DProfiler
---------
 Disk Throughput Profiler

 Copyright 2011-2012 (c) Felipe Franciosi <[email protected]>

 Disclaimer:
  This program is free software: you can redistribute it and/or modify
  it under the terms of the GNU General Public License as published by
  the Free Software Foundation, either version 3 of the License, or
  (at your option) any later version.

  This program is distributed in the hope that it will be useful,
  but WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  GNU General Public License for more details.

  You should have received a copy of the GNU General Public License
  along with this program.  If not, see <http://www.gnu.org/licenses/>.


 Description:
  This program reads or writes to an entire block device and records
  the time (in us) it took to process a certain amount of data.
  It features variable block sizes and grouping of results. This allows
  easier comparison of different experiments. For example, it is possible
  to have the average throughput at every 32MB of data by reading 32MB
  blocks (-b $[ 32 * 1048576] -g 1), 16MB blocks (-b $[ 16 * 1048576 ] -g 2)
  or even 1MB blocks (-b 1048576 -g 32).

  The output is in the form: O T, where:
   O - Offset (in bytes) where the operation was performed
   T - Time (us) to complete the operation

   The time is printed in microseconds and includes *only* the time spent
   during the I/O operations.


 Compiling:
  gcc -o dprofiler dprofiler.c -lrt -Wall -O3 -s


 Usage:
  ./dprofiler < -r | -w | -i > < -d dev_name >
             [ -hmyv[v] ] [ -b <buf_size> ] [ -g <grp_size> ] [ -o <datafile> ]

     One of the following operation modes must be selected:
     -r             Read from device.
     -w             Write to device (*CAUTION* This will overwrite the device).
     -i             Print information on the device and quit.
                    (default is -i)

     The device must be specified as follows:
     -d <dev_name>  For example: /dev/sdb

     The following options are available for both reading and writing:
     -b <buf_size>  Buffer size, in bytes.
                    (default is 1048576)
     -g <grp_size>  Amount of samples to group together in the output report.
     -o <datafile>  Name of the file to write the results.
                    (preferably on a different device than the one profiled)
     -h             Print the program usage.
     -m             Set output to MB/s instead of microseconds.
     -y             Open device with O_SYNC (see open(2) man page).
     -v             Print details of the device to stderr.
     -vv            Same as above, plus report to stdout apart from <datafile>.
                    (if no datafile is desired, use -o /dev/null)
 

 Changelog:
  2012.04.17 - Felipe Franciosi
   - Fixed compatibility parameters.
   - Minor adjustments to the code.
   - Added the "-m" parameter (output in MB/s).
   - Added the "-i" parameter.
   - Times are actually printed in 'us'.

  2012.02.22 - Felipe Franciosi
   - Fixed compatibility parameters.
   - Made it available on github.

  2012.01.24 - Felipe Franciosi
   - Times are now printed in 'ms' for compatibility with yngwie.

  2011.11.01 - Felipe Franciosi
   - Got rid of LARGEFILE64_SOURCE and started using FILE_OFFSET_BITS.
   - Fixed data types to 64 bit architectures compatilibity.
   - Print usage to stderr.
   - Replaced perror()s to correct locations.
   - Minor corrections.
   - Made "-d" mandatory.

  2011.10.06 - Felipe Franciosi
   - Program creation.
   - Recycled old code into this project.




 TODO:
  - Implement random seeks (read/write).
  - Implement random content writes (currently only write zeros).