The test class 'G' reads and writes to the same static variables in its
constructor, destructor and call operator. When threads are
constructed using `std::thread t((G()))` there is a race condition between the
destruction of the temporary and the execution of `G::operator()()`.
The fix is to simply create the input before creating the thread.
llvm-svn: 233946
Summary:
The summary of the bug, provided by Stephan T. Lavavej:
In shared_timed_mutex::try_lock_until() (line 195 in 3.6.0), you need to deliver a notification. The scenario is:
* There are N threads holding the shared lock.
* One thread calls try_lock_until() to attempt to acquire the exclusive lock. It sets the "I want to write" bool/bit, then waits for the N readers to drain away.
* K more threads attempt to acquire the shared lock, but they notice that someone said "I want to write", so they block on a condition_variable.
* At least one of the N readers is stubborn and doesn't release the shared lock.
* The wannabe-writer times out, gives up, and unsets the "I want to write" bool/bit.
At this point, a notification (it needs to be notify_all) must be delivered to the condition_variable that the K wannabe-readers are waiting on. Otherwise, they can block forever without waking up.
Reviewers: mclow.lists, jyasskin
Reviewed By: jyasskin
Subscribers: jyasskin, cfe-commits
Differential Revision: http://reviews.llvm.org/D8796
llvm-svn: 233944
Summary:
In certain cases vector can use memcpy to construct a range of elements at the back of the vector. We currently don't do this resulting in terrible code gen in non-optimized mode and a
very large slowdown compared to libstdc++.
This patch adds a `__construct_forward_range(Allocator, Iter, Iter, _Ptr&)` and `__construct_forward_range(Allocator, Tp*, Tp*, Tp*&)` functions to `allocator_traits` which act similarly to the existing `__construct_forward(...)` functions.
This patch also changes vectors `__construct_at_end(Iter, Iter)` to be `__construct_at_end(Iter, Iter, SizeType)` where SizeType is the size of the range. `__construct_at_end(Iter, Iter, SizeType)` now calls `allocator_traits<Tp>::__construct_forward_range(...)`.
This patch is based off the design of `__swap_out_circular_buffer(...)` which uses `allocator_traits<Tp>::__construct_forward(...)`.
On my machine this code performs 4x better than the current implementation when tested against `std::vector<int>`.
Reviewers: howard.hinnant, titus, kcc, mclow.lists
Reviewed By: mclow.lists
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D8109
llvm-svn: 233711
Summary:
This patch adds configuration to CMake and LIT for running the libc++ test-suite to generate code coverage.
To use code coverage use following instructions.
* Find the clang resource dir using `$CXX -print-search-dirs`. Let <library-dir> be the first library search directory.
* `cmake <regular-options> -DLIBCXX_GENERATE_COVERAGE=ON -DLIBCXX_COVERAGE_LIBRARY=<library-dir>/lib/<platform>/libclang_rt.profile.a <source>`
* `make cxx`
* `make check-libcxx`
* `make generate-libcxx-coverage`
The reason I want this patch upstreamed is so I can setup a bot that generates code coverage and posts in online for every revision.
Reviewers: mclow.lists, jroelofs, danalbert
Reviewed By: danalbert
Differential Revision: http://reviews.llvm.org/D8716
llvm-svn: 233669
Summary:
Currently the conversion check does not take place in a context where access control SFINAE is applied. This patch changes the context of the test expression so that SFINAE occurs if access control does not permit the conversion.
Related bug: https://llvm.org/bugs/show_bug.cgi?id=22771
Reviewers: mclow.lists, rsmith, dim
Reviewed By: dim
Subscribers: dim, rodrigc, emaste, cfe-commits
Differential Revision: http://reviews.llvm.org/D8461
llvm-svn: 233552
The idea behind Nuxi CloudABI is that it is targeted at (but not limited to)
running networked services in a sandboxed environment. The model behind stdin,
stdout and stderr is strongly focused on interactive tools in a command shell.
CloudABI does not support the notion of stdin and stdout, as 'standard
input/output' does not apply to services. The concept of stderr does makes
sense though, as services do need some mechanism to log error messages in a
uniform way.
This patch extends libc++ in such a way that std::cin and std::cout and the
associated <cstdio>/<cwchar> functions can be disabled through the flags
_LIBCPP_HAS_NO_STDIN and _LIBCPP_HAS_NO_STDOUT, respectively. At the same time
it attempts to clean up src/iostream.cpp a bit. Instead of using a single array
of mbstate_t objects and hardcoding the array indices, it creates separate
objects that declared next to the iostream objects and their buffers. The code
is also restructured by interleaving the construction and setup of c* and wc*
objects. That way it is more obvious that this is done identically.
The c* and wc* objects already have separate unit tests. Make use of this fact
by adding XFAILs in case libcpp-has-no-std* is set. That way the tests work in
both directions. If stdin or stdout is disabled, these tests will therefore
test for the absence of c* and wc*.
Differential Revision: http://reviews.llvm.org/D8340
llvm-svn: 233275
If we want to add support for making std::cin and std::cout optional, it
is impractical to have all of the mbstate_t objects in one array. This
would mean that if std::cin and std::cout are omitted, the state_types
array is only used partially.
Solve this by using separate global variables. These are placed right
next to the iostream object and the buffer, meaning we can easily #ifdef
them away.
Differential Revision: http://reviews.llvm.org/D8359
llvm-svn: 233274
The time_put test doesn't seem to work on Linux and CloudABI. For Linux
we already have an XFAIL. Closer inspection seems to reveal that this
test does not pass for a couple of reasons.
First of all, the tm_yday field is set to an invalid value. The
strftime() function doesn't behave consistently across platforms in case
the values in the tm structure are incoherent. Fix up this field to have
the value 121, which corresponds with tm_mday, tm_mon and tm_year. This
of course affects the output of time_put for some modifiers, so update
the tests accordingly.
Second, some of the tests actually use modifiers that are only present
on BSD derived systems. They are not part of the C standard/POSIX.
Simply remove them.
Finally, some of the tests actually use invalid modifiers, causing a
malformed format string to be passed to strftime(). Remove these tests
as well.
Differential Revision: http://reviews.llvm.org/D8349
llvm-svn: 233262
Summary:
Add symbol checking scripts for extracting a list of symbols from shared libraries and for comparing symbol lists for differences.
Reviewers: mclow.lists, danalbert, EricWF
Reviewed By: EricWF
Subscribers: majnemer, emaste, cfe-commits
Differential Revision: http://reviews.llvm.org/D4946
llvm-svn: 232855
Summary: This patch also fixes one test case that failed in the library version of is_convertible.
Reviewers: mclow.lists, EricWF
Reviewed By: EricWF
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D8456
llvm-svn: 232764
Summary:
Clean up all the different possible CMake options for specifying the ABI include paths into one CMake option named `LIBCXX_CXX_ABI_INCLUDE_PATHS`.
The documentation has been updated to reflect this change.
For the next week I have added explicit errors if any of the old flags is used. These errors inform users of the change and the new option to use.
Before committing the change I will announce this change on cfe-dev.
Reviewers: danalbert, mclow.lists
Reviewed By: danalbert, mclow.lists
Subscribers: jroelofs, cbergstrom, cfe-commits
Differential Revision: http://reviews.llvm.org/D5039
llvm-svn: 232762
Summary:
This patch changes std::function to use allocator_traits to rebind the allocator instead of allocator itself.
It also changes most of the tests to use `bare_allocator` where possible instead of `test_allocator`.
Reviewers: mclow.lists
Reviewed By: mclow.lists
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D8391
llvm-svn: 232686
1) <cstdlib> header should define std::abs([int|long|long long])
functions. They use "using ::abs" to import these functions (which are
declared in <stdlib.h>) into std namespace.
2) <cmath> header should define std::abs([float|double|long double])
function. If we try define new functions in std namespace, then it
will cause compile error in <cstdlib> because "using ::abs" will try
import not only [int|long|long long] functions, but also
[float|double|long double] which are defined in <math.h> header on
solaris.
Patch by C Bergstrom.
llvm-svn: 232641
Interleave the code for narrow and wide character streams. This makes it
more obvious that the two pieces of code are identical. Furthermore, it
makes it easier to conditionally compile support for certain streams, as
less #ifdef blocks are needed.
Differential Revision: http://reviews.llvm.org/D8342
Reviewed by: marshall
llvm-svn: 232516
Summary:
This patch adds the `<experimental/tuple>` header (almost) as specified in the latest draft of the library fundamentals TS.
The main changes in this patch are:
1. Added variable template `tuple_size_v`
2. Added function `apply(Func &&, Tuple &&)`.
3. Changed `__invoke` to be `_LIBCPP_CONSTEXPR_AFTER_CXX11`.
The `apply(...)` implementation uses `__invoke` to invoke the given function. `__invoke` already provides the required functionality. Using `__invoke` also allows `apply` to be used on pointers to member function/objects as an extension. In order to facilitate this `__invoke` has to be marked `constexpr`.
Test Plan:
Each new feature was tested.
The test cases for `tuple_size_v` are as follows:
1. tuple_size_v.pass.cpp
- Check `tuple_size_v` on cv qualified tuples, pairs and arrays.
2. tuple_size_v.fail.cpp
- Test on reference type.
3. tuple_size_v_2.fail.cpp
- Test on non-tuple
4. tuple_size_v_3.fail.cpp
- Test on pointer type.
The test cases for tuple.apply are as follows:
1. arg_type.pass.cpp
- Ensure that ref/pointer/cv qualified types are properly passed.
2. constexpr_types.pass.cpp
- Ensure constexpr evaluation of apply is possible for `tuple` and `pair`.
3. extended_types.pass.cpp
- Test apply on function types permitted by extension.
4. large_arity.pass.cpp
- Test that apply can evaluated on tuples and arrays with large sizes.
5. ref_qualifiers.pass.cpp
- Test that apply respects ref qualified functions.
6. return_type.pass.cpp
- Test that apply returns the proper type.
7. types.pass.cpp
- Test apply on function types as required by LFTS.
Reviewers: mclow.lists
Reviewed By: mclow.lists
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D4512
llvm-svn: 232515
Summary:
There is no reason to guard `tuple_size`, `tuple_element` and `get<I>(...)` for pair and array inside of `<__tuple>` so that they are only available when we have variadic templates.
This requires there be redundant declarations and definitions. It also makes it easy to get things wrong.
For example the following code should compile (and does in c++11).
```
#define _LIBCPP_HAS_NO_VARIADICS
#include <array>
int main()
{
static_assert((std::tuple_size<std::array<int, 10> volatile>::value == 10), "");
}
```
This patch lifts the non-variadic parts of `tuple_size`, `tuple_types`, and `get<I>(...)` to the top of `<__tuple>` where they don't require variadic templates. This patch also removes `<__tuple_03>` because there is no longer a need for it.
Reviewers: danalbert, K-ballo, mclow.lists
Reviewed By: mclow.lists
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D7774
llvm-svn: 232492
In one of the ostream tests we attempt to validate whether the output of
%p is correct. This is actually outside the scope of libc++, for the
%reason that the format of %p is implementation defined. Change the test
%to validate that the output of %p is non-empty and is different when
%given two unequal addresses.
Differential Revision: http://reviews.llvm.org/D8354
Reviewed by: marshall
llvm-svn: 232390
We already have a definition for the Czech locale name in
platform_support.h. Use this one instead.
While there, respect the common format of the tests. For most other
tests it's the case that test_iterators.h is placed right underneath the
other #includes (without an empty line). platform_support.h is included
after an empty line.
llvm-svn: 232383
According to POSIX, *abs() and *div() are allowed to be macros (in
addition to being functions). Make sure we undefine these, so that
std::*abs() and std::*div() work as expected.
llvm-svn: 232379
The rest of the test uses the #defines for the locale names properly. In
this single spot we do hardcode the string. This causes this test to
fail on CloudABI, where this locale is called en_US.UTF-8@UTC.
llvm-svn: 232365
Though common, there is no requirement that fenv_t and fexcept_t are
structure and integer types, respectively. fexcept_t is a structure on
CloudABI.
llvm-svn: 232329
Systems like FreeBSD's Capsicum and Nuxi CloudABI apply the concept of
capability-based security on the way processes can interact with the
filesystem API. It is no longer possible to interact with the VFS
through calls like open(), unlink(), rename(), etc. Instead, processes
are only allowed to interact with files and directories to which they
have been granted access. The *at() functions can be used for this
purpose.
This change adds a new config switch called
_LIBCPP_HAS_NO_GLOBAL_FILESYSTEM_NAMESPACE. If set, all functionality
that requires the global filesystem namespace will be disabled. More
concretely:
- fstream's open() function will be removed.
- cstdio will no longer pull in fopen(), rename(), etc.
- The test suite's get_temp_file_name() will be removed. This will cause
all tests that use the global filesystem namespace to break, but will
at least make all the other tests run (as get_temp_file_name will not
build anyway).
It is important to mention that this change will make fstream rather
useless on those systems for now. Still, I'd rather not have fstream
disabled entirely, as it is of course possible to come up with an
extension for fstream that would allow access to local filesystem
namespaces (e.g., by adding an openat() member function).
Differential revision: http://reviews.llvm.org/D8194
Reviewed by: jroelofs (thanks!)
llvm-svn: 232049
This basically reverts the revert in r216508, and fixes a few more cases while
I'm at it. Reading my commit message on that commit again, I think it's bupkis.
http://reviews.llvm.org/D8237
llvm-svn: 231940
On a new platform that I am working on
(https://github.com/NuxiNL/cloudlibc) I am not implementing the
cat{open,close,gets}() API, just like Android, Newlib, etc.
Instead of adding yet another operating system name to the #ifs,
introduce _LIBCPP_HAS_CATOPEN in include/__config. Also adjust the code
to only pull in nl_types.h when _LIBCPP_HAS_CATOPEN is set. We only
needed this header for the cat*() API.
Differential Revision: http://reviews.llvm.org/D8163
Reviewed by: marshall
llvm-svn: 231937
Summary:
The data files for any given test will be in the same directory as the
source with a file name that matches *.dat. To make these available to
tests running remotely (such as over adb or ssh), copy them into the
test's remote working directory.
Note that we will perform more copies than we actually need. The data
files in the directory may only be used by one of the tests, but will
be copied for all tests in the same directory.
This patch also moves the remote test binary into the working
directory (previously it was only invoked from the working directory
rather than existing in it).
Reviewers: EricWF, jroelofs
Reviewed By: jroelofs
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D8118
llvm-svn: 231864
Before I discovered that NetBSD provides a permanent handle to the C
locale called LC_C_LOCALE, I also added support for this to CloudABI
under the name LC_POSIX_LOCALE. I've renamed it to LC_C_LOCALE to
improve compatibility.
llvm-svn: 231780
CloudABI provides the _l() functions that are part of POSIX.1-2008, but
also the extensions that are available on systems like OS X and *BSD
(scanf_l, printf_l, etc).
llvm-svn: 231777
As CloudABI does not provide sysctl(), this header is not present. Make
thread.cpp build correctly (and pass all tests) by not including the header.
llvm-svn: 231768
There are a couple of places where libc++ prints log/error messages to
stdout on its own. This may of course interfere with the output
generated with applications. Log/error messages should be directed to
stderr instead.
Differential Revision: http://reviews.llvm.org/D8135
Reviewed by: marshall
llvm-svn: 231767
Nuxi CloudABI (https://github.com/NuxiNL/cloudlibc) does not allow
processes to access the global filesystem namespace. This breaks
random_device, as it attempts to use /dev/{u,}random. This change adds
support for arc4random(), which is present on CloudABI.
In my opinion it would also make sense to use arc4random() on other
operating systems, such as *BSD and Mac OS X, but I'd rather leave that
to the maintainers of the respective platforms. Switching to
arc4random() does change the ABI.
This change also attempts to make some cleanups to the code. It adds a
single #define for every random interface, instead of testing against
operating systems explicitly.
As discussed, also validate the token argument to be equal to
"/dev/urandom" on all systems that only provide pseudo-random numbers.
This should cause little to no breakage, as "/dev/urandom" is also the
default argument value.
Reviewed by: jfb
Differential Revision: http://reviews.llvm.org/D8134
llvm-svn: 231764
Summary: Fix suggested by @mclow.lists on D8109. Store the size of the un-poisoned vector upon construction instead of calculating it later.
Reviewers: titus, mclow.lists, kcc, EricWF
Reviewed By: EricWF
Subscribers: mclow.lists, cfe-commits
Differential Revision: http://reviews.llvm.org/D8172
llvm-svn: 231729
The _Pp typedef in __tree<_Tp, _Compare, _Allocator>::__count_multi()
isn't used anywhere, so adding _LIBCPP_UNUSED is unecessary.
Differential Revision: http://reviews.llvm.org/D8140
llvm-svn: 231705
On CloudABI we should append the timezone name to the end of the locale
(e.g., nl_NL.UTF-8@Europe/Amsterdam). By fixing the locale names we can
already let a lot of locale related tests pass.
llvm-svn: 231649
The PrefixExecutor wasn't passing the exe_path down the chain, so the
command was overriding that, the work_dir was being passed as the
command, and so on.
I've cleaned up a few pylint issues while I was here.
llvm-svn: 231496
Summary:
This patch add the CMake option `LIBCXX_ENABLE_STATIC_ABI_LIBRARY` which, when enabled, will link libc++ against the static version of the ABI library.
Reviewers: mclow.lists, jroelofs, danalbert
Reviewed By: danalbert
Subscribers: compnerd, cfe-commits
Differential Revision: http://reviews.llvm.org/D8017
llvm-svn: 231076
MSVCRT 12.0 introduces better compatibility for C99. This includes a number of
math routines that were previously undefined. Use the crtversion.h header to
detect the version of MSVCRT being targeted and avoid re-declaring the
variables.
Since copysign has been introduced in MSVCRT, importing the definition via using
makes it difficult to provide overloads (due to minor differences between
throw () and noexcept. Avoid defining the overloads on newer MSVCRT
targets.
llvm-svn: 230867
Executors can be specified at configure time by using the -DLIBCXX_EXECUTOR=""
option. Examples include:
$ cmake <other_flags> -DLIBCXX_EXECUTOR="TimeoutExecutor(30,LocalExecutor())"
This runs individual tests with a maximum duration
$ cmake <other_flags> -DLIBCXX_EXECUTOR="SSHExecutor('hostname','username')"
This runs tests on a remote target, using scp to shuttle binaries to the
target, and ssh to invoke commands there.
$ cmake <other_flags> -DLIBCXX_EXECUTOR="PrefixExecutor('/path/to/run/script',LocalExecutor())"
This assumes the script knows how to copy run the executables passed to it,
and allows for the ultimate control. This is useful for running things
inside emulators like Valgrind & QEMU.
TODO: This doesn't claim to support ShTest tests yet, that will take a bit more
thought & finagling (I'm still not sure how to orchestrate copy-in for those cases.
I've also punted on what to do about tests that read data files. The testsuite
has several tests that need to read *.dat files placed next to them, and
currently those aren't copied over when using, say, an SSHExecutor. The
affected tests are:
libc++ :: std/input.output/file.streams/fstreams/filebuf.virtuals/pbackfail.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/filebuf.virtuals/underflow.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.assign/member_swap.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.assign/move_assign.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.assign/nonmember_swap.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.cons/move.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.cons/pointer.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.cons/string.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.members/close.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.members/open_pointer.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.members/open_string.pass.cpp
libc++ :: std/input.output/file.streams/fstreams/ifstream.members/rdbuf.pass.cpp
libc++ :: std/localization/locales/locale.convenience/conversions/conversions.buffer/pbackfail.pass.cpp
libc++ :: std/localization/locales/locale.convenience/conversions/conversions.buffer/underflow.pass.cpp
Note: One thing to watch out for when using the SSHExecutor for cross-testing is
that you'll also want to specify a TargetInfo object (so that the host's
features aren't used for available-features checks and flags setup).
http://reviews.llvm.org/D7380
llvm-svn: 230592
Summary: Newlib supports ctype differently from other platforms, this patch teaches libc++ about yet another platform that does ctype differently.
Reviewers: jroelofs
Subscribers: cfe-commits, danalbert, EricWF, jvoung, jfb, mclow.lists
Differential Revision: http://reviews.llvm.org/D7888
llvm-svn: 230557
Summary:
GCC emits a pretty amusing warning when there are apostrophes in a #warning:
```warning: invalid suffix on literal; C++11 requires a space between literal and identifier [-Wliteral-suffix]```
Reword the warning to avoid this, and be more consistent with other warnings in libc++.
Reviewers: danalbert
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7818
llvm-svn: 230298
Summary:
Currently parts of the SFINAE on tuples default constructor always gets evaluated even when the default constructor is never called or instantiated. This can cause a hard compile error when a tuple is created with types that do not have a default constructor. Below is a self contained example using a pair like class. This code will not compile but probably should.
```
#include <type_traits>
template <class T>
struct IllFormedDefaultImp {
IllFormedDefaultImp(T x) : value(x) {}
constexpr IllFormedDefaultImp() {}
T value;
};
typedef IllFormedDefaultImp<int &> IllFormedDefault;
template <class T, class U>
struct pair
{
template <bool Dummy = true,
class = typename std::enable_if<
std::is_default_constructible<T>::value
&& std::is_default_constructible<U>::value
&& Dummy>::type
>
constexpr pair() : first(), second() {}
pair(T const & t, U const & u) : first(t), second(u) {}
T first;
U second;
};
int main()
{
int x = 1;
IllFormedDefault v(x);
pair<IllFormedDefault, IllFormedDefault> p(v, v);
}
```
One way to fix this is to use `Dummy` in a more involved way in the constructor SFINAE. The following patch fixes these sorts of hard compile errors for tuple.
Reviewers: mclow.lists, rsmith, K-ballo, EricWF
Reviewed By: EricWF
Subscribers: ldionne, cfe-commits
Differential Revision: http://reviews.llvm.org/D7569
llvm-svn: 230120
Summary:
libc++abi2.exp should be used whenever `cxxabi.h` defines `_LIBCPPABI_VERSION`. This macro was added to libc++abi in 2012 in r149632. For this reason we should use libc++abi2.exp as default unless otherwise specified.
Also when building against an in-tree libc++abi we definitely want to use libc++abi2.exp.
I would love to know what OSX was the last to use libc++abi.exp but I can only test on 10.9.
Reviewers: danalbert, mclow.lists, EricWF
Reviewed By: EricWF
Subscribers: meadori, cfe-commits
Differential Revision: http://reviews.llvm.org/D7773
llvm-svn: 230119
Summary: No declaration for the type `tuple` is given in c++03 or c++98 modes. Mark all tests that use the actual `tuple` type as UNSUPPORTED.
Reviewers: jroelofs, mclow.lists, danalbert
Reviewed By: danalbert
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D5956
llvm-svn: 229808
Summary:
This patch adds a lit option to enable color diagnostics when either `--param=color_diagnostics` is passed to LIT or `LIBCXX_COLOR_DIAGNOSTICS` is present in the environment.
My only concern with this patch is that GCC and Clang take different flags and that only GCC 4.9 and greater support `-fdiagnostics-color=always`
Does anybody have objections to this going in?
Reviewers: jroelofs, danalbert
Reviewed By: danalbert
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D7729
llvm-svn: 229707
Summary:
This patch introduces some black magic to detect const and volatile qualified function types such as `void () const`.
The patch works in the following way:
We first rule out any type that satisfies on of the following. These restrictions are important so that the test below works properly.
* `is_class<_Tp>::value`
* `is_union<_Tp>::value`
* `is_void<_Tp>::value`
* `is_reference<_Tp>::value`
* `__is_nullptr_t<_Tp>::value`
If none of the above is true we perform overload resolution on `__source<_Tp>(0)` to determine the return type.
* If `_Tp&` is well-formed we select `_Tp& __source(int)`. `_Tp&` is only ill formed for cv void types and cv/ref qualified function types.
* Otherwise we select `__dummy_type __source(...)`. Since we know `_Tp` cannot be void then it must be a function type.
let `R` be the returned from `__source<_Tp>(0)`.
We perform overload resolution on `__test<_Tp>(R)`.
* If `R` is `__dummy_type` we call `true_type __test(__dummy_type)`.
* if `R` is `_Tp&` and `_Tp&` decays to `_Tp*` we call `true_type __test(_Tp*)`. Only references to function types decay to a pointer of the same type.
* In all other cases we call `false_type __test(...)`.
`__source<_Tp>(0)` will try and form `_Tp&` in the return type. if `_Tp&` is not well formed the return type of `__source<_Tp>(0)` will be dummy type. `_Tp&` is only ill-formed for cv/ref qualified function types (and void which is dealt with elsewhere).
This fixes PR20084 - http://llvm.org/bugs/show_bug.cgi?id=20084
Reviewers: rsmith, K-ballo, mclow.lists
Reviewed By: mclow.lists
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D7573
llvm-svn: 229696
Marshall Clow