#include #include #include #include #include #define EIGHTBYTE_BIT (8 * sizeof(uint64_t)) uint8_t eightbyte_get_nibble(const uint64_t source, const uint8_t nibble_index, const bool data_is_big_endian) { return eightbyte_get_bitfield(source, NIBBLE_SIZE * nibble_index, NIBBLE_SIZE, data_is_big_endian); } uint8_t eightbyte_get_byte(uint64_t source, const uint8_t byte_index, const bool data_is_big_endian) { if(data_is_big_endian) { source = __builtin_bswap64(source); } return (source >> (EIGHTBYTE_BIT - ((byte_index + 1) * CHAR_BIT))) & 0xFF; } // TODO is this funciton necessary anymore? is it any faster for uint64_t than // get_bitfield(data[], ...)? is the performance better on a 32 bit platform // like the PIC32? uint64_t eightbyte_get_bitfield(uint64_t source, const uint16_t offset, const uint16_t bit_count, const bool data_is_big_endian) { int startByte = offset / CHAR_BIT; int endByte = (offset + bit_count - 1) / CHAR_BIT; if(!data_is_big_endian) { source = __builtin_bswap64(source); } uint8_t* bytes = (uint8_t*)&source; uint64_t ret = bytes[startByte]; if(startByte != endByte) { // The lowest byte address contains the most significant bit. uint8_t i; for(i = startByte + 1; i <= endByte; i++) { ret = ret << 8; ret = ret | bytes[i]; } } ret >>= 8 - find_end_bit(offset + bit_count); return ret & bitmask(bit_count); } bool eightbyte_set_bitfield(uint64_t* destination, uint64_t value, const uint16_t offset, const uint16_t bit_count) { if(value > bitmask(bit_count)) { return false; } int shiftDistance = EIGHTBYTE_BIT - offset - bit_count; value <<= shiftDistance; *destination &= ~(bitmask(bit_count) << shiftDistance); *destination |= value; return true; }