[Bug c++/65062] New: [4.9 Regression] Miscompilation: invalid C++ mangled name for global variable

[Bug c++/65062] New: [4.9 Regression] Miscompilation: invalid C++ mangled name for global variable

[thiago at kde dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65062

            Bug ID: 65062
           Summary: [4.9 Regression] Miscompilation: invalid C++ mangled
                    name for global variable
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: thiago at kde dot org

Code:

template <class>
void tf()
{
  extern void (*qt_memfill32)();
  qt_memfill32();
}

void f() {
  tf<int>();
}

When compiled with GCC 5.0, the object file has a reference to
"_Z12qt_memfill32", which doesn't seem to me to be a valid C++ mangled name. At
the very least, it's an ABI compatibility break with previous versions of GCC,
other compilers and with itself.

This symbol is defined in another .o as "qt_memfill32", even when compiled with
GCC 5.0.

Tests:

gcc-5.0 -O2 -S -o - test.cpp:
        jmp     *_Z12qt_memfill32(%rip)

gcc-4.9 -O2 -S -o - test.cpp:
        movq    qt_memfill32(%rip), %rax
        jmp     *%rax

clang-3.6 -O2 -S -o - test.cpp:
        jmpq    *qt_memfill32(%rip)  # TAILCALL

icpc -O2 -S -o - test.cpp:
        movq      qt_memfill32(%rip), %rax
        jmp       *%rax

[Bug c++/65062] [4.9 Regression] Miscompilation: invalid C++ mangled name for global variable

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65062

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |DUPLICATE

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
it was just fixed yesterday, see bug 64898.  Please check the latest GCC before
filing a bug especially during stage 4.

*** This bug has been marked as a duplicate of bug 64898 ***

[Bug c++/65062] [4.9 Regression] Miscompilation: invalid C++ mangled name for global variable

[thiago at kde dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65062

--- Comment #2 from Thiago Macieira <thiago at kde dot org> ---
(In reply to Andrew Pinski from comment #1)
> it was just fixed yesterday, see bug 64898.  Please check the latest GCC
> before filing a bug especially during stage 4.
> 
> *** This bug has been marked as a duplicate of bug 64898 ***

Sorry, I did rebuild two days ago when this started showing up. I must have
missed the fix by a few hours, more or less (again).