od and byte order

Objectives

1. Understand and use the program od on *nix systems.
2. Understand that the byte order in memory might not match the byte order in standard two's complement.

Notes

• References
  • man od and man hexdump
  • wikipedia.
• It is reasonable to look at files on the system in something other than "ascii" mode.
  • Mostly files you work with contain ascii data.
  • But not all.
• Create a file, letters.dat, with the letters in it.
  • cat letters.dat, what does this show?
  • Is there a way I could look at ascii values, the values in the file, not the character interpretation?
• Introduction od
  • This is an old unix utility that allows us the examine the contents of a file in different formats.
  • od -c letters.dat
  • The -c option displays in "ascii", note the \n
• od -x letters.dat
  • the -x says to print in hex format.
  • But why are the letters partially reversed?
• Intel processors are little-endian
  • Memory is just an array of bytes.
  • M[0] stores a byte, M[1] stores a byte
  • A character, is 8 bits or two hex digits.
    • So this is stored in a single byte.
    • in the previous od, we saw 6261 6463
    • And this represents ba dc
    • So the individual bytes are fine,
    • But the byte order is wrong (strange, challenging, ...).
  • As we discussed earlier, in Intel we often think of a word as 16 bits or 4 bytes. (this is not necessarily true, but ...)
  • Little endian means:
    • Intel represents the "first" byte or MSB at position n and the LSB or least byete as byte 0
    • Or at 0 from the start is the LSB, and n from the start is the msb.

      Address + 0	Address + 1
      LSB	MSB
      'b'	'a'

  • od --endian=big -x letters.dat
    • Will switch the byte ordering for a big-endian presentation.

      Address + 0	Address + 1
      LSB	MSB
      'a'	'b'

  • little endian reads small left to right
  • Big endian reads big right to left.
  • And this is somewhat backwards from our nature.
  • (wikipedia).
• I have some code to examine this.
  • By default c++ writes in ascii (ie convert to ascii and write the value)
  • If I open a stream with ifstream inFile(fileName, ios::binary); it will read binary not ascii.
  • I need to do this with inFile.read(char * location, size_t size);
    • It will read in size characters and store them starting at location.
  • inFile.clear() will remove the eof flag.
  • inFile.seekg(0,ios::beg) will reset the file pointer to the beginning of the file.
  • a << b shifts a to the left b times.
  • This is equivelent to multiplying by pow(2,b)
  • reader.cpp.
  • writeShort.cpp is a program that produces data.
• Note, we get the same character output
• Reading in shorts, we get a transposition, ie C++ knows what to do
• There is another program called hexdump that you might investigate.