Like other uses of ALLOW_RETRIES, this test tried to verify that an API
returned "quickly" but quick is not safe to define given slow and/or
busy machines.
Instead, we now verify that these "wait" APIs actually wait, which the
old test did not.
* improve coverage in `span`'s "conversion from `std::array`" test, while eliminating MSVC diagnostics about `testConstructorArray<T>() && testConstructorArray<const T, T>()` being redundant when `T` is already `const`.
* Remove use of `is_assignable` that triggers UB due to an insufficiently-complete type argument in `std::function`'s assignment operator test.
* Don't test that `shared_ptr` initialization from an rvalue triggers the lvalue aliasing constructor on non-libc++; this is not the case for Standard Libraries that implement LWG-2996. (Ditto, I'd simply remove this but it's your library ;).)
Differential Revision: https://reviews.llvm.org/D80030
We already set it using -rpath when linking test executables, and using
DYLD_LIBRARY_PATH causes problems when running other commands that
shouldn't run against the just-built libc++ (e.g. `ls` in a ShTest).
rdar://63241847
Since we're using the new testing format, DYLD_LIBRARY_PATH is not passed
to the compiler -- it's only passed to the programs we run as an argument
to the %{exec} substitution.
This is already handled by setting cxx_runtime_root instead -- I don't
see a reason to have two ways of setting the runtime path of the library
we're running against.
Because of Python's funny scoping rules with lambdas, we were always
using the value of `macro` as set in the last iteration of the loop.
This problem was introduced by e7bdfba4f0.
Otherwise, specifying (for example) the libc++.dylib from macos10.13
but the libc++abi.dylib from macos10.12 would end up adding library
paths for both the 10.12 and 10.13 dylibs, which would each contain
a copy of both libc++abi.dylib and libc++.dylib. By using a separate
directory for libc++.dylib and libc++abi.dylib, those do not conflict
anymore.
The back-deployment roots were updated to match this change.
This implements the relaxed requirements on the std::array constructors of span,
where the type only needs to be convertible to the element type of the span.
Note that the previous tests were not sufficient, as the const array<T, n> constructor
was only tested for compile time and the array<T, N> only during runtime.
Restructure the tests so that we can test conversions as well as both constructors.
Differential Revision: https://reviews.llvm.org/D75706
The default constructor of a static span requires _Extent == 0 so
SFINAE it out rather than using a static_assert
Differential Revision: https://reviews.llvm.org/D71994
unistd.h isn't guaranteed to exist when the target isn't Windows, in
particular if the target is bare-metal (i.e. no operating system).
Handle this by using __has_include instead, though in
filesystem/operations.cpp we already unconditionally include it so
just remove the extra include.
Differential Revision: https://reviews.llvm.org/D79784
These two tests do not use the "thread sleeps X milliseconds" pattern
that other libcxx tests use, so all we can do in order to remove
ALLOW_RETRIES workaround is remove the assumption that measuring the
"quick" return of `wait()` is possible (it is not). Let the test harness
verify overall that `wait()` does not hang.
As a bonus, have the spin-waiting threads `yield()`, which is what well
behaved code should do.
When grepping for unused features in the test suite, we will now find
those features and where they are defined, as opposed to thinking they
are dead features.
This test tried to verify that "wait()" returned quickly but "quick" is
impossible to define given a busy and/or slow system.
Instead, I've refactored the test to verify that `wait()` actually
waits which the old test did not verify.
Summary:
This LWG issue states that the result of `year_month_day_last::day()` is implementation defined if `ok()` is `false`.
However, from user perspective, calling `day()` in this situation will lead to a (possibly difficult to find) crash.
Hence, I have added an assertion to warn user at least when assertions are enabled.
I am however not aware of the libc++ stand on the desired behaviour.
Reviewers: ldionne, mclow.lists, EricWF, #libc
Reviewed By: ldionne, #libc
Subscribers: christof, dexonsmith, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D70346
Implements P0414R2:
* Adds support for array types in std::shared_ptr.
* Adds reinterpret_pointer_cast for shared_ptr.
Re-committing now that the leaking tests are fixed.
Differential Revision: https://reviews.llvm.org/D62259
Operating systems are best effort by default, so we cannot assume that
sleep-like APIs return as soon as we'd like.
Even if a sleep-like API returns when we want it to, the potential for
preemption means that attempts to measure time are subject to delays.
Implements P0414R2:
* Adds support for array types in std::shared_ptr.
* Adds reinterpret_pointer_cast for shared_ptr.
Differential Revision: https://reviews.llvm.org/D62259
Operating systems are best effort by default, so we cannot assume that
sleep-like APIs return as soon as we'd like.
Even if a sleep-like API returns when we want it to, the potential for
preemption means that attempts to measure time are subject to delays.
The challenge with measuring time in tests is that slow and/or busy
machines can cause tests to fail in unexpected ways. After this change,
three tests should be much more robust. The only remaining and tiny race
that I can think of is preemption after `--countDown`. That being said,
the race isn't fixable because the standard library doesn't provide a
way to count threads that are waiting to acquire a lock.
Reviewers: ldionne, EricWF, howard.hinnant, mclow.lists, #libc
Reviewed By: ldionne, #libc
Subscribers: dexonsmith, jfb, broadwaylamb, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D79406
This commit should will break libc++ without local submodule visibility, but
the LLVM+modules bots are now all using this mode. Before the Green Dragon
LLDB bot was failing to compile with a libc++ built with this commit as LSV
was disabled on macOS.
Original summary:
libc++ is careful to not fracture overload sets. When one overload
is visible to a user, all of them should be. Anything less causes
subtle bugs and ODR violations.
Previously, in order to support ::abs and ::div being supplied by
both <cmath> and <cstdlib> we had to do awful things that make
<math.h> and <stdlib.h> have header cycles and be non-modular.
This really breaks with modules.
Specifically the problem was that in C++ ::abs introduces overloads
for floating point numbers, these overloads forward to ::fabs,
which are defined in math.h. Therefore ::abs needed to be in math.h
too. But this required stdlib.h to include math.h and math.h to
include stdlib.h.
To avoid these problems the definitions have been moved to stddef.h
(which math includes), and the floating point overloads of ::abs
have been changed to call __builtin_fabs, which both Clang and GCC
support.
This change adds test coverage for the `codecvt<char16_t, char8_t, mbstate_t>` and `codecvt<char32_t, char8_t, mbstate_t>` ctype facets added to the C++20 WD by [P0482R6](https://wg21.link/P0428R6). Note that libc++ does not implement these facets despite implementing the remainder of P0482, presumably for ABI reasons, so these tests are marked `UNSUPPORTED: libc++`.
The ostream operator<< is currently broken for std::complex with
specified field widths.
This patch a partial revert of c3478eff7a (reviewed as D71214),
restoring the correct behavior.
Differential Revision: https://reviews.llvm.org/D78816
clock_gettime is documented to be available when _POSIX_TIMERS is
defined. Add a check for this.
Differential Revision: https://reviews.llvm.org/D79305
These tests fail due to a couple of changes to `move_iterator` for C++20:
* `move_iterator<I>::operator++(int)` returns `void` in C++20 if `I` doesn't model `forward_iterator`.
* `move_iterator<I>::reference` is calculated in C++20, so `I` must actually have an `operator*() const`.
Differential Revision: https://reviews.llvm.org/D79343
This patch adds deduction guides to <memory> to allow deducing
construction of shared_ptrs from unique_ptrs, and from weak_ptrs
and vice versa, as specified by C++17.
Differential Revision: https://reviews.llvm.org/D69603
This patch adds `_VSTD::` to some calls to `make_pair` inside the
implementations of searchers, to prevent things exploding if there is
a make_pair in an associated namespace of a user-defined type.
https://godbolt.org/z/xAFG98
Differential Revision: https://reviews.llvm.org/D72640
When running on remote hosts, we need the whole `echo 123 | %t.exe` command
to run on the remote host. Thus, we need to escape the pipe to make sure
the command is treated as `{ echo 123 | %t.exe } > %t.out` instead of
`{ echo 123 } | %t.exe > %t.out`m where only `echo 123` is run on the
remote host.
Running `export` when there is no environment variable to export will
cause the environment on the remote host to be printed. We don't want
that, so don't run any `export` command on the host when there's no env.
Since c0cd106fcc, we add __config_site macro defines to the compiler
command line whether we are building with modules or not. This means
that the modules tests are expected to fail on single-threaded systems
whether we build with modules or not.
Summary:
Instead of storing `static_test_env` (with all the symlinks) in the repo, we create it on the fly to be cross-toolchain-friendly. The primary use case for this are Windows-hosted cross-toolchains. Windows doesn't really have a concept of symlinks. So, when the monorepo is cloned, those symlinks turn to ordinary text files. Previously, if we cross-compiled libc++ for some symlink-friendly system (e. g. Linux) and ran tests on the target system, some tests would fail. This patch makes them pass.
Reviewers: ldionne, #libc
Reviewed By: ldionne, #libc
Subscribers: EricWF, dexonsmith, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D78200
Instead of storing static_test_env (with all the symlinks) in the repo,
we create it on the fly to be cross-toolchain-friendly. The primary
use case for this are Windows-hosted cross-toolchains. Windows doesn't
really have a concept of symlinks. So, when the monorepo is cloned,
those symlinks turn to ordinary text files. Previously, if we
cross-compiled libc++ for some symlink-friendly system (e. g. Linux) and
ran tests on the target system, some tests would fail. This patch makes
them pass.
Differential Revision: https://reviews.llvm.org/D78200
This commit removes minor features of the test suite that I've never
seen used and that are basically just a maintenance burden:
- color_diagnostics: Diagnostics are colored by default when running
from a terminal, and not colored otherwise. This is the right behavior.
Being able to tweak this has minor value, and could be achieved by
modifying the %{compile_flags} instead if absolutely needed.
- ccache: This can be achieved by using a wrapper for the %{cxx}
substitution.
- _dump_macros_verbose is just a dead function now.
When building with modules, always enable local submodule visibility.
It used to be disabled on Apple platforms, but it seems like we want
to use the same flags on Apple and Linux now (see https://reviews.llvm.org/D74892).
This commit migrates some of the Lit features from config.py to the new
DSL. This simplifies config.py and is a first step towards defining all
the features using the DSL instead of the complex logic in config.py.
Differential Revision: https://reviews.llvm.org/D78382
The tests were disabled under ASAN/MSAN because old Clangs were very
slow to build the test cases. However, I checked with the Clang used
on our build bots and the tests are not slow to build anymore, so the
tests can be re-enabled.
Android doesn't have a libgcc_s and uses libgcc instead, so adjust the
build accordingly. This matches compiler-rt's build setup. libc++abi and
libunwind were already checking for libgcc but in a different context.
This change makes them search only for libgcc on Android now, but the
code to link against libgcc if it were present was already there.
Reviewed By: #libc, #libc_abi, #libunwind, rprichard, srhines
Differential Revision: https://reviews.llvm.org/D78787
Summary:
pr42199
When using regex_constants::match_prev_avail, it is defined that
--first is valid, and match_not_bol and match_not_bow should be
ignored. At the moment these flags are not ignored. This fixis that.
Reviewers: ldionne, miyuki, EricWF, mclow.lists, #libc
Reviewed By: ldionne, miyuki, #libc
Subscribers: broadwaylamb, dexonsmith, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D75622
The libcxx.util utilities don't work properly, and we should remove them
when we get rid of compiler.py. In particular, libcxx.util.to_string
appears to be completely broken.
On Windows, quoting with single-quotes is both unnecessary and incorrect.
If the arguments are properly handled by Lit (which they are), quoting
should not be necessary.
The internal Lit shell requires the current working directory to exist.
This didn't show up locally because the directories were already created
by previous runs of the tests.
Tests that require support for Clang-verify are already marked as such
explicitly by their extension, which is .verify.cpp. Requiring the use
of an explicit Lit feature is, after thought, not really helpful.
This is a change in design: we have been bitten in the past by tests not
being enabled when we thought they were. However, the issue was mostly
with file extensions being ignored. The fix for that is not to blindly
require explicit features all the time, but instead to report all files
that are in the suite but that don't match any known test format. This
can be implemented in a follow-up patch.
This reverts commit 51a60ed14c, since the test still doesn't pass on
Windows. Marking the test as UNSUPORTED on Windows again until I've
figured out the problem.
Instead of completely disabling the tests on Apple, which makes them
disabled on all platforms we test (and hence useless), this commit
disables only the assertions that actually fail. I also created a
bug report to track re-enabling them (https://llvm.org/PR45739).
These two locale tests are disabled because they were said to "pass in
an uncontrolled manner on Apple platforms". This commit re-enables them
to see what that means, and whether that is still relevant on the
platforms we test.
Before this commit, the tests were either XFAILed or UNSUPPORTED on
Apple and Linux, which is pretty much all the systems we support. If
the tests truly don't work anywhere, they should be removed instead.
Since 88af3ddb1e, libc++ will prefer Python 3 when available. It is
available on Apple platforms, so subprocess.check_output will return
bytes instead of str. This lead to comparisons against str to be false,
and the MacOS platform not being detected properly.
This allows defining Lit features that can be enabled or disabled based
on compiler support, and parameters that are passed on the command line.
The main benefits are:
- Feature detection is entirely based on the substitutions provided in
the TestingConfig object, which is simpler and decouples it from the
complicated compiler emulation infrastructure.
- The syntax is declarative, which makes it easy to see what features
and parameters are accepted by the test suite. This is significantly
less entangled than the current config.py logic.
- Since feature detection is based on substitutions, it works really
well on top of the new format, and custom Lit configurations can be
created easily without being based on `config.py`.
This commit is a reapplication of 6d58030c8c, which was reverted in
8f24c4b72f because it broke Python 3 support. This re-application
supports Python 3.
Differential Revision: https://reviews.llvm.org/D78381
This relands this commit as it broke the LLDB bot the first time it landed.
See also the discussion on https://reviews.llvm.org/rG82b47b2978405f802a33b00d046e6f18ef6a47be
Since D74892 this code should now also work on macOS.
Original description:
libc++ is careful to not fracture overload sets. When one overload
is visible to a user, all of them should be. Anything less causes
subtle bugs and ODR violations.
Previously, in order to support ::abs and ::div being supplied by
both <cmath> and <cstdlib> we had to do awful things that make
<math.h> and <stdlib.h> have header cycles and be non-modular.
This really breaks with modules.
Specifically the problem was that in C++ ::abs introduces overloads
for floating point numbers, these overloads forward to ::fabs,
which are defined in math.h. Therefore ::abs needed to be in math.h
too. But this required stdlib.h to include math.h and math.h to
include stdlib.h.
To avoid these problems the definitions have been moved to stddef.h
(which math includes), and the floating point overloads of ::abs
have been changed to call __builtin_fabs, which both Clang and GCC
support.
Defining the nested types `reference` and `iterator_concept` of `reverse_iterator<I>` necessarily requires `I` to be complete in C++20. These tests that verify that `std::map<int, X>::reverse_iterator` can be instantiated when `X` is incomplete are going to have a bad time.
Differential Revision: https://reviews.llvm.org/D78944
Instead of using the libc++ headers provided alongside the toolchain,
use those in the sibling libcxx directory that we know is checked out.
Before the days of the monorepo, we couldn't assume that the libc++
repository was present when building libcxxabi. Since we can now make
that assumption, it's always better to use the version of libc++ that
is in lockstep with libc++abi, to avoid subtle bugs.
This allows defining Lit features that can be enabled or disabled based
on compiler support, and parameters that are passed on the command line.
The main benefits are:
- Feature detection is entirely based on the substitutions provided in
the TestingConfig object, which is simpler and decouples it from the
complicated compiler emulation infrastructure.
- The syntax is declarative, which makes it easy to see what features
and parameters are accepted by the test suite. This is significantly
less entangled than the current config.py logic.
- Since feature detection is based on substitutions, it works really
well on top of the new format, and custom Lit configurations can be
created easily without being based on `config.py`.
Differential Revision: https://reviews.llvm.org/D78381
The runtime for Blocks may not be available even though the Blocks
language extension _is_ available. Instead of potentially failing,
this commit is much more conservative and assumes the runtime for
Blocks is only provided on Apple platforms.
Differential Revision: https://reviews.llvm.org/D78757
The introduction of LIBCXX_HAS_MERGED_TYPEINFO_NAMES_DEFAULT changed
the default from =1 (assuming merged typeinfos) to =0 (not assuming
merged typeinfos) on all platforms where at least one other __config_site
macro is defined.
This commit explicitly enables the assumption of merged typeinfo names
on Apple platform to restore the previous behavior, at least until the
underlying issue has been fixed.
Instead of the ad-hoc #define _LIBCXX_DYNAMIC_FALLBACK, provide an option
to enable the setting when building libc++abi. Also use the occasion to
rename the option to something slightly more descriptive.
Note that in the future, it would be great to simply remove this option
altogether. However, in the meantime, it seems better to have it be an
official option than something ad-hoc.
Instead of having different names for the same Lit feature accross code
bases, use the same name everywhere. This NFC commit is in preparation
for a refactor where all three projects will be using the same Lit
feature detection logic, and hence it won't be convenient to use
different names for the feature.
Differential Revision: https://reviews.llvm.org/D78370
It turns out that all this time, we've actually been building without
assertions enabled in the dylib. This commit updates the Apple CMake
cache to make it consistent with reality.
When the new libc++ test format was enabled, warnings were accidentally
dropped cause they were not part of the %{compile_flags} substitution.
This commit adds them back, however `-Werror` is only used for non-verify
tests (cause it doesn't make sense for verify tests).
This commit is a re-application of 20fd624380, which was reverted in
5ec6fdb058 because it broke the C++03 bot. This failure should have
been fixed in b4fb705e77.
When the new libc++ test format was enabled, warnings were accidentally
dropped cause they were not part of the %{compile_flags} substitution.
This commit adds them back, however `-Werror` is only used for non-verify
tests (cause it doesn't make sense for verify tests).
Summary:
This patch add the dataflow option to LLVM_USE_SANITIZER and documents
it.
Tested via check-cxx (wip to fix the errors).
Reviewers: morehouse, #libc!
Subscribers: mgorny, cfe-commits, libcxx-commits
Tags: #clang, #libc
Differential Revision: https://reviews.llvm.org/D78390
Summary:
When I read this on the website it looks like the `--` in the used font turns
into an em dash. I updated this with inline literal mark up so the `--` will
remain obvious.
Reviewers: EricWF, #libc!
Subscribers: libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D78405
We previously tried re-exporting symbols that didn't exist when
exceptions were disabled. Note that building libc++abi without
exceptions still doesn't work when linking against the default-provided
libSystem.dylib, because it transitively depends on libobjc.dylib,
and that requires __gxx_personality_v0. But building libc++abi
with exceptions and libc++ without exceptions does work.
This was originally committed as f8452ddfcc and reverted in 7cb1aa9d93.
The issue was that shell builtins were being escaped too, and apparently
Bash won't execute a builtin when it is quoted e.g. '!'. Instead, it
thinks it's a command and it can't find it.
Re-committing the change with that issue fixed.
Cherrypick the upstream fix commit a77d5f7 onto llvm/utils/benchmark
and libcxx/utils/google-benchmark.
This fixes LLVM's 32-bit RISC-V compilation, and the issues
mentioned in https://github.com/google/benchmark/pull/955
An additional cherrypick of ecc1685 fixes some minor formatting
issues introduced by the preceding commit.
Differential Revision: https://reviews.llvm.org/D78084
Instead of creating Lit features for all __config_site macros automatically,
only do so for macros that generate features actually used in the test
suite. This makes it easier to know which ones are supported by the test
suite at a glance.
Note that the `libcpp-abi-version-vN` is dropped altogether, but it
wasn't used anywhere.
These substitutions are strongly tied to the operation of the test
format, so it makes sense to have them defined by the test format
instead of the Lit configuration. They should be defined regardless
of which configuration is in use.
Marked unsupported for C++03 and C++11 since this test uses alias
declarations, and at least one C++03 bot was failing with
-Wc++11-extensions.
Change-Id: I8c3a579edd7eb83e0bc74e85d116b68f22400161
The new format requires using an external shell, and as we transition
and we can simplify config.py as we transition to the new format. Also,
frankly, I'd be quite surprised if that setting was still working anyway
because we have several .sh.cpp tests that likely don't work in Lit's
internal shell.
When running the tests through `lit` directly instead of through `check-cxx`,
it is required to manually build the `cxx` (and often `cxx_experimental`)
targets. Instead of having to do that manually, this commit adds a new
target `check-cxx-deps` that does that for you.
When building these tests without modules in the whole test suite,
the __config_site macro definitions are not included anymore in the
%{compile_flags}. This causes the _LIBCPP_HAS_NO_THREADS define not
to be picked up, and the test to XPASS on single-threaded systems.
This is a stop-gap measure to fix the build bots, however the proper
solution would be to always pass the __config_site defines as compiler
macros, whether we build with modules or not.
Avoid using <sys/types.h> in those tests so that we can run them on
non-AIX systems (otherwise this test is basically dead-code on all
the build bots I'm aware of). Also, split up the test to allow using
.compile.pass.cpp tests instead of .sh.cpp tests, since that is the
last test referencing the %{compile} substitution explicitly.
Instead of being ShTests that use clang-verify (and without the proper
REQUIRES annotation), create .verify.cpp tests instead with the right
REQUIRES annotation.
By renaming .fail.cpp tests that don't need clang-verify to .compile.fail.cpp,
the new test format will not try to compile these tests with clang-verify,
and the old test format will work just the same. However, this allows
removing a workaround that requires parsing each test looking for
clang-verify markup.
After this change, a .fail.cpp test should always have clang-verify markup.
When clang-verify is not supported by the compiler, we will just check that
these tests fail to compile. When clang-verify is supported, these tests
will be compiled with clang-verify whether they have markup or not (so
they should have markup, or they will fail).
This simplifies the test suite and also ensures that all of our .fail.cpp
tests provide clang-verify markup. If it's impossible for a test to have
clang-verify markup, it can be moved to a .compile.fail.cpp test, which
are unconditionally just checked for compilation failure.
With this patch, .verify.cpp tests explicitly require clang-verify, but
no other test types require clang-verify out of the box. This will allow
making several .fail.cpp tests that don't have any clang-verify markup
to be just .compile.fail.cpp tests, which in turn should allow removing
a long standing workaround that requires parsing tests to detect whether
they have any clang-verify markup in them.
There are no such tests in the libc++ test suite, and I want to move
away from `.fail.cpp` tests (in favour of something else) too, which
require a workaround.
Instead of spamming a bunch of available features that are not actually
used anywhere, only set those that are actually used in the test suite.
In the future, this should probably be based on the target triple only,
with the ability to have wildcards in the triple.
Remove mentions of the ValgrindExecutor, which doesn't exist. That
executor is literally nowhere in the code base, so this is dead code
as far as we're concerned.
Also, inline a one-liner function that was called exactly once.
The libc++ test suite currently defines several features that are not
used anywhere in the tests, or that are redundant with other features.
For the purpose of simplifying config.py and to ease the bring up of a
new configuration, this commit removes some of these features:
- rename dylib-has-no-filesystem to c++filesystem-disabled, which exists
- rename apple-darwin to just darwin, which is already set
- remove useless setting of libstdc++, which is already set correctly
- remove libcpp-abi-unstable, which is not used anywhere
- remove the glibc-XXX features, which are not used anywhere
Summary: We discovered that the compiler may chose not to inline the operator=, which leads to an expensive extra stack frame. This change makes __assign_no_alias always tail called.
Reviewers: EricWF, #libc!
Subscribers: libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D77913
The libc++ test suite has a lot of old Lit features used to XFAIL tests
and mark them as UNSUPPORTED. Many of them are to workaround problems on
old compilers or old platforms. As time goes by, it is good to go and
clean those up to simplify the configuration of the test suite, and also
to reflect the testing reality. It's not useful to have markup that gives
the impression that e.g. clang-3.3 is supported, when we don't really
test on it anymore (and hence several new tests probably don't have the
necessary markup on them).
The use of the `&& ...` fold expression in std::array's deduction guides
recursively builds a set of binary operator expressions of depth N where
`N` is the number of elements in the initializer.
This is problematic because arrays may be large, and instantiation
depth is limited.
This patch addresses the issue by flattening the SFINAE using
the existing `__all` type trait.
Patch from Christopher Di Bella (cjdb@google.com)
Reviewed as https://reviews.llvm.org/D74291
Adds `std::same_as` to libc++. Since there aren't clang-format rules for
//requires-expressions//, I'll need to disable the formatter in certain areas.
Our documentation for running LIT is basically wrong, since it doesn't
mention `llvm-lit`, and nothing works without it. Note that this
documentation improvement reflects the current reality outside of
the recent efforts on improving the test suite. My goal is to document
the current way of running the test suite, so that everybody agrees on
how things currently work. As the configuration system for libc++ gets
simplified, we can come back and keep this documentation up to date to
reflect those changes.
Differential Revision: https://reviews.llvm.org/D77673
Summary: 5ade17e broke __is_pointer for Objective-C pointer types. This patch fixes the builtin and re-applies the change to type_traits.
Tags: #clang, #libc
Differential Revision: https://reviews.llvm.org/D77519
std::optional requires C++17. Instead of only running the test above
C++14, this commit removes uses of std::optional from the test (it's
already tested in block.objc.pass.mm).
Broke builders that emit different diagnostics. e.g.:
error: 'warning' diagnostics seen but not expected:
Line 13: alias declarations are a C++11 extension
Line 20: alias declarations are a C++11 extension
This reverts commit ff87813715.
It has never been used, and it actually doesn't really work because it
assumes that the target supports Python. Instead, it's better to just
use `!` since we're running ShTests in system shells anyway.
Instead of creating a temporary directory inside /tmp and running the
tests there, use a directory name based on LIT's %t substitution. This
has the benefit of not hitting /tmp so much (which is slow on some
filesystems). It also has the benefit that `ninja -C build clean` will
automatically remove the artifacts even if a test somehow failed to
remove its temporary directory (I've seen this happen when CTRL-C is
received).
This allows both the old and the new testing formats to handle these
tests with modules enabled.
We also include the modules flags in the %{flags} substitution, which
means that .sh.cpp tests in the old format and all tests in the new
format will use modules flags when enabled.
Otherwise, files don't link when using a GNU linker, which is more
sensitive on the order of the source file relative to the various
linked libraries. See http://c-faq.com/lib/libsearch.html for an
explanation of the problem.
Both test formats are equivalent, so this *should* not be a problem.
We've fixed a couple of failures uncovered by the first time we tried
making the switch, so this new attempt should go even farther.
If failures are noticed, it should be fine to revert this commit, but
please give a heads up afterwards so we know to address the issues!
Also note that it is still possible to use the old format by passing
`--param=use_old_format=True` when running Lit for the time being.
Summary:
This is a patch that Android has been carrying in its tree for several
years. This patch upstreams the existing ABI.
There's some historical cruft here. __regex_word used to be a part of
regex_traits rather than ctype_base. Bionic also used to use its own
ctype implementation because the libc++ builtin one wasn't available
yet. Bionic's ctype masks were 8 bits wide and already saturated, so a
wider type needed to be used for the regex mask, and the existing
value was already used so Android needed to specify its own.
Since then Android has migrated to the builtin ctype implementation
and this patch probably should have been dropped then. Unfortunately
that was not noticed at the time, so now we need to keep this to
maintain the current ABI.
Reviewers: EricWF, #libc, ldionne
Reviewed By: #libc, ldionne
Subscribers: dexonsmith, ldionne, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D76171
The LitConfig is shared across the whole test suite. However, since
enabling recursive expansion can be a breaking change for some test
suites, it's important to confine the setting to test suites that
enable it explicitly.
Note that other issues were raised with the way recursiveExpansionLimit
operates. However, this commit simply moves the setting to the right
place -- the mechanism by which it works can be improved independently.
Differential Revision: https://reviews.llvm.org/D77415
We had a workaround because GCC 5 does not evaluate static assertions
that are dependent on template parameters. This commit removes the
workaround and marks the corresponding tests as unsupported with GCC 5.
This has the benefit of bringing the new and the old test formats closer
without having to carry a workaround for an old compiler in the new
test format.
Otherwise, we're missing some flags like the flags that are used by
sanitizer builds and the 32-bit builds. In the long term, I think it
would be better to have only %{compile_flags} and %{link_flags}, but
for the benefit of adopting the new format by default, I think it's OK
to add %{flags} to it.
On Windows, we must make sure to close the temporary tar file before we
try to scp it.
This is an alternative approach to https://reviews.llvm.org/D77500.
This reverts commit 1580c76c4a.
This causes libcxx/selftest/newformat/sh.cpp/substitutions.sh.cpp to
fail with a linker error on Fedora 31 (x86-64) release (no assert)
builds.
Both test formats are equivalent, so this *should* not be a problem.
However, I'm taking advantage of the week-end to test this and see if
there are any failures. If so, it should be fine to revert this until
the failures have been addressed.
For the time being, it is still possible to use the old format by passing
`--param=use_old_format=True` when running Lit.
Summary:
Allow users to simultaneously enable address and undefined behavior
sanitizers, in the same manner that LLVM's 'HandleLLVMOptions.cmake'
allows.
Prior to this patch, `cmake -DLLVM_USE_SANITIZER="Address;Undefined"`
would succeed and the build would build most of the LLVM project with
`-fsanitize=address,undefined`, but a warning would be printed by
libcxx's CMake, and the build would use neither sanitizer. This
patch results in no warning being printed, and both sanitizers are used
in building libcxx.
Reviewers: jroelofs, EricWF, ldionne, #libc!
Subscribers: mgorny, dexonsmith, llvm-commits, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D77466
David Zarzycki