LLVM dropped support for Visual Studio versions older than 2015 quite
some time ago, so I consider it safe to drop libc++'s support for older
CRTs. The CRT in Visual Studio 2015 provides a lot of previously missing
functions, so targeting it requires less special casing.
Differential Revision: https://reviews.llvm.org/D31798
llvm-svn: 299743
Method specializations don't get exported even if there's an exported
extern template instantiation on Windows. Explicitly mark the methods
for export. They're already exported on Linux and Darwin, so there's no
ABI change on those platforms.
llvm-svn: 299348
This is effectively a partial revert of r278356, which started inlining
basic_string::__init. Even if we want to help the compiler along with
an inlinehint, we shouldn't hamstring it by forcing it to inline all the
time.
Libc++ uses always_inline widely as a limit-the-ABI-hack, but since
__init is already on the dylib boundary, it makes no sense here and just
harms the debugging experience at -O0.
rdar://problem/31013102
llvm-svn: 299290
r145698 introduced _LIBCPP_HAS_NO_STRONG_ENUMS by copy-pasting the
__has_feature check from objc_arc_weak/_LIBCPP_HAS_OBJC_ARC_WEAK, and
accidentally started defining _LIBCPP_HAS_NO_STRONG_ENUMS whenever
__has_feature(objc_arc_weak). This is totally bogus, and means that
Libc++ thinks Objective-C++ compilations with -fobjc-arc don't have
strong enums.
Delete the accidental line.
I thought about adding a test, but it would be entirely duplicative of
the patch (if has-feature strong enums, check that has-no-strong-enums
is not defined).
llvm-svn: 299236
Summary:
bad_function_call is currently an empty class, so any object files using
that class will end up with their own copy of its typeinfo, typeinfo
name and vtable, leading to unnecessary duplication that has to be
resolved by the dynamic linker. Instead, give bad_function_call a key
function and put a definition for that key function in libc++ itself, to
centralize the typeinfo and vtable.
This is consistent with the behavior for other exception classes. The
key functions are defined in libc++ rather than libc++abi since the
class is defined in the libc++ versioning namespace, so ABI
compatibility with libstdc++ is not a concern.
Guard this change behind an ABI macro, since it isn't backwards
compatible (i.e., clients built against the new libc++ headers wouldn't
be able to run against an older libc++ library).
Reviewers: mclow.lists, EricWF
Subscribers: mgorny, cfe-commits
Differential Revision: https://reviews.llvm.org/D27387
llvm-svn: 298937
Summary:
By manipulating a local variable in the loop, when the loop can
be optimized away (due to no non-trivial destructors), this lets
it be fully optimized away and we modify the __end_ separately.
This results in a substantial improvement in the generated code.
Prior to this change, this would be generated (on x86_64):
movq (%rdi), %rdx
movq 8(%rdi), %rcx
cmpq %rdx, %rcx
je LBB2_2
leaq -12(%rcx), %rax
subq %rdx, %rax
movabsq $-6148914691236517205, %rdx ## imm = 0xAAAAAAAAAAAAAAAB
mulq %rdx
shrq $3, %rdx
notq %rdx
leaq (%rdx,%rdx,2), %rax
leaq (%rcx,%rax,4), %rax
movq %rax, 8(%rdi)
And after:
movq (%rdi), %rax
movq %rax, 8(%rdi)
This brings this in line with what other implementations do.
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D25241
llvm-svn: 298601
basic_string::replace() has the below line
__sz += __n2 - __n1;
which fails overflow checks if __n1 > __n2, as the negative result
from the subtraction then overflows the original __sz when added to
it.
This behavior is valid as unsigned integer overflow is defined to wrap
around the maximum value and that produces the correct final value for
__sz. Therefore, we disable this check on this function.
llvm-svn: 297355
Put proper guards around _LIBCPP_METHOD_TEMPLATE_IMPLICIT_INSTANTIATION_VIS.
No functional change on non-Windows. Avoids incorrect macro redefinition
on Windows.
llvm-svn: 297330
r296565 attempted to add better diagnostics when an unordered container
is instantiated with a hash that doesn't meet the Hash requirements.
However I mistakenly checked the wrong set of requirements. Specifically
it checked if the hash met the requirements for specializations of
std::hash. However these requirements are stricter than the generic
Hash requirements.
This patch fixes the assertions to only check the Hash requirements.
llvm-svn: 296919
Summary:
Most classes annotated with _LIBCPP_TYPE_VIS need to have at least some
of their members exported, otherwise we have a lot of link errors when
linking against a libc++ built with hidden visibility. This also makes
_LIBCPP_TYPE_VIS be consistent across platforms, since on Windows it
already exports members.
With this change made, any template methods of a class marked
_LIBCPP_TYPE_VIS will also get default visibility when instantiatied,
which is not desirable for clients of libc++ headers who wish to control
their visibility; this is the same issue as PR30642. Annotate all
problematic methods with an explicit visibility specifier to avoid this.
The problematic methods were found by running bad-visibility-finder [1]
against the libc++ headers after making the _LIBCPP_TYPE_VIS change. The
small methods were marked for inlining; the larger ones hidden.
[1] https://github.com/smeenai/bad-visibility-finder
Reviewers: mclow.lists, EricWF
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D25208
llvm-svn: 296732
When building libc++ with hidden visibility, we want explicit template
instantiations to export members. This is consistent with existing
Windows behavior, and is necessary for clients to be able to link
against a hidden visibility built libc++ without running into lots of
missing symbols.
An unfortunate side effect, however, is that any template methods of a
class with an explicit instantiation will get default visibility when
instantiated, unless the methods are explicitly marked inline or hidden
visibility. This is not desirable for clients of libc++ headers who wish
to control their visibility, and led to PR30642.
Annotate all problematic methods with an explicit visibility specifier
to avoid this. The problematic methods were found by running
https://github.com/smeenai/bad-visibility-finder against the libc++
headers after making the _LIBCPP_EXTERN_TEMPLATE_TYPE_VIS change. The
methods were marked with the new _LIBCPP_METHOD_TEMPLATE_IMPLICIT_INSTANTIATION_VIS
macro, which was created for this purpose.
It should be noted that _LIBCPP_EXTERN_TEMPLATE_TYPE_VIS was originally
intended to expand to default visibility, and was changed to expanding
to default type visibility to fix PR30642. The visibility macro
documentation was not updated accordingly, however, so this change makes
the macro consistent with its documentation again, while explicitly
fixing the methods which resulted in that PR.
Differential Revision: https://reviews.llvm.org/D29157
llvm-svn: 296731
D29157 will make explicit template instantiations expand to default
visibility, at which point these method templates will need to be
explicitly marked hidden visibility to avoid leaking into other DSOs.
Unfortunately, because of clang PR32114, they must be marked inline (in
conjunction with `-fvisibility-inlines-hidden`) to actually hide them,
since clang doesn't respect the hidden visibility annotation.
Since this involves an ABI change, mark these methods inline in a
separate change, so that the ABI changes can be reviewed separately and
verified to be safe.
Differential Revision: https://reviews.llvm.org/D30523
llvm-svn: 296729
This patch adds a static assertion that the specified hash meets
the requirements of an enabled hash, and it ensures that the static
assertion is evaluated before __compressed_pair is instantiated.
That way the static assertion diagnostic is emitted first.
llvm-svn: 296565
This patch fixes llvm.org/PR32097 by using the __is_abstract
builtin type-trait instead of the previous library-only implementation.
All supported compilers provide this trait. I've tested as far
back as Clang 3.2, GCC 4.6 and MSVC trunk.
llvm-svn: 296561
This reverts SVN r295329. Although `__libcpp_thread_sleep_for` should
be alertable, the implementation causes a large regression in the test
suite. Add a FIXME item there for now to get the test suite in a better
state before attempting to fix that behaviour.
llvm-svn: 295561
When building with MSVCRT, we need to manually provide the type
promoting overloads to allow the correct type deduced invocation for
signbit(Int) and fpclassify(int).
llvm-svn: 295559
GCC 7.0.1 started warning that __attribute__((__strong__)) is depricated.
This patch switches to using inline namespace with GCC instead. I believe
this wasn't done originally in order to support older GCC versions w/o
support for inline namespaces, or because earlier versions of GCC warned
users that the STL was using an inline namespace (even though it shouldn't affect users).
However I believe all of the above problems are gone for GCC 4.9 and greater.
Therefore switching to using inline namespaces instead of using __strong__
is the most correct behavior.
llvm-svn: 295428
Summary:
This patch implements [P0003R5](http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p0003r5.html) which removes exception specifications from C++17.
The only changes to the library are removing `set_unexpected`, `get_unexpected`, `unexpected`, and `unexpected_handler`. These functions can be re-enabled in C++17 using `_LIBCPP_ENABLE_CXX17_REMOVED_UNEXPECTED_FUNCTIONS`.
@mclow.lists what do you think about removing stuff is this way?
Reviewers: mclow.lists
Reviewed By: mclow.lists
Subscribers: mclow.lists, cfe-commits
Differential Revision: https://reviews.llvm.org/D28172
llvm-svn: 295406
Summary:
This patch fixes http://llvm.org/PR31938. The description below is copy/pasted from the bug:
The standard says:
template<class charT, class traits = char_traits<charT>,
class Allocator = allocator<charT>>
class basic_string {
using value_type = typename traits::char_type;
// ...
basic_string(const charT* s, const Allocator& a = Allocator());
};
libc++ actually chooses to declare the constructor as
basic_string(const value_type* s, const Allocator& a = Allocator());
The implicit deduction guides from class template argument deduction make what was previously an implementation detail visible:
std::basic_string s = "foo"; // error, can't deduce charT.
The constructor in question is in the libc++ DSO, but fortunately it looks like fixing this will not result in an ABI break.
@rsmith How does this look? I did more than just the constructors mentioned in the PR, but IDK how far to take it.
Reviewers: mclow.lists, rsmith
Reviewed By: rsmith
Subscribers: cfe-commits, rsmith
Differential Revision: https://reviews.llvm.org/D29863
llvm-svn: 295393
On Windows, we were using `Sleep` which is not alertable. This means
that if the thread was used for a user APC or WinProc handling and
thread::sleep was used, we could potentially dead lock. Use `SleepEx`
with an alertable sleep, resuming until the time has expired if we are
awoken early.
llvm-svn: 295329
clang cannot properly handle __declspec and __attribute__ on classes
right now. This prevents the shared_mutex tests from working. Disable
the use of the annotation on COFF targets.
llvm-svn: 294958
