implementation of std::is_polymorphic does this:
template <class _Tp> struct __is_polymorphic1 : public _Tp {};
... and that g++ rejects this if _Tp has an inaccessible virtual destructor
(because __is_polymorphic1<_Tp> would have a deleted virtual destructor
overriding _Tp's non-deleted destructor). Clang was failing to reject this;
I've fixed that in r178563, but that causes libc++'s corresponding test
case to fail with both clang and gcc when using the fallback
implementation. The fallback code also incorrectly rejects final types.
The attached patch fixes the fallback implementation of is_polymorphic; we
now use dynamic_cast's detection of polymorphic class types rather than
trying to determine if adding a virtual function makes the type larger:
enable_if<sizeof((_Tp*)dynamic_cast<const volatile
void*>(declval<_Tp*>())) != 0, ...>
Two things of note here:
* the (_Tp*) cast is necessary to work around bugs in Clang and g++ where
we otherwise don't instantiate the dynamic_cast (filed as PR15656)
* the 'const volatile' is here to treat is_polymorphic<cv T> as true for a
polymorphic class type T -- my reading of the standard suggests this is
incorrect, but it matches our builtin __is_polymorphic and gcc
llvm-svn: 178576
- As of this commit, the test suite should now fully pass on both darwin11 and
darwin12 when testing against either a locally built libc++ or the system libc++.
llvm-svn: 174478
integers which remain unused and are subsequently leaked, so the test
fail when run under valgrind. Unless I'm overlooking a subtle reason
why they are needed I think they can be removed, allowing these tests
to pass under valgrind. The attached patch removes the variables. If
there is a reason for them to exist, I can change this to just delete
them at the end of the test.
llvm-svn: 161195
section in libc++. This requires a recompiled dylib. Failure to rebuild
the dylib will result in a link-time error if and only if the functions from
[util.smartptr.shared.atomic] are used.
The implementation is not lock free. After considerable thought, I know of no
way to make the implementation lock free. Ideas welcome along that front. But
changing the ABI of shared_ptr is not on the table at this point.
The mutex used to lock these function is encapsulated by std::__sp_mut. The
only thing the client knows about std::__sp_mut is that it has a void* data
member, can't be constructed, and has lock and unlock members. Within the
binary __sp_mut is currently implemented as a pointer to a std::mutex. That can
change in the future without disturbing the ABI (as long as sizeof(__sp_mut)
remains constant.
I specifically did not make __sp_mut a spin lock as I have a pathological
distrust of spin locks. Testing on OS X reveals that the use of std::mutex in
this role is not a large performance penalty as long as the contention for the
mutex is low (more likely to get the lock than to have to wait). In the future
we can still make __sp_mut a spin lock if that is what is desired (without ABI
damage).
The dylib contains 16 __sp_mut's to be chosen based on the hash of the address
of the shared_ptr. The constant 16 is a ball-park reasonable space/time
tradeoff.
std::hash<T*> was changed to call __murmur2_or_cityhash, instead of the identity
function. I had thought we had already done this, but I was mistaken.
All of this is under #if __has_feature(cxx_atomic) even though the
implementation is not lock free, because the signatures require access to
std::memory_order, which is currently available only under
__has_feature(cxx_atomic).
llvm-svn: 160940
between C++03 and C++0x and its effect on exceptions, and another two to
not test move construction when rvalue references are not available.
llvm-svn: 135445
a cv-qualifier rvalue reference to the type, e.g.,
template <class _Tp> char __test(const volatile typename remove_reference<_Tp>::type&&);
The use of this function signature rather than the more
straightforward one used in C++98/03 mode, e.g.,
template <class _Tp> char __test(_Tp);
is broken in two ways:
1) An rvalue reference cannot bind to lvalues, so is_convertible<X&,
X&>::value would be false. This breaks two of the unique_ptr tests
on Clang and GCC >= 4.5. Prior GCC's seem to have allowed rvalue
references to bind to lvalues, allowing this bug to slip in.
2) By adding cv-qualifiers to the type we're converting to, we get
some incorrect "true" results for, e.g., is_convertible<const X&, X&>::value.
llvm-svn: 124166
Marshall Clow