Thanks for the clarification. You're right! There's a segfault. I apologize for that. We get very few requests for standalone MTCP, and so we don't do as careful a job on testing. I'll look at this, and hopefully have it fixed shortly. I'll get back to you.
Best, - Gene On Mon, Apr 23, 2012 at 05:49:53PM +0100, Frederic B. wrote: > Hi Gene, > > The source file I'm running is in the dmtcp package, under directory mtcp, > file testmtcp2.c > 1. I admit I didn't do the test with dmtcp as I'm interested in mtcp only. > I'll do it tomorrow morning just to be sure it's an mtcp issue. > 2. I'm using dmtcp version 1.2.4 > 3. With the code checkedout from the svn repository, I get a segfault : > [13983] mtcp_restart_nolibc.c:232 mtcp_restoreverything: > error 22 unmapping from 0 to 0x111000 > Segmentation fault (core dumped) > > > Thanks, > Fred > > > > ________________________________ > De : Gene Cooperman <[email protected]> > À : Frederic B. <[email protected]> > Cc : "[email protected]" <[email protected]> > Envoyé le : Lundi 23 avril 2012 17h43 > Objet : Re: [Dmtcp-forum] Checkpoint problem with mtcp and test program > testmtcp2 > > Hi Fred, > Can you send us the code that you're running with MTPC, > so that we can test it locally? ESRCH means that MTCP thought that > the extra thread is still alive, but the thread has died. > I have two other small questions: > 1. I presume that this works correctly with standard DMTCP. The only > issue is the standalone MTCP case. Is this correct? > 2. Which version of DMTCP were you using? It it's not too much trouble, > could you checkout the svn and try it with that. To check out DMTCP: > svn co https://dmtcp.svn.sourceforge.net/svnroot/dmtcp/trunk dmtcp-svn > > Best, > - Gene > > On Mon, Apr 23, 2012 at 03:42:03PM +0100, Frederic B. wrote: > > Hello all, > > > > I hope my question won't be out of scope... I'm evaluating the mtcp module > > (no need to checkpoint distributed processes for the time being) and I have > > an execution error with the basic example provided : testmtcp2.c > > > > To search for the stupid thing I might have done, I've simplified this file. > > I'm starting only 1 thread and wait 5 seconds before doing a checkpoint. > > > > The problem I encounter at checkpoint time is that "tgkill" function called > > to stop threads before the checkpoint returns -1 with errno set to ESRCH for > > the 2nd thread (not the main task). I do not understand how it can happen > > as the thread exists and is running (and still runs after checkpoint). > > Did I miss a limitation ? (I'm really not fluent in assembly, I can't give > > you more details about INLINE_SYSCALL behavior). > > > > My configuration : > > Redhat 6.2 (32b) > > gcc version 4.4.6 > > > > Thanks, > > Fred > > > ------------------------------------------------------------------------------ > > For Developers, A Lot Can Happen In A Second. > > Boundary is the first to Know...and Tell You. > > Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! > > http://p.sf.net/sfu/Boundary-d2dvs2 > > > _______________________________________________ > > Dmtcp-forum mailing list > > [email protected] > > https://lists.sourceforge.net/lists/listinfo/dmtcp-forum ------------------------------------------------------------------------------ For Developers, A Lot Can Happen In A Second. Boundary is the first to Know...and Tell You. Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! http://p.sf.net/sfu/Boundary-d2dvs2 _______________________________________________ Dmtcp-forum mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dmtcp-forum
