Question on Data Byte and Word Manipulation

Follow

Steve Tulk

• February 01, 2010 06:59

I am used to using Keil C51 on 8051 8 bit CPU's. Keil treats 16 bit and 32 bit data types as "Big Endian". The Luminary (TI) Cortex-M3 parts are hard locked to "Little Endian".

 

My problem is this:   In my 32 bit data, I regularly need to access specific 8 bit or 16 bit sections of the 32 bit data. I can do this with on the 8051 with Keil using a pointer such as:

ByteWanted = *(char*)&LongData;         // Gets me Most Significant Byte

or

ByteWanted = *(((char*)&LongData)+3);   // Gets me Least Significant Byte

 

If I try this on the Little Endian Cortex then I get the opposite byte. This has caused me a big portable code issue.

 

My question is this: Is there an easier way to access the individual bytes using Crossworks, "C" and/or Arm architecture? I could Mask and Shift but that will be CPU Cycle intensive. I could re-write it to index my bytes in the opposite way to what I have been doing but then I would have the same problem as I have now if I tried to port it to a different architecture.

 

Any suggestions?

 

Thanks in Advance

Steve

0

• 

  Michael Freeman

  • February 02, 2010 04:39

  Use shift operator >>

   

  ByteWanted = (unsigned char)((LongData >> 0) & 0xff); //Always gets least significant byte

   

  ByteWanted = (unsigned char)((LongData >> 24) & 0xff); //Always gets most significant byte 

   

  This works no mater what endian the CPU is.

   

   

   

  Here are the macros I use:

  #define word_get_low_byte(val) ((unsigned char)((val) & 0xFF))

   

  #define word_get_high_byte(val) ((unsigned char)(((val)>>8) & 0xFF))

   

  #define dword_get_low_byte(val) ((unsigned char)((val) & 0xFF))

   

  #define dword_get_mlb_byte(val) ((unsigned char)(((val)>>8) & 0xFF))

   

  #define dword_get_mhb_byte(val) ((unsigned char)(((val)>>16) & 0xFF))

   

  #define dword_get_high_byte(val) ((unsigned char)(((val)>>24) & 0xFF))

   

   

   

   

  0

  Comment actions Permalink
• 

  Graham Cole

  • February 02, 2010 09:29

  Shift and AND is the way to go. It is fully portable and, in the case of ARM, it is efficiently handled - the ARM can shift by any number of bits in a single instruction.

  0

  Comment actions Permalink
• 

  Steve Tulk

  • February 02, 2010 09:55

  Thank You both for the replies. I was not aware that ARM did single cycle shifts of any size. That being the case, it is no doubt the way to go.

   

  Steve

  0

  Comment actions Permalink

Please sign in to leave a comment.