This document is the reference manual for the LLVM test suite. It documents the structure of the LLVM test suite, the tools needed to use it, and how to add and run tests.
In order to use the LLVM test suite, you will need all of the software required to build LLVM, plus the following:
Darwin (Mac OS X) developers can simplify the installation of Expect and tcl by using fink. fink install expect will install both. Alternatively, Darwinports users can use sudo port install expect to install Expect and tcl.
The tests are located in two separate CVS modules. The basic feature and regression tests are in the main "llvm" module under the directory llvm/test. A more comprehensive test suite that includes whole programs in C and C++ is in the llvm-test module. This module should be checked out to the llvm/projects directory. When you configure the llvm module, the llvm-test module will be automatically configured. Alternatively, you can configure the llvm-test module manually.
To run all of the simple tests in LLVM using DejaGNU, use the master Makefile in the llvm/test directory:
% gmake -C llvm/testor
% gmake check
To run only a subdirectory of tests in llvm/test using DejaGNU (ie. Regression/Transforms), just set the TESTSUITE variable to the path of the subdirectory (relative to llvm/test):
% gmake -C llvm/test TESTSUITE=Regression/Transforms
Note: If you are running the tests with objdir != subdir, you must have run the complete testsuite before you can specify a subdirectory.
To run the comprehensive test suite (tests that compile and execute whole programs), run the llvm-test tests:
% cd llvm/projects % cvs co llvm-test % cd llvm-test % ./configure --with-llvmsrc=$LLVM_SRC_ROOT --with-llvmobj=$LLVM_OBJ_ROOT % gmake
The LLVM test suite contains two major categories of tests: code fragments and whole programs. Code fragments are in the llvm module under the llvm/test directory. The whole programs test suite is in the llvm-test module under the main directory.
Code fragments 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.
Code fragments are not complete programs, and they are never executed to determine correct behavior.
These code fragment tests are located in the llvm/test/Features and llvm/test/Regression directories.
Whole Programs 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.
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.
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.
All "whole program" tests are located in the llvm-test CVS module.
Each type of test in the LLVM test suite has its own directory. The major subtrees of the test suite directory tree are as follows:
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:
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.
The llvm-test CVS module contains 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.
In addition for testing correctness, the llvm-test 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.
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.
The MultiSource directory contains subdirectories which contain entire programs with multiple source files. Large benchmarks and whole applications go here.
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 presence and location of these external programs is configured by the llvm-test configure script.
The LLVM test suite is partially driven by DejaGNU and partially driven by GNU Make. Specifically, the Features and Regression tests are all driven by DejaGNU. The llvm-test module is currently driven by a set of Makefiles.
The DejaGNU structure is very simple, but does require some information to be set. This information is gathered via configure and is written to a file, site.exp in llvm/test. The llvm/test Makefile does this work for you.
In order for DejaGNU to work, each directory of tests must have a dg.exp file. DejaGNU 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. It simply loads a Tcl library (test/lib/llvm.exp) and calls the llvm_runtests 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 dg.exp file.
The llvm-runtests function lookas at each file that is passed to it and gathers any lines together that match "RUN:". This 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 llvm-runtests function will issue an error and the test will fail.
RUN lines are specified in the comments of the test program using the keyword RUN followed by a colon, and lastly the command (pipeline) to execute. Together, these lines form the "script" that llvm-runtests 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 look like a shell script, they are not. RUN lines are interpreted directly by the Tcl exec 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.
Each RUN line is executed on its own, distinct from other lines unless its last character is \. 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 \ are concatenated until a RUN line that doesn't end in \ is found. This concatenated set or 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.
Below is an example of legal RUN lines in a .ll file:
; RUN: llvm-as < %s | llvm-dis > %t1 ; RUN: llvm-dis < %s.bc-13 > %t2 ; RUN: diff %t1 %t2
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 Tcl exec command and the tutorial. The major differences are:
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 ' or " so they will get passed to the invoked program. For example:
... | grep 'find this string'
This will fail because the ' characters are passed to grep. This would instruction grep to look for 'find in the files this and string'. To avoid this use curly braces to tell Tcl that it should treat everything enclosed as one value. So our example would become:
... | grep {find this string}
Additionally, the characters [ and ] 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:
... | grep bb[2-8]
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:
... | grep {bb\[2-8\]}
Finally, if you need to pass the \ character down to a program, then it must be doubled. This is another Tcl special character. So, suppose you had:
... | grep 'i32\*'
This will fail to match what you want (a pointer to i32). First, the ' do not get stripped off. Second, the \ gets stripped off by Tcl so what grep sees is: 'i32*'. That's not likely to match anything. To resolve this you must use \\ and the {}, like this:
... | grep {i32\\*}
With a RUN line there are a number of substitutions that are permitted. In general, any Tcl variable that is available in the substitute function (in test/lib/llvm.exp) 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.
Here are the available variable names. The alternate syntax is listed in parentheses.To add more variables, two things need to be changed. First, add a line in the test/Makefile that creates the site.exp file. This will "set" the variable as a global in the site.exp file. Second, in the test/lib/llvm.exp 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.
To make RUN line writing easier, there are several shell scripts located in the llvm/test/Scripts directory. For example:
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 XFAIL: 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 DejaGnu. 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 or llvmgcc version number 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 XFAIL: *. When matching the llvm-gcc version, you can specify the major (e.g. 3) or full version (i.e. 3.4) number. Here is an example of an XFAIL line:
; XFAIL: darwin,sun,llvmgcc4
To make the output more useful, the llvm_runtest 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 numer 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.
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.
As mentioned previously, the llvm-test module provides three types of tests: MultiSource, SingleSource, and External. 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.
In addition to the regular "whole program" tests, the llvm-test module also provides a mechanism for compiling the programs in different ways. If the variable TEST is defined on the gmake command line, the test system will include a Makefile named TEST.<value of TEST variable>.Makefile. This Makefile can modify build rules to yield different results.
For example, the LLVM nightly tester uses TEST.nightly.Makefile to create the nightly test reports. To run the nightly tests, run gmake TEST=nightly.
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.
Note, when configuring the llvm-test module, you might want to specify the following configuration options:
First, all tests are executed within the LLVM object directory tree. They are not executed inside of the LLVM source tree. This is because the test suite creates temporary files during execution.
The master Makefile in llvm/test is capable of running only the DejaGNU driven tests. By default, it will run all of these tests.
To run only the DejaGNU driven tests, run gmake at the command line in llvm/test. To run a specific directory of tests, use the TESTSUITE variable.
For example, to run the Regression tests, type gmake TESTSUITE=Regression in llvm/tests.
Note that there are no Makefiles in llvm/test/Features and llvm/test/Regression. You must use DejaGNU from the llvm/test directory to run them.
To run the llvm-test suite, you need to use the following steps:
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).
To make a specialized test (use one of the
llvm-test/TEST.<type>.Makefiles), just run:
gmake TEST=<type> test
For example, you could run the
nightly tester tests using the following commands:
% cd llvm/projects/llvm-test % gmake TEST=nightly test
Regardless of which test you're running, the results are printed on standard output and standard error. You can redirect these results to a file if you choose.
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 DejaGNU, the result for such tests will be XFAIL (eXpected FAILure). In this way, you can tell the difference between an expected and unexpected failure.
The tests in llvm-test 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 <program> FAILED message will be displayed. This will help you separate benign warnings from actual test failures.
Assuming you can run llvm-test, (e.g. "gmake TEST=nightly report" 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.
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 statistic to your pass, which will tally counts of things you care about.
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, an "llvm-test/TEST.XXX.Makefile" fragment (where XXX is the name of your test) and an "llvm-test/TEST.XXX.report" file that indicates how to format the output into a table. There are many example reports of various levels of sophistication included with llvm-test, and the framework is very general.
If you are interested in testing an optimization pass, check out the "libcalls" test as an example. It can be run like this:
% cd llvm/projects/llvm-test/MultiSource/Benchmarks # or some other level % make TEST=libcalls report
This will do a bunch of stuff, then eventually print a table like this:
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 | * | ...
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.
The source for this is in llvm-test/TEST.libcalls.*. The format is pretty simple: the Makefile indicates how to run the test (in this case, "opt -simplify-libcalls -stats"), 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.
The LLVM Nightly Testers automatically check out an LLVM tree, build it, run the "nightly" program test (described above), run all of the feature and regression tests, delete the checked out tree, and then submit the results to http://llvm.org/nightlytest/. After test results are submitted to http://llvm.org/nightlytest/, they are processed and displayed on the tests page. An email to llvm-testresults@cs.uiuc.edu summarizing the results is also generated. This testing scheme is designed to ensure that programs don't break as well as keep track of LLVM's progress over time.
If you'd like to set up an instance of the nightly tester to run on your machine, take a look at the comments at the top of the utils/NewNightlyTest.pl file. If you decide to set up a nightly tester please choose a unique nickname and invoke utils/NewNightlyTest.pl with the "-nickname [yournickname]" command line option. We usually run it from a crontab entry that looks like this:
5 3 * * * $HOME/llvm/utils/NewNightlyTest.pl -parallel -nickname Nickname \
$CVSROOT $HOME/buildtest $HOME/cvs/testresults
Or, you can create a shell script to encapsulate the running of the script. The optimized x86 Linux nightly test is run from just such a script:
#!/bin/bash BASE=/proj/work/llvm/nightlytest export CVSROOT=:pserver:anon@llvm.org:/var/cvs/llvm export BUILDDIR=$BASE/build export WEBDIR=$BASE/testresults export LLVMGCCDIR=/proj/work/llvm/cfrontend/install export PATH=/proj/install/bin:$LLVMGCCDIR/bin:$PATH export LD_LIBRARY_PATH=/proj/install/lib cd $BASE cp /proj/work/llvm/llvm/utils/NewNightlyTest.pl . nice ./NewNightlyTest.pl -nice -release -verbose -parallel -enable-linscan \ -nickname NightlyTester -noexternals 2>&1 > output.log
It is also possible to specify the the location your nightly test results are submitted. You can do this by passing the command line option "-submit-server [server_address]" and "-submit-script [script_on_server]" to utils/NewNightlyTest.pl. For example, to submit to the llvm.org nightly test results page, you would invoke the nightly test script with "-submit-server llvm.org -submit-script /nightlytest/NightlyTestAccept.cgi". If these options are not specified, the nightly test script sends the results to the llvm.org nightly test results page.
Take a look at the NewNightlyTest.pl file to see what all of the flags and strings do. If you start running the nightly tests, please let us know. Thanks!