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 MSG_paje_output(). It
76 generates an input to <a href="http://www-id.imag.fr/Logiciels/paje/">Paje</a>.
79 <a href="Paje_MSG_screenshot.jpg"><img src="Paje_MSG_screenshot_thn.jpg"></a>
83 Visualization with Paje can be seen as a kind of postmortem
84 analysis. However, as soon as you start playing with big simulations,
85 you'll realize that processing such output is kind of tricky. There is
86 so much generic information that it is hard to find the information
89 As a matter of fact, logging really depends on simulations (e.g. what
90 kind of events is important...). That is why we do not propose a big
91 dump of your whole simulation (it would slow everything down) but give
92 you neat tools to structure you logs. Have a look at \ref XBT_log. In
93 fact, rather than a post-mortem analysis, you may want to do it on the
94 fly. The process you are running can do whatever you want. Have you
95 thought about adding a global structure where you directly compute the
96 information that are really important rather than writing everything
97 down and then processing huge files?
99 \subsection faq_C Argh! Do I really have to code in C?
101 Up until now, there is no binding for other languages. If you use C++,
102 you should be able to use the SimGrid library as a standard C library
103 and everything should work fine (simply <i>link</i> against this
104 library; recompiling SimGrid with a C++ compiler won't work and it
105 wouldn't help if you could).
107 In fact, we are currently working on Java bindings of MSG to allow
108 all the undergrad students of the world to use this tool. This is a
109 little more tricky than I would have expected, but the work is moving
110 fast forward [2006/05/13]. More languages are evaluated, but for now,
111 we do not feel a real demand for any other language. Please speak up!
113 \section faq_installation Installing the SimGrid library
115 Many people have been asking me questions on how to use SimGrid. Quite
116 often, the questions were not really about SimGrid but on the
117 installation process. This section is intended to help people that are
118 not familiar with compiling C files under UNIX. If you follow these
119 instructions and still have some troubles, drop an e-mail to
120 <simgrid-user@lists.gforge.inria.fr>.
122 \subsection faq_compiling Compiling SimGrid from a stable archive
124 First of all, you need to download the latest version of SimGrid from
125 <a href="http://gforge.inria.fr/frs/?group_id=12">here</a>.
126 Suppose you have uncompressed SimGrid in some temporary location of
127 your home directory (say <tt>/home/joe/tmp/simgrid-3.0.1 </tt>). The
128 simplest way to use SimGrid is to install it in your home
129 directory. Change your directory to
130 <tt>/home/joe/tmp/simgrid-3.0.1</tt> and type
133 ./configure --prefix=$HOME
138 If at some point, something fails, check the section \ref faq_trouble_compil .
139 If it does not help, you can report this problem to the
140 list but, please, avoid sending a laconic mail like "There is a problem. Is it
141 okay?". Send the config.log file which is automatically generated by
142 configure. Try to capture both the standard output and the error output of the
143 <tt>make</tt> command with <tt>script</tt>. There is no way for us to help you
144 without the relevant bits of information.
146 Now, the following directory should have been created :
148 \li <tt>/home/joe/doc/simgrid/html/</tt>
149 \li <tt>/home/joe/lib/</tt>
150 \li <tt>/home/joe/include/</tt>
152 SimGrid is not a binary, it is a library. Both a static and a dynamic
153 version are available. Here is what you can find if you try a <tt>ls
156 \verbatim libsimgrid.a libsimgrid.la libsimgrid.so libsimgrid.so.0 libsimgrid.so.0.0.1
159 Thus, there is two ways to link your program with SimGrid:
160 \li Either you use the static version, e.g
161 \verbatim gcc libsimgrid.a -o MainProgram MainProgram.c
163 In this case, all the SimGrid functions are directly
164 included in <tt>MainProgram</tt> (hence a bigger binary).
165 \li Either you use the dynamic version (the preferred method)
166 \verbatim gcc -lsimgrid -o MainProgram MainProgram.c
168 In this case, the SimGrid functions are not included in
169 <tt>MainProgram</tt> and you need to set your environment
170 variable in such a way that <tt>libsimgrid.so</tt> will be
171 found at runtime. This can be done by adding the following
172 line in your .bashrc (if you use bash and if you have
173 installed the SimGrid libraries in your home directory):
174 \verbatim export LD_LIBRARY_PATH=$HOME/lib/:$LD_LIBRARY_PATH
177 \section faq_installationCmake Installing the SimGrid library with cmake
178 This section required cmake tools with a 2.6 version at least. You can
179 also try to install ccmake which is a graphical tool for cmake. Then
180 download the last version of simgrid <a href="http://gforge.inria.fr/frs/?group_id=12">here</a>
181 or checkout the svn with this command line .
182 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk
185 Then open a terminal to simgrid directory.
187 \subsection faq_PARTI-0 Using Cmake (generality)
189 Cmake is a tool to generate makefiles on different plateforms. Here are descibed more usefull command line.
192 "cmake ./" configure the project ( note1 )
193 "make" build all tagets
194 "make test" test all targets and summarize
195 "make package" make the distrib ( note2 )
196 "make install-simgrid" install the project (doc/ lib/ include/)
197 "make clean" clean all targets
199 \li note1 : for options see section \ref faq_PARTII.
200 \li note2 : doc is needed see section \ref faq_PARTIII.
202 \subsection faq_PARTI-1 Using Ctest (generality)
204 Ctest is a tool embeded by cmake and can be link to a dashbord (like cdash).
207 "ctest" launch only tests
208 "ctest -D Experimental" for -->|configure
215 \subsection faq_PARTI-2 How to install
217 Install lib from a svn repository
220 cmake -Denable_maintainer_mode=on ./
225 Install lib from a distrib repository
228 cmake -Dprefix=/home/navarrop/Bureau/install_simgrid ./
233 \subsection faq_PARTII Cmake options
235 Regarding options you can configure manually or with an executable
237 Compilation with command line
240 "cmake -D[name]=[value] ... ./"
242 [name] disable_gtnets [value] ON/OFF or TRUE/FALSE or 1/0
243 disable_java ON/OFF or TRUE/FALSE or 1/0
244 disable_lua ON/OFF or TRUE/FALSE or 1/0
245 disable_ruby ON/OFF or TRUE/FALSE or 1/0
247 enable_compile_optimizations ON/OFF or TRUE/FALSE or 1/0
248 enable_compile_warnings ON/OFF or TRUE/FALSE or 1/0
249 enable_maintainer_mode ON/OFF or TRUE/FALSE or 1/0
251 supernovae ON/OFF or TRUE/FALSE or 1/0
253 gtnets_path <path_to_gtnets_directory>
254 prefix <path_to_install_directory>
255 with_context auto/ucontext/pthread/window
258 Compilation with ccmake gui tool
261 "ccmake ./" then follow instructions.
264 \subsection faq_PARTIII Explaination of options
266 -->disable_gtnets : set to true implie that user doesn't want to use gtnets.
268 -->disable_java : set to true implie that user doesn't want to add java langage into simgrid compilation.
270 -->disable_lua : set to true implie that user doesn't want to add lua langage into simgrid compilation.
272 -->disable_ruby : set to true implie that user doesn't want to add ruby langage into simgrid compilation.
274 -->enable_compile_optimizations : add flags "-O3 -finline-functions -funroll-loops -fno-strict-aliasing"
276 -->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"
278 -->enable_maintainer_mode : set to true make doc and remake files with flex flexml.
281 /src/gras/DataDesc/ddt_parse.yy.c
282 /src/surf/simgrid_dtd.c
284 /src/simdag/dax_dtd.c
285 /include/surf/simgrid_dtd.h
286 /include/xbt/graphxml.h
287 /src/simdag/dax_dtd.h
289 -->supernovae : set to true make one file for each lib and compile with those generated files.
292 /src/supernovae_gras.c
293 /src/supernovae_smpi.c
295 -->gtnets_path : Path to gtnets install directory (ex /usr)
297 -->prefix : Path where are installed lib/ doc/ and include/ directories (ex /usr/local)
299 -->with context : specify which context the user wants to use.
301 \subsection faq_exemple Use and compilation result
304 navarrop@caraja:~$ cd Bureau/simgrid-trunk/
305 navarrop@caraja:~/Bureau/simgrid-trunk$ cmake ./
307 GTnetS doesn't works : set -Ddisable_gtnets=on <-|some warnings are printed
308 with_context auto change to ucontext <-|
309 (skaddr) <--info (needed)
310 (sksize) <--info (needed)
311 Make : src/simgrid.jar with : /usr/bin/javac <--info (if java)
312 Make examples/java with : /usr/bin/javac <--info (if java)
314 Configuration of package `simgrid' (revision 7209M) on arch (=4):
315 BUILDNAME : UCONTEXT <-- name of the compilation regarding to cdash
316 SITE : Linux_Ubuntu 9.10_x86_64 <-- distribution of the local machine regarding to cdash
318 Compiler: c++ : /usr/bin/c++
319 version: c++ (Ubuntu 4.4.1-4ubuntu9) 4.4.1
321 version: gcc (Ubuntu 4.4.1-4ubuntu9) 4.4.1
327 Context backend: ucontext
337 Simgrid dependencies: -ldl -llua5.1
338 Gras dependencies: pthread
342 USER_PREFIX: /usr/local
343 INSTALL_PREFIX: /usr/local
347 -- Build files have been written to: /home/navarrop/Bureau/simgrid-trunk
349 Here all options are checked and printed. If it doesn't match with your configuration it is probably due to a wrong configuration.
351 \subsection faq_compiling_java Java bindings don't get compiled
353 The configure script detects automatically whether you have the
354 softwares needed to use the Java bindings or not. At the end of the
355 configure, you can see the configuration picked by the script, which
356 should look similar to
357 \verbatim Configuration of package simgrid' (version 3.3.4-svn) on
360 Compiler: gcc (version: )
362 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
366 Context backend: ucontext
373 In this example, Java backends won't be compiled.
375 On Debian-like systems (which includes ubuntu), you need the following
376 packages: sun-java6-jdk libgcj10-dev. If you cannot find the
377 libgcj10-dev, try another version, like libgcj9-dev (on Ubuntu before
378 9.10) or libgcj11-dev (not released yet, but certainly one day).
379 Please note that you need to activate the contrib and non-free
380 repositories in Debian, and the universe ones in Ubuntu. Java comes at
383 \subsection faq_compiling_snapshoot SimGrid development snapshots
385 We have very high standards on software quality, and we are reluctant releasing
386 a stable release as long as there is still some known bug in the code base. In
387 addition, we added quite an extensive test base, making sure that we correctly
388 test the most important parts of the tool.
390 As an unfortunate conclusion, there may be some time between the stable
391 releases. If you want to benefit from the most recent features we introduced,
392 but don't want to take the risk of an untested version from the SVN, then
393 development snapshots are done for you.
395 These are pre-releases of SimGrid that still fail some tests about features
396 that almost nobody use, or on platforms not being in our core target (which is
397 Linux, Mac, other Unixes and Windows, from the most important to the less
398 one). That means that using this development releases should be safe for most
401 These archives can be found on
402 <a href="http://www.loria.fr/~quinson/simgrid.html">this web page</a>. Once you
403 got the lastest archive, you can compile it just like any archive (see above).
405 \subsection faq_compiling_svn Compiling SimGrid from the SVN
407 The project development takes place in the SVN, where all changes are
408 committed when they happen. Then every once in a while, we make sure that the
409 code quality meets our standard and release an archive from the code in the
410 SVN. We afterward go back to the development in the SVN. So, if you need a
411 recently added feature and can afford some little problem with the stability
412 of the lastest features, you may want to use the SVN version instead of a
415 For that, you first need to get the "simgrid" module from
416 <a href="http://gforge.inria.fr/scm/?group_id=12">here</a>.
418 You won't find any <tt>configure</tt> and a few other things
419 (<tt>Makefile.in</tt>'s, documentation, ...) will be missing as well. The
420 reason for that is that all these files have to be regenerated using the
421 latest versions of <tt>autoconf</tt>, <tt>libtool</tt>, <tt>automake</tt>
422 (>1.9) and <tt>doxygen</tt> (>1.4). To generate the <tt>configure</tt> and
423 the <tt>Makefile.in</tt>'s, you just have to launch the <tt>bootstrap</tt>
424 command that resides in the top of the source tree. Then just follow the
425 instructions of Section \ref faq_compiling.
427 We insist on the fact that you really need the latest versions of
428 autoconf, automake and libtool. Doing this step on exotic architectures/systems
429 (i.e. anything different from a recent linux distribution) may be
430 ... uncertain. If you need to compile the SVN version on a machine where all these
431 dependencies are not met, the easiest is to do <tt>make dist</tt> in the SVN
432 directory of another machine where all dependencies are met. It will create an
433 archive you may deploy on other sites just as a regular stable release.
435 In summary, the following commands will checkout the SVN, regenerate the
436 configure script and friends, configure SimGrid and build it.
438 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk simgrid
441 ./configure --enable-maintainer-mode --prefix=<where to install SimGrid>
444 Then, if you want to install SimGrid on the current box, just do:
445 \verbatim make install \endverbatim
447 If you want to build an snapshot of the SVN to deploy it on another box (for
448 example because the other machine don't have the autotools), do:
449 \verbatim make dist \endverbatim
451 Moreover, you should never call the autotools manually since you must run
452 them in a specific order with specific arguments. Most of the times, the
453 makefiles will automatically call the tools for you. When it's not possible
454 (such as the first time you checkout the SVN), use the ./bootstrap command
455 to call them explicitly.
458 \subsection faq_setting_MSG Setting up your own MSG code
460 Do not build your simulator by modifying the SimGrid examples. Go
461 outside the SimGrid source tree and create your own working directory
462 (say <tt>/home/joe/SimGrid/MyFirstScheduler/</tt>).
464 Suppose your simulation has the following structure (remember it is
465 just an example to illustrate a possible way to compile everything;
466 feel free to organize it as you want).
468 \li <tt>sched.h</tt>: a description of the core of the
469 scheduler (i.e. which functions are can be used by the
470 agents). For example we could find the following functions
471 (master, forwarder, slave).
473 \li <tt>sched.c</tt>: a C file including <tt>sched.h</tt> and
474 implementing the core of the scheduler. Most of these
475 functions use the MSG functions defined in section \ref
478 \li <tt>masterslave.c</tt>: a C file with the main function, i.e.
479 the MSG initialization (MSG_global_init()), the platform
480 creation (e.g. with MSG_create_environment()), the
481 deployment phase (e.g. with MSG_function_register() and
482 MSG_launch_application()) and the call to
485 To compile such a program, we suggest to use the following
486 Makefile. It is a generic Makefile that we have used many times with
487 our students when we teach the C language.
491 masterslave: masterslave.o sched.o
493 INSTALL_PATH = $$HOME
495 PEDANTIC_PARANOID_FREAK = -O0 -Wshadow -Wcast-align \
496 -Waggregate-return -Wmissing-prototypes -Wmissing-declarations \
497 -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations \
498 -Wmissing-noreturn -Wredundant-decls -Wnested-externs \
499 -Wpointer-arith -Wwrite-strings -finline-functions
500 REASONABLY_CAREFUL_DUDE = -Wall
501 NO_PRAYER_FOR_THE_WICKED = -w -O2
502 WARNINGS = $(REASONABLY_CAREFUL_DUDE)
503 CFLAGS = -g $(WARNINGS)
505 INCLUDES = -I$(INSTALL_PATH)/include
506 DEFS = -L$(INSTALL_PATH)/lib/
507 LDADD = -lm -lsimgrid
511 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) $^ $(LIBS) $(LDADD) -o $@
514 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) -c -o $@ $<
517 rm -f $(BIN_FILES) *.o *~
523 The first two lines indicates what should be build when typing make
524 (<tt>masterslave</tt>) and of which files it is to be made of
525 (<tt>masterslave.o</tt> and <tt>sched.o</tt>). This makefile assumes
526 that you have set up correctly your <tt>LD_LIBRARY_PATH</tt> variable
527 (look, there is a <tt>LDADD = -lm -lsimgrid</tt>). If you prefer using
528 the static version, remove the <tt>-lsimgrid</tt> and add a
529 <tt>$(INSTALL_PATH)/lib/libsimgrid.a</tt> on the next line, right
530 after the <tt>LIBS = </tt>.
532 More generally, if you have never written a Makefile by yourself, type
533 in a terminal : <tt>info make</tt> and read the introduction. The
534 previous example should be enough for a first try but you may want to
535 perform some more complex compilations...
537 \subsection faq_setting_GRAS Setting up your own GRAS code
539 If you use the GRAS interface instead of the MSG one, then previous section
540 is not the better source of information. Instead, you should check the GRAS
541 tutorial in general, and the \ref GRAS_tut_tour_setup in particular.
544 \section faq_howto Feature related questions
546 \subsection faq_MIA "Could you please add (your favorite feature here) to SimGrid?"
548 Here is the deal. The whole SimGrid project (MSG, SURF, GRAS, ...) is
549 meant to be kept as simple and generic as possible. We cannot add
550 functions for everybody's needs when these functions can easily be
551 built from the ones already in the API. Most of the time, it is
552 possible and when it was not possible we always have upgraded the API
553 accordingly. When somebody asks us a question like "How to do that?
554 Is there a function in the API to simply do this?", we're always glad
555 to answer and help. However if we don't need this code for our own
556 need, there is no chance we're going to write it... it's your job! :)
557 The counterpart to our answers is that once you come up with a neat
558 implementation of this feature (task duplication, RPC, thread
559 synchronization, ...), you should send it to us and we will be glad to
560 add it to the distribution. Thus, other people will take advantage of
561 it (and we don't have to answer this question again and again ;).
563 You'll find in this section a few "Missing In Action" features. Many
564 people have asked about it and we have given hints on how to simply do
565 it with MSG. Feel free to contribute...
567 \subsection faq_MIA_MSG MSG features
569 \subsubsection faq_MIA_examples I want some more complex MSG examples!
571 Many people have come to ask me a more complex example and each time,
572 they have realized afterward that the basics were in the previous three
575 Of course they have often been needing more complex functions like
576 MSG_process_suspend(), MSG_process_resume() and
577 MSG_process_isSuspended() (to perform synchronization), or
578 MSG_task_Iprobe() and MSG_process_sleep() (to avoid blocking
579 receptions), or even MSG_process_create() (to design asynchronous
580 communications or computations). But the examples are sufficient to
583 We know. We should add some more examples, but not really some more
584 complex ones... We should add some examples that illustrate some other
585 functionalists (like how to simply encode asynchronous
586 communications, RPC, process migrations, thread synchronization, ...)
587 and we will do it when we will have a little bit more time. We have
588 tried to document the examples so that they are understandable. Tell
589 us if something is not clear and once again feel free to participate!
592 \subsubsection faq_MIA_taskdup Missing in action: MSG Task duplication/replication
594 There is no task duplication in MSG. When you create a task, you can
595 process it or send it somewhere else. As soon as a process has sent
596 this task, he doesn't have this task anymore. It's gone. The receiver
597 process has got the task. However, you could decide upon receiving to
598 create a "copy" of a task but you have to handle by yourself the
599 semantic associated to this "duplication".
601 As we already told, we prefer keeping the API as simple as
602 possible. This kind of feature is rather easy to implement by users
603 and the semantic you associate really depends on people. Having a
604 *generic* task duplication mechanism is not that trivial (in
605 particular because of the data field). That is why I would recommand
606 that you write it by yourself even if I can give you advice on how to
609 You have the following functions to get informations about a task:
610 MSG_task_get_name(), MSG_task_get_compute_duration(),
611 MSG_task_get_remaining_computation(), MSG_task_get_data_size(),
612 and MSG_task_get_data().
614 You could use a dictionary (#xbt_dict_t) of dynars (#xbt_dynar_t). If
615 you still don't see how to do it, please come back to us...
617 \subsubsection faq_MIA_asynchronous I want to do asynchronous communications in MSG
619 Up until now, there is no asynchronous communications in MSG. However,
620 you can create as many process as you want so you should be able to do
621 whatever you want... I've written a queue module to help implementing
622 some asynchronous communications at low cost (creating thousands of
623 process only to handle communications may be problematic in term of
624 performance at some point). I'll add it in the distribution asap.
626 \subsubsection faq_MIA_thread_synchronization I need to synchronize my MSG processes
628 You obviously cannot use pthread_mutexes of pthread_conds. The best
629 thing would be to propose similar structures. Unfortunately, we
630 haven't found time to do it yet. However you can try to play with
631 MSG_process_suspend() and MSG_process_resume(). You can even do some
632 synchronization with fake communications (using MSG_task_get(),
633 MSG_task_put() and MSG_task_Iprobe()).
635 \subsubsection faq_MIA_host_load Where is the get_host_load function hidden in MSG?
637 There is no such thing because its semantic wouldn't be really
638 clear. Of course, it is something about the amount of host throughput,
639 but there is as many definition of "host load" as people asking for
640 this function. First, you have to remember that resource availability
641 may vary over time, which make any load notion harder to define.
643 It may be instantaneous value or an average one. Moreover it may be only the
644 power of the computer, or may take the background load into account, or may
645 even take the currently running tasks into account. In some SURF models,
646 communications have an influence on computational power. Should it be taken
649 First of all, it's near to impossible to predict the load beforehands in the
650 simulator since it depends on too much parameters (background load
651 variation, bandwidth sharing algorithmic complexity) some of them even being
652 not known beforehands (other task starting at the same time). So, getting
653 this information is really hard (just like in real life). It's not just that
654 we want MSG to be as painful as real life. But as it is in some way
655 realistic, we face some of the same problems as we would face in real life.
657 How would you do it for real? The most common option is to use something
658 like NWS that performs active probes. The best solution is probably to do
659 the same within MSG, as in next code snippet. It is very close from what you
660 would have to do out of the simulator, and thus gives you information that
661 you could also get in real settings to not hinder the realism of your
665 double get_host_load() {
666 m_task_t task = MSG_task_create("test", 0.001, 0, NULL);
667 double date = MSG_get_clock();
669 MSG_task_execute(task);
670 date = MSG_get_clock() - date;
671 MSG_task_destroy(task);
676 Of course, it may not match your personal definition of "host load". In this
677 case, please detail what you mean on the mailing list, and we will extend
678 this FAQ section to fit your taste if possible.
680 \subsubsection faq_MIA_communication_time How can I get the *real* communication time?
682 Communications are synchronous and thus if you simply get the time
683 before and after a communication, you'll only get the transmission
684 time and the time spent to really communicate (it will also take into
685 account the time spent waiting for the other party to be
686 ready). However, getting the *real* communication time is not really
687 hard either. The following solution is a good starting point.
692 m_task_t task = MSG_task_create("Task", task_comp_size, task_comm_size,
693 calloc(1,sizeof(double)));
694 *((double*) task->data) = MSG_get_clock();
695 MSG_task_put(task, slaves[i % slaves_count], PORT_22);
696 INFO0("Send completed");
701 m_task_t task = NULL;
704 time1 = MSG_get_clock();
705 a = MSG_task_get(&(task), PORT_22);
706 time2 = MSG_get_clock();
707 if(time1<*((double *)task->data))
708 time1 = *((double *) task->data);
709 INFO1("Communication time : \"%f\" ", time2-time1);
711 MSG_task_destroy(task);
716 \subsection faq_MIA_SimDag SimDag related questions
718 \subsubsection faq_SG_comm Implementing communication delays between tasks.
720 A classic question of SimDag newcomers is about how to express a
721 communication delay between tasks. The thing is that in SimDag, both
722 computation and communication are seen as tasks. So, if you want to
723 model a data dependency between two DAG tasks t1 and t2, you have to
724 create 3 SD_tasks: t1, t2 and c and add dependencies in the following
728 SD_task_dependency_add(NULL, NULL, t1, c);
729 SD_task_dependency_add(NULL, NULL, c, t2);
732 This way task t2 cannot start before the termination of communication c
733 which in turn cannot start before t1 ends.
735 When creating task c, you have to associate an amount of data (in bytes)
736 corresponding to what has to be sent by t1 to t2.
738 Finally to schedule the communication task c, you have to build a list
739 comprising the workstations on which t1 and t2 are scheduled (w1 and w2
740 for example) and build a communication matrix that should look like
743 \subsubsection faq_SG_DAG How to implement a distributed dynamic scheduler of DAGs.
745 Distributed is somehow "contagious". If you start making distributed
746 decisions, there is no way to handle DAGs directly anymore (unless I
747 am missing something). You have to encode your DAGs in term of
748 communicating process to make the whole scheduling process
749 distributed. Here is an example of how you could do that. Assume T1
750 has to be done before T2.
753 int your_agent(int argc, char *argv[] {
755 T1 = MSG_task_create(...);
756 T2 = MSG_task_create(...);
760 if(cond) MSG_task_execute(T1);
762 if((MSG_task_get_remaining_computation(T1)=0.0) && (you_re_in_a_good_mood))
765 /* do something else */
771 If you decide that the distributed part is not that much important and that
772 DAG is really the level of abstraction you want to work with, then you should
773 give a try to \ref SD_API.
775 \subsection faq_MIA_generic Generic features
777 \subsubsection faq_more_processes Increasing the amount of simulated processes
779 Here are a few tricks you can apply if you want to increase the amount
780 of processes in your simulations.
782 - <b>A few thousands of simulated processes</b> (soft tricks)\n
783 SimGrid can use either pthreads library or the UNIX98 contextes. On
784 most systems, the number of pthreads is limited and then your
785 simulation may be limited for a stupid reason. This is especially
786 true with the current linux pthreads, and I cannot get more than
787 2000 simulated processes with pthreads on my box. The UNIX98
788 contexts allow me to raise the limit to 25,000 simulated processes
790 The <tt>--with-context</tt> option of the <tt>./configure</tt>
791 script allows you to choose between UNIX98 contextes
792 (<tt>--with-context=ucontext</tt>) and the pthread version
793 (<tt>--with-context=pthread</tt>). The default value is ucontext
794 when the script detect a working UNIX98 context implementation. On
795 Windows boxes, the provided value is discarded and an adapted
796 version is picked up.\n\n
797 We experienced some issues with contextes on some rare systems
798 (solaris 8 and lower or old alpha linuxes comes to mind). The main
799 problem is that the configure script detect the contextes as being
800 functional when it's not true. If you happen to use such a system,
801 switch manually to the pthread version, and provide us with a good
802 patch for the configure script so that it is done automatically ;)
804 - <b>Hundred thousands of simulated processes</b> (hard-core tricks)\n
805 As explained above, SimGrid can use UNIX98 contextes to represent
806 and handle the simulated processes. Thanks to this, the main
807 limitation to the number of simulated processes becomes the
808 available memory.\n\n
809 Here are some tricks I had to use in order to run a token ring
810 between 25,000 processes on my laptop (1Gb memory, 1.5Gb swap).\n
811 - First of all, make sure your code runs for a few hundreds
812 processes before trying to push the limit. Make sure it's
813 valgrind-clean, ie that valgrind does not report neither memory
814 error nor memory leaks. Indeed, numerous simulated processes
815 result in *fat* simulation hindering debugging.
816 - It was really boring to write 25,000 entries in the deployment
817 file, so I wrote a little script
818 <tt>examples/gras/mutual_exclusion/simple_token/make_deployment.pl</tt>, which you may
819 want to adapt to your case. You could also think about hijacking
820 the SURFXML parser (have look at \ref faq_flexml_bypassing).
821 - The deployment file became quite big, so I had to do what is in
822 the FAQ entry \ref faq_flexml_limit
823 - Each UNIX98 context has its own stack entry. As debugging this is
824 quite hairly, the default value is a bit overestimated so that
825 user don't get into trouble about this. You want to tune this
826 size to increse the number of processes. This is the
827 <tt>STACK_SIZE</tt> define in
828 <tt>src/xbt/xbt_context_sysv.c</tt>, which is 128kb by default.
829 Reduce this as much as you can, but be warned that if this value
830 is too low, you'll get a segfault. The token ring example, which
831 is quite simple, runs with 40kb stacks.
832 - You may tweak the logs to reduce the stack size further. When
833 logging something, we try to build the string to display in a
834 char array on the stack. The size of this array is constant (and
835 equal to XBT_LOG_BUFF_SIZE, defined in include/xbt/log/h). If the
836 string is too large to fit this buffer, we move to a dynamically
837 sized buffer. In which case, we have to traverse one time the log
838 event arguments to compute the size we need for the buffer,
839 malloc it, and traverse the argument list again to do the actual
841 The idea here is to move XBT_LOG_BUFF_SIZE to 1, forcing the logs
842 to use a dynamic array each time. This allows us to lower further
843 the stack size at the price of some performance loss...\n
844 This allowed me to run the reduce the stack size to ... 4k. Ie,
845 on my 1Gb laptop, I can run more than 250,000 processes!
847 \subsubsection faq_MIA_batch_scheduler Is there a native support for batch schedulers in SimGrid?
849 No, there is no native support for batch schedulers and none is
850 planned because this is a very specific need (and doing it in a
851 generic way is thus very hard). However some people have implemented
852 their own batch schedulers. Vincent Garonne wrote one during his PhD
853 and put his code in the contrib directory of our SVN so that other can
854 keep working on it. You may find inspiring ideas in it.
856 \subsubsection faq_MIA_checkpointing I need a checkpointing thing
858 Actually, it depends on whether you want to checkpoint the simulation, or to
859 simulate checkpoints.
861 The first one could help if your simulation is a long standing process you
862 want to keep running even on hardware issues. It could also help to
863 <i>rewind</i> the simulation by jumping sometimes on an old checkpoint to
864 cancel recent calculations.\n
865 Unfortunately, such thing will probably never exist in SG. One would have to
866 duplicate all data structures because doing a rewind at the simulator level
867 is very very hard (not talking about the malloc free operations that might
868 have been done in between). Instead, you may be interested in the Libckpt
869 library (http://www.cs.utk.edu/~plank/plank/www/libckpt.html). This is the
870 checkpointing solution used in the condor project, for example. It makes it
871 easy to create checkpoints (at the OS level, creating something like core
872 files), and rerunning them on need.
874 If you want to simulate checkpoints instead, it means that you want the
875 state of an executing task (in particular, the progress made towards
876 completion) to be saved somewhere. So if a host (and the task executing on
877 it) fails (cf. #MSG_HOST_FAILURE), then the task can be restarted
878 from the last checkpoint.\n
880 Actually, such a thing does not exists in SimGrid either, but it's just
881 because we don't think it is fundamental and it may be done in the user code
882 at relatively low cost. You could for example use a watcher that
883 periodically get the remaining amount of things to do (using
884 MSG_task_get_remaining_computation()), or fragment the task in smaller
887 \subsection faq_platform Platform building and Dynamic resources
889 \subsubsection faq_platform_example Where can I find SimGrid platform files?
891 There is several little examples in the archive, in the examples/msg
892 directory. From time to time, we are asked for other files, but we
893 don't have much at hand right now.
895 You should refer to the Platform Description Archive
896 (http://pda.gforge.inria.fr) project to see the other platform file we
897 have available, as well as the Simulacrum simulator, meant to generate
898 SimGrid platforms using all classical generation algorithms.
900 \subsubsection faq_platform_alnem How can I automatically map an existing platform?
902 We are working on a project called ALNeM (Application-Level Network
903 Mapper) which goal is to automatically discover the topology of an
904 existing network. Its output will be a platform description file
905 following the SimGrid syntax, so everybody will get the ability to map
906 their own lab network (and contribute them to the catalog project).
907 This tool is not ready yet, but it move quite fast forward. Just stay
910 \subsubsection faq_platform_synthetic Generating synthetic but realistic platforms
912 The third possibility to get a platform file (after manual or
913 automatic mapping of real platforms) is to generate synthetic
914 platforms. Getting a realistic result is not a trivial task, and
915 moreover, nobody is really able to define what "realistic" means when
916 speaking of topology files. You can find some more thoughts on this
918 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">slides</a>.
920 If you are looking for an actual tool, there we have a little tool to
921 annotate Tiers-generated topologies. This perl-script is in
922 <tt>tools/platform_generation/</tt> directory of the SVN. Dinda et Al.
923 released a very comparable tool, and called it GridG.
925 \subsubsection faq_SURF_dynamic Expressing dynamic resource availability in platform files
927 A nice feature of SimGrid is that it enables you to seamlessly have
928 resources whose availability change over time. When you build a
929 platform, you generally declare hosts like that:
932 <host id="host A" power="100.00"/>
935 If you want the availability of "host A" to change over time, the only
936 thing you have to do is change this definition like that:
939 <host id="host A" power="100.00" availability_file="trace_A.txt" state_file="trace_A_failure.txt"/>
942 For hosts, availability files are expressed in fraction of available
943 power. Let's have a look at what "trace_A.txt" may look like:
952 At time 0, our host will deliver 100 flop/s. At time 11.0, it will
953 deliver only 50 flop/s until time 20.0 where it will will start
954 delivering 90 flop/s. Last at time 21.0 (20.0 plus the periodicity
955 1.0), we'll be back to the beginning and it will deliver 100 flop/s.
957 Now let's look at the state file:
964 A negative value means "off" while a positive one means "on". At time
965 1.0, the host is on. At time 1.0, it is turned off and at time 2.0, it
966 is turned on again until time 12 (2.0 plus the periodicity 10.0). It
967 will be turned on again at time 13.0 until time 23.0, and so on.
969 Now, let's look how the same kind of thing can be done for network
970 links. A usual declaration looks like:
973 <link id="LinkA" bandwidth="10.0" latency="0.2"/>
976 You have at your disposal the following options: bandwidth_file,
977 latency_file and state_file. The only difference with hosts is that
978 bandwidth_file and latency_file do not express fraction of available
979 power but are expressed directly in bytes per seconds and seconds.
981 \subsubsection faq_platform_multipath How to express multipath routing in platform files?
983 It is unfortunately impossible to express the fact that there is more
984 than one routing path between two given hosts. Let's consider the
985 following platform file:
988 <route src="A" dst="B">
991 <route src="B" dst="C">
994 <route src="A" dst="C">
999 Although it is perfectly valid, it does not mean that data traveling
1000 from A to C can either go directly (using link 3) or through B (using
1001 links 1 and 2). It simply means that the routing on the graph is not
1002 trivial, and that data do not following the shortest path in number of
1003 hops on this graph. Another way to say it is that there is no implicit
1004 in these routing descriptions. The system will only use the routes you
1005 declare (such as <route src="A" dst="C"><link:ctn
1006 id="3"/></route>), without trying to build new routes by aggregating
1009 You are also free to declare platform where the routing is not
1010 symmetric. For example, add the following to the previous file:
1013 <route src="C" dst="A">
1019 This makes sure that data from C to A go through B where data from A
1020 to C go directly. Don't worry about realism of such settings since
1021 we've seen ways more weird situation in real settings (in fact, that's
1022 the realism of very regular platforms which is questionable, but
1023 that's another story).
1025 \subsubsection faq_flexml_bypassing Bypassing the XML parser with your own C functions
1027 So you want to bypass the XML files parser, uh? Maybe doing some parameter
1028 sweep experiments on your simulations or so? This is possible, and
1029 it's not even really difficult (well. Such a brutal idea could be
1030 harder to implement). Here is how it goes.
1032 For this, you have to first remember that the XML parsing in SimGrid is done
1033 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
1034 you want to bypass the parser, you need to provide some code mimicking what
1035 it does and replacing it in its interactions with the SURF code. So, let's
1036 have a look at these interactions.
1038 FleXML parser are close to classical SAX parsers. It means that a
1039 well-formed SimGrid platform XML file might result in the following
1042 - start "platform_description" with attribute version="2"
1043 - start "host" with attributes id="host1" power="1.0"
1045 - start "host" with attributes id="host2" power="2.0"
1047 - start "link" with ...
1049 - start "route" with ...
1050 - start "link:ctn" with ...
1053 - end "platform_description"
1055 The communication from the parser to the SURF code uses two means:
1056 Attributes get copied into some global variables, and a surf-provided
1057 function gets called by the parser for each event. For example, the event
1058 - start "host" with attributes id="host1" power="1.0"
1060 let the parser do something roughly equivalent to:
1062 strcpy(A_host_id,"host1");
1067 In SURF, we attach callbacks to the different events by initializing the
1068 pointer functions to some the right surf functions. Since there can be
1069 more than one callback attached to the same event (if more than one
1070 model is in use, for example), they are stored in a dynar. Example in
1071 workstation_ptask_L07.c:
1073 /* Adding callback functions */
1074 surf_parse_reset_parser();
1075 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
1076 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
1077 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
1078 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
1079 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
1080 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
1082 /* Parse the file */
1083 surf_parse_open(file);
1084 xbt_assert1((!surf_parse()), "Parse error in %s", file);
1088 So, to bypass the FleXML parser, you need to write your own version of the
1089 surf_parse function, which should do the following:
1090 - Fill the A_<tag>_<attribute> variables with the wanted values
1091 - Call the corresponding STag_<tag>_fun function to simulate tag start
1092 - Call the corresponding ETag_<tag>_fun function to simulate tag end
1093 - (do the same for the next set of values, and loop)
1095 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
1097 surf_parse = surf_parse_bypass_environment;
1098 MSG_create_environment(NULL);
1099 surf_parse = surf_parse_bypass_application;
1100 MSG_launch_application(NULL);
1103 A set of macros are provided at the end of
1104 include/surf/surfxml_parse.h to ease the writing of the bypass
1105 functions. An example of this trick is distributed in the file
1106 examples/msg/masterslave/masterslave_bypass.c
1108 \subsection faq_simgrid_configuration Changing SimGrid's behavior
1110 A number of options can be given at runtime to change the default
1111 SimGrid behavior. In particular, you can change the default cpu and
1114 \subsubsection faq_simgrid_configuration_gtnets Using GTNetS
1116 It is possible to use a packet-level network simulator
1117 instead of the default flow-based simulation. You may want to use such
1118 an approach if you have doubts about the validity of the default model
1119 or if you want to perform some validation experiments. At the moment,
1120 we support the GTNetS simulator (it is still rather experimental
1121 though, so leave us a message if you play with it).
1125 To enable GTNetS model inside SimGrid it is needed to patch the GTNetS simulator source code
1126 and build/install it from scratch
1129 - <b>Download and enter the recent downloaded GTNetS directory</b>
1132 svn checkout svn://scm.gforge.inria.fr/svn/simgrid/contrib/trunk/GTNetS/
1137 - <b>Use the following commands to unzip and patch GTNetS package to work within SimGrid.</b>
1140 unzip gtnets-current.zip
1141 tar zxvf gtnets-current-patch.tgz
1143 cat ../00*.patch | patch -p1
1146 - <b>OPTIONALLY</b> you can use a patch for itanium 64bit processor family.
1149 cat ../AMD64-FATAL-Removed-DUL_SIZE_DIFF-Added-fPIC-compillin.patch | patch -p1
1152 - <b>Compile GTNetS</b>
1154 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.
1158 ln -sf Makefile.linux Makefile
1164 - <b>NOTE</b> A lot of warnings are expected but the application should compile
1165 just fine. If the makefile insists in compiling some QT libraries
1166 please try a make clean before asking for help.
1169 - <b>To compile optimized version</b>
1176 - <b>Installing GTNetS</b>
1178 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.
1181 ln -sf /<absolute_path>/gtnets_current/libgtsim-debug.so /<userhome>/usr/lib/libgtnets.so
1182 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/<userhome>/usr/lib/libgtnets.so
1183 mkdir /<userhome>/usr/include/gtnets
1184 cp -fr SRC/*.h /<userhome>/usr/include/gtnets
1188 - <b>Enable GTNetS support in SimGrid</b>
1191 ./configure --with-gtnets=/<userhome>/usr
1194 - <b>Once you have followed all the instructions for compiling and
1195 installing successfully you can activate this feature at
1196 runntime with the following options:</b>
1199 cd simgrid/example/msg/
1205 - <b>Or try the GTNetS model dogbone example with</b>
1208 gtnets/gtnets gtnets/onelink-p.xml gtnets/onelink-d.xml --cfg=network_model:GTNets
1212 A long version of this <a href="http://gforge.inria.fr/docman/view.php/12/6283/GTNetS HowTo.html">HowTo</a> it is available
1215 More about GTNetS simulator at <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/index.html">GTNetS Website</a>
1219 The patches provided by us worked successfully with GTNetS found
1220 <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/software/gtnets-current.zip">here</a>,
1221 dated from 12th June 2008. Due to the discontinuing development of
1222 GTNetS it is impossible to precise a version number. We STRONGLY recommend you
1223 to download and install the GTNetS version found in SimGrid repository as explained above.
1228 \subsubsection faq_simgrid_configuration_alternate_network Using alternative flow models
1230 The default simgrid network model uses a max-min based approach as
1231 explained in the research report
1232 <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>.
1233 Other models have been proposed and implemented since then (see for example
1234 <a href="http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf">Accuracy Study and Improvement of Network Simulation in the SimGrid Framework</a>)
1235 and can be activated at runtime. For example:
1237 ./mycode platform.xml deployment.xml --cfg=workstation_model:compound --cfg=network_model:LV08 -cfg=cpu_model:Cas01
1240 Possible models for the network are currently "Constant", "CM02",
1241 "LegrandVelho", "GTNets", Reno", "Reno2", "Vegas". Others will
1242 probably be added in the future and many of the previous ones are
1243 experimental and are likely to disappear without notice...
1245 \section faq_troubleshooting Troubleshooting
1247 \subsection faq_trouble_lib_compil SimGrid compilation and installation problems
1249 \subsubsection faq_trouble_lib_config ./configure fails!
1251 We know only one reason for the configure to fail:
1253 - <b>You are using a broken build environment</b>\n
1254 If symptom is that configure complains about gcc not being able to build
1255 executables, you are probably missing the libc6-dev package. Damn Ubuntu.
1257 If you experience other kind of issue, please get in touch with us. We are
1258 always interested in improving our portability to new systems.
1260 \subsubsection faq_trouble_distcheck Dude! "make check" fails on my machine!
1262 Don't assume we never run this target, because we do. Check
1263 http://bob.loria.fr:8010 if you don't believe us.
1265 There is several reasons which may cause the make check to fail on your
1268 - <b>You are using a broken libc (probably concerning the contextes)</b>.\n
1269 The symptom is that the "make check" fails within the examples/msg directory.\n
1270 By default, SimGrid uses something called ucontexts. This is part of the
1271 libc, but it's quite undertested. For example, some (old) versions of the
1272 glibc on alpha do not implement these functions, but provide the stubs
1273 (which return ENOSYS: not implemented). It may fool our detection mechanism
1274 and leads to segfaults. There is not much we can do to fix the bug.
1275 A workaround is to compile with --with-context=pthread to avoid
1276 ucontext completely. You'll be a bit more limited in the number
1277 of simulated processes you can start concurrently, but 5000
1278 processes is still enough for most purposes, isn't it?\n
1279 This limitation is the reason why we insist on using this piece of ...
1280 software even if it's so troublesome.\n
1281 <b>=> use --with-pthread on AMD64 architecture that do not have an
1282 ultra-recent libc.</b>
1284 - <b>There is a bug in SimGrid we aren't aware of</b>.\n
1285 If none of the above apply, please drop us a mail on the mailing list so
1286 that we can check it out. Make sure to read \ref faq_bugrepport
1289 \subsection faq_trouble_compil User code compilation problems
1291 \subsubsection faq_trouble_err_logcat "gcc: _simgrid_this_log_category_does_not_exist__??? undeclared (first use in this function)"
1293 This is because you are using the log mecanism, but you didn't created
1294 any default category in this file. You should refer to \ref XBT_log
1295 for all the details, but you simply forgot to call one of
1296 XBT_LOG_NEW_DEFAULT_CATEGORY() or XBT_LOG_NEW_DEFAULT_SUBCATEGORY().
1298 \subsubsection faq_trouble_pthreadstatic "gcc: undefined reference to pthread_key_create"
1300 This indicates that one of the library SimGrid depends on (libpthread
1301 here) was missing on the linking command line. Dependencies of
1302 libsimgrid are expressed directly in the dynamic library, so it's
1303 quite impossible that you see this message when doing dynamic linking.
1305 If you compile your code statically (and if you use a pthread version
1306 of SimGrid -- see \ref faq_more_processes), you must absolutely
1307 specify <tt>-lpthread</tt> on the linker command line. As usual, this should
1308 come after <tt>-lsimgrid</tt> on this command line.
1310 \subsection faq_trouble_errors Runtime error messages
1312 \subsubsection faq_flexml_limit "surf_parse_lex: Assertion `next limit' failed."
1314 This is because your platform file is too big for the parser.
1316 Actually, the message comes directly from FleXML, the technology on top of
1317 which the parser is built. FleXML has the bad idea of fetching the whole
1318 document in memory before parsing it. And moreover, the memory buffer size
1319 must be determined at compilation time.
1321 We use a value which seems big enough for our need without bloating the
1322 simulators footprints. But of course your mileage may vary. In this case,
1323 just edit src/surf/surfxml.l modify the definition of
1324 FLEXML_BUFFERSTACKSIZE. E.g.
1327 #define FLEXML_BUFFERSTACKSIZE 1000000000
1330 Then recompile and everything should be fine, provided that your version of
1331 Flex is recent enough (>= 2.5.31). If not the compilation process should
1334 A while ago, we worked on FleXML to reduce a bit its memory consumption, but
1335 these issues remain. There is two things we should do:
1337 - use a dynamic buffer instead of a static one so that the only limit
1338 becomes your memory, not a stupid constant fixed at compilation time
1339 (maybe not so difficult).
1340 - change the parser so that it does not need to get the whole file in
1341 memory before parsing
1342 (seems quite difficult, but I'm a complete newbe wrt flex stuff).
1344 These are changes to FleXML itself, not SimGrid. But since we kinda hijacked
1345 the development of FleXML, I can grant you that any patches would be really
1346 welcome and quickly integrated.
1348 <b>Update:</b> A new version of FleXML (1.7) was released. Most of the work
1349 was done by William Dowling, who use it in his own work. The good point is
1350 that it now use a dynamic buffer, and that the memory usage was greatly
1351 improved. The downside is that William also changed some things internally,
1352 and it breaks the hack we devised to bypass the parser, as explained in
1353 \ref faq_flexml_bypassing. Indeed, this is not a classical usage of the
1354 parser, and Will didn't imagine that we may have used (and even documented)
1355 such a crude usage of FleXML. So, we now have to repair the bypassing
1356 functionality to use the lastest FleXML version and fix the memory usage in
1359 \subsubsection faq_trouble_gras_transport GRAS spits networking error messages
1361 Gras, on real platforms, naturally use regular sockets to communicate. They
1362 are deeply hidden in the gras abstraction, but when things go wrong, you may
1363 get some weird error messages. Here are some example, with the probable
1366 - <b>Transport endpoint is not connected</b>: several processes try to open
1367 a server socket on the same port number of the same machine. This is
1368 naturally bad and each process should pick its own port number for this.\n
1369 Maybe, you just have some processes remaining from a previous experiment
1371 Killing them may help, but again if you kill -KILL them, you'll have to
1372 wait for a while: they didn't close there sockets properly and the system
1373 needs a while to notice that this port is free again.
1375 - <b>Socket closed by remote side</b>: if the remote process is not
1376 supposed to close the socket at this point, it may be dead.
1378 - <b>Connection reset by peer</b>: I found this on Internet about this
1379 error. I think it's what's happening here, too:\n
1380 <i>This basically means that a network error occurred while the client was
1381 receiving data from the server. But what is really happening is that the
1382 server actually accepts the connection, processes the request, and sends
1383 a reply to the client. However, when the server closes the socket, the
1384 client believes that the connection has been terminated abnormally
1385 because the socket implementation sends a TCP reset segment telling the
1386 client to throw away the data and report an error.\n
1387 Sometimes, this problem is caused by not properly closing the
1388 input/output streams and the socket connection. Make sure you close the
1389 input/output streams and socket connection properly. If everything is
1390 closed properly, however, and the problem persists, you can work around
1391 it by adding a one-second sleep before closing the streams and the
1392 socket. This technique, however, is not reliable and may not work on all
1394 Since GRAS sockets are closed properly (repeat after me: there is no bug
1395 in GRAS), it is either that you are closing your sockets on server side
1396 before the client get a chance to read them (use gras_os_sleep() to delay
1397 the server), or the server died awfully before the client got the data.
1399 \subsubsection faq_trouble_errors_big_fat_warning I'm told that my XML files are too old.
1401 The format of the XML platform description files is sometimes
1402 improved. For example, we decided to change the units used in SimGrid
1403 from MBytes, MFlops and seconds to Bytes, Flops and seconds to ease
1404 people exchanging small messages. We also reworked the route
1405 descriptions to allow more compact descriptions.
1407 That is why the XML files are versionned using the 'version' attribute
1408 of the root tag. Currently, it should read:
1410 <platform version="2">
1413 If your files are too old, you can use the simgrid_update_xml.pl
1414 script which can be found in the tools directory of the archive.
1416 \subsection faq_trouble_valgrind Valgrind-related and other debugger issues
1418 If you don't, you really should use valgrind to debug your code, it's
1421 \subsubsection faq_trouble_vg_longjmp longjmp madness in valgrind
1423 This is when valgrind starts complaining about longjmp things, just like:
1425 \verbatim ==21434== Conditional jump or move depends on uninitialised value(s)
1426 ==21434== at 0x420DBE5: longjmp (longjmp.c:33)
1428 ==21434== Use of uninitialised value of size 4
1429 ==21434== at 0x420DC3A: __longjmp (__longjmp.S:48)
1432 This is the sign that you didn't used the exception mecanism well. Most
1433 probably, you have a <tt>return;</tt> somewhere within a <tt>TRY{}</tt>
1434 block. This is <b>evil</b>, and you must not do this. Did you read the section
1437 \subsubsection faq_trouble_vg_libc Valgrind spits tons of errors about backtraces!
1439 It may happen that valgrind, the memory debugger beloved by any decent C
1440 programmer, spits tons of warnings like the following :
1441 \verbatim ==8414== Conditional jump or move depends on uninitialised value(s)
1442 ==8414== at 0x400882D: (within /lib/ld-2.3.6.so)
1443 ==8414== by 0x414EDE9: (within /lib/tls/i686/cmov/libc-2.3.6.so)
1444 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
1445 ==8414== by 0x414F937: _dl_open (in /lib/tls/i686/cmov/libc-2.3.6.so)
1446 ==8414== by 0x4150F4C: (within /lib/tls/i686/cmov/libc-2.3.6.so)
1447 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
1448 ==8414== by 0x415102D: __libc_dlopen_mode (in /lib/tls/i686/cmov/libc-2.3.6.so)
1449 ==8414== by 0x412D6B9: backtrace (in /lib/tls/i686/cmov/libc-2.3.6.so)
1450 ==8414== by 0x8076446: xbt_dictelm_get_ext (dict_elm.c:714)
1451 ==8414== by 0x80764C1: xbt_dictelm_get (dict_elm.c:732)
1452 ==8414== by 0x8079010: xbt_cfg_register (config.c:208)
1453 ==8414== by 0x806821B: MSG_config (msg_config.c:42)
1456 This problem is somewhere in the libc when using the backtraces and there is
1457 very few things we can do ourselves to fix it. Instead, here is how to tell
1458 valgrind to ignore the error. Add the following to your ~/.valgrind.supp (or
1459 create this file on need). Make sure to change the obj line according to
1460 your personnal mileage (change 2.3.6 to the actual version you are using,
1461 which you can retrieve with a simple "ls /lib/ld*.so").
1464 name: Backtrace madness
1466 obj:/lib/ld-2.3.6.so
1471 fun:__libc_dlopen_mode
1474 Then, you have to specify valgrind to use this suppression file by passing
1475 the <tt>--suppressions=$HOME/.valgrind.supp</tt> option on the command line.
1476 You can also add the following to your ~/.bashrc so that it gets passed
1477 automatically. Actually, it passes a bit more options to valgrind, and this
1478 happen to be my personnal settings. Check the valgrind documentation for
1481 \verbatim export VALGRIND_OPTS="--leak-check=yes --leak-resolution=high --num-callers=40 --tool=memcheck --suppressions=$HOME/.valgrind.supp" \endverbatim
1483 \subsubsection faq_trouble_backtraces Truncated backtraces
1485 When debugging SimGrid, it's easier to pass the
1486 --disable-compiler-optimization flag to the configure if valgrind or
1487 gdb get fooled by the optimization done by the compiler. But you
1488 should remove these flag when everything works before going in
1489 production (before launching your 1252135 experiments), or everything
1490 will run only one half of the true SimGrid potential.
1492 \subsection faq_deadlock There is a deadlock in my code!!!
1494 Unfortunately, we cannot debug every code written in SimGrid. We
1495 furthermore believe that the framework provides ways enough
1496 information to debug such informations yourself. If the textual output
1497 is not enough, Make sure to check the \ref faq_visualization FAQ entry to see
1498 how to get a graphical one.
1500 Now, if you come up with a really simple example that deadlocks and
1501 you're absolutely convinced that it should not, you can ask on the
1502 list. Just be aware that you'll be severely punished if the mistake is
1503 on your side... We have plenty of FAQ entries to redact and new
1504 features to implement for the impenitents! ;)
1506 \subsection faq_surf_network_latency I get weird timings when I play with the latencies.
1508 OK, first of all, remember that units should be Bytes, Flops and
1509 Seconds. If you don't use such units, some SimGrid constants (e.g. the
1510 SG_TCP_CTE_GAMMA constant used in most network models) won't have the
1511 right unit and you'll end up with weird results.
1513 Here is what happens with a single transfer of size L on a link
1514 (bw,lat) when nothing else happens.
1517 0-----lat--------------------------------------------------t
1518 |-----|**** real_bw =min(bw,SG_TCP_CTE_GAMMA/(2*lat)) *****|
1521 In more complex situations, this min is the solution of a complex
1522 max-min linear system. Have a look
1523 <a href="http://lists.gforge.inria.fr/pipermail/simgrid-devel/2006-April/thread.html">here</a>
1524 and read the two threads "Bug in SURF?" and "Surf bug not
1525 fixed?". You'll have a few other examples of such computations. You
1526 can also read "A Network Model for Simulation of Grid Application" by
1527 Henri Casanova and Loris Marchal to have all the details. The fact
1528 that the real_bw is smaller than bw is easy to understand. The fact
1529 that real_bw is smaller than SG_TCP_CTE_GAMMA/(2*lat) is due to the
1530 window-based congestion mechanism of TCP. With TCP, you can't exploit
1531 your huge network capacity if you don't have a good round-trip-time
1532 because of the acks...
1534 Anyway, what you get is t=lat + L/min(bw,SG_TCP_CTE_GAMMA/(2*lat)).
1536 * if I you set (bw,lat)=(100 000 000, 0.00001), you get t = 1.00001 (you fully
1538 * if I you set (bw,lat)=(100 000 000, 0.0001), you get t = 1.0001 (you're on the
1540 * if I you set (bw,lat)=(100 000 000, 0.001), you get t = 10.001 (ouch!)
1542 This bound on the effective bandwidth of a flow is not the only thing
1543 that may make your result be unexpected. For example, two flows
1544 competing on a saturated link receive an amount of bandwidth inversely
1545 proportional to their round trip time.
1547 \subsection faq_bugrepport So I've found a bug in SimGrid. How to report it?
1549 We do our best to make sure to hammer away any bugs of SimGrid, but this is
1550 still an academic project so please be patient if/when you find bugs in it.
1551 If you do, the best solution is to drop an email either on the simgrid-user
1552 or the simgrid-devel mailing list and explain us about the issue. You can
1553 also decide to open a formal bug report using the
1554 <a href="https://gforge.inria.fr/tracker/?atid=165&group_id=12&func=browse">relevant
1555 interface</a>. You need to login on the server to get the ability to submit
1558 We will do our best to solve any problem repported, but you need to help us
1559 finding the issue. Just telling "it segfault" isn't enough. Telling "It
1560 segfaults when running the attached simulator" doesn't really help either.
1561 You may find the following article interesting to see how to repport
1562 informative bug repports:
1563 http://www.chiark.greenend.org.uk/~sgtatham/bugs.html (it is not SimGrid
1564 specific at all, but it's full of good advices).
1566 \author Arnaud Legrand (arnaud.legrand::imag.fr)
1567 \author Martin Quinson (martin.quinson::loria.fr)
1572 ******************************************************************
1573 * OLD CRUFT NOT USED ANYMORE *
1574 ******************************************************************
1577 \subsection faq_crosscompile Cross-compiling a Windows DLL of SimGrid from linux
1579 At the moment, we do not distribute Windows pre-compiled version of SimGrid
1580 because the support for this platform is still experimental. We know that
1581 some parts of the GRAS environment do not work, and we think that the others
1582 environments (MSG and SD) have good chances to work, but we didn't test
1583 ourselves. This section explains how we generate the SimGrid DLL so that you
1584 can build it for yourself. First of all, you need to have a version more
1585 recent than 3.1 (ie, a SVN version as time of writting).
1587 In order to cross-compile the package to windows from linux, you need to
1588 install mingw32 (minimalist gnu win32). On Debian, you can do so by
1589 installing the packages mingw32 (compiler), mingw32-binutils (linker and
1590 so), mingw32-runtime.
1592 You can use the VPATH support of configure to compile at the same time for
1593 linux and windows without dupplicating the source nor cleaning the tree
1594 between each. Just run bootstrap (if you use the SVN) to run the autotools.
1595 Then, create a linux and a win directories. Then, type:
1596 \verbatim cd linux; ../configure --srcdir=.. <usual configure flags>; make; cd ..
1597 cd win; ../configure --srcdir=.. --host=i586-mingw32msvc <flags>; make; cd ..
1599 The trick to VPATH builds is to call configure from another directory,
1600 passing it an extra --srcdir argument to tell it where all the sources are.
1601 It will understand you want to use VPATH. Then, the trick to cross-compile
1602 is simply to add a --host argument specifying the target you want to build
1603 for. The i586-mingw32msvc string is what you have to pass to use the mingw32
1604 environment as distributed in Debian.
1606 After that, you can run all make targets from both directories, and test
1607 easily that what you change for one arch does not break the other one.
1609 It is possible that this VPATH build thing breaks from time to time in the
1610 SVN since it's quite fragile, but it's granted to work in any released
1611 version. If you experience problems, drop us a mail.
1613 Another possible source of issue is that at the moment, building the
1614 examples request to use the gras_stub_generator tool, which is a compiled
1615 program, not a script. In cross-compilation, you need to cross-execute with
1616 wine for example, which is not really pleasant. We are working on this, but
1617 in the meanwhile, simply don't build the examples in cross-compilation
1618 (<tt>cd src</tt> before running make).
1620 Program (cross-)compiled with mingw32 do request an extra DLL at run-time to be
1621 usable. For example, if you want to test your build with wine, you should do
1622 the following to put this library where wine looks for DLLs.
1624 cp /usr/share/doc/mingw32-runtime/mingwm10.dll.gz ~/.wine/c/windows/system/
1625 gunzip ~/.wine/c/windows/system/mingwm10.dll.gz
1628 The DLL is built in src/.libs, and installed in the <i>prefix</i>/bin directory
1629 when you run make install.
1631 If you want to use it in a native project on windows, you need to use
1632 simgrid.dll and mingwm10.dll. For each DLL, you need to build .def file
1633 under linux (listing the defined symbols), and convert it into a .lib file
1634 under windows (specifying this in a way that windows compilers like). To
1635 generate the def files, run (under linux):
1636 \verbatim echo "LIBRARY libsimgrid-0.dll" > simgrid.def
1637 echo EXPORTS >> simgrid.def
1638 nm libsimgrid-0.dll | grep ' T _' | sed 's/.* T _//' >> simgrid.def
1639 nm libsimgrid-0.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> simgrid.def
1641 echo "LIBRARY mingwm10.dll" > mingwm10.def
1642 echo EXPORTS >> mingwm10.def
1643 nm mingwm10.dll | grep ' T _' | sed 's/.* T _//' >> mingwm10.def
1644 nm mingwm10.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> mingwm10.def
1647 To create the import .lib files, use the <tt>lib</tt> windows tool (from
1648 MSVC) the following way to produce simgrid.lib and mingwm10.lib
1649 \verbatim lib /def:simgrid.def
1650 lib /def:mingwm10.def
1653 If you happen to use Borland C Builder, the right command line is the
1654 following (note that you don't need any file.def to get this working).
1655 \verbatim implib simgrid.lib libsimgrid-0.dll
1656 implib mingwm10.lib mingwm10.dll
1659 Then, set the following parameters in Visual C++ 2005:
1660 Linker -> Input -> Additional dependencies = simgrid.lib mingwm10.lib
1662 Just in case you wonder how to generate a DLL from libtool in another
1663 project, we added -no-undefined to any lib*_la_LDFLAGS variables so that
1664 libtool accepts to generate a dynamic library under windows. Then, to make
1665 it true, we pass any dependencies (such as -lws2 under windows or -lpthread
1666 on need) on the linking line. Passing such deps is a good idea anyway so
1667 that they get noted in the library itself, avoiding the users to know about
1668 our dependencies and put them manually on their compilation line. Then we
1669 added the AC_LIBTOOL_WIN32_DLL macro just before AC_PROG_LIBTOOL in the
1670 configure.ac. It means that we exported any symbols which need to be.
1671 Nowadays, functions get automatically exported, so we don't need to load our
1672 header files with tons of __declspec(dllexport) cruft. We only need to do so
1673 for data, but there is no public data in SimGrid so we are good.