https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117415
Bug ID: 117415
Summary: Bogus execution count for assignment to *func()
Product: gcc
Version: 15.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: gcov-profile
Assignee: unassigned at gcc dot gnu.org
Reporter: wentaoz5 at illinois dot edu
Target Milestone: ---
I hit the issue when dealing with
https://github.com/apache/httpd/blob/2.4.62/server/mpm_unix.c#L901. After
reduction, I ended up with two variants with slightly different triggering
conditions. See comments in code for details.
How to reproduce:
$ gcc --version
gcc (GCC) 15.0.0 20240923 (experimental)
Copyright (C) 2024 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
$ cat > test.c << 'EOF'
int arr[10];
int *foo() {
return arr;
}
int bar() {
return 0;
}
int main(void) {
int *a = arr;
int zero = 0;
if (zero) {} else {
// Variant 1 triggering conditions:
//
// 1. RHS of assignment is a function call
// 2. The first function name and assignment are put in one line;
// the second function name is put in the other line
// 3. Inside if(){...}, else{...}, for(){...} etc
*foo()=
bar();
}
// Variant 2 triggering conditions:
//
// 1. RHS of assignment is a dereference
// 2. The function name and assignment are put in one line; the second
// dereference is put in the other line
// 3. The operand of dereference is referenced elsewhere
*foo()=
*a;
&a;
return 0;
}
EOF
$ gcc --coverage test.c -o test
$ ./test
$ gcov test
$ cat test.c.gcov
...
2: 22: *foo()=
1: 23: bar();
...
2: 32: *foo()=
1: 33: *a;
...
Line 22 and 32 are reported to executed twice, which is hard to interpret.
