Programming Forums > Application Development > C++

rsnd · Nov 14th, 2005, 6:41 PM

Well...HELP!!!

Just have 2 questions

1. What are some of the efficient ways of playing around with byte ordering? e.g. changing 0xFFCCDDEE to 0xEEDDCCFF.

2. I've made a program that parses intercepted data between a server and a client program to model client/server state. For convenience I've made a custom class that buffers the recived data and parses them. Its sortof like the "StreamBuffer" class in Java where you go .nextInt() or whatever and it progresses pointer giving you the values etc... Well, the problem is, the byte ordering looks reversed! for things like ints! e.g. for an unsigned int, 4 = 0x00000004...but the data I get reads 0x04000000! Initially I thaught its to do with a bug with the interception method ive impimented. But It's all good. Because I dont get problems with strings byte ordering. My suspicion is that it uses one of c++'s "standard" streaming classes...which upon works putes bytes in that order! Or they are just the way things are put set into memory! I think its the former! So I just want to know what you think. Or which "standard" class the programmer might have used to do the buffering. (sorry...I am not vary good with default library classes...I just look up the platform sdk to custom do things my way).

Thanks in advance

Animatronic · Nov 14th, 2005, 7:39 PM

There are the standard family of functions (for Berkeley sockets) like:

htons >> host to network short
htonl >> host to network long
ntohs >> network to host short
ntohl >> network to host long

etc.. for converting to/from network byte order, just add this to your buffer class.

rsnd · Nov 17th, 2005, 11:22 PM

well...but my question was...how do i do it myself

andro · Nov 18th, 2005, 12:25 AM

This code may or may not help you - I'm not sure if it's exactly what you're looking for :p

(Toggle Plain Text)

#include <stdio.h>
 
int main();
unsigned long convert(long num);
 
int main()
{
        unsigned long lendian = 1234;
        unsigned long bendian = convert(lendian);
 
        printf("Value: %lu\n", lendian );
 
        printf("Converted: %lu\n", bendian);
 
        lendian = convert(bendian);
        printf("Converted back: %lu\n", lendian);
 
        return 0;
}
 
unsigned long convert(long num)
{
        long returnval;
        char *p_num  = (char *) &num;
        char *p_returnval = (char *) &returnval;
 
        int i, j;
        for (i = 0, j = 3; i <4; i++, j--)
        {
                p_returnval[i] = p_num[j];
        }
 
 
        return returnval;
}

Animatronic · Nov 18th, 2005, 4:43 AM

The point is that you dont do the convertion yourself. Doing it yourself means you need to know the endianess of your system whilst using the standard functions you dont. Also I doubt your'll find a better implementation than the standard ones.

DaWei · Nov 18th, 2005, 6:24 AM

I'm not sure you realize that there IS an endianess issue. Not all machines arrange the individual bytes of a group in the same order. The common desktop Intel machines put the least significant byte first ("little endian"). Others put the msb first, in the way that we typically write our numbers (most western cultures). There are also in-between mixes. In the world of networking, one uses "network order." One calls the aforementioned functions as an interface. If no adjustment is necessary, your system's version will make none; if it is, it will.

rsnd · Nov 20th, 2005, 10:16 PM

Thanks everyone

I understand the issue now.
Anyways...will this work if I just add it to my partially working class?

(Toggle Plain Text)

	unsigned int GetUINT(){
		unsigned int temp=0;
		temp= (*(buffer+pointer++))<<0;
		temp|= (*(buffer+pointer++))<<8;
		temp|= (*(buffer+pointer++))<<16;
		temp|= (*(buffer+pointer++))<<24;
		return temp;
	}

have read access the the buffer.

DaWei · Nov 21st, 2005, 6:37 AM

If you know the underlying nature of your particular platform you can get away with coercions of various types. If you don't, you'll just make yourself trouble. That's why they make the functions. Your shifting of an entity by zero bits raises some question. If you want the practice, find out how your machine stores floats, then peel it apart and convert it yourself. If you're trying to conform to network byte order as a matter of course, use the standard functions.

lectricpharaoh · Nov 21st, 2005, 11:17 PM

Quote:

Originally Posted by rsnd

(Toggle Plain Text)

unsigned int GetUINT(){
	unsigned int temp=0;
	temp= (*(buffer+pointer++))<<0;
	temp|= (*(buffer+pointer++))<<8;
	temp|= (*(buffer+pointer++))<<16;
	temp|= (*(buffer+pointer++))<<24;
	return temp;
}

This might work. Then again, it will not be as efficient as using the standard functions. On the Intel 80x86 CPUs from the 486 onwards, there is a BSWAP machine instruction that can do this quite quickly. I'd expect other modern CPUs to have a similar instruction. If you can use inline assembly, this might be an option for you. Note that BSWAP works on a 32-bit value; if you want to use it on a 16-bit value, you'll have to shift the register right 16 places.

Some example code follows.

(Toggle Plain Text)

; uses NASM syntax- you can adjust if needed
; for 32-bit values
mov eax, [value]
bswap eax
mov [value], eax
; for 16-bit values
mov ax, [value]
bswap eax
shr eax, 16
mov [value], ax

Nov 14th, 2005, 6:41 PM	#1
rsnd Hobbyist Programmer Join Date: Jun 2005 Location: Helltown Posts: 162 Rep Power: 4	byte reversing & streaming Well...HELP!!! Just have 2 questions 1. What are some of the efficient ways of playing around with byte ordering? e.g. changing 0xFFCCDDEE to 0xEEDDCCFF. 2. I've made a program that parses intercepted data between a server and a client program to model client/server state. For convenience I've made a custom class that buffers the recived data and parses them. Its sortof like the "StreamBuffer" class in Java where you go .nextInt() or whatever and it progresses pointer giving you the values etc... Well, the problem is, the byte ordering looks reversed! for things like ints! e.g. for an unsigned int, 4 = 0x00000004...but the data I get reads 0x04000000! Initially I thaught its to do with a bug with the interception method ive impimented. But It's all good. Because I dont get problems with strings byte ordering. My suspicion is that it uses one of c++'s "standard" streaming classes...which upon works putes bytes in that order! Or they are just the way things are put set into memory! I think its the former! So I just want to know what you think. Or which "standard" class the programmer might have used to do the buffering. (sorry...I am not vary good with default library classes...I just look up the platform sdk to custom do things my way). Thanks in advance __________________ Spread your wings and fly! Chicken!

Nov 17th, 2005, 11:22 PM	#3
rsnd Hobbyist Programmer Join Date: Jun 2005 Location: Helltown Posts: 162 Rep Power: 4	well...but my question was...how do i do it myself __________________ Spread your wings and fly! Chicken!

Nov 18th, 2005, 12:25 AM	#4
andro Professional Programmer Join Date: Oct 2005 Location: California Posts: 310 Rep Power: 3	This code may or may not help you - I'm not sure if it's exactly what you're looking for :p (Toggle Plain Text) #include <stdio.h> int main(); unsigned long convert(long num); int main() { unsigned long lendian = 1234; unsigned long bendian = convert(lendian); printf("Value: %lu\n", lendian ); printf("Converted: %lu\n", bendian); lendian = convert(bendian); printf("Converted back: %lu\n", lendian); return 0; } unsigned long convert(long num) { long returnval; char p_num = (char ) &num; char p_returnval = (char ) &returnval; int i, j; for (i = 0, j = 3; i <4; i++, j--) { p_returnval[i] = p_num[j]; } return returnval; } #include <stdio.h> int main(); unsigned long convert(long num); int main() { unsigned long lendian = 1234; unsigned long bendian = convert(lendian); printf("Value: %lu\n", lendian ); printf("Converted: %lu\n", bendian); lendian = convert(bendian); printf("Converted back: %lu\n", lendian); return 0; } unsigned long convert(long num) { long returnval; char p_num = (char ) &num; char p_returnval = (char ) &returnval; int i, j; for (i = 0, j = 3; i <4; i++, j--) { p_returnval[i] = p_num[j]; } return returnval; }

Nov 18th, 2005, 6:24 AM	#6
DaWei Resident Grouch Join Date: Jun 2005 Posts: 6,453 Rep Power: 10	I'm not sure you realize that there IS an endianess issue. Not all machines arrange the individual bytes of a group in the same order. The common desktop Intel machines put the least significant byte first ("little endian"). Others put the msb first, in the way that we typically write our numbers (most western cultures). There are also in-between mixes. In the world of networking, one uses "network order." One calls the aforementioned functions as an interface. If no adjustment is necessary, your system's version will make none; if it is, it will. __________________ Abstraction doesn't make it impossible to write bad code; it makes it possible to write superior code. Contributor's Corner: Grumpy on C++ Exceptions DaWei on Pointers

Nov 20th, 2005, 10:16 PM	#7
rsnd Hobbyist Programmer Join Date: Jun 2005 Location: Helltown Posts: 162 Rep Power: 4	Thanks everyone I understand the issue now. Anyways...will this work if I just add it to my partially working class? (Toggle Plain Text) unsigned int GetUINT(){ unsigned int temp=0; temp= ((buffer+pointer++))<<0; temp\|= ((buffer+pointer++))<<8; temp\|= ((buffer+pointer++))<<16; temp\|= ((buffer+pointer++))<<24; return temp; } unsigned int GetUINT(){ unsigned int temp=0; temp= ((buffer+pointer++))<<0; temp\|= ((buffer+pointer++))<<8; temp\|= ((buffer+pointer++))<<16; temp\|= ((buffer+pointer++))<<24; return temp; } have read access the the buffer. __________________ Spread your wings and fly! Chicken!

Nov 14th, 2005, 7:39 PM	#2
Animatronic Programmer Join Date: Jun 2005 Posts: 99 Rep Power: 4	There are the standard family of functions (for Berkeley sockets) like: htons >> host to network short htonl >> host to network long ntohs >> network to host short ntohl >> network to host long etc.. for converting to/from network byte order, just add this to your buffer class.

Nov 18th, 2005, 4:43 AM	#5
Animatronic Programmer Join Date: Jun 2005 Posts: 99 Rep Power: 4	The point is that you dont do the convertion yourself. Doing it yourself means you need to know the endianess of your system whilst using the standard functions you dont. Also I doubt your'll find a better implementation than the standard ones.

Nov 21st, 2005, 6:37 AM	#8
DaWei Resident Grouch Join Date: Jun 2005 Posts: 6,453 Rep Power: 10	If you know the underlying nature of your particular platform you can get away with coercions of various types. If you don't, you'll just make yourself trouble. That's why they make the functions. Your shifting of an entity by zero bits raises some question. If you want the practice, find out how your machine stores floats, then peel it apart and convert it yourself. If you're trying to conform to network byte order as a matter of course, use the standard functions. __________________ Abstraction doesn't make it impossible to write bad code; it makes it possible to write superior code. Contributor's Corner: Grumpy on C++ Exceptions DaWei on Pointers

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Programming Forums > Application Development > C++

byte reversing & streaming