32-bit powerpc provides a 64 bit time_t type and older ppc64 systems
provide time_t as a floating point type. This caused problems when building
operations.cpp since operations.cpp contained compile time tests for conversions
between time_t and filesystem time type.
When these tests failed they caused the libc++ build to fail as well. This is unfortunate.
This patch moves the tests out of the source file and into the test suite. It also
expands the tests to allow testing of the weird time_t configurations on all platforms.
llvm-svn: 307461
This patch speculatively implements the PR for LWG 2937, which fixes
two issues with equivalent.
(1) It makes equivalent("dne", "exists") an error. Previously only
equivalent("dne", "dne") was an error and the former case was not (it returned false).
Now equivalent reports an error when either input doesn't exist.
(2) It makes equivalent(p1, p2) well-formed when `is_other(p1) && is_other(p2)`.
Previously this was an error, but there is seemingly no reason why it should be on POSIX system.
llvm-svn: 307117
Most of filesystem/path.cpp uses string_view_t. This fixes the two spots
that use string_view directly.
https://reviews.llvm.org/D34332
llvm-svn: 305661
Previously the explicit instantiation for this was in locale.cpp,
but that didn't make much sense. This patch creates a new vector.cpp
source file to contain the explicit instantiation.
llvm-svn: 305442
Summary:
This patch improves how libc++ handles min/max macros within the headers. Previously libc++ would undef them and emit a warning.
This patch changes libc++ to use `#pragma push_macro` to save the macro before undefining it, and `#pragma pop_macro` to restore the macros and the end of the header.
Reviewers: mclow.lists, bcraig, compnerd, EricWF
Reviewed By: EricWF
Subscribers: cfe-commits, krytarowski
Differential Revision: https://reviews.llvm.org/D33080
llvm-svn: 304357
Summary:
This patch refactors and tries to remove as much of the Windows support headers as possible. This is needed because they currently introduce super weird include cycles and dependencies between STL and libc headers.
The changes in this patch are:
* remove `support/win32/support.h` completely. The required parts have either been moved into `support/win32/msvc_support.h` (for `MSVC` only helpers not needed by Clang), or directly into their respective `foo.h` headers.
* Combine `locale_win32.h` and `locale_mgmt_win32.h` into a single headers, this header should only be included within `__locale` or `locale` to avoid include cycles.
* Remove the unneeded parts of `limits_win32.h` and re-name it to `limits_msvc_win32.h` since it's only needed by Clang.
I've tested this patch using Clang on Windows, but I suspect it might technically regress our non-existent support for MSVC. Is somebody able to double check?
This refactor is needed to support upcoming fixes to `<locale>` on Windows.
Reviewers: bcraig, rmaprath, compnerd, EricWF
Reviewed By: EricWF
Subscribers: majnemer, cfe-commits
Differential Revision: https://reviews.llvm.org/D32988
llvm-svn: 302727
Previously <locale> used std::unique_ptr<remove_ptr<locale_t>, locale-mgmt-function>
as a scope guard for (A) creating new locales, and (B) setting the thread specific locale
in RAII safe manner.
However using unique_ptr has some problems, first it requires that locale_t is a pointer
type, which may not be the case (Windows will need a non-pointer locale_t type that emulates _locale_t).
The second problem is that users of the guards had to supply the locale management function to the custom
deleter at every call site. However these locale management functions don't exist natively Windows, making
a good Windows implementation of locale more difficult.
This patch creates distinct and simply RAII guards that replace unique_ptr. These guards handle calling
the correct locale management function so that callers don't have too. This simplification will
aid in upcoming Windows fixes.
llvm-svn: 302474
Summary:
This patch implements exception_ptr on Windows using the `__ExceptionPtrFoo` functions provided by MSVC.
The `__ExceptionPtrFoo` functions are defined inside the C++ standard library, `msvcprt`, which is unfortunate because it requires libc++ to link to the MSVC STL. However this doesn't seem to cause any immediate problems. However to be safe I kept all usages within the libc++ dylib so that user programs wouldn't have to link to MSVCPRT as well.
Note there are still 2 outstanding exception_ptr/nested_exception test failures.
* `current_exception.pass.cpp` needs to be rewritten for the Windows exception_ptr semantics which copy the exception every time.
* `rethrow_if_nested.pass.cpp` need investigation. It hits a stack overflow, likely from recursion.
This patch also gets most of the `<future>` tests passing as well.
Reviewers: mclow.lists, compnerd, bcraig, rmaprath, majnemer, BillyONeal, STL_MSFT
Subscribers: mgorny, cfe-commits
Differential Revision: https://reviews.llvm.org/D32927
llvm-svn: 302393
_LIBCPP_ABI_MICROSOFT is more appropriate to use here, since the
conditionals are controlling Microsoft mangling. It wasn't used
originally since it didn't exist at the time.
llvm-svn: 300743
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
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:
Currently both libc++ and libc++abi provide definitions for operator new/delete. However I believe this is incorrect and that one or the other should offer them.
This patch adds the CMake option `-DLIBCXX_ENABLE_NEW_DELETE_DEFINITIONS` which defaults no `ON` unless `-DLIBCXXABI_ENABLE_NEW_DELETE_DEFINITIONS=ON` is specified.
Reviewers: mclow.lists, mehdi_amini, dexonsmith, danalbert, smeenai, mgorny, rmaprath
Reviewed By: mehdi_amini
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D30516
llvm-svn: 296802
This recommits r294707 with additional fixes. The main difference is
libc++ now correctly builds without any ABI library.
exception.cpp is a bloody mess. It's full of confusing #ifdef branches for
each different ABI library we support, and it's getting unmaintainable.
This patch breaks down exception.cpp into multiple different header files,
roughly one per implementation. Additionally it moves the definitions of
exceptions in new.cpp into the correct implementation header.
This patch also removes an unmaintained libc++abi configuration.
This configuration may still be used by Apple internally but there
are no other possible users. If it turns out that Apple still uses
this configuration internally I will re-add it in a later commit.
See http://llvm.org/PR31904.
llvm-svn: 294730
exception.cpp is a bloody mess. It's full of confusing #ifdef branches for
each different ABI library we support, and it's getting unmaintainable.
This patch breaks down exception.cpp into multiple different header files,
roughly one per implementation. Additionally it moves the definitions of
exceptions in new.cpp into the correct implementation header.
This patch also removes an unmaintained libc++abi configuration.
This configuration may still be used by Apple internally but there
are no other possible users. If it turns out that Apple still uses
this configuration internally I will re-add it in a later commit.
See http://llvm.org/PR31904.
llvm-svn: 294707
Different platforms implement the wait/sleep functions in difrerent ways.
It makes sense to externalize this into the threading API.
Differential revision: https://reviews.llvm.org/D29630
Reviewers: EricWF, joerg
llvm-svn: 294573
Recently I turned on libc++'s debug mode assertions when
CMake is configured with -DLIBCXX_ENABLE_ASSERTIONS=ON. This
change exposed assertion failures caused by bugs in filesystem.
This patch fixes those failures.
The first bug was that `PathParser` was using front()/back()
on empty string views in order to get the address of the character.
However this is UB on empty strings. Those operations now use data()
to obtain the pointer.
The second bug was that directory_iterator attempted to capture errno when it
was unset and there was an assertion to detect this.
llvm-svn: 294360
This really should get identified properly by the compiler to convert to
a NVRO, but compress the code anyways. This makes the implementation
identical to directory_iterator.cpp
llvm-svn: 294270
N4100 states that an error shall be reported if
`!exists(p) || !is_directory(p)`. We were missing the first half of the
conditional. Invert the error and normal code paths to make the code
easier to follow.
llvm-svn: 294127
It is my opinion that libc++ should never use `<cassert>`, including in the `dylib`.
This patch remove all uses of `assert` from within libc++ and replaces most of them with `_LIBCPP_ASSERT` instead.
Additionally this patch turn `LIBCXX_ENABLE_ASSERTIONS` off by default,
because the standard library should not be aborting user programs unless explicitly asked to.
llvm-svn: 294107
This adds a basic first cut implementation for directory_iterator on
Windows. It uses the FindFirstFile/FindNextFile which has the same
restrictions as opendir/readdir where there exists a TOCTOU race
condition.
llvm-svn: 293531
Microsoft's SAL has a `__deref` macro which results in a compilation
failure when building the filesystem module on Windows. Rename the
member function internally to avoid the conflict.
llvm-svn: 293449
This reverts commit r292883. Unfortunately <string_view> uses
_LIBCPP_ASSERT in a way which is not compatible with the C++11 dylib
build. I'll investigate more tomorrow.
llvm-svn: 292923
Summary:
It is my opinion that libc++ should never use `<cassert>`, including in the `dylib`. This patch remove all uses of `assert` from within libc++ and replaces most of them with `_LIBCPP_ASSERT` instead.
Additionally this patch turn `LIBCXX_ENABLE_ASSERTIONS` off by default, because the standard library should not be aborting user programs unless explicitly asked to.
Reviewers: mclow.lists, compnerd, smeenai
Reviewed By: mclow.lists
Subscribers: mgorny, cfe-commits
Differential Revision: https://reviews.llvm.org/D29063
llvm-svn: 292883
Summary:
This patch disables the aligned new/delet overloads on Apple platforms without `posix_memalign`. This fixes libc++.dylib build regressions on such platforms.
This fixes http://llvm.org/PR31448.
This patch should also be merged into the 4.0 release branch
Reviewers: mclow.lists, rsmith, dexonsmith, jeremyhu
Reviewed By: mclow.lists
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D28931
llvm-svn: 292564
In order to allow inlining of previously out-of-line functions without an ABI break
libc++ provides legacy definitions in the dylib that old programs can
continue to use. Unfortunatly Windows link.exe detects this hack and diagnoses the duplicate
definitions.
This patch disable the duplicate definitions on Windows by adding an ABI option
which disables all "legacy out-of-line symbols"
llvm-svn: 292190
Moves hot functions such as atomic add into the memory header file
so that they can be inlined, which brings performance benefits.
Patch by Kevin Hu, Aditya Kumar, Sebastian Pop
Differential Revision: https://reviews.llvm.org/D24991
llvm-svn: 292184
This patch contains multiple cleanups and fixes to better support building on
Windows.
* [Test] Fix handling of library runtime search paths by correctly adding them
to the PATH variable when running the tests.
* [Test] Don't explicitly force "--target=i686-pc-windows" when running the
test suite. Clang++ seems to deduce the correct target.
* [Test] Fix `.sh.cpp` tests on Windows by properly escaping flags used in
shell commands. Specifically windows style paths which included spaces
were causing these tests to fail.
* [CMake] Add "vcruntime" to the list of supported C++ ABI libraries in CMake, and
teach the test suite how to handle it. For now libc++ defaults to using
"vcruntime" on Windows except when libc++abi is in tree; That is probably
a bug and should be changed to always use vcruntime, at least for now.
* [Misc] Move the "c++-build" include directory to the libc++ binary dir
instead of the top level project dir and rename it "c++build". This is just
misc cleanup. Libc++ shouldn't be creating internal build files and directories
at the top-level projects root.
* [Misc] Build type_info's destructor when building for MSVC. This is a temporary
work around to prevent link errors until we have a proper type_info
implementation.
llvm-svn: 292157
Attempting to pair an `_aligned_malloc` with a regular free causes heap
corruption. Pairing with `_aligned_free` is required instead.
Makes the following libc++ tests pass on Windows:
```
std/language.support/support.dynamic/new.delete/new.delete.array/new_align_val_t.pass.cpp
std/language.support/support.dynamic/new.delete/new.delete.array/new_align_val_t_nothrow.pass.cpp
std/language.support/support.dynamic/new.delete/new.delete.single/new_align_val_t.pass.cpp
std/language.support/support.dynamic/new.delete/new.delete.single/new_align_val_t_nothrow.pass.cpp
```
Differential Revision: https://reviews.llvm.org/D28512
llvm-svn: 291743
Use CLOCK_UPTIME_RAW in case clock_gettime is available on Darwin.
On Apple platforms only CLOCK_UPTIME_RAW or mach_absolute_time are able
to time functions in the nanosecond range. Thus, they are the only
acceptable implementations of steady_clock.
Differential Revision: https://reviews.llvm.org/D27429
rdar://problem/29449467
llvm-svn: 291466
ELAST should point to the last valid error string value. However,
`_sys_nerr` provides the number of elements in the errlist array. Since
the index is 0-based, this is off-by-one. Adjust it accordingly.
Thanks to David Majnemer for catching this!
llvm-svn: 291336
Summary:
On Windows the identifier `__deallocate` is defined as a macro by one of the Windows system headers. Previously libc++ worked around this by `#undef __deallocate` and generating a warning. However this causes the WIN32 version of `__threading_support` to always generate a warning on Windows. This is not OK.
This patch renames all usages of `__deallocate` internally as to not conflict with the macro.
Reviewers: mclow.lists, majnemer, rnk, rsmith, smeenai, compnerd
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D28426
llvm-svn: 291332
Windows does not provide an implementation of `nanosleep`. Round up the
time duration to the nearest ms and use `Sleep`. Although this may
over-sleep, there is no hard real-time guarantee on the wake, so
sleeping a bit more is better than under-sleeping as it within the
specification.
llvm-svn: 291331
This patch refactors the compiler detection done in `__config` by creating a
set of `_LIBCPP_COMPILER_<TYPE>` macros. The goal of this patch is to make
it easier to detect what compiler is being used outside of `__config`.
Additionally this patch removes workarounds for GCC in `__bit_reference`. I
tested GCC 4.8 and 4.9 without the workaround and neither seemed to need it
anymore.
llvm-svn: 291286
Split out the recursive and non-recursive mutex. This split is needed
for platforms which may use differing types for the two mutex (e.g.
Win32 threads).
llvm-svn: 291145
In the C++ standard `std::pointer_safety` is defined
as a C++11 strongly typed enum. However libc++ currently defines
it as a class type which simulates a C++11 enumeration. This
can be detected in valid C++ code.
This patch introduces an the _LIBCPP_ABI_POINTER_SAFETY_ENUM_TYPE ABI option.
When defined `std::pointer_safety` is implemented as an enum type.
Unfortunatly this also means it can no longer be provided as an extension
in C++03.
Additionally this patch moves the definition for `get_pointer_safety()`
out of the dylib, and into the headers. New usages of `get_pointer_safety()`
will now use the inline version instead of the dylib version. However in
order to keep the dylib ABI compatible the old definition is explicitly
compiled into it.
llvm-svn: 291046
The name _LIBCPP_TYPE_VIS_ONLY is no longer accurate because both
_LIBCPP_TYPE_VIS and _LIBCPP_TYPE_VIS_ONLY expand to
__attribute__((__type_visibility__)) with Clang. The only remaining difference
is that _LIBCPP_TYPE_VIS_ONLY can be applied to templates whereas
_LIBCPP_TYPE_VIS cannot (due to dllimport/dllexport not being allowed on
templates).
This patch renames _LIBCPP_TYPE_VIS_ONLY to _LIBCPP_TEMPLATE_VIS.
llvm-svn: 291035
A typo and missing header inclusion was obscured by the litany of user
defined literal warnings. This fixes the detection of ELAST on windows.
llvm-svn: 290941
Replace the use of _WIN32 in libc++. Replace most use with a C runtime
check _LIBCPP_MSVCRT or the new _LIBCPP_WIN32 to indicate that we are
using the Win32 API. Use a new _LIBCPP_WCHAR_IS_UCS2 to indicate that we
are on an environment that has a short wchar_t.
llvm-svn: 290910
This patch re-commits a previous attempt to support building libc++ w/o
an ABI library. That patch was originally reverted because:
1) It forgot to teach the test suite about "default" ABI libraries.
2) Some LLVM builders don't clear the CMake cache between builds. The previous
patch caused those builders to fail since their old cache entry for
LIBCXX_CXX_ABI="" is no longer valid.
The updated patch addresses both issues. It works around (2) by adding
a hack to force the builders to update their cache entries. The hack will
be removed shortly once all LLVM builders have run.
Original commit message
-----------------------
Typically libc++ uses libc++abi or libcxxrt to provide the ABI and runtime bits
of the C++ STL. However we also support building w/o an ABI library entirely.
This patch fixes building libc++ w/o an ABI library (and incorporates the
`~type_info()` fix in D28211).
The main changes in this patch are:
1) Add `-DLIBCXX_CXX_ABI=default` instead of using the empty string to mean "default".
2) Fix CMake bits which treated "none" as "default" on OS X.
3) Teach the source files to respect `-D_LIBCPP_BUILDING_HAS_NO_ABI_LIBRARY`.
4) Define ~type_info() when _LIBCPP_BUILDING_HAS_NO_ABI_LIBRARY is defined.
Unfortunately this patch doesn't help clean up the macro mess that we use to
configure for different ABI libraries.
llvm-svn: 290849
Currently libc++ compiles a special version of error_category()
into the dylib. This definition is no longer needed, and doesn't
work on Windows due to dllimport/dllexport semantics.
For those reasons this patch introduces an option to
disable/enable this definition. By default the definition
is provided in ABI v1 except on windows. This patch
also addresses D28210.
llvm-svn: 290840
Typically libc++ uses libc++abi or libcxxrt to provide the ABI and runtime bits
of the C++ STL. However we also support building w/o an ABI library entirely.
This patch fixes building libc++ w/o an ABI library (and incorporates the
`~type_info()` fix in D28211).
The main changes in this patch are:
1) Add `-DLIBCXX_CXX_ABI=default` instead of using the empty string to mean "default".
2) Fix CMake bits which treated "none" as "default" on OS X.
3) Teach the source files to respect `-D_LIBCPP_BUILDING_HAS_NO_ABI_LIBRARY`.
4) Define ~type_info() when _LIBCPP_BUILDING_HAS_NO_ABI_LIBRARY is defined.
Unfortunately this patch doesn't help clean up the macro mess that we use to
configure for different ABI libraries.
llvm-svn: 290839
The locale structures have been made opaque in CRT 14+. This currently
prevents building libc++ for Windows. We can re-enable this in the
future when we have replicated the structure to access the private field
for the name (unless there exists a better supported mechanism to query
the name of a locale given the locale_t).
llvm-svn: 290835
As pointed out by Howard, this is actually 134774 days (* 24 * 3600),
and therefore seconds, not 100ns units. Adjust the units to reflect
reality.
llvm-svn: 290824
Visual C++ 14 and newer split msvcrt into msvcrt and ucrt with flavours
of the ucrt for different environments. This changed the access to the
ctype table by introducing the `__pctype_func` and `__pwctype_func`
accessors. Use this rather than directly accessing `_ctype` which
allows us to be safer in threaded situations by going through the libc
locking.
llvm-svn: 290823
Drawing some inspiration from code from Bill O'Neal as pointed out by
Howard, rework the code to avoid an overflow in the duration. Adjust
the style to match libc++ style as well.
Create a local typedef for the FILETIME duration (100-ns units). Use
this to define the difference between the NT and the UNIX epochs (which
previously overflowed due to the representation limits due to the
bouncing to ns). Return the FILETIME duration biased by the NT-to-UNIX
epoch conversion.
Use of the custom duration makes it easier to read and reason about the
code.
llvm-svn: 290806
system_clock::now is not entirely straight forward on Windows, which
does not have a clock_gettime function.
GetSystemTimeAsFileTime gives us the value relative to the NT epoch (Jan
1 1601) rather than the Unix epoch (Jan 1 1970). However, this function
has a low resolution (~10ms). Newer versions of Windows provide
GetSystemTimePreciseAsFileTime which gives us a much more accurate time
(<1us). Unfortunately, the latter is only available on Windows 8+ when
targeting desktop apps.
llvm-svn: 290803
When building libc++ without threading, strerror_r is not used. Define
the code only when threading is enabled. This allows us to build
system_error for Windows, which ATM doesn't build with threading.
llvm-svn: 290791
This patch implements changes to allow _LIBCPP_ASSERT to throw on failure
instead of aborting. The main changes needed to do this are:
1. Change _LIBCPP_ASSERT to call a handler via a replacable function pointer
instead of calling abort directly. Additionally this patch implements two
handler functions, one which aborts and another that throws an exception.
2. Add _NOEXCEPT_DEBUG macro for disabling noexcept spec on function which
contain _LIBCPP_ASSERT. This is required in order to prevent assertion
failures throwing through a noexcept function. This macro has no effect
unless _LIBCPP_DEBUG_USE_EXCEPTIONS is defined.
Having a non-aborting _LIBCPP_ASSERT is very important to allow sane testing of
debug mode. Currently we can only have one test case per file, since the test
case will cause the program to abort. Testing debug mode this way would require
thousands of test files, most of which would be 95% boiler plate. I don't think
this is a feasible strategy. Fortunately using a throwing debug handler solves
these issues.
Additionally this patch rewrites the documentation for debug mode.
llvm-svn: 290651
Summary:
The underlying C locales provide the `thousands_sep` and `decimal_point` as strings, possible with more than one character. We currently don't handle this case even for `wchar_t`.
This patch properly converts the mbs -> wide character for `moneypunct_byname<wchar_t>`. For the `moneypunct_byname<char>` case we attempt to narrow the WC and if that fails we also attempt to translate it to some reasonable value. For example we translate U00A0 (non-breaking space) into U0020 (regular space). If none of these conversions succeed then we simply allow the base class to provide a fallback value.
Reviewers: mclow.lists, EricWF
Subscribers: vangyzen, george.burgess.iv, cfe-commits
Differential Revision: https://reviews.llvm.org/D24218
llvm-svn: 289347
The function definitions being guarded by the pragma were all static, so
they wouldn't be exported anyway. In any case, we should prefer the
visibility macros. No functional change.
Differential Revision: https://reviews.llvm.org/D26940
llvm-svn: 287768
The code cannot currently link when using libsupc++ with the
LIBCXX_ENABLE_STATIC_ABI_LIBRARY option.
This change ifdef's out the the destructor and 'what' function for
bad_array_length and bad_array_new_length when GLIBCXX is defined.
The constructors that are left in are the only functions not being provided by
libsupc++ itself, and follows the same pattern that was used to ifdef bad_alloc.
Testing was done on a Linux x86_64 host using GCC 5.4 and libc++ from ToT.
I see no change to the test results when using libsup++ or libstdc++ without
LIBCXX_ENABLE_STATIC_ABI_LIBRARY. When using libsupc++ with
LIBCXX_ENABLE_STATIC_ABI_LIBRARY it will now build and test results are the
same as those without the option specified.
Reviewed as https://reviews.llvm.org/D26186
llvm-svn: 287388
This is a generalization of `_LIBCPP_NEW_DELETE_VIS`; the new macro name
captures the semantics better, and also allows us to get rid of the
`_WIN32` check in `include/new`. No functional change.
Differential Revision: https://reviews.llvm.org/D26702
llvm-svn: 287164
This patch fixes a performance bug when constructing or appending to a path
from a string or c-string. Previously we called 'push_back' to append every
single character. This caused multiple re-allocation and copies when at most
one reallocation is necessary. The new behavior is to simply call
`string::append` so it can correctly handle reallocation.
For large strings this change is a ~4x improvement. This also makes our path
faster to construct than libstdc++'s.
llvm-svn: 285530
This patch entirely rewrites the parsing logic for paths. Unlike the previous
implementation this one stores information about the current state; For example
if we are in a trailing separator or a root separator. This avoids the need for
extra lookahead (and extra work) when incrementing or decrementing an iterator.
Roughly this gives us a 15% speedup over the previous implementation.
Unfortunately this implementation is still a lot slower than libstdc++'s.
Because libstdc++ pre-parses and splits the path upon construction their
iterators are trivial to increment/decrement. This makes libc++ lazy parsing
100x slower than libstdc++. However the pre-parsing libstdc++ causes a ton
of extra and unneeded allocations when constructing the string. For example
`path("/foo/bar/")` would require at least 5 allocations with libstdc++
whereas libc++ uses only one. The non-allocating behavior is much preferable
when you consider filesystem usages like 'exists("/foo/bar/")'.
Even then libc++'s path seems to be twice as slow to simply construct compared
to libstdc++. More investigation is needed about this.
llvm-svn: 285526
This patch does two seperate things. First it adds a file called
"__libcpp_version" which only contains the current libc++ version
(currently 4000). This file is not intended for use as a header. This file
is used by Clang in order to easily determine the installed libc++ version.
This allows Clang to enable/disable certain language features only when the
library supports them.
The second change is the addition of _LIBCPP_LIBRARY_VERSION macro, which
returns the version of the installed dylib since it may be different than
the headers.
llvm-svn: 285382
Summary:
`__libcpp_refstring` currently has two different definitions. First there is the complete definition in `<__refstring>` but there is also a second in `<stdexcept>`. The historical reason for this split is because both libc++ and libc++abi need to see the inline definitions of __libcpp_refstrings methods, but the `<stdexcept>` header doesn't. However this is an ODR violation and breaks the modules build.
This patch fixes the issue by creating a single class definition in `<stdexcept>` and changing `<__refstring>` to contain only the inline method definitions. This way both `libcxx/src/stdexcept.cpp` and `libcxxabi/src/stdexcept.cpp` see the same declaration in `<stdexcept>` and definitions in `<__refstring>`
Reviewers: mclow.lists, EricWF
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D25603
llvm-svn: 285100
Convert the Solaris xlocale.c compatibility library from plain C to C++
in order to fix the build failures caused by the addition of -std=c++11
to LIBCXX_COMPILE_FLAGS. The additional flag got propagated to the C
file, resulting in error with strict compilers.
Differential Revision: https://reviews.llvm.org/D25431
llvm-svn: 284494
Fix the iswxdigit_l() function prototype to take wint_t parameter
instead of incorrect wchar_t.
Differential Revision: https://reviews.llvm.org/D25431
llvm-svn: 284493
Summary:
This patch implements the library side of P0035R4. The implementation is thanks to @rsmith.
In addition to the C++17 implementation, the library implementation can be explicitly turned on using `-faligned-allocation` in all dialects.
Reviewers: mclow.lists, rsmith
Subscribers: rsmith, cfe-commits
Differential Revision: https://reviews.llvm.org/D25591
llvm-svn: 284206
These functions were removed from the dylib sometime between the 3.9 release
and now. This patch manually exports them to re-gain ABI compatibility.
llvm-svn: 284193
Summary:
Adapt implementation of Library Fundamentals TS optional into an implementation of N4606 optional.
- Update relational operators per http://wg21.link/P0307
- Update to requirements of http://wg21.link/P0032
- Extension: Implement trivial copy/move construction/assignment for `optional<T>` when `T` is trivially copyable.
Audit P/Rs for optional LWG issues:
- 2756 "C++ WP optional<T> should 'forward' T's implicit conversions" Implemented, which also resolves 2753 "Optional's constructors and assignments need constraints" (modulo my refusal to explicitly delete the move operations, which is a design error that I'm working on correcting in the 2756 P/R).
- 2736 "nullopt_t insufficiently constrained" Already conforming. I've added a test ensuring that `nullopt_t` is not copy-initializable from an empty braced-init-list, which I believe is the root intent of the issue, to avoid regression.
- 2740 "constexpr optional<T>::operator->" Already conforming.
- 2746 "Inconsistency between requirements for emplace between optional and variant" No P/R, but note that the author's '"suggested resolution" is already implemented.
- 2748 "swappable traits for optionals" Already conforming.
- 2753 "Optional's constructors and assignments need constraints" Implemented.
Most of the work for this patch was done by Casey Carter @ Microsoft. Thank you Casey!
Reviewers: mclow.lists, CaseyCarter, EricWF
Differential Revision: https://reviews.llvm.org/D22741
llvm-svn: 283980
This patch is largely thanks to Casey Carter @ Microsoft. He did the initial
work of porting our experimental implementation and tests over to namespace
std.
llvm-svn: 283977
Fuchsia is a new operating system which uses musl as the standard
C library, libc++ and libc++abi as the C++ standard library.
Differential Revision: https://reviews.llvm.org/D25414
llvm-svn: 283788
This patch applies the _LIBCPP_SAFE_STATIC attribute to internal globals,
most of which are locking primitives, in order to ensure that they can
safely be used during program startup.
This patch also fixes an unsafe static init issue with the global locks
used to implement atomic operations on shared pointers. Previously the
locks were initialized using a dynamically initialized pointer, so it was
possible that the pointer was uninitialized.
llvm-svn: 282640
The ::stat struct on Linux, FreeBSD, and OS X provides the access and
modification times as an instance of 'timespec', which has a nanosecond
resolution. The 'st_mtime' and 'st_atime' members simply reference the 'tv_sec'
value of the timespec struct. This patch changes 'last_write_time(...)' so that
it extracts both the seconds and nanoseconds values of the last modification
time, providing a more accurate implementation of 'last_write_time(...)'.
Additionally this patch fixes a possible signed integer overflow bug. The
'file_time_type' type cannot represent all possible values returned by
the filesystem. Attempting to construct a 'file_time_type' from one of these
values is undefined behavior. This patch avoids that UB by detecting possible
overflows before the conversion.
llvm-svn: 282634
Summary:
This patch has been a long time coming (Thanks @eugenis). It changes `_LIBCPP_INLINE_VISIBILITY` to use `__attribute__((internal_linkage))` instead of `__attribute__((visibility("hidden"), always_inline))`.
The point of `_LIBCPP_INLINE_VISIBILITY` is to prevent inline functions from being exported from both the libc++ library and from user libraries. This helps libc++ better manage it's ABI.
Previously this was done by forcing inlining and modifying the symbols visibility. However inlining isn't guaranteed and symbol visibility only affects shared libraries making this an imperfect solution. `internal_linkage` improves this situation by making all symbols local to the TU they are emitted in, regardless of inlining or visibility. IIRC the effect of applying `__attribute__((internal_linkage))` to an inline function is the same as applying `static`.
For more information about the attribute see: http://lists.llvm.org/pipermail/cfe-dev/2015-October/045580.html
Most of the work for this patch was done by @eugenis.
Reviewers: mclow.lists, eugenis
Subscribers: eugenis, cfe-commits
Differential Revision: https://reviews.llvm.org/D24642
llvm-svn: 282345
On Windows, marking an `extern template class` declaration as exported
actually forces an instantiation, which is not the desired behavior.
Instead, the actual explicit instantiations need to be exported.
Differential Revision: https://reviews.llvm.org/D24679
llvm-svn: 281925
Summary:
None of these checks are specific to Android devices. If libc++ was
used with Bionic on a normal Linux system these checks would still be
needed.
Reviewers: mclow.lists, EricWF
Subscribers: compnerd, tberghammer, danalbert, srhines, cfe-commits
Differential Revision: https://reviews.llvm.org/D24690
llvm-svn: 281921
When `_LIBCPP_NO_EXCEPTIONS` is defined, we end up with compile errors
when targeting MSVCRT:
* Code includes `<new>`
* `<new>` includes `<cstdlib>` in order to get `abort`
* `<cstdlib>` includes `<stdlib.h>`, _before_ the `using ::abort`
* `<stdlib.h>` includes `locale_win32.h`
* `locale_win32.h` includes `<memory>`
* `<memory>` includes `<stdexcept>`
* `<stdexcept>` includes `<cstdlib` for `abort`, but that inclusion gets
(correctly) ignored because of header guards
* `<stdexcept>` references `_VSTD::abort`, which isn't declared
The easiest solution is to make `locale_win32.h` not include `<memory>`,
by removing the use of `unique_ptr` and manually restoring the locale
instead.
Differential Revision: https://reviews.llvm.org/D24374
llvm-svn: 281641
call_once is using relaxed atomic load to perform double-checked locking, which contains a data race. The fast-path load has to be an acquire atomic load.
Differential Revision: https://reviews.llvm.org/D24028
llvm-svn: 280621
Summary:
This patch allows threads not created using `std::thread` to use `std::notify_all_at_thread_exit` by ensuring the TL state has been initialized within `std::notify_all_at_thread_exit`.
Additionally this patch "fixes" a potential oddity in `__thread_local_pointer::reset(pointer)`, which would previously delete the old thread local data. However there should *never* be old thread local data because pthread *should* null it out on thread exit. Unfortunately it's possible that pthread failed to do this according to the spec:
>
> Upon key creation, the value NULL shall be associated with the new key in all active threads. Upon thread creation, the value NULL shall be associated with all defined keys in the new thread.
>
> An optional destructor function may be associated with each key value. At thread exit, if a key value has a non-NULL destructor pointer, and the thread has a non-NULL value associated with that key, the value of the key is set to NULL, and then the function pointed to is called with the previously associated value as its sole argument. The order of destructor calls is unspecified if more than one destructor exists for a thread when it exits.
>
> If, after all the destructors have been called for all non-NULL values with associated destructors, there are still some non-NULL values with associated destructors, then the process is repeated. If, after at least {PTHREAD_DESTRUCTOR_ITERATIONS} iterations of destructor calls for outstanding non-NULL values, there are still some non-NULL values with associated destructors, implementations may stop calling destructors, or they may continue calling destructors until no non-NULL values with associated destructors exist, even though this might result in an infinite loop.
However if pthread fails to delete the value it is probably incorrect for us to do it. Destroying the value performs all of the "at thread exit" actions registered with it but we are way past "at thread exit".
Reviewers: mclow.lists, bcraig, EricWF
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D24159
llvm-svn: 280588
Summary:
Currently a number of GCC warnings are emitted when building libc++. This patch fixes or ignores all of them. The primary changes are:
* Work around strict aliasing issues in `typeinfo::hash_code()` by using __attribute__((may_alias)). However I think a non-aliasing `hash_code()` implementation is possible. Further investigation needed.
* Add `_LIBCPP_UNREACHABLE()` to switch in `strstream.cpp` to avoid -Wpotentially-uninitialized.
* Fix -Wunused-value warning in `__all` by adding a void cast.
* Ignore -Wattributes for now. There are a number of real attribute issues when using GCC but enabling the warning is too noisy.
* Ignore -Wliteral-suffix since it warns about the use of reserved identifiers. Note Only GCC 7.0 supports disabling this warning.
* Ignore -Wc++14-compat since it warns about the sized new/delete overloads.
Reviewers: EricWF
Differential Revision: https://reviews.llvm.org/D24003
llvm-svn: 280007
If the last destruction is uncontended, skip the atomic store on
__shared_weak_owners_. This shifts some costs from normal
shared_ptr usage to weak_ptr uses.
https://reviews.llvm.org/D22470
llvm-svn: 277357
The end pointer should point to one past the end of the newly allocated
buffer.
rdar://problem/24265174
Differential Revision: http://reviews.llvm.org/D20334
llvm-svn: 274132
This patch makes the bind placeholders in std::placeholders both (1) const and
(2) constexpr (See below).
This is technically a breaking change for any code using the placeholders
outside of std::bind and depending on them being non-const. However I don't
think this will break any real world code.
(1) Previously the placeholders were non-const extern globals in all
dialects. This patch changes these extern globals to be const in all dialects.
Since the cv-qualifiers don't participate in name mangling for globals this
is an ABI compatible change.
(2) Make the placeholders constexpr in C++11 and beyond. Although LWG 2488 only
applies to C++17 I don't see any reason not to backport this change.
llvm-svn: 273824
This changes how filesystem::permissions(p, perms) handles symlinks. Previously
symlinks were not resolved by default instead only getting resolved when
"perms::resolve_symlinks" was used. After this change symlinks are resolved
by default and perms::symlink_nofollow must be given to change this.
This issue has not yet been moved to Ready status, and I will revert if it
doesn't get moved at the current meeting. However I feel confident that it
will and it's nice to have implementations when moving issues.
llvm-svn: 273328
* Fix non-null violation in strstream.cpp
Overflow was calling memcpy with a null parameter and a size of 0.
* Fix std/atomics/atomics.flag/ tests:
a.test_and_set() was reading from an uninitialized atomic, but wasn't
using the value. The tests now clear the flag before performing the
first test_and_set. This allows UBSAN to test that clear doesn't read
an invalid value.
* Fix std/experimental/algorithms/alg.random.sample/sample.pass.cpp
The tests were dereferencing a past-the-end pointer to an array so that
they could do pointer arithmetic with it. Instead of dereference the iterator
I changed the tests to use the special 'base()' test iterator method.
* Add -fno-sanitize=float-divide-by-zero to suppress division by zero UBSAN diagnostics.
The tests that cause float division by zero are explicitly aware that they
are doing that. Since this is well defined for IEEE floats suppress the warnings
for now.
llvm-svn: 273107
* Fix passing a negative number as either tv_usec or tv_nsec. When file_time_type
is negative and has a non-zero sub-second value we subtract 1 from tv_sec
and make the sub-second duration positive.
* Detect and report when 'file_time_type' cannot be represented by time_t. This
happens when using large/small file_time_type values with a 32 bit time_t.
There is more work to be done in the implementation. It should start to use
stat's st_mtim or st_mtimeval if it's provided as an extension. That way
we can provide a better resolution.
llvm-svn: 273103
Summary:
Currently the implementation of [util.smartptr.shared.atomic] is provided only when using Clang, and not with GCC. This is a relic of not having a GCC implementation of <atomic>, even though <atomic> isn't actually used in the implementation. This patch enables support for atomic shared_ptr functions when using GCC.
Note that this is not a header only change. Previously only Clang builds of libc++.so would provide the required symbols. There is no reason for this restriction.
After this change both Clang and GCC builds should be binary compatible with each other WRT these symbols.
Reviewers: mclow.lists, rmaprath, EricWF
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D21407
llvm-svn: 273076
Currently 4 tests are failing on the ARM buildbot. To try and diagnose each
of the failures this patch does the following:
1) path.itr/iterator.pass.cpp
* Temporarily print iteration sequence to see where its failing.
2) path.native.obs/string_alloc.pass.cpp
* Remove test that ::new is not called when constructing a short string
that requires a conversion. Since during the conversion global locale
objects might be constructed.
3) fs.op.funcs/space.pass.cpp
* Explicitly use uintmax_t in the implementation of space, hopefully
preventing possible overflows.
* Add additional tests that check for overflow is the calculation of the
space_info values.
* Add additional tests for the values returned from statfvs.
4) fs.op.funcs/last_write_time.pass.cpp
* No changes made yet.
llvm-svn: 273075
r273060 didn't completely fix the issues in recursive_directory_iterator and
the tests. This patch follows up with more fixes
* Fix bug where recursive_directory_iterator::increment(ec) did not reset
the error code if no failure occurred.
* Fix bad assertion in the recursive_directory_iterator::increment(ec) test
that would only fire for certain iteration orders.
llvm-svn: 273070
There are two fixes in this patch:
* Fix bug where the constructor of recursive_directory_iterator did not reset
the error code if no failure occurred.
* Fix tests were dependent on the iteration order of the test directories.
llvm-svn: 273060
Add the completed std::experimental::filesystem implementation and tests.
The implementation supports C++11 or newer.
The TS is built as part of 'libc++experimental.a'. Users of the TS need to
manually link this library. Building and testing the TS can be disabled using
the CMake option '-DLIBCXX_ENABLE_FILESYSTEM=OFF'.
Currently 'libc++experimental.a' is not installed by default. To turn on the
installation of the library use '-DLIBCXX_INSTALL_EXPERIMENTAL_LIBRARY=ON'.
llvm-svn: 273034
CloudABI has gained the mblen_l() function in the meantime that does
properly return whether the character set has shift-states (read:
never).
llvm-svn: 272886
Summary:
Android didn't gain GNU's strerror_r until Marshmallow. If we're
building libc++ against something older (we build the NDK library
against the oldest release we support, currently Gingerbread), fall
back to the POSIX version.
Reviewers: mclow.lists, EricWF
Subscribers: tberghammer, danalbert, srhines, cfe-commits
Differential Revision: http://reviews.llvm.org/D21402
llvm-svn: 272827
Summary:
system_error::message() uses `strerror` for the generic and system categories. This function is not thread safe.
The fix is to use `strerror_r`. It has been available since 2001 for GNU libc and since BSD 4.4 on FreeBSD/OS X.
On platforms with GNU libc the extended version is used which always returns a valid string, even if an error occurs.
In single-threaded builds `strerror` is still used.
See https://llvm.org/bugs/show_bug.cgi?id=25598
Reviewers: majnemer, mclow.lists
Subscribers: erik65536, cfe-commits, emaste
Differential Revision: http://reviews.llvm.org/D20903
llvm-svn: 272633
Some pthread implementations do not like being called pthead_join()
with the pthread_t argument set to 0, and causes a segfault. This
patch fixes this issue by validating the pthread_t argument before
invoking pthread_join().
NFC.
Differential revision: http://reviews.llvm.org/D20929
Change-Id: Ief817c57bd0e1f43cbaa03061e02417d6a180c38
Reviewers: EricWF
llvm-svn: 271634
This patch extracts out all the pthread dependencies of libcxx into the
new header __threading_support. The motivation is to make it easy to
re-target libcxx into platforms that do not support pthread.
Original patch from Fulvio Esposito (fulvio.esposito@outlook.com) - D11781
Applied with tweaks - D19412
Change-Id: I301111f0075de93dd8129416e06babc195aa936b
llvm-svn: 268734
Summary:
Out-of-line symbols for <experimental/...> headers are not ABI or API stable and cannot live in the 'libc++.dylib'. Currently they have nowhere to live. I would like to add a new library target `libc++experimental.a` to fix this.
Previously I had suggested different libraries for different TS's (`libc++filesystem.a`, 'libc++LFTS.a`, ect). I no longer think this is the right approach.
Instead `c++experimental` will hold *all* TS implementations as a single monolithic library. I see two main benefits to this:
1. Users only have to know about and manually link one library.
2. It makes it easy to implement TS's with one or two out-of-line symbols. (Ex. PMRs)
`c++experimental` provides NO ABI compatibility. Symbols can freely be added/removed/changed without concern for ABI stability.
I will add documentation for this after landing this patch (but before adding anything to it).
`c++experimental` only builds as a static library. By default CMake will build/test this library but will *NOT* install it.
This patch adds the CMake and LIT logic needed to build/test the new library. Once this lands I plan on using it to implement parts of `<experimental/memory_resource>`.
Reviewers: mclow.lists
Subscribers: cfe-commits, theraven, krememek, dexonsmith, bcraig, beanz, danalbert
Differential Revision: http://reviews.llvm.org/D19856
llvm-svn: 268443
Instead of checking _LIBCPP_LOCALE_L_EXTENSIONS all over, instead check it
once, and define the various *_l symbols once. The private redirector symbol
names are all prefixed with _libcpp_* so that they won't conflict with user
symbols, and so they won't conflict with future C library symbols. In
particular, glibc likes providing private symbols such as __locale_t, so we
should follow a different naming pattern (like _libcpp_*) to avoid problems
on that front.
Tested on Linux with glibc. Hoping for the best on OSX and the various BSDs.
http://reviews.llvm.org/D17456
llvm-svn: 263016
Instead of excluding all known operating systems that are not derived from BSD,
I now include all operating systems that claim to be derived from BSD.
Hopefully, that will make it so that this check doesn't need to change for
every new operating system that comes along.
http://reviews.llvm.org/D16634
llvm-svn: 259193
Summary:
Also, there are no exported character type tables from Musl so we have to
Fallback to the standard functions. This reduces the number of libcxx's
test-suite failures down to ~130 for MIPS. Most of the remaining failures
come from the atomics (due to the lack of 8-byte atomic-ops in MIPS32) and
thread tests.
Reviewers: mclow.lists, EricWF, dalias, jroelofs
Subscribers: tberghammer, danalbert, srhines, cfe-commits
Differential Revision: http://reviews.llvm.org/D14926
llvm-svn: 253972
Fixes a small omission in libcxx that prevents libcxx being built when
-DLIBCXX_ENABLE_EXCEPTIONS=0 is specified.
This patch adds XFAILS to all those tests that are currently failing
on the new -fno-exceptions library variant. Follow-up patches will
update the tests (progressively) to cope with the new library variant.
Change-Id: I4b801bd8d8e4fe7193df9e55f39f1f393a8ba81a
llvm-svn: 252598
Summary:
This patch adds the LIBCXX_LIBC_IS_MUSL cmake option to allow the
building of libcxx with the Musl C library. The option is necessary as
Musl does not provide any predefined macro in order to test for its
presence, like GLIBC. Most of the changes specify the correct path to
choose through the various #if/#else constructs in the locale code.
Depends on D13407.
Reviewers: mclow.lists, jroelofs, EricWF
Subscribers: jfb, tberghammer, danalbert, srhines, cfe-commits
Differential Revision: http://reviews.llvm.org/D13673
llvm-svn: 252457
This was put in to get libc++ building without libcxxabi. We now have
macros that show that we are building against libcxxabi so use that
instead. This guards against existing but broken cxxabi.h headers on the
system.
llvm-svn: 250507
Summary:
In rL241532, atomic_support.h was added, which provides handling of
atomic operations for libc++. When atomic builtins are not available,
it emits a warning about being unsupported, but it still provides a
number of stubs for the required functions.
However, it misses a stub for `__libcpp_relaxed_store()`. Add it, by
using the same implementation as for `__libcpp_atomic_store()`.
(Note that I encountered this on arm-freebsd, which still defaults to
armv4, and does not have the runtime libcalls to support atomic
builtins. For now, I have simply disabled using them.)
Reviewers: mclow.lists, EricWF
Subscribers: theraven, cfe-commits, jroelofs, majnemer, aemerson
Differential Revision: http://reviews.llvm.org/D13051
llvm-svn: 248313
Summary:
After putting this question up on cfe-dev I have decided that it would be best to allow the use of `<atomic>` in C++03. Although static initialization is a concern the syntax required to get it is C++11 only. Meaning that C++11 constant static initialization cannot silently break in C++03, it will always cause a syntax error. Furthermore `ATOMIC_VAR_INIT` and `ATOMIC_FLAG_INIT` remain defined in C++03 even though they cannot be used because C++03 usages will cause better error messages.
The main change in this patch is to replace `__has_feature(cxx_atomic)`, which only returns true when C++ >= 11, to `__has_extension(c_atomic)` which returns true whenever clang supports the required atomic builtins.
This patch adds the following macros:
* `_LIBCPP_HAS_C_ATOMIC_IMP` - Defined on clang versions which provide the C `_Atomic` keyword.
* `_LIBCPP_HAS_GCC_ATOMIC_IMP` - Defined on GCC > 4.7. We must use the fallback atomic implementation.
* `_LIBCPP_HAS_NO_ATOMIC_HEADER` - Defined when it is not safe to include `<atomic>`.
`_LIBCPP_HAS_C_ATOMIC_IMP` and `_LIBCPP_HAS_GCC_ATOMIC_IMP` are mutually exclusive, only one should be defined. If neither is defined then `<atomic>` is not implemented and including `<atomic>` will issue an error.
Reviewers: chandlerc, jroelofs, mclow.lists
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D11555
llvm-svn: 245463
Summary:
In some places in libc++ we need to use the `__atomic_*` builtins. This patch adds a header that provides access to those builtins in a uniform way from within the dylib source.
If the compiler building the dylib does not support these builtins then a warning is issued.
Only relaxed loads are needed within the headers. A singe function to do these relaxed loads has been added to `<memory>`.
This patch applies the new atomic builtins to `__shared_count` and `call_once`.
Reviewers: mclow.lists
Subscribers: majnemer, jroelofs, cfe-commits
Differential Revision: http://reviews.llvm.org/D10406
llvm-svn: 241532
After r241454 landed, libc++'s locale code compiles on CloudABI, with
the exception of the following two bits:
- CloudABI doesn't have setlocale(), as the C library does not keep
track of any global state. The global locale is always set to "C".
Disable the call to setlocale() on this system.
- Similarly, mbtowc_l() is also not present, as it is also not
thread-safe. As CloudABI does not support state-dependent encodings,
simply disable that part of the logic.
The locale code now compiles out of the box on CloudABI.
Differential Revision: http://reviews.llvm.org/D10729
Reviewed by: jroelofs
llvm-svn: 241455
The __cloc() function is only present in case the environment does not
provide a way to refer to the C locale using a compile-time constant
expression. _LIBCPP_GET_C_LOCALE seems to be defined unconditionally.
This improves compilation of the locale code on CloudABI.
Differential Revision: http://reviews.llvm.org/D10690
Reviewed by: jroelofs
llvm-svn: 241454
Within the shared state methods do not unlock the lock guards manually. This
could cause a race condition where the shared state is destroyed before the
method is complete.
llvm-svn: 239577
The changes in src/exception.cpp and cmake/Modules/HandleLibCXXABI.cmake fix a
bug when building libc++ with GCC. Because GCC does not support __has_include
we need to explicitly tell it that we are building against libc++abi via the
preprocessor definition `LIBCXX_BUILDING_LIBCXXABI`.
The changes in include/ratio are to work around CWG defect
1712 (constexpr variable template declarations). GCC 4.8 and before has not
adopted the resolution to this defect.
The changes in include/exception work around an issue where is_final is used
without it being defined in type_traits.
llvm-svn: 237767
The system_clock::now() function currently uses gettimeofday(). The
problem with gettimeofday() is that it is an obsolete XSI function,
hence unavailable on CloudABI. See:
http://pubs.opengroup.org/onlinepubs/9699919799/functions/gettimeofday.html
Change this code to use clock_gettime() with CLOCK_REALTIME instead,
which is more consistent, as clock_gettime() is already used for
steady_clock.
A previous version of this change actually attempted to change
system_clock::duration, but I reverted this part as it breaks the
existing ABI.
Differential Revision: http://reviews.llvm.org/D8253
Approved by: jroelofs
llvm-svn: 237390
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
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
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
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: 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: NaCl shouldn't include sysctl.h when trying to determine std:🧵:hardware_concurrency, it should instead use sysconf(_SC_NPROCESSORS_ONLN) through unistd.h. No test needs to be changed, since hardware_concurrency.pass.cpp already tests that std:🧵:hardware_concurrency > 0.
Test Plan: make check-libcxx
Reviewers: dschuff, danalbert
Subscribers: jfb, cfe-commits
Differential Revision: http://reviews.llvm.org/D6470
llvm-svn: 223128
Summary:
The NaCl sandbox doesn't allow opening files under /dev, but it offers an API which provides the same capabilities. This is the same random device emulation that nacl_io performs for POSIX support, but nacl_io is an optional library so libc++ can't assume that device emulation will be performed. Note that NaCl only supports /dev/urandom, not /dev/random.
This patch also cleans up some of the preprocessor #endif, and fixes the test for Win32 (it accepts any token, and would therefore never throw regardless of the token provided).
Test Plan: ninja check-libcxx
Reviewers: dschuff, mclow.lists, danalbert
Subscribers: jfb, cfe-commits
Differential Revision: http://reviews.llvm.org/D6442
llvm-svn: 223068
Since the initialization of the pointer happens across the libc library boundry
MSAN will not know the pointer was initialized. This fixes MSAN failures in
test/strings/string.conversions.
llvm-svn: 222052
If you're crazy enough to want this sort of thing, then add
-D_LIBCPP_HAS_NO_THREADS to your CXXFLAGS and
--param=additiona_features=libcpp-has-no-threads to your lit commnad line.
http://reviews.llvm.org/D3969
llvm-svn: 217271
After discussing implementing more tests for this with @danalbert & @mclow, I
realized this change is not correct.
The C++ standard requires do_is() to behave as if it were a loop that checked
is(). Furthermore, it requires is() to check "The first form returns the result
of the expression (M & m) != 0; i.e., true if the character has the
characteristics specified"... which the reverted patch definitely does not
conform to. Even further, furthermore, this requires that ctype's mask be an
actual bitmask, unlike what android and newlib provide for _ctype_.
Fixing the original bug that instigated this patch remains TBD.
llvm-svn: 216508
Turning off explicit template instantiation leads to a pretty
significant build time and code size cost. We're better off dealing
with ABI incompatibility issues that come up in a less heavy handed
way.
This reverts commit r189610.
llvm-svn: 215740
__get_classname() and __bracket_expression were assuming that
char_class_type was ctype_base::mask rather than using
regex_traits<_CharT>::char_class_type.
This change allows char_class_type to be defined to something other than
ctype_base::mask so that the implementation will still work for
platforms with an 8-bit ctype mask (such as Android and OpenBSD).
llvm-svn: 214201
Signals may result in nanosleep returning with only some of the
requested sleeping performed.
Utilize nanosleep's "time-remaining" out parameter to continue sleeping
when this occurs.
llvm-svn: 210210
Make sure we appropriately retry calls to read if the return result is
less than what we asked for.
Additionally, check and handle IO errors: EINTR results in the read
operation getting restarted; other errors turn into exceptions.
llvm-svn: 210061
random_device::random_device(const string&) wrongly assumes that open
can only validly return a file descriptor greater than zero.
This results in random_device believing that it didn't successfully open
the device causing it to throw in it's constructor, this ends up leaking
a file descriptor.
The fix is simple, don't error on file descriptors which are zero.
llvm-svn: 210060
[syserr.errcat.objects]p4 specifies that
system_category().default_error_condition(ev) map to
error_condition(posv, generic_category()) if ev could map to a POSIX
errno.
Linux reserves up to and including 4095 for errno values, use this as a
bound.
This fixes syserr.errcat.objects/system_category.pass.cpp on Linux.
llvm-svn: 209795
compatibility to libstdc++. Move the implementation into a header for
easier sharing with libc++abi. Merge a number of improvements from that
version. Provide a POD definition for <stdexcept>'s public use to avoid
cast dances. Discussed with Marshall Clow.
llvm-svn: 207695
This is as straightforward as it sounds, a renamed from shared_mutex to
shared_timed_mutex.
Note that libcxx .dylib and .so files built with c++14 support need to
be rebuilt.
llvm-svn: 204078
The __sync_add_and_fetch() builtin parameter is volatile but clang has
'different' type checking and ends up accepting this code.
Undo the C++ cast from r198505 to get libc++/LLVM building with g++ while this
is investigated.
llvm-svn: 199494
We should check defined(__clang__) before the usage of the
clang diagnostic pragmas.
The [-Wswitch] warning in src/future.cpp should be ignored.
As the result, the equivalent GCC pragma is added.
llvm-svn: 197314
functions in src/support/win32/locale_win32.cpp and locale_win32.h,
calling upon vsnprintf for which there is a MingW correct alternative.
Note! __USE_MINGW_ANSI_STDIO is not modified in this patch. In order to
use the __mingw version it must be defined before including the MingW
headers.
llvm-svn: 195044
libsupc++ does not implement the dependent EH ABI and the
functionality it uses to implement std::exception_ptr (which it
declares as an alias of std::__exception_ptr::exception_ptr) is not
directly exported to clients. So we have little choice but to hijack
std::__exception_ptr::exception_ptr's (which fortunately has the
same layout as our std::exception_ptr) copy constructor, assignment
operator and destructor (which are part of its stable ABI), and its
rethrow_exception(std::__exception_ptr::exception_ptr) function.
Also, remove some out of date comments.
Differential Revision: http://llvm-reviews.chandlerc.com/D1826
llvm-svn: 192076
Linking against libstdc++, rather than libsupc++, is probably better
for people who need to link against clients of libstdc++. Because
libsupc++ is provided only as a static library, its globals are not
shared between the static library and the copy linked into libstdc++.
This has been found to cause at least one test failure.
This also removes a number of symbols which were multiply defined
between libstdc++ and libc++, only when linking with libstdc++.
Differential Revision: http://llvm-reviews.chandlerc.com/D1825
llvm-svn: 192075
The remaining multiple definitions were flushed out by attempting to
link libsupc++ and libc++ into the same executable with --whole-archive,
e.g.
clang++ -I../llvm/projects/libcxx/include -nodefaultlibs -Wl,--whole-archive lib/libc++.a /usr/lib/gcc/x86_64-linux-gnu/4.6/libsupc++.a -Wl,--no-whole-archive -lgcc -lgcc_s -lc -lpthread -lrt
(The same technique was used to flush out multiple definitions in
libstdc++.)
Differential Revision: http://llvm-reviews.chandlerc.com/D1824
llvm-svn: 192074
The issue this patch seeks to address is that MS's compiler (cl.exe) doesn't support the __attribute__((__weak__)) or __atribute__((__visibility__("default")) syntax; so a solution must be found where cl.exe doesn't see this syntax.
This patch seeks to solve this problem by changing code patterned like this:
__attribute__((__weak__, __visibility__("default")))
void* operator new(size_t size, const std::nothrow_t&) _NOEXCEPT { /*snip*/; return p; }
to code like this:
_LIBCPP_WEAK
void* operator new(size_t size, const std::nothrow_t&) _NOEXCEPT { return p; }
Howard: Thanks for all the comments regarding the default visibility
tag on the definition. I agree it isn't needed, and that there are lots
of other places where it is missing. That being said, I'm not wanting
to rock the boat on that issue right now. So I've added it back to the
definition via _LIBCPP_FUNC_VIS. A later pass dedicated just to this
issue can bring things in to a consistent state one way or the other.
Note that we do not want to have the exact same attributes on the
declaration and defintion in this case. The declaration should not be
marked weak, whereas the definition should (which is what G M's patch
did). I've fully tested on OS X to ensure that the resultant attribute
syntax actually works.
llvm-svn: 192007
libsupc++ in typeinfo.cpp, bringing it into agreement with
exception.cpp. This fixes link errors due to duplicate symbols from
this translation unit.
llvm-svn: 191397
1. I had been detecting and trapping iterator == and \!= among iterators
in different containers as an error. But the trapping itself is actually
an error.
Consider:
#include <iostream>
#include <vector>
#include <algorithm>
template <class C>
void
display(const C& c)
{
std::cout << "{";
bool first = true;
for (const auto& x : c)
{
if (\!first)
std::cout << ", ";
first = false;
std::cout << x;
}
std::cout << "}\n";
}
int
main()
{
typedef std::vector<int> V;
V v1 = {1, 3, 5};
V v2 = {2, 4, 6};
display(v1);
display(v2);
V::iterator i = std::find(v1.begin(), v1.end(), 1);
V::iterator j = std::find(v2.begin(), v2.end(), 2);
if (*i == *j)
i = j; // perfectly legal
// ...
if (i \!= j) // the only way to check
v2.push_back(*i);
display(v1);
display(v2);
}
It is legal to assign an iterator from one container to another of the
same type. This is required to work. One might want to test whether or
not such an assignment had been made. The way one performs such a check
is using the iterator's ==, \!= operator. This is a logical and necessary
function and does not constitute an error.
2. I had a header circular dependence bug when _LIBCPP_DEBUG2 is defined.
This caused a problem in several of the libc++ tests.
Fixed.
3. There is a serious problem when _LIBCPP_DEBUG2=1 at the moment in that
std::basic_string is inoperable. std::basic_string uses __wrap_iterator
to implement its iterators. __wrap_iterator has been rigged up in debug
mode to support vector. But string hasn't been rigged up yet. This means
that one gets false positives when using std::string in debug mode. I've
upped std::string's priority in www/debug_mode.html.
llvm-svn: 187636
MSVC-specific, MSVCRT-specific, or Windows-specific. Because Clang can
also define _MSC_VER, and MSVCRT is not necessarily the only C runtime,
these macros should not be used interchangeably.
This patch divides all Windows-related bits into the aforementioned
categories. Two new macros are introduced:
- _LIBCPP_MSVC: Defined when compiling with MSVC. Detected using
_MSC_VER, excluding Clang.
- _LIBCPP_MSVCRT: Defined when using the Microsoft CRT. This is the default
when _WIN32 is defined.
This leaves _WIN32 for code using the Windows API.
This also corrects the spelling of _LIBCP_HAS_IS_BASE_OF to _LIBCPP_HAS_IS_BASE_OF.
Nico, please prepare a patch for CREDITS.TXT, thanks.
llvm-svn: 187593
code to specify what version of POSIX the system should provide. If
you want to check what version of POSIX is actually available, you're
supposed to test _POSIX_VERSION.
However, since sysconf() has been in POSIX since 1995, it's probably
safe to assume it's available on any system with a C++11 compiler,
especially if _SC_NPROCESSORS_ONLN is defined too. So no point in a
complicated preprocessor rule if just we unconditionally include
<unistd.h> (on non-Windows systems).
Also, I've added a #warning for to help porters detect when a suitable
implementation isn't detected at compile-time.
Howard: Matthew, can you patch CREDITS.TXT? Thanks.
llvm-svn: 185275
template typename deductions on swap<> (used in string.cpp). Use
decltype(errno) to replicate the type and qualifier information for holding the
errno value. Because errno is expected to be assignable, there is no need to
use typename std::remove_const<decltype(errno)>::type to hold the value.
llvm-svn: 173172
building against libsupc++ as the functions for which they are used are provided
by libsupc++. Simply preprocess them away when building against libsupc++.
llvm-svn: 173165
inlined. These do not need to be always-inlined for ABI stability because they are not exported beyond this source due to the unnamed namespace.
Also simplified use of the Wmissing-field-initializers pragma as was done for clang.
llvm-svn: 171202
functions to protect against duration and time_point overflow. Since
we're about to wait anyway, we can afford to spend a few more cycles on
this checking. I purposefully did not treat the timed try_locks with
overflow checking. This fixes
http://llvm.org/bugs/show_bug.cgi?id=13721 . I'm unsure if the standard
needs clarification in this area, or if this is simply QOI. The
<chrono> facilities were never intended to overflow check, but just to
not overflow if durations stayed within +/- 292 years.
llvm-svn: 162925
__time_get_storage<char> to match the initialization behavior in
__time_get_storage<wchar>. Without the initialization, valgrind
reports errors in the subsequent calls to strftime_l.
llvm-svn: 161196
localization/locale.categories/category.collate/category.ctype/locale.ctype.byname/is_1.pass.cpp
and scan_is.pass.cpp. The tests fail when the character class being
tested is compound, like ctype_base::alnum or ctype_base::graph,
because the existing series of conditionals in do_is an do_scan_is
will abort too early. For instance, if the character class being
tested is alnum, and the character is numeric, do_is will return false
because iswalpha_l will return false, 'result' becomes false, and the
'true' result from the later call to iswdigit_l ends up being ignored
. A similar problem exists in do_scan_is.
llvm-svn: 161192
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
based on the LIBCXXRT and _LIBCPPABI_VERSION defines, but those files
do not currently include <cxxabi.h> in the non __APPLE__ case. The
attached patch updates those files so that for non __APPLE__ builds
<cxxabi.h> is included if available or if LIBCXXRT is set. I'm
modeling this on the recent updates to exception.cpp.
llvm-svn: 160790
out of the box on Linux systems. If you're building against libc++abi, you
still need to make sure it can find <cxxabi.h> so it knows not to export
symbols which libc++abi provides.
llvm-svn: 155091
Solaris not providing some of the locales that the test suite uses.
Note: This depends on an xlocale (partial) implementation for Solaris and a
couple of fixed standard headers. These will be committed to a branch later
today.
llvm-svn: 151720
libc++ now mostly works on FreeBSD with libcxxrt and this patch applied to the base system:
http://lists.freebsd.org/pipermail/freebsd-hackers/attachments/20110920/e666632c/xlocale-0001.obj
Summary of tests on FreeBSD:
****************************************************
Results for /root/libcxx/test:
using FreeBSD clang version 3.0 (trunk 135360) 20110717
Target: x86_64-unknown-freebsd9.0
Thread model: posix
with -std=c++0x -stdlib=libc++ -I/root/libcxx/include -L/root/libcxx/build/lib
----------------------------------------------------
sections without tests : 1
sections with failures : 48
sections without failures: 1015
+ ----
total number of sections : 1064
----------------------------------------------------
number of tests failed : 145
number of tests passed : 4179
+ ----
total number of tests : 4324
****************************************************
(Many due to this clang version not supporting C++ atomics)
More fixes to follow...
llvm-svn: 140245
Weiming Zhao