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 lastest framework.
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 limitated 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 begining 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 Vizualization 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 informations that it is hard to find the information
89 As a matter of fact, loging 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 informations 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
132 \verbatim./configure --prefix=$HOME
137 If at some point, something fails, check the section "\ref
138 faq_trouble_compil". If it does not help, you can report this problem to the
139 list but, please, avoid sending a laconic mail like "There is a problem. Is it
140 okay?". Send the config.log file which is automatically generated by
141 configure. Try to capture both the standard output and the error output of the
142 <tt>make</tt> command with <tt>script</tt>. There is no way for us to help you
143 without the relevant bits of information.
145 Now, the following directory should have been created :
147 \li <tt>/home/joe/doc/simgrid/html/</tt>
148 \li <tt>/home/joe/lib/</tt>
149 \li <tt>/home/joe/include/</tt>
151 SimGrid is not a binary, it is a library. Both a static and a dynamic
152 version are available. Here is what you can find if you try a <tt>ls
155 \verbatim libsimgrid.a libsimgrid.la libsimgrid.so libsimgrid.so.0 libsimgrid.so.0.0.1
158 Thus, there is two ways to link your program with SimGrid:
159 \li Either you use the static version, e.g
160 \verbatim gcc libsimgrid.a -o MainProgram MainProgram.c
162 In this case, all the SimGrid functions are directly
163 included in <tt>MainProgram</tt> (hence a bigger binary).
164 \li Either you use the dynamic version (the preferred method)
165 \verbatim gcc -lsimgrid -o MainProgram MainProgram.c
167 In this case, the SimGrid functions are not included in
168 <tt>MainProgram</tt> and you need to set your environment
169 variable in such a way that <tt>libsimgrid.so</tt> will be
170 found at runtime. This can be done by adding the following
171 line in your .bashrc (if you use bash and if you have
172 installed the SimGrid libraries in your home directory):
173 \verbatim export LD_LIBRARY_PATH=$HOME/lib/:$LD_LIBRARY_PATH
176 \subsection faq_compiling_snapshoot SimGrid development snapshots
178 We have very high standards on software quality, and we are reluctant releasing
179 a stable release as long as there is still some known bug in the code base. In
180 addition, we added quite an extensive test base, making sure that we correctly
181 test the most important parts of the tool.
183 As an infortunate conclusion, there may be some time between the stable
184 releases. If you want to benefit from the most recent features we introduced,
185 but don't want to take the risk of an untested version from the SVN, then
186 development snapshots are done for you.
188 These are pre-releases of SimGrid that still fail some tests about features
189 that almost nobody use, or on platforms not being in our core target (which is
190 linux, mac, other unixes and windows, from the most important to the less
191 one). That means that using this development releases should be safe for most
194 These archives can be found on
195 <a href="http://www.loria.fr/~quinson/simgrid.html">this web page</a>. Once you
196 got the lastest archive, you can compile it just like any archive (see above).
198 \subsection faq_compiling_svn Compiling SimGrid from the SVN
200 The project development takes place in the svn, where all changes are
201 commited when they happen. Then every once in a while, we make sure that the
202 code quality meets our standard and release an archive from the code in the
203 SVN. We afterward go back to the development in the SVN. So, if you need a
204 recently added feature and can afford some little problem with the stability
205 of the lastest features, you may want to use the SVN version instead of a
208 For that, you first need to get the "simgrid" module from
209 <a href="http://gforge.inria.fr/scm/?group_id=12">here</a>.
211 You won't find any <tt>configure</tt> and a few other things
212 (<tt>Makefile.in</tt>'s, documentation, ...) will be missing as well. The
213 reason for that is that all these files have to be regenerated using the
214 latest versions of <tt>autoconf</tt>, <tt>libtool</tt>, <tt>automake</tt>
215 (>1.9) and <tt>doxygen</tt> (>1.4). To generate the <tt>configure</tt> and
216 the <tt>Makefile.in</tt>'s, you just have to launch the <tt>bootstrap</tt>
217 command that resides in the top of the source tree. Then just follow the
218 instructions of Section \ref faq_compiling.
220 We insist on the fact that you really need the latest versions of
221 autoconf, automake and libtool. Doing this step on exotic architectures/systems
222 (i.e. anything different from a recent linux distribution) may be
223 ... uncertain. If you need to compile the SVN version on a machine where all these
224 dependencies are not met, the easiest is to do <tt>make dist</tt> in the SVN
225 dir of another machine where all dependencies are met. It will create an
226 archive you may deploy on other sites just as a regular stable release.
228 In summary, the following commands will checkout the SVN, regenerate the
229 configure script and friends, configure SimGrid and build it.
231 \verbatim svn checkout svn://scm.gforge.inria.fr/svn/simgrid/simgrid/trunk simgrid
234 ./configure --enable-maintainer-mode --prefix=<where to install SimGrid>
237 Then, if you want to install SimGrid on the current box, just do:
238 \verbatim make install \endverbatim
240 If you want to build an snapshot of the SVN to deploy it on another box (for
241 example because the other machine don't have the autotools), do:
242 \verbatim make dist \endverbatim
244 Moreover, you should never call the autotools manually since you must run
245 them in a specific order with specific arguments. Most of the times, the
246 makefiles will automatically call the tools for you. When it's not possible
247 (such as the first time you checkout the SVN), use the ./bootstrap command
248 to call them explicitely.
251 \subsection faq_setting_MSG Setting up your own MSG code
253 Do not build your simulator by modifying the SimGrid examples. Go
254 outside the SimGrid source tree and create your own working directory
255 (say <tt>/home/joe/SimGrid/MyFirstScheduler/</tt>).
257 Suppose your simulation has the following structure (remember it is
258 just an example to illustrate a possible way to compile everything;
259 feel free to organize it as you want).
261 \li <tt>sched.h</tt>: a description of the core of the
262 scheduler (i.e. which functions are can be used by the
263 agents). For example we could find the following functions
264 (master, forwarder, slave).
266 \li <tt>sched.c</tt>: a C file including <tt>sched.h</tt> and
267 implementing the core of the scheduler. Most of these
268 functions use the MSG functions defined in section \ref
271 \li <tt>masterslave.c</tt>: a C file with the main function, i.e.
272 the MSG initialization (MSG_global_init()), the platform
273 creation (e.g. with MSG_create_environment()), the
274 deployment phase (e.g. with MSG_function_register() and
275 MSG_launch_application()) and the call to
278 To compile such a program, we suggest to use the following
279 Makefile. It is a generic Makefile that we have used many times with
280 our students when we teach the C language.
284 masterslave: masterslave.o sched.o
286 INSTALL_PATH = $$HOME
288 PEDANTIC_PARANOID_FREAK = -O0 -Wshadow -Wcast-align \
289 -Waggregate-return -Wmissing-prototypes -Wmissing-declarations \
290 -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations \
291 -Wmissing-noreturn -Wredundant-decls -Wnested-externs \
292 -Wpointer-arith -Wwrite-strings -finline-functions
293 REASONABLY_CAREFUL_DUDE = -Wall
294 NO_PRAYER_FOR_THE_WICKED = -w -O2
295 WARNINGS = $(REASONABLY_CAREFUL_DUDE)
296 CFLAGS = -g $(WARNINGS)
298 INCLUDES = -I$(INSTALL_PATH)/include
299 DEFS = -L$(INSTALL_PATH)/lib/
300 LDADD = -lm -lsimgrid
304 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) $^ $(LIBS) $(LDADD) -o $@
307 $(CC) $(INCLUDES) $(DEFS) $(CFLAGS) -c -o $@ $<
310 rm -f $(BIN_FILES) *.o *~
316 The first two lines indicates what should be build when typing make
317 (<tt>masterslave</tt>) and of which files it is to be made of
318 (<tt>masterslave.o</tt> and <tt>sched.o</tt>). This makefile assumes
319 that you have set up correctly your <tt>LD_LIBRARY_PATH</tt> variable
320 (look, there is a <tt>LDADD = -lm -lsimgrid</tt>). If you prefer using
321 the static version, remove the <tt>-lsimgrid</tt> and add a
322 <tt>$(INSTALL_PATH)/lib/libsimgrid.a</tt> on the next line, right
323 after the <tt>LIBS = </tt>.
325 More generally, if you have never written a Makefile by yourself, type
326 in a terminal : <tt>info make</tt> and read the introduction. The
327 previous example should be enough for a first try but you may want to
328 perform some more complex compilations...
330 \subsection faq_setting_GRAS Setting up your own GRAS code
332 If you use the GRAS interface instead of the MSG one, then previous section
333 is not the better source of information. Instead, you should check the GRAS
334 tutorial in general, and the \ref GRAS_tut_tour_setup in particular.
337 \section faq_howto Feature related questions
339 \subsection faq_MIA "Could you please add (your favorite feature here) to SimGrid?"
341 Here is the deal. The whole SimGrid project (MSG, SURF, GRAS, ...) is
342 meant to be kept as simple and generic as possible. We cannot add
343 functions for everybody's need when these functions can easily be
344 built from the ones already in the API. Most of the time, it is
345 possible and when it was not possible we always have upgraded the API
346 accordingly. When somebody asks us a question like "How to do that?
347 Is there a function in the API to simply do this?", we're always glad
348 to answer and help. However if we don't need this code for our own
349 need, there is no chance we're going to write it... it's your job! :)
350 The counterpart to our answers is that once you come up with a neat
351 implementation of this feature (task duplication, RPC, thread
352 synchronization, ...), you should send it to us and we will be glad to
353 add it to the distribution. Thus, other people will take advantage of
354 it (and we don't have to answer this question again and again ;).
356 You'll find in this section a few "Missing In Action" features. Many
357 people have asked about it and we have given hints on how to simply do
358 it with MSG. Feel free to contribute...
360 \subsection faq_MIA_MSG MSG features
362 \subsubsection faq_MIA_examples I want some more complex MSG examples!
364 Many people have come to ask me a more complex example and each time,
365 they have realized afterward that the basics were in the previous three
368 Of course they have often been needing more complex functions like
369 MSG_process_suspend(), MSG_process_resume() and
370 MSG_process_isSuspended() (to perform synchronization), or
371 MSG_task_Iprobe() and MSG_process_sleep() (to avoid blocking
372 receptions), or even MSG_process_create() (to design asynchronous
373 communications or computations). But the examples are sufficient to
376 We know. We should add some more examples, but not really some more
377 complex ones... We should add some examples that illustrate some other
378 functionalities (like how to simply encode asynchronous
379 communications, RPC, process migrations, thread synchronization, ...)
380 and we will do it when we will have a little bit more time. We have
381 tried to document the examples so that they are understandable. Tell
382 us if something is not clear and once again feel free to participate!
385 \subsubsection faq_MIA_taskdup Missing in action: MSG Task duplication/replication
387 There is no task duplication in MSG. When you create a task, you can
388 process it or send it somewhere else. As soon as a process has sent
389 this task, he doesn't have this task anymore. It's gone. The receiver
390 process has got the task. However, you could decide upon receiving to
391 create a "copy" of a task but you have to handle by yourself the
392 semantic associated to this "duplication".
394 As we already told, we prefer keeping the API as simple as
395 possible. This kind of feature is rather easy to implement by users
396 and the semantic you associate really depends on people. Having a
397 *generic* task duplication mechanism is not that trivial (in
398 particular because of the data field). That is why I would recommand
399 that you write it by yourself even if I can give you advice on how to
402 You have the following functions to get informations about a task:
403 MSG_task_get_name(), MSG_task_get_compute_duration(),
404 MSG_task_get_remaining_computation(), MSG_task_get_data_size(),
405 and MSG_task_get_data().
407 You could use a dictionnary (#xbt_dict_t) of dynars (#xbt_dynar_t). If
408 you still don't see how to do it, please come back to us...
410 \subsubsection faq_MIA_asynchronous I want to do asynchronous communications in MSG
412 Up until now, there is no asynchronous communications in MSG. However,
413 you can create as many process as you want so you should be able to do
414 whatever you want... I've written a queue module to help implementing
415 some asynchronous communications at low cost (creating thousands of
416 process only to handle communications may be problematic in term of
417 performance at some point). I'll add it in the distribution asap.
419 \subsubsection faq_MIA_thread_synchronization I need to synchronize my MSG processes
421 You obviously cannot use pthread_mutexes of pthread_conds. The best
422 thing would be to propose similar structures. Unfortunately, we
423 haven't found time to do it yet. However you can try to play with
424 MSG_process_suspend() and MSG_process_resume(). You can even do some
425 synchronization with fake communications (using MSG_task_get(),
426 MSG_task_put() and MSG_task_Iprobe()).
428 \subsubsection faq_MIA_host_load Where is the get_host_load function hidden in MSG?
430 There is no such thing because its semantic wouldn't be really
431 clear. Of course, it is something about the amount of host throughput,
432 but there is as many definition of "host load" as people asking for
433 this function. First, you have to remember that resource availability
434 may vary over time, which make any load notion harder to define.
436 It may be instantaneous value or an average one. Moreover it may be only the
437 power of the computer, or may take the background load into account, or may
438 even take the currently running tasks into account. In some SURF models,
439 communications have an influence on computational power. Should it be taken
442 First of all, it's near to impossible to predict the load beforehands in the
443 simulator since it depends on too much parameters (background load
444 variation, bandwidth sharing algorithmic complexity) some of them even being
445 not known beforehands (other task starting at the same time). So, getting
446 this information is really hard (just like in real life). It's not just that
447 we want MSG to be as painful as real life. But as it is in some way
448 realistic, we face some of the same problems as we would face in real life.
450 How would you do it for real? The most common option is to use something
451 like NWS that performs active probes. The best solution is probably to do
452 the same within MSG, as in next code snippet. It is very close from what you
453 would have to do out of the simulator, and thus gives you information that
454 you could also get in real settings to not hinder the realism of your
458 double get_host_load() {
459 m_task_t task = MSG_task_create("test", 0.001, 0, NULL);
460 double date = MSG_get_clock();
462 MSG_task_execute(task);
463 date = MSG_get_clock() - date;
464 MSG_task_destroy(task);
469 Of course, it may not match your personal definition of "host load". In this
470 case, please detail what you mean on the mailing list, and we will extend
471 this FAQ section to fit your taste if possible.
473 \subsubsection faq_MIA_communication_time How can I get the *real* communication time?
475 Communications are synchronous and thus if you simply get the time
476 before and after a communication, you'll only get the transmission
477 time and the time spent to really communicate (it will also take into
478 account the time spent waiting for the other party to be
479 ready). However, getting the *real* communication time is not really
480 hard either. The following solution is a good starting point.
485 m_task_t task = MSG_task_create("Task", task_comp_size, task_comm_size,
486 calloc(1,sizeof(double)));
487 *((double*) task->data) = MSG_get_clock();
488 MSG_task_put(task, slaves[i % slaves_count], PORT_22);
489 INFO0("Send completed");
494 m_task_t task = NULL;
497 time1 = MSG_get_clock();
498 a = MSG_task_get(&(task), PORT_22);
499 time2 = MSG_get_clock();
500 if(time1<*((double *)task->data))
501 time1 = *((double *) task->data);
502 INFO1("Communication time : \"%f\" ", time2-time1);
504 MSG_task_destroy(task);
509 \subsection faq_MIA_SimDag SimDag related questions
511 \subsubsection faq_SG_comm Implementing communication delays between tasks.
513 A classic question of SimDag newcommers is about how to express a
514 communication delay between tasks. The thing is that in SimDag, both
515 computation and communication are seen as tasks. So, if you want to
516 model a data dependency between two DAG tasks t1 and t2, you have to
517 create 3 SD_tasks: t1, t2 and c and add dependencies in the following
521 SD_task_dependency_add(NULL, NULL, t1, c);
522 SD_task_dependency_add(NULL, NULL, c, t2);
525 This way task t2 cannot start before the termination of communication c
526 which in turn cannot start before t1 ends.
528 When creating task c, you have to associate an amount of data (in bytes)
529 corresponding to what has to be sent by t1 to t2.
531 Finally to schedule the communication task c, you have to build a list
532 comprising the workstations on which t1 and t2 are scheduled (w1 and w2
533 for example) and build a communication matrix that should look like
536 \subsubsection faq_SG_DAG How to implement a distributed dynamic scheduler of DAGs.
538 Distributed is somehow "contagious". If you start making distributed
539 decisions, there is no way to handle DAGs directly anymore (unless I
540 am missing something). You have to encode your DAGs in term of
541 communicating process to make the whole scheduling process
542 distributed. Here is an example of how you could do that. Assume T1
543 has to be done before T2.
546 int your_agent(int argc, char *argv[] {
548 T1 = MSG_task_create(...);
549 T2 = MSG_task_create(...);
553 if(cond) MSG_task_execute(T1);
555 if((MSG_task_get_remaining_computation(T1)=0.0) && (you_re_in_a_good_mood))
558 /* do something else */
564 If you decide that the distributed part is not that much important and that
565 DAG is really the level of abstraction you want to work with, then you should
566 give a try to \ref SD_API.
568 \subsection faq_MIA_generic Generic features
570 \subsubsection faq_more_processes Increasing the amount of simulated processes
572 Here are a few tricks you can apply if you want to increase the amount
573 of processes in your simulations.
575 - <b>A few thousands of simulated processes</b> (soft tricks)\n
576 SimGrid can use either pthreads library or the UNIX98 contextes. On
577 most systems, the number of pthreads is limited and then your
578 simulation may be limited for a stupid reason. This is especially
579 true with the current linux pthreads, and I cannot get more than
580 2000 simulated processes with pthreads on my box. The UNIX98
581 contexts allow me to raise the limit to 25,000 simulated processes
583 The <tt>--with-context</tt> option of the <tt>./configure</tt>
584 script allows you to choose between UNIX98 contextes
585 (<tt>--with-context=ucontext</tt>) and the pthread version
586 (<tt>--with-context=pthread</tt>). The default value is ucontext
587 when the script detect a working UNIX98 context implementation. On
588 Windows boxes, the provided value is discarded and an adapted
589 version is picked up.\n\n
590 We experienced some issues with contextes on some rare systems
591 (solaris 8 and lower or old alpha linuxes comes to mind). The main
592 problem is that the configure script detect the contextes as being
593 functional when it's not true. If you happen to use such a system,
594 switch manually to the pthread version, and provide us with a good
595 patch for the configure script so that it is done automatically ;)
597 - <b>Hundred thousands of simulated processes</b> (hard-core tricks)\n
598 As explained above, SimGrid can use UNIX98 contextes to represent
599 and handle the simulated processes. Thanks to this, the main
600 limitation to the number of simulated processes becomes the
601 available memory.\n\n
602 Here are some tricks I had to use in order to run a token ring
603 between 25,000 processes on my laptop (1Gb memory, 1.5Gb swap).\n
604 - First of all, make sure your code runs for a few hundreds
605 processes before trying to push the limit. Make sure it's
606 valgrind-clean, ie that valgrind does not report neither memory
607 error nor memory leaks. Indeed, numerous simulated processes
608 result in *fat* simulation hindering debugging.
609 - It was really boring to write 25,000 entries in the deployment
610 file, so I wrote a little script
611 <tt>examples/gras/mutual_exclusion/simple_token/make_deployment.pl</tt>, which you may
612 want to adapt to your case. You could also think about hijacking
613 the SURFXML parser (have look at \ref faq_flexml_bypassing).
614 - The deployment file became quite big, so I had to do what is in
615 the FAQ entry \ref faq_flexml_limit
616 - Each UNIX98 context has its own stack entry. As debugging this is
617 quite hairly, the default value is a bit overestimated so that
618 user don't get into trouble about this. You want to tune this
619 size to increse the number of processes. This is the
620 <tt>STACK_SIZE</tt> define in
621 <tt>src/xbt/xbt_context_sysv.c</tt>, which is 128kb by default.
622 Reduce this as much as you can, but be warned that if this value
623 is too low, you'll get a segfault. The token ring example, which
624 is quite simple, runs with 40kb stacks.
625 - You may tweak the logs to reduce the stack size further. When
626 logging something, we try to build the string to display in a
627 char array on the stack. The size of this array is constant (and
628 equal to XBT_LOG_BUFF_SIZE, defined in include/xbt/log/h). If the
629 string is too large to fit this buffer, we move to a dynamically
630 sized buffer. In which case, we have to traverse one time the log
631 event arguments to compute the size we need for the buffer,
632 malloc it, and traverse the argument list again to do the actual
634 The idea here is to move XBT_LOG_BUFF_SIZE to 1, forcing the logs
635 to use a dynamic array each time. This allows us to lower further
636 the stack size at the price of some performance loss...\n
637 This allowed me to run the reduce the stack size to ... 4k. Ie,
638 on my 1Gb laptop, I can run more than 250,000 processes!
640 \subsubsection faq_MIA_batch_scheduler Is there a native support for batch schedulers in SimGrid?
642 No, there is no native support for batch schedulers and none is
643 planned because this is a very specific need (and doing it in a
644 generic way is thus very hard). However some people have implemented
645 their own batch schedulers. Vincent Garonne wrote one during his PhD
646 and put his code in the contrib directory of our SVN so that other can
647 keep working on it. You may find inspinring ideas in it.
649 \subsubsection faq_MIA_checkpointing I need a checkpointing thing
651 Actually, it depends on whether you want to checkpoint the simulation, or to
652 simulate checkpoints.
654 The first one could help if your simulation is a long standing process you
655 want to keep running even on hardware issues. It could also help to
656 <i>rewind</i> the simulation by jumping sometimes on an old checkpoint to
657 cancel recent calculations.\n
658 Unfortunately, such thing will probably never exist in SG. One would have to
659 duplicate all data structures because doing a rewind at the simulator level
660 is very very hard (not talking about the malloc free operations that might
661 have been done in between). Instead, you may be interested in the Libckpt
662 library (http://www.cs.utk.edu/~plank/plank/www/libckpt.html). This is the
663 checkpointing solution used in the condor project, for example. It makes it
664 easy to create checkpoints (at the OS level, creating something like core
665 files), and rerunning them on need.
667 If you want to simulate checkpoints instead, it means that you want the
668 state of an executing task (in particular, the progress made towards
669 completion) to be saved somewhere. So if a host (and the task executing on
670 it) fails (cf. #MSG_HOST_FAILURE), then the task can be restarted
671 from the last checkpoint.\n
673 Actually, such a thing does not exists in SimGrid either, but it's just
674 because we don't think it is fundamental and it may be done in the user code
675 at relatively low cost. You could for example use a watcher that
676 periodically get the remaining amount of things to do (using
677 MSG_task_get_remaining_computation()), or fragment the task in smaller
680 \subsection faq_platform Platform building and Dynamic resources
682 \subsubsection faq_platform_example Where can I find SimGrid platform files?
684 There is several little examples in the archive, in the examples/msg
685 directory. From time to time, we are asked for other files, but we
686 don't have any at hand right now.
688 We do have a description of the Grid'5000 platform, but because of
689 some flaws in the current formalism, this file is actually 500Mb,
690 which is ways too big to be used in the parser. We have an internship
691 currently working on xml syntax improvements which will allow to
692 reduce the size of this file and release it.
694 Once it's done, we plan to model manually other existing platforms the
695 same way we did for G5K and also release those files in some sort of
696 SimGrid platform catalog project.
698 \subsubsection faq_platform_alnem How can I automatically map an existing platform?
700 We are working on a project called ALNeM (Application-Level Network
701 Mapper) which goal is to automatically discover the topology of an
702 existing network. Its output will be a platform description file
703 following the SimGrid syntax, so everybody will get the ability to map
704 their own lab network (and contribute them to the catalog project).
705 This tool is not ready yet, but it move quite fast forward. Just stay
708 \subsubsection faq_platform_synthetic Generating synthetic but realistic platforms
710 The third possibility to get a platform file (after manual or
711 automatic mapping of real platforms) is to generate synthetic
712 platforms. Getting a realistic result is not a trivial task, and
713 moreover, nobody is really able to define what "realistic" means when
714 speaking of topology files. You can find some more thoughts on this
716 <a href="http://graal.ens-lyon.fr/~alegrand/articles/Simgrid-Introduction.pdf">slides</a>.
718 If you are looking for an actual tool, there we have a little tool to
719 annotate Tiers-generated topologies. This perl-script is in
720 <tt>tools/platform_generation/</tt> directory of the SVN. Dinda et Al.
721 released a very comparable tool, and called it GridG.
723 \subsubsection faq_SURF_dynamic Expressing dynamic resource availability in platform files
725 A nice feature of SimGrid is that it enables you to seamlessly have
726 resources whose availability change over time. When you build a
727 platform, you generally declare hosts like that:
730 <host id="host A" power="100.00"/>
733 If you want the availability of "host A" to change over time, the only
734 thing you have to do is change this definition like that:
737 <host id="host A" power="100.00" availability_file="trace_A.txt" state_file="trace_A_failure.txt"/>
740 For hosts, availability files are expressed in fraction of available
741 power. Let's have a look at what "trace_A.txt" may look like:
750 At time 0, our host will deliver 100 flop/s. At time 11.0, it will
751 deliver only 50 flop/s until time 20.0 where it will will start
752 delivering 90 flop/s. Last at time 21.0 (20.0 plus the periodicity
753 1.0), we'll be back to the beginning and it will deliver 100 flop/s.
755 Now let's look at the state file:
762 A negative value means "off" while a positive one means "on". At time
763 1.0, the host is on. At time 1.0, it is turned off and at time 2.0, it
764 is turned on again until time 12 (2.0 plus the periodicity 10.0). It
765 will be turned on again at time 13.0 until time 23.0, and so on.
767 Now, let's look how the same kind of thing can be done for network
768 links. A usual declaration looks like:
771 <link id="LinkA" bandwidth="10.0" latency="0.2"/>
774 You have at your disposal the following options: bandwidth_file,
775 latency_file and state_file. The only difference with hosts is that
776 bandwidth_file and latency_file do not express fraction of available
777 power but are expressed directly in bytes per seconds and seconds.
779 \subsubsection faq_platform_multipath How to express multipath routing in platform files?
781 It is unfortunately impossible to express the fact that there is more
782 than one routing path between two given hosts. Let's consider the
783 following platform file:
786 <route src="A" dst="B">
789 <route src="B" dst="C">
792 <route src="A" dst="C">
797 Althrough it is perfectly valid, it does not mean that data traveling
798 from A to C can either go directly (using link 3) or through B (using
799 links 1 and 2). It simply means that the routing on the graph is not
800 trivial, and that data do not following the shortest path in number of
801 hops on this graph. Another way to say it is that there is no implicit
802 in these routing descriptions. The system will only use the routes you
803 declare (such as <route src="A" dst="C"><link:ctn
804 id="3"/></route>), without trying to build new routes by aggregating
807 You are also free to declare platform where the routing is not
808 symetric. For example, add the following to the previous file:
811 <route src="C" dst="A">
817 This makes sure that data from C to A go through B where data from A
818 to C go directly. Don't worry about realism of such settings since
819 we've seen ways more weird situation in real settings (in fact, that's
820 the realism of very regular platforms which is questionable, but
821 that's another story).
823 \subsubsection faq_flexml_bypassing Bypassing the XML parser with your own C functions
825 So you want to bypass the XML files parser, uh? Maybe doin some parameter
826 sweep experiments on your simulations or so? This is possible, and
827 it's not even really difficult (well. Such a brutal idea could be
828 harder to implement). Here is how it goes.
830 For this, you have to first remember that the XML parsing in SimGrid is done
831 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
832 you want to bypass the parser, you need to provide some code mimicking what
833 it does and replacing it in its interactions with the SURF code. So, let's
834 have a look at these interactions.
836 FleXML parser are close to classical SAX parsers. It means that a
837 well-formed SimGrid platform XML file might result in the following
840 - start "platform_description" with attribute version="2"
841 - start "host" with attributes id="host1" power="1.0"
843 - start "host" with attributes id="host2" power="2.0"
845 - start "link" with ...
847 - start "route" with ...
848 - start "link:ctn" with ...
851 - end "platform_description"
853 The communication from the parser to the SURF code uses two means:
854 Attributes get copied into some global variables, and a surf-provided
855 function gets called by the parser for each event. For example, the event
856 - start "host" with attributes id="host1" power="1.0"
858 let the parser do something roughly equivalent to:
860 strcpy(A_host_id,"host1");
865 In SURF, we attach callbacks to the different events by initializing the
866 pointer functions to some the right surf functions. Since there can be
867 more than one callback attached to the same event (if more than one
868 model is in use, for example), they are stored in a dynar. Example in
869 workstation_ptask_L07.c:
871 /* Adding callback functions */
872 surf_parse_reset_parser();
873 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
874 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
875 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
876 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
877 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
878 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
881 surf_parse_open(file);
882 xbt_assert1((!surf_parse()), "Parse error in %s", file);
886 So, to bypass the FleXML parser, you need to write your own version of the
887 surf_parse function, which should do the following:
888 - Fill the A_<tag>_<attribute> variables with the wanted values
889 - Call the corresponding STag_<tag>_fun function to simulate tag start
890 - Call the corresponding ETag_<tag>_fun function to simulate tag end
891 - (do the same for the next set of values, and loop)
893 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
895 surf_parse = surf_parse_bypass_environment;
896 MSG_create_environment(NULL);
897 surf_parse = surf_parse_bypass_application;
898 MSG_launch_application(NULL);
901 A set of macros are provided at the end of
902 include/surf/surfxml_parse.h to ease the writing of the bypass
903 functions. An example of this trick is distributed in the file
904 examples/msg/masterslave/masterslave_bypass.c
906 \section faq_troubleshooting Troubleshooting
908 \subsection faq_trouble_lib_compil SimGrid compilation and installation problems
910 \subsubsection faq_trouble_lib_config ./configure fails!
912 We now only one reason for the configure to fail:
914 - <b>You are using a borken build environment</b>\n
915 If symptom is that configure complains about gcc not being able to build
916 executables, you are probably missing the libc6-dev package. Damn Ubuntu.
918 If you experience other kind of issue, please get in touch with us. We are
919 always interested in improving our portability to new systems.
921 \subsubsection faq_trouble_distcheck Dude! "make check" fails on my machine!
923 Don't assume we never run this target, because we do. Check
924 http://bob.loria.fr:8010 if you don't believe us.
926 There is several reasons which may cause the make check to fail on your
929 - <b>You are using a borken libc (probably concerning the contextes)</b>.\n
930 The symptom is that the "make check" fails within the examples/msg directory.\n
931 By default, SimGrid uses something called ucontexts. This is part of the
932 libc, but it's quite undertested. For example, some (old) versions of the
933 glibc on alpha do not implement these functions, but provide the stubs
934 (which return ENOSYS: not implemented). It may fool our detection mecanism
935 and leads to segfaults. There is not much we can do to fix the bug.
936 A workaround is to compile with --with-context=pthread to avoid
937 ucontext completely. You'll be a bit more limitated in the number
938 of simulated processes you can start concurently, but 5000
939 processes is still enough for most purposes, isn't it?\n
940 This limitation is the reason why we insist on using this piece of ...
941 software even if it's so troublesome.\n
942 <b>=> use --with-pthread on AMD64 architecture that do not have an
943 ultra-recent libc.</b>
945 - <b>There is a bug in SimGrid we aren't aware of</b>.\n
946 If none of the above apply, please drop us a mail on the mailing list so
947 that we can check it out. Make sure to read \ref faq_bugrepport
950 \subsection faq_trouble_compil User code compilation problems
952 \subsubsection faq_trouble_err_logcat "gcc: _simgrid_this_log_category_does_not_exist__??? undeclared (first use in this function)"
954 This is because you are using the log mecanism, but you didn't created
955 any default category in this file. You should refer to \ref XBT_log
956 for all the details, but you simply forgot to call one of
957 XBT_LOG_NEW_DEFAULT_CATEGORY() or XBT_LOG_NEW_DEFAULT_SUBCATEGORY().
959 \subsubsection faq_trouble_pthreadstatic "gcc: undefinded reference to pthread_key_create"
961 This indicates that one of the library SimGrid depends on (libpthread
962 here) was missing on the linking command line. Dependencies of
963 libsimgrid are expressed directly in the dynamic library, so it's
964 quite impossible that you see this message when doing dynamic linking.
966 If you compile your code statically (and if you use a pthread version
967 of SimGrid -- see \ref faq_more_processes), you must absolutely
968 specify <tt>-lpthread</tt> on the linker command line. As usual, this should
969 come after <tt>-lsimgrid</tt> on this command line.
971 \subsection faq_trouble_errors Runtime error messages
973 \subsubsection faq_flexml_limit "surf_parse_lex: Assertion `next limit' failed."
975 This is because your platform file is too big for the parser.
977 Actually, the message comes directly from FleXML, the technology on top of
978 which the parser is built. FleXML has the bad idea of fetching the whole
979 document in memory before parsing it. And moreover, the memory buffer size
980 must be determinded at compilation time.
982 We use a value which seems big enough for our need without bloating the
983 simulators footprints. But of course your mileage may vary. In this case,
984 just edit src/surf/surfxml.l modify the definition of
985 FLEXML_BUFFERSTACKSIZE. E.g.
988 #define FLEXML_BUFFERSTACKSIZE 1000000000
991 Then recompile and everything should be fine, provided that your version of
992 Flex is recent enough (>= 2.5.31). If not the compilation process should
995 A while ago, we worked on FleXML to reduce a bit its memory consumtion, but
996 these issues remain. There is two things we should do:
998 - use a dynamic buffer instead of a static one so that the only limit
999 becomes your memory, not a stupid constant fixed at compilation time
1000 (maybe not so difficult).
1001 - change the parser so that it does not need to get the whole file in
1002 memory before parsing
1003 (seems quite difficult, but I'm a complete newbe wrt flex stuff).
1005 These are changes to FleXML itself, not SimGrid. But since we kinda hijacked
1006 the development of FleXML, I can grant you that any patches would be really
1007 welcome and quickly integrated.
1009 <b>Update:</b> A new version of FleXML (1.7) was released. Most of the work
1010 was done by William Dowling, who use it in his own work. The good point is
1011 that it now use a dynamic buffer, and that the memory usage was greatly
1012 improved. The downside is that William also changed some things internally,
1013 and it breaks the hack we devised to bypass the parser, as explained in
1014 \ref faq_flexml_bypassing. Indeed, this is not a classical usage of the
1015 parser, and Will didn't imagine that we may have used (and even documented)
1016 such a crude usage of FleXML. So, we now have to repare the bypassing
1017 functionnality to use the lastest FleXML version and fix the memory usage in
1020 \subsubsection faq_trouble_gras_transport GRAS spits networking error messages
1022 Gras, on real platforms, naturally use regular sockets to communicate. They
1023 are deeply hiden in the gras abstraction, but when things go wrong, you may
1024 get some weird error messages. Here are some example, with the probable
1027 - <b>Transport endpoint is not connected</b>: several processes try to open
1028 a server socket on the same port number of the same machine. This is
1029 naturally bad and each process should pick its own port number for this.\n
1030 Maybe, you just have some processes remaining from a previous experiment
1032 Killing them may help, but again if you kill -KILL them, you'll have to
1033 wait for a while: they didn't close there sockets properly and the system
1034 needs a while to notice that this port is free again.
1036 - <b>Socket closed by remote side</b>: if the remote process is not
1037 supposed to close the socket at this point, it may be dead.
1039 - <b>Connection reset by peer</b>: I found this on internet about this
1040 error. I think it's what's happening here, too:\n
1041 <i>This basically means that a network error occurred while the client was
1042 receiving data from the server. But what is really happening is that the
1043 server actually accepts the connection, processes the request, and sends
1044 a reply to the client. However, when the server closes the socket, the
1045 client believes that the connection has been terminated abnormally
1046 because the socket implementation sends a TCP reset segment telling the
1047 client to throw away the data and report an error.\n
1048 Sometimes, this problem is caused by not properly closing the
1049 input/output streams and the socket connection. Make sure you close the
1050 input/output streams and socket connection properly. If everything is
1051 closed properly, however, and the problem persists, you can work around
1052 it by adding a one-second sleep before closing the streams and the
1053 socket. This technique, however, is not reliable and may not work on all
1055 Since GRAS sockets are closed properly (repeat after me: there is no bug
1056 in GRAS), it is either that you are closing your sockets on server side
1057 before the client get a chance to read them (use gras_os_sleep() to delay
1058 the server), or the server died awfully before the client got the data.
1060 \subsubsection faq_trouble_errors_big_fat_warning I'm told that my XML files are too old.
1062 The format of the XML platform description files is sometimes
1063 improved. For example, we decided to change the units used in SimGrid
1064 from MBytes, MFlops and seconds to Bytes, Flops and seconds to ease
1065 people exchanging small messages. We also reworked the route
1066 descriptions to allow more compact descriptions.
1068 That is why the XML files are versionned using the 'version' attribute
1069 of the root tag. Currently, it should read:
1071 <platform version="2">
1074 If your files are too old, you can use the simgrid_update_xml.pl
1075 script which can be found in the tools directory of the archive.
1077 \subsection faq_trouble_valgrind Valgrind-related and other debugger issues
1079 If you don't, you really should use valgrind to debug your code, it's
1082 \subsubsection faq_trouble_vg_longjmp longjmp madness in valgrind
1084 This is when valgrind starts complaining about longjmp things, just like:
1086 \verbatim ==21434== Conditional jump or move depends on uninitialised value(s)
1087 ==21434== at 0x420DBE5: longjmp (longjmp.c:33)
1089 ==21434== Use of uninitialised value of size 4
1090 ==21434== at 0x420DC3A: __longjmp (__longjmp.S:48)
1093 This is the sign that you didn't used the exception mecanism well. Most
1094 probably, you have a <tt>return;</tt> somewhere within a <tt>TRY{}</tt>
1095 block. This is <b>evil</b>, and you must not do this. Did you read the section
1098 \subsubsection faq_trouble_vg_libc Valgrind spits tons of errors about backtraces!
1100 It may happen that valgrind, the memory debugger beloved by any decent C
1101 programmer, spits tons of warnings like the following :
1102 \verbatim ==8414== Conditional jump or move depends on uninitialised value(s)
1103 ==8414== at 0x400882D: (within /lib/ld-2.3.6.so)
1104 ==8414== by 0x414EDE9: (within /lib/tls/i686/cmov/libc-2.3.6.so)
1105 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
1106 ==8414== by 0x414F937: _dl_open (in /lib/tls/i686/cmov/libc-2.3.6.so)
1107 ==8414== by 0x4150F4C: (within /lib/tls/i686/cmov/libc-2.3.6.so)
1108 ==8414== by 0x400B105: (within /lib/ld-2.3.6.so)
1109 ==8414== by 0x415102D: __libc_dlopen_mode (in /lib/tls/i686/cmov/libc-2.3.6.so)
1110 ==8414== by 0x412D6B9: backtrace (in /lib/tls/i686/cmov/libc-2.3.6.so)
1111 ==8414== by 0x8076446: xbt_dictelm_get_ext (dict_elm.c:714)
1112 ==8414== by 0x80764C1: xbt_dictelm_get (dict_elm.c:732)
1113 ==8414== by 0x8079010: xbt_cfg_register (config.c:208)
1114 ==8414== by 0x806821B: MSG_config (msg_config.c:42)
1117 This problem is somewhere in the libc when using the backtraces and there is
1118 very few things we can do ourselves to fix it. Instead, here is how to tell
1119 valgrind to ignore the error. Add the following to your ~/.valgrind.supp (or
1120 create this file on need). Make sure to change the obj line according to
1121 your personnal mileage (change 2.3.6 to the actual version you are using,
1122 which you can retrieve with a simple "ls /lib/ld*.so").
1125 name: Backtrace madness
1127 obj:/lib/ld-2.3.6.so
1132 fun:__libc_dlopen_mode
1135 Then, you have to specify valgrind to use this suppression file by passing
1136 the <tt>--suppressions=$HOME/.valgrind.supp</tt> option on the command line.
1137 You can also add the following to your ~/.bashrc so that it gets passed
1138 automatically. Actually, it passes a bit more options to valgrind, and this
1139 happen to be my personnal settings. Check the valgrind documentation for
1142 \verbatim export VALGRIND_OPTS="--leak-check=yes --leak-resolution=high --num-callers=40 --tool=memcheck --suppressions=$HOME/.valgrind.supp" \endverbatim
1144 \subsubsection faq_trouble_backtraces Truncated backtraces
1146 When debugging SimGrid, it's easier to pass the
1147 --disable-compiler-optimization flag to the configure if valgrind or
1148 gdb get fooled by the optimization done by the compiler. But you
1149 should remove these flages when everything works before going in
1150 production (before launching your 1252135 experiments), or everything
1151 will run only one half of the true SimGrid potential.
1153 \subsection faq_deadlock There is a deadlock in my code!!!
1155 Unfortunately, we cannot debug every code written in SimGrid. We
1156 furthermore believe that the framework provides ways enough
1157 information to debug such informations yourself. If the textual output
1158 is not enough, Make sure to check the \ref faq_visualization FAQ entry to see
1159 how to get a graphical one.
1161 Now, if you come up with a really simple example that deadlocks and
1162 you're absolutely convinced that it should not, you can ask on the
1163 list. Just be aware that you'll be severely punished if the mistake is
1164 on your side... We have plenty of FAQ entries to redact and new
1165 features to implement for the impenitents! ;)
1167 \subsection faq_surf_network_latency I get weird timings when I play with the latencies.
1169 OK, first of all, remember that units should be Bytes, Flops and
1170 Seconds. If you don't use such units, some SimGrid constants (e.g. the
1171 SG_TCP_CTE_GAMMA constant used in most network models) won't have the
1172 right unit and you'll end up with weird results.
1174 Here is what happens with a single transfer of size L on a link
1175 (bw,lat) when nothing else happens.
1178 0-----lat--------------------------------------------------t
1179 |-----|**** real_bw =min(bw,SG_TCP_CTE_GAMMA/(2*lat)) *****|
1182 In more complex situations, this min is the solution of a complex
1183 max-min linear system. Have a look
1184 <a href="http://lists.gforge.inria.fr/pipermail/simgrid-devel/2006-April/thread.html">here</a>
1185 and read the two threads "Bug in SURF?" and "Surf bug not
1186 fixed?". You'll have a few other examples of such computations. You
1187 can also read "A Network Model for Simulation of Grid Application" by
1188 Henri Casanova and Loris Marchal to have all the details. The fact
1189 that the real_bw is smaller than bw is easy to understand. The fact
1190 that real_bw is smaller than SG_TCP_CTE_GAMMA/(2*lat) is due to the
1191 window-based congestion mechanism of TCP. With TCP, you can't exploit
1192 your huge network capacity if you don't have a good round-trip-time
1193 because of the acks...
1195 Anyway, what you get is t=lat + L/min(bw,SG_TCP_CTE_GAMMA/(2*lat)).
1197 * if I you set (bw,lat)=(100 000 000, 0.00001), you get t = 1.00001 (you fully
1199 * if I you set (bw,lat)=(100 000 000, 0.0001), you get t = 1.0001 (you're on the
1201 * if I you set (bw,lat)=(100 000 000, 0.001), you get t = 10.001 (ouch!)
1203 This bound on the effective bandwidth of a flow is not the only thing
1204 that may make your result be unexpected. For example, two flows
1205 competing on a saturated link receive an amount of bandwidth inversely
1206 proportional to their round trip time.
1208 \subsection faq_bugrepport So I've found a bug in SimGrid. How to report it?
1210 We do our best to make sure to hammer away any bugs of SimGrid, but this is
1211 still an academic project so please be patient if/when you find bugs in it.
1212 If you do, the best solution is to drop an email either on the simgrid-user
1213 or the simgrid-devel mailing list and explain us about the issue. You can
1214 also decide to open a formal bug report using the
1215 <a href="https://gforge.inria.fr/tracker/?atid=165&group_id=12&func=browse">relevant
1216 interface</a>. You need to login on the server to get the ability to submit
1219 We will do our best to solve any problem repported, but you need to help us
1220 finding the issue. Just telling "it segfault" isn't enough. Telling "It
1221 segfaults when running the attached simulator" doesn't really help either.
1222 You may find the following article interesting to see how to repport
1223 informative bug repports:
1224 http://www.chiark.greenend.org.uk/~sgtatham/bugs.html (it is not SimGrid
1225 specific at all, but it's full of good advices).
1227 \author Arnaud Legrand (arnaud.legrand::imag.fr)
1228 \author Martin Quinson (martin.quinson::loria.fr)
1233 ******************************************************************
1234 * OLD CRUFT NOT USED ANYMORE *
1235 ******************************************************************
1238 \subsection faq_crosscompile Cross-compiling a Windows DLL of SimGrid from linux
1240 At the moment, we do not distribute Windows pre-compiled version of SimGrid
1241 because the support for this platform is still experimental. We know that
1242 some parts of the GRAS environment do not work, and we think that the others
1243 environments (MSG and SD) have good chances to work, but we didn't test
1244 ourselves. This section explains how we generate the SimGrid DLL so that you
1245 can build it for yourself. First of all, you need to have a version more
1246 recent than 3.1 (ie, a SVN version as time of writting).
1248 In order to cross-compile the package to windows from linux, you need to
1249 install mingw32 (minimalist gnu win32). On Debian, you can do so by
1250 installing the packages mingw32 (compiler), mingw32-binutils (linker and
1251 so), mingw32-runtime.
1253 You can use the VPATH support of configure to compile at the same time for
1254 linux and windows without dupplicating the source nor cleaning the tree
1255 between each. Just run bootstrap (if you use the SVN) to run the autotools.
1256 Then, create a linux and a win directories. Then, type:
1257 \verbatim cd linux; ../configure --srcdir=.. <usual configure flags>; make; cd ..
1258 cd win; ../configure --srcdir=.. --host=i586-mingw32msvc <flags>; make; cd ..
1260 The trick to VPATH builds is to call configure from another directory,
1261 passing it an extra --srcdir argument to tell it where all the sources are.
1262 It will understand you want to use VPATH. Then, the trick to cross-compile
1263 is simply to add a --host argument specifying the target you want to build
1264 for. The i586-mingw32msvc string is what you have to pass to use the mingw32
1265 environment as distributed in Debian.
1267 After that, you can run all make targets from both directories, and test
1268 easily that what you change for one arch does not break the other one.
1270 It is possible that this VPATH build thing breaks from time to time in the
1271 SVN since it's quite fragile, but it's granted to work in any released
1272 version. If you experience problems, drop us a mail.
1274 Another possible source of issue is that at the moment, building the
1275 examples request to use the gras_stub_generator tool, which is a compiled
1276 program, not a script. In cross-compilation, you need to cross-execute with
1277 wine for example, which is not really pleasant. We are working on this, but
1278 in the meanwhile, simply don't build the examples in cross-compilation
1279 (<tt>cd src</tt> before running make).
1281 Program (cross-)compiled with mingw32 do request an extra DLL at run-time to be
1282 usable. For example, if you want to test your build with wine, you should do
1283 the following to put this library where wine looks for DLLs.
1285 cp /usr/share/doc/mingw32-runtime/mingwm10.dll.gz ~/.wine/c/windows/system/
1286 gunzip ~/.wine/c/windows/system/mingwm10.dll.gz
1289 The DLL is builded in src/.libs, and installed in the <i>prefix</i>/bin directory
1290 when you run make install.
1292 If you want to use it in a native project on windows, you need to use
1293 simgrid.dll and mingwm10.dll. For each DLL, you need to build .def file
1294 under linux (listing the defined symbols), and convert it into a .lib file
1295 under windows (specifying this in a way that windows compilers like). To
1296 generate the def files, run (under linux):
1297 \verbatim echo "LIBRARY libsimgrid-0.dll" > simgrid.def
1298 echo EXPORTS >> simgrid.def
1299 nm libsimgrid-0.dll | grep ' T _' | sed 's/.* T _//' >> simgrid.def
1300 nm libsimgrid-0.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> simgrid.def
1302 echo "LIBRARY mingwm10.dll" > mingwm10.def
1303 echo EXPORTS >> mingwm10.def
1304 nm mingwm10.dll | grep ' T _' | sed 's/.* T _//' >> mingwm10.def
1305 nm mingwm10.dll | grep ' D _' | sed 's/.* D _//' | sed 's/$/ DATA/' >> mingwm10.def
1308 To create the import .lib files, use the <tt>lib</tt> windows tool (from
1309 MSVC) the following way to produce simgrid.lib and mingwm10.lib
1310 \verbatim lib /def:simgrid.def
1311 lib /def:mingwm10.def
1314 If you happen to use Borland C Builder, the right command line is the
1315 following (note that you don't need any file.def to get this working).
1316 \verbatim implib simgrid.lib libsimgrid-0.dll
1317 implib mingwm10.lib mingwm10.dll
1320 Then, set the following parameters in Visual C++ 2005:
1321 Linker -> Input -> Additional dependencies = simgrid.lib mingwm10.lib
1323 Just in case you wonder how to generate a DLL from libtool in another
1324 project, we added -no-undefined to any lib*_la_LDFLAGS variables so that
1325 libtool accepts to generate a dynamic library under windows. Then, to make
1326 it true, we pass any dependencies (such as -lws2 under windows or -lpthread
1327 on need) on the linking line. Passing such deps is a good idea anyway so
1328 that they get noted in the library itself, avoiding the users to know about
1329 our dependencies and put them manually on their compilation line. Then we
1330 added the AC_LIBTOOL_WIN32_DLL macro just before AC_PROG_LIBTOOL in the
1331 configure.ac. It means that we exported any symbols which need to be.
1332 Nowadays, functions get automatically exported, so we don't need to load our
1333 header files with tons of __declspec(dllexport) cruft. We only need to do so
1334 for data, but there is no public data in SimGrid so we are good.