llvm-6502/docs/TestingGuide.html
2011-10-31 11:21:59 +00:00

1187 lines
47 KiB
HTML

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
"http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<title>LLVM Testing Infrastructure Guide</title>
<link rel="stylesheet" href="llvm.css" type="text/css">
</head>
<body>
<h1>
LLVM Testing Infrastructure Guide
</h1>
<ol>
<li><a href="#overview">Overview</a></li>
<li><a href="#requirements">Requirements</a></li>
<li><a href="#org">LLVM testing infrastructure organization</a>
<ul>
<li><a href="#regressiontests">Regression tests</a></li>
<li><a href="#testsuite">Test suite</a></li>
<li><a href="#debuginfotests">Debugging Information tests</a></li>
</ul>
</li>
<li><a href="#quick">Quick start</a>
<ul>
<li><a href="#quickregressiontests">Regression tests</a></li>
<li><a href="#quicktestsuite">Test suite</a></li>
<li><a href="#quickdebuginfotests">Debugging Information tests</a></li>
</ul>
</li>
<li><a href="#rtstructure">Regression test structure</a>
<ul>
<li><a href="#rtcustom">Writing new regression tests</a></li>
<li><a href="#FileCheck">The FileCheck utility</a></li>
<li><a href="#rtvars">Variables and substitutions</a></li>
<li><a href="#rtfeatures">Other features</a></li>
</ul>
</li>
<li><a href="#testsuitestructure">Test suite structure</a></li>
<li><a href="#testsuiterun">Running the test suite</a>
<ul>
<li><a href="#testsuiteexternal">Configuring External Tests</a></li>
<li><a href="#testsuitetests">Running different tests</a></li>
<li><a href="#testsuiteoutput">Generating test output</a></li>
<li><a href="#testsuitecustom">Writing custom tests for test-suite</a></li>
</ul>
</li>
</ol>
<div class="doc_author">
<p>Written by John T. Criswell, Daniel Dunbar, Reid Spencer, and Tanya Lattner</p>
</div>
<!--=========================================================================-->
<h2><a name="overview">Overview</a></h2>
<!--=========================================================================-->
<div>
<p>This document is the reference manual for the LLVM testing infrastructure. It
documents the structure of the LLVM testing infrastructure, the tools needed to
use it, and how to add and run tests.</p>
</div>
<!--=========================================================================-->
<h2><a name="requirements">Requirements</a></h2>
<!--=========================================================================-->
<div>
<p>In order to use the LLVM testing infrastructure, you will need all of the
software required to build LLVM, as well
as <a href="http://python.org">Python</a> 2.4 or later.</p>
</div>
<!--=========================================================================-->
<h2><a name="org">LLVM testing infrastructure organization</a></h2>
<!--=========================================================================-->
<div>
<p>The LLVM testing infrastructure contains two major categories of tests:
regression tests and whole programs. The regression tests are contained inside
the LLVM repository itself under <tt>llvm/test</tt> and are expected to always
pass -- they should be run before every commit. The whole programs tests are
referred to as the "LLVM test suite" and are in the <tt>test-suite</tt> module
in subversion.
</p>
<!-- _______________________________________________________________________ -->
<h3><a name="regressiontests">Regression tests</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>The regression tests are small pieces of code that test a specific feature of
LLVM or trigger a specific bug in LLVM. They are usually written in LLVM
assembly language, but can be written in other languages if the test targets a
particular language front end (and the appropriate <tt>--with-llvmgcc</tt>
options were used at <tt>configure</tt> time of the <tt>llvm</tt> module). These
tests are driven by the 'lit' testing tool, which is part of LLVM.</p>
<p>These code fragments are not complete programs. The code generated
from them is never executed to determine correct behavior.</p>
<p>These code fragment tests are located in the <tt>llvm/test</tt>
directory.</p>
<p>Typically when a bug is found in LLVM, a regression test containing
just enough code to reproduce the problem should be written and placed
somewhere underneath this directory. In most cases, this will be a small
piece of LLVM assembly language code, often distilled from an actual
application or benchmark.</p>
</div>
<!-- _______________________________________________________________________ -->
<h3><a name="testsuite">Test suite</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>The test suite contains whole programs, which are pieces of
code which can be compiled and linked into a stand-alone program that can be
executed. These programs are generally written in high level languages such as
C or C++, but sometimes they are written straight in LLVM assembly.</p>
<p>These programs are compiled and then executed using several different
methods (native compiler, LLVM C backend, LLVM JIT, LLVM native code generation,
etc). The output of these programs is compared to ensure that LLVM is compiling
the program correctly.</p>
<p>In addition to compiling and executing programs, whole program tests serve as
a way of benchmarking LLVM performance, both in terms of the efficiency of the
programs generated as well as the speed with which LLVM compiles, optimizes, and
generates code.</p>
<p>The test-suite is located in the <tt>test-suite</tt> Subversion module.</p>
</div>
<!-- _______________________________________________________________________ -->
<h3><a name="debuginfotests">Debugging Information tests</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>The test suite contains tests to check quality of debugging information.
The test are written in C based languages or in LLVM assembly language. </p>
<p>These tests are compiled and run under a debugger. The debugger output
is checked to validate of debugging information. See README.txt in the
test suite for more information . This test suite is located in the
<tt>debuginfo-tests</tt> Subversion module. </p>
</div>
</div>
<!--=========================================================================-->
<h2><a name="quick">Quick start</a></h2>
<!--=========================================================================-->
<div>
<p>The tests are located in two separate Subversion modules. The regressions
tests are in the main "llvm" module under the directory
<tt>llvm/test</tt> (so you get these tests for free with the main llvm tree).
The more comprehensive test suite that includes whole
programs in C and C++ is in the <tt>test-suite</tt> module. This module should
be checked out to the <tt>llvm/projects</tt> directory (don't use another name
than the default "test-suite", for then the test suite will be run every time
you run <tt>make</tt> in the main <tt>llvm</tt> directory).
When you <tt>configure</tt> the <tt>llvm</tt> module,
the <tt>test-suite</tt> directory will be automatically configured.
Alternatively, you can configure the <tt>test-suite</tt> module manually.</p>
<!-- _______________________________________________________________________ -->
<h3><a name="quickregressiontests">Regression tests</a></h3>
<div>
<!-- _______________________________________________________________________ -->
<p>To run all of the LLVM regression tests, use master Makefile in
the <tt>llvm/test</tt> directory:</p>
<div class="doc_code">
<pre>
% gmake -C llvm/test
</pre>
</div>
<p>or</p>
<div class="doc_code">
<pre>
% gmake check
</pre>
</div>
<p>If you have <a href="http://clang.llvm.org/">Clang</a> checked out and built,
you can run the LLVM and Clang tests simultaneously using:</p>
<p>or</p>
<div class="doc_code">
<pre>
% gmake check-all
</pre>
</div>
<p>To run the tests with Valgrind (Memcheck by default), just append
<tt>VG=1</tt> to the commands above, e.g.:</p>
<div class="doc_code">
<pre>
% gmake check VG=1
</pre>
</div>
<p>To run individual tests or subsets of tests, you can use the 'llvm-lit'
script which is built as part of LLVM. For example, to run the
'Integer/BitCast.ll' test by itself you can run:</p>
<div class="doc_code">
<pre>
% llvm-lit ~/llvm/test/Integer/BitCast.ll
</pre>
</div>
<p>or to run all of the ARM CodeGen tests:</p>
<div class="doc_code">
<pre>
% llvm-lit ~/llvm/test/CodeGen/ARM
</pre>
</div>
<p>For more information on using the 'lit' tool, see 'llvm-lit --help' or the
'lit' man page.</p>
</div>
<!-- _______________________________________________________________________ -->
<h3><a name="quicktestsuite">Test suite</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>To run the comprehensive test suite (tests that compile and execute whole
programs), first checkout and setup the <tt>test-suite</tt> module:</p>
<div class="doc_code">
<pre>
% cd llvm/projects
% svn co http://llvm.org/svn/llvm-project/test-suite/trunk test-suite
% cd ..
% ./configure --with-llvmgccdir=$LLVM_GCC_DIR
</pre>
</div>
<p>where <tt>$LLVM_GCC_DIR</tt> is the directory where
you <em>installed</em> llvm-gcc, not its src or obj
dir. The <tt>--with-llvmgccdir</tt> option assumes that
the <tt>llvm-gcc-4.2</tt> module was configured with
<tt>--program-prefix=llvm-</tt>, and therefore that the C and C++
compiler drivers are called <tt>llvm-gcc</tt> and <tt>llvm-g++</tt>
respectively. If this is not the case,
use <tt>--with-llvmgcc</tt>/<tt>--with-llvmgxx</tt> to specify each
executable's location.</p>
<p>Then, run the entire test suite by running make in the <tt>test-suite</tt>
directory:</p>
<div class="doc_code">
<pre>
% cd projects/test-suite
% gmake
</pre>
</div>
<p>Usually, running the "nightly" set of tests is a good idea, and you can also
let it generate a report by running:</p>
<div class="doc_code">
<pre>
% cd projects/test-suite
% gmake TEST=nightly report report.html
</pre>
</div>
<p>Any of the above commands can also be run in a subdirectory of
<tt>projects/test-suite</tt> to run the specified test only on the programs in
that subdirectory.</p>
</div>
<!-- _______________________________________________________________________ -->
<h3><a name="quickdebuginfotests">Debugging Information tests</a></h3>
<div>
<!-- _______________________________________________________________________ -->
<div>
<p> To run debugging information tests simply checkout the tests inside
clang/test directory. </p>
<div class="doc_code">
<pre>
%cd clang/test
% svn co http://llvm.org/svn/llvm-project/debuginfo-tests/trunk debuginfo-tests
</pre>
</div>
<p> These tests are already set up to run as part of clang regression tests.</p>
</div>
</div>
</div>
<!--=========================================================================-->
<h2><a name="rtstructure">Regression test structure</a></h2>
<!--=========================================================================-->
<div>
<p>The LLVM regression tests are driven by 'lit' and are located in
the <tt>llvm/test</tt> directory.
<p>This directory contains a large array of small tests
that exercise various features of LLVM and to ensure that regressions do not
occur. The directory is broken into several sub-directories, each focused on
a particular area of LLVM. A few of the important ones are:</p>
<ul>
<li><tt>Analysis</tt>: checks Analysis passes.</li>
<li><tt>Archive</tt>: checks the Archive library.</li>
<li><tt>Assembler</tt>: checks Assembly reader/writer functionality.</li>
<li><tt>Bitcode</tt>: checks Bitcode reader/writer functionality.</li>
<li><tt>CodeGen</tt>: checks code generation and each target.</li>
<li><tt>Features</tt>: checks various features of the LLVM language.</li>
<li><tt>Linker</tt>: tests bitcode linking.</li>
<li><tt>Transforms</tt>: tests each of the scalar, IPO, and utility
transforms to ensure they make the right transformations.</li>
<li><tt>Verifier</tt>: tests the IR verifier.</li>
</ul>
<!-- _______________________________________________________________________ -->
<h3><a name="rtcustom">Writing new regression tests</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>The regression test structure is very simple, but does require some
information to be set. This information is gathered via <tt>configure</tt> and
is written to a file, <tt>lit.site.cfg</tt>
in <tt>llvm/test</tt>. The <tt>llvm/test</tt> Makefile does this work for
you.</p>
<p>In order for the regression tests to work, each directory of tests must
have a <tt>dg.exp</tt> file. Lit looks for this file to determine how to
run the tests. This file is just a Tcl script and it can do anything you want,
but we've standardized it for the LLVM regression tests. If you're adding a
directory of tests, just copy <tt>dg.exp</tt> from another directory to get
running. The standard <tt>dg.exp</tt> simply loads a Tcl library
(<tt>test/lib/llvm.exp</tt>) and calls the <tt>llvm_runtests</tt> function
defined in that library with a list of file names to run. The names are
obtained by using Tcl's glob command. Any directory that contains only
directories does not need the <tt>dg.exp</tt> file.</p>
<p>The <tt>llvm-runtests</tt> function looks at each file that is passed to
it and gathers any lines together that match "RUN:". These are the "RUN" lines
that specify how the test is to be run. So, each test script must contain
RUN lines if it is to do anything. If there are no RUN lines, the
<tt>llvm-runtests</tt> function will issue an error and the test will
fail.</p>
<p>RUN lines are specified in the comments of the test program using the
keyword <tt>RUN</tt> followed by a colon, and lastly the command (pipeline)
to execute. Together, these lines form the "script" that
<tt>llvm-runtests</tt> executes to run the test case. The syntax of the
RUN lines is similar to a shell's syntax for pipelines including I/O
redirection and variable substitution. However, even though these lines
may <i>look</i> like a shell script, they are not. RUN lines are interpreted
directly by the Tcl <tt>exec</tt> command. They are never executed by a
shell. Consequently the syntax differs from normal shell script syntax in a
few ways. You can specify as many RUN lines as needed.</p>
<p>lit performs substitution on each RUN line to replace LLVM tool
names with the full paths to the executable built for each tool (in
$(LLVM_OBJ_ROOT)/$(BuildMode)/bin). This ensures that lit does not
invoke any stray LLVM tools in the user's path during testing.</p>
<p>Each RUN line is executed on its own, distinct from other lines unless
its last character is <tt>\</tt>. This continuation character causes the RUN
line to be concatenated with the next one. In this way you can build up long
pipelines of commands without making huge line lengths. The lines ending in
<tt>\</tt> are concatenated until a RUN line that doesn't end in <tt>\</tt> is
found. This concatenated set of RUN lines then constitutes one execution.
Tcl will substitute variables and arrange for the pipeline to be executed. If
any process in the pipeline fails, the entire line (and test case) fails too.
</p>
<p> Below is an example of legal RUN lines in a <tt>.ll</tt> file:</p>
<div class="doc_code">
<pre>
; RUN: llvm-as &lt; %s | llvm-dis &gt; %t1
; RUN: llvm-dis &lt; %s.bc-13 &gt; %t2
; RUN: diff %t1 %t2
</pre>
</div>
<p>As with a Unix shell, the RUN: lines permit pipelines and I/O redirection
to be used. However, the usage is slightly different than for Bash. To check
what's legal, see the documentation for the
<a href="http://www.tcl.tk/man/tcl8.5/TclCmd/exec.htm#M2">Tcl exec</a>
command and the
<a href="http://www.tcl.tk/man/tcl8.5/tutorial/Tcl26.html">tutorial</a>.
The major differences are:</p>
<ul>
<li>You can't do <tt>2&gt;&amp;1</tt>. That will cause Tcl to write to a
file named <tt>&amp;1</tt>. Usually this is done to get stderr to go through
a pipe. You can do that in tcl with <tt>|&amp;</tt> so replace this idiom:
<tt>... 2&gt;&amp;1 | grep</tt> with <tt>... |&amp; grep</tt></li>
<li>You can only redirect to a file, not to another descriptor and not from
a here document.</li>
<li>tcl supports redirecting to open files with the @ syntax but you
shouldn't use that here.</li>
</ul>
<p>There are some quoting rules that you must pay attention to when writing
your RUN lines. In general nothing needs to be quoted. Tcl won't strip off any
quote characters so they will get passed to the invoked program. For
example:</p>
<div class="doc_code">
<pre>
... | grep 'find this string'
</pre>
</div>
<p>This will fail because the ' characters are passed to grep. This would
instruction grep to look for <tt>'find</tt> in the files <tt>this</tt> and
<tt>string'</tt>. To avoid this use curly braces to tell Tcl that it should
treat everything enclosed as one value. So our example would become:</p>
<div class="doc_code">
<pre>
... | grep {find this string}
</pre>
</div>
<p>Additionally, the characters <tt>[</tt> and <tt>]</tt> are treated
specially by Tcl. They tell Tcl to interpret the content as a command to
execute. Since these characters are often used in regular expressions this can
have disastrous results and cause the entire test run in a directory to fail.
For example, a common idiom is to look for some basicblock number:</p>
<div class="doc_code">
<pre>
... | grep bb[2-8]
</pre>
</div>
<p>This, however, will cause Tcl to fail because its going to try to execute
a program named "2-8". Instead, what you want is this:</p>
<div class="doc_code">
<pre>
... | grep {bb\[2-8\]}
</pre>
</div>
<p>Finally, if you need to pass the <tt>\</tt> character down to a program,
then it must be doubled. This is another Tcl special character. So, suppose
you had:
<div class="doc_code">
<pre>
... | grep 'i32\*'
</pre>
</div>
<p>This will fail to match what you want (a pointer to i32). First, the
<tt>'</tt> do not get stripped off. Second, the <tt>\</tt> gets stripped off
by Tcl so what grep sees is: <tt>'i32*'</tt>. That's not likely to match
anything. To resolve this you must use <tt>\\</tt> and the <tt>{}</tt>, like
this:</p>
<div class="doc_code">
<pre>
... | grep {i32\\*}
</pre>
</div>
<p>If your system includes GNU <tt>grep</tt>, make sure
that <tt>GREP_OPTIONS</tt> is not set in your environment. Otherwise,
you may get invalid results (both false positives and false
negatives).</p>
</div>
<!-- _______________________________________________________________________ -->
<h3><a name="FileCheck">The FileCheck utility</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>A powerful feature of the RUN: lines is that it allows any arbitrary commands
to be executed as part of the test harness. While standard (portable) unix
tools like 'grep' work fine on run lines, as you see above, there are a lot
of caveats due to interaction with Tcl syntax, and we want to make sure the
run lines are portable to a wide range of systems. Another major problem is
that grep is not very good at checking to verify that the output of a tools
contains a series of different output in a specific order. The FileCheck
tool was designed to help with these problems.</p>
<p>FileCheck (whose basic command line arguments are described in <a
href="http://llvm.org/cmds/FileCheck.html">the FileCheck man page</a> is
designed to read a file to check from standard input, and the set of things
to verify from a file specified as a command line argument. A simple example
of using FileCheck from a RUN line looks like this:</p>
<div class="doc_code">
<pre>
; RUN: llvm-as &lt; %s | llc -march=x86-64 | <b>FileCheck %s</b>
</pre>
</div>
<p>This syntax says to pipe the current file ("%s") into llvm-as, pipe that into
llc, then pipe the output of llc into FileCheck. This means that FileCheck will
be verifying its standard input (the llc output) against the filename argument
specified (the original .ll file specified by "%s"). To see how this works,
lets look at the rest of the .ll file (after the RUN line):</p>
<div class="doc_code">
<pre>
define void @sub1(i32* %p, i32 %v) {
entry:
; <b>CHECK: sub1:</b>
; <b>CHECK: subl</b>
%0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v)
ret void
}
define void @inc4(i64* %p) {
entry:
; <b>CHECK: inc4:</b>
; <b>CHECK: incq</b>
%0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1)
ret void
}
</pre>
</div>
<p>Here you can see some "CHECK:" lines specified in comments. Now you can see
how the file is piped into llvm-as, then llc, and the machine code output is
what we are verifying. FileCheck checks the machine code output to verify that
it matches what the "CHECK:" lines specify.</p>
<p>The syntax of the CHECK: lines is very simple: they are fixed strings that
must occur in order. FileCheck defaults to ignoring horizontal whitespace
differences (e.g. a space is allowed to match a tab) but otherwise, the contents
of the CHECK: line is required to match some thing in the test file exactly.</p>
<p>One nice thing about FileCheck (compared to grep) is that it allows merging
test cases together into logical groups. For example, because the test above
is checking for the "sub1:" and "inc4:" labels, it will not match unless there
is a "subl" in between those labels. If it existed somewhere else in the file,
that would not count: "grep subl" matches if subl exists anywhere in the
file.</p>
<!-- _______________________________________________________________________ -->
<h4>
<a name="FileCheck-check-prefix">The FileCheck -check-prefix option</a>
</h4>
<div>
<p>The FileCheck -check-prefix option allows multiple test configurations to be
driven from one .ll file. This is useful in many circumstances, for example,
testing different architectural variants with llc. Here's a simple example:</p>
<div class="doc_code">
<pre>
; RUN: llvm-as &lt; %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \
; RUN: | <b>FileCheck %s -check-prefix=X32</b>
; RUN: llvm-as &lt; %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \
; RUN: | <b>FileCheck %s -check-prefix=X64</b>
define &lt;4 x i32&gt; @pinsrd_1(i32 %s, &lt;4 x i32&gt; %tmp) nounwind {
%tmp1 = insertelement &lt;4 x i32&gt; %tmp, i32 %s, i32 1
ret &lt;4 x i32&gt; %tmp1
; <b>X32:</b> pinsrd_1:
; <b>X32:</b> pinsrd $1, 4(%esp), %xmm0
; <b>X64:</b> pinsrd_1:
; <b>X64:</b> pinsrd $1, %edi, %xmm0
}
</pre>
</div>
<p>In this case, we're testing that we get the expected code generation with
both 32-bit and 64-bit code generation.</p>
</div>
<!-- _______________________________________________________________________ -->
<h4>
<a name="FileCheck-CHECK-NEXT">The "CHECK-NEXT:" directive</a>
</h4>
<div>
<p>Sometimes you want to match lines and would like to verify that matches
happen on exactly consecutive lines with no other lines in between them. In
this case, you can use CHECK: and CHECK-NEXT: directives to specify this. If
you specified a custom check prefix, just use "&lt;PREFIX&gt;-NEXT:". For
example, something like this works as you'd expect:</p>
<div class="doc_code">
<pre>
define void @t2(&lt;2 x double&gt;* %r, &lt;2 x double&gt;* %A, double %B) {
%tmp3 = load &lt;2 x double&gt;* %A, align 16
%tmp7 = insertelement &lt;2 x double&gt; undef, double %B, i32 0
%tmp9 = shufflevector &lt;2 x double&gt; %tmp3,
&lt;2 x double&gt; %tmp7,
&lt;2 x i32&gt; &lt; i32 0, i32 2 &gt;
store &lt;2 x double&gt; %tmp9, &lt;2 x double&gt;* %r, align 16
ret void
; <b>CHECK:</b> t2:
; <b>CHECK:</b> movl 8(%esp), %eax
; <b>CHECK-NEXT:</b> movapd (%eax), %xmm0
; <b>CHECK-NEXT:</b> movhpd 12(%esp), %xmm0
; <b>CHECK-NEXT:</b> movl 4(%esp), %eax
; <b>CHECK-NEXT:</b> movapd %xmm0, (%eax)
; <b>CHECK-NEXT:</b> ret
}
</pre>
</div>
<p>CHECK-NEXT: directives reject the input unless there is exactly one newline
between it an the previous directive. A CHECK-NEXT cannot be the first
directive in a file.</p>
</div>
<!-- _______________________________________________________________________ -->
<h4>
<a name="FileCheck-CHECK-NOT">The "CHECK-NOT:" directive</a>
</h4>
<div>
<p>The CHECK-NOT: directive is used to verify that a string doesn't occur
between two matches (or the first match and the beginning of the file). For
example, to verify that a load is removed by a transformation, a test like this
can be used:</p>
<div class="doc_code">
<pre>
define i8 @coerce_offset0(i32 %V, i32* %P) {
store i32 %V, i32* %P
%P2 = bitcast i32* %P to i8*
%P3 = getelementptr i8* %P2, i32 2
%A = load i8* %P3
ret i8 %A
; <b>CHECK:</b> @coerce_offset0
; <b>CHECK-NOT:</b> load
; <b>CHECK:</b> ret i8
}
</pre>
</div>
</div>
<!-- _______________________________________________________________________ -->
<h4>
<a name="FileCheck-Matching">FileCheck Pattern Matching Syntax</a>
</h4>
<div>
<p>The CHECK: and CHECK-NOT: directives both take a pattern to match. For most
uses of FileCheck, fixed string matching is perfectly sufficient. For some
things, a more flexible form of matching is desired. To support this, FileCheck
allows you to specify regular expressions in matching strings, surrounded by
double braces: <b>{{yourregex}}</b>. Because we want to use fixed string
matching for a majority of what we do, FileCheck has been designed to support
mixing and matching fixed string matching with regular expressions. This allows
you to write things like this:</p>
<div class="doc_code">
<pre>
; CHECK: movhpd <b>{{[0-9]+}}</b>(%esp), <b>{{%xmm[0-7]}}</b>
</pre>
</div>
<p>In this case, any offset from the ESP register will be allowed, and any xmm
register will be allowed.</p>
<p>Because regular expressions are enclosed with double braces, they are
visually distinct, and you don't need to use escape characters within the double
braces like you would in C. In the rare case that you want to match double
braces explicitly from the input, you can use something ugly like
<b>{{[{][{]}}</b> as your pattern.</p>
</div>
<!-- _______________________________________________________________________ -->
<h4>
<a name="FileCheck-Variables">FileCheck Variables</a>
</h4>
<div>
<p>It is often useful to match a pattern and then verify that it occurs again
later in the file. For codegen tests, this can be useful to allow any register,
but verify that that register is used consistently later. To do this, FileCheck
allows named variables to be defined and substituted into patterns. Here is a
simple example:</p>
<div class="doc_code">
<pre>
; CHECK: test5:
; CHECK: notw <b>[[REGISTER:%[a-z]+]]</b>
; CHECK: andw {{.*}}<b>[[REGISTER]]</b>
</pre>
</div>
<p>The first check line matches a regex (<tt>%[a-z]+</tt>) and captures it into
the variables "REGISTER". The second line verifies that whatever is in REGISTER
occurs later in the file after an "andw". FileCheck variable references are
always contained in <tt>[[ ]]</tt> pairs, are named, and their names can be
formed with the regex "<tt>[a-zA-Z][a-zA-Z0-9]*</tt>". If a colon follows the
name, then it is a definition of the variable, if not, it is a use.</p>
<p>FileCheck variables can be defined multiple times, and uses always get the
latest value. Note that variables are all read at the start of a "CHECK" line
and are all defined at the end. This means that if you have something like
"<tt>CHECK: [[XYZ:.*]]x[[XYZ]]</tt>" that the check line will read the previous
value of the XYZ variable and define a new one after the match is performed. If
you need to do something like this you can probably take advantage of the fact
that FileCheck is not actually line-oriented when it matches, this allows you to
define two separate CHECK lines that match on the same line.
</p>
</div>
</div>
<!-- _______________________________________________________________________ -->
<h3><a name="rtvars">Variables and substitutions</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>With a RUN line there are a number of substitutions that are permitted. In
general, any Tcl variable that is available in the <tt>substitute</tt>
function (in <tt>test/lib/llvm.exp</tt>) can be substituted into a RUN line.
To make a substitution just write the variable's name preceded by a $.
Additionally, for compatibility reasons with previous versions of the test
library, certain names can be accessed with an alternate syntax: a % prefix.
These alternates are deprecated and may go away in a future version.
</p>
<p>Here are the available variable names. The alternate syntax is listed in
parentheses.</p>
<dl style="margin-left: 25px">
<dt><b>$test</b> (%s)</dt>
<dd>The full path to the test case's source. This is suitable for passing
on the command line as the input to an llvm tool.</dd>
<dt><b>$srcdir</b></dt>
<dd>The source directory from where the "<tt>make check</tt>" was run.</dd>
<dt><b>objdir</b></dt>
<dd>The object directory that corresponds to the <tt>$srcdir</tt>.</dd>
<dt><b>subdir</b></dt>
<dd>A partial path from the <tt>test</tt> directory that contains the
sub-directory that contains the test source being executed.</dd>
<dt><b>srcroot</b></dt>
<dd>The root directory of the LLVM src tree.</dd>
<dt><b>objroot</b></dt>
<dd>The root directory of the LLVM object tree. This could be the same
as the srcroot.</dd>
<dt><b>path</b><dt>
<dd>The path to the directory that contains the test case source. This is
for locating any supporting files that are not generated by the test, but
used by the test.</dd>
<dt><b>tmp</b></dt>
<dd>The path to a temporary file name that could be used for this test case.
The file name won't conflict with other test cases. You can append to it if
you need multiple temporaries. This is useful as the destination of some
redirected output.</dd>
<dt><b>target_triplet</b> (%target_triplet)</dt>
<dd>The target triplet that corresponds to the current host machine (the one
running the test cases). This should probably be called "host".<dd>
<dt><b>link</b> (%link)</dt>
<dd>This full link command used to link LLVM executables. This has all the
configured -I, -L and -l options.</dd>
<dt><b>shlibext</b> (%shlibext)</dt>
<dd>The suffix for the host platforms share library (dll) files. This
includes the period as the first character.</dd>
</dl>
<p>To add more variables, two things need to be changed. First, add a line in
the <tt>test/Makefile</tt> that creates the <tt>site.exp</tt> file. This will
"set" the variable as a global in the site.exp file. Second, in the
<tt>test/lib/llvm.exp</tt> file, in the substitute proc, add the variable name
to the list of "global" declarations at the beginning of the proc. That's it,
the variable can then be used in test scripts.</p>
</div>
<!-- _______________________________________________________________________ -->
<h3><a name="rtfeatures">Other Features</a></h3>
<!-- _______________________________________________________________________ -->
<div>
<p>To make RUN line writing easier, there are several shell scripts located
in the <tt>llvm/test/Scripts</tt> directory. This directory is in the PATH
when running tests, so you can just call these scripts using their name. For
example:</p>
<dl>
<dt><b>ignore</b></dt>
<dd>This script runs its arguments and then always returns 0. This is useful
in cases where the test needs to cause a tool to generate an error (e.g. to
check the error output). However, any program in a pipeline that returns a
non-zero result will cause the test to fail. This script overcomes that
issue and nicely documents that the test case is purposefully ignoring the
result code of the tool</dd>
<dt><b>not</b></dt>
<dd>This script runs its arguments and then inverts the result code from
it. Zero result codes become 1. Non-zero result codes become 0. This is
useful to invert the result of a grep. For example "not grep X" means
succeed only if you don't find X in the input.</dd>
</dl>
<p>Sometimes it is necessary to mark a test case as "expected fail" or XFAIL.
You can easily mark a test as XFAIL just by including <tt>XFAIL: </tt> on a
line near the top of the file. This signals that the test case should succeed
if the test fails. Such test cases are counted separately by the testing tool. To
specify an expected fail, use the XFAIL keyword in the comments of the test
program followed by a colon and one or more regular expressions (separated by
a comma). The regular expressions allow you to XFAIL the test conditionally by
host platform. The regular expressions following the : are matched against the
target triplet for the host machine. If there is a match, the test is expected
to fail. If not, the test is expected to succeed. To XFAIL everywhere just
specify <tt>XFAIL: *</tt>. Here is an example of an <tt>XFAIL</tt> line:</p>
<div class="doc_code">
<pre>
; XFAIL: darwin,sun
</pre>
</div>
<p>To make the output more useful, the <tt>llvm_runtest</tt> function wil
scan the lines of the test case for ones that contain a pattern that matches
PR[0-9]+. This is the syntax for specifying a PR (Problem Report) number that
is related to the test case. The number after "PR" specifies the LLVM bugzilla
number. When a PR number is specified, it will be used in the pass/fail
reporting. This is useful to quickly get some context when a test fails.</p>
<p>Finally, any line that contains "END." will cause the special
interpretation of lines to terminate. This is generally done right after the
last RUN: line. This has two side effects: (a) it prevents special
interpretation of lines that are part of the test program, not the
instructions to the test case, and (b) it speeds things up for really big test
cases by avoiding interpretation of the remainder of the file.</p>
</div>
</div>
<!--=========================================================================-->
<h2><a name="testsuitestructure">Test suite Structure</a></h2>
<!--=========================================================================-->
<div>
<p>The <tt>test-suite</tt> module contains a number of programs that can be compiled
with LLVM and executed. These programs are compiled using the native compiler
and various LLVM backends. The output from the program compiled with the
native compiler is assumed correct; the results from the other programs are
compared to the native program output and pass if they match.</p>
<p>When executing tests, it is usually a good idea to start out with a subset of
the available tests or programs. This makes test run times smaller at first and
later on this is useful to investigate individual test failures. To run some
test only on a subset of programs, simply change directory to the programs you
want tested and run <tt>gmake</tt> there. Alternatively, you can run a different
test using the <tt>TEST</tt> variable to change what tests or run on the
selected programs (see below for more info).</p>
<p>In addition for testing correctness, the <tt>test-suite</tt> directory also
performs timing tests of various LLVM optimizations. It also records
compilation times for the compilers and the JIT. This information can be
used to compare the effectiveness of LLVM's optimizations and code
generation.</p>
<p><tt>test-suite</tt> tests are divided into three types of tests: MultiSource,
SingleSource, and External.</p>
<ul>
<li><tt>test-suite/SingleSource</tt>
<p>The SingleSource directory contains test programs that are only a single
source file in size. These are usually small benchmark programs or small
programs that calculate a particular value. Several such programs are grouped
together in each directory.</p></li>
<li><tt>test-suite/MultiSource</tt>
<p>The MultiSource directory contains subdirectories which contain entire
programs with multiple source files. Large benchmarks and whole applications
go here.</p></li>
<li><tt>test-suite/External</tt>
<p>The External directory contains Makefiles for building code that is external
to (i.e., not distributed with) LLVM. The most prominent members of this
directory are the SPEC 95 and SPEC 2000 benchmark suites. The <tt>External</tt>
directory does not contain these actual tests, but only the Makefiles that know
how to properly compile these programs from somewhere else. The presence and
location of these external programs is configured by the test-suite
<tt>configure</tt> script.</p></li>
</ul>
<p>Each tree is then subdivided into several categories, including applications,
benchmarks, regression tests, code that is strange grammatically, etc. These
organizations should be relatively self explanatory.</p>
<p>Some tests are known to fail. Some are bugs that we have not fixed yet;
others are features that we haven't added yet (or may never add). In the
regression tests, the result for such tests will be XFAIL (eXpected FAILure).
In this way, you can tell the difference between an expected and unexpected
failure.</p>
<p>The tests in the test suite have no such feature at this time. If the
test passes, only warnings and other miscellaneous output will be generated. If
a test fails, a large &lt;program&gt; FAILED message will be displayed. This
will help you separate benign warnings from actual test failures.</p>
</div>
<!--=========================================================================-->
<h2><a name="testsuiterun">Running the test suite</a></h2>
<!--=========================================================================-->
<div>
<p>First, all tests are executed within the LLVM object directory tree. They
<i>are not</i> executed inside of the LLVM source tree. This is because the
test suite creates temporary files during execution.</p>
<p>To run the test suite, you need to use the following steps:</p>
<ol>
<li><tt>cd</tt> into the <tt>llvm/projects</tt> directory in your source tree.
</li>
<li><p>Check out the <tt>test-suite</tt> module with:</p>
<div class="doc_code">
<pre>
% svn co http://llvm.org/svn/llvm-project/test-suite/trunk test-suite
</pre>
</div>
<p>This will get the test suite into <tt>llvm/projects/test-suite</tt>.</p>
</li>
<li><p>Configure and build <tt>llvm</tt>.</p></li>
<li><p>Configure and build <tt>llvm-gcc</tt>.</p></li>
<li><p>Install <tt>llvm-gcc</tt> somewhere.</p></li>
<li><p><em>Re-configure</em> <tt>llvm</tt> from the top level of
each build tree (LLVM object directory tree) in which you want
to run the test suite, just as you do before building LLVM.</p>
<p>During the <em>re-configuration</em>, you must either: (1)
have <tt>llvm-gcc</tt> you just built in your path, or (2)
specify the directory where your just-built <tt>llvm-gcc</tt> is
installed using <tt>--with-llvmgccdir=$LLVM_GCC_DIR</tt>.</p>
<p>You must also tell the configure machinery that the test suite
is available so it can be configured for your build tree:</p>
<div class="doc_code">
<pre>
% cd $LLVM_OBJ_ROOT ; $LLVM_SRC_ROOT/configure [--with-llvmgccdir=$LLVM_GCC_DIR]
</pre>
</div>
<p>[Remember that <tt>$LLVM_GCC_DIR</tt> is the directory where you
<em>installed</em> llvm-gcc, not its src or obj directory.]</p>
</li>
<li><p>You can now run the test suite from your build tree as follows:</p>
<div class="doc_code">
<pre>
% cd $LLVM_OBJ_ROOT/projects/test-suite
% make
</pre>
</div>
</li>
</ol>
<p>Note that the second and third steps only need to be done once. After you
have the suite checked out and configured, you don't need to do it again (unless
the test code or configure script changes).</p>
<!-- _______________________________________________________________________ -->
<h3>
<a name="testsuiteexternal">Configuring External Tests</a>
</h3>
<!-- _______________________________________________________________________ -->
<div>
<p>In order to run the External tests in the <tt>test-suite</tt>
module, you must specify <i>--with-externals</i>. This
must be done during the <em>re-configuration</em> step (see above),
and the <tt>llvm</tt> re-configuration must recognize the
previously-built <tt>llvm-gcc</tt>. If any of these is missing or
neglected, the External tests won't work.</p>
<dl>
<dt><i>--with-externals</i></dt>
<dt><i>--with-externals=&lt;<tt>directory</tt>&gt;</i></dt>
</dl>
This tells LLVM where to find any external tests. They are expected to be
in specifically named subdirectories of &lt;<tt>directory</tt>&gt;.
If <tt>directory</tt> is left unspecified,
<tt>configure</tt> uses the default value
<tt>/home/vadve/shared/benchmarks/speccpu2000/benchspec</tt>.
Subdirectory names known to LLVM include:
<dl>
<dt>spec95</dt>
<dt>speccpu2000</dt>
<dt>speccpu2006</dt>
<dt>povray31</dt>
</dl>
Others are added from time to time, and can be determined from
<tt>configure</tt>.
</div>
<!-- _______________________________________________________________________ -->
<h3>
<a name="testsuitetests">Running different tests</a>
</h3>
<!-- _______________________________________________________________________ -->
<div>
<p>In addition to the regular "whole program" tests, the <tt>test-suite</tt>
module also provides a mechanism for compiling the programs in different ways.
If the variable TEST is defined on the <tt>gmake</tt> command line, the test system will
include a Makefile named <tt>TEST.&lt;value of TEST variable&gt;.Makefile</tt>.
This Makefile can modify build rules to yield different results.</p>
<p>For example, the LLVM nightly tester uses <tt>TEST.nightly.Makefile</tt> to
create the nightly test reports. To run the nightly tests, run <tt>gmake
TEST=nightly</tt>.</p>
<p>There are several TEST Makefiles available in the tree. Some of them are
designed for internal LLVM research and will not work outside of the LLVM
research group. They may still be valuable, however, as a guide to writing your
own TEST Makefile for any optimization or analysis passes that you develop with
LLVM.</p>
</div>
<!-- _______________________________________________________________________ -->
<h3>
<a name="testsuiteoutput">Generating test output</a>
</h3>
<!-- _______________________________________________________________________ -->
<div>
<p>There are a number of ways to run the tests and generate output. The most
simple one is simply running <tt>gmake</tt> with no arguments. This will
compile and run all programs in the tree using a number of different methods
and compare results. Any failures are reported in the output, but are likely
drowned in the other output. Passes are not reported explicitely.</p>
<p>Somewhat better is running <tt>gmake TEST=sometest test</tt>, which runs
the specified test and usually adds per-program summaries to the output
(depending on which sometest you use). For example, the <tt>nightly</tt> test
explicitely outputs TEST-PASS or TEST-FAIL for every test after each program.
Though these lines are still drowned in the output, it's easy to grep the
output logs in the Output directories.</p>
<p>Even better are the <tt>report</tt> and <tt>report.format</tt> targets
(where <tt>format</tt> is one of <tt>html</tt>, <tt>csv</tt>, <tt>text</tt> or
<tt>graphs</tt>). The exact contents of the report are dependent on which
<tt>TEST</tt> you are running, but the text results are always shown at the
end of the run and the results are always stored in the
<tt>report.&lt;type&gt;.format</tt> file (when running with
<tt>TEST=&lt;type&gt;</tt>).
The <tt>report</tt> also generate a file called
<tt>report.&lt;type&gt;.raw.out</tt> containing the output of the entire test
run.
</div>
<!-- _______________________________________________________________________ -->
<h3>
<a name="testsuitecustom">Writing custom tests for the test suite</a>
</h3>
<!-- _______________________________________________________________________ -->
<div>
<p>Assuming you can run the test suite, (e.g. "<tt>gmake TEST=nightly report</tt>"
should work), it is really easy to run optimizations or code generator
components against every program in the tree, collecting statistics or running
custom checks for correctness. At base, this is how the nightly tester works,
it's just one example of a general framework.</p>
<p>Lets say that you have an LLVM optimization pass, and you want to see how
many times it triggers. First thing you should do is add an LLVM
<a href="ProgrammersManual.html#Statistic">statistic</a> to your pass, which
will tally counts of things you care about.</p>
<p>Following this, you can set up a test and a report that collects these and
formats them for easy viewing. This consists of two files, a
"<tt>test-suite/TEST.XXX.Makefile</tt>" fragment (where XXX is the name of your
test) and a "<tt>test-suite/TEST.XXX.report</tt>" file that indicates how to
format the output into a table. There are many example reports of various
levels of sophistication included with the test suite, and the framework is very
general.</p>
<p>If you are interested in testing an optimization pass, check out the
"libcalls" test as an example. It can be run like this:<p>
<div class="doc_code">
<pre>
% cd llvm/projects/test-suite/MultiSource/Benchmarks # or some other level
% make TEST=libcalls report
</pre>
</div>
<p>This will do a bunch of stuff, then eventually print a table like this:</p>
<div class="doc_code">
<pre>
Name | total | #exit |
...
FreeBench/analyzer/analyzer | 51 | 6 |
FreeBench/fourinarow/fourinarow | 1 | 1 |
FreeBench/neural/neural | 19 | 9 |
FreeBench/pifft/pifft | 5 | 3 |
MallocBench/cfrac/cfrac | 1 | * |
MallocBench/espresso/espresso | 52 | 12 |
MallocBench/gs/gs | 4 | * |
Prolangs-C/TimberWolfMC/timberwolfmc | 302 | * |
Prolangs-C/agrep/agrep | 33 | 12 |
Prolangs-C/allroots/allroots | * | * |
Prolangs-C/assembler/assembler | 47 | * |
Prolangs-C/bison/mybison | 74 | * |
...
</pre>
</div>
<p>This basically is grepping the -stats output and displaying it in a table.
You can also use the "TEST=libcalls report.html" target to get the table in HTML
form, similarly for report.csv and report.tex.</p>
<p>The source for this is in test-suite/TEST.libcalls.*. The format is pretty
simple: the Makefile indicates how to run the test (in this case,
"<tt>opt -simplify-libcalls -stats</tt>"), and the report contains one line for
each column of the output. The first value is the header for the column and the
second is the regex to grep the output of the command for. There are lots of
example reports that can do fancy stuff.</p>
</div>
</div>
<!-- *********************************************************************** -->
<hr>
<address>
<a href="http://jigsaw.w3.org/css-validator/check/referer"><img
src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
<a href="http://validator.w3.org/check/referer"><img
src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
John T. Criswell, Daniel Dunbar, Reid Spencer, and Tanya Lattner<br>
<a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Last modified: $Date$
</address>
</body>
</html>