LAM/MPI logo

LAM/MPI Development Mailing List Archives

  |   Home   |   Download   |   Documentation   |   FAQ   |   all just in this list

From: Jeff Squyres \(jsquyres\) (jsquyres_at_[hidden])
Date: 2006-05-01 07:20:36


The BLCR web site is http://ftg.lbl.gov/CheckpointRestart/CheckpointRestart.shtml. At the very bottom of that web page is an address where you can send questions about BLCR itself. Additionally, we wrote a paper about the BLCR integration in LAM/MPI: http://www.lam-mpi.org/papers/lacsi2003/

________________________________

        From: lam-devel-bounces_at_[hidden] [mailto:lam-devel-bounces_at_[hidden]] On Behalf Of Mars Lenjoy
        Sent: Monday, May 01, 2006 2:03 AM
        To: LAM/MPI development issues
        Subject: Re: [lam-devel] how to cause a checkpoint after register (src code)?
        
        
        well, that's make sence!
        thanks, Brian,
        
        so, in another words, the LAM registers the callback and then using cr_checkpoint to fire that, the process is done by the blcr kernal module.
        that's the answer what i want to know.
        
        could you tell me what's blcr mailing list?
        i didn't find that in the web. :(
        
        happy labour day!
        
        Brian Barrett <brbarret_at_[hidden]> wrote:

                On Apr 29, 2006, at 9:44 PM, Mars Lenjoy wrote:
                
> the BLCR has been implemented by registering a callback function
> and then request that!
>
> my question is that, after the register is done, how to cause the
> checkpoint?
>
> for example:
>
> === lam-7.1.2/share/ssi/crlam/blcr/src/ssi_crlam_blcr.c L153 ===
> /*
> * Register the checkpoint signal handler
> */
> cr_thread_callback_id = cr_register_callback(crlam_thread_callback,
> cr_thread_callback_arg,
> CR_THREAD_CONTEXT);
> cr_signal_callback_id = cr_register_callback(crlam_signal_callback,
> cr_signal_callback_arg,
> CR_SIGNAL_CONTEXT);
> ==================================================
>
                
                I'm not exactly sure what you are asking. cr_register_callback() is
                part of the BLCR API. It will call the specified callback when the
                process receives a checkpoint request. We don't really care how that
                callback is fired -- that's the problem of BLCR. The checkpoint
                requests are delivered to the specified process by the cr_checkpoint
                command that is part of the BLCR package.
                
                If you are interested in the internals of the BLCR implementation,
                you should contact the blcr mailing list. If I missed your question
                entirely, please let me know.
                
                
                Brian
                
                --
                Brian Barrett
                LAM/MPI developer and all around nice guy
                Have a LAM/MPI day: http://www.lam-mpi.org/
                
                
                _______________________________________________
                lam-devel mailing list
                lam-devel_at_[hidden]
                http://www.lam-mpi.org/mailman/listinfo.cgi/lam-devel
                

        
________________________________

        Love cheap thrills? Enjoy PC-to-Phone calls to 30+ countries <http://us.rd.yahoo.com/mail_us/taglines/postman9/*http://us.rd.yahoo.com/evt=39666/*http://messenger.yahoo.com/> for just 2¢/min with Yahoo! Messenger with Voice.