a
709 <div class="doc_code">
712 ; CHECK: notw <b>[[REGISTER:%[a-z]+]]</b>
713 ; CHECK: andw {{.*}}<b>[[REGISTER]]</b>
717 <p>The first check line matches a regex (<tt>%[a-z]+</tt>) and captures it into
718 the variables "REGISTER". The second line verifies that whatever is in REGISTER
719 occurs later in the file after an "andw". FileCheck variable references are
720 always contained in <tt>[[ ]]</tt> pairs, are named, and their names can be
721 formed with the regex "<tt>[a-zA-Z][a-zA-Z0-9]*</tt>". If a colon follows the
722 name, then it is a definition of the variable, if not, it is a use.</p>
724 <p>FileCheck variables can be defined multiple times, and uses always get the
725 latest value. Note that variables are all read at the start of a "CHECK" line
726 and are all defined at the end. This means that if you have something like
727 "<tt>CHECK: [[XYZ:.*]]x[[XYZ]]</tt>" that the check line will read the previous
728 value of the XYZ variable and define a new one after the match is performed. If
729 you need to do something like this you can probably take advantage of the fact
730 that FileCheck is not actually line-oriented when it matches, this allows you to
731 define two separate CHECK lines that match on the same line.
736 <!-- _______________________________________________________________________ -->
737 <div class="doc_subsection"><a name="rtvars">Variables and
738 substitutions</a></div>
739 <!-- _______________________________________________________________________ -->
740 <div class="doc_text">
741 <p>With a RUN line there are a number of substitutions that are permitted. In
742 general, any Tcl variable that is available in the <tt>substitute</tt>
743 function (in <tt>test/lib/llvm.exp</tt>) can be substituted into a RUN line.
744 To make a substitution just write the variable's name preceded by a $.
745 Additionally, for compatibility reasons with previous versions of the test
746 library, certain names can be accessed with an alternate syntax: a % prefix.
747 These alternates are deprecated and may go away in a future version.
749 <p>Here are the available variable names. The alternate syntax is listed in
752 <dl style="margin-left: 25px">
753 <dt><b>$test</b> (%s)</dt>
754 <dd>The full path to the test case's source. This is suitable for passing
755 on the command line as the input to an llvm tool.</dd>
757 <dt><b>$srcdir</b></dt>
758 <dd>The source directory from where the "<tt>make check</tt>" was run.</dd>
760 <dt><b>objdir</b></dt>
761 <dd>The object directory that corresponds to the <tt>$srcdir</tt>.</dd>
763 <dt><b>subdir</b></dt>
764 <dd>A partial path from the <tt>test</tt> directory that contains the
765 sub-directory that contains the test source being executed.</dd>
767 <dt><b>srcroot</b></dt>
768 <dd>The root directory of the LLVM src tree.</dd>
770 <dt><b>objroot</b></dt>
771 <dd>The root directory of the LLVM object tree. This could be the same
775 <dd>The path to the directory that contains the test case source. This is
776 for locating any supporting files that are not generated by the test, but
777 used by the test.</dd>
780 <dd>The path to a temporary file name that could be used for this test case.
781 The file name won't conflict with other test cases. You can append to it if
782 you need multiple temporaries. This is useful as the destination of some
783 redirected output.</dd>
785 <dt><b>llvmlibsdir</b> (%llvmlibsdir)</dt>
786 <dd>The directory where the LLVM libraries are located.</dd>
788 <dt><b>target_triplet</b> (%target_triplet)</dt>
789 <dd>The target triplet that corresponds to the current host machine (the one
790 running the test cases). This should probably be called "host".<dd>
792 <dt><b>llvmgcc</b> (%llvmgcc)</dt>
793 <dd>The full path to the <tt>llvm-gcc</tt> executable as specified in the
794 configured LLVM environment</dd>
796 <dt><b>llvmgxx</b> (%llvmgxx)</dt>
797 <dd>The full path to the <tt>llvm-gxx</tt> executable as specified in the
798 configured LLVM environment</dd>
800 <dt><b>gccpath</b></dt>
801 <dd>The full path to the C compiler used to <i>build </i> LLVM. Note that
802 this might not be gcc.</dd>
804 <dt><b>gxxpath</b></dt>
805 <dd>The full path to the C++ compiler used to <i>build </i> LLVM. Note that
806 this might not be g++.</dd>
808 <dt><b>compile_c</b> (%compile_c)</dt>
809 <dd>The full command line used to compile LLVM C source code. This has all
810 the configured -I, -D and optimization options.</dd>
812 <dt><b>compile_cxx</b> (%compile_cxx)</dt>
813 <dd>The full command used to compile LLVM C++ source code. This has
814 all the configured -I, -D and optimization options.</dd>
816 <dt><b>link</b> (%link)</dt>
817 <dd>This full link command used to link LLVM executables. This has all the
818 configured -I, -L and -l options.</dd>
820 <dt><b>shlibext</b> (%shlibext)</dt>
821 <dd>The suffix for the host platforms share library (dll) files. This
822 includes the period as the first character.</dd>
824 <p>To add more variables, two things need to be changed. First, add a line in
825 the <tt>test/Makefile</tt> that creates the <tt>site.exp</tt> file. This will
826 "set" the variable as a global in the site.exp file. Second, in the
827 <tt>test/lib/llvm.exp</tt> file, in the substitute proc, add the variable name
828 to the list of "global" declarations at the beginning of the proc. That's it,
829 the variable can then be used in test scripts.</p>
832 <!-- _______________________________________________________________________ -->
833 <div class="doc_subsection"><a name="rtfeatures">Other Features</a></div>
834 <!-- _______________________________________________________________________ -->
835 <div class="doc_text">
836 <p>To make RUN line writing easier, there are several shell scripts located
837 in the <tt>llvm/test/Scripts</tt> directory. This directory is in the PATH
838 when running tests, so you can just call these scripts using their name. For
841 <dt><b>ignore</b></dt>
842 <dd>This script runs its arguments and then always returns 0. This is useful
843 in cases where the test needs to cause a tool to generate an error (e.g. to
844 check the error output). However, any program in a pipeline that returns a
845 non-zero result will cause the test to fail. This script overcomes that
846 issue and nicely documents that the test case is purposefully ignoring the
847 result code of the tool</dd>
850 <dd>This script runs its arguments and then inverts the result code from
851 it. Zero result codes become 1. Non-zero result codes become 0. This is
852 useful to invert the result of a grep. For example "not grep X" means
853 succeed only if you don't find X in the input.</dd>
856 <p>Sometimes it is necessary to mark a test case as "expected fail" or XFAIL.
857 You can easily mark a test as XFAIL just by including <tt>XFAIL: </tt> on a
858 line near the top of the file. This signals that the test case should succeed
859 if the test fails. Such test cases are counted separately by the testing tool. To
860 specify an expected fail, use the XFAIL keyword in the comments of the test
861 program followed by a colon and one or more regular expressions (separated by
862 a comma). The regular expressions allow you to XFAIL the test conditionally by
863 host platform. The regular expressions following the : are matched against the
864 target triplet for the host machine. If there is a match, the test is expected
865 to fail. If not, the test is expected to succeed. To XFAIL everywhere just
866 specify <tt>XFAIL: *</tt>. Here is an example of an <tt>XFAIL</tt> line:</p>
868 <div class="doc_code">
874 <p>To make the output more useful, the <tt>llvm_runtest</tt> function wil
875 scan the lines of the test case for ones that contain a pattern that matches
876 PR[0-9]+. This is the syntax for specifying a PR (Problem Report) number that
877 is related to the test case. The number after "PR" specifies the LLVM bugzilla
878 number. When a PR number is specified, it will be used in the pass/fail
879 reporting. This is useful to quickly get some context when a test fails.</p>
881 <p>Finally, any line that contains "END." will cause the special
882 interpretation of lines to terminate. This is generally done right after the
883 last RUN: line. This has two side effects: (a) it prevents special
884 interpretation of lines that are part of the test program, not the
885 instructions to the test case, and (b) it speeds things up for really big test
886 cases by avoiding interpretation of the remainder of the file.</p>
890 <!--=========================================================================-->
891 <div class="doc_section"><a name="testsuitestructure">Test suite
893 <!--=========================================================================-->
895 <div class="doc_text">
897 <p>The <tt>test-suite</tt> module contains a number of programs that can be compiled
898 with LLVM and executed. These programs are compiled using the native compiler
899 and various LLVM backends. The output from the program compiled with the
900 native compiler is assumed correct; the results from the other programs are
901 compared to the native program output and pass if they match.</p>
903 <p>When executing tests, it is usually a good idea to start out with a subset of
904 the available tests or programs. This makes test run times smaller at first and
905 later on this is useful to investigate individual test failures. To run some
906 test only on a subset of programs, simply change directory to the programs you
907 want tested and run <tt>gmake</tt> there. Alternatively, you can run a different
908 test using the <tt>TEST</tt> variable to change what tests or run on the
909 selected programs (see below for more info).</p>
911 <p>In addition for testing correctness, the <tt>test-suite</tt> directory also
912 performs timing tests of various LLVM optimizations. It also records
913 compilation times for the compilers and the JIT. This information can be
914 used to compare the effectiveness of LLVM's optimizations and code
917 <p><tt>test-suite</tt> tests are divided into three types of tests: MultiSource,
918 SingleSource, and External.</p>
921 <li><tt>test-suite/SingleSource</tt>
922 <p>The SingleSource directory contains test programs that are only a single
923 source file in size. These are usually small benchmark programs or small
924 programs that calculate a particular value. Several such programs are grouped
925 together in each directory.</p></li>
927 <li><tt>test-suite/MultiSource</tt>
928 <p>The MultiSource directory contains subdirectories which contain entire
929 programs with multiple source files. Large benchmarks and whole applications
932 <li><tt>test-suite/External</tt>
933 <p>The External directory contains Makefiles for building code that is external
934 to (i.e., not distributed with) LLVM. The most prominent members of this
935 directory are the SPEC 95 and SPEC 2000 benchmark suites. The <tt>External</tt>
936 directory does not contain these actual tests, but only the Makefiles that know
937 how to properly compile these programs from somewhere else. The presence and
938 location of these external programs is configured by the test-suite
939 <tt>configure</tt> script.</p></li>
942 <p>Each tree is then subdivided into several categories, including applications,
943 benchmarks, regression tests, code that is strange grammatically, etc. These
944 organizations should be relatively self explanatory.</p>
946 <p>Some tests are known to fail. Some are bugs that we have not fixed yet;
947 others are features that we haven't added yet (or may never add). In the
948 regression tests, the result for such tests will be XFAIL (eXpected FAILure).
949 In this way, you can tell the difference between an expected and unexpected
952 <p>The tests in the test suite have no such feature at this time. If the
953 test passes, only warnings and other miscellaneous output will be generated. If
954 a test fails, a large <program> FAILED message will be displayed. This
955 will help you separate benign warnings from actual test failures.</p>
959 <!--=========================================================================-->
960 <div class="doc_section"><a name="testsuiterun">Running the test suite</a></div>
961 <!--=========================================================================-->
963 <div class="doc_text">
965 <p>First, all tests are executed within the LLVM object directory tree. They
966 <i>are not</i> executed inside of the LLVM source tree. This is because the
967 test suite creates temporary files during execution.</p>
969 <p>To run the test suite, you need to use the following steps:</p>
972 <li><tt>cd</tt> into the <tt>llvm/projects</tt> directory in your source tree.
975 <li><p>Check out the <tt>test-suite</tt> module with:</p>
977 <div class="doc_code">
979 % svn co http://llvm.org/svn/llvm-project/test-suite/trunk test-suite
982 <p>This will get the test suite into <tt>llvm/projects/test-suite</tt>.</p>
984 <li><p>Configure and build <tt>llvm</tt>.</p></li>
985 <li><p>Configure and build <tt>llvm-gcc</tt>.</p></li>
986 <li><p>Install <tt>llvm-gcc</tt> somewhere.</p></li>
987 <li><p><em>Re-configure</em> <tt>llvm</tt> from the top level of
988 each build tree (LLVM object directory tree) in which you want
989 to run the test suite, just as you do before building LLVM.</p>
990 <p>During the <em>re-configuration</em>, you must either: (1)
991 have <tt>llvm-gcc</tt> you just built in your path, or (2)
992 specify the directory where your just-built <tt>llvm-gcc</tt> is
993 installed using <tt>--with-llvmgccdir=$LLVM_GCC_DIR</tt>.</p>
994 <p>You must also tell the configure machinery that the test suite
995 is available so it can be configured for your build tree:</p>
996 <div class="doc_code">
998 % cd $LLVM_OBJ_ROOT ; $LLVM_SRC_ROOT/configure [--with-llvmgccdir=$LLVM_GCC_DIR]
1001 <p>[Remember that <tt>$LLVM_GCC_DIR</tt> is the directory where you
1002 <em>installed</em> llvm-gcc, not its src or obj directory.]</p>
1005 <li><p>You can now run the test suite from your build tree as follows:</p>
1006 <div class="doc_code">
1008 % cd $LLVM_OBJ_ROOT/projects/test-suite
1014 <p>Note that the second and third steps only need to be done once. After you
1015 have the suite checked out and configured, you don't need to do it again (unless
1016 the test code or configure script changes).</p>
1020 <!-- _______________________________________________________________________ -->
1021 <div class="doc_subsection">
1022 <a name="testsuiteexternal">Configuring External Tests</a></div>
1023 <!-- _______________________________________________________________________ -->
1025 <div class="doc_text">
1026 <p>In order to run the External tests in the <tt>test-suite</tt>
1027 module, you must specify <i>--with-externals</i>. This
1028 must be done during the <em>re-configuration</em> step (see above),
1029 and the <tt>llvm</tt> re-configuration must recognize the
1030 previously-built <tt>llvm-gcc</tt>. If any of these is missing or
1031 neglected, the External tests won't work.</p>
1033 <dt><i>--with-externals</i></dt>
1034 <dt><i>--with-externals=<<tt>directory</tt>></i></dt>
1036 This tells LLVM where to find any external tests. They are expected to be
1037 in specifically named subdirectories of <<tt>directory</tt>>.
1038 If <tt>directory</tt> is left unspecified,
1039 <tt>configure</tt> uses the default value
1040 <tt>/home/vadve/shared/benchmarks/speccpu2000/benchspec</tt>.
1041 Subdirectory names known to LLVM include:
1044 <dt>speccpu2000</dt>
1045 <dt>speccpu2006</dt>
1048 Others are added from time to time, and can be determined from
1052 <!-- _______________________________________________________________________ -->
1053 <div class="doc_subsection">
1054 <a name="testsuitetests">Running different tests</a></div>
1055 <!-- _______________________________________________________________________ -->
1056 <div class="doc_text">
1057 <p>In addition to the regular "whole program" tests, the <tt>test-suite</tt>
1058 module also provides a mechanism for compiling the programs in different ways.
1059 If the variable TEST is defined on the <tt>gmake</tt> command line, the test system will
1060 include a Makefile named <tt>TEST.<value of TEST variable>.Makefile</tt>.
1061 This Makefile can modify build rules to yield different results.</p>
1063 <p>For example, the LLVM nightly tester uses <tt>TEST.nightly.Makefile</tt> to
1064 create the nightly test reports. To run the nightly tests, run <tt>gmake
1065 TEST=nightly</tt>.</p>
1067 <p>There are several TEST Makefiles available in the tree. Some of them are
1068 designed for internal LLVM research and will not work outside of the LLVM
1069 research group. They may still be valuable, however, as a guide to writing your
1070 own TEST Makefile for any optimization or analysis passes that you develop with
1075 <!-- _______________________________________________________________________ -->
1076 <div class="doc_subsection">
1077 <a name="testsuiteoutput">Generating test output</a></div>
1078 <!-- _______________________________________________________________________ -->
1079 <div class="doc_text">
1080 <p>There are a number of ways to run the tests and generate output. The most
1081 simple one is simply running <tt>gmake</tt> with no arguments. This will
1082 compile and run all programs in the tree using a number of different methods
1083 and compare results. Any failures are reported in the output, but are likely
1084 drowned in the other output. Passes are not reported explicitely.</p>
1086 <p>Somewhat better is running <tt>gmake TEST=sometest test</tt>, which runs
1087 the specified test and usually adds per-program summaries to the output
1088 (depending on which sometest you use). For example, the <tt>nightly</tt> test
1089 explicitely outputs TEST-PASS or TEST-FAIL for every test after each program.
1090 Though these lines are still drowned in the output, it's easy to grep the
1091 output logs in the Output directories.</p>
1093 <p>Even better are the <tt>report</tt> and <tt>report.format</tt> targets
1094 (where <tt>format</tt> is one of <tt>html</tt>, <tt>csv</tt>, <tt>text</tt> or
1095 <tt>graphs</tt>). The exact contents of the report are dependent on which
1096 <tt>TEST</tt> you are running, but the text results are always shown at the
1097 end of the run and the results are always stored in the
1098 <tt>report.<type>.format</tt> file (when running with
1099 <tt>TEST=<type></tt>).
1101 The <tt>report</tt> also generate a file called
1102 <tt>report.<type>.raw.out</tt> containing the output of the entire test
1106 <!-- _______________________________________________________________________ -->
1107 <div class="doc_subsection">
1108 <a name="testsuitecustom">Writing custom tests for the test suite</a></div>
1109 <!-- _______________________________________________________________________ -->
1111 <div class="doc_text">
1113 <p>Assuming you can run the test suite, (e.g. "<tt>gmake TEST=nightly report</tt>"
1114 should work), it is really easy to run optimizations or code generator
1115 components against every program in the tree, collecting statistics or running
1116 custom checks for correctness. At base, this is how the nightly tester works,
1117 it's just one example of a general framework.</p>
1119 <p>Lets say that you have an LLVM optimization pass, and you want to see how
1120 many times it triggers. First thing you should do is add an LLVM
1121 <a href="ProgrammersManual.html#Statistic">statistic</a> to your pass, which
1122 will tally counts of things you care about.</p>
1124 <p>Following this, you can set up a test and a report that collects these and
1125 formats them for easy viewing. This consists of two files, a
1126 "<tt>test-suite/TEST.XXX.Makefile</tt>" fragment (where XXX is the name of your
1127 test) and a "<tt>test-suite/TEST.XXX.report</tt>" file that indicates how to
1128 format the output into a table. There are many example reports of various
1129 levels of sophistication included with the test suite, and the framework is very
1132 <p>If you are interested in testing an optimization pass, check out the
1133 "libcalls" test as an example. It can be run like this:<p>
1135 <div class="doc_code">
1137 % cd llvm/projects/test-suite/MultiSource/Benchmarks # or some other level
1138 % make TEST=libcalls report
1142 <p>This will do a bunch of stuff, then eventually print a table like this:</p>
1144 <div class="doc_code">
1146 Name | total | #exit |
1148 FreeBench/analyzer/analyzer | 51 | 6 |
1149 FreeBench/fourinarow/fourinarow | 1 | 1 |
1150 FreeBench/neural/neural | 19 | 9 |
1151 FreeBench/pifft/pifft | 5 | 3 |
1152 MallocBench/cfrac/cfrac | 1 | * |
1153 MallocBench/espresso/espresso | 52 | 12 |
1154 MallocBench/gs/gs | 4 | * |
1155 Prolangs-C/TimberWolfMC/timberwolfmc | 302 | * |
1156 Prolangs-C/agrep/agrep | 33 | 12 |
1157 Prolangs-C/allroots/allroots | * | * |
1158 Prolangs-C/assembler/assembler | 47 | * |
1159 Prolangs-C/bison/mybison | 74 | * |
1164 <p>This basically is grepping the -stats output and displaying it in a table.
1165 You can also use the "TEST=libcalls report.html" target to get the table in HTML
1166 form, similarly for report.csv and report.tex.</p>
1168 <p>The source for this is in test-suite/TEST.libcalls.*. The format is pretty
1169 simple: the Makefile indicates how to run the test (in this case,
1170 "<tt>opt -simplify-libcalls -stats</tt>"), and the report contains one line for
1171 each column of the output. The first value is the header for the column and the
1172 second is the regex to grep the output of the command for. There are lots of
1173 example reports that can do fancy stuff.</p>
1177 <!-- *********************************************************************** -->
1181 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1182 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
1183 <a href="http://validator.w3.org/check/referer"><img
1184 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
1186 John T. Criswell, Daniel Dunbar, Reid Spencer, and Tanya Lattner<br>
1187 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
1188 Last modified: $Date$