CODEWITHSUNDEEP
c
serial8
serial8

Reputation: 60

C, small int to larger uint conversion

When converting from a small int (negative value) to a larger uint what should the value be? Example:

int8_t small_int = 0x80;
uint16_t big_uint = (uint16_t)small_int;   // result should be 0xFF80 or 0x0080?

I was expecting 0x80, from the C standard like this: enter image description here

0x80 + 0xFFFF = 0x0079, then add 1 gives 0x0080.

But my simulator gives the result for big_uint as 0xFF80. I tried this for 8 => 16, 16 => 32 and 32 => 64. All come out the same (i.e. sign extended) Is this because the compiler actually does this:

uint16_t big_uint = (uint16_t)((int16_t)small_int);

If so could someone clarify what 6.3.1.3 part 2 actually means with a simple example please?

Upvotes: 1

Views: 300

Answers (2)

Eric Postpischil
Eric Postpischil

Reputation: 224546

int8_t is a two’s complement type, so the bits 10000000 in it represent −128.

For uint16_t, one more than the maximum value that can be represented in uint16_t is one more than 65,535 (hexadecimal FFFF), which is 65,536 (hexadecimal 10000).

−128 plus 65,536 is 65,408, which is FF80 in hexadecimal.

Backing up, though, there is a problem in your first declaration: int8_t small_int = 0x80;. int8_t can represent values from −128 to +127. But 0x80 represents the value 128. It does not fit. When it is used to initialize an int8_t, a conversion is performed. Per C 2018 6.3.1.3 3, the conversion is implementation-defined (when converting an unrepresentable value to a signed integer type). It is common for compilers to wrap modulo the number of values representable in the type (256), but this is not guaranteed by the C standard. Properly, your definition should be int8_t small_int = -128; or int8_t small_int = -0x80;.

Upvotes: 2

gnasher729
gnasher729

Reputation: 52632

The value 0x80, stored in a signed 8 bit int, has a value of -128. So the new value is the old value -128, repeatedly adding 0x10000, giving 0xff80.

There is no intermediate conversion to int16_t. It’s not needed. The adding is done with the real mathematical values.

Upvotes: 1

Related Questions