2021-07-06 10:39:01 -04:00
.. _using-libcxx:
2015-08-22 19:40:49 +00:00
============
Using libc++
============
.. contents ::
:local:
2021-07-06 10:39:01 -04:00
Usually, libc++ is packaged and shipped by a vendor through some delivery vehicle
(operating system distribution, SDK, toolchain, etc) and users don't need to do
anything special in order to use the library.
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
This page contains information about configuration knobs that can be used by
users when they know libc++ is used by their toolchain, and how to use libc++
when it is not the default library used by their toolchain.
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
Using a different version of the C++ Standard
=============================================
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
Libc++ implements the various versions of the C++ Standard. Changing the version of
the standard can be done by passing `` -std=c++XY `` to the compiler. Libc++ will
automatically detect what Standard is being used and will provide functionality that
matches that Standard in the library.
2015-08-22 19:40:49 +00:00
.. code-block :: bash
2021-07-06 10:39:01 -04:00
$ clang++ -std=c++17 test.cpp
2019-03-21 16:21:09 +00:00
2021-07-06 10:39:01 -04:00
.. warning ::
Using `` -std=c++XY `` with a version of the Standard that has not been ratified yet
is considered unstable. Libc++ reserves the right to make breaking changes to the
library until the standard has been ratified.
2019-03-21 16:21:09 +00:00
Implement <filesystem>
This patch implements the <filesystem> header and uses that
to provide <experimental/filesystem>.
Unlike other standard headers, the symbols needed for <filesystem>
have not yet been placed in libc++.so. Instead they live in the
new libc++fs.a library. Users of filesystem are required to link this
library. (Also note that libc++experimental no longer contains the
definition of <experimental/filesystem>, which now requires linking libc++fs).
The reason for keeping <filesystem> out of the dylib for now is that
it's still somewhat experimental, and the possibility of requiring an
ABI breaking change is very real. In the future the symbols will likely
be moved into the dylib, or the dylib will be made to link libc++fs automagically).
Note that moving the symbols out of libc++experimental may break user builds
until they update to -lc++fs. This should be OK, because the experimental
library provides no stability guarantees. However, I plan on looking into
ways we can force libc++experimental to automagically link libc++fs.
In order to use a single implementation and set of tests for <filesystem>, it
has been placed in a special `__fs` namespace. This namespace is inline in
C++17 onward, but not before that. As such implementation is available
in C++11 onward, but no filesystem namespace is present "directly", and
as such name conflicts shouldn't occur in C++11 or C++14.
llvm-svn: 338093
2018-07-27 03:07:09 +00:00
2022-07-19 10:44:06 -04:00
Enabling experimental C++ Library features
==========================================
2015-08-22 19:40:49 +00:00
2022-07-19 10:44:06 -04:00
Libc++ provides implementations of some experimental features. Experimental features
are either Technical Specifications (TSes) or official features that were voted to
the Standard but whose implementation is not complete or stable yet in libc++. Those
are disabled by default because they are neither API nor ABI stable. However, the
2022-07-20 10:42:04 -04:00
`` -fexperimental-library `` compiler flag can be defined to turn those features on.
2016-05-03 22:32:08 +00:00
2023-09-12 11:02:17 -04:00
The following features are currently considered experimental and are only provided
when `` -fexperimental-library `` is passed:
* The parallel algorithms library (`` <execution> `` and the associated algorithms)
* `` std::stop_token `` , `` std::stop_source `` and `` std::stop_callback ``
2023-06-19 13:32:58 +01:00
* `` std::jthread ``
2023-09-12 11:02:17 -04:00
* `` std::chrono::tzdb `` and related time zone functionality
2023-09-12 11:39:17 -04:00
* `` std::ranges::join_view ``
2023-09-12 11:02:17 -04:00
2016-05-03 22:32:08 +00:00
.. warning ::
2022-07-20 10:42:04 -04:00
Experimental libraries are experimental.
2022-07-19 10:44:06 -04:00
* The contents of the `` <experimental/...> `` headers and the associated static
2016-05-03 22:32:08 +00:00
library will not remain compatible between versions.
* No guarantees of API or ABI stability are provided.
2021-07-06 10:39:01 -04:00
* When the standardized version of an experimental feature is implemented,
2019-06-11 14:48:40 +00:00
the experimental feature is removed two releases after the non-experimental
version has shipped. The full policy is explained :ref: `here <experimental features>` .
2015-08-22 19:40:49 +00:00
2022-07-20 10:42:04 -04:00
.. note ::
On compilers that do not support the `` -fexperimental-library `` flag, users can
define the `` _LIBCPP_ENABLE_EXPERIMENTAL `` macro and manually link against the
appropriate static library (usually shipped as `` libc++experimental.a `` ) to get
access to experimental library features.
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
Using libc++ when it is not the system default
==============================================
On systems where libc++ is provided but is not the default, Clang provides a flag
called `` -stdlib= `` that can be used to decide which standard library is used.
Using `` -stdlib=libc++ `` will select libc++:
2015-08-22 19:40:49 +00:00
.. code-block :: bash
2021-07-06 10:39:01 -04:00
$ clang++ -stdlib=libc++ test.cpp
2015-08-22 19:40:49 +00:00
2021-09-07 12:55:48 -04:00
On systems where libc++ is the library in use by default such as macOS and FreeBSD,
this flag is not required.
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
.. _alternate libcxx:
Using a custom built libc++
===========================
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
Most compilers provide a way to disable the default behavior for finding the
standard library and to override it with custom paths. With Clang, this can
be done with:
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
.. code-block :: bash
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
$ clang++ -nostdinc++ -nostdlib++ \
-isystem <install>/include/c++/v1 \
-L <install>/lib \
-Wl,-rpath,<install>/lib \
-lc++ \
test.cpp
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
The option `` -Wl,-rpath,<install>/lib `` adds a runtime library search path,
which causes the system's dynamic linker to look for libc++ in `` <install>/lib ``
whenever the program is loaded.
2015-08-22 19:40:49 +00:00
2021-07-06 10:39:01 -04:00
GCC does not support the `` -nostdlib++ `` flag, so one must use `` -nodefaultlibs ``
instead. Since that removes all the standard system libraries and not just libc++,
the system libraries must be re-added manually. For example:
2015-08-22 19:40:49 +00:00
.. code-block :: bash
2021-07-06 10:39:01 -04:00
$ g++ -nostdinc++ -nodefaultlibs \
-isystem <install>/include/c++/v1 \
-L <install>/lib \
-Wl,-rpath,<install>/lib \
-lc++ -lc++abi -lm -lc -lgcc_s -lgcc \
test.cpp
2016-01-20 01:26:30 +00:00
GDB Pretty printers for libc++
2021-07-06 10:39:01 -04:00
==============================
2016-01-20 01:26:30 +00:00
2021-07-06 10:39:01 -04:00
GDB does not support pretty-printing of libc++ symbols by default. However, libc++ does
provide pretty-printers itself. Those can be used as:
2016-01-20 01:26:30 +00:00
2021-07-06 10:39:01 -04:00
.. code-block :: bash
2016-01-20 01:26:30 +00:00
2021-07-06 10:39:01 -04:00
$ gdb -ex "source <libcxx>/utils/gdb/libcxx/printers.py" \
-ex "python register_libcxx_printer_loader()" \
<args>
2016-11-13 23:00:30 +00:00
[libcxx] adds an include-what-you-use (IWYU) mapping file
This makes it possible for programmers to run IWYU and get more accurate
standard library inclusions. Prior to this commit, the following program
would be transformed thusly:
```cpp
// Before
#include <algorithm>
#include <vector>
void f() {
auto v = std::vector{0, 1};
std::find(std::ranges::begin(v), std::ranges::end(v), 0);
}
```
```cpp
// After
#include <__algorithm/find.h>
#include <__ranges/access.h>
#include <vector>
...
```
There are two ways to fix this issue: to use [comment pragmas](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUPragmas.md)
on every private include, or to write a canonical [mapping file](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUMappings.md)
that provides the tool with a manual on how libc++ is laid out. Due to
the complexity of libc++, this commit opts for the latter, to maximise
correctness and minimise developer burden.
To mimimise developer updates to the file, it makes use of wildcards
that match everything within listed subdirectories. A script has also
been added to ensure that the mapping is always fresh in CI, and makes
the process a single step.
Finally, documentation has been added to inform users that IWYU is
supported, and what they need to do in order to leverage the mapping
file.
Closes #56937.
Differential Revision: https://reviews.llvm.org/D138189
2022-11-17 07:36:37 +00:00
.. _include-what-you-use:
include-what-you-use (IWYU)
===========================
2023-08-15 12:14:27 -07:00
libc++ provides an IWYU `mapping file <https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUMappings.md> `_ ,
[libcxx] adds an include-what-you-use (IWYU) mapping file
This makes it possible for programmers to run IWYU and get more accurate
standard library inclusions. Prior to this commit, the following program
would be transformed thusly:
```cpp
// Before
#include <algorithm>
#include <vector>
void f() {
auto v = std::vector{0, 1};
std::find(std::ranges::begin(v), std::ranges::end(v), 0);
}
```
```cpp
// After
#include <__algorithm/find.h>
#include <__ranges/access.h>
#include <vector>
...
```
There are two ways to fix this issue: to use [comment pragmas](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUPragmas.md)
on every private include, or to write a canonical [mapping file](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUMappings.md)
that provides the tool with a manual on how libc++ is laid out. Due to
the complexity of libc++, this commit opts for the latter, to maximise
correctness and minimise developer burden.
To mimimise developer updates to the file, it makes use of wildcards
that match everything within listed subdirectories. A script has also
been added to ensure that the mapping is always fresh in CI, and makes
the process a single step.
Finally, documentation has been added to inform users that IWYU is
supported, and what they need to do in order to leverage the mapping
file.
Closes #56937.
Differential Revision: https://reviews.llvm.org/D138189
2022-11-17 07:36:37 +00:00
which drastically improves the accuracy of the tool when using libc++. To use the mapping file with
IWYU, you should run the tool like so:
.. code-block :: bash
2023-08-19 13:30:55 +02:00
$ include-what-you-use -Xiwyu --mapping_file=/path/to/libcxx/include/libcxx.imp file.cpp
[libcxx] adds an include-what-you-use (IWYU) mapping file
This makes it possible for programmers to run IWYU and get more accurate
standard library inclusions. Prior to this commit, the following program
would be transformed thusly:
```cpp
// Before
#include <algorithm>
#include <vector>
void f() {
auto v = std::vector{0, 1};
std::find(std::ranges::begin(v), std::ranges::end(v), 0);
}
```
```cpp
// After
#include <__algorithm/find.h>
#include <__ranges/access.h>
#include <vector>
...
```
There are two ways to fix this issue: to use [comment pragmas](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUPragmas.md)
on every private include, or to write a canonical [mapping file](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUMappings.md)
that provides the tool with a manual on how libc++ is laid out. Due to
the complexity of libc++, this commit opts for the latter, to maximise
correctness and minimise developer burden.
To mimimise developer updates to the file, it makes use of wildcards
that match everything within listed subdirectories. A script has also
been added to ensure that the mapping is always fresh in CI, and makes
the process a single step.
Finally, documentation has been added to inform users that IWYU is
supported, and what they need to do in order to leverage the mapping
file.
Closes #56937.
Differential Revision: https://reviews.llvm.org/D138189
2022-11-17 07:36:37 +00:00
2023-08-15 12:14:27 -07:00
If you would prefer to not use that flag, then you can replace `` /path/to/include-what-you-use/share/libcxx.imp ``
[libcxx] adds an include-what-you-use (IWYU) mapping file
This makes it possible for programmers to run IWYU and get more accurate
standard library inclusions. Prior to this commit, the following program
would be transformed thusly:
```cpp
// Before
#include <algorithm>
#include <vector>
void f() {
auto v = std::vector{0, 1};
std::find(std::ranges::begin(v), std::ranges::end(v), 0);
}
```
```cpp
// After
#include <__algorithm/find.h>
#include <__ranges/access.h>
#include <vector>
...
```
There are two ways to fix this issue: to use [comment pragmas](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUPragmas.md)
on every private include, or to write a canonical [mapping file](https://github.com/include-what-you-use/include-what-you-use/blob/master/docs/IWYUMappings.md)
that provides the tool with a manual on how libc++ is laid out. Due to
the complexity of libc++, this commit opts for the latter, to maximise
correctness and minimise developer burden.
To mimimise developer updates to the file, it makes use of wildcards
that match everything within listed subdirectories. A script has also
been added to ensure that the mapping is always fresh in CI, and makes
the process a single step.
Finally, documentation has been added to inform users that IWYU is
supported, and what they need to do in order to leverage the mapping
file.
Closes #56937.
Differential Revision: https://reviews.llvm.org/D138189
2022-11-17 07:36:37 +00:00
file with the libc++-provided `` libcxx.imp `` file.
2016-11-13 23:00:30 +00:00
2023-07-14 16:58:15 -07:00
.. _termination-handler:
Overriding the default termination handler
==========================================
When the library wants to terminate due to an unforeseen condition (such as a hardening assertion
failure), the program is aborted through a special verbose termination function. The library provides
a default function that prints an error message and calls `` std::abort() `` . Note that this function is
provided by the static or shared library, so it is only available when deploying to a platform where
the compiled library is sufficiently recent. On older platforms, the program will terminate in an
unspecified unsuccessful manner, but the quality of diagnostics won't be great.
2023-01-09 16:59:59 -05:00
However, users can also override that mechanism at two different levels. First, the mechanism can be
2023-07-14 16:58:15 -07:00
overridden at compile time by defining the `` _LIBCPP_VERBOSE_ABORT(format, args...) `` variadic macro.
2023-01-09 16:59:59 -05:00
When that macro is defined, it will be called with a format string as the first argument, followed by
a series of arguments to format using printf-style formatting. Compile-time customization may be
2023-07-14 16:58:15 -07:00
useful to get precise control over code generation, however it is also inconvenient to use in
2023-01-09 16:59:59 -05:00
some cases. Indeed, compile-time customization of the verbose termination function requires that all
translation units be compiled with a consistent definition for `` _LIBCPP_VERBOSE_ABORT `` to avoid ODR
violations, which can add complexity in the build system of users.
Otherwise, if compile-time customization is not necessary, link-time customization of the handler is also
possible, similarly to how replacing `` operator new `` works. This mechanism trades off fine-grained control
2023-07-14 16:58:15 -07:00
over the call site where the termination is initiated in exchange for better ergonomics. Link-time
customization is done by simply defining the following function in exactly one translation unit of your
program:
[libc++] Add a lightweight overridable assertion handler
This patch adds a lightweight assertion handler mechanism that can be
overriden at link-time in a fashion similar to `operator new`.
This is a third take on https://llvm.org/D121123 (which allowed customizing
the assertion handler at compile-time), and https://llvm.org/D119969
(which allowed customizing the assertion handler at runtime only).
This approach is, I think, the best of all three explored approaches.
Indeed, replacing the assertion handler in user code is ergonomic,
yet we retain the ability to provide a custom assertion handler when
deploying to older platforms that don't have a default handler in
the dylib.
As-is, this patch provides a pretty good amount of backwards compatibility
with the previous debug mode:
- Code that used to set _LIBCPP_DEBUG=0 in order to get basic assertions
in their code will still get basic assertions out of the box, but
those assertions will be using the new assertion handler support.
- Code that was previously compiled with references to __libcpp_debug_function
and friends will work out-of-the-box, no changes required. This is
because we provide the same symbols in the dylib as we used to.
- Code that used to set a custom __libcpp_debug_function will stop
compiling, because we don't provide that declaration anymore. Users
will have to migrate to the new way of setting a custom assertion
handler, which is extremely easy. I suspect that pool of users is
very limited, so breaking them at compile-time is probably acceptable.
The main downside of this approach is that code being compiled with
assertions enabled but deploying to an older platform where the assertion
handler didn't exist yet will fail to compile. However users can easily
fix the problem by providing a custom assertion handler and defining
the _LIBCPP_AVAILABILITY_CUSTOM_ASSERTION_HANDLER_PROVIDED macro to
let the library know about the custom handler. In a way, this is
actually a feature because it avoids a load-time error that one would
otherwise get when trying to run the code on the older target.
Differential Revision: https://reviews.llvm.org/D121478
2022-03-03 17:37:03 -05:00
.. code-block :: cpp
2022-07-25 13:43:47 -04:00
void __libcpp_verbose_abort(char const* format, ...)
[libc++] Add a lightweight overridable assertion handler
This patch adds a lightweight assertion handler mechanism that can be
overriden at link-time in a fashion similar to `operator new`.
This is a third take on https://llvm.org/D121123 (which allowed customizing
the assertion handler at compile-time), and https://llvm.org/D119969
(which allowed customizing the assertion handler at runtime only).
This approach is, I think, the best of all three explored approaches.
Indeed, replacing the assertion handler in user code is ergonomic,
yet we retain the ability to provide a custom assertion handler when
deploying to older platforms that don't have a default handler in
the dylib.
As-is, this patch provides a pretty good amount of backwards compatibility
with the previous debug mode:
- Code that used to set _LIBCPP_DEBUG=0 in order to get basic assertions
in their code will still get basic assertions out of the box, but
those assertions will be using the new assertion handler support.
- Code that was previously compiled with references to __libcpp_debug_function
and friends will work out-of-the-box, no changes required. This is
because we provide the same symbols in the dylib as we used to.
- Code that used to set a custom __libcpp_debug_function will stop
compiling, because we don't provide that declaration anymore. Users
will have to migrate to the new way of setting a custom assertion
handler, which is extremely easy. I suspect that pool of users is
very limited, so breaking them at compile-time is probably acceptable.
The main downside of this approach is that code being compiled with
assertions enabled but deploying to an older platform where the assertion
handler didn't exist yet will fail to compile. However users can easily
fix the problem by providing a custom assertion handler and defining
the _LIBCPP_AVAILABILITY_CUSTOM_ASSERTION_HANDLER_PROVIDED macro to
let the library know about the custom handler. In a way, this is
actually a feature because it avoids a load-time error that one would
otherwise get when trying to run the code on the older target.
Differential Revision: https://reviews.llvm.org/D121478
2022-03-03 17:37:03 -05:00
This mechanism is similar to how one can replace the default definition of `` operator new ``
and `` operator delete `` . For example:
.. code-block :: cpp
// In HelloWorldHandler.cpp
2022-07-25 13:43:47 -04:00
#include <version> // must include any libc++ header before defining the function (C compatibility headers excluded)
[libc++] Add a lightweight overridable assertion handler
This patch adds a lightweight assertion handler mechanism that can be
overriden at link-time in a fashion similar to `operator new`.
This is a third take on https://llvm.org/D121123 (which allowed customizing
the assertion handler at compile-time), and https://llvm.org/D119969
(which allowed customizing the assertion handler at runtime only).
This approach is, I think, the best of all three explored approaches.
Indeed, replacing the assertion handler in user code is ergonomic,
yet we retain the ability to provide a custom assertion handler when
deploying to older platforms that don't have a default handler in
the dylib.
As-is, this patch provides a pretty good amount of backwards compatibility
with the previous debug mode:
- Code that used to set _LIBCPP_DEBUG=0 in order to get basic assertions
in their code will still get basic assertions out of the box, but
those assertions will be using the new assertion handler support.
- Code that was previously compiled with references to __libcpp_debug_function
and friends will work out-of-the-box, no changes required. This is
because we provide the same symbols in the dylib as we used to.
- Code that used to set a custom __libcpp_debug_function will stop
compiling, because we don't provide that declaration anymore. Users
will have to migrate to the new way of setting a custom assertion
handler, which is extremely easy. I suspect that pool of users is
very limited, so breaking them at compile-time is probably acceptable.
The main downside of this approach is that code being compiled with
assertions enabled but deploying to an older platform where the assertion
handler didn't exist yet will fail to compile. However users can easily
fix the problem by providing a custom assertion handler and defining
the _LIBCPP_AVAILABILITY_CUSTOM_ASSERTION_HANDLER_PROVIDED macro to
let the library know about the custom handler. In a way, this is
actually a feature because it avoids a load-time error that one would
otherwise get when trying to run the code on the older target.
Differential Revision: https://reviews.llvm.org/D121478
2022-03-03 17:37:03 -05:00
2022-07-25 13:43:47 -04:00
void std::__libcpp_verbose_abort(char const* format, ...) {
2023-09-25 13:40:16 -04:00
std::va_list list;
2022-07-25 13:19:51 -04:00
va_start(list, format);
std::vfprintf(stderr, format, list);
va_end(list);
[libc++] Add a lightweight overridable assertion handler
This patch adds a lightweight assertion handler mechanism that can be
overriden at link-time in a fashion similar to `operator new`.
This is a third take on https://llvm.org/D121123 (which allowed customizing
the assertion handler at compile-time), and https://llvm.org/D119969
(which allowed customizing the assertion handler at runtime only).
This approach is, I think, the best of all three explored approaches.
Indeed, replacing the assertion handler in user code is ergonomic,
yet we retain the ability to provide a custom assertion handler when
deploying to older platforms that don't have a default handler in
the dylib.
As-is, this patch provides a pretty good amount of backwards compatibility
with the previous debug mode:
- Code that used to set _LIBCPP_DEBUG=0 in order to get basic assertions
in their code will still get basic assertions out of the box, but
those assertions will be using the new assertion handler support.
- Code that was previously compiled with references to __libcpp_debug_function
and friends will work out-of-the-box, no changes required. This is
because we provide the same symbols in the dylib as we used to.
- Code that used to set a custom __libcpp_debug_function will stop
compiling, because we don't provide that declaration anymore. Users
will have to migrate to the new way of setting a custom assertion
handler, which is extremely easy. I suspect that pool of users is
very limited, so breaking them at compile-time is probably acceptable.
The main downside of this approach is that code being compiled with
assertions enabled but deploying to an older platform where the assertion
handler didn't exist yet will fail to compile. However users can easily
fix the problem by providing a custom assertion handler and defining
the _LIBCPP_AVAILABILITY_CUSTOM_ASSERTION_HANDLER_PROVIDED macro to
let the library know about the custom handler. In a way, this is
actually a feature because it avoids a load-time error that one would
otherwise get when trying to run the code on the older target.
Differential Revision: https://reviews.llvm.org/D121478
2022-03-03 17:37:03 -05:00
std::abort();
}
// In HelloWorld.cpp
#include <vector>
int main() {
std::vector<int> v;
2023-07-14 16:58:15 -07:00
int& x = v[0]; // Your termination function will be called here if hardening is enabled.
[libc++] Add a lightweight overridable assertion handler
This patch adds a lightweight assertion handler mechanism that can be
overriden at link-time in a fashion similar to `operator new`.
This is a third take on https://llvm.org/D121123 (which allowed customizing
the assertion handler at compile-time), and https://llvm.org/D119969
(which allowed customizing the assertion handler at runtime only).
This approach is, I think, the best of all three explored approaches.
Indeed, replacing the assertion handler in user code is ergonomic,
yet we retain the ability to provide a custom assertion handler when
deploying to older platforms that don't have a default handler in
the dylib.
As-is, this patch provides a pretty good amount of backwards compatibility
with the previous debug mode:
- Code that used to set _LIBCPP_DEBUG=0 in order to get basic assertions
in their code will still get basic assertions out of the box, but
those assertions will be using the new assertion handler support.
- Code that was previously compiled with references to __libcpp_debug_function
and friends will work out-of-the-box, no changes required. This is
because we provide the same symbols in the dylib as we used to.
- Code that used to set a custom __libcpp_debug_function will stop
compiling, because we don't provide that declaration anymore. Users
will have to migrate to the new way of setting a custom assertion
handler, which is extremely easy. I suspect that pool of users is
very limited, so breaking them at compile-time is probably acceptable.
The main downside of this approach is that code being compiled with
assertions enabled but deploying to an older platform where the assertion
handler didn't exist yet will fail to compile. However users can easily
fix the problem by providing a custom assertion handler and defining
the _LIBCPP_AVAILABILITY_CUSTOM_ASSERTION_HANDLER_PROVIDED macro to
let the library know about the custom handler. In a way, this is
actually a feature because it avoids a load-time error that one would
otherwise get when trying to run the code on the older target.
Differential Revision: https://reviews.llvm.org/D121478
2022-03-03 17:37:03 -05:00
}
2022-07-25 13:43:47 -04:00
Also note that the verbose termination function should never return. Since assertions in libc++
catch undefined behavior, your code will proceed with undefined behavior if your function is called
and does return.
Furthermore, exceptions should not be thrown from the function. Indeed, many functions in the
library are `` noexcept `` , and any exception thrown from the termination function will result
in `` std::terminate `` being called.
2016-11-13 23:00:30 +00:00
Libc++ Configuration Macros
===========================
Libc++ provides a number of configuration macros which can be used to enable
2023-07-14 16:58:15 -07:00
or disable extended libc++ behavior, including enabling hardening or thread
safety annotations.
2016-11-13 23:00:30 +00:00
**_LIBCPP_ENABLE_THREAD_SAFETY_ANNOTATIONS** :
This macro is used to enable -Wthread-safety annotations on libc++'s
2021-04-17 20:34:06 +05:30
`` std::mutex `` and `` std::lock_guard `` . By default, these annotations are
2016-11-13 23:00:30 +00:00
disabled and must be manually enabled by the user.
2016-12-05 19:40:12 +00:00
2023-07-14 16:58:15 -07:00
**_LIBCPP_ENABLE_HARDENED_MODE** :
2023-09-12 12:01:36 -07:00
This macro is used to enable the :ref: `hardened mode <using-hardening-modes>` .
**_LIBCPP_ENABLE_SAFE_MODE** :
This macro is used to enable the :ref: `safe mode <using-hardening-modes>` .
2023-07-14 16:58:15 -07:00
**_LIBCPP_ENABLE_DEBUG_MODE** :
2023-09-12 12:01:36 -07:00
This macro is used to enable the :ref: `debug mode <using-hardening-modes>` .
2023-07-14 16:58:15 -07:00
2016-12-05 19:40:12 +00:00
**_LIBCPP_DISABLE_VISIBILITY_ANNOTATIONS** :
This macro is used to disable all visibility annotations inside libc++.
Defining this macro and then building libc++ with hidden visibility gives a
build of libc++ which does not export any symbols, which can be useful when
building statically for inclusion into another library.
2016-12-08 23:57:08 +00:00
2017-01-13 22:02:08 +00:00
**_LIBCPP_DISABLE_ADDITIONAL_DIAGNOSTICS** :
This macro disables the additional diagnostics generated by libc++ using the
`diagnose_if` attribute. These additional diagnostics include checks for:
2018-12-06 21:46:17 +00:00
* Giving `set` , `map` , `multiset` , `multimap` and their `unordered_`
counterparts a comparator which is not const callable.
* Giving an unordered associative container a hasher that is not const
callable.
2017-01-13 22:02:08 +00:00
2017-10-09 19:25:17 +00:00
**_LIBCPP_NO_VCRUNTIME** :
Microsoft's C and C++ headers are fairly entangled, and some of their C++
headers are fairly hard to avoid. In particular, `vcruntime_new.h` gets pulled
in from a lot of other headers and provides definitions which clash with
libc++ headers, such as `nothrow_t` (note that `nothrow_t` is a struct, so
there's no way for libc++ to provide a compatible definition, since you can't
have multiple definitions).
By default, libc++ solves this problem by deferring to Microsoft's vcruntime
headers where needed. However, it may be undesirable to depend on vcruntime
headers, since they may not always be available in cross-compilation setups,
or they may clash with other headers. The `_LIBCPP_NO_VCRUNTIME` macro
prevents libc++ from depending on vcruntime headers. Consequently, it also
prevents libc++ headers from being interoperable with vcruntime headers (from
the aforementioned clashes), so users of this macro are promising to not
attempt to combine libc++ headers with the problematic vcruntime headers. This
macro also currently prevents certain `operator new` /`operator delete`
replacement scenarios from working, e.g. replacing `operator new` and
expecting a non-replaced `operator new[]` to call the replaced `operator new` .
2022-09-01 12:02:58 +02:00
**_LIBCPP_DISABLE_NODISCARD_EXT** :
This macro disables library-extensions of `` [[nodiscard]] `` .
2022-08-19 15:41:56 +02:00
See :ref: `Extended Applications of [[nodiscard]] <nodiscard extension>` for more information.
[libc++] Add _LIBCPP_ENABLE_NODISCARD and _LIBCPP_NODISCARD_EXT to allow pre-C++2a [[nodiscard]]
Summary:
The `[[nodiscard]]` attribute is intended to help users find bugs where
function return values are ignored when they shouldn't be. After C++17 the
C++ standard has started to declared such library functions as `[[nodiscard]]`.
However, this application is limited and applies only to dialects after C++17.
Users who want help diagnosing misuses of STL functions may desire a more
liberal application of `[[nodiscard]]`.
For this reason libc++ provides an extension that does just that! The
extension must be enabled by defining `_LIBCPP_ENABLE_NODISCARD`. The extended
applications of `[[nodiscard]]` takes two forms:
1. Backporting `[[nodiscard]]` to entities declared as such by the
standard in newer dialects, but not in the present one.
2. Extended applications of `[[nodiscard]]`, at the libraries discretion,
applied to entities never declared as such by the standard.
Users may also opt-out of additional applications `[[nodiscard]]` using
additional macros.
Applications of the first form, which backport `[[nodiscard]]` from a newer
dialect may be disabled using macros specific to the dialect it was added. For
example `_LIBCPP_DISABLE_NODISCARD_AFTER_CXX17`.
Applications of the second form, which are pure extensions, may be disabled
by defining `_LIBCPP_DISABLE_NODISCARD_EXT`.
This patch was originally written by me (Roman Lebedev),
then but then reworked by Eric Fiselier.
Reviewers: mclow.lists, thakis, EricWF
Reviewed By: thakis, EricWF
Subscribers: llvm-commits, mclow.lists, lebedev.ri, EricWF, rjmccall, Quuxplusone, cfe-commits, christof
Differential Revision: https://reviews.llvm.org/D45179
llvm-svn: 342808
2018-09-22 17:54:48 +00:00
2019-03-12 20:10:06 +00:00
**_LIBCPP_DISABLE_DEPRECATION_WARNINGS** :
This macro disables warnings when using deprecated components. For example,
using `std::auto_ptr` when compiling in C++11 mode will normally trigger a
warning saying that `std::auto_ptr` is deprecated. If the macro is defined,
no warning will be emitted. By default, this macro is not defined.
[libc++] Add _LIBCPP_ENABLE_NODISCARD and _LIBCPP_NODISCARD_EXT to allow pre-C++2a [[nodiscard]]
Summary:
The `[[nodiscard]]` attribute is intended to help users find bugs where
function return values are ignored when they shouldn't be. After C++17 the
C++ standard has started to declared such library functions as `[[nodiscard]]`.
However, this application is limited and applies only to dialects after C++17.
Users who want help diagnosing misuses of STL functions may desire a more
liberal application of `[[nodiscard]]`.
For this reason libc++ provides an extension that does just that! The
extension must be enabled by defining `_LIBCPP_ENABLE_NODISCARD`. The extended
applications of `[[nodiscard]]` takes two forms:
1. Backporting `[[nodiscard]]` to entities declared as such by the
standard in newer dialects, but not in the present one.
2. Extended applications of `[[nodiscard]]`, at the libraries discretion,
applied to entities never declared as such by the standard.
Users may also opt-out of additional applications `[[nodiscard]]` using
additional macros.
Applications of the first form, which backport `[[nodiscard]]` from a newer
dialect may be disabled using macros specific to the dialect it was added. For
example `_LIBCPP_DISABLE_NODISCARD_AFTER_CXX17`.
Applications of the second form, which are pure extensions, may be disabled
by defining `_LIBCPP_DISABLE_NODISCARD_EXT`.
This patch was originally written by me (Roman Lebedev),
then but then reworked by Eric Fiselier.
Reviewers: mclow.lists, thakis, EricWF
Reviewed By: thakis, EricWF
Subscribers: llvm-commits, mclow.lists, lebedev.ri, EricWF, rjmccall, Quuxplusone, cfe-commits, christof
Differential Revision: https://reviews.llvm.org/D45179
llvm-svn: 342808
2018-09-22 17:54:48 +00:00
2017-02-17 03:25:08 +00:00
C++17 Specific Configuration Macros
-----------------------------------
**_LIBCPP_ENABLE_CXX17_REMOVED_FEATURES** :
This macro is used to re-enable all the features removed in C++17. The effect
is equivalent to manually defining each macro listed below.
2017-02-17 03:30:25 +00:00
**_LIBCPP_ENABLE_CXX17_REMOVED_AUTO_PTR** :
2021-05-24 18:36:17 -04:00
This macro is used to re-enable `auto_ptr` .
**_LIBCPP_ENABLE_CXX17_REMOVED_BINDERS** :
This macro is used to re-enable the `binder1st` , `binder2nd` ,
`pointer_to_unary_function` , `pointer_to_binary_function` , `mem_fun_t` ,
`mem_fun1_t` , `mem_fun_ref_t` , `mem_fun1_ref_t` , `const_mem_fun_t` ,
`const_mem_fun1_t` , `const_mem_fun_ref_t` , and `const_mem_fun1_ref_t`
class templates, and the `bind1st` , `bind2nd` , `mem_fun` , `mem_fun_ref` ,
and `ptr_fun` functions.
**_LIBCPP_ENABLE_CXX17_REMOVED_RANDOM_SHUFFLE** :
This macro is used to re-enable the `random_shuffle` algorithm.
**_LIBCPP_ENABLE_CXX17_REMOVED_UNEXPECTED_FUNCTIONS** :
This macro is used to re-enable `set_unexpected` , `get_unexpected` , and
`unexpected` .
[libc++] Add _LIBCPP_ENABLE_NODISCARD and _LIBCPP_NODISCARD_EXT to allow pre-C++2a [[nodiscard]]
Summary:
The `[[nodiscard]]` attribute is intended to help users find bugs where
function return values are ignored when they shouldn't be. After C++17 the
C++ standard has started to declared such library functions as `[[nodiscard]]`.
However, this application is limited and applies only to dialects after C++17.
Users who want help diagnosing misuses of STL functions may desire a more
liberal application of `[[nodiscard]]`.
For this reason libc++ provides an extension that does just that! The
extension must be enabled by defining `_LIBCPP_ENABLE_NODISCARD`. The extended
applications of `[[nodiscard]]` takes two forms:
1. Backporting `[[nodiscard]]` to entities declared as such by the
standard in newer dialects, but not in the present one.
2. Extended applications of `[[nodiscard]]`, at the libraries discretion,
applied to entities never declared as such by the standard.
Users may also opt-out of additional applications `[[nodiscard]]` using
additional macros.
Applications of the first form, which backport `[[nodiscard]]` from a newer
dialect may be disabled using macros specific to the dialect it was added. For
example `_LIBCPP_DISABLE_NODISCARD_AFTER_CXX17`.
Applications of the second form, which are pure extensions, may be disabled
by defining `_LIBCPP_DISABLE_NODISCARD_EXT`.
This patch was originally written by me (Roman Lebedev),
then but then reworked by Eric Fiselier.
Reviewers: mclow.lists, thakis, EricWF
Reviewed By: thakis, EricWF
Subscribers: llvm-commits, mclow.lists, lebedev.ri, EricWF, rjmccall, Quuxplusone, cfe-commits, christof
Differential Revision: https://reviews.llvm.org/D45179
llvm-svn: 342808
2018-09-22 17:54:48 +00:00
2022-07-07 19:07:03 +02:00
C++20 Specific Configuration Macros
-----------------------------------
2021-05-24 18:36:17 -04:00
**_LIBCPP_ENABLE_CXX20_REMOVED_FEATURES** :
This macro is used to re-enable all the features removed in C++20. The effect
is equivalent to manually defining each macro listed below.
**_LIBCPP_ENABLE_CXX20_REMOVED_ALLOCATOR_MEMBERS** :
This macro is used to re-enable redundant members of `allocator<T>` ,
including `pointer` , `reference` , `rebind` , `address` , `max_size` ,
`construct` , `destroy` , and the two-argument overload of `allocate` .
2022-06-15 10:55:55 +02:00
**_LIBCPP_ENABLE_CXX20_REMOVED_ALLOCATOR_VOID_SPECIALIZATION** :
This macro is used to re-enable the library-provided specializations of
`allocator<void>` and `allocator<const void>` .
Use it in conjunction with `_LIBCPP_ENABLE_CXX20_REMOVED_ALLOCATOR_MEMBERS`
to ensure that removed members of `allocator<void>` can be accessed.
2021-05-25 14:34:18 -04:00
**_LIBCPP_ENABLE_CXX20_REMOVED_BINDER_TYPEDEFS** :
This macro is used to re-enable the `argument_type` , `result_type` ,
`first_argument_type` , and `second_argument_type` members of class
templates such as `plus` , `logical_not` , `hash` , and `owner_less` .
2021-05-24 18:36:17 -04:00
**_LIBCPP_ENABLE_CXX20_REMOVED_NEGATORS** :
This macro is used to re-enable `not1` , `not2` , `unary_negate` ,
and `binary_negate` .
**_LIBCPP_ENABLE_CXX20_REMOVED_RAW_STORAGE_ITERATOR** :
This macro is used to re-enable `raw_storage_iterator` .
2021-07-02 17:08:36 +00:00
**_LIBCPP_ENABLE_CXX20_REMOVED_TYPE_TRAITS** :
2021-07-06 10:39:01 -04:00
This macro is used to re-enable `is_literal_type` , `is_literal_type_v` ,
2021-07-02 17:08:36 +00:00
`result_of` and `result_of_t` .
[libc++] Add _LIBCPP_ENABLE_NODISCARD and _LIBCPP_NODISCARD_EXT to allow pre-C++2a [[nodiscard]]
Summary:
The `[[nodiscard]]` attribute is intended to help users find bugs where
function return values are ignored when they shouldn't be. After C++17 the
C++ standard has started to declared such library functions as `[[nodiscard]]`.
However, this application is limited and applies only to dialects after C++17.
Users who want help diagnosing misuses of STL functions may desire a more
liberal application of `[[nodiscard]]`.
For this reason libc++ provides an extension that does just that! The
extension must be enabled by defining `_LIBCPP_ENABLE_NODISCARD`. The extended
applications of `[[nodiscard]]` takes two forms:
1. Backporting `[[nodiscard]]` to entities declared as such by the
standard in newer dialects, but not in the present one.
2. Extended applications of `[[nodiscard]]`, at the libraries discretion,
applied to entities never declared as such by the standard.
Users may also opt-out of additional applications `[[nodiscard]]` using
additional macros.
Applications of the first form, which backport `[[nodiscard]]` from a newer
dialect may be disabled using macros specific to the dialect it was added. For
example `_LIBCPP_DISABLE_NODISCARD_AFTER_CXX17`.
Applications of the second form, which are pure extensions, may be disabled
by defining `_LIBCPP_DISABLE_NODISCARD_EXT`.
This patch was originally written by me (Roman Lebedev),
then but then reworked by Eric Fiselier.
Reviewers: mclow.lists, thakis, EricWF
Reviewed By: thakis, EricWF
Subscribers: llvm-commits, mclow.lists, lebedev.ri, EricWF, rjmccall, Quuxplusone, cfe-commits, christof
Differential Revision: https://reviews.llvm.org/D45179
llvm-svn: 342808
2018-09-22 17:54:48 +00:00
2021-07-06 10:39:01 -04:00
[libc++] Add _LIBCPP_ENABLE_NODISCARD and _LIBCPP_NODISCARD_EXT to allow pre-C++2a [[nodiscard]]
Summary:
The `[[nodiscard]]` attribute is intended to help users find bugs where
function return values are ignored when they shouldn't be. After C++17 the
C++ standard has started to declared such library functions as `[[nodiscard]]`.
However, this application is limited and applies only to dialects after C++17.
Users who want help diagnosing misuses of STL functions may desire a more
liberal application of `[[nodiscard]]`.
For this reason libc++ provides an extension that does just that! The
extension must be enabled by defining `_LIBCPP_ENABLE_NODISCARD`. The extended
applications of `[[nodiscard]]` takes two forms:
1. Backporting `[[nodiscard]]` to entities declared as such by the
standard in newer dialects, but not in the present one.
2. Extended applications of `[[nodiscard]]`, at the libraries discretion,
applied to entities never declared as such by the standard.
Users may also opt-out of additional applications `[[nodiscard]]` using
additional macros.
Applications of the first form, which backport `[[nodiscard]]` from a newer
dialect may be disabled using macros specific to the dialect it was added. For
example `_LIBCPP_DISABLE_NODISCARD_AFTER_CXX17`.
Applications of the second form, which are pure extensions, may be disabled
by defining `_LIBCPP_DISABLE_NODISCARD_EXT`.
This patch was originally written by me (Roman Lebedev),
then but then reworked by Eric Fiselier.
Reviewers: mclow.lists, thakis, EricWF
Reviewed By: thakis, EricWF
Subscribers: llvm-commits, mclow.lists, lebedev.ri, EricWF, rjmccall, Quuxplusone, cfe-commits, christof
Differential Revision: https://reviews.llvm.org/D45179
llvm-svn: 342808
2018-09-22 17:54:48 +00:00
Libc++ Extensions
=================
This section documents various extensions provided by libc++, how they're
provided, and any information regarding how to use them.
.. _nodiscard extension:
Extended applications of `` [[nodiscard]] ``
------------------------------------------
The `` [[nodiscard]] `` attribute is intended to help users find bugs where
function return values are ignored when they shouldn't be. After C++17 the
C++ standard has started to declared such library functions as `` [[nodiscard]] `` .
However, this application is limited and applies only to dialects after C++17.
Users who want help diagnosing misuses of STL functions may desire a more
liberal application of `` [[nodiscard]] `` .
For this reason libc++ provides an extension that does just that! The
2022-09-01 12:02:58 +02:00
extension is enabled by default and can be disabled by defining `` _LIBCPP_DISABLE_NODISCARD_EXT `` .
The extended applications of `` [[nodiscard]] `` takes two forms:
[libc++] Add _LIBCPP_ENABLE_NODISCARD and _LIBCPP_NODISCARD_EXT to allow pre-C++2a [[nodiscard]]
Summary:
The `[[nodiscard]]` attribute is intended to help users find bugs where
function return values are ignored when they shouldn't be. After C++17 the
C++ standard has started to declared such library functions as `[[nodiscard]]`.
However, this application is limited and applies only to dialects after C++17.
Users who want help diagnosing misuses of STL functions may desire a more
liberal application of `[[nodiscard]]`.
For this reason libc++ provides an extension that does just that! The
extension must be enabled by defining `_LIBCPP_ENABLE_NODISCARD`. The extended
applications of `[[nodiscard]]` takes two forms:
1. Backporting `[[nodiscard]]` to entities declared as such by the
standard in newer dialects, but not in the present one.
2. Extended applications of `[[nodiscard]]`, at the libraries discretion,
applied to entities never declared as such by the standard.
Users may also opt-out of additional applications `[[nodiscard]]` using
additional macros.
Applications of the first form, which backport `[[nodiscard]]` from a newer
dialect may be disabled using macros specific to the dialect it was added. For
example `_LIBCPP_DISABLE_NODISCARD_AFTER_CXX17`.
Applications of the second form, which are pure extensions, may be disabled
by defining `_LIBCPP_DISABLE_NODISCARD_EXT`.
This patch was originally written by me (Roman Lebedev),
then but then reworked by Eric Fiselier.
Reviewers: mclow.lists, thakis, EricWF
Reviewed By: thakis, EricWF
Subscribers: llvm-commits, mclow.lists, lebedev.ri, EricWF, rjmccall, Quuxplusone, cfe-commits, christof
Differential Revision: https://reviews.llvm.org/D45179
llvm-svn: 342808
2018-09-22 17:54:48 +00:00
1. Backporting `` [[nodiscard]] `` to entities declared as such by the
standard in newer dialects, but not in the present one.
2021-04-05 14:56:03 -04:00
2. Extended applications of `` [[nodiscard]] `` , at the library's discretion,
2023-10-24 05:03:57 -07:00
applied to entities never declared as such by the standard. You can find
all such applications by grepping for `` _LIBCPP_NODISCARD_EXT `` .
[libc++] Support int8_t and uint8_t in integer distributions as an extension
In D125283, we ensured that integer distributions would not compile when
used with arbitrary unsupported types. This effectively enforced what
the Standard mentions here: http://eel.is/c++draft/rand#req.genl-1.5.
However, this also had the effect of breaking some users that were
using integer distributions with unsupported types like int8_t. Since we
already support using __int128_t in those distributions, it is reasonable
to also support smaller types like int8_t and its unsigned variant. This
commit implements that, adds tests and documents the extension. Note that
we voluntarily don't add support for instantiating these distributions
with bool and char, since those are not integer types. However, it is
trivial to replace uses of these random distributions on char using int8_t.
It is also interesting to note that in the process of adding tests
for smaller types, I discovered that our distributions sometimes don't
provide as faithful a distribution when instantiated with smaller types,
so I had to relax a couple of tests. In particular, we do a really bad
job at implementing the negative binomial, geometric and poisson distributions
for small types. I think this all boils down to the algorithm we use in
std::poisson_distribution, however I am running out of time to investigate
that and changing the algorithm would be an ABI break (which might be
reasonable).
As part of this patch, I also added a mitigation for a very likely
integer overflow bug we were hitting in our tests in negative_binomial_distribution.
I also filed http://llvm.org/PR56656 to track fixing the problematic
distributions with int8_t and uint8_t.
Supersedes D125283.
Differential Revision: https://reviews.llvm.org/D126823
2022-06-01 15:25:14 -04:00
2022-07-07 20:02:07 +02:00
Extended integral type support
------------------------------
2022-08-10 17:34:45 -04:00
Several platforms support types that are not specified in the Standard, such as
the 128-bit integral types `` __int128_t `` and `` __uint128_t `` . As an extension,
libc++ does a best-effort attempt to support these types like other integral
types, by supporting them notably in:
2022-07-07 20:02:07 +02:00
* `` <bits> ``
* `` <charconv> ``
* `` <functional> ``
* `` <type_traits> ``
* `` <format> ``
* `` <random> ``
2022-07-15 07:42:17 +02:00
2022-08-10 17:34:45 -04:00
Additional types supported in random distributions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The `C++ Standard <http://eel.is/c++draft/rand#req.genl-1.5> `_ mentions that instantiating several random number
distributions with types other than `` short `` , `` int `` , `` long `` , `` long long `` , and their unsigned versions is
undefined. As an extension, libc++ supports instantiating `` binomial_distribution `` , `` discrete_distribution `` ,
`` geometric_distribution `` , `` negative_binomial_distribution `` , `` poisson_distribution `` , and `` uniform_int_distribution ``
with `` int8_t `` , `` __int128_t `` and their unsigned versions.
2022-07-15 07:42:17 +02:00
Extensions to `` <format> ``
--------------------------
The exposition only type `` basic-format-string `` and its typedefs
`` format-string `` and `` wformat-string `` became `` basic_format_string `` ,
`` format_string `` , and `` wformat_string `` in C++23. Libc++ makes these types
available in C++20 as an extension.
2023-02-21 17:33:56 +01:00
For padding Unicode strings the `` format `` library relies on the Unicode
Standard. Libc++ retroactively updates the Unicode Standard in older C++
versions. This allows the library to have better estimates for newly introduced
Unicode code points, without requiring the user to use the latest C++ version
in their code base.
2023-06-27 18:39:59 +02:00
2022-04-03 17:43:52 +02:00
In C++26 formatting pointers gained a type `` P `` and allows to use
zero-padding. These options have been retroactively applied to C++20.
2023-08-20 12:30:57 +02:00
Extensions to the C++23 modules `` std `` and `` std.compat ``
----------------------------------------------------------
Like other major implementations, libc++ provides C++23 modules `` std `` and
`` std.compat `` in C++20 as an extension"
[libcxx] Allow string to use SSO in constant evaluation. (#66576)
Previously, libcxx forced all strings created during constant evaluation
to point to allocated memory. That was done due to implementation
difficultites, but it turns out not to be necessary. This patch permits
the use of SSO strings during constant evaluation, and also simplifies
the implementation.
This does have a downside in terms of enabling users to accidentally
write non-portable code, however, which I've documented in
UsingLibcxx.rst.
In particular, whether `constinit std::string x = "...";` will
successfully compile now depends on whether the string is smaller than
the SSO capacity -- in libc++, up to 22 bytes on 64-bit platforms, and
up to 10 bytes on 32-bit platforms. By comparison, libstdc++ and MSVC
have an SSO capacity of 15 bytes, except that in libstdc++,
constant-initialized strings cannot be used as function-locals because
the object contains a pointer to itself.
Closes #68434
2023-10-10 14:31:47 -04:00
Constant-initialized std::string
--------------------------------
As an implementation-specific optimization, `` std::basic_string `` (`` std::string `` ,
`` std::wstring `` , etc.) may either store the string data directly in the object, or else store a
pointer to heap-allocated memory, depending on the length of the string.
As of C++20, the constructors are now declared `` constexpr `` , which permits strings to be used
during constant-evaluation time. In libc++, as in other common implementations, it is also possible
to constant-initialize a string object (e.g. via declaring a variable with `` constinit `` or
`` constexpr `` ), but, only if the string is short enough to not require a heap allocation. Reliance
upon this should be discouraged in portable code, as the allowed length differs based on the
standard-library implementation and also based on whether the platform uses 32-bit or 64-bit
pointers.
.. code-block :: cpp
// Non-portable: 11-char string works on 64-bit libc++, but not on 32-bit.
constinit std::string x = "hello world";
// Prefer to use string_view, or remove constinit/constexpr from the variable definition:
constinit std::string_view x = "hello world";
std::string_view y = "hello world";
2023-07-09 17:54:11 +02:00
.. _turning-off-asan:
[ASan][libcxx] A way to turn off annotations for containers with a specific allocator
This revision is part of our efforts to support container annotations with (almost) every allocator.
That patch is necessary to enable support for most annotations (D136765). Without a way to turn off annotations, it's hard to use ASan with area allocators (no calls to destructors).
This is an answer to a request about it. This patch provides a solution to the aforementioned issue by introducing a new template structure `__asan_annotate_container_with_allocator`, which allows the disabling of container annotations for a specific allocator.
This patch also introduces `_LIBCPP_HAS_ASAN_CONTAINER_ANNOTATIONS_FOR_ALL_ALLOCATORS` FTM.
To turn off annotations, it is sufficient to create a template specialization with a false value using a [Unary Type Trait](https://en.cppreference.com/w/cpp/types/integral_constant).
The proposed structure is being used in the code enabling annotations for all allocators in `std::vector`, `std::basic_string`, and `std::deque`. (D136765 D146214 D146815)
Possibility to do it was added to ASan API in rGdd1b7b797a116eed588fd752fbe61d34deeb24e4 commit.
For context on not calling a destructor, look at https://eel.is/c++draft/basic.life#5 and notes there, you may also read a discussion in D136765.
Reviewed By: ldionne, philnik, #libc, hans
Spies: EricWF, mikhail.ramalho, #sanitizers, libcxx-commits, hans, vitalybuka
Differential Revision: https://reviews.llvm.org/D145628
2023-05-04 14:16:06 -07:00
Turning off ASan annotation in containers
2023-06-27 18:39:59 +02:00
-----------------------------------------
[ASan][libcxx] A way to turn off annotations for containers with a specific allocator
This revision is part of our efforts to support container annotations with (almost) every allocator.
That patch is necessary to enable support for most annotations (D136765). Without a way to turn off annotations, it's hard to use ASan with area allocators (no calls to destructors).
This is an answer to a request about it. This patch provides a solution to the aforementioned issue by introducing a new template structure `__asan_annotate_container_with_allocator`, which allows the disabling of container annotations for a specific allocator.
This patch also introduces `_LIBCPP_HAS_ASAN_CONTAINER_ANNOTATIONS_FOR_ALL_ALLOCATORS` FTM.
To turn off annotations, it is sufficient to create a template specialization with a false value using a [Unary Type Trait](https://en.cppreference.com/w/cpp/types/integral_constant).
The proposed structure is being used in the code enabling annotations for all allocators in `std::vector`, `std::basic_string`, and `std::deque`. (D136765 D146214 D146815)
Possibility to do it was added to ASan API in rGdd1b7b797a116eed588fd752fbe61d34deeb24e4 commit.
For context on not calling a destructor, look at https://eel.is/c++draft/basic.life#5 and notes there, you may also read a discussion in D136765.
Reviewed By: ldionne, philnik, #libc, hans
Spies: EricWF, mikhail.ramalho, #sanitizers, libcxx-commits, hans, vitalybuka
Differential Revision: https://reviews.llvm.org/D145628
2023-05-04 14:16:06 -07:00
`` __asan_annotate_container_with_allocator `` is a customization point to allow users to disable
`Address Sanitizer annotations for containers <https://github.com/google/sanitizers/wiki/AddressSanitizerContainerOverflow> `_ for specific allocators. This may be necessary for allocators that access allocated memory.
This customization point exists only when `` _LIBCPP_HAS_ASAN_CONTAINER_ANNOTATIONS_FOR_ALL_ALLOCATORS `` Feature Test Macro is defined.
For allocators not running destructors, it is also possible to `bulk-unpoison memory <https://github.com/google/sanitizers/wiki/AddressSanitizerManualPoisoning> `_ instead of disabling annotations altogether.
The struct may be specialized for user-defined allocators. It is a `Cpp17UnaryTypeTrait <http://eel.is/c++draft/type.traits#meta.rqmts> `_ with a base characteristic of `` true_type `` if the container is allowed to use annotations and `` false_type `` otherwise.
The annotations for a `` user_allocator `` can be disabled like this:
.. code-block :: cpp
#ifdef _LIBCPP_HAS_ASAN_CONTAINER_ANNOTATIONS_FOR_ALL_ALLOCATORS
template <class T>
struct std::__asan_annotate_container_with_allocator<user_allocator<T>> : std::false_type {};
#endif
Why may I want to turn it off?
2023-06-27 18:39:59 +02:00
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[ASan][libcxx] A way to turn off annotations for containers with a specific allocator
This revision is part of our efforts to support container annotations with (almost) every allocator.
That patch is necessary to enable support for most annotations (D136765). Without a way to turn off annotations, it's hard to use ASan with area allocators (no calls to destructors).
This is an answer to a request about it. This patch provides a solution to the aforementioned issue by introducing a new template structure `__asan_annotate_container_with_allocator`, which allows the disabling of container annotations for a specific allocator.
This patch also introduces `_LIBCPP_HAS_ASAN_CONTAINER_ANNOTATIONS_FOR_ALL_ALLOCATORS` FTM.
To turn off annotations, it is sufficient to create a template specialization with a false value using a [Unary Type Trait](https://en.cppreference.com/w/cpp/types/integral_constant).
The proposed structure is being used in the code enabling annotations for all allocators in `std::vector`, `std::basic_string`, and `std::deque`. (D136765 D146214 D146815)
Possibility to do it was added to ASan API in rGdd1b7b797a116eed588fd752fbe61d34deeb24e4 commit.
For context on not calling a destructor, look at https://eel.is/c++draft/basic.life#5 and notes there, you may also read a discussion in D136765.
Reviewed By: ldionne, philnik, #libc, hans
Spies: EricWF, mikhail.ramalho, #sanitizers, libcxx-commits, hans, vitalybuka
Differential Revision: https://reviews.llvm.org/D145628
2023-05-04 14:16:06 -07:00
There are a few reasons why you may want to turn off annotations for an allocator.
Unpoisoning may not be an option, if (for example) you are not maintaining the allocator.
* You are using allocator, which does not call destructor during deallocation.
* You are aware that memory allocated with an allocator may be accessed, even when unused by container.
[libcxx] Fix using std::wcout/wcin on Windows with streams configured in wide mode
On Windows, the underlying file descriptors for stdout/stdin/stderr
can be reconfigured to wide mode. In the default (narrow) mode, the
charset usually isn't utf8 (as libcxx assumes), but normally a locale
specific codepage (where each codepage only can represent a small
subset of unicode characters).
By configuring the stdout file descriptor to wide mode, the user can
output wchar_t based strings without convesion to the narrow charset.
Within libcxx, don't try to use codecvt to convert this to a narrow
character encoding, but output these strings as such with fputwc.
In wide mode, such strings could be output directly with fwrite too,
but if the file descriptor hasn't been configured in wide mode, that
breaks the output (which currently works reasonably). By always
outputting one character at a time with fputwc, it works regardless
of mode of the stdout file descriptor.
For the narrow output stream, std::cout, outputting (via fwrite)
does fail when the file descriptor is set to wide mode. This matches
how it behaves with both MS STL and GNU libstdc++ too, so this is
probably acceptable.
This fixes https://github.com/llvm/llvm-project/issues/46646, and
the downstream bugs https://github.com/mstorsjo/llvm-mingw/issues/145
and https://github.com/mstorsjo/llvm-mingw/issues/222.
Differential Revision: https://reviews.llvm.org/D146398
2023-03-15 12:11:28 +02:00
Platform specific behavior
==========================
Windows
-------
The `` stdout `` , `` stderr `` , and `` stdin `` file streams can be placed in
Unicode mode by a suitable call to `` _setmode() `` . When in this mode,
the sequence of bytes read from, or written to, these streams is interpreted
as a sequence of little-endian `` wchar_t `` elements. Thus, use of
`` std::cout `` , `` std::cerr `` , or `` std::cin `` with streams in Unicode mode
will not behave as they usually do since bytes read or written won't be
interpreted as individual `` char `` elements. However, `` std::wcout `` ,
`` std::wcerr `` , and `` std::wcin `` will behave as expected.
Wide character stream such as `` std::wcin `` or `` std::wcout `` imbued with a
locale behave differently than they otherwise do. By default, wide character
streams don't convert wide characters but input/output them as is. If a
specific locale is imbued, the IO with the underlying stream happens with
regular `` char `` elements, which are converted to/from wide characters
according to the locale. Note that this doesn't behave as expected if the
stream has been set in Unicode mode.