1 /*! \page faq Frequently Asked Questions
3 \htmlinclude .FAQ.doc.toc
5 \section faq_simgrid I'm new to SimGrid. I have some questions. Where should I start?
7 You are at the right place... Having a look to these
8 <a href="http://www.loria.fr/~quinson/articles/simgrid-tutorial.pdf">the tutorial slides</a>
9 (or to these <a href="http://graal.ens-lyon.fr/~alegrand/articles/slides_g5k_simul.pdf">old slides</a>,
11 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">"obsolete" slides</a>)
12 may give you some insights on what SimGrid can help you to do and what
13 are its limitations. Then you definitely should read the \ref
14 MSG_examples. The \ref GRAS_tut can also help you.
16 If you are stuck at any point and if this FAQ cannot help you, please drop us a
17 mail to the user mailing list: <simgrid-user@lists.gforge.inria.fr>.
19 \subsection faq_interfaces What is the difference between MSG, SimDag, and GRAS? Do they serve the same purpose?
21 It depend on how you define "purpose", I guess ;)
23 They all allow you to build a prototype of application which you can run
24 within the simulator afterward. They all share the same simulation kernel,
25 which is the core of the SimGrid project. They differ by the way you express
28 With SimDag, you express your code as a collection of interdependent
29 parallel tasks. So, in this model, applications can be seen as a DAG of
30 tasks. This is the interface of choice for people wanting to port old
31 code designed for SimGrid v1 or v2 to the framework current version.
33 With both GRAS and MSG, your application is seen as a set of communicating
34 processes, exchanging data by the way of messages and performing computation
37 The difference between both is that MSG is somehow easier to use, but GRAS
38 is not limited to the simulator. Once you're done writing your GRAS code,
39 you can run your code both in the simulator or on a real platform. For this,
40 there is two implementations of the GRAS interface, one for simulation, one
41 for real execution. So, you just have to relink your code to chose one of
44 \subsection faq_generic First steps with SimGrid
46 If you decide to go for the MSG interface, please read carefully the
47 \ref MSG_examples. You'll find in \ref MSG_ex_master_slave a very
48 simple consisting of a master (that owns a bunch of tasks and
49 distributes them) , some slaves (that process tasks whenever they
50 receive one) and some forwarder agents (that simply pass the tasks
51 they receive to some slaves).
53 If you decide to go for the GRAS interface, you should definitively
54 read the \ref GRAS_tut. The first section constitutes an introduction
55 to the tool and presents the model we use. The second section
56 constitutes a complete step-by-step tutorial building a distributed
57 application from the beginning and exemplifying most of the GRAS
58 features in the process. The last section groups some HOWTOS
59 highlighting a given feature of the framework in a more concise way.
61 If you decide to go for another interface, I'm afraid your only sources
62 of information will be the source code and the mailing lists...
64 \subsection faq_visualization Visualizing and analyzing the results
66 It is sometime convenient to "see" how the agents are behaving. If you
67 like colors, you can use <tt>tools/MSG_visualization/colorize.pl </tt>
68 as a filter to your MSG outputs. It works directly with INFO. Beware,
69 INFO() prints on stderr. Do not forget to redirect if you want to
70 filter (e.g. with bash):
72 ./msg_test small_platform.xml small_deployment.xml 2>&1 | ../../tools/MSG_visualization/colorize.pl
75 We also have a more graphical output. Have a look at section \ref faq_tracing.
77 \subsection faq_C Argh! Do I really have to code in C?
79 Up until now, there is no binding for other languages. If you use C++,
80 you should be able to use the SimGrid library as a standard C library
81 and everything should work fine (simply <i>link</i> against this
82 library; recompiling SimGrid with a C++ compiler won't work and it
83 wouldn't help if you could).
85 In fact, we are currently working on Java bindings of MSG to allow
86 all the undergrad students of the world to use this tool. This is a
87 little more tricky than I would have expected, but the work is moving
88 fast forward [2006/05/13]. More languages are evaluated, but for now,
89 we do not feel a real demand for any other language. Please speak up!
91 \section faq_cmake Installing the SimGrid library with Cmake (since V3.4)
93 \subsection faq_intro Some generalitty
95 \subsubsection faq_intro1 What is Cmake?
97 CMake is a family of tools designed to build, test and package software. CMake is used to control the software compilation process using simple platform and compiler independent configuration files. CMake generates native makefiles and workspaces that can be used in the compiler environment of your choice. For more information see official web site <a href="http://www.cmake.org/">here</a>.
99 \subsubsection faq_intro2 Why cmake?
101 CMake permits to developers to compil projects on different plateforms. Then many tools are embedded like ctest for making test, a link to cdash for vizualise results but also test coverage and bug reports.
103 \subsubsection faq_intro3 What cmake need?
105 CMake needs some prerequists like :
110 \li c, c++ and java compiler regards to developers
111 \li ccmake for graphical used of CMake
112 \li cmake <a href="http://www.cmake.org/cmake/resources/software.html">(download page)</a>
115 \li cmake 2.8.3 <a href="http://www.cmake.org/files/v2.8/cmake-2.8.3-win32-x86.exe">(download page)</a>
116 \li perl strawberry <a href="http://www.strawberryperl.com/download/5.12.2.0/strawberry-perl-5.12.2.0.msi">(download page)</a>
117 \li pcre-7.0 <a href="http://sourceforge.net/projects/gnuwin32/files/pcre/7.0/pcre-7.0.exe/download">(download page)</a>
119 \subsubsection faq_cmakeoption1 Liste of options
122 "cmake -D[name]=[value] ... ./"
124 [name] enable_gtnets [value] ON/OFF or TRUE/FALSE or 1/0
125 enable_java ON/OFF or TRUE/FALSE or 1/0
126 enable_lua ON/OFF or TRUE/FALSE or 1/0
127 enable_ruby ON/OFF or TRUE/FALSE or 1/0
128 enable_compile_optimizations ON/OFF or TRUE/FALSE or 1/0
129 enable_compile_warnings ON/OFF or TRUE/FALSE or 1/0
130 enable_smpi ON/OFF or TRUE/FALSE or 1/0
131 enable_maintainer_mode ON/OFF or TRUE/FALSE or 1/0
132 enable_supernovae ON/OFF or TRUE/FALSE or 1/0
133 enable_tracing ON/OFF or TRUE/FALSE or 1/0
134 enable_coverage ON/OFF or TRUE/FALSE or 1/0
135 enable_memcheck ON/OFF or TRUE/FALSE or 1/0
136 enable_model-checking ON/OFF or TRUE/FALSE or 1/0
137 gtnets_path <path_to_gtnets_directory>
138 CMAKE_INSTALL_PREFIX <path_to_install_directory>
139 pipol_user <pipol_username>
142 \subsubsection faq_cmakeoption2 Options explaination
144 \li enable_gtnets : set to true implie that user wants to use gtnets.
146 \li enable_java : set to true implie that user wants to add java langage into simgrid compilation.
148 \li enable_lua : set to true implie that user wants to add lua langage into simgrid compilation.
150 \li enable_ruby : set to true implie that user wants to add ruby langage into simgrid compilation.
152 \li enable_compile_optimizations : add flags "-O3 -finline-functions -funroll-loops -fno-strict-aliasing"
154 \li enable_compile_warnings : add flags "-Wall -Wunused -Wmissing-prototypes -Wmissing-declarations -Wpointer-arith -Wchar-subscripts -Wcomment -Wformat -Wwrite-strings -Wno-unused-function -Wno-unused-parameter -Wno-strict-aliasing -Wno-format-nonliteral -Werror"
156 \li enable_smpi : Set to true if you want to use smpi lib. Actually on simgrid v3.4.1 Mac doesn't support lib smpi.
158 \li enable_maintainer_mode : set to true it remakes some files.
160 \li enable_supernovae : set to true make one file for each lib and compile with those generated files.
162 \li enable_tracing : To enable the generation of simulation traces for visualization
164 \li enable_coverage : When set to true this option enable code coverage by setting -fprofile-arcs -ftest-coverage flags.
166 \li enable_memcheck : When set to true this option enable tests for memcheck.
168 \li enable_model-checking : Enable the model checking when set to true.
170 \li gtnets_path : Path to gtnets install directory (ex /usr)
172 \li CMAKE_INSTALL_PREFIX : Path where are installed lib/ doc/ and include/ directories (ex /usr/local)
174 \li pipol_user : specify your pipol username if you want to use the pipol-remote command.
176 \subsubsection faq_cmakeoption3 Initialisation
178 Those options are initialized the first time you launch "cmake ." whithout specified option.
188 enable_compile_optimizations on
189 enable_compile_warnings off
190 enable_maintainer_mode off
193 enable_model-checking off
194 CMAKE_INSTALL_PREFIX /usr/local
199 \subsubsection faq_cmakeoption4 Option's cache and how to reset?
201 When options have been set they are keep into a cache file named "CMakeCache.txt". So if you want
202 reset values you just delete this file located to the project directory.
204 \subsection faq_cmakecompilation Cmake compilation
206 \subsubsection faq_cmakecompilation1 With command line.
209 cmake -D[name]=[value] ... ./
216 cmake -G"Unix Makefiles" -D[name]=[value] ... ./
220 \subsubsection faq_cmakecompilation2 With ccmake tool.
225 Then follow instructions.
227 \subsubsection faq_cmakecompilation2bis Build out of source.
229 As cmake generate many files used for compilation, we recommand to make a build directory.
230 For examples you can make :
233 "navarrop@caraja:~/Developments$ cd simgrid/"
234 "navarrop@caraja:~/Developments/simgrid$ mkdir build_directory"
235 "navarrop@caraja:~/Developments/simgrid$ cd build_directory/"
236 "navarrop@caraja:~/Developments/simgrid/build_directory$ cmake ../"
237 "navarrop@caraja:~/Developments/simgrid/build_directory$ make"
240 Or complety out of sources :
243 "navarrop@caraja:~/Developments$ mkdir build_dir"
244 "navarrop@caraja:~/Developments$ cd build_dir/"
245 "navarrop@caraja:~/Developments/build_dir$ cmake ../simgrid/"
246 "navarrop@caraja:~/Developments/build_dir$ make"
249 Those two kind of compilation permit to delete files created by compilation easier.
251 \subsubsection faq_cmakecompilation3 Resume of command line
255 cmake <path> configure the project
256 make build all targets
257 make VERBOSE=1 build all targets and print build command lines
258 make check test all targets and summarize
259 make dist make the distrib
260 make distcheck check the dist (make + make dist + make check)
261 make install install the project (doc/ bin/ lib/ include/)
262 make uninstall uninstall the project (doc/ bin/ lib/ include/)
263 make clean clean all targets
264 make simgrid_documentation Create simgrid documentation
267 When the project have been succesfully compiling and build you can make tests.
271 ctest launch only tests
273 ctest -D Continuous(Start|Update|Configure|Build)
274 ctest -D Continuous(Test|Coverage|MemCheck|Submit)
275 ctest -D Experimental
276 ctest -D Experimental(Start|Update|Configure|Build)
277 ctest -D Experimental(Test|Coverage|MemCheck|Submit)
279 ctest -D Nightly(Start|Update|Configure|Build)
280 ctest -D Nightly(Test|Coverage|MemCheck|Submit)
281 ctest -D NightlyMemoryCheck
284 If you want to test before make a commit you can simply make "ctest -D Experimental" and then you can visualize results submitted into Cdash. <a href="http://cdash.inria.fr/CDash/index.php?project=Simgrid">(Go to Cdash site)</a>.
286 \subsection faq_cmakeinstall How to install with cmake?
288 \subsubsection faq_cmakeinstall1 From svn.
292 cmake -Denable_maintainer_mode=on -DCMAKE_INSTALL_PREFIX=/home/navarrop/Bureau/install_simgrid ./
300 cmake -G"Unix Makefiles" -DCMAKE_INSTALL_PREFIX=C:\simgrid_install ./
305 \subsubsection faq_cmakeinstall2 From a distrib
308 For version 3.4.1 and 3.4
309 cmake -Dprefix=/home/navarrop/Bureau/install_simgrid ./
313 cmake -DCMAKE_INSTALL_PREFIX=/home/navarrop/Bureau/install_simgrid ./
318 \subsection faq_cmakeWHATisInstall What is installed by cmake?
320 \subsubsection faq_cmakeWHATisInstallBIN CMAKE_INSTALL_PREFIX/bin
332 \subsubsection faq_cmakeWHATisInstallDOC CMAKE_INSTALL_PREFIX/doc
337 \subsubsection faq_cmakeWHATisInstallINCLUDE CMAKE_INSTALL_PREFIX/include
353 \subsubsection faq_cmakeWHATisInstallLIB CMAKE_INSTALL_PREFIX/lib
358 libsimgrid.so -> libsimgrid.so.3.5
359 libgras.so -> libgras.so.3.5
360 libsmpi.so -> libsmpi.so.3.5
361 lua/5.1/simgrid.so -> ../../libsimgrid.so
362 ruby/1.9.0/x86_64-linux/libsimgrid.so -> ../../../libsimgrid.so
363 ruby/1.9.0/x86_64-linux/simgrid.rb
365 \subsection faq_cmakehowto How to modified sources files for developers
367 \subsubsection faq_cmakehowto1 Add an executable or examples.
369 If you want make an executable you have to create a CMakeList.txt to the src directory.
370 You must specified where to create the executable, source list, dependencies and the name of the binary.
373 cmake_minimum_required(VERSION 2.6)
375 set(EXECUTABLE_OUTPUT_PATH "./")
376 set(LIBRARY_OUTPUT_PATH "${CMAKE_HOME_DIRECTORY}/lib")
378 add_executable(get_sender get_sender.c) #add_executable(<name_of_target> <src list>)
380 ### Add definitions for compile
381 target_link_libraries(get_sender simgrid m pthread) #target_link_libraries(<name_of_targe> <dependencies>)
384 Then you have to modified <project/directory>/buildtools/Cmake/MakeExeLib.cmake and add
387 add_subdirectory(${CMAKE_HOME_DIRECTORY}/<path_where_is_CMakeList.txt>)
390 \subsubsection faq_cmakehowto2 Delete/add sources to lib.
392 If you want modified, add or delete source files from a library you have to edit <project/directory>/buildtools/Cmake/DefinePackages.cmake
396 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/MsgException.java
397 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/JniException.java
398 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/NativeException.java
399 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/HostNotFoundException.java
400 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/ProcessNotFoundException.java
401 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Msg.java
402 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Process.java
403 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Host.java
404 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Task.java
405 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/MsgNative.java
406 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/ApplicationHandler.java
407 ${CMAKE_HOME_DIRECTORY}/src/java/simgrid/msg/Sem.java
411 \subsubsection faq_cmakehowto3 Add test
413 If you want modified, add or delete tests you have to edit <project/directory>/buildtools/Cmake/AddTests.cmake
414 with this function : ADD_TEST(<name> <bin> <ARGS>)
417 add_test(test-simdag-1 ${CMAKE_HOME_DIRECTORY}/testsuite/simdag/sd_test --cfg=path:${CMAKE_HOME_DIRECTORY}/testsuite/simdag small_platform_variable.xml)
420 \subsection faq_PIPOL Pipol-remote
422 Now we offer the possibility to test your local sources on pipol platforms before a commit. Of course you have to be user of pipol <a href="https://pipol.inria.fr/users/">(Account request)</a> cause you need to give your pipol_username to cmake. Here is a list of available systems :
424 amd64_kvm-linux-debian-lenny
425 amd64_kvm-linux-debian-testing
427 amd64-linux-centos-5.dd.gz
428 amd64-linux-debian-etch.dd.gz
429 amd64-linux-debian-lenny.dd.gz
430 amd64-linux-debian-testing.dd.gz
431 amd64-linux-fedora-core10.dd.gz
432 amd64-linux-fedora-core11.dd.gz
433 amd64-linux-fedora-core12.dd.gz
434 amd64-linux-fedora-core13.dd.gz
435 amd64-linux-fedora-core7.dd.gz
436 amd64-linux-fedora-core8.dd.gz
437 amd64-linux-fedora-core9.dd.gz
438 amd64-linux-mandriva-2007_springs_powerpack.dd.gz
439 amd64-linux-mandriva-2009_powerpack.dd.gz
440 amd64-linux-opensuse-11.dd.gz
441 amd64-linux-redhatEL-5.0.dd.gz
442 amd64-linux-suse-LES10.dd.gz
443 amd64-linux-ubuntu-feisty.dd.gz
444 amd64-linux-ubuntu-hardy.dd.gz
445 amd64-linux-ubuntu-intrepid.dd.gz
446 amd64-linux-ubuntu-jaunty.dd.gz
447 amd64-linux-ubuntu-karmic.dd.gz
448 amd64-linux-ubuntu-lucid.dd.gz
449 amd64-unix-freebsd-7.dd.gz
450 amd64-windows-server-2003-64bits.dd.gz
451 amd64-windows-server-2008-64bits.dd.gz
452 i386_kvm-linux-debian-lenny
453 i386_kvm-linux-debian-testing
454 i386_kvm-linux-fedora-core13
455 i386_kvm-windows-xp-pro-sp3
456 i386-linux-centos-5.dd.gz
457 i386-linux-debian-etch.dd.gz
458 i386-linux-debian-lenny.dd.gz
459 i386-linux-debian-testing.dd.gz
460 i386-linux-fedora-core10.dd.gz
461 i386-linux-fedora-core11.dd.gz
462 i386-linux-fedora-core12.dd.gz
463 i386-linux-fedora-core13.dd.gz
464 i386-linux-fedora-core7.dd.gz
465 i386-linux-fedora-core8.dd.gz
466 i386-linux-fedora-core9.dd.gz
467 i386-linux-mandriva-2007_springs_powerpack.dd.gz
468 i386-linux-mandriva-2009_powerpack.dd.gz
469 i386-linux-opensuse-11.dd.gz
470 i386-linux-redhatEL-5.0.dd.gz
471 i386-linux-suse-LES10.dd.gz
472 i386-linux-ubuntu-feisty.dd.gz
473 i386-linux-ubuntu-hardy.dd.gz
474 i386-linux-ubuntu-intrepid.dd.gz
475 i386-linux-ubuntu-jaunty.dd.gz
476 i386-linux-ubuntu-karmic.dd.gz
477 i386-linux-ubuntu-lucid.dd.gz
478 i386_mac-mac-osx-server-leopard.dd.gz
479 i386-unix-freebsd-7.dd.gz
480 i386-unix-opensolaris-10.dd.gz
481 i386-unix-opensolaris-11.dd.gz
482 i386-unix-solaris-10.dd.gz
483 ia64-linux-debian-lenny.dd
484 ia64-linux-fedora-core9.dd
485 ia64-linux-redhatEL-5.0.dd
486 x86_64_mac-mac-osx-server-snow-leopard.dd.gz
487 x86_mac-mac-osx-server-snow-leopard.dd.gz
490 Two kind of uses are possible :
492 This command copy your source and execute a configure then a build and finish with tests.
493 bob@caraja:~/Developments/simgrid/tmp_build$ make <name_of_image>
495 This command copy your source and execute a \"ctest -D Experimental\" and submit the result to cdash.
496 bob@caraja:~/Developments/simgrid/tmp_build$ make <name_of_image>_experimental
498 All commands are resumed with :
500 bob@caraja:~/Developments/simgrid/tmp_build$ make pipol_experimental_list_images
501 bob@caraja:~/Developments/simgrid/tmp_build$ make pipol_test_list_images
504 \section faq_installation Installing the SimGrid library with Autotools (valid until V3.3.4)
506 Many people have been asking me questions on how to use SimGrid. Quite
507 often, the questions were not really about SimGrid but on the
508 installation process. This section is intended to help people that are
509 not familiar with compiling C files under UNIX. If you follow these
510 instructions and still have some troubles, drop an e-mail to
511 <simgrid-user@lists.gforge.inria.fr>.
513 \subsection faq_compiling Compiling SimGrid from a stable archive
515 First of all, you need to download the latest version of SimGrid from
516 <a href="http://gforge.inria.fr/frs/?group_id=12">here</a>.
517 Suppose you have uncompressed SimGrid in some temporary location of
518 your home directory (say <tt>/home/joe/tmp/simgrid-3.0.1 </tt>). The
519 simplest way to use SimGrid is to install it in your home
520 directory. Change your directory to
521 <tt>/home/joe/tmp/simgrid-3.0.1</tt> and type
524 ./configure --prefix=$HOME
529 If at some point, something fails, check the section \ref faq_trouble_compil .
530 If it does not help, you can report this problem to the
531 list but, please, avoid sending a laconic mail like "There is a problem. Is it
532 okay?". Send the config.log file which is automatically generated by
533 configure. Try to capture both the standard output and the error output of the
534 <tt>make</tt> command with <tt>script</tt>. There is no way for us to help you
535 without the relevant bits of information.
537 Now, the following directory should have been created :
539 \li <tt>/home/joe/doc/simgrid/html/</tt>
540 \li <tt>/home/joe/lib/</tt>
541 \li <tt>/home/joe/include/</tt>
543 SimGrid is not a binary, it is a library. Both a static and a dynamic
544 version are available. Here is what you can find if you try a <tt>ls
547 \verbatim libsimgrid.a libsimgrid.la libsimgrid.so libsimgrid.so.0 libsimgrid.so.0.0.1
550 Thus, there is two ways to link your program with SimGrid:
551 \li Either you use the static version, e.g
552 \verbatim gcc libsimgrid.a -o MainProgram MainProgram.c
554 In this case, all the SimGrid functions are directly
555 included in <tt>MainProgram</tt> (hence a bigger binary).
556 \li Either you use the dynamic version (the preferred method)
557 \verbatim gcc -lsimgrid -o MainProgram MainProgram.c
559 In this case, the SimGrid functions are not included in
560 <tt>MainProgram</tt> and you need to set your environment
561 variable in such a way that <tt>libsimgrid.so</tt> will be
562 found at runtime. This can be done by adding the following
563 line in your .bashrc (if you use bash and if you have
564 installed the SimGrid libraries in your home directory):
565 \verbatim export LD_LIBRARY_PATH=$HOME/lib/:$LD_LIBRARY_PATH
568 \subsection faq_compiling_java Java bindings don't get compiled
570 The configure script detects automatically whether you have the
571 softwares needed to use the Java bindings or not. At the end of the
572 configure, you can see the configuration picked by the script, which
573 should look similar to
574 \verbatim Configuration of package simgrid' (version 3.3.4-svn) on
577 Compiler: gcc (version: )
579 CFlags: -O3 -finline-functions -funroll-loops -fno-strict-aliasing -Wall -Wunused -Wmissing-prototypes -Wmissing-declarations -Wpointer-arith -Wchar-subscripts -Wcomment -Wformat -Wwrite-strings -Wno-unused-function -Wno-unused-parameter -Wno-strict-aliasing -Wno-format-nonliteral -Werror -g3
583 Context backend: ucontext
590 In this example, Java backends won't be compiled.
592 On Debian-like systems (which includes ubuntu), you need the following
593 packages: sun-java6-jdk libgcj10-dev. If you cannot find the
594 libgcj10-dev, try another version, like libgcj9-dev (on Ubuntu before
595 9.10) or libgcj11-dev (not released yet, but certainly one day).
596 Please note that you need to activate the contrib and non-free
597 repositories in Debian, and the universe ones in Ubuntu. Java comes at
600 \subsection faq_compiling_snapshoot SimGrid development snapshots
602 We have very high standards on software quality, and we are reluctant releasing
603 a stable release as long as there is still some known bug in the code base. In
604 addition, we added quite an extensive test base, making sure that we correctly
605 test the most important parts of the tool.
607 As an unfortunate conclusion, there may be some time between the stable
608 releases. If you want to benefit from the most recent features we introduced,
609 but don't want to take the risk of an untested version from the SVN, then
610 development snapshots are done for you.
612 These are pre-releases of SimGrid that still fail some tests about features
613 that almost nobody use, or on platforms not being in our core target (which is
614 Linux, Mac, other Unixes and Windows, from the most important to the less
615 one). That means that using this development releases should be safe for most
618 These archives can be found on
619 <a href="http://www.loria.fr/~quinson/simgrid.html">this web page</a>. Once you
620 got the lastest archive, you can compile it just like any archive (see above).
622 \subsection faq_compiling_svn Compiling SimGrid from the SVN
624 The project development takes place in the SVN, where all changes are
625 committed when they happen. Then every once in a while, we make sure that the
626 code quality meets our standard and release an archive from the code in the
627 SVN. We afterward go back to the development in the SVN. So, if you need a
628 recently added feature and can afford some little problem with the stability
629 of the lastest features, you may want to use the SVN version instead of a
632 For that, you first need to get the "simgrid" module from
633 <a href="http://gforge.inria.fr/scm/?group_id=12">here</a>.
635 You won't find any <tt>configure</tt> and a few other things
636 (<tt>Makefile.in</tt>'s, documentation, ...) will be missing as well. The
637 reason for that is that all these files have to be regenerated using the
638 latest versions of <tt>autoconf</tt>, <tt>libtool</tt>, <tt>automake</tt>
639 (>1.9) and <tt>doxygen</tt> (>1.4). To generate the <tt>configure</tt> and
640 the <tt>Makefile.in</tt>'s, you just have to launch the <tt>bootstrap</tt>
641 command that resides in the top of the source tree. Then just follow the
642 instructions of Section \ref faq_compiling.
644 We insist on the fact that you really need the latest versions of
645 autoconf, automake and libtool. Doing this step on exotic architectures/systems
646 (i.e. anything different from a recent linux distribution) may be
647 ... uncertain. If you need to compile the SVN version on a machine where all these
648 dependencies are not met, the easiest is to do <tt>make dist</tt> in the SVN
649 directory of another machine where all dependencies are met. It will create an
650 archive you may deploy on other sites just as a regular stable release.
652 In summary, the following commands will checkout the SVN, regenerate the
653 configure script and friends, configure SimGrid and build it.
655 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk simgrid
658 ./configure --enable-maintainer-mode --prefix=<where to install SimGrid>
661 Then, if you want to install SimGrid on the current box, just do:
662 \verbatim make install \endverbatim
664 If you want to build an snapshot of the SVN to deploy it on another box (for
665 example because the other machine don't have the autotools), do:
666 \verbatim make dist \endverbatim
668 Moreover, you should never call the autotools manually since you must run
669 them in a specific order with specific arguments. Most of the times, the
670 makefiles will automatically call the tools for you. When it's not possible
671 (such as the first time you checkout the SVN), use the ./bootstrap command
672 to call them explicitly.
675 \subsection faq_setting_MSG Setting up your own MSG code
677 Do not build your simulator by modifying the SimGrid examples. Go
678 outside the SimGrid source tree and create your own working directory
679 (say <tt>/home/joe/SimGrid/MyFirstScheduler/</tt>).
681 Suppose your simulation has the following structure (remember it is
682 just an example to illustrate a possible way to compile everything;
683 feel free to organize it as you want).
685 \li <tt>sched.h</tt>: a description of the core of the
686 scheduler (i.e. which functions are can be used by the
687 agents). For example we could find the following functions
688 (master, forwarder, slave).
690 \li <tt>sched.c</tt>: a C file including <tt>sched.h</tt> and
691 implementing the core of the scheduler. Most of these
692 functions use the MSG functions defined in section \ref
695 \li <tt>masterslave.c</tt>: a C file with the main function, i.e.
696 the MSG initialization (MSG_global_init()), the platform
697 creation (e.g. with MSG_create_environment()), the
698 deployment phase (e.g. with MSG_function_register() and
699 MSG_launch_application()) and the call to
702 To compile such a program, we suggest to use the following
703 Makefile. It is a generic Makefile that we have used many times with
704 our students when we teach the C language.
708 masterslave: masterslave.o sched.o
710 INSTALL_PATH = $$HOME
712 PEDANTIC_PARANOID_FREAK = -O0 -Wshadow -Wcast-align \
713 -Waggregate-return -Wmissing-prototypes -Wmissing-declarations \
714 -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations \
715 -Wmissing-noreturn -Wredundant-decls -Wnested-externs \
716 -Wpointer-arith -Wwrite-strings -finline-functions
717 REASONABLY_CAREFUL_DUDE = -Wall
718 NO_PRAYER_FOR_THE_WICKED = -w -O2
719 WARNINGS = $(REASONABLY_CAREFUL_DUDE)
720 CFLAGS = -g $(WARNINGS)
722 INCLUDES = -I$(INSTALL_PATH)/include
723 DEFS = -L$(INSTALL_PATH)/lib/
724 LDADD = -lm -lsimgrid
728 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) $^ $(LIBS) $(LDADD) -o $@
731 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) -c -o $@ $<
734 rm -f $(BIN_FILES) *.o *~
740 The first two lines indicates what should be build when typing make
741 (<tt>masterslave</tt>) and of which files it is to be made of
742 (<tt>masterslave.o</tt> and <tt>sched.o</tt>). This makefile assumes
743 that you have set up correctly your <tt>LD_LIBRARY_PATH</tt> variable
744 (look, there is a <tt>LDADD = -lm -lsimgrid</tt>). If you prefer using
745 the static version, remove the <tt>-lsimgrid</tt> and add a
746 <tt>$(INSTALL_PATH)/lib/libsimgrid.a</tt> on the next line, right
747 after the <tt>LIBS = </tt>.
749 More generally, if you have never written a Makefile by yourself, type
750 in a terminal : <tt>info make</tt> and read the introduction. The
751 previous example should be enough for a first try but you may want to
752 perform some more complex compilations...
754 \subsection faq_setting_GRAS Setting up your own GRAS code
756 If you use the GRAS interface instead of the MSG one, then previous section
757 is not the better source of information. Instead, you should check the GRAS
758 tutorial in general, and the \ref GRAS_tut_tour_setup in particular.
760 \section faq_howto Feature related questions
762 \subsection faq_MIA "Could you please add (your favorite feature here) to SimGrid?"
764 Here is the deal. The whole SimGrid project (MSG, SURF, GRAS, ...) is
765 meant to be kept as simple and generic as possible. We cannot add
766 functions for everybody's needs when these functions can easily be
767 built from the ones already in the API. Most of the time, it is
768 possible and when it was not possible we always have upgraded the API
769 accordingly. When somebody asks us a question like "How to do that?
770 Is there a function in the API to simply do this?", we're always glad
771 to answer and help. However if we don't need this code for our own
772 need, there is no chance we're going to write it... it's your job! :)
773 The counterpart to our answers is that once you come up with a neat
774 implementation of this feature (task duplication, RPC, thread
775 synchronization, ...), you should send it to us and we will be glad to
776 add it to the distribution. Thus, other people will take advantage of
777 it (and we don't have to answer this question again and again ;).
779 You'll find in this section a few "Missing In Action" features. Many
780 people have asked about it and we have given hints on how to simply do
781 it with MSG. Feel free to contribute...
783 \subsection faq_MIA_MSG MSG features
785 \subsubsection faq_MIA_examples I want some more complex MSG examples!
787 Many people have come to ask me a more complex example and each time,
788 they have realized afterward that the basics were in the previous three
791 Of course they have often been needing more complex functions like
792 MSG_process_suspend(), MSG_process_resume() and
793 MSG_process_isSuspended() (to perform synchronization), or
794 MSG_task_Iprobe() and MSG_process_sleep() (to avoid blocking
795 receptions), or even MSG_process_create() (to design asynchronous
796 communications or computations). But the examples are sufficient to
799 We know. We should add some more examples, but not really some more
800 complex ones... We should add some examples that illustrate some other
801 functionalists (like how to simply encode asynchronous
802 communications, RPC, process migrations, thread synchronization, ...)
803 and we will do it when we will have a little bit more time. We have
804 tried to document the examples so that they are understandable. Tell
805 us if something is not clear and once again feel free to participate!
808 \subsubsection faq_MIA_taskdup Missing in action: MSG Task duplication/replication
810 There is no task duplication in MSG. When you create a task, you can
811 process it or send it somewhere else. As soon as a process has sent
812 this task, he doesn't have this task anymore. It's gone. The receiver
813 process has got the task. However, you could decide upon receiving to
814 create a "copy" of a task but you have to handle by yourself the
815 semantic associated to this "duplication".
817 As we already told, we prefer keeping the API as simple as
818 possible. This kind of feature is rather easy to implement by users
819 and the semantic you associate really depends on people. Having a
820 *generic* task duplication mechanism is not that trivial (in
821 particular because of the data field). That is why I would recommand
822 that you write it by yourself even if I can give you advice on how to
825 You have the following functions to get informations about a task:
826 MSG_task_get_name(), MSG_task_get_compute_duration(),
827 MSG_task_get_remaining_computation(), MSG_task_get_data_size(),
828 and MSG_task_get_data().
830 You could use a dictionary (#xbt_dict_t) of dynars (#xbt_dynar_t). If
831 you still don't see how to do it, please come back to us...
833 \subsubsection faq_MIA_asynchronous I want to do asynchronous communications in MSG
835 In the past (version <= 3.4), there was no function to perform asynchronous communications.
836 It could easily be implemented by creating new process when needed though. Since version 3.5,
837 we have introduced the following functions:
846 We refer you to the description of these functions for more details on their usage as well
847 as to the exemple section on \ref MSG_ex_asynchronous_communications.
849 \subsubsection faq_MIA_thread_synchronization I need to synchronize my MSG processes
851 You obviously cannot use pthread_mutexes of pthread_conds since we handle every
852 scheduling related decision within SimGrid.
854 In the past (version <=3.3.4) you could do it by playing with
855 MSG_process_suspend() and MSG_process_resume() or with fake communications (using MSG_task_get(),
856 MSG_task_put() and MSG_task_Iprobe()).
858 Since version 3.4, you can use classical synchronization structures. See page \ref XBT_synchro or simply check in
859 include/xbt/synchro_core.h.
861 \subsubsection faq_MIA_host_load Where is the get_host_load function hidden in MSG?
863 There is no such thing because its semantic wouldn't be really
864 clear. Of course, it is something about the amount of host throughput,
865 but there is as many definition of "host load" as people asking for
866 this function. First, you have to remember that resource availability
867 may vary over time, which make any load notion harder to define.
869 It may be instantaneous value or an average one. Moreover it may be only the
870 power of the computer, or may take the background load into account, or may
871 even take the currently running tasks into account. In some SURF models,
872 communications have an influence on computational power. Should it be taken
875 First of all, it's near to impossible to predict the load beforehands in the
876 simulator since it depends on too much parameters (background load
877 variation, bandwidth sharing algorithmic complexity) some of them even being
878 not known beforehands (other task starting at the same time). So, getting
879 this information is really hard (just like in real life). It's not just that
880 we want MSG to be as painful as real life. But as it is in some way
881 realistic, we face some of the same problems as we would face in real life.
883 How would you do it for real? The most common option is to use something
884 like NWS that performs active probes. The best solution is probably to do
885 the same within MSG, as in next code snippet. It is very close from what you
886 would have to do out of the simulator, and thus gives you information that
887 you could also get in real settings to not hinder the realism of your
891 double get_host_load() {
892 m_task_t task = MSG_task_create("test", 0.001, 0, NULL);
893 double date = MSG_get_clock();
895 MSG_task_execute(task);
896 date = MSG_get_clock() - date;
897 MSG_task_destroy(task);
902 Of course, it may not match your personal definition of "host load". In this
903 case, please detail what you mean on the mailing list, and we will extend
904 this FAQ section to fit your taste if possible.
906 \subsubsection faq_MIA_communication_time How can I get the *real* communication time?
908 Communications are synchronous and thus if you simply get the time
909 before and after a communication, you'll only get the transmission
910 time and the time spent to really communicate (it will also take into
911 account the time spent waiting for the other party to be
912 ready). However, getting the *real* communication time is not really
913 hard either. The following solution is a good starting point.
918 m_task_t task = MSG_task_create("Task", task_comp_size, task_comm_size,
919 calloc(1,sizeof(double)));
920 *((double*) task->data) = MSG_get_clock();
921 MSG_task_put(task, slaves[i % slaves_count], PORT_22);
922 XBT_INFO("Send completed");
927 m_task_t task = NULL;
930 time1 = MSG_get_clock();
931 a = MSG_task_get(&(task), PORT_22);
932 time2 = MSG_get_clock();
933 if(time1<*((double *)task->data))
934 time1 = *((double *) task->data);
935 XBT_INFO("Communication time : \"%f\" ", time2-time1);
937 MSG_task_destroy(task);
942 \subsection faq_MIA_SimDag SimDag related questions
944 \subsubsection faq_SG_comm Implementing communication delays between tasks.
946 A classic question of SimDag newcomers is about how to express a
947 communication delay between tasks. The thing is that in SimDag, both
948 computation and communication are seen as tasks. So, if you want to
949 model a data dependency between two DAG tasks t1 and t2, you have to
950 create 3 SD_tasks: t1, t2 and c and add dependencies in the following
954 SD_task_dependency_add(NULL, NULL, t1, c);
955 SD_task_dependency_add(NULL, NULL, c, t2);
958 This way task t2 cannot start before the termination of communication c
959 which in turn cannot start before t1 ends.
961 When creating task c, you have to associate an amount of data (in bytes)
962 corresponding to what has to be sent by t1 to t2.
964 Finally to schedule the communication task c, you have to build a list
965 comprising the workstations on which t1 and t2 are scheduled (w1 and w2
966 for example) and build a communication matrix that should look like
969 \subsubsection faq_SG_DAG How to implement a distributed dynamic scheduler of DAGs.
971 Distributed is somehow "contagious". If you start making distributed
972 decisions, there is no way to handle DAGs directly anymore (unless I
973 am missing something). You have to encode your DAGs in term of
974 communicating process to make the whole scheduling process
975 distributed. Here is an example of how you could do that. Assume T1
976 has to be done before T2.
979 int your_agent(int argc, char *argv[] {
981 T1 = MSG_task_create(...);
982 T2 = MSG_task_create(...);
986 if(cond) MSG_task_execute(T1);
988 if((MSG_task_get_remaining_computation(T1)=0.0) && (you_re_in_a_good_mood))
991 /* do something else */
997 If you decide that the distributed part is not that much important and that
998 DAG is really the level of abstraction you want to work with, then you should
999 give a try to \ref SD_API.
1001 \subsection faq_MIA_generic Generic features
1003 \subsubsection faq_more_processes Increasing the amount of simulated processes
1005 Here are a few tricks you can apply if you want to increase the amount
1006 of processes in your simulations.
1008 - <b>A few thousands of simulated processes</b> (soft tricks)\n
1009 SimGrid can use either pthreads library or the UNIX98 contextes. On
1010 most systems, the number of pthreads is limited and then your
1011 simulation may be limited for a stupid reason. This is especially
1012 true with the current linux pthreads, and I cannot get more than
1013 2000 simulated processes with pthreads on my box. The UNIX98
1014 contexts allow me to raise the limit to 25,000 simulated processes
1016 The <tt>--with-context</tt> option of the <tt>./configure</tt>
1017 script allows you to choose between UNIX98 contextes
1018 (<tt>--with-context=ucontext</tt>) and the pthread version
1019 (<tt>--with-context=pthread</tt>). The default value is ucontext
1020 when the script detect a working UNIX98 context implementation. On
1021 Windows boxes, the provided value is discarded and an adapted
1022 version is picked up.\n\n
1023 We experienced some issues with contextes on some rare systems
1024 (solaris 8 and lower or old alpha linuxes comes to mind). The main
1025 problem is that the configure script detect the contextes as being
1026 functional when it's not true. If you happen to use such a system,
1027 switch manually to the pthread version, and provide us with a good
1028 patch for the configure script so that it is done automatically ;)
1030 - <b>Hundred thousands of simulated processes</b> (hard-core tricks)\n
1031 As explained above, SimGrid can use UNIX98 contextes to represent
1032 and handle the simulated processes. Thanks to this, the main
1033 limitation to the number of simulated processes becomes the
1034 available memory.\n\n
1035 Here are some tricks I had to use in order to run a token ring
1036 between 25,000 processes on my laptop (1Gb memory, 1.5Gb swap).\n
1037 - First of all, make sure your code runs for a few hundreds
1038 processes before trying to push the limit. Make sure it's
1039 valgrind-clean, ie that valgrind does not report neither memory
1040 error nor memory leaks. Indeed, numerous simulated processes
1041 result in *fat* simulation hindering debugging.
1042 - It was really boring to write 25,000 entries in the deployment
1043 file, so I wrote a little script
1044 <tt>examples/gras/mutual_exclusion/simple_token/make_deployment.pl</tt>, which you may
1045 want to adapt to your case. You could also think about hijacking
1046 the SURFXML parser (have look at \ref faq_flexml_bypassing).
1047 - The deployment file became quite big, so I had to do what is in
1048 the FAQ entry \ref faq_flexml_limit
1049 - Each UNIX98 context has its own stack entry. As debugging this is
1050 quite hairly, the default value is a bit overestimated so that
1051 user don't get into trouble about this. You want to tune this
1052 size to increse the number of processes. This is the
1053 <tt>STACK_SIZE</tt> define in
1054 <tt>src/xbt/xbt_context_sysv.c</tt>, which is 128kb by default.
1055 Reduce this as much as you can, but be warned that if this value
1056 is too low, you'll get a segfault. The token ring example, which
1057 is quite simple, runs with 40kb stacks.
1058 - You may tweak the logs to reduce the stack size further. When
1059 logging something, we try to build the string to display in a
1060 char array on the stack. The size of this array is constant (and
1061 equal to XBT_LOG_BUFF_SIZE, defined in include/xbt/log/h). If the
1062 string is too large to fit this buffer, we move to a dynamically
1063 sized buffer. In which case, we have to traverse one time the log
1064 event arguments to compute the size we need for the buffer,
1065 malloc it, and traverse the argument list again to do the actual
1067 The idea here is to move XBT_LOG_BUFF_SIZE to 1, forcing the logs
1068 to use a dynamic array each time. This allows us to lower further
1069 the stack size at the price of some performance loss...\n
1070 This allowed me to run the reduce the stack size to ... 4k. Ie,
1071 on my 1Gb laptop, I can run more than 250,000 processes!
1073 \subsubsection faq_MIA_batch_scheduler Is there a native support for batch schedulers in SimGrid?
1075 No, there is no native support for batch schedulers and none is
1076 planned because this is a very specific need (and doing it in a
1077 generic way is thus very hard). However some people have implemented
1078 their own batch schedulers. Vincent Garonne wrote one during his PhD
1079 and put his code in the contrib directory of our SVN so that other can
1080 keep working on it. You may find inspiring ideas in it.
1082 \subsubsection faq_MIA_checkpointing I need a checkpointing thing
1084 Actually, it depends on whether you want to checkpoint the simulation, or to
1085 simulate checkpoints.
1087 The first one could help if your simulation is a long standing process you
1088 want to keep running even on hardware issues. It could also help to
1089 <i>rewind</i> the simulation by jumping sometimes on an old checkpoint to
1090 cancel recent calculations.\n
1091 Unfortunately, such thing will probably never exist in SG. One would have to
1092 duplicate all data structures because doing a rewind at the simulator level
1093 is very very hard (not talking about the malloc free operations that might
1094 have been done in between). Instead, you may be interested in the Libckpt
1095 library (http://www.cs.utk.edu/~plank/plank/www/libckpt.html). This is the
1096 checkpointing solution used in the condor project, for example. It makes it
1097 easy to create checkpoints (at the OS level, creating something like core
1098 files), and rerunning them on need.
1100 If you want to simulate checkpoints instead, it means that you want the
1101 state of an executing task (in particular, the progress made towards
1102 completion) to be saved somewhere. So if a host (and the task executing on
1103 it) fails (cf. #MSG_HOST_FAILURE), then the task can be restarted
1104 from the last checkpoint.\n
1106 Actually, such a thing does not exists in SimGrid either, but it's just
1107 because we don't think it is fundamental and it may be done in the user code
1108 at relatively low cost. You could for example use a watcher that
1109 periodically get the remaining amount of things to do (using
1110 MSG_task_get_remaining_computation()), or fragment the task in smaller
1113 \subsection faq_platform Platform building and Dynamic resources
1115 \subsubsection faq_platform_example Where can I find SimGrid platform files?
1117 There is several little examples in the archive, in the examples/msg
1118 directory. From time to time, we are asked for other files, but we
1119 don't have much at hand right now.
1121 You should refer to the Platform Description Archive
1122 (http://pda.gforge.inria.fr) project to see the other platform file we
1123 have available, as well as the Simulacrum simulator, meant to generate
1124 SimGrid platforms using all classical generation algorithms.
1126 \subsubsection faq_platform_alnem How can I automatically map an existing platform?
1128 We are working on a project called ALNeM (Application-Level Network
1129 Mapper) which goal is to automatically discover the topology of an
1130 existing network. Its output will be a platform description file
1131 following the SimGrid syntax, so everybody will get the ability to map
1132 their own lab network (and contribute them to the catalog project).
1133 This tool is not ready yet, but it move quite fast forward. Just stay
1136 \subsubsection faq_platform_synthetic Generating synthetic but realistic platforms
1138 The third possibility to get a platform file (after manual or
1139 automatic mapping of real platforms) is to generate synthetic
1140 platforms. Getting a realistic result is not a trivial task, and
1141 moreover, nobody is really able to define what "realistic" means when
1142 speaking of topology files. You can find some more thoughts on this
1144 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">slides</a>.
1146 If you are looking for an actual tool, there we have a little tool to
1147 annotate Tiers-generated topologies. This perl-script is in
1148 <tt>tools/platform_generation/</tt> directory of the SVN. Dinda et Al.
1149 released a very comparable tool, and called it GridG.
1151 \subsubsection faq_SURF_multicore Modeling multi-core resources
1153 There is currently no native support for multi-core or SMP machines in
1154 SimGrid. We are currently working on it, but coming up with the right
1155 model is very hard: Cores share caches and bus to access memory and
1156 thus interfere with each others. Memory contention is a crucial
1157 component of multi-core modeling.
1159 In the meanwhile, some user-level tricks can reveal sufficient for
1160 you. For example, you may model each core by a CPU and add some very
1161 high speed links between them. This complicates a bit the user code
1162 since you have to remember that when you assign something to a (real)
1163 host, it can be any of the (fake) hosts representing the cores of a
1164 given machine. For that, you can use the prop tag of the XML files as
1165 follows. Your code should then look at the ‘machine’ property
1166 associated with each workstation, and run parallel tasks over all
1167 cores of the machine.
1170 <host id="machine0/core0" power="91500E6">
1171 <prop id="machine" value="machine0"/>
1172 <prop id="core" value="0"/>
1174 <host id="machine0/core1" power="91500E6">
1175 <prop id="machine" value="machine0"/>
1176 <prop id="core" value="1"/>
1182 \subsubsection faq_SURF_dynamic Modeling dynamic resource availability
1184 A nice feature of SimGrid is that it enables you to seamlessly have
1185 resources whose availability change over time. When you build a
1186 platform, you generally declare hosts like that:
1189 <host id="host A" power="100.00"/>
1192 If you want the availability of "host A" to change over time, the only
1193 thing you have to do is change this definition like that:
1196 <host id="host A" power="100.00" availability_file="trace_A.txt" state_file="trace_A_failure.txt"/>
1199 For hosts, availability files are expressed in fraction of available
1200 power. Let's have a look at what "trace_A.txt" may look like:
1209 At time 0, our host will deliver 100 flop/s. At time 11.0, it will
1210 deliver only 50 flop/s until time 20.0 where it will will start
1211 delivering 90 flop/s. Last at time 21.0 (20.0 plus the periodicity
1212 1.0), we'll be back to the beginning and it will deliver 100 flop/s.
1214 Now let's look at the state file:
1221 A negative value means "off" while a positive one means "on". At time
1222 1.0, the host is on. At time 1.0, it is turned off and at time 2.0, it
1223 is turned on again until time 12 (2.0 plus the periodicity 10.0). It
1224 will be turned on again at time 13.0 until time 23.0, and so on.
1226 Now, let's look how the same kind of thing can be done for network
1227 links. A usual declaration looks like:
1230 <link id="LinkA" bandwidth="10.0" latency="0.2"/>
1233 You have at your disposal the following options: bandwidth_file,
1234 latency_file and state_file. The only difference with hosts is that
1235 bandwidth_file and latency_file do not express fraction of available
1236 power but are expressed directly in bytes per seconds and seconds.
1238 \subsubsection faq_platform_multipath How to express multipath routing in platform files?
1240 It is unfortunately impossible to express the fact that there is more
1241 than one routing path between two given hosts. Let's consider the
1242 following platform file:
1245 <route src="A" dst="B">
1248 <route src="B" dst="C">
1251 <route src="A" dst="C">
1256 Although it is perfectly valid, it does not mean that data traveling
1257 from A to C can either go directly (using link 3) or through B (using
1258 links 1 and 2). It simply means that the routing on the graph is not
1259 trivial, and that data do not following the shortest path in number of
1260 hops on this graph. Another way to say it is that there is no implicit
1261 in these routing descriptions. The system will only use the routes you
1262 declare (such as <route src="A" dst="C"><link:ctn
1263 id="3"/></route>), without trying to build new routes by aggregating
1266 You are also free to declare platform where the routing is not
1267 symmetric. For example, add the following to the previous file:
1270 <route src="C" dst="A">
1276 This makes sure that data from C to A go through B where data from A
1277 to C go directly. Don't worry about realism of such settings since
1278 we've seen ways more weird situation in real settings (in fact, that's
1279 the realism of very regular platforms which is questionable, but
1280 that's another story).
1282 \subsubsection faq_flexml_bypassing Bypassing the XML parser with your own C functions
1284 So you want to bypass the XML files parser, uh? Maybe doing some parameter
1285 sweep experiments on your simulations or so? This is possible, and
1286 it's not even really difficult (well. Such a brutal idea could be
1287 harder to implement). Here is how it goes.
1289 For this, you have to first remember that the XML parsing in SimGrid is done
1290 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
1291 you want to bypass the parser, you need to provide some code mimicking what
1292 it does and replacing it in its interactions with the SURF code. So, let's
1293 have a look at these interactions.
1295 FleXML parser are close to classical SAX parsers. It means that a
1296 well-formed SimGrid platform XML file might result in the following
1299 - start "platform_description" with attribute version="2"
1300 - start "host" with attributes id="host1" power="1.0"
1302 - start "host" with attributes id="host2" power="2.0"
1304 - start "link" with ...
1306 - start "route" with ...
1307 - start "link:ctn" with ...
1310 - end "platform_description"
1312 The communication from the parser to the SURF code uses two means:
1313 Attributes get copied into some global variables, and a surf-provided
1314 function gets called by the parser for each event. For example, the event
1315 - start "host" with attributes id="host1" power="1.0"
1317 let the parser do something roughly equivalent to:
1319 strcpy(A_host_id,"host1");
1324 In SURF, we attach callbacks to the different events by initializing the
1325 pointer functions to some the right surf functions. Since there can be
1326 more than one callback attached to the same event (if more than one
1327 model is in use, for example), they are stored in a dynar. Example in
1328 workstation_ptask_L07.c:
1330 /* Adding callback functions */
1331 surf_parse_reset_parser();
1332 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
1333 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
1334 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
1335 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
1336 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
1337 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
1339 /* Parse the file */
1340 surf_parse_open(file);
1341 xbt_assert(!surf_parse(), "Parse error in %s", file);
1345 So, to bypass the FleXML parser, you need to write your own version of the
1346 surf_parse function, which should do the following:
1347 - Fill the A_<tag>_<attribute> variables with the wanted values
1348 - Call the corresponding STag_<tag>_fun function to simulate tag start
1349 - Call the corresponding ETag_<tag>_fun function to simulate tag end
1350 - (do the same for the next set of values, and loop)
1352 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
1354 surf_parse = surf_parse_bypass_environment;
1355 MSG_create_environment(NULL);
1356 surf_parse = surf_parse_bypass_application;
1357 MSG_launch_application(NULL);
1360 A set of macros are provided at the end of
1361 include/surf/surfxml_parse.h to ease the writing of the bypass
1362 functions. An example of this trick is distributed in the file
1363 examples/msg/masterslave/masterslave_bypass.c
1365 \subsection faq_simgrid_configuration Changing SimGrid's behavior
1367 A number of options can be given at runtime to change the default
1368 SimGrid behavior. In particular, you can change the default cpu and
1371 \subsubsection faq_simgrid_configuration_fullduplex Using Fullduplex
1373 Experimental fullduplex support is now available on the svn branch. In order to fullduple to work your platform must have two links for each pair
1374 of interconnected hosts, see an example here:
1376 simgrid_svn_sources/exemples/msg/gtnets/fullduplex-p.xml
1379 Using fullduplex support ongoing and incoming communication flows are
1380 treated independently for most models. The exception is the LV08 model which
1381 adds 0.05 of usage on the opposite direction for each new created flow. This
1382 can be useful to simulate some important TCP phenomena such as ack compression.
1384 Running a fullduplex example:
1386 cd simgrid_svn_sources/exemples/msg/gtnets
1387 ./gtnets fullduplex-p.xml fullduplex-d.xml --cfg=fullduplex:1
1394 \subsubsection faq_simgrid_configuration_gtnets Using GTNetS
1396 It is possible to use a packet-level network simulator
1397 instead of the default flow-based simulation. You may want to use such
1398 an approach if you have doubts about the validity of the default model
1399 or if you want to perform some validation experiments. At the moment,
1400 we support the GTNetS simulator (it is still rather experimental
1401 though, so leave us a message if you play with it).
1405 To enable GTNetS model inside SimGrid it is needed to patch the GTNetS simulator source code
1406 and build/install it from scratch
1409 - <b>Download and enter the recent downloaded GTNetS directory</b>
1412 svn checkout svn://scm.gforge.inria.fr/svn/simgrid/contrib/trunk/GTNetS/
1417 - <b>Use the following commands to unzip and patch GTNetS package to work within SimGrid.</b>
1420 unzip gtnets-current.zip
1421 tar zxvf gtnets-current-patch.tgz
1423 cat ../00*.patch | patch -p1
1426 - <b>OPTIONALLY</b> you can use a patch for itanium 64bit processor family.
1429 cat ../AMD64-FATAL-Removed-DUL_SIZE_DIFF-Added-fPIC-compillin.patch | patch -p1
1432 - <b>Compile GTNetS</b>
1434 Due to portability issues it is possible that GTNetS does not compile in your architecture. The patches furnished in SimGrid SVN repository are intended for use in Linux architecture only. Unfortunately, we do not have the time, the money, neither the manpower to guarantee GTNetS portability. We advice you to use one of GTNetS communication channel to get more help in compiling GTNetS.
1438 ln -sf Makefile.linux Makefile
1444 - <b>NOTE</b> A lot of warnings are expected but the application should compile
1445 just fine. If the makefile insists in compiling some QT libraries
1446 please try a make clean before asking for help.
1449 - <b>To compile optimized version</b>
1456 - <b>Installing GTNetS</b>
1458 It is important to put the full path of your libgtsim-xxxx.so file when creating the symbolic link. Replace < userhome > by some path you have write access to.
1461 ln -sf /<absolute_path>/gtnets_current/libgtsim-debug.so /<userhome>/usr/lib/libgtnets.so
1462 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/<userhome>/usr/lib/libgtnets.so
1463 mkdir /<userhome>/usr/include/gtnets
1464 cp -fr SRC/*.h /<userhome>/usr/include/gtnets
1468 - <b>Enable GTNetS support in SimGrid</b>
1470 In order to enable gtnets with simgrid you have to give where is gtnets. (path to \<gtnets_path\>/lib and \<gtnets_path\>/include)
1473 Since v3.4 (with cmake)
1474 cmake . -Dgtnets_path=/<userhome>/usr
1476 Until v3.4 (with autotools)
1477 ./configure --with-gtnets=/<userhome>/usr
1480 - <b>Once you have followed all the instructions for compiling and
1481 installing successfully you can activate this feature at
1482 runntime with the following options:</b>
1485 Since v3.4 (with cmake)
1490 Until v3.4 (with autotools)
1491 cd simgrid/example/msg/
1497 - <b>Or try the GTNetS model dogbone example with</b>
1500 gtnets/gtnets gtnets/onelink-p.xml gtnets/onelink-d.xml --cfg=network_model:GTNets
1504 A long version of this <a href="http://gforge.inria.fr/docman/view.php/12/6283/GTNetS HowTo.html">HowTo</a> it is available
1507 More about GTNetS simulator at <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/index.html">GTNetS Website</a>
1511 The patches provided by us worked successfully with GTNetS found
1512 <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/software/gtnets-current.zip">here</a>,
1513 dated from 12th June 2008. Due to the discontinuing development of
1514 GTNetS it is impossible to precise a version number. We STRONGLY recommend you
1515 to download and install the GTNetS version found in SimGrid repository as explained above.
1520 \subsubsection faq_simgrid_configuration_alternate_network Using alternative flow models
1522 The default simgrid network model uses a max-min based approach as
1523 explained in the research report
1524 <a href="ftp://ftp.ens-lyon.fr/pub/LIP/Rapports/RR/RR2002/RR2002-40.ps.gz">A Network Model for Simulation of Grid Application</a>.
1525 Other models have been proposed and implemented since then (see for example
1526 <a href="http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf">Accuracy Study and Improvement of Network Simulation in the SimGrid Framework</a>)
1527 and can be activated at runtime. For example:
1529 ./mycode platform.xml deployment.xml --cfg=workstation/model:compound --cfg=network/model:LV08 -cfg=cpu/model:Cas01
1532 Possible models for the network are currently "Constant", "CM02",
1533 "LegrandVelho", "GTNets", Reno", "Reno2", "Vegas". Others will
1534 probably be added in the future and many of the previous ones are
1535 experimental and are likely to disappear without notice... To know the
1536 list of the currently implemented models, you should use the
1537 --help-models command line option.
1540 ./masterslave_forwarder ../small_platform.xml deployment_masterslave.xml --help-models
1541 Long description of the workstation models accepted by this simulator:
1542 CLM03: Default workstation model, using LV08 and CM02 as network and CPU
1543 compound: Workstation model allowing you to use other network and CPU models
1544 ptask_L07: Workstation model with better parallel task modeling
1545 Long description of the CPU models accepted by this simulator:
1546 Cas01_fullupdate: CPU classical model time=size/power
1547 Cas01: Variation of Cas01_fullupdate with partial invalidation optimization of lmm system. Should produce the same values, only faster
1548 CpuTI: Variation of Cas01 with also trace integration. Should produce the same values, only faster if you use availability traces
1549 Long description of the network models accepted by this simulator:
1550 Constant: Simplistic network model where all communication take a constant time (one second)
1551 CM02: Realistic network model with lmm_solve and no correction factors
1552 LV08: Realistic network model with lmm_solve and these correction factors: latency*=10.4, bandwidth*=.92, S=8775
1553 Reno: Model using lagrange_solve instead of lmm_solve (experts only)
1554 Reno2: Model using lagrange_solve instead of lmm_solve (experts only)
1555 Vegas: Model using lagrange_solve instead of lmm_solve (experts only)
1558 \subsection faq_tracing Tracing Simulations for Visualization
1560 The trace visualization is widely used to observe and understand the behavior
1561 of parallel applications and distributed algorithms. Usually, this is done in a
1562 two-step fashion: the user instruments the application and the traces are
1563 analyzed after the end of the execution. The visualization itself can highlights
1564 unexpected behaviors, bottlenecks and sometimes can be used to correct
1565 distributed algorithms. The SimGrid team has instrumented the library
1566 in order to let users trace their simulations and analyze them. This part of the
1567 user manual explains how the tracing-related features can be enabled and used
1568 during the development of simulators using the SimGrid library.
1570 \subsubsection faq_tracing_howitworks How it works
1572 For now, the SimGrid library is instrumented so users can trace the <b>platform
1573 utilization</b> using the MSG, SimDAG and SMPI interface. This means that the tracing will
1574 register how much power is used for each host and how much bandwidth is used for
1575 each link of the platform. The idea with this type of tracing is to observe the
1576 overall view of resources utilization in the first place, especially the
1577 identification of bottlenecks, load-balancing among hosts, and so on.
1579 The idea of the tracing facilities is to give SimGrid users to possibility to
1580 classify MSG and SimDAG tasks by category, tracing the platform utilization
1581 (hosts and links) for each of the categories. For that,
1582 the tracing interface enables the declaration of categories and a function to
1583 mark a task with a previously declared category. <em>The tasks that are not
1584 classified according to a category are not traced</em>. Even if the user
1585 does not specify any category, the simulations can still be traced in terms
1586 of resource utilization by using a special parameter that is detailed below.
1588 \subsubsection faq_tracing_enabling Enabling using CMake
1590 With the sources of SimGrid, it is possible to enable the tracing
1591 using the parameter <b>-Denable_tracing=ON</b> when the cmake is executed.
1592 The section \ref faq_tracing_functions describes all the functions available
1593 when this Cmake options is activated. These functions will have no effect
1594 if SimGrid is configured without this option (they are wiped-out by the
1598 $ cmake -Denable_tracing=ON .
1602 \subsubsection faq_tracing_functions Tracing Functions
1604 \li <b>\c TRACE_category (const char *category)</b>: This function should be used
1605 to define a user category. The category can be used to differentiate the tasks
1606 that are created during the simulation (for example, tasks from server1,
1607 server2, or request tasks, computation tasks, communication tasks).
1608 All resource utilization (host power and link bandwidth) will be
1609 classified according to the task category. Tasks that do not belong to a
1610 category are not traced.
1612 \li <b>\c TRACE_msg_set_task_category (m_task_t task, const char *category)</b>:
1613 This function should be called after the creation of a MSG task, to define the
1614 category of that task. The first parameter \c task must contain a task that was
1615 created with the function \c MSG_task_create. The second parameter
1616 \c category must contain a category that was previously defined by the function
1619 \li <b>\c TRACE_sd_set_task_category (SD_task_t task, const char *category)</b>:
1620 This function should be called after the creation of a SimDAG task, to define the
1621 category of that task. The first parameter \c task must contain a task that was
1622 created with the function \c MSG_task_create. The second parameter
1623 \c category must contain a category that was previously defined by the function
1626 \li <b>\c TRACE_host_variable_declare (const char *variable)</b>:
1627 Declare a user variable that will be associated to hosts. A variable can
1628 be used to trace user variables such as the number of tasks in a server,
1629 the number of clients in an application, and so on.
1631 \li <b>\c TRACE_host_variable_[set|add|sub] (const char *variable, double
1633 Set the value of a given user variable. It is important to remind that
1634 the value of this variable is always associated to the host. The host
1635 that will be used when these functions are called is the one returned by
1636 the function \c MSG_host_self().
1638 \subsubsection faq_tracing_options Tracing configuration Options
1640 These are the options accepted by the tracing system of SimGrid:
1645 It activates the tracing system and register the simulation platform
1646 in the trace file. You have to enable this option to others take effect.
1651 It activates the categorized resource utilization tracing. It should
1652 be enabled if tracing categories are used by this simulator.
1655 tracing/uncategorized
1657 It activates the uncategorized resource utilization tracing. Use it if
1658 this simulator do not use tracing categories and resource use have to be
1662 tracing/platform/method
1664 It changes the way resource utilization (categorized or not) is traced
1665 inside the simulation core. Method 'a' (default) traces all updates defined
1666 by the CPU/network model of a given resource. Depending on the interface used
1667 by this simulator (MSG, SMPI, SimDAG), the default method can generate large
1668 trace files. Method 'b' tries to make smaller tracefiles using clever updates,
1669 without losing details of resource utilization. Method 'c' generates even
1670 smaller files by doing time integration during the simulation, but it loses
1671 precision. If this last method is used, the smallest timeslice used in the
1672 tracefile analysis must be bigger than the smaller resource utilization. If
1673 unsure, do not change this option.
1678 A file with this name will be created to register the simulation. The file
1679 is in the Paje format and can be analyzed using Triva or Paje visualization
1680 tools. More information can be found in these webpages:
1681 <a href="http://triva.gforge.inria.fr/">http://triva.gforge.inria.fr/</a>
1682 <a href="http://paje.sourceforge.net/">http://paje.sourceforge.net/</a>
1687 This option only has effect if this simulator is SMPI-based. Traces the MPI
1688 interface and generates a trace that can be analyzed using Gantt-like
1689 visualizations. Every MPI function (implemented by SMPI) is transformed in a
1690 state, and point-to-point communications can be analyzed with arrows.
1695 This option only has effect if this simulator is SMPI-based. The processes
1696 are grouped by the hosts where they were executed.
1701 This option only has effect if this simulator is MSG-based. It traces the
1702 behavior of all categorized MSG tasks, grouping them by hosts.
1707 This option only has effect if this simulator is MSG-based. It traces the
1708 behavior of all categorized MSG processes, grouping them by hosts. This option
1709 can be used to track process location if this simulator has process migration.
1714 This experimental option only has effect if this simulator is MSG-based.
1715 It traces the communication volume of MSG send/receive.
1717 \subsubsection faq_tracing_example Example of Instrumentation
1719 A simplified example using the tracing mandatory functions.
1722 int main (int argc, char **argv)
1724 MSG_global_init (&argc, &argv);
1726 //note that TRACE_start must be called after MSG_global_init
1727 TRACE_category_with_color ("request", "1 0 0");
1728 TRACE_category_with_color ("computation", "0.3 1 0.4");
1729 TRACE_category ("finalize");
1731 //(... after deployment ...)
1733 m_task_t req1 = MSG_task_create("1st_request_task", 10, 10, NULL);
1734 m_task_t req2 = MSG_task_create("2nd_request_task", 10, 10, NULL);
1735 m_task_t req3 = MSG_task_create("3rd_request_task", 10, 10, NULL);
1736 m_task_t req4 = MSG_task_create("4th_request_task", 10, 10, NULL);
1737 TRACE_msg_set_task_category (req1, "request");
1738 TRACE_msg_set_task_category (req2, "request");
1739 TRACE_msg_set_task_category (req3, "request");
1740 TRACE_msg_set_task_category (req4, "request");
1742 m_task_t comp = MSG_task_create ("comp_task", 100, 100, NULL);
1743 TRACE_msg_set_task_category (comp, "computation");
1745 m_task_t finalize = MSG_task_create ("finalize", 0, 0, NULL);
1746 TRACE_msg_set_task_category (finalize, "finalize");
1755 \subsubsection faq_tracing_analyzing Analyzing the SimGrid Traces
1757 The SimGrid library, during an instrumented simulation, creates a trace file in
1758 the Paje file format that contains the platform utilization for the simulation
1759 that was executed. The visualization analysis of this file is performed with the
1760 visualization tool <a href="http://triva.gforge.inria.fr">Triva</a>, with
1761 special configurations tunned to SimGrid needs. This part of the documentation
1762 explains how to configure and use Triva to analyse a SimGrid trace file.
1764 - <b>Installing Triva</b>: the tool is available in the INRIAGforge,
1765 at <a href="http://triva.gforge.inria.fr">http://triva.gforge.inria.fr</a>.
1766 Use the following command to get the sources, and then check the file
1767 <i>INSTALL</i>. This file contains instructions to install
1768 the tool's dependencies in a Ubuntu/Debian Linux. The tool can also
1769 be compiled in MacOSes natively, check <i>INSTALL.mac</i> file.
1771 $ svn checkout svn://scm.gforge.inria.fr/svn/triva
1776 - <b>Executing Triva</b>: a binary called <i>Triva</i> is available after the
1777 installation (you can execute it passing <em>--help</em> to check its
1778 options). If the triva binary is not available after following the
1779 installation instructions, you may want to execute the following command to
1780 initialize the GNUstep environment variables. We strongly recommend that you
1781 use the latest GNUstep packages, and not the packages available through apt-get
1782 in Ubuntu/Debian packaging systems. If you install GNUstep using the latest
1783 available packages, you can execute this command:
1785 $ source /usr/GNUstep/System/Library/Makefiles/GNUstep.sh
1787 You should be able to see this output after the installation of triva:
1789 $ ./Triva.app/Triva --help
1790 Usage: Triva [OPTIONS...] TRACE0 [TRACE1]
1791 Trace Analysis through Visualization
1794 --ti_frequency {double} Animation: frequency of updates
1795 --ti_hide Hide the TimeInterval window
1796 --ti_forward {double} Animation: value to move time-slice
1797 --ti_apply Apply the configuration
1798 --ti_update Update on slider change
1799 --ti_animate Start animation
1800 --ti_start {double} Start of time slice
1801 --ti_size {double} Size of time slice
1803 --comparison Compare Trace Files (Experimental)
1804 --graph Configurable Graph
1805 --list Print Trace Type Hierarchy
1806 --hierarchy Export Trace Type Hierarchy (dot)
1807 --stat Trace Statistics and Memory Utilization
1808 --instances List All Trace Entities
1809 --linkview Link View (Experimental)
1810 --treemap Squarified Treemap
1811 --merge Merge Trace Files (Experimental)
1812 --check Check Trace File Integrity
1814 --gc_conf {file} Graph Configuration in Property List Format
1815 --gc_apply Apply the configuration
1816 --gc_hide Hide the GraphConfiguration window
1818 Triva expects that the user choose one of the available options
1819 (currently <em>--graph</em> or <em>--treemap</em> for a visualization analysis)
1820 and the trace file from the simulation.
1822 - <b>Understanding Triva - time-slice</b>: the analysis of a trace file using
1823 the tool always takes into account the concept of the <em>time-slice</em>.
1824 This concept means that what is being visualized in the screen is always
1825 calculated considering a specific time frame, with its beggining and end
1826 timestamp. The time-slice is configured by the user and can be changed
1827 dynamically through the window called <em>Time Interval</em> that is opened
1828 whenever a trace file is being analyzed. The next figure depicts the time-slice
1829 configuration window.
1830 In the top of the window, in the space named <i>Trace Time</i>,
1831 the two fields show the beggining of the trace (which usually starts in 0) and
1832 the end (that depends on the time simulated by SimGrid). The middle of the
1833 window, in the square named <i>Time Slice Configuration</i>, contains the
1834 aspects related to the time-slice, including its <i>start</i> and its
1835 <i>size</i>. The gray rectangle in the bottom of this part indicates the
1836 <i>current time-slice</i> that is considered for the drawings. If the checkbox
1837 <i>Update Drawings on Sliders Change</i> is not selected, the button
1838 <i>Apply</i> must be clicked in order to inform triva that the
1839 new time-slice must be considered. The bottom part of the window, in the space
1840 indicated by the square <i>Time Slice Animation</i> can be used to advance
1841 the time-frame automatically. The user configures the amount of time that the
1842 time-frame will forward and how frequent this update will happen. Once this is
1843 configured, the user clicks the <i>Play</i> button in order to see the dynamic
1844 changes on the drawings.
1847 <a href="triva-time_interval.png" border=0><img src="triva-time_interval.png" width="50%" border=0></a>
1850 <b>Remarks:</b> when the trace has too many hosts or links, the computation to
1851 take into account a new time-slice can be expensive. When this happens, the
1852 <i>Frequency</i> parameter, but also updates caused by change on configurations
1853 when the checkbox <i>Update Drawings on Sliders
1854 Change</i> is selected will not be followed.
1856 - <b>Understanding Triva - graph</b>: this part of the documention explains how
1857 to analyze the traces using the graph view of Triva, when the user executes
1858 the tool passing <em>--graph</em> as parameter. Triva opens three windows when
1859 this parameter is used: the <i>Time Interval</i> window (previously described),
1860 the <i>Graph Representation</i> window, and the <em>Graph Configuration</em>
1861 window. The Graph Representation is the window where drawings take place.
1862 Initially, it is completely white waiting for a proper graph configuration input
1863 by the user. We start the description of this type of analysis by describing the
1864 <i>Graph Configuration</i> window (depicted below). By using a particular
1865 configuration, triva
1866 can be used to customize the graph drawing according to
1867 the SimGrid trace that was created with user-specific categories. Before delving
1868 into the details of this customization, let us first explain the major parts of
1869 the graph configuration window. The buttons located in the top-right corner can
1870 be used to delete, copy and create a new configuration. The checkbox in the
1871 top-middle part of the window indicates if the configuration typed in the
1872 textfield is syntactically correct (we are using the non-XML
1873 <a href="http://en.wikipedia.org/wiki/Property_list">Property List Format</a> to
1874 describe the configuration). The pop-up button located on the top-left corner
1875 indicates the selected configuration (the user can have multiple graph
1876 configurations). The bottom-left text field contains the name of the current
1877 configuration (updates on this field must be followed by typing enter on the
1878 keyboard to take into account the name change). The bottom-right <em>Apply</em>
1879 button activates the current configuration, resulting on an update on the graph
1883 <a href="triva-graph_configuration.png" border=0><img src="triva-graph_configuration.png" width="50%" border=0></a>
1886 <b>Basic SimGrid Configuration</b>: The figure shows in the big textfield the
1887 basic configuration that should be used during the analysis of a SimGrid trace
1888 file. The basic logic of the configuration is as follows:
1894 The nodes of the graph will be created based on the <i>node</i> parameter, which
1895 in this case is the different <em>"HOST"</em>s of the platform
1896 used to simulate. The <i>edge</i> parameter indicates that the edges of the
1897 graph will be created based on the <em>"LINK"</em>s of the platform. After the
1898 definition of these two parameters, the configuration must detail how
1899 <em>HOST</em>s and <em>LINK</em>s should be drawn. For that, the configuration
1900 must have an entry for each of the types used. For <em>HOST</em>, as basic
1901 configuration, we have:
1908 The parameter <em>size</em> indicates which variable from the trace file will be
1909 used to define the size of the node HOST in the visualization. If the simulation
1910 was executed with availability traces, the size of the nodes will be changed
1911 according to these traces. The parameter <em>scale</em> indicates if the value
1912 of the variable is <em>global</em> or <em>local</em>. If it is global, the value
1913 will be relative to the power of all other hosts, if it is local, the value will
1914 be relative locally.
1915 For <em>LINK</em> we have:
1925 For the types specified in the <em>edge</em> parameter (such as <em>LINK</em>),
1926 the configuration must contain two additional parameters: <em>src</em> and
1927 <em>dst</em> that are used to properly identify which nodes this edge is
1928 connecting. The values <em>source</em> and <em>destination</em> are always present
1929 in the SimGrid trace file and should not be changed in the configuration. The
1930 parameter <em>size</em> for the LINK, in this case, is configured as the
1931 variable <em>bandwidth</em>, with a <em>global</em> scale. The scale meaning
1932 here is exactly the same used for nodes. The last parameter is the GraphViz
1933 algorithm used to calculate the position of the nodes in the graph
1936 graphviz-algorithm = neato;
1939 <b>Customizing the Graph Representation</b>: triva is capable to handle
1940 a customized graph representation based on the variables present in the trace
1941 file. In the case of SimGrid, every time a category is created for tasks, two
1942 variables in the trace file are defined: one to indicate node utilization (how
1943 much power was used by that task category), and another to indicate link
1944 utilization (how much bandwidth was used by that category). For instance, if the
1945 user declares a category named <i>request</i>, there will be variables named
1946 <b>p</b><i>request</i> and a <b>b</b><i>request</i> (<b>p</b> for power and
1947 <b>b</b> for bandwidth). It is important to notice that the variable
1948 <i>prequest</i> in this case is only available for HOST, and
1949 <i>brequest</i> is only available for LINK. <b>Example</b>: suppose there are
1950 two categories for tasks: request and compute. To create a customized graph
1951 representation with a proportional separation of host and link utilization, use
1952 as configuration for HOST and LINK this:
1961 values = (prequest, pcomputation);
1974 values = (brequest, bcomputation);
1978 Where <i>sep_host</i> contains a composition of type <i>separation</i> where
1979 its max size is the <i>power</i> of the host and the variables <i>prequest</i>
1980 and <i>pcomputation</i> are drawn proportionally to the size of the HOST. And
1981 <i>sep_link</i> is also a separation where max is defined as the
1982 <i>bandwidth</i> of the link, and the variables <i>brequest</i> and
1983 <i>bcomputation</i> are drawn proportionally within a LINK.
1984 <i>This configuration enables the analysis of resource utilization by MSG tasks,
1985 and the identification of load-balancing issues, network bottlenecks, for
1987 <b>Other compositions</b>: besides <i>separation</i>, it is possible to use
1988 other types of compositions, such as gradients, and colors, like this:
1993 values = (numberOfTasks);
1997 values = (is_server);
2000 Where <i>gra_host</i> creates a gradient within a node of the graph, using a
2001 global scale and using as value a variable called <i>numberOfTasks</i>, that
2002 could be declared by the user using the optional tracing functions of SimGrid.
2003 If scale is global, the max and min value for the gradient will be equal to the
2004 max and min numberOfTasks among all hosts, and if scale is local, the max and
2005 min value based on the value of numberOfTasks locally in each host.
2006 And <i>color_host</i> composition draws a square based on a positive value of
2007 the variable <i>is_server</i>, that could also be defined by the user using the
2008 SimGrid tracing functions. \n
2009 <b>The Graph Visualization</b>: The next figure shows a graph visualization of a
2010 given time-slice of the masterslave_forwarder example (present in the SimGrid
2011 sources). The red color indicates tasks from the <i>compute</i> category. This
2012 visualization was generated with the following configuration:
2025 values = (pcompute, pfinalize);
2037 values = (bcompute, bfinalize);
2040 graphviz-algorithm = neato;
2045 <a href="triva-graph_visualization.png" border=0><img src="triva-graph_visualization.png" width="50%" border=0></a>
2049 - <b>Understading Triva - colors</b>: An important issue when using Triva is how
2050 to define colors. To do that, we have to know which variables are defined in
2051 the trace file generated by the SimGrid library. The parameter <em>--list</em>
2052 lists the variables for a given trace file:
2054 $ Triva -l masterslave_forwarder.trace
2072 We can see that HOST has seven variables (from power to pfinalize) and LINK has
2073 four (from bandwidth to bfinalize). To define a red color for the
2074 <i>pcompute</i> and <i>bcompute</i> (which are defined based on user category
2075 <i>compute</i>), execute:
2077 $ defaults write Triva 'pcompute Color' '1 0 0'
2078 $ defaults write Triva 'bcompute Color' '1 0 0'
2080 Where the three numbers in each line are the RGB color with values from 0 to 1.
2082 \subsection faq_modelchecking Model-Checking
2083 \subsubsection faq_modelchecking_howto How to use it
2084 To enable the experimental SimGrid model-checking support the program should
2085 be executed with the command line argument
2089 Properties are expressed as assertions using the function
2091 void MC_assert(int prop);
2094 \subsection faq_binding_lua Lua Binding
2095 Most of Simgrid modules require a good level in C programming, since simgrid is used to be as standard C library.
2096 Sometime users prefer using some kind of « easy scripts » or a language easier to code with, for their works,
2097 which avoid dealing with C errors, and sometime an important gain of time.
2098 Besides Java Binding, Lua and Ruby bindings are available since version 3.4 of Simgrid
2099 for MSG Module, and we are currenlty working on bindings for other modules.
2102 \subsubsection faq_binding_lua_about What is lua ?
2103 Lua is a lightweight, reflective, imperative and functional programming language,
2104 designed as a scripting language with extensible semantics as a primary goal (see official web site <a href="http://www.lua.org">here</a>).
2105 \subsubsection faq_binding_lua_why Why lua ?
2106 Lua is a fast, portable and powerful script language, quite simple to use for developpers.
2107 it combines procedural features with powerful data description facilities,
2108 by using a simple, yet powerful, mechanism of tables.
2109 Lua has a relatively simple C API compared to other scripting languages,
2110 and accordingly it provides a robust, easy to use it.
2111 \subsubsection faq_binding_lua_simgrid How to use lua in Simgrid ?
2112 Actually, the use of lua in Simgrid is quite simple, you have just to follow the same steps as coding with C in Simgird :
2113 - Coding functions coresponding to each process
2114 - loading the platforme/deployment XML file that describe the environment of simulation
2115 - and … Running the Simulation.
2117 \dontinclude lua/master_slave.lua
2118 \subsubsection faq_binding_lua_example_master_slave Master/Slave Example
2121 \until end_of_master
2122 we mainly use simgrid.Task.new(task_name,computation_size,communication_size) to create our MSG Task,
2123 then simgrid.Task.send(task,alias) to send it.
2124 we use also simgrid.Task.name(task), to get the task's name.
2128 Here, we see the use of simgrid.Task.recv(alias) to receive a task with a specific alias,
2129 this function return directly the task recevied.
2131 \li Set Environmenet and run application
2132 \until simgrid.clean()
2134 \subsubsection faq_binding_lua_example_data Exchanging Data
2135 You can also exchange data between Process using lua. for that, you have to deal with lua task as a table,
2136 since lua is based itself on a mechanism of tables,
2137 so you can exchange any kind of data (tables, matrix, strings,…) between process via tasks.
2141 task = simgrid.Task.new("data_task",task_comp,task_comm);
2142 task['matrix'] = my_matrix;
2143 task['table'] = my_table;
2144 task['message'] = "Hello from (Lua || Simgrid ) !! "
2146 simgrid.Task.send(task,alias)
2148 After creating task, we associate to it various kind of data with a specific key (string in this case)
2149 to distinguish between data variables. The receiver will use this key to access easily to datas.
2152 \li Receiver processe
2154 task = simgrid.Task.recv(alias);
2155 sender_matrix = task['matrix'];
2156 sender_table = task['table'];
2157 sender_message = task['message']
2160 Note that in lua, both sender and receiver share the same lua task.
2161 So that the receiver could joint data directly on the received task without sending it back.
2162 You can find a complet example (matrix multiplication case) in the file example/lua/mult_matrix.lua.
2165 \subsubsection faq_binding_lua_example_bypass Bypass XML
2166 maybe you wonder if there is a way to bypass the XML files,
2167 and describe your platform directly from the code, with lua bindings it's Possible !! how ?
2168 We provide some additional (tricky?) functions in lua that allows you to set up your own platform without using the XML files
2169 ( this can be useful for large platforms, so a simple for loop will avoid you to deal with an annoying XML File ;) )
2172 \li set Routing mode
2174 simgrid.AS.new{id="AS0",mode="Full"};
2179 simgrid.Host.new{id="Tremblay",power=98095000};
2180 simgrid.Host.new{id="Jupiter",power=76296000};
2181 simgrid.Host.new{id="Fafard",power=76296000};
2182 simgrid.Host.new{id="Ginette",power=48492000};
2183 simgrid.Host.new{id="Bourassa",power=48492000};
2185 we use simgrid.Host.new{id=id_host,power=power_host} to instanciate our hosts.
2190 simgrid.Link.new{id=i,bandwidth=252750+ i*768,latency=0.000270544+i*0.087}; -- some crazy values ;)
2193 we used simgrid.Link.new{id=link_id,bandwidth=bw,latency=lat} with a simple for loop to create all links we need (much easier than XML hein ?)
2197 -- simgrid.Route.new(src_id,des_id,links_nb,links_list)
2198 simgrid.Route.new("Tremblay","Jupiter",1,{"1"});
2199 simgrid.Route.new("Tremblay","Fafard",6,{"0","1","2","3","4","8"});
2200 simgrid.Route.new("Tremblay","Ginette",3,{"3","4","5"});
2201 simgrid.Route.new("Tremblay","Bourassa",7,{"0","1","3","2","4","6","7"});
2203 simgrid.Route.new("Jupiter","Tremblay",1,{"1"});
2204 simgrid.Route.new("Jupiter","Fafard",7,{"0","1","2","3","4","8","9"});
2205 simgrid.Route.new("Jupiter","Ginette",4,{"3","4","5","9"});
2206 simgrid.Route.new("Jupiter","Bourassa",8,{"0","1","2","3","4","6","7","9"});
2209 for each host you have to specify which route to choose to access to the rest of hosts connected in the grid.
2213 simgrid.register_platform();
2215 Don't forget to register your platform, that SURF callbacks starts their work ;)
2219 simgrid.Host.setFunction("Tremblay","Master",4,{"20","550000000","1000000","4"});
2220 simgrid.Host.setFunction("Bourassa","Slave",1,{"0"});
2221 simgrid.Host.setFunction("Jupiter","Slave",1,{"1"});
2222 simgrid.Host.setFunction("Fafard","Slave",1,{"2"});
2223 simgrid.Host.setFunction("Ginette","Slave",1,{"3"});
2225 you don't need to use a deployment XML file, thanks to simgrid.Host.setFunction(host_id,function,args_number,args_list)
2226 you can associate functions for each host with arguments if needed .
2230 simgrid.register_application();
2232 Yes, Here too you have to resgiter your application before running the simulation.
2234 the full example is distributed in the file examples/lua/master_slave_bypass.lua
2236 \subsection faq_binding_ruby Ruby Binding
2239 \subsubsection faq_binding_ruby_simgrid Use Ruby in Simgrid
2240 Since v3.4, the use of <a href="http://ruby-lang.org">ruby</a> in simgrid is available for the MSG Module.
2241 you can find almost all MSG functionalities in Ruby code, that allows you to set up your environment, manage tasks between hosts and run the simulation.
2243 \dontinclude ruby/MasterSlave.rb
2244 \subsubsection faq_binding_ruby_example Master/Slave Ruby Application
2245 for each process method(master and slave in this example), you have to associate a ruby class, that should inherit from <i>MSG::Process</i> ruby class,
2246 with a 'main' function that describe the behaviour of the process during the simulation.
2254 \until end_of_master
2256 the class MSG::Task contains methods that allows the management of the native MSG tasks.
2257 in master ruby code we used :
2258 - <i>MSG::Task.new(task_name,compute_size,communication_size)</i> : to instanciate a new task.
2259 - <i>MSG::Task.send(mailbox)</i> : to send the task via a mailbox alias.
2260 - <i>MSG::Task.name</i> : to get the task's name.
2264 to receive a task, we use the method <i>MSG::Task.receive(mailbox)</i> that return a MSG:Task object (received task).
2269 - <i>MSG.createEnvironment(platform_file)</i> : set up the environment
2270 - <i>MSG.deployApplication(deployment_file)</i> : load the deployment file description.
2271 - <i>MSG.run</i> : run the simulation
2273 \subsubsection faq_binding_ruby_data Exchanging data
2274 ruby bindings provides two ways to exchange data between ruby processes.
2275 \li MSG::Task.join & MSG::Task.data \br
2277 the MSG::Task class contains 2 methods that allows a data exchange between 2 process.
2279 -<i>MSG::Task.join</i> : makes possible to join any kind of ruby data within a task.
2283 myTable <<1<<-2<<45<<67<<87<<76<<89<<56<<78<<3<<-4<<99
2284 # Creates and send Task With the Table inside
2285 task = MSG::Task.new("quicksort_task",taskComputeSize, taskCommunicationSize);
2290 -<i>MSG::Task.data</i> : to access to the data contained into the task.
2293 task = MSG::Task.receive(recv_mailbox.to_s)
2295 quicksort(table,0,table.size-1)
2298 you can find a complet example illustrating the use of those methods in file /example/ruby/Quicksort.rb
2302 another 'object-oriented' way to do it, is to make your own 'task' class that inherit from MSG::Task ,
2303 and contains data you want to deal with, the only 'tricky' thing is that "the initializer" method has no effect !
2305 the use of some getter/setter methods would be the simple way to manage your data :)
2307 class PingPongTask < MSG::Task
2308 # The initialize method has no effect
2318 you can find an example of use in file example/ruby/PingPong.rb
2320 \section faq_troubleshooting Troubleshooting
2322 \subsection faq_trouble_lib_compil SimGrid compilation and installation problems
2324 \subsubsection faq_trouble_lib_config ./configure fails!
2326 We know only one reason for the configure to fail:
2328 - <b>You are using a broken build environment</b>\n
2329 If symptom is that configure complains about gcc not being able to build
2330 executables, you are probably missing the libc6-dev package. Damn Ubuntu.
2332 If you experience other kind of issue, please get in touch with us. We are
2333 always interested in improving our portability to new systems.
2335 \subsubsection faq_trouble_distcheck Dude! "make check" fails on my machine!
2337 Don't assume we never run this target, because we do. Check
2338 http://bob.loria.fr:8010 if you don't believe us.
2340 There is several reasons which may cause the make check to fail on your
2343 - <b>You are using a broken libc (probably concerning the contextes)</b>.\n
2344 The symptom is that the "make check" fails within the examples/msg directory.\n
2345 By default, SimGrid uses something called ucontexts. This is part of the
2346 libc, but it's quite undertested. For example, some (old) versions of the
2347 glibc on alpha do not implement these functions, but provide the stubs
2348 (which return ENOSYS: not implemented). It may fool our detection mechanism
2349 and leads to segfaults. There is not much we can do to fix the bug.
2350 A workaround is to compile with --with-context=pthread to avoid
2351 ucontext completely. You'll be a bit more limited in the number
2352 of simulated processes you can start concurrently, but 5000
2353 processes is still enough for most purposes, isn't it?\n
2354 This limitation is the reason why we insist on using this piece of ...
2355 software even if it's so troublesome.\n
2356 <b>=> use --with-pthread on AMD64 architecture that do not have an
2357 ultra-recent libc.</b>
2359 - <b>There is a bug in SimGrid we aren't aware of</b>.\n
2360 If none of the above apply, please drop us a mail on the mailing list so
2361 that we can check it out. Make sure to read \ref faq_bugrepport
2364 \subsection faq_trouble_compil User code compilation problems
2366 \subsubsection faq_trouble_err_logcat "gcc: _simgrid_this_log_category_does_not_exist__??? undeclared (first use in this function)"
2368 This is because you are using the log mecanism, but you didn't created
2369 any default category in this file. You should refer to \ref XBT_log
2370 for all the details, but you simply forgot to call one of
2371 XBT_LOG_NEW_DEFAULT_CATEGORY() or XBT_LOG_NEW_DEFAULT_SUBCATEGORY().
2373 \subsubsection faq_trouble_pthreadstatic "gcc: undefined reference to pthread_key_create"
2375 This indicates that one of the library SimGrid depends on (libpthread
2376 here) was missing on the linking command line. Dependencies of
2377 libsimgrid are expressed directly in the dynamic library, so it's
2378 quite impossible that you see this message when doing dynamic linking.
2380 If you compile your code statically (and if you use a pthread version
2381 of SimGrid -- see \ref faq_more_processes), you must absolutely
2382 specify <tt>-lpthread</tt> on the linker command line. As usual, this should
2383 come after <tt>-lsimgrid</tt> on this command line.
2385 \subsection faq_trouble_errors Runtime error messages
2387 \subsubsection faq_flexml_limit "surf_parse_lex: Assertion `next limit' failed."
2389 This is because your platform file is too big for the parser.
2391 Actually, the message comes directly from FleXML, the technology on top of
2392 which the parser is built. FleXML has the bad idea of fetching the whole
2393 document in memory before parsing it. And moreover, the memory buffer size
2394 must be determined at compilation time.
2396 We use a value which seems big enough for our need without bloating the
2397 simulators footprints. But of course your mileage may vary. In this case,
2398 just edit src/surf/surfxml.l modify the definition of
2399 FLEXML_BUFFERSTACKSIZE. E.g.
2402 #define FLEXML_BUFFERSTACKSIZE 1000000000
2405 Then recompile and everything should be fine, provided that your version of
2406 Flex is recent enough (>= 2.5.31). If not the compilation process should
2409 A while ago, we worked on FleXML to reduce a bit its memory consumption, but
2410 these issues remain. There is two things we should do:
2412 - use a dynamic buffer instead of a static one so that the only limit
2413 becomes your memory, not a stupid constant fixed at compilation time
2414 (maybe not so difficult).
2415 - change the parser so that it does not need to get the whole file in
2416 memory before parsing
2417 (seems quite difficult, but I'm a complete newbe wrt flex stuff).
2419 These are changes to FleXML itself, not SimGrid. But since we kinda hijacked
2420 the development of FleXML, I can grant you that any patches would be really
2421 welcome and quickly integrated.
2423 <b>Update:</b> A new version of FleXML (1.7) was released. Most of the work
2424 was done by William Dowling, who use it in his own work. The good point is
2425 that it now use a dynamic buffer, and that the memory usage was greatly
2426 improved. The downside is that William also changed some things internally,
2427 and it breaks the hack we devised to bypass the parser, as explained in
2428 \ref faq_flexml_bypassing. Indeed, this is not a classical usage of the
2429 parser, and Will didn't imagine that we may have used (and even documented)
2430 such a crude usage of FleXML. So, we now have to repair the bypassing
2431 functionality to use the lastest FleXML version and fix the memory usage in
2434 \subsubsection faq_trouble_gras_transport GRAS spits networking error messages
2436 Gras, on real platforms, naturally use regular sockets to communicate. They
2437 are deeply hidden in the gras abstraction, but when things go wrong, you may
2438 get some weird error messages. Here are some example, with the probable
2441 - <b>Transport endpoint is not connected</b>: several processes try to open
2442 a server socket on the same port number of the same machine. This is
2443 naturally bad and each process should pick its own port number for this.\n
2444 Maybe, you just have some processes remaining from a previous experiment
2446 Killing them may help, but again if you kill -KILL them, you'll have to
2447 wait for a while: they didn't close there sockets properly and the system
2448 needs a while to notice that this port is free again.
2450 - <b>Socket closed by remote side</b>: if the remote process is not
2451 supposed to close the socket at this point, it may be dead.
2453 - <b>Connection reset by peer</b>: I found this on Internet about this
2454 error. I think it's what's happening here, too:\n
2455 <i>This basically means that a network error occurred while the client was
2456 receiving data from the server. But what is really happening is that the
2457 server actually accepts the connection, processes the request, and sends
2458 a reply to the client. However, when the server closes the socket, the
2459 client believes that the connection has been terminated abnormally
2460 because the socket implementation sends a TCP reset segment telling the
2461 client to throw away the data and report an error.\n
2462 Sometimes, this problem is caused by not properly closing the
2463 input/output streams and the socket connection. Make sure you close the
2464 input/output streams and socket connection properly. If everything is
2465 closed properly, however, and the problem persists, you can work around
2466 it by adding a one-second sleep before closing the streams and the
2467 socket. This technique, however, is not reliable and may not work on all
2469 Since GRAS sockets are closed properly (repeat after me: there is no bug
2470 in GRAS), it is either that you are closing your sockets on server side
2471 before the client get a chance to read them (use gras_os_sleep() to delay
2472 the server), or the server died awfully before the client got the data.
2474 \subsubsection faq_trouble_errors_big_fat_warning I'm told that my XML files are too old.
2476 The format of the XML platform description files is sometimes
2477 improved. For example, we decided to change the units used in SimGrid
2478 from MBytes, MFlops and seconds to Bytes, Flops and seconds to ease
2479 people exchanging small messages. We also reworked the route
2480 descriptions to allow more compact descriptions.
2482 That is why the XML files are versionned using the 'version' attribute
2483 of the root tag. Currently, it should read:
2485 <platform version="2">
2488 If your files are too old, you can use the simgrid_update_xml.pl
2489 script which can be found in the tools directory of the archive.
2491 \subsection faq_trouble_valgrind Valgrind-related and other debugger issues
2493 If you don't, you really should use valgrind to debug your code, it's
2496 \subsubsection faq_trouble_vg_longjmp longjmp madness in valgrind
2498 This is when valgrind starts complaining about longjmp things, just like:
2500 \verbatim ==21434== Conditional jump or move depends on uninitialised value(s)
2501 ==21434== at 0x420DBE5: longjmp (longjmp.c:33)
2503 ==21434== Use of uninitialised value of size 4
2504 ==21434== at 0x420DC3A: __longjmp (__longjmp.S:48)
2507 This is the sign that you didn't used the exception mecanism well. Most
2508 probably, you have a <tt>return;</tt> somewhere within a <tt>TRY{}</tt>
2509 block. This is <b>evil</b>, and you must not do this. Did you read the section
2512 \subsubsection faq_trouble_vg_libc Valgrind spits tons of errors about backtraces!
2514 It may happen that valgrind, the memory debugger beloved by any decent C
2515 programmer, spits tons of warnings like the following :
2516 \verbatim ==8414== Conditional jump or move depends on uninitialised value(s)
2517 ==8414== at 0x400882D: (within /lib/ld-2.3.6.so)
2518 ==8414== by 0x414EDE9: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2519 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2520 ==8414== by 0x414F937: _dl_open (in /lib/tls/i686/cmov/libc-2.3.6.so)
2521 ==8414== by 0x4150F4C: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2522 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2523 ==8414== by 0x415102D: __libc_dlopen_mode (in /lib/tls/i686/cmov/libc-2.3.6.so)
2524 ==8414== by 0x412D6B9: backtrace (in /lib/tls/i686/cmov/libc-2.3.6.so)
2525 ==8414== by 0x8076446: xbt_dictelm_get_ext (dict_elm.c:714)
2526 ==8414== by 0x80764C1: xbt_dictelm_get (dict_elm.c:732)
2527 ==8414== by 0x8079010: xbt_cfg_register (config.c:208)
2528 ==8414== by 0x806821B: MSG_config (msg_config.c:42)
2531 This problem is somewhere in the libc when using the backtraces and there is
2532 very few things we can do ourselves to fix it. Instead, here is how to tell
2533 valgrind to ignore the error. Add the following to your ~/.valgrind.supp (or
2534 create this file on need). Make sure to change the obj line according to
2535 your personnal mileage (change 2.3.6 to the actual version you are using,
2536 which you can retrieve with a simple "ls /lib/ld*.so").
2539 name: Backtrace madness
2541 obj:/lib/ld-2.3.6.so
2546 fun:__libc_dlopen_mode
2549 Then, you have to specify valgrind to use this suppression file by passing
2550 the <tt>--suppressions=$HOME/.valgrind.supp</tt> option on the command line.
2551 You can also add the following to your ~/.bashrc so that it gets passed
2552 automatically. Actually, it passes a bit more options to valgrind, and this
2553 happen to be my personnal settings. Check the valgrind documentation for
2556 \verbatim export VALGRIND_OPTS="--leak-check=yes --leak-resolution=high --num-callers=40 --tool=memcheck --suppressions=$HOME/.valgrind.supp" \endverbatim
2558 \subsubsection faq_trouble_backtraces Truncated backtraces
2560 When debugging SimGrid, it's easier to pass the
2561 --disable-compiler-optimization flag to the configure if valgrind or
2562 gdb get fooled by the optimization done by the compiler. But you
2563 should remove these flag when everything works before going in
2564 production (before launching your 1252135 experiments), or everything
2565 will run only one half of the true SimGrid potential.
2567 \subsection faq_deadlock There is a deadlock in my code!!!
2569 Unfortunately, we cannot debug every code written in SimGrid. We
2570 furthermore believe that the framework provides ways enough
2571 information to debug such informations yourself. If the textual output
2572 is not enough, Make sure to check the \ref faq_visualization FAQ entry to see
2573 how to get a graphical one.
2575 Now, if you come up with a really simple example that deadlocks and
2576 you're absolutely convinced that it should not, you can ask on the
2577 list. Just be aware that you'll be severely punished if the mistake is
2578 on your side... We have plenty of FAQ entries to redact and new
2579 features to implement for the impenitents! ;)
2581 \subsection faq_surf_network_latency I get weird timings when I play with the latencies.
2583 OK, first of all, remember that units should be Bytes, Flops and
2584 Seconds. If you don't use such units, some SimGrid constants (e.g. the
2585 SG_TCP_CTE_GAMMA constant used in most network models) won't have the
2586 right unit and you'll end up with weird results.
2588 Here is what happens with a single transfer of size L on a link
2589 (bw,lat) when nothing else happens.
2592 0-----lat--------------------------------------------------t
2593 |-----|**** real_bw =min(bw,SG_TCP_CTE_GAMMA/(2*lat)) *****|
2596 In more complex situations, this min is the solution of a complex
2597 max-min linear system. Have a look
2598 <a href="http://lists.gforge.inria.fr/pipermail/simgrid-devel/2006-April/thread.html">here</a>
2599 and read the two threads "Bug in SURF?" and "Surf bug not
2600 fixed?". You'll have a few other examples of such computations. You
2601 can also read "A Network Model for Simulation of Grid Application" by
2602 Henri Casanova and Loris Marchal to have all the details. The fact
2603 that the real_bw is smaller than bw is easy to understand. The fact
2604 that real_bw is smaller than SG_TCP_CTE_GAMMA/(2*lat) is due to the
2605 window-based congestion mechanism of TCP. With TCP, you can't exploit
2606 your huge network capacity if you don't have a good round-trip-time
2607 because of the acks...
2609 Anyway, what you get is t=lat + L/min(bw,SG_TCP_CTE_GAMMA/(2*lat)).
2611 * if I you set (bw,lat)=(100 000 000, 0.00001), you get t = 1.00001 (you fully
2613 * if I you set (bw,lat)=(100 000 000, 0.0001), you get t = 1.0001 (you're on the
2615 * if I you set (bw,lat)=(100 000 000, 0.001), you get t = 10.001 (ouch!)
2617 This bound on the effective bandwidth of a flow is not the only thing
2618 that may make your result be unexpected. For example, two flows
2619 competing on a saturated link receive an amount of bandwidth inversely
2620 proportional to their round trip time.
2622 \subsection faq_bugrepport So I've found a bug in SimGrid. How to report it?
2624 We do our best to make sure to hammer away any bugs of SimGrid, but this is
2625 still an academic project so please be patient if/when you find bugs in it.
2626 If you do, the best solution is to drop an email either on the simgrid-user
2627 or the simgrid-devel mailing list and explain us about the issue. You can
2628 also decide to open a formal bug report using the
2629 <a href="https://gforge.inria.fr/tracker/?atid=165&group_id=12&func=browse">relevant
2630 interface</a>. You need to login on the server to get the ability to submit
2633 We will do our best to solve any problem repported, but you need to help us
2634 finding the issue. Just telling "it segfault" isn't enough. Telling "It
2635 segfaults when running the attached simulator" doesn't really help either.
2636 You may find the following article interesting to see how to repport
2637 informative bug repports:
2638 http://www.chiark.greenend.org.uk/~sgtatham/bugs.html (it is not SimGrid
2639 specific at all, but it's full of good advices).
2641 \author Arnaud Legrand (arnaud.legrand::imag.fr)
2642 \author Martin Quinson (martin.quinson::loria.fr)
2647 ******************************************************************
2648 * OLD CRUFT NOT USED ANYMORE *
2649 ******************************************************************
2652 \subsection faq_crosscompile Cross-compiling a Windows DLL of SimGrid from linux
2654 At the moment, we do not distribute Windows pre-compiled version of SimGrid
2655 because the support for this platform is still experimental. We know that
2656 some parts of the GRAS environment do not work, and we think that the others
2657 environments (MSG and SD) have good chances to work, but we didn't test
2658 ourselves. This section explains how we generate the SimGrid DLL so that you
2659 can build it for yourself. First of all, you need to have a version more
2660 recent than 3.1 (ie, a SVN version as time of writting).
2662 In order to cross-compile the package to windows from linux, you need to
2663 install mingw32 (minimalist gnu win32). On Debian, you can do so by
2664 installing the packages mingw32 (compiler), mingw32-binutils (linker and
2665 so), mingw32-runtime.
2667 You can use the VPATH support of configure to compile at the same time for
2668 linux and windows without dupplicating the source nor cleaning the tree
2669 between each. Just run bootstrap (if you use the SVN) to run the autotools.
2670 Then, create a linux and a win directories. Then, type:
2671 \verbatim cd linux; ../configure --srcdir=.. <usual configure flags>; make; cd ..
2672 cd win; ../configure --srcdir=.. --host=i586-mingw32msvc <flags>; make; cd ..
2674 The trick to VPATH builds is to call configure from another directory,
2675 passing it an extra --srcdir argument to tell it where all the sources are.
2676 It will understand you want to use VPATH. Then, the trick to cross-compile
2677 is simply to add a --host argument specifying the target you want to build
2678 for. The i586-mingw32msvc string is what you have to pass to use the mingw32
2679 environment as distributed in Debian.
2681 After that, you can run all make targets from both directories, and test
2682 easily that what you change for one arch does not break the other one.
2684 It is possible that this VPATH build thing breaks from time to time in the
2685 SVN since it's quite fragile, but it's granted to work in any released
2686 version. If you experience problems, drop us a mail.
2688 Another possible source of issue is that at the moment, building the
2689 examples request to use the gras_stub_generator tool, which is a compiled
2690 program, not a script. In cross-compilation, you need to cross-execute with
2691 wine for example, which is not really pleasant. We are working on this, but
2692 in the meanwhile, simply don't build the examples in cross-compilation
2693 (<tt>cd src</tt> before running make).
2695 Program (cross-)compiled with mingw32 do request an extra DLL at run-time to be
2696 usable. For example, if you want to test your build with wine, you should do
2697 the following to put this library where wine looks for DLLs.
2699 cp /usr/share/doc/mingw32-runtime/mingwm10.dll.gz ~/.wine/c/windows/system/
2700 gunzip ~/.wine/c/windows/system/mingwm10.dll.gz
2703 The DLL is built in src/.libs, and installed in the <i>prefix</i>/bin directory
2704 when you run make install.
2706 If you want to use it in a native project on windows, you need to use
2707 simgrid.dll and mingwm10.dll. For each DLL, you need to build .def file
2708 under linux (listing the defined symbols), and convert it into a .lib file
2709 under windows (specifying this in a way that windows compilers like). To
2710 generate the def files, run (under linux):
2711 \verbatim echo "LIBRARY libsimgrid-0.dll" > simgrid.def
2712 echo EXPORTS >> simgrid.def
2713 nm libsimgrid-0.dll | grep ' T _' | sed 's/.* T _//' >> simgrid.def
2714 nm libsimgrid-0.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> simgrid.def
2716 echo "LIBRARY mingwm10.dll" > mingwm10.def
2717 echo EXPORTS >> mingwm10.def
2718 nm mingwm10.dll | grep ' T _' | sed 's/.* T _//' >> mingwm10.def
2719 nm mingwm10.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> mingwm10.def
2722 To create the import .lib files, use the <tt>lib</tt> windows tool (from
2723 MSVC) the following way to produce simgrid.lib and mingwm10.lib
2724 \verbatim lib /def:simgrid.def
2725 lib /def:mingwm10.def
2728 If you happen to use Borland C Builder, the right command line is the
2729 following (note that you don't need any file.def to get this working).
2730 \verbatim implib simgrid.lib libsimgrid-0.dll
2731 implib mingwm10.lib mingwm10.dll
2734 Then, set the following parameters in Visual C++ 2005:
2735 Linker -> Input -> Additional dependencies = simgrid.lib mingwm10.lib
2737 Just in case you wonder how to generate a DLL from libtool in another
2738 project, we added -no-undefined to any lib*_la_LDFLAGS variables so that
2739 libtool accepts to generate a dynamic library under windows. Then, to make
2740 it true, we pass any dependencies (such as -lws2 under windows or -lpthread
2741 on need) on the linking line. Passing such deps is a good idea anyway so
2742 that they get noted in the library itself, avoiding the users to know about
2743 our dependencies and put them manually on their compilation line. Then we
2744 added the AC_LIBTOOL_WIN32_DLL macro just before AC_PROG_LIBTOOL in the
2745 configure.ac. It means that we exported any symbols which need to be.
2746 Nowadays, functions get automatically exported, so we don't need to load our
2747 header files with tons of __declspec(dllexport) cruft. We only need to do so
2748 for data, but there is no public data in SimGrid so we are good.