https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120095
Bug ID: 120095
Summary: ICE in convert_mode_scalar when casting double
(*)(double) to int (*)(float) and calling it with a
large argument
Product: gcc
Version: 15.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: mario.rodriguezb1 at um dot es
Target Milestone: ---
GCC crashes with an internal compiler error when casting standard math
functions like cos and fabs (with type double (*)(double)) to a mismatched
function pointer type int (*)(float)
```
#include <limits.h>
#include <stdlib.h>
#include <stdio.h>
#include <math.h>
int main(void) {
int (*c)(float);
if (c = (int(*)(float)) cos)
puts("success");
if (c = (int(*)(float)) fabs)
puts("failure");
if (c && c((float) INT_MAX))
abort();
return EXIT_SUCCESS;
}
```
```
<source>:11:12: internal compiler error: in convert_mode_scalar, at expr.cc:336
11 | if (c && c((float) INT_MAX))
| ^~~~~~~~~~~~~~~~~~
0x20bcdd5 diagnostic_context::diagnostic_impl(rich_location*,
diagnostic_metadata const*, diagnostic_option_id, char const*, __va_list_tag
(*) [1], diagnostic_t)
???:0
0x20ce326 internal_error(char const*, ...)
???:0
0x7a56c6 fancy_abort(char const*, int, char const*)
???:0
0xa5111d store_expr(tree_node*, rtx_def*, int, bool, bool)
???:0
```
It happens since 4.9 to last version using any optimization flag, -O1, -O2...
https://gcc.godbolt.org/z/T4aK6vooW
