I think you can use rcS file to run your alpha program and create checkpoints. For instance, if you want to create a checkpoint after the OS has booted up but before your program starts, the rcS file should look like:
#!/bin/sh cd /path/to/program/directory /sbin/m5 dumpstats /sbin/m5 resetstats ./your-alpha-program /sbin/m5 exit If you don't want to simulate the entire program but only part of it, you will need to insert M5 pseudo instructions at the beginning and end of your program range of interest, and recompile your program. For details, you can take a look at UT Austin group technical report "Running PARSEC 2.1 on M5" Page 8 about "M5 Hooks". Lide On Tue, Nov 2, 2010 at 8:30 AM, Veydan Wu <[email protected]> wrote: > Hi, Lide. Thanks for your reply. The reason I thought it is problematic is > that I cannot see the prompt of OS when executing the m5term. So I cannot > use it to run programs. The linux booting displays showed for only several > lines and then stop there; it is in the middle of booting and it could not > reach the prompt. Do you know what the reason maybe? I created the > checkpoint simply by "m5 check" in the prompt, there should be nothing wrong > here. It is really confusing. > > On Mon, Nov 1, 2010 at 1:20 PM, Lide Duan <[email protected]> wrote: > >> I also encountered this kind of warnings, but I think they are just >> warnings instead of errors, so the simulation should run without problems. >> How do you know "it could not enter the OS"? From my experience, after >> restoring from a checkpoint, the m5term simulated terminal displays >> something like: >> ==== m5 slave terminal: Terminal 0 ==== >> and no more output will be shown unless your alpha program outputs >> something (which can be very slow if detailed model is used). The Linux >> booting displays won't show in m5term now since they were already shown >> before the checkpoint was created. Finally, to verify that your simulation >> is running, you can simply stop the simulation (via Ctrl+C) after a while >> and check the stats output. >> >> Lide >> >> On Mon, Nov 1, 2010 at 9:54 AM, Veydan Wu <[email protected]> wrote: >> >>> Hi, all. I am trying to use checkpoint in ALPHA_FS mode. I followed the >>> instruction of the website, invoking "*m5 checkpoint*" after booting the >>> simulator. It created a checkpoint named "*cpt.76220053480500*". When I >>> tried to restored it later by typing "*build/ALPHA_FS/m5.fast >>> configs/example/fs.py -r 1*", the following errors came out: >>> * >>> Listening for system connection on port 3456 >>> 0: system.remote_gdb.listener: listening for remote gdb on port 7000 >>> Restoring from checkpoint >>> warn: Not unserializing 'system.cpu': no section found in checkpoint. >>> For more information see: http://www.m5sim.org/warn/dc3e87f8 >>> warn: Not unserializing 'system.cpu.tracer': no section found in >>> checkpoint. >>> For more information see: http://www.m5sim.org/warn/dc3e87f8 >>> warn: Not unserializing 'system.cpu.itb': no section found in checkpoint. >>> For more information see: http://www.m5sim.org/warn/dc3e87f8 >>> warn: Not unserializing 'system.cpu.interrupts': no section found in >>> checkpoint. >>> For more information see: http://www.m5sim.org/warn/dc3e87f8 >>> warn: Not unserializing 'system.cpu.dtb': no section found in checkpoint. >>> For more information see: http://www.m5sim.org/warn/dc3e87f8 >>> **** REAL SIMULATION **** >>> info: Entering event queue @ 76220053480500. Starting simulation...* >>> >>> After that I run the m5term but it could not enter the OS, it just stuck >>> there (I could not see the shell prompt). Was there anything wrong in my >>> operation above? I hope someone can give me a hint. Thanks. >>> >>> >>> >>> veyday >>> >>> >>> _______________________________________________ >>> m5-users mailing list >>> [email protected] >>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >>> >> >> >> _______________________________________________ >> m5-users mailing list >> [email protected] >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >> > > > > -- > Regards, > > Weidan Wu > > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >
_______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
