https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121479
Bug ID: 121479
Summary: The return value of __builtin_constant_p() is
inconsistent at different optimization levels
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: njuwy at smail dot nju.edu.cn
Target Milestone: ---
gcc version:
gcc version 16.0.0 20250804 (experimental) (GCC)
Compiler Explorer: https://godbolt.org/z/Gnxfvqvza
src:
#include <math.h>
#include <errno.h>
#include <stdlib.h>
#include <stdio.h>
int foo(double *data, size_t len) {
int status = 0;
double sum = 0.0;
size_t i = 0;
while (i < len) {
double val = data[i];
if (isinf(val) || isnan(val)) {
errno = EDOM;
status = -2;
break;
}
sum += fabs(val);
i++;
}
printf("In foo, __builtin_constant_p(len) = %d\n",
__builtin_constant_p(len));
return status;
}
int main(void) {
double values[] = { -1.5, 2.0, -3.25 };
size_t len = sizeof(values) / sizeof(values[0]);
int result = foo(values, len);
printf("In main, __builtin_constant_p(len) = %d\n",
__builtin_constant_p(len));
return abs(result);
}
issue:
at -O0, -O1, -Os, -O2, -O3
the results of __builtin_constant_p(len) in foo() are：
0,0,0,1,1
the results of __builtin_constant_p(len) in main() are：
0,1,1,1,1
Is this a expected bahavior or miscompilation bug?
