How to check data redundancy on float32 variable in C?-CodePudding

I would like to check data corruption of a variable in some function. In general, if it is a unsigned integer variable i can easily check like below method

sender side:

uint8 actual_data = 23;
uint8 send_data = ~actual_data; /*bitwise NOT*/
send_data_to_other(send_data, actual_data);

receiver side:

void receive_data_to_other(uint8 received_data, uint8 actual_data)
{
   uint8 check_data = ~received_data;
   if(check_data == actual_data)
   {
     printf("data is correct");
   }
   else
   {
     printf("something wrong")
   }
}

But suppose if i am using float32 data,

eg: float32 actual_data = 23.22f;

then i can not perform bitwise NOT(~) for float32 variables. Any one Suggest me how can we check the data verification for float32 variable in C?

CodePudding user response：

Assuming sizeof (float) == sizeof (uint32_t):

    uint32_t data;
    uint32_t ndata;

    memcpy(&data, &f, sizeof(data));
    ndata = ~data;

But of course, your send function will need to send larger chunks of data.

To "reconstruct" float you will need to copy data in the opposite direction.

float toFloat(uint32_t data)
{
    float f;

    memcpy(&f, &data, sizeof(f));
    return f;
}

Do not worry about memcpy overhead as this function is very well known to the compiler and optimizing compilers will optimize out the call.

CodePudding user response：

The binary representation of floating point numbers uses scientific notation instead of normal binary.

See this https://www.doc.ic.ac.uk/~eedwards/compsys/float/