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_installation Installing the SimGrid library
93 Many people have been asking me questions on how to use SimGrid. Quite
94 often, the questions were not really about SimGrid but on the
95 installation process. This section is intended to help people that are
96 not familiar with compiling C files under UNIX. If you follow these
97 instructions and still have some troubles, drop an e-mail to
98 <simgrid-user@lists.gforge.inria.fr>.
100 \subsection faq_compiling Compiling SimGrid from a stable archive
102 First of all, you need to download the latest version of SimGrid from
103 <a href="http://gforge.inria.fr/frs/?group_id=12">here</a>.
104 Suppose you have uncompressed SimGrid in some temporary location of
105 your home directory (say <tt>/home/joe/tmp/simgrid-3.0.1 </tt>). The
106 simplest way to use SimGrid is to install it in your home
107 directory. Change your directory to
108 <tt>/home/joe/tmp/simgrid-3.0.1</tt> and type
111 ./configure --prefix=$HOME
116 If at some point, something fails, check the section \ref faq_trouble_compil .
117 If it does not help, you can report this problem to the
118 list but, please, avoid sending a laconic mail like "There is a problem. Is it
119 okay?". Send the config.log file which is automatically generated by
120 configure. Try to capture both the standard output and the error output of the
121 <tt>make</tt> command with <tt>script</tt>. There is no way for us to help you
122 without the relevant bits of information.
124 Now, the following directory should have been created :
126 \li <tt>/home/joe/doc/simgrid/html/</tt>
127 \li <tt>/home/joe/lib/</tt>
128 \li <tt>/home/joe/include/</tt>
130 SimGrid is not a binary, it is a library. Both a static and a dynamic
131 version are available. Here is what you can find if you try a <tt>ls
134 \verbatim libsimgrid.a libsimgrid.la libsimgrid.so libsimgrid.so.0 libsimgrid.so.0.0.1
137 Thus, there is two ways to link your program with SimGrid:
138 \li Either you use the static version, e.g
139 \verbatim gcc libsimgrid.a -o MainProgram MainProgram.c
141 In this case, all the SimGrid functions are directly
142 included in <tt>MainProgram</tt> (hence a bigger binary).
143 \li Either you use the dynamic version (the preferred method)
144 \verbatim gcc -lsimgrid -o MainProgram MainProgram.c
146 In this case, the SimGrid functions are not included in
147 <tt>MainProgram</tt> and you need to set your environment
148 variable in such a way that <tt>libsimgrid.so</tt> will be
149 found at runtime. This can be done by adding the following
150 line in your .bashrc (if you use bash and if you have
151 installed the SimGrid libraries in your home directory):
152 \verbatim export LD_LIBRARY_PATH=$HOME/lib/:$LD_LIBRARY_PATH
155 \subsection faq_compiling_java Java bindings don't get compiled
157 The configure script detects automatically whether you have the
158 softwares needed to use the Java bindings or not. At the end of the
159 configure, you can see the configuration picked by the script, which
160 should look similar to
161 \verbatim Configuration of package simgrid' (version 3.3.4-svn) on
164 Compiler: gcc (version: )
166 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
170 Context backend: ucontext
177 In this example, Java backends won't be compiled.
179 On Debian-like systems (which includes ubuntu), you need the following
180 packages: sun-java6-jdk libgcj10-dev. If you cannot find the
181 libgcj10-dev, try another version, like libgcj9-dev (on Ubuntu before
182 9.10) or libgcj11-dev (not released yet, but certainly one day).
183 Please note that you need to activate the contrib and non-free
184 repositories in Debian, and the universe ones in Ubuntu. Java comes at
187 \subsection faq_compiling_snapshoot SimGrid development snapshots
189 We have very high standards on software quality, and we are reluctant releasing
190 a stable release as long as there is still some known bug in the code base. In
191 addition, we added quite an extensive test base, making sure that we correctly
192 test the most important parts of the tool.
194 As an unfortunate conclusion, there may be some time between the stable
195 releases. If you want to benefit from the most recent features we introduced,
196 but don't want to take the risk of an untested version from the SVN, then
197 development snapshots are done for you.
199 These are pre-releases of SimGrid that still fail some tests about features
200 that almost nobody use, or on platforms not being in our core target (which is
201 Linux, Mac, other Unixes and Windows, from the most important to the less
202 one). That means that using this development releases should be safe for most
205 These archives can be found on
206 <a href="http://www.loria.fr/~quinson/simgrid.html">this web page</a>. Once you
207 got the lastest archive, you can compile it just like any archive (see above).
209 \subsection faq_compiling_svn Compiling SimGrid from the SVN
211 The project development takes place in the SVN, where all changes are
212 committed when they happen. Then every once in a while, we make sure that the
213 code quality meets our standard and release an archive from the code in the
214 SVN. We afterward go back to the development in the SVN. So, if you need a
215 recently added feature and can afford some little problem with the stability
216 of the lastest features, you may want to use the SVN version instead of a
219 For that, you first need to get the "simgrid" module from
220 <a href="http://gforge.inria.fr/scm/?group_id=12">here</a>.
222 You won't find any <tt>configure</tt> and a few other things
223 (<tt>Makefile.in</tt>'s, documentation, ...) will be missing as well. The
224 reason for that is that all these files have to be regenerated using the
225 latest versions of <tt>autoconf</tt>, <tt>libtool</tt>, <tt>automake</tt>
226 (>1.9) and <tt>doxygen</tt> (>1.4). To generate the <tt>configure</tt> and
227 the <tt>Makefile.in</tt>'s, you just have to launch the <tt>bootstrap</tt>
228 command that resides in the top of the source tree. Then just follow the
229 instructions of Section \ref faq_compiling.
231 We insist on the fact that you really need the latest versions of
232 autoconf, automake and libtool. Doing this step on exotic architectures/systems
233 (i.e. anything different from a recent linux distribution) may be
234 ... uncertain. If you need to compile the SVN version on a machine where all these
235 dependencies are not met, the easiest is to do <tt>make dist</tt> in the SVN
236 directory of another machine where all dependencies are met. It will create an
237 archive you may deploy on other sites just as a regular stable release.
239 In summary, the following commands will checkout the SVN, regenerate the
240 configure script and friends, configure SimGrid and build it.
242 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk simgrid
245 ./configure --enable-maintainer-mode --prefix=<where to install SimGrid>
248 Then, if you want to install SimGrid on the current box, just do:
249 \verbatim make install \endverbatim
251 If you want to build an snapshot of the SVN to deploy it on another box (for
252 example because the other machine don't have the autotools), do:
253 \verbatim make dist \endverbatim
255 Moreover, you should never call the autotools manually since you must run
256 them in a specific order with specific arguments. Most of the times, the
257 makefiles will automatically call the tools for you. When it's not possible
258 (such as the first time you checkout the SVN), use the ./bootstrap command
259 to call them explicitly.
262 \subsection faq_setting_MSG Setting up your own MSG code
264 Do not build your simulator by modifying the SimGrid examples. Go
265 outside the SimGrid source tree and create your own working directory
266 (say <tt>/home/joe/SimGrid/MyFirstScheduler/</tt>).
268 Suppose your simulation has the following structure (remember it is
269 just an example to illustrate a possible way to compile everything;
270 feel free to organize it as you want).
272 \li <tt>sched.h</tt>: a description of the core of the
273 scheduler (i.e. which functions are can be used by the
274 agents). For example we could find the following functions
275 (master, forwarder, slave).
277 \li <tt>sched.c</tt>: a C file including <tt>sched.h</tt> and
278 implementing the core of the scheduler. Most of these
279 functions use the MSG functions defined in section \ref
282 \li <tt>masterslave.c</tt>: a C file with the main function, i.e.
283 the MSG initialization (MSG_global_init()), the platform
284 creation (e.g. with MSG_create_environment()), the
285 deployment phase (e.g. with MSG_function_register() and
286 MSG_launch_application()) and the call to
289 To compile such a program, we suggest to use the following
290 Makefile. It is a generic Makefile that we have used many times with
291 our students when we teach the C language.
295 masterslave: masterslave.o sched.o
297 INSTALL_PATH = $$HOME
299 PEDANTIC_PARANOID_FREAK = -O0 -Wshadow -Wcast-align \
300 -Waggregate-return -Wmissing-prototypes -Wmissing-declarations \
301 -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations \
302 -Wmissing-noreturn -Wredundant-decls -Wnested-externs \
303 -Wpointer-arith -Wwrite-strings -finline-functions
304 REASONABLY_CAREFUL_DUDE = -Wall
305 NO_PRAYER_FOR_THE_WICKED = -w -O2
306 WARNINGS = $(REASONABLY_CAREFUL_DUDE)
307 CFLAGS = -g $(WARNINGS)
309 INCLUDES = -I$(INSTALL_PATH)/include
310 DEFS = -L$(INSTALL_PATH)/lib/
311 LDADD = -lm -lsimgrid
315 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) $^ $(LIBS) $(LDADD) -o $@
318 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) -c -o $@ $<
321 rm -f $(BIN_FILES) *.o *~
327 The first two lines indicates what should be build when typing make
328 (<tt>masterslave</tt>) and of which files it is to be made of
329 (<tt>masterslave.o</tt> and <tt>sched.o</tt>). This makefile assumes
330 that you have set up correctly your <tt>LD_LIBRARY_PATH</tt> variable
331 (look, there is a <tt>LDADD = -lm -lsimgrid</tt>). If you prefer using
332 the static version, remove the <tt>-lsimgrid</tt> and add a
333 <tt>$(INSTALL_PATH)/lib/libsimgrid.a</tt> on the next line, right
334 after the <tt>LIBS = </tt>.
336 More generally, if you have never written a Makefile by yourself, type
337 in a terminal : <tt>info make</tt> and read the introduction. The
338 previous example should be enough for a first try but you may want to
339 perform some more complex compilations...
341 \subsection faq_setting_GRAS Setting up your own GRAS code
343 If you use the GRAS interface instead of the MSG one, then previous section
344 is not the better source of information. Instead, you should check the GRAS
345 tutorial in general, and the \ref GRAS_tut_tour_setup in particular.
347 \section faq_cmake CMAKE
349 \subsection faq_intro Some generalitty
351 \subsubsection faq_intro1 What is Cmake?
353 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>.
355 \subsubsection faq_intro2 Why cmake?
357 CMake permits to developers to compil projects on different plateform. Then many tools are embedded like ctest for making test, a link to cdash for vizualise results but also test coverage and bug reports.
359 \subsubsection faq_intro3 What cmake need?
361 CMake needs some prerequists like :
363 \li c, c++ and java compiler regards to developers
364 \li ccmake for graphical used of CMake
365 \li cmake <a href="http://www.cmake.org/cmake/resources/software.html">(download page)</a>
367 \subsubsection faq_intro4 Cmake vs Autotools...
371 \subsection faq_cmakeoption Cmake options
373 \subsubsection faq_cmakeoption1 Liste of options
376 "cmake -D[name]=[value] ... ./"
378 [name] enable_gtnets [value] ON/OFF or TRUE/FALSE or 1/0
379 enable_java ON/OFF or TRUE/FALSE or 1/0
380 enable_lua ON/OFF or TRUE/FALSE or 1/0
381 enable_ruby ON/OFF or TRUE/FALSE or 1/0
382 enable_compile_optimizations ON/OFF or TRUE/FALSE or 1/0
383 enable_compile_warnings ON/OFF or TRUE/FALSE or 1/0
384 enable_maintainer_mode ON/OFF or TRUE/FALSE or 1/0
385 enable_supernovae ON/OFF or TRUE/FALSE or 1/0
386 enable_tracing ON/OFF or TRUE/FALSE or 1/0
387 enable_coverage ON/OFF or TRUE/FALSE or 1/0
388 enable_memcheck ON/OFF or TRUE/FALSE or 1/0
389 enable_print_message ON/OFF or TRUE/FALSE or 1/0
391 gtnets_path <path_to_gtnets_directory>
392 prefix <path_to_install_directory>
393 BIBTEX2HTML <path_to_bibtex2html>
394 with_context auto/ucontext/pthread/window
397 \subsubsection faq_cmakeoption2 Options explaination
399 \li enable_gtnets : set to true implie that user wants to use gtnets.
401 \li enable_java : set to true implie that user wants to add java langage into simgrid compilation.
403 \li enable_lua : set to true implie that user wants to add lua langage into simgrid compilation.
405 \li enable_ruby : set to true implie that user wants to add ruby langage into simgrid compilation.
407 \li enable_compile_optimizations : add flags "-O3 -finline-functions -funroll-loops -fno-strict-aliasing"
409 \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"
411 \li enable_maintainer_mode : set to true it remakes some files.
413 include/surf/simgrid_dtd.h
414 include/xbt/graphxml.h
423 src/xbt_strbuff_unit.c
426 src/xbt_synchro_unit.c
427 src/simgrid_units_main.c
433 src/surf/simgrid_dtd.c
434 src/surf/simgrid_dtd.l
439 src/gras/DataDesc/ddt_parse.yy.c
441 \li enable_supernovae : set to true make one file for each lib and compile with those generated files.
444 /src/supernovae_gras.c
445 /src/supernovae_smpi.c
448 \li enable_tracing : To enable the generation of simulation traces for visualization
450 \li enable_coverage : When set to true this option enable code coverage by setting -fprofile-arcs -ftest-coverage flags.
452 \li enable_memcheck : When set to true this option enable tests for memcheck.
454 \li enable_print_message : This option when enable permits to see variables from gras_config.h
456 \li gtnets_path : Path to gtnets install directory (ex /usr)
458 \li prefix : Path where are installed lib/ doc/ and include/ directories (ex /usr/local)
460 \li BIBTEX2HTML : Path where is installed bibtex2html.
462 \li with context : specify which context the user wants to use.
464 \subsubsection faq_cmakeoption3 Initialisation
466 Those options are initialized the first time you launch \"cmake ./\" whithout specified option.
473 enable_compile_optimizations off
474 enable_compile_warnings off
475 enable_maintainer_mode off
476 enable_supernovae off
480 enable_print_message off
488 \subsubsection faq_cmakeoption4 Option's cache and how to reset?
490 When options have been set they are keep into a cache file named \"CMakeCache.txt\". So if you want
491 reset values you just delete this file located to the project directory.
493 \subsection faq_cmakecompilation Cmake compilation
495 \subsubsection faq_cmakecompilation1 With command line.
498 cmake -D[name]=[value] ... ./
502 \subsubsection faq_cmakecompilation2 With ccmake tool.
507 Then follow instructions.
509 \subsubsection faq_cmakecompilation3 Resume of command line
513 cmake ./ configure the project
514 make build all targets
515 make VERBOSE=1 build all targets and print build command lines
516 make test test all targets and summarize
517 make dist make the distrib
518 make distcheck check the dist (make + make dist + make test)
519 make install-simgrid install the project (doc/ lib/ include/)
520 make uninstall uninstall the project (doc/ lib/ include/)
521 make clean clean all targets
522 make java-clean clean files created by java option
523 make doc-clean clean files created for making doc
524 make supernovae-clean clean supernovae files
525 make maintainer-clean clean maintainer files
526 make all-clean execute the 5 upper clean command
527 make html Create simgrid documentation
528 make maintainer-clean Remove all files generated by mainainer mode
531 When the project have been succesfully compiling and build you can make tests.
535 ctest launch only tests
537 ctest -D Continuous(Start|Update|Configure|Build)
538 ctest -D Continuous(Test|Coverage|MemCheck|Submit)
539 ctest -D Experimental
540 ctest -D Experimental(Start|Update|Configure|Build)
541 ctest -D Experimental(Test|Coverage|MemCheck|Submit)
543 ctest -D Nightly(Start|Update|Configure|Build)
544 ctest -D Nightly(Test|Coverage|MemCheck|Submit)
545 ctest -D NightlyMemoryCheck
548 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>.
550 \subsubsection faq_cmakecompilation4 Examples for different mode.
554 cmake -Denable_maintainer_mode=on ./
555 \verbatim GTnetS doesn't works : set -Ddisable_gtnets=on
556 with_context auto change to ucontext
559 Make : src/simgrid.jar with : /usr/bin/javac
560 Make examples/java with : /usr/bin/javac
562 Configuration of package `simgrid' (revision 7228M) on arch (=4):
564 SITE : Linux_Ubuntu 9.10_x86_64
566 Compiler: c++ : /usr/bin/c++
567 version: c++ (Ubuntu 4.4.1-4ubuntu9) 4.4.1
569 version: gcc (Ubuntu 4.4.1-4ubuntu9) 4.4.1
575 Context backend: ucontext
585 Simgrid dependencies: dl -llua5.1
586 Gras dependencies: pthread
590 USER_PREFIX: /usr/local
591 INSTALL_PREFIX: /usr/local
595 -- Build files have been written to: /home/navarrop/Bureau/simgrid-trunk
600 cmake -Dsupernovae=on ./
601 \verbatim GTnetS doesn't works : set -Ddisable_gtnets=on
602 with_context auto change to ucontext
605 Make : src/simgrid.jar with : /usr/bin/javac
606 Make examples/java with : /usr/bin/javac
608 Configuration of package `simgrid' (revision 7228M) on arch (=4):
609 BUILDNAME : SUPERNOVAE
610 SITE : Linux_Ubuntu 9.10_x86_64
612 Compiler: c++ : /usr/bin/c++
613 version: c++ (Ubuntu 4.4.1-4ubuntu9) 4.4.1
615 version: gcc (Ubuntu 4.4.1-4ubuntu9) 4.4.1
617 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
621 Context backend: ucontext
631 Simgrid dependencies: dl -llua5.1
632 Gras dependencies: pthread
636 USER_PREFIX: /usr/local
637 INSTALL_PREFIX: /usr/local
641 -- Build files have been written to: /home/navarrop/Bureau/simgrid-trunk
646 cmake -Dgtnets_path=/home/navarrop/Bureau/usr/ ./
647 \verbatim with_context auto change to ucontext
650 Make : src/simgrid.jar with : /usr/bin/javac
651 Make examples/java with : /usr/bin/javac
653 Configuration of package `simgrid' (revision 7228M) on arch (=4):
655 SITE : Linux_Ubuntu 9.10_x86_64
657 Compiler: c++ : /usr/bin/c++
658 version: c++ (Ubuntu 4.4.1-4ubuntu9) 4.4.1
660 version: gcc (Ubuntu 4.4.1-4ubuntu9) 4.4.1
662 CFlags: -L/home/navarrop/Bureau/usr/lib -I/home/navarrop/Bureau/usr/include/gtnets -g3
663 CPPFlags: -L/home/navarrop/Bureau/usr/lib -I/home/navarrop/Bureau/usr/include/gtnets
666 Context backend: ucontext
668 path: /home/navarrop/Bureau/usr
676 Simgrid dependencies: dl -llua5.1 -lgtnets
677 Gras dependencies: pthread
681 USER_PREFIX: /usr/local
682 INSTALL_PREFIX: /usr/local
684 INFO -->> Take care to have export LD_LIBRARY_PATH before run make command for make examples with gtnets
685 copy and paste : export LD_LIBRARY_PATH=/home/navarrop/Bureau/usr/lib/:$LD_LIBRARY_PATH
690 -- Build files have been written to: /home/navarrop/Bureau/simgrid-trunk
693 \subsection faq_cmakeinstall How to install with cmake?
695 \subsubsection faq_cmakeinstall1 From svn.
698 cmake -Denable_maintainer_mode=on -Dprefix=/home/navarrop/Bureau/install_simgrid ./
703 \subsubsection faq_cmakeinstall2 From a distrib
706 cmake -Dprefix=/home/navarrop/Bureau/install_simgrid ./
711 \subsection faq_screenshot Screenshot
714 navarrop@caraja:~$ cd Bureau/simgrid-trunk/
715 navarrop@caraja:~/Bureau/simgrid-trunk$ cmake ./
717 GTnetS doesn't works : set -Ddisable_gtnets=on <-|some warnings are printed
718 with_context auto change to ucontext <-|
719 (skaddr) <--info (needed)
720 (sksize) <--info (needed)
721 Make : src/simgrid.jar with : /usr/bin/javac <--info (if java)
722 Make examples/java with : /usr/bin/javac <--info (if java)
724 Configuration of package `simgrid' (revision 7209M) on arch (=4):
725 BUILDNAME : UCONTEXT <-- name of the compilation regarding to cdash
726 SITE : Linux_Ubuntu 9.10_x86_64 <-- distribution of the local machine regarding to cdash
728 Compiler: c++ : /usr/bin/c++
729 version: c++ (Ubuntu 4.4.1-4ubuntu9) 4.4.1
731 version: gcc (Ubuntu 4.4.1-4ubuntu9) 4.4.1
737 Context backend: ucontext
747 Simgrid dependencies: -ldl -llua5.1
748 Gras dependencies: pthread
752 USER_PREFIX: /usr/local
753 INSTALL_PREFIX: /usr/local
757 -- Build files have been written to: /home/navarrop/Bureau/simgrid-trunk
759 Here all options are checked and printed. If it doesn't match with your configuration
760 it is probably due to a wrong configuration.
762 \subsection faq_cmakehowto How to modified sources files for developers
764 \subsubsection faq_cmakehowto1 Add an executable or examples.
766 If you want make an executable you have to create a CMakeList.txt to the src directory.
767 You must specified where to create the executable, source list, dependencies and the name of the binary.
770 cmake_minimum_required(VERSION 2.6)
772 set(EXECUTABLE_OUTPUT_PATH "./")
773 set(LIBRARY_OUTPUT_PATH "${PROJECT_DIRECTORY}/lib")
775 add_executable(get_sender get_sender.c) #add_executable(<name_of_target> <src list>)
777 ### Add definitions for compile
778 target_link_libraries(get_sender simgrid m pthread -fprofile-arcs) #target_link_libraries(<name_of_targe> <dependencies>)
781 Then you have to modified <project/directory>/buildtools/Cmake/src/CMakeMakeExeLib.txt and add
784 add_subdirectory(${PROJECT_DIRECTORY}/<path_where_is_CMakeList.txt>)
787 \subsubsection faq_cmakehowto2 Delete/add sources to lib.
789 If you want modified, add or delete source files from a library you have to edit <project/directory>/buildtools/Cmake/src/CMakeDefinePackages.txt
793 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/MsgException.java
794 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/JniException.java
795 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/NativeException.java
796 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/HostNotFoundException.java
797 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/ProcessNotFoundException.java
798 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/Msg.java
799 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/Process.java
800 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/Host.java
801 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/Task.java
802 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/MsgNative.java
803 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/ApplicationHandler.java
804 ${PROJECT_DIRECTORY}/src/java/simgrid/msg/Sem.java
808 \subsubsection faq_cmakehowto3 Add test
810 If you want modified, add or delete tests you have to edit <project/directory>/buildtools/Cmake/src/CMakeTest.txt
811 with this function : ADD_TEST(<name> <bin> <ARGS>)
814 add_test(test-simdag-1 ${PROJECT_DIRECTORY}/testsuite/simdag/sd_test --cfg=path:${PROJECT_DIRECTORY}/testsuite/simdag small_platform_variable.xml)
817 \subsection faq_cmakeExplain Explaination of sources files for cmake
821 Those files are the "main parts". One located at the project directory call all the cmake sources files. The others
822 are little projects called by the first for make examples.
824 \li CMakeCompleteInFiles.txt
826 Complete all .in files and define Variables for h files
830 This file make the html documentation.
832 \li CMakeMakeExeLib.txt
834 Here are callled all "CMakeLists.txt" for make executables and libraries.
836 \li CMakePrintArgs.txt
838 This file is called at the end of the build for summarize environment variables.
840 \li CMakeDefinePackages.txt
842 Here is defined sources packages for compiling libs.
846 Defined flags which are used for compiling sources.
848 \li CMakeSupernovae.txt
850 Here are made files for the supernovae mode.
854 Here is defined packages for install simgrid and make a distribution.
856 \li CMakeMaintainerMode.txt
858 Part where are generated source files for maintainer mode.
862 Here are defined options and initialized values.
866 All tests are listed.
868 \li CTestConfig.cmake
870 Properties which link tests with dashboard.
872 \subsection faq_cmakeList List of files added for cmake
874 Here is a list of files involved into cmake build (relative to project directory path) :
879 ./buildtools/Cmake/src/CMakeCompleteInFiles.txt
880 ./buildtools/Cmake/src/CMakeDocs.txt
881 ./buildtools/Cmake/src/CMakeMakeExeLib.txt
882 ./buildtools/Cmake/src/CMakePrintArgs.txt
883 ./buildtools/Cmake/src/CMakeDefinePackages.txt
884 ./buildtools/Cmake/src/CMakeFlags.txt
885 ./buildtools/Cmake/src/CMakeSupernovae.txt
886 ./buildtools/Cmake/src/CMakeDistrib.txt
887 ./buildtools/Cmake/src/CMakeMaintainerMode.txt
888 ./buildtools/Cmake/src/CMakeOption.txt
889 ./buildtools/Cmake/src/CMakeTest.txt
890 ./buildtools/Cmake/src/CTestConfig.cmake
892 Test files for define properties :
893 ./buildtools/Cmake/prog_test/prog_GRAS_ARCH.c
894 ./buildtools/Cmake/prog_test/prog_max_size.c
895 ./buildtools/Cmake/prog_test/prog_sem_init.c
896 ./buildtools/Cmake/prog_test/prog_stackgrowth.c
897 ./buildtools/Cmake/prog_test/prog_vsnprintf.c
898 ./buildtools/Cmake/prog_test/prog_AC_CHECK_MCSC.c
899 ./buildtools/Cmake/prog_test/prog_GRAS_CHECK_STRUCT_COMPACTION.c
900 ./buildtools/Cmake/prog_test/prog_mutex_timedlock.c
901 ./buildtools/Cmake/prog_test/prog_sem_timedwait.c
902 ./buildtools/Cmake/prog_test/prog_stacksetup.c
903 ./buildtools/Cmake/prog_test/prog_getline.c
904 ./buildtools/Cmake/prog_test/prog_gtnets.cpp
905 ./buildtools/Cmake/prog_test/prog_printf_null.c
906 ./buildtools/Cmake/prog_test/prog_snprintf.c
907 ./buildtools/Cmake/prog_test/prog_va_copy.c
909 CMakeLists for each binaries or examples:
912 ./teshsuite/gras/empty_main/CMakeLists.txt
913 ./teshsuite/gras/small_sleep/CMakeLists.txt
914 ./teshsuite/gras/datadesc/CMakeLists.txt
915 ./teshsuite/gras/msg_handle/CMakeLists.txt
916 ./teshsuite/simdag/CMakeLists.txt
917 ./teshsuite/simdag/partask/CMakeLists.txt
918 ./teshsuite/simdag/platforms/CMakeLists.txt
919 ./teshsuite/simdag/network/CMakeLists.txt
920 ./teshsuite/simdag/network/mxn/CMakeLists.txt
921 ./teshsuite/simdag/network/p2p/CMakeLists.txt
922 ./teshsuite/xbt/CMakeLists.txt
923 ./teshsuite/msg/CMakeLists.txt
924 ./tools/gras/CMakeLists.txt
925 ./tools/tesh/CMakeLists.txt
926 ./testsuite/simdag/CMakeLists.txt
927 ./testsuite/xbt/CMakeLists.txt
928 ./testsuite/surf/CMakeLists.txt
929 ./examples/gras/properties/CMakeLists.txt
930 ./examples/gras/ping/CMakeLists.txt
931 ./examples/gras/pmm/CMakeLists.txt
932 ./examples/gras/mmrpc/CMakeLists.txt
933 ./examples/gras/synchro/CMakeLists.txt
934 ./examples/gras/timer/CMakeLists.txt
935 ./examples/gras/mutual_exclusion/simple_token/CMakeLists.txt
936 ./examples/gras/spawn/CMakeLists.txt
937 ./examples/gras/chrono/CMakeLists.txt
938 ./examples/gras/rpc/CMakeLists.txt
939 ./examples/gras/all2all/CMakeLists.txt
940 ./examples/simdag/properties/CMakeLists.txt
941 ./examples/simdag/CMakeLists.txt
942 ./examples/simdag/metaxml/CMakeLists.txt
943 ./examples/simdag/dax/CMakeLists.txt
944 ./examples/smpi/CMakeLists.txt
945 ./examples/amok/bandwidth/CMakeLists.txt
946 ./examples/amok/saturate/CMakeLists.txt
947 ./examples/msg/priority/CMakeLists.txt
948 ./examples/msg/properties/CMakeLists.txt
949 ./examples/msg/migration/CMakeLists.txt
950 ./examples/msg/gtnets/CMakeLists.txt
951 ./examples/msg/parallel_task/CMakeLists.txt
952 ./examples/msg/trace/CMakeLists.txt
953 ./examples/msg/suspend/CMakeLists.txt
954 ./examples/msg/masterslave/CMakeLists.txt
955 ./examples/msg/actions/CMakeLists.txt
956 ./examples/msg/sendrecv/CMakeLists.txt
959 \section faq_howto Feature related questions
961 \subsection faq_MIA "Could you please add (your favorite feature here) to SimGrid?"
963 Here is the deal. The whole SimGrid project (MSG, SURF, GRAS, ...) is
964 meant to be kept as simple and generic as possible. We cannot add
965 functions for everybody's needs when these functions can easily be
966 built from the ones already in the API. Most of the time, it is
967 possible and when it was not possible we always have upgraded the API
968 accordingly. When somebody asks us a question like "How to do that?
969 Is there a function in the API to simply do this?", we're always glad
970 to answer and help. However if we don't need this code for our own
971 need, there is no chance we're going to write it... it's your job! :)
972 The counterpart to our answers is that once you come up with a neat
973 implementation of this feature (task duplication, RPC, thread
974 synchronization, ...), you should send it to us and we will be glad to
975 add it to the distribution. Thus, other people will take advantage of
976 it (and we don't have to answer this question again and again ;).
978 You'll find in this section a few "Missing In Action" features. Many
979 people have asked about it and we have given hints on how to simply do
980 it with MSG. Feel free to contribute...
982 \subsection faq_MIA_MSG MSG features
984 \subsubsection faq_MIA_examples I want some more complex MSG examples!
986 Many people have come to ask me a more complex example and each time,
987 they have realized afterward that the basics were in the previous three
990 Of course they have often been needing more complex functions like
991 MSG_process_suspend(), MSG_process_resume() and
992 MSG_process_isSuspended() (to perform synchronization), or
993 MSG_task_Iprobe() and MSG_process_sleep() (to avoid blocking
994 receptions), or even MSG_process_create() (to design asynchronous
995 communications or computations). But the examples are sufficient to
998 We know. We should add some more examples, but not really some more
999 complex ones... We should add some examples that illustrate some other
1000 functionalists (like how to simply encode asynchronous
1001 communications, RPC, process migrations, thread synchronization, ...)
1002 and we will do it when we will have a little bit more time. We have
1003 tried to document the examples so that they are understandable. Tell
1004 us if something is not clear and once again feel free to participate!
1007 \subsubsection faq_MIA_taskdup Missing in action: MSG Task duplication/replication
1009 There is no task duplication in MSG. When you create a task, you can
1010 process it or send it somewhere else. As soon as a process has sent
1011 this task, he doesn't have this task anymore. It's gone. The receiver
1012 process has got the task. However, you could decide upon receiving to
1013 create a "copy" of a task but you have to handle by yourself the
1014 semantic associated to this "duplication".
1016 As we already told, we prefer keeping the API as simple as
1017 possible. This kind of feature is rather easy to implement by users
1018 and the semantic you associate really depends on people. Having a
1019 *generic* task duplication mechanism is not that trivial (in
1020 particular because of the data field). That is why I would recommand
1021 that you write it by yourself even if I can give you advice on how to
1024 You have the following functions to get informations about a task:
1025 MSG_task_get_name(), MSG_task_get_compute_duration(),
1026 MSG_task_get_remaining_computation(), MSG_task_get_data_size(),
1027 and MSG_task_get_data().
1029 You could use a dictionary (#xbt_dict_t) of dynars (#xbt_dynar_t). If
1030 you still don't see how to do it, please come back to us...
1032 \subsubsection faq_MIA_asynchronous I want to do asynchronous communications in MSG
1034 Up until now, there is no asynchronous communications in MSG. However,
1035 you can create as many process as you want so you should be able to do
1036 whatever you want... I've written a queue module to help implementing
1037 some asynchronous communications at low cost (creating thousands of
1038 process only to handle communications may be problematic in term of
1039 performance at some point). I'll add it in the distribution asap.
1041 \subsubsection faq_MIA_thread_synchronization I need to synchronize my MSG processes
1043 You obviously cannot use pthread_mutexes of pthread_conds. The best
1044 thing would be to propose similar structures. Unfortunately, we
1045 haven't found time to do it yet. However you can try to play with
1046 MSG_process_suspend() and MSG_process_resume(). You can even do some
1047 synchronization with fake communications (using MSG_task_get(),
1048 MSG_task_put() and MSG_task_Iprobe()).
1050 \subsubsection faq_MIA_host_load Where is the get_host_load function hidden in MSG?
1052 There is no such thing because its semantic wouldn't be really
1053 clear. Of course, it is something about the amount of host throughput,
1054 but there is as many definition of "host load" as people asking for
1055 this function. First, you have to remember that resource availability
1056 may vary over time, which make any load notion harder to define.
1058 It may be instantaneous value or an average one. Moreover it may be only the
1059 power of the computer, or may take the background load into account, or may
1060 even take the currently running tasks into account. In some SURF models,
1061 communications have an influence on computational power. Should it be taken
1064 First of all, it's near to impossible to predict the load beforehands in the
1065 simulator since it depends on too much parameters (background load
1066 variation, bandwidth sharing algorithmic complexity) some of them even being
1067 not known beforehands (other task starting at the same time). So, getting
1068 this information is really hard (just like in real life). It's not just that
1069 we want MSG to be as painful as real life. But as it is in some way
1070 realistic, we face some of the same problems as we would face in real life.
1072 How would you do it for real? The most common option is to use something
1073 like NWS that performs active probes. The best solution is probably to do
1074 the same within MSG, as in next code snippet. It is very close from what you
1075 would have to do out of the simulator, and thus gives you information that
1076 you could also get in real settings to not hinder the realism of your
1080 double get_host_load() {
1081 m_task_t task = MSG_task_create("test", 0.001, 0, NULL);
1082 double date = MSG_get_clock();
1084 MSG_task_execute(task);
1085 date = MSG_get_clock() - date;
1086 MSG_task_destroy(task);
1087 return (0.001/date);
1091 Of course, it may not match your personal definition of "host load". In this
1092 case, please detail what you mean on the mailing list, and we will extend
1093 this FAQ section to fit your taste if possible.
1095 \subsubsection faq_MIA_communication_time How can I get the *real* communication time?
1097 Communications are synchronous and thus if you simply get the time
1098 before and after a communication, you'll only get the transmission
1099 time and the time spent to really communicate (it will also take into
1100 account the time spent waiting for the other party to be
1101 ready). However, getting the *real* communication time is not really
1102 hard either. The following solution is a good starting point.
1107 m_task_t task = MSG_task_create("Task", task_comp_size, task_comm_size,
1108 calloc(1,sizeof(double)));
1109 *((double*) task->data) = MSG_get_clock();
1110 MSG_task_put(task, slaves[i % slaves_count], PORT_22);
1111 INFO0("Send completed");
1116 m_task_t task = NULL;
1119 time1 = MSG_get_clock();
1120 a = MSG_task_get(&(task), PORT_22);
1121 time2 = MSG_get_clock();
1122 if(time1<*((double *)task->data))
1123 time1 = *((double *) task->data);
1124 INFO1("Communication time : \"%f\" ", time2-time1);
1126 MSG_task_destroy(task);
1131 \subsection faq_MIA_SimDag SimDag related questions
1133 \subsubsection faq_SG_comm Implementing communication delays between tasks.
1135 A classic question of SimDag newcomers is about how to express a
1136 communication delay between tasks. The thing is that in SimDag, both
1137 computation and communication are seen as tasks. So, if you want to
1138 model a data dependency between two DAG tasks t1 and t2, you have to
1139 create 3 SD_tasks: t1, t2 and c and add dependencies in the following
1143 SD_task_dependency_add(NULL, NULL, t1, c);
1144 SD_task_dependency_add(NULL, NULL, c, t2);
1147 This way task t2 cannot start before the termination of communication c
1148 which in turn cannot start before t1 ends.
1150 When creating task c, you have to associate an amount of data (in bytes)
1151 corresponding to what has to be sent by t1 to t2.
1153 Finally to schedule the communication task c, you have to build a list
1154 comprising the workstations on which t1 and t2 are scheduled (w1 and w2
1155 for example) and build a communication matrix that should look like
1158 \subsubsection faq_SG_DAG How to implement a distributed dynamic scheduler of DAGs.
1160 Distributed is somehow "contagious". If you start making distributed
1161 decisions, there is no way to handle DAGs directly anymore (unless I
1162 am missing something). You have to encode your DAGs in term of
1163 communicating process to make the whole scheduling process
1164 distributed. Here is an example of how you could do that. Assume T1
1165 has to be done before T2.
1168 int your_agent(int argc, char *argv[] {
1170 T1 = MSG_task_create(...);
1171 T2 = MSG_task_create(...);
1175 if(cond) MSG_task_execute(T1);
1177 if((MSG_task_get_remaining_computation(T1)=0.0) && (you_re_in_a_good_mood))
1178 MSG_task_execute(T2)
1180 /* do something else */
1186 If you decide that the distributed part is not that much important and that
1187 DAG is really the level of abstraction you want to work with, then you should
1188 give a try to \ref SD_API.
1190 \subsection faq_MIA_generic Generic features
1192 \subsubsection faq_more_processes Increasing the amount of simulated processes
1194 Here are a few tricks you can apply if you want to increase the amount
1195 of processes in your simulations.
1197 - <b>A few thousands of simulated processes</b> (soft tricks)\n
1198 SimGrid can use either pthreads library or the UNIX98 contextes. On
1199 most systems, the number of pthreads is limited and then your
1200 simulation may be limited for a stupid reason. This is especially
1201 true with the current linux pthreads, and I cannot get more than
1202 2000 simulated processes with pthreads on my box. The UNIX98
1203 contexts allow me to raise the limit to 25,000 simulated processes
1205 The <tt>--with-context</tt> option of the <tt>./configure</tt>
1206 script allows you to choose between UNIX98 contextes
1207 (<tt>--with-context=ucontext</tt>) and the pthread version
1208 (<tt>--with-context=pthread</tt>). The default value is ucontext
1209 when the script detect a working UNIX98 context implementation. On
1210 Windows boxes, the provided value is discarded and an adapted
1211 version is picked up.\n\n
1212 We experienced some issues with contextes on some rare systems
1213 (solaris 8 and lower or old alpha linuxes comes to mind). The main
1214 problem is that the configure script detect the contextes as being
1215 functional when it's not true. If you happen to use such a system,
1216 switch manually to the pthread version, and provide us with a good
1217 patch for the configure script so that it is done automatically ;)
1219 - <b>Hundred thousands of simulated processes</b> (hard-core tricks)\n
1220 As explained above, SimGrid can use UNIX98 contextes to represent
1221 and handle the simulated processes. Thanks to this, the main
1222 limitation to the number of simulated processes becomes the
1223 available memory.\n\n
1224 Here are some tricks I had to use in order to run a token ring
1225 between 25,000 processes on my laptop (1Gb memory, 1.5Gb swap).\n
1226 - First of all, make sure your code runs for a few hundreds
1227 processes before trying to push the limit. Make sure it's
1228 valgrind-clean, ie that valgrind does not report neither memory
1229 error nor memory leaks. Indeed, numerous simulated processes
1230 result in *fat* simulation hindering debugging.
1231 - It was really boring to write 25,000 entries in the deployment
1232 file, so I wrote a little script
1233 <tt>examples/gras/mutual_exclusion/simple_token/make_deployment.pl</tt>, which you may
1234 want to adapt to your case. You could also think about hijacking
1235 the SURFXML parser (have look at \ref faq_flexml_bypassing).
1236 - The deployment file became quite big, so I had to do what is in
1237 the FAQ entry \ref faq_flexml_limit
1238 - Each UNIX98 context has its own stack entry. As debugging this is
1239 quite hairly, the default value is a bit overestimated so that
1240 user don't get into trouble about this. You want to tune this
1241 size to increse the number of processes. This is the
1242 <tt>STACK_SIZE</tt> define in
1243 <tt>src/xbt/xbt_context_sysv.c</tt>, which is 128kb by default.
1244 Reduce this as much as you can, but be warned that if this value
1245 is too low, you'll get a segfault. The token ring example, which
1246 is quite simple, runs with 40kb stacks.
1247 - You may tweak the logs to reduce the stack size further. When
1248 logging something, we try to build the string to display in a
1249 char array on the stack. The size of this array is constant (and
1250 equal to XBT_LOG_BUFF_SIZE, defined in include/xbt/log/h). If the
1251 string is too large to fit this buffer, we move to a dynamically
1252 sized buffer. In which case, we have to traverse one time the log
1253 event arguments to compute the size we need for the buffer,
1254 malloc it, and traverse the argument list again to do the actual
1256 The idea here is to move XBT_LOG_BUFF_SIZE to 1, forcing the logs
1257 to use a dynamic array each time. This allows us to lower further
1258 the stack size at the price of some performance loss...\n
1259 This allowed me to run the reduce the stack size to ... 4k. Ie,
1260 on my 1Gb laptop, I can run more than 250,000 processes!
1262 \subsubsection faq_MIA_batch_scheduler Is there a native support for batch schedulers in SimGrid?
1264 No, there is no native support for batch schedulers and none is
1265 planned because this is a very specific need (and doing it in a
1266 generic way is thus very hard). However some people have implemented
1267 their own batch schedulers. Vincent Garonne wrote one during his PhD
1268 and put his code in the contrib directory of our SVN so that other can
1269 keep working on it. You may find inspiring ideas in it.
1271 \subsubsection faq_MIA_checkpointing I need a checkpointing thing
1273 Actually, it depends on whether you want to checkpoint the simulation, or to
1274 simulate checkpoints.
1276 The first one could help if your simulation is a long standing process you
1277 want to keep running even on hardware issues. It could also help to
1278 <i>rewind</i> the simulation by jumping sometimes on an old checkpoint to
1279 cancel recent calculations.\n
1280 Unfortunately, such thing will probably never exist in SG. One would have to
1281 duplicate all data structures because doing a rewind at the simulator level
1282 is very very hard (not talking about the malloc free operations that might
1283 have been done in between). Instead, you may be interested in the Libckpt
1284 library (http://www.cs.utk.edu/~plank/plank/www/libckpt.html). This is the
1285 checkpointing solution used in the condor project, for example. It makes it
1286 easy to create checkpoints (at the OS level, creating something like core
1287 files), and rerunning them on need.
1289 If you want to simulate checkpoints instead, it means that you want the
1290 state of an executing task (in particular, the progress made towards
1291 completion) to be saved somewhere. So if a host (and the task executing on
1292 it) fails (cf. #MSG_HOST_FAILURE), then the task can be restarted
1293 from the last checkpoint.\n
1295 Actually, such a thing does not exists in SimGrid either, but it's just
1296 because we don't think it is fundamental and it may be done in the user code
1297 at relatively low cost. You could for example use a watcher that
1298 periodically get the remaining amount of things to do (using
1299 MSG_task_get_remaining_computation()), or fragment the task in smaller
1302 \subsection faq_platform Platform building and Dynamic resources
1304 \subsubsection faq_platform_example Where can I find SimGrid platform files?
1306 There is several little examples in the archive, in the examples/msg
1307 directory. From time to time, we are asked for other files, but we
1308 don't have much at hand right now.
1310 You should refer to the Platform Description Archive
1311 (http://pda.gforge.inria.fr) project to see the other platform file we
1312 have available, as well as the Simulacrum simulator, meant to generate
1313 SimGrid platforms using all classical generation algorithms.
1315 \subsubsection faq_platform_alnem How can I automatically map an existing platform?
1317 We are working on a project called ALNeM (Application-Level Network
1318 Mapper) which goal is to automatically discover the topology of an
1319 existing network. Its output will be a platform description file
1320 following the SimGrid syntax, so everybody will get the ability to map
1321 their own lab network (and contribute them to the catalog project).
1322 This tool is not ready yet, but it move quite fast forward. Just stay
1325 \subsubsection faq_platform_synthetic Generating synthetic but realistic platforms
1327 The third possibility to get a platform file (after manual or
1328 automatic mapping of real platforms) is to generate synthetic
1329 platforms. Getting a realistic result is not a trivial task, and
1330 moreover, nobody is really able to define what "realistic" means when
1331 speaking of topology files. You can find some more thoughts on this
1333 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">slides</a>.
1335 If you are looking for an actual tool, there we have a little tool to
1336 annotate Tiers-generated topologies. This perl-script is in
1337 <tt>tools/platform_generation/</tt> directory of the SVN. Dinda et Al.
1338 released a very comparable tool, and called it GridG.
1340 \subsubsection faq_SURF_dynamic Expressing dynamic resource availability in platform files
1342 A nice feature of SimGrid is that it enables you to seamlessly have
1343 resources whose availability change over time. When you build a
1344 platform, you generally declare hosts like that:
1347 <host id="host A" power="100.00"/>
1350 If you want the availability of "host A" to change over time, the only
1351 thing you have to do is change this definition like that:
1354 <host id="host A" power="100.00" availability_file="trace_A.txt" state_file="trace_A_failure.txt"/>
1357 For hosts, availability files are expressed in fraction of available
1358 power. Let's have a look at what "trace_A.txt" may look like:
1367 At time 0, our host will deliver 100 flop/s. At time 11.0, it will
1368 deliver only 50 flop/s until time 20.0 where it will will start
1369 delivering 90 flop/s. Last at time 21.0 (20.0 plus the periodicity
1370 1.0), we'll be back to the beginning and it will deliver 100 flop/s.
1372 Now let's look at the state file:
1379 A negative value means "off" while a positive one means "on". At time
1380 1.0, the host is on. At time 1.0, it is turned off and at time 2.0, it
1381 is turned on again until time 12 (2.0 plus the periodicity 10.0). It
1382 will be turned on again at time 13.0 until time 23.0, and so on.
1384 Now, let's look how the same kind of thing can be done for network
1385 links. A usual declaration looks like:
1388 <link id="LinkA" bandwidth="10.0" latency="0.2"/>
1391 You have at your disposal the following options: bandwidth_file,
1392 latency_file and state_file. The only difference with hosts is that
1393 bandwidth_file and latency_file do not express fraction of available
1394 power but are expressed directly in bytes per seconds and seconds.
1396 \subsubsection faq_platform_multipath How to express multipath routing in platform files?
1398 It is unfortunately impossible to express the fact that there is more
1399 than one routing path between two given hosts. Let's consider the
1400 following platform file:
1403 <route src="A" dst="B">
1406 <route src="B" dst="C">
1409 <route src="A" dst="C">
1414 Although it is perfectly valid, it does not mean that data traveling
1415 from A to C can either go directly (using link 3) or through B (using
1416 links 1 and 2). It simply means that the routing on the graph is not
1417 trivial, and that data do not following the shortest path in number of
1418 hops on this graph. Another way to say it is that there is no implicit
1419 in these routing descriptions. The system will only use the routes you
1420 declare (such as <route src="A" dst="C"><link:ctn
1421 id="3"/></route>), without trying to build new routes by aggregating
1424 You are also free to declare platform where the routing is not
1425 symmetric. For example, add the following to the previous file:
1428 <route src="C" dst="A">
1434 This makes sure that data from C to A go through B where data from A
1435 to C go directly. Don't worry about realism of such settings since
1436 we've seen ways more weird situation in real settings (in fact, that's
1437 the realism of very regular platforms which is questionable, but
1438 that's another story).
1440 \subsubsection faq_flexml_bypassing Bypassing the XML parser with your own C functions
1442 So you want to bypass the XML files parser, uh? Maybe doing some parameter
1443 sweep experiments on your simulations or so? This is possible, and
1444 it's not even really difficult (well. Such a brutal idea could be
1445 harder to implement). Here is how it goes.
1447 For this, you have to first remember that the XML parsing in SimGrid is done
1448 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
1449 you want to bypass the parser, you need to provide some code mimicking what
1450 it does and replacing it in its interactions with the SURF code. So, let's
1451 have a look at these interactions.
1453 FleXML parser are close to classical SAX parsers. It means that a
1454 well-formed SimGrid platform XML file might result in the following
1457 - start "platform_description" with attribute version="2"
1458 - start "host" with attributes id="host1" power="1.0"
1460 - start "host" with attributes id="host2" power="2.0"
1462 - start "link" with ...
1464 - start "route" with ...
1465 - start "link:ctn" with ...
1468 - end "platform_description"
1470 The communication from the parser to the SURF code uses two means:
1471 Attributes get copied into some global variables, and a surf-provided
1472 function gets called by the parser for each event. For example, the event
1473 - start "host" with attributes id="host1" power="1.0"
1475 let the parser do something roughly equivalent to:
1477 strcpy(A_host_id,"host1");
1482 In SURF, we attach callbacks to the different events by initializing the
1483 pointer functions to some the right surf functions. Since there can be
1484 more than one callback attached to the same event (if more than one
1485 model is in use, for example), they are stored in a dynar. Example in
1486 workstation_ptask_L07.c:
1488 /* Adding callback functions */
1489 surf_parse_reset_parser();
1490 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
1491 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
1492 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
1493 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
1494 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
1495 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
1497 /* Parse the file */
1498 surf_parse_open(file);
1499 xbt_assert1((!surf_parse()), "Parse error in %s", file);
1503 So, to bypass the FleXML parser, you need to write your own version of the
1504 surf_parse function, which should do the following:
1505 - Fill the A_<tag>_<attribute> variables with the wanted values
1506 - Call the corresponding STag_<tag>_fun function to simulate tag start
1507 - Call the corresponding ETag_<tag>_fun function to simulate tag end
1508 - (do the same for the next set of values, and loop)
1510 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
1512 surf_parse = surf_parse_bypass_environment;
1513 MSG_create_environment(NULL);
1514 surf_parse = surf_parse_bypass_application;
1515 MSG_launch_application(NULL);
1518 A set of macros are provided at the end of
1519 include/surf/surfxml_parse.h to ease the writing of the bypass
1520 functions. An example of this trick is distributed in the file
1521 examples/msg/masterslave/masterslave_bypass.c
1523 \subsection faq_simgrid_configuration Changing SimGrid's behavior
1525 A number of options can be given at runtime to change the default
1526 SimGrid behavior. In particular, you can change the default cpu and
1529 \subsubsection faq_simgrid_configuration_gtnets Using GTNetS
1531 It is possible to use a packet-level network simulator
1532 instead of the default flow-based simulation. You may want to use such
1533 an approach if you have doubts about the validity of the default model
1534 or if you want to perform some validation experiments. At the moment,
1535 we support the GTNetS simulator (it is still rather experimental
1536 though, so leave us a message if you play with it).
1540 To enable GTNetS model inside SimGrid it is needed to patch the GTNetS simulator source code
1541 and build/install it from scratch
1544 - <b>Download and enter the recent downloaded GTNetS directory</b>
1547 svn checkout svn://scm.gforge.inria.fr/svn/simgrid/contrib/trunk/GTNetS/
1552 - <b>Use the following commands to unzip and patch GTNetS package to work within SimGrid.</b>
1555 unzip gtnets-current.zip
1556 tar zxvf gtnets-current-patch.tgz
1558 cat ../00*.patch | patch -p1
1561 - <b>OPTIONALLY</b> you can use a patch for itanium 64bit processor family.
1564 cat ../AMD64-FATAL-Removed-DUL_SIZE_DIFF-Added-fPIC-compillin.patch | patch -p1
1567 - <b>Compile GTNetS</b>
1569 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.
1573 ln -sf Makefile.linux Makefile
1579 - <b>NOTE</b> A lot of warnings are expected but the application should compile
1580 just fine. If the makefile insists in compiling some QT libraries
1581 please try a make clean before asking for help.
1584 - <b>To compile optimized version</b>
1591 - <b>Installing GTNetS</b>
1593 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.
1596 ln -sf /<absolute_path>/gtnets_current/libgtsim-debug.so /<userhome>/usr/lib/libgtnets.so
1597 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/<userhome>/usr/lib/libgtnets.so
1598 mkdir /<userhome>/usr/include/gtnets
1599 cp -fr SRC/*.h /<userhome>/usr/include/gtnets
1603 - <b>Enable GTNetS support in SimGrid</b>
1606 ./configure --with-gtnets=/<userhome>/usr
1609 - <b>Once you have followed all the instructions for compiling and
1610 installing successfully you can activate this feature at
1611 runntime with the following options:</b>
1614 cd simgrid/example/msg/
1620 - <b>Or try the GTNetS model dogbone example with</b>
1623 gtnets/gtnets gtnets/onelink-p.xml gtnets/onelink-d.xml --cfg=network_model:GTNets
1627 A long version of this <a href="http://gforge.inria.fr/docman/view.php/12/6283/GTNetS HowTo.html">HowTo</a> it is available
1630 More about GTNetS simulator at <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/index.html">GTNetS Website</a>
1634 The patches provided by us worked successfully with GTNetS found
1635 <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/software/gtnets-current.zip">here</a>,
1636 dated from 12th June 2008. Due to the discontinuing development of
1637 GTNetS it is impossible to precise a version number. We STRONGLY recommend you
1638 to download and install the GTNetS version found in SimGrid repository as explained above.
1643 \subsubsection faq_simgrid_configuration_alternate_network Using alternative flow models
1645 The default simgrid network model uses a max-min based approach as
1646 explained in the research report
1647 <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>.
1648 Other models have been proposed and implemented since then (see for example
1649 <a href="http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf">Accuracy Study and Improvement of Network Simulation in the SimGrid Framework</a>)
1650 and can be activated at runtime. For example:
1652 ./mycode platform.xml deployment.xml --cfg=workstation_model:compound --cfg=network_model:LV08 -cfg=cpu_model:Cas01
1655 Possible models for the network are currently "Constant", "CM02",
1656 "LegrandVelho", "GTNets", Reno", "Reno2", "Vegas". Others will
1657 probably be added in the future and many of the previous ones are
1658 experimental and are likely to disappear without notice...
1660 \subsection faq_tracing Tracing Simulations for Visualization
1662 The trace visualization is widely used to observe and understand the behavior
1663 of parallel applications and distributed algorithms. Usually, this is done in a
1664 two-step fashion: the user instruments the application and the traces are
1665 analyzed after the end of the execution. The visualization itself can highlights
1666 unexpected behaviors, bottlenecks and sometimes can be used to correct
1667 distributed algorithms. The SimGrid team is currently instrumenting the library
1668 in order to let users trace their simulations and analyze them. This part of the
1669 user manual explains how the tracing-related features can be enabled and used
1670 during the development of simulators using the SimGrid library.
1672 \subsubsection faq_tracing_howitworks How it works
1674 For now, the SimGrid library is instrumented so users can trace the <b>platform
1675 utilization</b> using the MSG interface. This means that the tracing will
1676 register how much power is used for each host and how much bandwidth is used for
1677 each link of the platform. The idea with this type of tracing is to observe the
1678 overall view of resources utilization in the first place, especially the
1679 identification of bottlenecks, load-balancing among hosts, and so on.
1681 The idea of the instrumentation is to classify the MSG tasks by category,
1683 the platform utilization (hosts and links) for each of the categories. For that,
1684 the tracing interface enables the declaration of categories and a function to
1685 mark a task with a previously declared category. <em>The tasks that are not
1686 classified according to a category are not traced</em>.
1688 \subsubsection faq_tracing_enabling Enabling using CMake
1690 With the sources of SimGrid, it is possible to enable the tracing
1691 using the parameter <b>-Dtracing=on</b> when the cmake is executed.
1692 The section \ref faq_tracing_functions describes all the functions available
1693 when this Cmake options is activated. These functions will have no effect
1694 if SimGrid is configured without this option (they are wiped-out by the
1698 $ cmake -Dtracing=on .
1702 \subsubsection faq_tracing_functions Tracing Functions
1704 \subsubsubsection Mandatory Functions
1706 \li <b>\c TRACE_start (const char *filename)</b>: This is the first function to
1707 be called. It receives a single argument as parameter that contains the name of
1708 the file that will hold the trace in the end of the simulation. It returns 0 if
1709 everything was properly initialized, 1 otherwise. All trace functions called
1710 before TRACE_start do nothing.
1712 \li <b>\c TRACE_category (const char *category)</b>: This function should be used
1713 to define a user category. The category can be used to differentiate the tasks
1714 that are created during the simulation (for example, tasks from server1,
1715 server2, or request tasks, computation tasks, communication tasks).
1716 All resource utilization (host power and link bandwidth) will be
1717 classified according to the task category. Tasks that do not belong to a
1718 category are not traced.
1720 \li <b>\c TRACE_msg_set_task_category (m_task_t task, const char *category)</b>:
1721 This function should be called after the creation of a task, to define the
1722 category of that task. The first parameter \c task must contain a task that was
1723 created with the function \c MSG_task_create. The second parameter
1724 \c category must contain a category that was previously defined by the function
1727 \li <b>\c TRACE_end ()</b>: This is the last function to be called. It closes
1728 the trace file and stops the tracing of the simulation. All tracing will be
1729 completely disabled after the calling this function. Although we recommend
1730 the use of this function somewhere in the end of program, it can be used
1731 anywhere in the code. This function returns 0 if everything is ok, 1 otherwise.
1733 \subsubsubsection Optional Functions
1735 \li <b>\c TRACE_host_variable_declare (const char *variable)</b>:
1736 Declare a user variable that will be associated to hosts. A variable can
1737 be used to trace user variables such as the number of tasks in a server,
1738 the number of clients in an application, and so on.
1740 \li <b>\c TRACE_host_variable_[set|add|sub] (const char *variable, double
1742 Set the value of a given user variable. It is important to remind that
1743 the value of this variable is always associated to the host. The host
1744 that will be used when these functions are called is the one returned by
1745 the function \c MSG_host_self().
1747 \subsubsection faq_tracing_example Example of Instrumentation
1749 A simplified example using the tracing mandatory functions.
1752 int main (int argc, char **argv)
1754 TRACE_start ("traced_simulation.trace");
1755 TRACE_category ("request");
1756 TRACE_category ("computation");
1757 TRACE_category ("finalize");
1759 MSG_global_init (&argc, &argv);
1761 //(... after deployment ...)
1763 m_task_t req1 = MSG_task_create("1st_request_task", 10, 10, NULL);
1764 m_task_t req2 = MSG_task_create("2nd_request_task", 10, 10, NULL);
1765 m_task_t req3 = MSG_task_create("3rd_request_task", 10, 10, NULL);
1766 m_task_t req4 = MSG_task_create("4th_request_task", 10, 10, NULL);
1767 TRACE_msg_set_task_category (req1, "request");
1768 TRACE_msg_set_task_category (req2, "request");
1769 TRACE_msg_set_task_category (req3, "request");
1770 TRACE_msg_set_task_category (req4, "request");
1772 m_task_t comp = MSG_task_create ("comp_task", 100, 100, NULL);
1773 TRACE_msg_set_task_category (comp, "computation");
1775 m_task_t finalize = MSG_task_create ("finalize", 0, 0, NULL);
1776 TRACE_msg_set_task_category (finalize, "finalize");
1787 \subsubsection faq_tracing_analyzing Analyzing the SimGrid Traces
1789 The SimGrid library, during an instrumented simulation, creates a trace file in
1790 the Paje file format that contains the platform utilization for the simulation
1791 that was executed. The visualization analysis of this file is performed with the
1792 visualization tool <a href="http://triva.gforge.inria.fr">Triva</a>, with
1793 special configurations tunned to SimGrid needs. This part of the documentation
1794 explains how to configure and use Triva to analyse a SimGrid trace file.
1796 - <b>Installing Triva</b>: the tool is available in the INRIAGforge,
1797 at <a href="http://triva.gforge.inria.fr">http://triva.gforge.inria.fr</a>.
1798 Use the following command to get the sources, and then check the file
1799 <i>INSTALL.simplified</i>. This file contains instructions to install
1800 the tool's dependencies in a Ubuntu/Debian Linux.
1802 $ svn checkout svn://scm.gforge.inria.fr/svn/triva
1804 $ cat INSTALL.simplified
1807 - <b>Executing Triva</b>: a binary called <i>Triva</i> is available after the
1808 installation (you can execute it passing <em>--help</em> to check its
1809 options). If the triva binary is not available after following the
1810 installation instructions, you may want to execute the following command to
1811 initialize the GNUstep environment variables (note that the location of the
1812 <i>GNUstep.sh</i> file may vary depending on your GNUstep installation - the
1813 command is known to work in Ubuntu and Debian Linux):
1815 $ source /usr/share/GNUstep/Makefiles/GNUstep.sh
1817 You should be able to see this output after the installation of triva:
1819 $ ./Triva.app/Triva --help
1820 Usage: Triva [OPTION...] TRACEFILE
1821 Trace Analysis through Visualization
1823 You need to use one of the following options:
1824 -g, --graph Graph Analysis
1825 -t, --treemap Treemap Analysis
1827 Other auxiliary options to check the trace file:
1828 -c, --check Check the integrity of trace file
1829 -h, --hierarchy Export the trace type hierarchy
1830 -l, --list List entity types
1832 -?, --help Give this help list
1833 --usage Give a short usage message
1835 Triva expects that the user choose one of the available options
1836 (currently <em>--graph</em> or <em>--treemap</em> for a visualization analysis)
1837 and the trace file from the simulation.
1839 - <b>Understanding Triva - time-slice</b>: the analysis of a trace file using
1840 the tool always takes into account the concept of the <em>time-slice</em>.
1841 This concept means that what is being visualized in the screen is always
1842 calculated considering a specific time frame, with its beggining and end
1843 timestamp. The time-slice is configured by the user and can be changed
1844 dynamically through the window called <em>Time Interval</em> that is opened
1845 whenever a trace file is being analyzed. The next figure depicts the time-slice
1846 configuration window.
1847 In the top of the window, in the space named <i>Trace Time</i>,
1848 the two fields show the beggining of the trace (which usually starts in 0) and
1849 the end (that depends on the time simulated by SimGrid). The middle of the
1850 window, in the square named <i>Time Slice Configuration</i>, contains the
1851 aspects related to the time-slice, including its <i>start</i> and its
1852 <i>size</i>. The gray rectangle in the bottom of this part indicates the
1853 <i>current time-slice</i> that is considered for the drawings. If the checkbox
1854 <i>Update Drawings on Sliders Change</i> is not selected, the button
1855 <i>Apply</i> must be clicked in order to inform triva that the
1856 new time-slice must be considered. The bottom part of the window, in the space
1857 indicated by the square <i>Time Slice Animation</i> can be used to advance
1858 the time-frame automatically. The user configures the amount of time that the
1859 time-frame will forward and how frequent this update will happen. Once this is
1860 configured, the user clicks the <i>Play</i> button in order to see the dynamic
1861 changes on the drawings.
1864 <a href="triva-time_interval.png" border=0><img src="triva-time_interval.png" width="50%" border=0></a>
1867 <b>Remarks:</b> when the trace has too many hosts or links, the computation to
1868 take into account a new time-slice can be expensive. When this happens, the
1869 <i>Frequency</i> parameter, but also updates caused by change on configurations
1870 when the checkbox <i>Update Drawings on Sliders
1871 Change</i> is selected will not be followed.
1873 - <b>Understanding Triva - graph</b>: this part of the documention explains how
1874 to analyze the traces using the graph view of Triva, when the user executes
1875 the tool passing <em>--graph</em> as parameter. Triva opens three windows when
1876 this parameter is used: the <i>Time Interval</i> window (previously described),
1877 the <i>Graph Representation</i> window, and the <em>Graph Configuration</em>
1878 window. The Graph Representation is the window where drawings take place.
1879 Initially, it is completely white waiting for a proper graph configuration input
1880 by the user. We start the description of this type of analysis by describing the
1881 <i>Graph Configuration</i> window (depicted below). By using a particular
1882 configuration, triva
1883 can be used to customize the graph drawing according to
1884 the SimGrid trace that was created with user-specific categories. Before delving
1885 into the details of this customization, let us first explain the major parts of
1886 the graph configuration window. The buttons located in the top-right corner can
1887 be used to delete, copy and create a new configuration. The checkbox in the
1888 top-middle part of the window indicates if the configuration typed in the
1889 textfield is syntactically correct (we are using the non-XML
1890 <a href="http://en.wikipedia.org/wiki/Property_list">Property List Format</a> to
1891 describe the configuration). The pop-up button located on the top-left corner
1892 indicates the selected configuration (the user can have multiple graph
1893 configurations). The bottom-left text field contains the name of the current
1894 configuration (updates on this field must be followed by typing enter on the
1895 keyboard to take into account the name change). The bottom-right <em>Apply</em>
1896 button activates the current configuration, resulting on an update on the graph
1900 <a href="triva-graph_configuration.png" border=0><img src="triva-graph_configuration.png" width="50%" border=0></a>
1903 <b>Basic SimGrid Configuration</b>: The figure shows in the big textfield the
1904 basic configuration that should be used during the analysis of a SimGrid trace
1905 file. The basic logic of the configuration is as follows:
1911 The nodes of the graph will be created based on the <i>node</i> parameter, which
1912 in this case is the different <em>"HOST"</em>s of the platform
1913 used to simulate. The <i>edge</i> parameter indicates that the edges of the
1914 graph will be created based on the <em>"LINK"</em>s of the platform. After the
1915 definition of these two parameters, the configuration must detail how
1916 <em>HOST</em>s and <em>LINK</em>s should be drawn. For that, the configuration
1917 must have an entry for each of the types used. For <em>HOST</em>, as basic
1918 configuration, we have:
1925 The parameter <em>size</em> indicates which variable from the trace file will be
1926 used to define the size of the node HOST in the visualization. If the simulation
1927 was executed with availability traces, the size of the nodes will be changed
1928 according to these traces. The parameter <em>scale</em> indicates if the value
1929 of the variable is <em>global</em> or <em>local</em>. If it is global, the value
1930 will be relative to the power of all other hosts, if it is local, the value will
1931 be relative locally.
1932 For <em>LINK</em> we have:
1942 For the types specified in the <em>edge</em> parameter (such as <em>LINK</em>),
1943 the configuration must contain two additional parameters: <em>src</em> and
1944 <em>dst</em> that are used to properly identify which nodes this edge is
1945 connecting. The values <em>SrcHost</em> and <em>DstHost</em> are always present
1946 in the SimGrid trace file and should not be changed in the configuration. The
1947 parameter <em>size</em> for the LINK, in this case, is configured as the
1948 variable <em>bandwidth</em>, with a <em>global</em> scale. The scale meaning
1949 here is exactly the same used for nodes. The last parameter is the GraphViz
1950 algorithm used to calculate the position of the nodes in the graph
1953 graphviz-algorithm = neato;
1956 <b>Customizing the Graph Representation</b>: triva is capable to handle
1957 a customized graph representation based on the variables present in the trace
1958 file. In the case of SimGrid, every time a category is created for tasks, two
1959 variables in the trace file are defined: one to indicate node utilization (how
1960 much power was used by that task category), and another to indicate link
1961 utilization (how much bandwidth was used by that category). For instance, if the
1962 user declares a category named <i>request</i>, there will be variables named
1963 <b>p</b><i>request</i> and a <b>b</b><i>request</i> (<b>p</b> for power and
1964 <b>b</b> for bandwidth). It is important to notice that the variable
1965 <i>prequest</i> in this case is only available for HOST, and
1966 <i>brequest</i> is only available for LINK. <b>Example</b>: suppose there are
1967 two categories for tasks: request and compute. To create a customized graph
1968 representation with a proportional separation of host and link utilization, use
1969 as configuration for HOST and LINK this:
1978 values = (prequest, pcomputation);
1991 values = (brequest, bcomputation);
1995 Where <i>sep_host</i> contains a composition of type <i>separation</i> where
1996 its max size is the <i>power</i> of the host and the variables <i>prequest</i>
1997 and <i>pcomputation</i> are drawn proportionally to the size of the HOST. And
1998 <i>sep_link</i> is also a separation where max is defined as the
1999 <i>bandwidth</i> of the link, and the variables <i>brequest</i> and
2000 <i>bcomputation</i> are drawn proportionally within a LINK.
2001 <i>This configuration enables the analysis of resource utilization by MSG tasks,
2002 and the identification of load-balancing issues, network bottlenecks, for
2004 <b>Other compositions</b>: besides <i>separation</i>, it is possible to use
2005 other types of compositions, such as gradients, and colors, like this:
2010 values = (numberOfTasks);
2014 values = (is_server);
2017 Where <i>gra_host</i> creates a gradient within a node of the graph, using a
2018 global scale and using as value a variable called <i>numberOfTasks</i>, that
2019 could be declared by the user using the optional tracing functions of SimGrid.
2020 If scale is global, the max and min value for the gradient will be equal to the
2021 max and min numberOfTasks among all hosts, and if scale is local, the max and
2022 min value based on the value of numberOfTasks locally in each host.
2023 And <i>color_host</i> composition draws a square based on a positive value of
2024 the variable <i>is_server</i>, that could also be defined by the user using the
2025 SimGrid tracing functions. \n
2026 <b>The Graph Visualization</b>: The next figure shows a graph visualization of a
2027 given time-slice of the masterslave_forwarder example (present in the SimGrid
2028 sources). The red color indicates tasks from the <i>compute</i> category. This
2029 visualization was generated with the following configuration:
2042 values = (pcompute, pfinalize);
2054 values = (bcompute, bfinalize);
2057 graphviz-algorithm = neato;
2062 <a href="triva-graph_visualization.png" border=0><img src="triva-graph_visualization.png" width="50%" border=0></a>
2066 - <b>Understading Triva - colors</b>: An important issue when using Triva is how
2067 to define colors. To do that, we have to know which variables are defined in
2068 the trace file generated by the SimGrid library. The parameter <em>--list</em>
2069 lists the variables for a given trace file:
2071 $ Triva -l masterslave_forwarder.trace
2089 We can see that HOST has seven variables (from power to pfinalize) and LINK has
2090 four (from bandwidth to bfinalize). To define a red color for the
2091 <i>pcompute</i> and <i>bcompute</i> (which are defined based on user category
2092 <i>compute</i>), execute:
2094 $ defaults write Triva 'pcompute Color' '1 0 0'
2095 $ defaults write Triva 'bcompute Color' '1 0 0'
2097 Where the three numbers in each line are the RGB color with values from 0 to 1.
2099 \section faq_troubleshooting Troubleshooting
2101 \subsection faq_trouble_lib_compil SimGrid compilation and installation problems
2103 \subsubsection faq_trouble_lib_config ./configure fails!
2105 We know only one reason for the configure to fail:
2107 - <b>You are using a broken build environment</b>\n
2108 If symptom is that configure complains about gcc not being able to build
2109 executables, you are probably missing the libc6-dev package. Damn Ubuntu.
2111 If you experience other kind of issue, please get in touch with us. We are
2112 always interested in improving our portability to new systems.
2114 \subsubsection faq_trouble_distcheck Dude! "make check" fails on my machine!
2116 Don't assume we never run this target, because we do. Check
2117 http://bob.loria.fr:8010 if you don't believe us.
2119 There is several reasons which may cause the make check to fail on your
2122 - <b>You are using a broken libc (probably concerning the contextes)</b>.\n
2123 The symptom is that the "make check" fails within the examples/msg directory.\n
2124 By default, SimGrid uses something called ucontexts. This is part of the
2125 libc, but it's quite undertested. For example, some (old) versions of the
2126 glibc on alpha do not implement these functions, but provide the stubs
2127 (which return ENOSYS: not implemented). It may fool our detection mechanism
2128 and leads to segfaults. There is not much we can do to fix the bug.
2129 A workaround is to compile with --with-context=pthread to avoid
2130 ucontext completely. You'll be a bit more limited in the number
2131 of simulated processes you can start concurrently, but 5000
2132 processes is still enough for most purposes, isn't it?\n
2133 This limitation is the reason why we insist on using this piece of ...
2134 software even if it's so troublesome.\n
2135 <b>=> use --with-pthread on AMD64 architecture that do not have an
2136 ultra-recent libc.</b>
2138 - <b>There is a bug in SimGrid we aren't aware of</b>.\n
2139 If none of the above apply, please drop us a mail on the mailing list so
2140 that we can check it out. Make sure to read \ref faq_bugrepport
2143 \subsection faq_trouble_compil User code compilation problems
2145 \subsubsection faq_trouble_err_logcat "gcc: _simgrid_this_log_category_does_not_exist__??? undeclared (first use in this function)"
2147 This is because you are using the log mecanism, but you didn't created
2148 any default category in this file. You should refer to \ref XBT_log
2149 for all the details, but you simply forgot to call one of
2150 XBT_LOG_NEW_DEFAULT_CATEGORY() or XBT_LOG_NEW_DEFAULT_SUBCATEGORY().
2152 \subsubsection faq_trouble_pthreadstatic "gcc: undefined reference to pthread_key_create"
2154 This indicates that one of the library SimGrid depends on (libpthread
2155 here) was missing on the linking command line. Dependencies of
2156 libsimgrid are expressed directly in the dynamic library, so it's
2157 quite impossible that you see this message when doing dynamic linking.
2159 If you compile your code statically (and if you use a pthread version
2160 of SimGrid -- see \ref faq_more_processes), you must absolutely
2161 specify <tt>-lpthread</tt> on the linker command line. As usual, this should
2162 come after <tt>-lsimgrid</tt> on this command line.
2164 \subsection faq_trouble_errors Runtime error messages
2166 \subsubsection faq_flexml_limit "surf_parse_lex: Assertion `next limit' failed."
2168 This is because your platform file is too big for the parser.
2170 Actually, the message comes directly from FleXML, the technology on top of
2171 which the parser is built. FleXML has the bad idea of fetching the whole
2172 document in memory before parsing it. And moreover, the memory buffer size
2173 must be determined at compilation time.
2175 We use a value which seems big enough for our need without bloating the
2176 simulators footprints. But of course your mileage may vary. In this case,
2177 just edit src/surf/surfxml.l modify the definition of
2178 FLEXML_BUFFERSTACKSIZE. E.g.
2181 #define FLEXML_BUFFERSTACKSIZE 1000000000
2184 Then recompile and everything should be fine, provided that your version of
2185 Flex is recent enough (>= 2.5.31). If not the compilation process should
2188 A while ago, we worked on FleXML to reduce a bit its memory consumption, but
2189 these issues remain. There is two things we should do:
2191 - use a dynamic buffer instead of a static one so that the only limit
2192 becomes your memory, not a stupid constant fixed at compilation time
2193 (maybe not so difficult).
2194 - change the parser so that it does not need to get the whole file in
2195 memory before parsing
2196 (seems quite difficult, but I'm a complete newbe wrt flex stuff).
2198 These are changes to FleXML itself, not SimGrid. But since we kinda hijacked
2199 the development of FleXML, I can grant you that any patches would be really
2200 welcome and quickly integrated.
2202 <b>Update:</b> A new version of FleXML (1.7) was released. Most of the work
2203 was done by William Dowling, who use it in his own work. The good point is
2204 that it now use a dynamic buffer, and that the memory usage was greatly
2205 improved. The downside is that William also changed some things internally,
2206 and it breaks the hack we devised to bypass the parser, as explained in
2207 \ref faq_flexml_bypassing. Indeed, this is not a classical usage of the
2208 parser, and Will didn't imagine that we may have used (and even documented)
2209 such a crude usage of FleXML. So, we now have to repair the bypassing
2210 functionality to use the lastest FleXML version and fix the memory usage in
2213 \subsubsection faq_trouble_gras_transport GRAS spits networking error messages
2215 Gras, on real platforms, naturally use regular sockets to communicate. They
2216 are deeply hidden in the gras abstraction, but when things go wrong, you may
2217 get some weird error messages. Here are some example, with the probable
2220 - <b>Transport endpoint is not connected</b>: several processes try to open
2221 a server socket on the same port number of the same machine. This is
2222 naturally bad and each process should pick its own port number for this.\n
2223 Maybe, you just have some processes remaining from a previous experiment
2225 Killing them may help, but again if you kill -KILL them, you'll have to
2226 wait for a while: they didn't close there sockets properly and the system
2227 needs a while to notice that this port is free again.
2229 - <b>Socket closed by remote side</b>: if the remote process is not
2230 supposed to close the socket at this point, it may be dead.
2232 - <b>Connection reset by peer</b>: I found this on Internet about this
2233 error. I think it's what's happening here, too:\n
2234 <i>This basically means that a network error occurred while the client was
2235 receiving data from the server. But what is really happening is that the
2236 server actually accepts the connection, processes the request, and sends
2237 a reply to the client. However, when the server closes the socket, the
2238 client believes that the connection has been terminated abnormally
2239 because the socket implementation sends a TCP reset segment telling the
2240 client to throw away the data and report an error.\n
2241 Sometimes, this problem is caused by not properly closing the
2242 input/output streams and the socket connection. Make sure you close the
2243 input/output streams and socket connection properly. If everything is
2244 closed properly, however, and the problem persists, you can work around
2245 it by adding a one-second sleep before closing the streams and the
2246 socket. This technique, however, is not reliable and may not work on all
2248 Since GRAS sockets are closed properly (repeat after me: there is no bug
2249 in GRAS), it is either that you are closing your sockets on server side
2250 before the client get a chance to read them (use gras_os_sleep() to delay
2251 the server), or the server died awfully before the client got the data.
2253 \subsubsection faq_trouble_errors_big_fat_warning I'm told that my XML files are too old.
2255 The format of the XML platform description files is sometimes
2256 improved. For example, we decided to change the units used in SimGrid
2257 from MBytes, MFlops and seconds to Bytes, Flops and seconds to ease
2258 people exchanging small messages. We also reworked the route
2259 descriptions to allow more compact descriptions.
2261 That is why the XML files are versionned using the 'version' attribute
2262 of the root tag. Currently, it should read:
2264 <platform version="2">
2267 If your files are too old, you can use the simgrid_update_xml.pl
2268 script which can be found in the tools directory of the archive.
2270 \subsection faq_trouble_valgrind Valgrind-related and other debugger issues
2272 If you don't, you really should use valgrind to debug your code, it's
2275 \subsubsection faq_trouble_vg_longjmp longjmp madness in valgrind
2277 This is when valgrind starts complaining about longjmp things, just like:
2279 \verbatim ==21434== Conditional jump or move depends on uninitialised value(s)
2280 ==21434== at 0x420DBE5: longjmp (longjmp.c:33)
2282 ==21434== Use of uninitialised value of size 4
2283 ==21434== at 0x420DC3A: __longjmp (__longjmp.S:48)
2286 This is the sign that you didn't used the exception mecanism well. Most
2287 probably, you have a <tt>return;</tt> somewhere within a <tt>TRY{}</tt>
2288 block. This is <b>evil</b>, and you must not do this. Did you read the section
2291 \subsubsection faq_trouble_vg_libc Valgrind spits tons of errors about backtraces!
2293 It may happen that valgrind, the memory debugger beloved by any decent C
2294 programmer, spits tons of warnings like the following :
2295 \verbatim ==8414== Conditional jump or move depends on uninitialised value(s)
2296 ==8414== at 0x400882D: (within /lib/ld-2.3.6.so)
2297 ==8414== by 0x414EDE9: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2298 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2299 ==8414== by 0x414F937: _dl_open (in /lib/tls/i686/cmov/libc-2.3.6.so)
2300 ==8414== by 0x4150F4C: (within /lib/tls/i686/cmov/libc-2.3.6.so)
2301 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
2302 ==8414== by 0x415102D: __libc_dlopen_mode (in /lib/tls/i686/cmov/libc-2.3.6.so)
2303 ==8414== by 0x412D6B9: backtrace (in /lib/tls/i686/cmov/libc-2.3.6.so)
2304 ==8414== by 0x8076446: xbt_dictelm_get_ext (dict_elm.c:714)
2305 ==8414== by 0x80764C1: xbt_dictelm_get (dict_elm.c:732)
2306 ==8414== by 0x8079010: xbt_cfg_register (config.c:208)
2307 ==8414== by 0x806821B: MSG_config (msg_config.c:42)
2310 This problem is somewhere in the libc when using the backtraces and there is
2311 very few things we can do ourselves to fix it. Instead, here is how to tell
2312 valgrind to ignore the error. Add the following to your ~/.valgrind.supp (or
2313 create this file on need). Make sure to change the obj line according to
2314 your personnal mileage (change 2.3.6 to the actual version you are using,
2315 which you can retrieve with a simple "ls /lib/ld*.so").
2318 name: Backtrace madness
2320 obj:/lib/ld-2.3.6.so
2325 fun:__libc_dlopen_mode
2328 Then, you have to specify valgrind to use this suppression file by passing
2329 the <tt>--suppressions=$HOME/.valgrind.supp</tt> option on the command line.
2330 You can also add the following to your ~/.bashrc so that it gets passed
2331 automatically. Actually, it passes a bit more options to valgrind, and this
2332 happen to be my personnal settings. Check the valgrind documentation for
2335 \verbatim export VALGRIND_OPTS="--leak-check=yes --leak-resolution=high --num-callers=40 --tool=memcheck --suppressions=$HOME/.valgrind.supp" \endverbatim
2337 \subsubsection faq_trouble_backtraces Truncated backtraces
2339 When debugging SimGrid, it's easier to pass the
2340 --disable-compiler-optimization flag to the configure if valgrind or
2341 gdb get fooled by the optimization done by the compiler. But you
2342 should remove these flag when everything works before going in
2343 production (before launching your 1252135 experiments), or everything
2344 will run only one half of the true SimGrid potential.
2346 \subsection faq_deadlock There is a deadlock in my code!!!
2348 Unfortunately, we cannot debug every code written in SimGrid. We
2349 furthermore believe that the framework provides ways enough
2350 information to debug such informations yourself. If the textual output
2351 is not enough, Make sure to check the \ref faq_visualization FAQ entry to see
2352 how to get a graphical one.
2354 Now, if you come up with a really simple example that deadlocks and
2355 you're absolutely convinced that it should not, you can ask on the
2356 list. Just be aware that you'll be severely punished if the mistake is
2357 on your side... We have plenty of FAQ entries to redact and new
2358 features to implement for the impenitents! ;)
2360 \subsection faq_surf_network_latency I get weird timings when I play with the latencies.
2362 OK, first of all, remember that units should be Bytes, Flops and
2363 Seconds. If you don't use such units, some SimGrid constants (e.g. the
2364 SG_TCP_CTE_GAMMA constant used in most network models) won't have the
2365 right unit and you'll end up with weird results.
2367 Here is what happens with a single transfer of size L on a link
2368 (bw,lat) when nothing else happens.
2371 0-----lat--------------------------------------------------t
2372 |-----|**** real_bw =min(bw,SG_TCP_CTE_GAMMA/(2*lat)) *****|
2375 In more complex situations, this min is the solution of a complex
2376 max-min linear system. Have a look
2377 <a href="http://lists.gforge.inria.fr/pipermail/simgrid-devel/2006-April/thread.html">here</a>
2378 and read the two threads "Bug in SURF?" and "Surf bug not
2379 fixed?". You'll have a few other examples of such computations. You
2380 can also read "A Network Model for Simulation of Grid Application" by
2381 Henri Casanova and Loris Marchal to have all the details. The fact
2382 that the real_bw is smaller than bw is easy to understand. The fact
2383 that real_bw is smaller than SG_TCP_CTE_GAMMA/(2*lat) is due to the
2384 window-based congestion mechanism of TCP. With TCP, you can't exploit
2385 your huge network capacity if you don't have a good round-trip-time
2386 because of the acks...
2388 Anyway, what you get is t=lat + L/min(bw,SG_TCP_CTE_GAMMA/(2*lat)).
2390 * if I you set (bw,lat)=(100 000 000, 0.00001), you get t = 1.00001 (you fully
2392 * if I you set (bw,lat)=(100 000 000, 0.0001), you get t = 1.0001 (you're on the
2394 * if I you set (bw,lat)=(100 000 000, 0.001), you get t = 10.001 (ouch!)
2396 This bound on the effective bandwidth of a flow is not the only thing
2397 that may make your result be unexpected. For example, two flows
2398 competing on a saturated link receive an amount of bandwidth inversely
2399 proportional to their round trip time.
2401 \subsection faq_bugrepport So I've found a bug in SimGrid. How to report it?
2403 We do our best to make sure to hammer away any bugs of SimGrid, but this is
2404 still an academic project so please be patient if/when you find bugs in it.
2405 If you do, the best solution is to drop an email either on the simgrid-user
2406 or the simgrid-devel mailing list and explain us about the issue. You can
2407 also decide to open a formal bug report using the
2408 <a href="https://gforge.inria.fr/tracker/?atid=165&group_id=12&func=browse">relevant
2409 interface</a>. You need to login on the server to get the ability to submit
2412 We will do our best to solve any problem repported, but you need to help us
2413 finding the issue. Just telling "it segfault" isn't enough. Telling "It
2414 segfaults when running the attached simulator" doesn't really help either.
2415 You may find the following article interesting to see how to repport
2416 informative bug repports:
2417 http://www.chiark.greenend.org.uk/~sgtatham/bugs.html (it is not SimGrid
2418 specific at all, but it's full of good advices).
2420 \author Arnaud Legrand (arnaud.legrand::imag.fr)
2421 \author Martin Quinson (martin.quinson::loria.fr)
2426 ******************************************************************
2427 * OLD CRUFT NOT USED ANYMORE *
2428 ******************************************************************
2431 \subsection faq_crosscompile Cross-compiling a Windows DLL of SimGrid from linux
2433 At the moment, we do not distribute Windows pre-compiled version of SimGrid
2434 because the support for this platform is still experimental. We know that
2435 some parts of the GRAS environment do not work, and we think that the others
2436 environments (MSG and SD) have good chances to work, but we didn't test
2437 ourselves. This section explains how we generate the SimGrid DLL so that you
2438 can build it for yourself. First of all, you need to have a version more
2439 recent than 3.1 (ie, a SVN version as time of writting).
2441 In order to cross-compile the package to windows from linux, you need to
2442 install mingw32 (minimalist gnu win32). On Debian, you can do so by
2443 installing the packages mingw32 (compiler), mingw32-binutils (linker and
2444 so), mingw32-runtime.
2446 You can use the VPATH support of configure to compile at the same time for
2447 linux and windows without dupplicating the source nor cleaning the tree
2448 between each. Just run bootstrap (if you use the SVN) to run the autotools.
2449 Then, create a linux and a win directories. Then, type:
2450 \verbatim cd linux; ../configure --srcdir=.. <usual configure flags>; make; cd ..
2451 cd win; ../configure --srcdir=.. --host=i586-mingw32msvc <flags>; make; cd ..
2453 The trick to VPATH builds is to call configure from another directory,
2454 passing it an extra --srcdir argument to tell it where all the sources are.
2455 It will understand you want to use VPATH. Then, the trick to cross-compile
2456 is simply to add a --host argument specifying the target you want to build
2457 for. The i586-mingw32msvc string is what you have to pass to use the mingw32
2458 environment as distributed in Debian.
2460 After that, you can run all make targets from both directories, and test
2461 easily that what you change for one arch does not break the other one.
2463 It is possible that this VPATH build thing breaks from time to time in the
2464 SVN since it's quite fragile, but it's granted to work in any released
2465 version. If you experience problems, drop us a mail.
2467 Another possible source of issue is that at the moment, building the
2468 examples request to use the gras_stub_generator tool, which is a compiled
2469 program, not a script. In cross-compilation, you need to cross-execute with
2470 wine for example, which is not really pleasant. We are working on this, but
2471 in the meanwhile, simply don't build the examples in cross-compilation
2472 (<tt>cd src</tt> before running make).
2474 Program (cross-)compiled with mingw32 do request an extra DLL at run-time to be
2475 usable. For example, if you want to test your build with wine, you should do
2476 the following to put this library where wine looks for DLLs.
2478 cp /usr/share/doc/mingw32-runtime/mingwm10.dll.gz ~/.wine/c/windows/system/
2479 gunzip ~/.wine/c/windows/system/mingwm10.dll.gz
2482 The DLL is built in src/.libs, and installed in the <i>prefix</i>/bin directory
2483 when you run make install.
2485 If you want to use it in a native project on windows, you need to use
2486 simgrid.dll and mingwm10.dll. For each DLL, you need to build .def file
2487 under linux (listing the defined symbols), and convert it into a .lib file
2488 under windows (specifying this in a way that windows compilers like). To
2489 generate the def files, run (under linux):
2490 \verbatim echo "LIBRARY libsimgrid-0.dll" > simgrid.def
2491 echo EXPORTS >> simgrid.def
2492 nm libsimgrid-0.dll | grep ' T _' | sed 's/.* T _//' >> simgrid.def
2493 nm libsimgrid-0.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> simgrid.def
2495 echo "LIBRARY mingwm10.dll" > mingwm10.def
2496 echo EXPORTS >> mingwm10.def
2497 nm mingwm10.dll | grep ' T _' | sed 's/.* T _//' >> mingwm10.def
2498 nm mingwm10.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> mingwm10.def
2501 To create the import .lib files, use the <tt>lib</tt> windows tool (from
2502 MSVC) the following way to produce simgrid.lib and mingwm10.lib
2503 \verbatim lib /def:simgrid.def
2504 lib /def:mingwm10.def
2507 If you happen to use Borland C Builder, the right command line is the
2508 following (note that you don't need any file.def to get this working).
2509 \verbatim implib simgrid.lib libsimgrid-0.dll
2510 implib mingwm10.lib mingwm10.dll
2513 Then, set the following parameters in Visual C++ 2005:
2514 Linker -> Input -> Additional dependencies = simgrid.lib mingwm10.lib
2516 Just in case you wonder how to generate a DLL from libtool in another
2517 project, we added -no-undefined to any lib*_la_LDFLAGS variables so that
2518 libtool accepts to generate a dynamic library under windows. Then, to make
2519 it true, we pass any dependencies (such as -lws2 under windows or -lpthread
2520 on need) on the linking line. Passing such deps is a good idea anyway so
2521 that they get noted in the library itself, avoiding the users to know about
2522 our dependencies and put them manually on their compilation line. Then we
2523 added the AC_LIBTOOL_WIN32_DLL macro just before AC_PROG_LIBTOOL in the
2524 configure.ac. It means that we exported any symbols which need to be.
2525 Nowadays, functions get automatically exported, so we don't need to load our
2526 header files with tons of __declspec(dllexport) cruft. We only need to do so
2527 for data, but there is no public data in SimGrid so we are good.
