_BitInt bit size reflection

The new _BitInt type is awesome! However, I have come across a super annoying issue: its tricky to infer from the type how many bits it actually has. This makes it hard to do things such as get the max or min possible value of a _BitInt. Maybe clang could lead the charge and add a solution for this?

[pinskia]

__builtin_popcountg ((typeof (x))-1) will work for unsigned _BitInt in GCC today and when clang adds __builtin_popcountg, it work then too. For signed it is a little tricky I think.

[jakubjelinek]

In GCC one can use __builtin_popcountg ((type) -1) if type is unsigned (standard unsigned type or bit-precise) and
__builtin_clrsbg ((type) -1) + 1 for signed (standard signed or bit-precise). So if you want a macro that will work with either,

#define bitwidth(x) ((__typeof (x)) -1 < 0 ? __builtin_clrsbg ((__typeof (x)) -1) + 1 : __builtin_popcountg ((__typeof (x)) -1))

or so. One can use (int) __builtin_stdc_count_zeros ((__typeof (x)) 0) instead of the __builtin_popcountg as well.

Here is a solution in C without using any builtins

#include <stddef.h>
#include <stdio.h>

int main() {
    _BitInt(15) test = 5;

    unsigned maxval = 1;
    for (typeof(test) cur = 1; maxval < ~0; cur <<= 1, ++maxval){
        if (cur > (typeof(test))(cur << 1)){
            ++maxval;
            break;
        }
    }
    printf("%d\n", maxval);
}

Pretty bad on the performance though. Probably a better way to do this, and making it consteval would help.

(credits to https://github.com/Tsche)

[nickdesaulniers]

when clang adds __builtin_popcountg, it work then too.

#82058
21d8332
clang-19 should have that builtin.

[llvmbot]

@llvm/issue-subscribers-clang-frontend

Author: Daniel Aimé Valcour (Journeyman1337)

The new _BitInt type is awesome! However, I have come across a super annoying issue: its tricky to infer from the type how many bits it actually has. This makes it hard to do things such as get the max or min possible value of a _BitInt. Maybe clang could lead the charge and add a solution for this?

[frederick-vs-ja]

See also #130429.