Unexpected results when creating 32-bit int from byte array

Question

I'm having problems creating a 32-bit integer from a 4-byte array. In the following complete, minimal, and verifiable example, converting a byte array containing 0x00, 0x04, 0x04F, 0x45 results in 4F45 (expected result would be 44F45). I must be overlooking something fundamental here.... can anyone see where I went wrong?

void setup()
{
 Serial.begin(9600);
 while(!Serial);
}
void loop()
{
 uint8_t bytes[4] = {0x00, 0x04, 0x4F, 0x45}; //00044F45
 int32_t theInt = byteArrayToInt32(bytes);
 Serial.println(theInt, HEX); //Print 4F45
 delay(250);
}
uint32_t byteArrayToInt32(uint8_t* bytes)
{
 uint32_t i = 0;
 i |= bytes[0] << 24;
 i |= bytes[1] << 16;
 i |= bytes[2] << 8;
 i |= bytes[3];
 return i;
}

• You'reAGitForNotUsingGit: @jsotola in what way?
• You'reAGitForNotUsingGit: @jsotola I went ahead re-tested with it as uint32_t, no difference :(

Answer 1

The problem is that you are trying to shift uint8_t more than 8 bits and then OR it with int32_t.

You have to cast the uint8_t to int32_t before doing the shift.

void setup()
{
 Serial.begin(9600);
 while(!Serial);
}
void loop()
{
 uint8_t bytes[4] = { 0x00, 0x04, 0x4F, 0x45}; //00044F45
 int32_t theInt = byteArrayToInt32(bytes);
 Serial.println(theInt, HEX); //Print 4F45
 delay(1000);
}
int32_t byteArrayToInt32(uint8_t* bytes)
{
 int32_t i = 0;
 i |= (int32_t) bytes[0] << 24;
 i |= (int32_t) bytes[1] << 16;
 i |= (int32_t) bytes[2] << 8;
 i |= (int32_t) bytes[3];
 return i;
}

• Edgar Bonet: Note that if the most significant bit of bytes[0] is set, then (int32_t) bytes[0] << 24 will shift it into the sign bit, which is undefined behavior. It is safer to cast to a uint32_t, which can in turn be safely assigned to an int32_t.
• You'reAGitForNotUsingGit: Hmm, so doing as you suggested fixes the problem, however I'm very perplexed as to why the left shift by 8 bits worked originally, then?
• jsotola: it appears that the left shift function uses a 16 bit buffer .... possibly because it uses the same code for 16 bit shifts