I have a function that generates a CRC check byte based on the content of any packet.The problem is in translating the function from C++ to C#
C++ code:
unsigned char GenerateCheckByte( char* packet, int length, unsigned long seed )
{
if( !packet ) return 0;
unsigned long checksum = 0xFFFFFFFF;
length &= 0x7FFF;
char* ptr = packet;
unsigned long moddedseed = seed << 8;
for( int i = 0; i < length; i++ )
checksum = ( checksum >> 8 ) ^ table[moddedseed + ( ( *(ptr++) ^ checksum ) & 0xFF )];
unsigned char result = ( (checksum>>24)&0xFF ) + ( (checksum>>8)&0xFF ) + ( (checksum>>16)&0xFF ) + ( checksum&0xFF );
return result;
}
the char*(packet) can also be defined as LPBYTE,the idea is that the value assigned to *packet is assigned to *ptr and as you see *ptr increases.Meaning a byte array is passed in and by increasing the value of the pointer it goes to the next byte.
I tried to do it in C# and failed many times.After some hard work I figured out some code,but i can't execute it :?
C# code
public static unsafe byte GenerateCheckByte(byte *packet, int length, UInt32 seed )
{
if (*packet == 0)
return 0;
UInt32 checksum = 0xFFFFFFFF;
length &= 0x7FFF;
byte *ptr = packet;
UInt32 moddedseed = seed << 8;
for (int i = 0; i < length; i++)
checksum = ( checksum >> 8 ) ^ Table.table[moddedseed + ( ( *(ptr++) ^ checksum ) & 0xFF )];
byte result = (byte)(( (checksum>>24)&0xFF ) + ( (checksum>>8)&0xFF ) + ( (checksum>>16)&0xFF ) + ( checksum&0xFF ));
return result;
}
It doesn't look that bad,but I can't call it
unsafe
{
packetBuffer[5] = Functions.GenerateCheckByte(&packetBuffer[0], 18, packet.seedCRC);
}
error: "You can only take the address of an unfixed expression inside of a fixed statement initializer"
Please note
packetbuffer in both C++ and C# application is byte[] packetBuffer = new byte[18];
You could make the method accept a byte array:
public static unsafe byte GenerateCheckByte(byte[] packetArray, int length, UInt32 seed)
{
fixed(byte *packet = packetArray)
{
... etc
}
}
It's better to keep the unsafe stuff hidden away as much as possible behind managed interfaces.
Then calling it would be easy:
packetBuffer[5] = Functions.GenerateCheckByte(packetBuffer, 18, ...
In fact, it would be better to write GenerateCheckByte
to operate on an array anyway, instead of delving into unsafe
techniques:
public static unsafe byte GenerateCheckByte(byte[] packet, int length, UInt32 seed )
{
if (packet == null)
throw new ArgumentNullException("packet"); // the right way in C#
UInt32 checksum = 0xFFFFFFFF;
length &= 0x7FFF;
UInt32 moddedseed = seed << 8;
for (int i = 0; i < length; i++)
checksum = ( checksum >> 8 ) ^ Table.table[moddedseed + ( ( packet[i] ^ checksum ) & 0xFF )];
byte result = (byte)(( (checksum>>24)&0xFF ) + ( (checksum>>8)&0xFF ) + ( (checksum>>16)&0xFF ) + ( checksum&0xFF ));
return result;
}
Write the simplest, safest implementation you can, and only mess with pointers if you find a bottleneck in profiling.
Are you just translating a lot of existing C/C++ into C#? There's little point doing that unless you get some new safety/maintainability from it. :)
You shouldn't have to use unsafe code at all. If you send in a byte array to the function, it can access it without using pointers.
I haven't tested the code, but it should be something like this:
byte GenerateCheckByte(byte[] packet, ulong seed) {
if (packet == null) return 0;
int length = packet.Length & 0x7FFF;
ulong checksum = 0xFFFFFFFF;
ulong moddedseed = seed << 8;
for (int i = 0; i < length; i++) {
checksum = (checksum >> 8) ^ table[moddedseed + ((packet[i] ^ checksum) & 0xFF)];
}
return (byte)(
((checksum >> 24) & 0xFF) +
((checksum >> 16) & 0xFF) +
((checksum >> 8) & 0xFF) +
(checksum & 0xFF)
);
}
You need to 'pin' the byte array into memory to use it as a byte*.
byte checksum;
fixed(byte* pPacketBuffer = packetBuffer)
{
checksum = Functions.GenerateCheckByte(pPacketBuffer, 18, packet.seedCRC)
}
packetBuffer[5] = checksum
References:
User contributions licensed under CC BY-SA 3.0