README.md

c_bitfields_integer_promotion

Little experiment to check how integer promotion affects bitfields.

The interest in this came after the discussion at https://lore.kernel.org/all/[email protected]/t/#u

Turns out that for bitfields bigger than sizeof(int) the behavior is compiler dependent.

c_bitfields_integer_promotion/main.c

Lines 1 to 33 in 55cff64

	#include <stdio.h>
	struct s1{
	unsigned int a: 20;
	unsigned int _: 12;
	};
	struct s2 {
	unsigned long long b: 40;
	unsigned long long _: 24;
	};
	int main()
	{
	struct s1 s1;
	struct s2 s2;
	s1.a = 2;
	s2.b = 2;
	unsigned int a = s1.a << 19;
	unsigned long long b = s2.b << 39;
	printf("0x%x\n", a); //both gcc and clang: 0x100000
	printf("0x%llx\n", b); //gcc: 0x0 clang: 0x10000000000
	return 0;
	}
	/*
	Debian clang version 18.1.8
	gcc (Debian 12.2.0-14) 12.2.0
	*/

A bunch of related links

https://gcc.gnu.org/onlinedocs/gcc/Structures-unions-enumerations-and-bit-fields-implementation.html

https://stackoverflow.com/questions/2647320/struct-bitfield-max-size-c99-c

https://www.gnu.org/software/gnu-c-manual/gnu-c-manual.html#Bit-Fields

https://archive.org/details/the-ansi-c-programming-language-by-brian-w.-kernighan-dennis-m.-ritchie.org/page/132/mode/2up

https://stackoverflow.com/questions/32529080/should-bit-fields-less-than-int-in-size-be-the-subject-of-integral-promotion

https://stackoverflow.com/questions/46073295/implicit-type-promotion-rules

https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2310.pdf