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=$HOME
138 If at some point, something fails, check the section "\ref
139 faq_trouble_compil". If it does not help, you can report this problem to the
140 list but, please, avoid sending a laconic mail like "There is a problem. Is it
141 okay?". Send the config.log file which is automatically generated by
142 configure. Try to capture both the standard output and the error output of the
143 <tt>make</tt> command with <tt>script</tt>. There is no way for us to help you
144 without the relevant bits of information.
146 Now, the following directory should have been created :
148 \li <tt>/home/joe/doc/simgrid/html/</tt>
149 \li <tt>/home/joe/lib/</tt>
150 \li <tt>/home/joe/include/</tt>
152 SimGrid is not a binary, it is a library. Both a static and a dynamic
153 version are available. Here is what you can find if you try a <tt>ls
156 \verbatim libsimgrid.a libsimgrid.la libsimgrid.so libsimgrid.so.0 libsimgrid.so.0.0.1
159 Thus, there is two ways to link your program with SimGrid:
160 \li Either you use the static version, e.g
161 \verbatim gcc libsimgrid.a -o MainProgram MainProgram.c
163 In this case, all the SimGrid functions are directly
164 included in <tt>MainProgram</tt> (hence a bigger binary).
165 \li Either you use the dynamic version (the preferred method)
166 \verbatim gcc -lsimgrid -o MainProgram MainProgram.c
168 In this case, the SimGrid functions are not included in
169 <tt>MainProgram</tt> and you need to set your environment
170 variable in such a way that <tt>libsimgrid.so</tt> will be
171 found at runtime. This can be done by adding the following
172 line in your .bashrc (if you use bash and if you have
173 installed the SimGrid libraries in your home directory):
174 \verbatim export LD_LIBRARY_PATH=$HOME/lib/:$LD_LIBRARY_PATH
177 \subsection faq_compiling_java Java bindings don't get compiled
179 The configure script detects automatically whether you have the
180 softwares needed to use the Java bindings or not. At the end of the
181 configure, you can see the configuration picked by the script, which
182 should look similar to
183 \verbatim Configuration of package simgrid' (version 3.3.4-svn) on
186 Compiler: gcc (version: )
188 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
192 Context backend: ucontext
199 In this example, Java backends won't be compiled.
201 On Debian-like systems (which includes ubuntu), you need the following
202 packages: sun-java6-jdk libgcj10-dev. If you cannot find the
203 libgcj10-dev, try another version, like libgcj9-dev (on Ubuntu before
204 9.10) or libgcj11-dev (not released yet, but certainly one day).
205 Please note that you need to activate the contrib and non-free
206 repositories in Debian, and the universe ones in Ubuntu. Java comes at
209 \subsection faq_compiling_snapshoot SimGrid development snapshots
211 We have very high standards on software quality, and we are reluctant releasing
212 a stable release as long as there is still some known bug in the code base. In
213 addition, we added quite an extensive test base, making sure that we correctly
214 test the most important parts of the tool.
216 As an unfortunate conclusion, there may be some time between the stable
217 releases. If you want to benefit from the most recent features we introduced,
218 but don't want to take the risk of an untested version from the SVN, then
219 development snapshots are done for you.
221 These are pre-releases of SimGrid that still fail some tests about features
222 that almost nobody use, or on platforms not being in our core target (which is
223 Linux, Mac, other Unixes and Windows, from the most important to the less
224 one). That means that using this development releases should be safe for most
227 These archives can be found on
228 <a href="http://www.loria.fr/~quinson/simgrid.html">this web page</a>. Once you
229 got the lastest archive, you can compile it just like any archive (see above).
231 \subsection faq_compiling_svn Compiling SimGrid from the SVN
233 The project development takes place in the SVN, where all changes are
234 committed when they happen. Then every once in a while, we make sure that the
235 code quality meets our standard and release an archive from the code in the
236 SVN. We afterward go back to the development in the SVN. So, if you need a
237 recently added feature and can afford some little problem with the stability
238 of the lastest features, you may want to use the SVN version instead of a
241 For that, you first need to get the "simgrid" module from
242 <a href="http://gforge.inria.fr/scm/?group_id=12">here</a>.
244 You won't find any <tt>configure</tt> and a few other things
245 (<tt>Makefile.in</tt>'s, documentation, ...) will be missing as well. The
246 reason for that is that all these files have to be regenerated using the
247 latest versions of <tt>autoconf</tt>, <tt>libtool</tt>, <tt>automake</tt>
248 (>1.9) and <tt>doxygen</tt> (>1.4). To generate the <tt>configure</tt> and
249 the <tt>Makefile.in</tt>'s, you just have to launch the <tt>bootstrap</tt>
250 command that resides in the top of the source tree. Then just follow the
251 instructions of Section \ref faq_compiling.
253 We insist on the fact that you really need the latest versions of
254 autoconf, automake and libtool. Doing this step on exotic architectures/systems
255 (i.e. anything different from a recent linux distribution) may be
256 ... uncertain. If you need to compile the SVN version on a machine where all these
257 dependencies are not met, the easiest is to do <tt>make dist</tt> in the SVN
258 directory of another machine where all dependencies are met. It will create an
259 archive you may deploy on other sites just as a regular stable release.
261 In summary, the following commands will checkout the SVN, regenerate the
262 configure script and friends, configure SimGrid and build it.
264 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk simgrid
267 ./configure --enable-maintainer-mode --prefix=<where to install SimGrid>
270 Then, if you want to install SimGrid on the current box, just do:
271 \verbatim make install \endverbatim
273 If you want to build an snapshot of the SVN to deploy it on another box (for
274 example because the other machine don't have the autotools), do:
275 \verbatim make dist \endverbatim
277 Moreover, you should never call the autotools manually since you must run
278 them in a specific order with specific arguments. Most of the times, the
279 makefiles will automatically call the tools for you. When it's not possible
280 (such as the first time you checkout the SVN), use the ./bootstrap command
281 to call them explicitly.
284 \subsection faq_setting_MSG Setting up your own MSG code
286 Do not build your simulator by modifying the SimGrid examples. Go
287 outside the SimGrid source tree and create your own working directory
288 (say <tt>/home/joe/SimGrid/MyFirstScheduler/</tt>).
290 Suppose your simulation has the following structure (remember it is
291 just an example to illustrate a possible way to compile everything;
292 feel free to organize it as you want).
294 \li <tt>sched.h</tt>: a description of the core of the
295 scheduler (i.e. which functions are can be used by the
296 agents). For example we could find the following functions
297 (master, forwarder, slave).
299 \li <tt>sched.c</tt>: a C file including <tt>sched.h</tt> and
300 implementing the core of the scheduler. Most of these
301 functions use the MSG functions defined in section \ref
304 \li <tt>masterslave.c</tt>: a C file with the main function, i.e.
305 the MSG initialization (MSG_global_init()), the platform
306 creation (e.g. with MSG_create_environment()), the
307 deployment phase (e.g. with MSG_function_register() and
308 MSG_launch_application()) and the call to
311 To compile such a program, we suggest to use the following
312 Makefile. It is a generic Makefile that we have used many times with
313 our students when we teach the C language.
317 masterslave: masterslave.o sched.o
319 INSTALL_PATH = $$HOME
321 PEDANTIC_PARANOID_FREAK = -O0 -Wshadow -Wcast-align \
322 -Waggregate-return -Wmissing-prototypes -Wmissing-declarations \
323 -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations \
324 -Wmissing-noreturn -Wredundant-decls -Wnested-externs \
325 -Wpointer-arith -Wwrite-strings -finline-functions
326 REASONABLY_CAREFUL_DUDE = -Wall
327 NO_PRAYER_FOR_THE_WICKED = -w -O2
328 WARNINGS = $(REASONABLY_CAREFUL_DUDE)
329 CFLAGS = -g $(WARNINGS)
331 INCLUDES = -I$(INSTALL_PATH)/include
332 DEFS = -L$(INSTALL_PATH)/lib/
333 LDADD = -lm -lsimgrid
337 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) $^ $(LIBS) $(LDADD) -o $@
340 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) -c -o $@ $<
343 rm -f $(BIN_FILES) *.o *~
349 The first two lines indicates what should be build when typing make
350 (<tt>masterslave</tt>) and of which files it is to be made of
351 (<tt>masterslave.o</tt> and <tt>sched.o</tt>). This makefile assumes
352 that you have set up correctly your <tt>LD_LIBRARY_PATH</tt> variable
353 (look, there is a <tt>LDADD = -lm -lsimgrid</tt>). If you prefer using
354 the static version, remove the <tt>-lsimgrid</tt> and add a
355 <tt>$(INSTALL_PATH)/lib/libsimgrid.a</tt> on the next line, right
356 after the <tt>LIBS = </tt>.
358 More generally, if you have never written a Makefile by yourself, type
359 in a terminal : <tt>info make</tt> and read the introduction. The
360 previous example should be enough for a first try but you may want to
361 perform some more complex compilations...
363 \subsection faq_setting_GRAS Setting up your own GRAS code
365 If you use the GRAS interface instead of the MSG one, then previous section
366 is not the better source of information. Instead, you should check the GRAS
367 tutorial in general, and the \ref GRAS_tut_tour_setup in particular.
370 \section faq_howto Feature related questions
372 \subsection faq_MIA "Could you please add (your favorite feature here) to SimGrid?"
374 Here is the deal. The whole SimGrid project (MSG, SURF, GRAS, ...) is
375 meant to be kept as simple and generic as possible. We cannot add
376 functions for everybody's needs when these functions can easily be
377 built from the ones already in the API. Most of the time, it is
378 possible and when it was not possible we always have upgraded the API
379 accordingly. When somebody asks us a question like "How to do that?
380 Is there a function in the API to simply do this?", we're always glad
381 to answer and help. However if we don't need this code for our own
382 need, there is no chance we're going to write it... it's your job! :)
383 The counterpart to our answers is that once you come up with a neat
384 implementation of this feature (task duplication, RPC, thread
385 synchronization, ...), you should send it to us and we will be glad to
386 add it to the distribution. Thus, other people will take advantage of
387 it (and we don't have to answer this question again and again ;).
389 You'll find in this section a few "Missing In Action" features. Many
390 people have asked about it and we have given hints on how to simply do
391 it with MSG. Feel free to contribute...
393 \subsection faq_MIA_MSG MSG features
395 \subsubsection faq_MIA_examples I want some more complex MSG examples!
397 Many people have come to ask me a more complex example and each time,
398 they have realized afterward that the basics were in the previous three
401 Of course they have often been needing more complex functions like
402 MSG_process_suspend(), MSG_process_resume() and
403 MSG_process_isSuspended() (to perform synchronization), or
404 MSG_task_Iprobe() and MSG_process_sleep() (to avoid blocking
405 receptions), or even MSG_process_create() (to design asynchronous
406 communications or computations). But the examples are sufficient to
409 We know. We should add some more examples, but not really some more
410 complex ones... We should add some examples that illustrate some other
411 functionalists (like how to simply encode asynchronous
412 communications, RPC, process migrations, thread synchronization, ...)
413 and we will do it when we will have a little bit more time. We have
414 tried to document the examples so that they are understandable. Tell
415 us if something is not clear and once again feel free to participate!
418 \subsubsection faq_MIA_taskdup Missing in action: MSG Task duplication/replication
420 There is no task duplication in MSG. When you create a task, you can
421 process it or send it somewhere else. As soon as a process has sent
422 this task, he doesn't have this task anymore. It's gone. The receiver
423 process has got the task. However, you could decide upon receiving to
424 create a "copy" of a task but you have to handle by yourself the
425 semantic associated to this "duplication".
427 As we already told, we prefer keeping the API as simple as
428 possible. This kind of feature is rather easy to implement by users
429 and the semantic you associate really depends on people. Having a
430 *generic* task duplication mechanism is not that trivial (in
431 particular because of the data field). That is why I would recommand
432 that you write it by yourself even if I can give you advice on how to
435 You have the following functions to get informations about a task:
436 MSG_task_get_name(), MSG_task_get_compute_duration(),
437 MSG_task_get_remaining_computation(), MSG_task_get_data_size(),
438 and MSG_task_get_data().
440 You could use a dictionary (#xbt_dict_t) of dynars (#xbt_dynar_t). If
441 you still don't see how to do it, please come back to us...
443 \subsubsection faq_MIA_asynchronous I want to do asynchronous communications in MSG
445 Up until now, there is no asynchronous communications in MSG. However,
446 you can create as many process as you want so you should be able to do
447 whatever you want... I've written a queue module to help implementing
448 some asynchronous communications at low cost (creating thousands of
449 process only to handle communications may be problematic in term of
450 performance at some point). I'll add it in the distribution asap.
452 \subsubsection faq_MIA_thread_synchronization I need to synchronize my MSG processes
454 You obviously cannot use pthread_mutexes of pthread_conds. The best
455 thing would be to propose similar structures. Unfortunately, we
456 haven't found time to do it yet. However you can try to play with
457 MSG_process_suspend() and MSG_process_resume(). You can even do some
458 synchronization with fake communications (using MSG_task_get(),
459 MSG_task_put() and MSG_task_Iprobe()).
461 \subsubsection faq_MIA_host_load Where is the get_host_load function hidden in MSG?
463 There is no such thing because its semantic wouldn't be really
464 clear. Of course, it is something about the amount of host throughput,
465 but there is as many definition of "host load" as people asking for
466 this function. First, you have to remember that resource availability
467 may vary over time, which make any load notion harder to define.
469 It may be instantaneous value or an average one. Moreover it may be only the
470 power of the computer, or may take the background load into account, or may
471 even take the currently running tasks into account. In some SURF models,
472 communications have an influence on computational power. Should it be taken
475 First of all, it's near to impossible to predict the load beforehands in the
476 simulator since it depends on too much parameters (background load
477 variation, bandwidth sharing algorithmic complexity) some of them even being
478 not known beforehands (other task starting at the same time). So, getting
479 this information is really hard (just like in real life). It's not just that
480 we want MSG to be as painful as real life. But as it is in some way
481 realistic, we face some of the same problems as we would face in real life.
483 How would you do it for real? The most common option is to use something
484 like NWS that performs active probes. The best solution is probably to do
485 the same within MSG, as in next code snippet. It is very close from what you
486 would have to do out of the simulator, and thus gives you information that
487 you could also get in real settings to not hinder the realism of your
491 double get_host_load() {
492 m_task_t task = MSG_task_create("test", 0.001, 0, NULL);
493 double date = MSG_get_clock();
495 MSG_task_execute(task);
496 date = MSG_get_clock() - date;
497 MSG_task_destroy(task);
502 Of course, it may not match your personal definition of "host load". In this
503 case, please detail what you mean on the mailing list, and we will extend
504 this FAQ section to fit your taste if possible.
506 \subsubsection faq_MIA_communication_time How can I get the *real* communication time?
508 Communications are synchronous and thus if you simply get the time
509 before and after a communication, you'll only get the transmission
510 time and the time spent to really communicate (it will also take into
511 account the time spent waiting for the other party to be
512 ready). However, getting the *real* communication time is not really
513 hard either. The following solution is a good starting point.
518 m_task_t task = MSG_task_create("Task", task_comp_size, task_comm_size,
519 calloc(1,sizeof(double)));
520 *((double*) task->data) = MSG_get_clock();
521 MSG_task_put(task, slaves[i % slaves_count], PORT_22);
522 INFO0("Send completed");
527 m_task_t task = NULL;
530 time1 = MSG_get_clock();
531 a = MSG_task_get(&(task), PORT_22);
532 time2 = MSG_get_clock();
533 if(time1<*((double *)task->data))
534 time1 = *((double *) task->data);
535 INFO1("Communication time : \"%f\" ", time2-time1);
537 MSG_task_destroy(task);
542 \subsection faq_MIA_SimDag SimDag related questions
544 \subsubsection faq_SG_comm Implementing communication delays between tasks.
546 A classic question of SimDag newcomers is about how to express a
547 communication delay between tasks. The thing is that in SimDag, both
548 computation and communication are seen as tasks. So, if you want to
549 model a data dependency between two DAG tasks t1 and t2, you have to
550 create 3 SD_tasks: t1, t2 and c and add dependencies in the following
554 SD_task_dependency_add(NULL, NULL, t1, c);
555 SD_task_dependency_add(NULL, NULL, c, t2);
558 This way task t2 cannot start before the termination of communication c
559 which in turn cannot start before t1 ends.
561 When creating task c, you have to associate an amount of data (in bytes)
562 corresponding to what has to be sent by t1 to t2.
564 Finally to schedule the communication task c, you have to build a list
565 comprising the workstations on which t1 and t2 are scheduled (w1 and w2
566 for example) and build a communication matrix that should look like
569 \subsubsection faq_SG_DAG How to implement a distributed dynamic scheduler of DAGs.
571 Distributed is somehow "contagious". If you start making distributed
572 decisions, there is no way to handle DAGs directly anymore (unless I
573 am missing something). You have to encode your DAGs in term of
574 communicating process to make the whole scheduling process
575 distributed. Here is an example of how you could do that. Assume T1
576 has to be done before T2.
579 int your_agent(int argc, char *argv[] {
581 T1 = MSG_task_create(...);
582 T2 = MSG_task_create(...);
586 if(cond) MSG_task_execute(T1);
588 if((MSG_task_get_remaining_computation(T1)=0.0) && (you_re_in_a_good_mood))
591 /* do something else */
597 If you decide that the distributed part is not that much important and that
598 DAG is really the level of abstraction you want to work with, then you should
599 give a try to \ref SD_API.
601 \subsection faq_MIA_generic Generic features
603 \subsubsection faq_more_processes Increasing the amount of simulated processes
605 Here are a few tricks you can apply if you want to increase the amount
606 of processes in your simulations.
608 - <b>A few thousands of simulated processes</b> (soft tricks)\n
609 SimGrid can use either pthreads library or the UNIX98 contextes. On
610 most systems, the number of pthreads is limited and then your
611 simulation may be limited for a stupid reason. This is especially
612 true with the current linux pthreads, and I cannot get more than
613 2000 simulated processes with pthreads on my box. The UNIX98
614 contexts allow me to raise the limit to 25,000 simulated processes
616 The <tt>--with-context</tt> option of the <tt>./configure</tt>
617 script allows you to choose between UNIX98 contextes
618 (<tt>--with-context=ucontext</tt>) and the pthread version
619 (<tt>--with-context=pthread</tt>). The default value is ucontext
620 when the script detect a working UNIX98 context implementation. On
621 Windows boxes, the provided value is discarded and an adapted
622 version is picked up.\n\n
623 We experienced some issues with contextes on some rare systems
624 (solaris 8 and lower or old alpha linuxes comes to mind). The main
625 problem is that the configure script detect the contextes as being
626 functional when it's not true. If you happen to use such a system,
627 switch manually to the pthread version, and provide us with a good
628 patch for the configure script so that it is done automatically ;)
630 - <b>Hundred thousands of simulated processes</b> (hard-core tricks)\n
631 As explained above, SimGrid can use UNIX98 contextes to represent
632 and handle the simulated processes. Thanks to this, the main
633 limitation to the number of simulated processes becomes the
634 available memory.\n\n
635 Here are some tricks I had to use in order to run a token ring
636 between 25,000 processes on my laptop (1Gb memory, 1.5Gb swap).\n
637 - First of all, make sure your code runs for a few hundreds
638 processes before trying to push the limit. Make sure it's
639 valgrind-clean, ie that valgrind does not report neither memory
640 error nor memory leaks. Indeed, numerous simulated processes
641 result in *fat* simulation hindering debugging.
642 - It was really boring to write 25,000 entries in the deployment
643 file, so I wrote a little script
644 <tt>examples/gras/mutual_exclusion/simple_token/make_deployment.pl</tt>, which you may
645 want to adapt to your case. You could also think about hijacking
646 the SURFXML parser (have look at \ref faq_flexml_bypassing).
647 - The deployment file became quite big, so I had to do what is in
648 the FAQ entry \ref faq_flexml_limit
649 - Each UNIX98 context has its own stack entry. As debugging this is
650 quite hairly, the default value is a bit overestimated so that
651 user don't get into trouble about this. You want to tune this
652 size to increse the number of processes. This is the
653 <tt>STACK_SIZE</tt> define in
654 <tt>src/xbt/xbt_context_sysv.c</tt>, which is 128kb by default.
655 Reduce this as much as you can, but be warned that if this value
656 is too low, you'll get a segfault. The token ring example, which
657 is quite simple, runs with 40kb stacks.
658 - You may tweak the logs to reduce the stack size further. When
659 logging something, we try to build the string to display in a
660 char array on the stack. The size of this array is constant (and
661 equal to XBT_LOG_BUFF_SIZE, defined in include/xbt/log/h). If the
662 string is too large to fit this buffer, we move to a dynamically
663 sized buffer. In which case, we have to traverse one time the log
664 event arguments to compute the size we need for the buffer,
665 malloc it, and traverse the argument list again to do the actual
667 The idea here is to move XBT_LOG_BUFF_SIZE to 1, forcing the logs
668 to use a dynamic array each time. This allows us to lower further
669 the stack size at the price of some performance loss...\n
670 This allowed me to run the reduce the stack size to ... 4k. Ie,
671 on my 1Gb laptop, I can run more than 250,000 processes!
673 \subsubsection faq_MIA_batch_scheduler Is there a native support for batch schedulers in SimGrid?
675 No, there is no native support for batch schedulers and none is
676 planned because this is a very specific need (and doing it in a
677 generic way is thus very hard). However some people have implemented
678 their own batch schedulers. Vincent Garonne wrote one during his PhD
679 and put his code in the contrib directory of our SVN so that other can
680 keep working on it. You may find inspiring ideas in it.
682 \subsubsection faq_MIA_checkpointing I need a checkpointing thing
684 Actually, it depends on whether you want to checkpoint the simulation, or to
685 simulate checkpoints.
687 The first one could help if your simulation is a long standing process you
688 want to keep running even on hardware issues. It could also help to
689 <i>rewind</i> the simulation by jumping sometimes on an old checkpoint to
690 cancel recent calculations.\n
691 Unfortunately, such thing will probably never exist in SG. One would have to
692 duplicate all data structures because doing a rewind at the simulator level
693 is very very hard (not talking about the malloc free operations that might
694 have been done in between). Instead, you may be interested in the Libckpt
695 library (http://www.cs.utk.edu/~plank/plank/www/libckpt.html). This is the
696 checkpointing solution used in the condor project, for example. It makes it
697 easy to create checkpoints (at the OS level, creating something like core
698 files), and rerunning them on need.
700 If you want to simulate checkpoints instead, it means that you want the
701 state of an executing task (in particular, the progress made towards
702 completion) to be saved somewhere. So if a host (and the task executing on
703 it) fails (cf. #MSG_HOST_FAILURE), then the task can be restarted
704 from the last checkpoint.\n
706 Actually, such a thing does not exists in SimGrid either, but it's just
707 because we don't think it is fundamental and it may be done in the user code
708 at relatively low cost. You could for example use a watcher that
709 periodically get the remaining amount of things to do (using
710 MSG_task_get_remaining_computation()), or fragment the task in smaller
713 \subsection faq_platform Platform building and Dynamic resources
715 \subsubsection faq_platform_example Where can I find SimGrid platform files?
717 There is several little examples in the archive, in the examples/msg
718 directory. From time to time, we are asked for other files, but we
719 don't have much at hand right now.
721 You should refer to the Platform Description Archive
722 (http://pda.gforge.inria.fr) project to see the other platform file we
723 have available, as well as the Simulacrum simulator, meant to generate
724 SimGrid platforms using all classical generation algorithms.
726 \subsubsection faq_platform_alnem How can I automatically map an existing platform?
728 We are working on a project called ALNeM (Application-Level Network
729 Mapper) which goal is to automatically discover the topology of an
730 existing network. Its output will be a platform description file
731 following the SimGrid syntax, so everybody will get the ability to map
732 their own lab network (and contribute them to the catalog project).
733 This tool is not ready yet, but it move quite fast forward. Just stay
736 \subsubsection faq_platform_synthetic Generating synthetic but realistic platforms
738 The third possibility to get a platform file (after manual or
739 automatic mapping of real platforms) is to generate synthetic
740 platforms. Getting a realistic result is not a trivial task, and
741 moreover, nobody is really able to define what "realistic" means when
742 speaking of topology files. You can find some more thoughts on this
744 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">slides</a>.
746 If you are looking for an actual tool, there we have a little tool to
747 annotate Tiers-generated topologies. This perl-script is in
748 <tt>tools/platform_generation/</tt> directory of the SVN. Dinda et Al.
749 released a very comparable tool, and called it GridG.
751 \subsubsection faq_SURF_dynamic Expressing dynamic resource availability in platform files
753 A nice feature of SimGrid is that it enables you to seamlessly have
754 resources whose availability change over time. When you build a
755 platform, you generally declare hosts like that:
758 <host id="host A" power="100.00"/>
761 If you want the availability of "host A" to change over time, the only
762 thing you have to do is change this definition like that:
765 <host id="host A" power="100.00" availability_file="trace_A.txt" state_file="trace_A_failure.txt"/>
768 For hosts, availability files are expressed in fraction of available
769 power. Let's have a look at what "trace_A.txt" may look like:
778 At time 0, our host will deliver 100 flop/s. At time 11.0, it will
779 deliver only 50 flop/s until time 20.0 where it will will start
780 delivering 90 flop/s. Last at time 21.0 (20.0 plus the periodicity
781 1.0), we'll be back to the beginning and it will deliver 100 flop/s.
783 Now let's look at the state file:
790 A negative value means "off" while a positive one means "on". At time
791 1.0, the host is on. At time 1.0, it is turned off and at time 2.0, it
792 is turned on again until time 12 (2.0 plus the periodicity 10.0). It
793 will be turned on again at time 13.0 until time 23.0, and so on.
795 Now, let's look how the same kind of thing can be done for network
796 links. A usual declaration looks like:
799 <link id="LinkA" bandwidth="10.0" latency="0.2"/>
802 You have at your disposal the following options: bandwidth_file,
803 latency_file and state_file. The only difference with hosts is that
804 bandwidth_file and latency_file do not express fraction of available
805 power but are expressed directly in bytes per seconds and seconds.
807 \subsubsection faq_platform_multipath How to express multipath routing in platform files?
809 It is unfortunately impossible to express the fact that there is more
810 than one routing path between two given hosts. Let's consider the
811 following platform file:
814 <route src="A" dst="B">
817 <route src="B" dst="C">
820 <route src="A" dst="C">
825 Although it is perfectly valid, it does not mean that data traveling
826 from A to C can either go directly (using link 3) or through B (using
827 links 1 and 2). It simply means that the routing on the graph is not
828 trivial, and that data do not following the shortest path in number of
829 hops on this graph. Another way to say it is that there is no implicit
830 in these routing descriptions. The system will only use the routes you
831 declare (such as <route src="A" dst="C"><link:ctn
832 id="3"/></route>), without trying to build new routes by aggregating
835 You are also free to declare platform where the routing is not
836 symmetric. For example, add the following to the previous file:
839 <route src="C" dst="A">
845 This makes sure that data from C to A go through B where data from A
846 to C go directly. Don't worry about realism of such settings since
847 we've seen ways more weird situation in real settings (in fact, that's
848 the realism of very regular platforms which is questionable, but
849 that's another story).
851 \subsubsection faq_flexml_bypassing Bypassing the XML parser with your own C functions
853 So you want to bypass the XML files parser, uh? Maybe doing some parameter
854 sweep experiments on your simulations or so? This is possible, and
855 it's not even really difficult (well. Such a brutal idea could be
856 harder to implement). Here is how it goes.
858 For this, you have to first remember that the XML parsing in SimGrid is done
859 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
860 you want to bypass the parser, you need to provide some code mimicking what
861 it does and replacing it in its interactions with the SURF code. So, let's
862 have a look at these interactions.
864 FleXML parser are close to classical SAX parsers. It means that a
865 well-formed SimGrid platform XML file might result in the following
868 - start "platform_description" with attribute version="2"
869 - start "host" with attributes id="host1" power="1.0"
871 - start "host" with attributes id="host2" power="2.0"
873 - start "link" with ...
875 - start "route" with ...
876 - start "link:ctn" with ...
879 - end "platform_description"
881 The communication from the parser to the SURF code uses two means:
882 Attributes get copied into some global variables, and a surf-provided
883 function gets called by the parser for each event. For example, the event
884 - start "host" with attributes id="host1" power="1.0"
886 let the parser do something roughly equivalent to:
888 strcpy(A_host_id,"host1");
893 In SURF, we attach callbacks to the different events by initializing the
894 pointer functions to some the right surf functions. Since there can be
895 more than one callback attached to the same event (if more than one
896 model is in use, for example), they are stored in a dynar. Example in
897 workstation_ptask_L07.c:
899 /* Adding callback functions */
900 surf_parse_reset_parser();
901 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
902 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
903 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
904 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
905 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
906 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
909 surf_parse_open(file);
910 xbt_assert1((!surf_parse()), "Parse error in %s", file);
914 So, to bypass the FleXML parser, you need to write your own version of the
915 surf_parse function, which should do the following:
916 - Fill the A_<tag>_<attribute> variables with the wanted values
917 - Call the corresponding STag_<tag>_fun function to simulate tag start
918 - Call the corresponding ETag_<tag>_fun function to simulate tag end
919 - (do the same for the next set of values, and loop)
921 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
923 surf_parse = surf_parse_bypass_environment;
924 MSG_create_environment(NULL);
925 surf_parse = surf_parse_bypass_application;
926 MSG_launch_application(NULL);
929 A set of macros are provided at the end of
930 include/surf/surfxml_parse.h to ease the writing of the bypass
931 functions. An example of this trick is distributed in the file
932 examples/msg/masterslave/masterslave_bypass.c
934 \subsection faq_simgrid_configuration Changing SimGrid's behavior
936 A number of options can be given at runtime to change the default
937 SimGrid behavior. In particular, you can change the default cpu and
940 \subsubsection faq_simgrid_configuration_gtnets Using GTNetS
942 It is possible to use a packet-level network simulator
943 instead of the default flow-based simulation. You may want to use such
944 an approach if you have doubts about the validity of the default model
945 or if you want to perform some validation experiments. At the moment,
946 we support the GTNetS simulator (it is still rather experimental
947 though, so leave us a message if you play with it).
951 To enable GTNetS model inside SimGrid it is needed to patch the GTNetS simulator source code
952 and build/install it from scratch
955 - <b>Download and enter the recent downloaded GTNetS directory</b>
958 svn checkout svn://scm.gforge.inria.fr/svn/simgrid/contrib/trunk/GTNetS/
963 - <b>Use the following commands to unzip and patch GTNetS package to work within SimGrid.</b>
966 unzip gtnets-current.zip
967 tar zxvf gtnets-current-patch.tgz
969 cat ../00*.patch | patch -p1
972 - <b>OPTIONALLY</b> you can use a patch for itanium 64bit processor family.
975 cat ../AMD64-FATAL-Removed-DUL_SIZE_DIFF-Added-fPIC-compillin.patch | patch -p1
978 - <b>Compile GTNetS</b>
980 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.
984 ln -sf Makefile.linux Makefile
990 - <b>NOTE</b> A lot of warnings are expected but the application should compile
991 just fine. If the makefile insists in compiling some QT libraries
992 please try a make clean before asking for help.
995 - <b>To compile optimized version</b>
1002 - <b>Installing GTNetS</b>
1004 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.
1007 ln -sf /<absolute_path>/gtnets_current/libgtsim-debug.so /<userhome>/usr/lib/libgtnets.so
1008 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/<userhome>/usr/lib/libgtnets.so
1009 mkdir /<userhome>/usr/include/gtnets
1010 cp -fr SRC/*.h /<userhome>/usr/include/gtnets
1014 - <b>Enable GTNetS support in SimGrid</b>
1017 ./configure --with-gtnets=/<userhome>/usr
1020 - <b>Once you have followed all the instructions for compiling and
1021 installing successfully you can activate this feature at
1022 runntime with the following options:</b>
1025 cd simgrid/example/msg/
1031 - <b>Or try the GTNetS model dogbone example with</b>
1034 gtnets/gtnets gtnets/onelink-p.xml gtnets/onelink-d.xml --cfg=network_model:GTNets
1038 A long version of this <a href="http://gforge.inria.fr/docman/view.php/12/6283/GTNetS HowTo.html">HowTo</a> it is available
1041 More about GTNetS simulator at <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/index.html">GTNetS Website</a>
1045 The patches provided by us worked successfully with GTNetS found
1046 <a href="http://www.ece.gatech.edu/research/labs/MANIACS/GTNetS/software/gtnets-current.zip">here</a>,
1047 dated from 12th June 2008. Due to the discontinuing development of
1048 GTNetS it is impossible to precise a version number. We STRONGLY recommend you
1049 to download and install the GTNetS version found in SimGrid repository as explained above.
1054 \subsubsection faq_simgrid_configuration_alternate_network Using alternative flow models
1056 The default simgrid network model uses a max-min based approach as
1057 explained in the research report
1058 <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>.
1059 Other models have been proposed and implemented since then (see for example
1060 <a href="http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf">Accuracy Study and Improvement of Network Simulation in the SimGrid Framework</a>)
1061 and can be activated at runtime. For example:
1063 ./mycode platform.xml deployment.xml --cfg=workstation_model:compound --cfg=network_model:LV08 -cfg=cpu_model:Cas01
1066 Possible models for the network are currently "Constant", "CM02",
1067 "LegrandVelho", "GTNets", Reno", "Reno2", "Vegas". Others will
1068 probably be added in the future and many of the previous ones are
1069 experimental and are likely to disappear without notice...
1071 \section faq_troubleshooting Troubleshooting
1073 \subsection faq_trouble_lib_compil SimGrid compilation and installation problems
1075 \subsubsection faq_trouble_lib_config ./configure fails!
1077 We know only one reason for the configure to fail:
1079 - <b>You are using a broken build environment</b>\n
1080 If symptom is that configure complains about gcc not being able to build
1081 executables, you are probably missing the libc6-dev package. Damn Ubuntu.
1083 If you experience other kind of issue, please get in touch with us. We are
1084 always interested in improving our portability to new systems.
1086 \subsubsection faq_trouble_distcheck Dude! "make check" fails on my machine!
1088 Don't assume we never run this target, because we do. Check
1089 http://bob.loria.fr:8010 if you don't believe us.
1091 There is several reasons which may cause the make check to fail on your
1094 - <b>You are using a broken libc (probably concerning the contextes)</b>.\n
1095 The symptom is that the "make check" fails within the examples/msg directory.\n
1096 By default, SimGrid uses something called ucontexts. This is part of the
1097 libc, but it's quite undertested. For example, some (old) versions of the
1098 glibc on alpha do not implement these functions, but provide the stubs
1099 (which return ENOSYS: not implemented). It may fool our detection mechanism
1100 and leads to segfaults. There is not much we can do to fix the bug.
1101 A workaround is to compile with --with-context=pthread to avoid
1102 ucontext completely. You'll be a bit more limited in the number
1103 of simulated processes you can start concurrently, but 5000
1104 processes is still enough for most purposes, isn't it?\n
1105 This limitation is the reason why we insist on using this piece of ...
1106 software even if it's so troublesome.\n
1107 <b>=> use --with-pthread on AMD64 architecture that do not have an
1108 ultra-recent libc.</b>
1110 - <b>There is a bug in SimGrid we aren't aware of</b>.\n
1111 If none of the above apply, please drop us a mail on the mailing list so
1112 that we can check it out. Make sure to read \ref faq_bugrepport
1115 \subsection faq_trouble_compil User code compilation problems
1117 \subsubsection faq_trouble_err_logcat "gcc: _simgrid_this_log_category_does_not_exist__??? undeclared (first use in this function)"
1119 This is because you are using the log mecanism, but you didn't created
1120 any default category in this file. You should refer to \ref XBT_log
1121 for all the details, but you simply forgot to call one of
1122 XBT_LOG_NEW_DEFAULT_CATEGORY() or XBT_LOG_NEW_DEFAULT_SUBCATEGORY().
1124 \subsubsection faq_trouble_pthreadstatic "gcc: undefined reference to pthread_key_create"
1126 This indicates that one of the library SimGrid depends on (libpthread
1127 here) was missing on the linking command line. Dependencies of
1128 libsimgrid are expressed directly in the dynamic library, so it's
1129 quite impossible that you see this message when doing dynamic linking.
1131 If you compile your code statically (and if you use a pthread version
1132 of SimGrid -- see \ref faq_more_processes), you must absolutely
1133 specify <tt>-lpthread</tt> on the linker command line. As usual, this should
1134 come after <tt>-lsimgrid</tt> on this command line.
1136 \subsection faq_trouble_errors Runtime error messages
1138 \subsubsection faq_flexml_limit "surf_parse_lex: Assertion `next limit' failed."
1140 This is because your platform file is too big for the parser.
1142 Actually, the message comes directly from FleXML, the technology on top of
1143 which the parser is built. FleXML has the bad idea of fetching the whole
1144 document in memory before parsing it. And moreover, the memory buffer size
1145 must be determined at compilation time.
1147 We use a value which seems big enough for our need without bloating the
1148 simulators footprints. But of course your mileage may vary. In this case,
1149 just edit src/surf/surfxml.l modify the definition of
1150 FLEXML_BUFFERSTACKSIZE. E.g.
1153 #define FLEXML_BUFFERSTACKSIZE 1000000000
1156 Then recompile and everything should be fine, provided that your version of
1157 Flex is recent enough (>= 2.5.31). If not the compilation process should
1160 A while ago, we worked on FleXML to reduce a bit its memory consumption, but
1161 these issues remain. There is two things we should do:
1163 - use a dynamic buffer instead of a static one so that the only limit
1164 becomes your memory, not a stupid constant fixed at compilation time
1165 (maybe not so difficult).
1166 - change the parser so that it does not need to get the whole file in
1167 memory before parsing
1168 (seems quite difficult, but I'm a complete newbe wrt flex stuff).
1170 These are changes to FleXML itself, not SimGrid. But since we kinda hijacked
1171 the development of FleXML, I can grant you that any patches would be really
1172 welcome and quickly integrated.
1174 <b>Update:</b> A new version of FleXML (1.7) was released. Most of the work
1175 was done by William Dowling, who use it in his own work. The good point is
1176 that it now use a dynamic buffer, and that the memory usage was greatly
1177 improved. The downside is that William also changed some things internally,
1178 and it breaks the hack we devised to bypass the parser, as explained in
1179 \ref faq_flexml_bypassing. Indeed, this is not a classical usage of the
1180 parser, and Will didn't imagine that we may have used (and even documented)
1181 such a crude usage of FleXML. So, we now have to repair the bypassing
1182 functionality to use the lastest FleXML version and fix the memory usage in
1185 \subsubsection faq_trouble_gras_transport GRAS spits networking error messages
1187 Gras, on real platforms, naturally use regular sockets to communicate. They
1188 are deeply hidden in the gras abstraction, but when things go wrong, you may
1189 get some weird error messages. Here are some example, with the probable
1192 - <b>Transport endpoint is not connected</b>: several processes try to open
1193 a server socket on the same port number of the same machine. This is
1194 naturally bad and each process should pick its own port number for this.\n
1195 Maybe, you just have some processes remaining from a previous experiment
1197 Killing them may help, but again if you kill -KILL them, you'll have to
1198 wait for a while: they didn't close there sockets properly and the system
1199 needs a while to notice that this port is free again.
1201 - <b>Socket closed by remote side</b>: if the remote process is not
1202 supposed to close the socket at this point, it may be dead.
1204 - <b>Connection reset by peer</b>: I found this on Internet about this
1205 error. I think it's what's happening here, too:\n
1206 <i>This basically means that a network error occurred while the client was
1207 receiving data from the server. But what is really happening is that the
1208 server actually accepts the connection, processes the request, and sends
1209 a reply to the client. However, when the server closes the socket, the
1210 client believes that the connection has been terminated abnormally
1211 because the socket implementation sends a TCP reset segment telling the
1212 client to throw away the data and report an error.\n
1213 Sometimes, this problem is caused by not properly closing the
1214 input/output streams and the socket connection. Make sure you close the
1215 input/output streams and socket connection properly. If everything is
1216 closed properly, however, and the problem persists, you can work around
1217 it by adding a one-second sleep before closing the streams and the
1218 socket. This technique, however, is not reliable and may not work on all
1220 Since GRAS sockets are closed properly (repeat after me: there is no bug
1221 in GRAS), it is either that you are closing your sockets on server side
1222 before the client get a chance to read them (use gras_os_sleep() to delay
1223 the server), or the server died awfully before the client got the data.
1225 \subsubsection faq_trouble_errors_big_fat_warning I'm told that my XML files are too old.
1227 The format of the XML platform description files is sometimes
1228 improved. For example, we decided to change the units used in SimGrid
1229 from MBytes, MFlops and seconds to Bytes, Flops and seconds to ease
1230 people exchanging small messages. We also reworked the route
1231 descriptions to allow more compact descriptions.
1233 That is why the XML files are versionned using the 'version' attribute
1234 of the root tag. Currently, it should read:
1236 <platform version="2">
1239 If your files are too old, you can use the simgrid_update_xml.pl
1240 script which can be found in the tools directory of the archive.
1242 \subsection faq_trouble_valgrind Valgrind-related and other debugger issues
1244 If you don't, you really should use valgrind to debug your code, it's
1247 \subsubsection faq_trouble_vg_longjmp longjmp madness in valgrind
1249 This is when valgrind starts complaining about longjmp things, just like:
1251 \verbatim ==21434== Conditional jump or move depends on uninitialised value(s)
1252 ==21434== at 0x420DBE5: longjmp (longjmp.c:33)
1254 ==21434== Use of uninitialised value of size 4
1255 ==21434== at 0x420DC3A: __longjmp (__longjmp.S:48)
1258 This is the sign that you didn't used the exception mecanism well. Most
1259 probably, you have a <tt>return;</tt> somewhere within a <tt>TRY{}</tt>
1260 block. This is <b>evil</b>, and you must not do this. Did you read the section
1263 \subsubsection faq_trouble_vg_libc Valgrind spits tons of errors about backtraces!
1265 It may happen that valgrind, the memory debugger beloved by any decent C
1266 programmer, spits tons of warnings like the following :
1267 \verbatim ==8414== Conditional jump or move depends on uninitialised value(s)
1268 ==8414== at 0x400882D: (within /lib/ld-2.3.6.so)
1269 ==8414== by 0x414EDE9: (within /lib/tls/i686/cmov/libc-2.3.6.so)
1270 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
1271 ==8414== by 0x414F937: _dl_open (in /lib/tls/i686/cmov/libc-2.3.6.so)
1272 ==8414== by 0x4150F4C: (within /lib/tls/i686/cmov/libc-2.3.6.so)
1273 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
1274 ==8414== by 0x415102D: __libc_dlopen_mode (in /lib/tls/i686/cmov/libc-2.3.6.so)
1275 ==8414== by 0x412D6B9: backtrace (in /lib/tls/i686/cmov/libc-2.3.6.so)
1276 ==8414== by 0x8076446: xbt_dictelm_get_ext (dict_elm.c:714)
1277 ==8414== by 0x80764C1: xbt_dictelm_get (dict_elm.c:732)
1278 ==8414== by 0x8079010: xbt_cfg_register (config.c:208)
1279 ==8414== by 0x806821B: MSG_config (msg_config.c:42)
1282 This problem is somewhere in the libc when using the backtraces and there is
1283 very few things we can do ourselves to fix it. Instead, here is how to tell
1284 valgrind to ignore the error. Add the following to your ~/.valgrind.supp (or
1285 create this file on need). Make sure to change the obj line according to
1286 your personnal mileage (change 2.3.6 to the actual version you are using,
1287 which you can retrieve with a simple "ls /lib/ld*.so").
1290 name: Backtrace madness
1292 obj:/lib/ld-2.3.6.so
1297 fun:__libc_dlopen_mode
1300 Then, you have to specify valgrind to use this suppression file by passing
1301 the <tt>--suppressions=$HOME/.valgrind.supp</tt> option on the command line.
1302 You can also add the following to your ~/.bashrc so that it gets passed
1303 automatically. Actually, it passes a bit more options to valgrind, and this
1304 happen to be my personnal settings. Check the valgrind documentation for
1307 \verbatim export VALGRIND_OPTS="--leak-check=yes --leak-resolution=high --num-callers=40 --tool=memcheck --suppressions=$HOME/.valgrind.supp" \endverbatim
1309 \subsubsection faq_trouble_backtraces Truncated backtraces
1311 When debugging SimGrid, it's easier to pass the
1312 --disable-compiler-optimization flag to the configure if valgrind or
1313 gdb get fooled by the optimization done by the compiler. But you
1314 should remove these flag when everything works before going in
1315 production (before launching your 1252135 experiments), or everything
1316 will run only one half of the true SimGrid potential.
1318 \subsection faq_deadlock There is a deadlock in my code!!!
1320 Unfortunately, we cannot debug every code written in SimGrid. We
1321 furthermore believe that the framework provides ways enough
1322 information to debug such informations yourself. If the textual output
1323 is not enough, Make sure to check the \ref faq_visualization FAQ entry to see
1324 how to get a graphical one.
1326 Now, if you come up with a really simple example that deadlocks and
1327 you're absolutely convinced that it should not, you can ask on the
1328 list. Just be aware that you'll be severely punished if the mistake is
1329 on your side... We have plenty of FAQ entries to redact and new
1330 features to implement for the impenitents! ;)
1332 \subsection faq_surf_network_latency I get weird timings when I play with the latencies.
1334 OK, first of all, remember that units should be Bytes, Flops and
1335 Seconds. If you don't use such units, some SimGrid constants (e.g. the
1336 SG_TCP_CTE_GAMMA constant used in most network models) won't have the
1337 right unit and you'll end up with weird results.
1339 Here is what happens with a single transfer of size L on a link
1340 (bw,lat) when nothing else happens.
1343 0-----lat--------------------------------------------------t
1344 |-----|**** real_bw =min(bw,SG_TCP_CTE_GAMMA/(2*lat)) *****|
1347 In more complex situations, this min is the solution of a complex
1348 max-min linear system. Have a look
1349 <a href="http://lists.gforge.inria.fr/pipermail/simgrid-devel/2006-April/thread.html">here</a>
1350 and read the two threads "Bug in SURF?" and "Surf bug not
1351 fixed?". You'll have a few other examples of such computations. You
1352 can also read "A Network Model for Simulation of Grid Application" by
1353 Henri Casanova and Loris Marchal to have all the details. The fact
1354 that the real_bw is smaller than bw is easy to understand. The fact
1355 that real_bw is smaller than SG_TCP_CTE_GAMMA/(2*lat) is due to the
1356 window-based congestion mechanism of TCP. With TCP, you can't exploit
1357 your huge network capacity if you don't have a good round-trip-time
1358 because of the acks...
1360 Anyway, what you get is t=lat + L/min(bw,SG_TCP_CTE_GAMMA/(2*lat)).
1362 * if I you set (bw,lat)=(100 000 000, 0.00001), you get t = 1.00001 (you fully
1364 * if I you set (bw,lat)=(100 000 000, 0.0001), you get t = 1.0001 (you're on the
1366 * if I you set (bw,lat)=(100 000 000, 0.001), you get t = 10.001 (ouch!)
1368 This bound on the effective bandwidth of a flow is not the only thing
1369 that may make your result be unexpected. For example, two flows
1370 competing on a saturated link receive an amount of bandwidth inversely
1371 proportional to their round trip time.
1373 \subsection faq_bugrepport So I've found a bug in SimGrid. How to report it?
1375 We do our best to make sure to hammer away any bugs of SimGrid, but this is
1376 still an academic project so please be patient if/when you find bugs in it.
1377 If you do, the best solution is to drop an email either on the simgrid-user
1378 or the simgrid-devel mailing list and explain us about the issue. You can
1379 also decide to open a formal bug report using the
1380 <a href="https://gforge.inria.fr/tracker/?atid=165&group_id=12&func=browse">relevant
1381 interface</a>. You need to login on the server to get the ability to submit
1384 We will do our best to solve any problem repported, but you need to help us
1385 finding the issue. Just telling "it segfault" isn't enough. Telling "It
1386 segfaults when running the attached simulator" doesn't really help either.
1387 You may find the following article interesting to see how to repport
1388 informative bug repports:
1389 http://www.chiark.greenend.org.uk/~sgtatham/bugs.html (it is not SimGrid
1390 specific at all, but it's full of good advices).
1392 \author Arnaud Legrand (arnaud.legrand::imag.fr)
1393 \author Martin Quinson (martin.quinson::loria.fr)
1398 ******************************************************************
1399 * OLD CRUFT NOT USED ANYMORE *
1400 ******************************************************************
1403 \subsection faq_crosscompile Cross-compiling a Windows DLL of SimGrid from linux
1405 At the moment, we do not distribute Windows pre-compiled version of SimGrid
1406 because the support for this platform is still experimental. We know that
1407 some parts of the GRAS environment do not work, and we think that the others
1408 environments (MSG and SD) have good chances to work, but we didn't test
1409 ourselves. This section explains how we generate the SimGrid DLL so that you
1410 can build it for yourself. First of all, you need to have a version more
1411 recent than 3.1 (ie, a SVN version as time of writting).
1413 In order to cross-compile the package to windows from linux, you need to
1414 install mingw32 (minimalist gnu win32). On Debian, you can do so by
1415 installing the packages mingw32 (compiler), mingw32-binutils (linker and
1416 so), mingw32-runtime.
1418 You can use the VPATH support of configure to compile at the same time for
1419 linux and windows without dupplicating the source nor cleaning the tree
1420 between each. Just run bootstrap (if you use the SVN) to run the autotools.
1421 Then, create a linux and a win directories. Then, type:
1422 \verbatim cd linux; ../configure --srcdir=.. <usual configure flags>; make; cd ..
1423 cd win; ../configure --srcdir=.. --host=i586-mingw32msvc <flags>; make; cd ..
1425 The trick to VPATH builds is to call configure from another directory,
1426 passing it an extra --srcdir argument to tell it where all the sources are.
1427 It will understand you want to use VPATH. Then, the trick to cross-compile
1428 is simply to add a --host argument specifying the target you want to build
1429 for. The i586-mingw32msvc string is what you have to pass to use the mingw32
1430 environment as distributed in Debian.
1432 After that, you can run all make targets from both directories, and test
1433 easily that what you change for one arch does not break the other one.
1435 It is possible that this VPATH build thing breaks from time to time in the
1436 SVN since it's quite fragile, but it's granted to work in any released
1437 version. If you experience problems, drop us a mail.
1439 Another possible source of issue is that at the moment, building the
1440 examples request to use the gras_stub_generator tool, which is a compiled
1441 program, not a script. In cross-compilation, you need to cross-execute with
1442 wine for example, which is not really pleasant. We are working on this, but
1443 in the meanwhile, simply don't build the examples in cross-compilation
1444 (<tt>cd src</tt> before running make).
1446 Program (cross-)compiled with mingw32 do request an extra DLL at run-time to be
1447 usable. For example, if you want to test your build with wine, you should do
1448 the following to put this library where wine looks for DLLs.
1450 cp /usr/share/doc/mingw32-runtime/mingwm10.dll.gz ~/.wine/c/windows/system/
1451 gunzip ~/.wine/c/windows/system/mingwm10.dll.gz
1454 The DLL is built in src/.libs, and installed in the <i>prefix</i>/bin directory
1455 when you run make install.
1457 If you want to use it in a native project on windows, you need to use
1458 simgrid.dll and mingwm10.dll. For each DLL, you need to build .def file
1459 under linux (listing the defined symbols), and convert it into a .lib file
1460 under windows (specifying this in a way that windows compilers like). To
1461 generate the def files, run (under linux):
1462 \verbatim echo "LIBRARY libsimgrid-0.dll" > simgrid.def
1463 echo EXPORTS >> simgrid.def
1464 nm libsimgrid-0.dll | grep ' T _' | sed 's/.* T _//' >> simgrid.def
1465 nm libsimgrid-0.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> simgrid.def
1467 echo "LIBRARY mingwm10.dll" > mingwm10.def
1468 echo EXPORTS >> mingwm10.def
1469 nm mingwm10.dll | grep ' T _' | sed 's/.* T _//' >> mingwm10.def
1470 nm mingwm10.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> mingwm10.def
1473 To create the import .lib files, use the <tt>lib</tt> windows tool (from
1474 MSVC) the following way to produce simgrid.lib and mingwm10.lib
1475 \verbatim lib /def:simgrid.def
1476 lib /def:mingwm10.def
1479 If you happen to use Borland C Builder, the right command line is the
1480 following (note that you don't need any file.def to get this working).
1481 \verbatim implib simgrid.lib libsimgrid-0.dll
1482 implib mingwm10.lib mingwm10.dll
1485 Then, set the following parameters in Visual C++ 2005:
1486 Linker -> Input -> Additional dependencies = simgrid.lib mingwm10.lib
1488 Just in case you wonder how to generate a DLL from libtool in another
1489 project, we added -no-undefined to any lib*_la_LDFLAGS variables so that
1490 libtool accepts to generate a dynamic library under windows. Then, to make
1491 it true, we pass any dependencies (such as -lws2 under windows or -lpthread
1492 on need) on the linking line. Passing such deps is a good idea anyway so
1493 that they get noted in the library itself, avoiding the users to know about
1494 our dependencies and put them manually on their compilation line. Then we
1495 added the AC_LIBTOOL_WIN32_DLL macro just before AC_PROG_LIBTOOL in the
1496 configure.ac. It means that we exported any symbols which need to be.
1497 Nowadays, functions get automatically exported, so we don't need to load our
1498 header files with tons of __declspec(dllexport) cruft. We only need to do so
1499 for data, but there is no public data in SimGrid so we are good.