[Bug middle-end/87296] [8/9/10 Regression] -Wstringop-overflow false positive due to using MEM_REF type of &MEM

[Bug middle-end/87296] [8/9/10 Regression] -Wstringop-overflow false positive due to using MEM_REF type of &MEM

[law at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87296

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sbergman at redhat dot com

--- Comment #8 from Jeffrey A. Law <law at redhat dot com> ---
*** Bug 92893 has been marked as a duplicate of this bug. ***

[Bug middle-end/87296] [8/9/10/11 Regression] -Wstringop-overflow false positive due to using MEM_REF type of &MEM

[mostlyaspambox at protonmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87296

mostlyaspambox at protonmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mostlyaspambox at protonmail dot c
                   |                            |om

--- Comment #9 from mostlyaspambox at protonmail dot com ---
I think I've also hit this or a variation of it with the following reproducer:

```
#include <string.h>
#include <algorithm>
#include <assert.h>
#include <array>

struct Base1
{
    char arr[5];
};

struct Base2
{
    char arr[25];
};

struct A : Base1, Base2
{
};

extern const char* str;

void f(A& a)
{
    Base1& base1 = a;
    std::fill(std::begin(base1.arr), std::end(base1.arr), ' ');
    Base2& base2 = a;
    strncpy(reinterpret_cast<char*>(&base2), str, 6);
}

int main()
{
    A a{};
    f(a);
}
```

If `a` is created inside `f` instead of passed in by reference, there is no
error.

When compiled with `g++ -std=c++11 -O2 -Wall -Wextra -fno-strict-aliasing
-fwrapv -fno-aggressive-loop-optimizations -fsanitize=undefined` on x86-64 with
gcc 8.1 this prints:

test.cpp:27:12: warning: 'char* strncpy(char*, const char*, size_t)' writing 6
bytes into a region of size 5 overflows the destination [-Wstringop-overflow=]

     strncpy(reinterpret_cast<char*>(&base2), str, 6);

     ~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

If `str` in the strncpy call is replaced with a string literal, there is the
same warning unless the literal has length 6 including the null terminator.

[Bug middle-end/87296] [8/9/10/11 Regression] -Wstringop-overflow false positive due to using MEM_REF type of &MEM

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87296

Martin Sebor <msebor at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
   Target Milestone|8.5                         |11.0
             Status|NEW                         |RESOLVED
             Blocks|                            |88443
           Assignee|unassigned at gcc dot gnu.org      |msebor at gcc dot gnu.org

--- Comment #10 from Martin Sebor <msebor at gcc dot gnu.org> ---
GCC 11 no longer warns for these test cases, most likely as a result of
r11-3827.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88443
[Bug 88443] [meta-bug] bogus/missing -Wstringop-overflow warnings

[Bug middle-end/87296] [8/9/10/11 Regression] -Wstringop-overflow false positive due to using MEM_REF type of &MEM

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87296

--- Comment #11 from Martin Sebor <msebor at gcc dot gnu.org> ---
I should note: The changes are too intrusive to backport.