https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108893
--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Jonny Grant from comment #0)
>
> void f(const char * const str) __attribute__((access(read_only, 1)));
> void f(const char * const str)
> {
> __builtin_puts(str);
> }
>
> int main()
> {
> const char * a = nullptr;
> f(a);
> }
>
> Tested on trunk in C++
> https://godbolt.org/z/rc8G6bePq
The goldbolt testcase does not match the above testcase. In fact using the
above testcase works and gives us the warning for both -fanalyzer and
-Wnonnull:
```
In function 'void f(const char*)',
inlined from 'int main()' at <source>:10:6:
<source>:4:19: warning: use of NULL where non-null expected [CWE-476]
[-Wanalyzer-null-argument]
4 | __builtin_puts(str);
| ~~~~~~~~~~~~~~^~~~~
'int main()': event 1
|
| 10 | f(a);
| | ^
| | |
| | (1) inlined call to 'f' from 'main'
|
+--> 'void f(const char*)': event 2
|
| 4 | __builtin_puts(str);
| | ~~~~~~~~~~~~~~^~~~~
| | |
| | (2) argument 1 NULL where non-null
expected
|
<built-in>: In function 'int main()':
<built-in>: note: argument 1 of 'int __builtin_puts(const char*)' must be
non-null
In function 'void f(const char*)',
inlined from 'int main()' at <source>:10:6:
<source>:4:19: warning: argument 1 null where non-null expected [-Wnonnull]
<source>:4:19: note: in a call to built-in function 'int __builtin_puts(const
char*)'
```
