From abdulrahman004 at gmail.com  Wed Sep  2 13:25:48 2009
From: abdulrahman004 at gmail.com (Abdul Rahman)
Date: Wed, 2 Sep 2009 10:25:48 -0700
Subject: [Disksim-users] Regarding simulation of events in ssd model disksim
Message-ID: <63e51aa0909021025v3487a06auc1f8e33ccd397ab2@mail.gmail.com>

I started using the ssd model of disksim simulator a couple of weeks back
and I have gone through the call stack of the functions that were called
after a request from the parameter file was read. I have debugged the
program to trace the function calls for a few requests that were handled, I
have a couple of doubts of how things are really simulated in the ssd
simulator. Firstly, I notice that the simulator handles function on a per
request basis. In the actual model requests come in all the time that are
buffered at the controller level to be handled later. Also, two or more
requests may be interleaved. I haven't seen how this behavior is simulated
in the model and its always like the call returns back to
disksim_simulate_event(event_count) function where the next request is
handled sequentially

Also, how is the time tracked for interleaved events. If there 4 requests
that were interleaved completely, the simulation time will be the time taken
for 1 request (rather than 4). So, where is the check being made to check if
an event's simulation time should be added to the total simulation time or
if it was a parallel operation. It might also be possible that an even was
partially interleaved , in which case only a fractin of time for that event
has to be added to the total simulation time instead of the complete time

Thanks
Abdul Rahman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.andrew.cmu.edu/pipermail/disksim-users/attachments/20090902/07666c80/attachment.html>

From hdiakos at csd.uoc.gr  Sun Sep  6 02:33:05 2009
From: hdiakos at csd.uoc.gr (hdiakos at csd.uoc.gr)
Date: Sun, 06 Sep 2009 09:33:05 +0300
Subject: [Disksim-users] new_disk.pl and dx_track errors
Message-ID: <20090906093305.hlqk5n9qssc0c808@mail.uoc.gr>

Hello all.

I was wondering if anyone has run the new_disk.pl or the dx_trac  
executable because when I run both I get this error:

Could not set FORCE_PACK_ID on device /dev/sdXX

Does anyone know the solution to that?I would appreciate it!

P.S.0:dx_trac runs as dx_stat -d <disk dev>
Is the disk parameter mandatory?Whenever I run it with the disk  
parameter I get an error:

Device XX...XX not defined or no R/W permissions.

Without I get the referenced FORCE_PACK_ID error.
Any ideas?






From hdiakos at csd.uoc.gr  Sun Sep  6 08:55:08 2009
From: hdiakos at csd.uoc.gr (hdiakos at csd.uoc.gr)
Date: Sun, 06 Sep 2009 15:55:08 +0300
Subject: [Disksim-users] Update:new_disk.pl and dx_track errors
Message-ID: <20090906155508.nih394fr440g488w@mail.uoc.gr>

After some debugging I found out that the error is generated from file  
build_scsi.c in line 153.Adding some diagnostic code the ioctl  
function returns -1 and I get the errno error description  
"Inappropiate ioctl for device".Here is the modified code:

   ictl_val = 0;
   int res;
   errno = 0;
   if ( 0 > ( res = ioctl(fd,SG_SET_FORCE_PACK_ID,&ictl_val) ) ) {
     printf("Printing FORCE_PACK_ID:%0x\n", SG_SET_FORCE_PACK_ID);
     printf("ioctl returned:%d ERRNO returned:%s\n", res, strerror(errno));
     error_handler("Could not set FORCE_PACK_ID  on device %s \n",devname);
   }

Does anyone have any idea as to what am I doing wrong?It perfectly  
finds devices such as sg0 which is a CD/DVD-ROM but it can't find any  
device labeled sdaX or sdbX...