Bit shifting causing intermittent bit corruption

[sdhunna21]

https://github.com/trinamic/TMC-API/blob/3da0cec86d2f14c3f1b6f29950df6fe2cc41624b/tmc/ic/TMC5160/TMC5160.c#L53

Sometimes when reading a spi value the upper 16 bits are all 1's. I noticed this line of code and changed it like so and the problem was resolved.

Shifting bits with types smaller than the shift can cause undefined behavour. At least on our processor, the msp430 it needs this fix.

	uint32_t v[5] = {0};
	for (uint32_t i = 0; i < 5; i++) {
		v[i] = data[i];
	}

	uint32_t value = (v[1] << 24) | (v[2] << 16) | (v[3] << 8) | (v[4]);
	return value;
	```

[fleutot]

Good catch!

You could also skip the uint32_t array, and the hard-coded shift sizes (untested):

uint32_t value = 0;
for (int i = 5 - 1; i >= 1; i--) {
    value <<= 8;
    value |= data[i];
}
return value;

[noah-GGS]

Same Issue with TMC5160 SPI:

int32_t readRegisterSPI(uint16_t icID, uint8_t address)
{
    uint8_t data[5] = { 0 };

    address = address & TMC5160_ADDRESS_MASK;

    // clear write bit
    data[0] = address;

    // Send the read request
    tmc5160_readWriteSPI(icID, &data[0], sizeof(data));

    // Rewrite address and clear write bit
    data[0] = address;

    // Send another request to receive the read reply
    tmc5160_readWriteSPI(icID, &data[0], sizeof(data));

    return ((uint32_t)data[1] << 24) | ((uint32_t) data[2] << 16) | ( data[3] <<  8) | ( data[4]);
}

should be more like:

    return ((int32_t)data[1] << 24) | 
           ((int32_t)data[2] << 16) | 
           ((int32_t)data[3] << 8) |   // Added cast
           ((int32_t)data[4]);      

please fix this to a more robust function. It took me ages to find out, whats wrong with my tmc5160.

Best regards