https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
--- Comment #8 from Jonathan Wakely ---
I've reported this as https://github.com/itanium-cxx-abi/cxx-abi/issues/119 but
I haven't tried to fix the spec, or fix G++.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
--- Comment #7 from Jonathan Wakely ---
Also, the note in that section of the ABI is wrong:
> (Note: if the usual array deallocation function takes two arguments,
> then it is a member function whose second argument is of type size_t.
That was
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
--- Comment #6 from Jonathan Wakely ---
Yes, I think the ABI needs fixing. In this example Foo has a trivial destructor
and Foo::operator delete[](void*, size_t, align_val_t) does not have two
parameters. According to the ABI, no cookie is needed
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
--- Comment #5 from Jonathan Wakely ---
I wonder if https://itanium-cxx-abi.github.io/cxx-abi/abi.html#array-cookies
needs to be updated for aligned new[] expressions, or if G++ is just not
accounring for them correctly.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
--- Comment #4 from Jonathan Wakely ---
#include
using std::size_t;
struct alignas(32) Foo
{
char x;
void * operator new[ ] (size_t s, std::align_val_t a)
{
void* p = aligned_alloc(static_cast(a), s);
__builtin_pr
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
--- Comment #3 from Jonathan Wakely ---
(In reply to Martin Liška from comment #2)
> I think it's a bug in libstdc++ and one can see it with valgrind:
But there's no error when compiled with clang and libstdc++, so that suggests
the problem is g
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
Martin Liška changed:
What|Removed |Added
CC||jwakely at redhat dot com
Assig
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98798
Martin Liška changed:
What|Removed |Added
Last reconfirmed||2021-01-25
Status|UNCONFIRMED
