Named '.amdhsa_code_object_version'. This directive sets the
e_ident[ABIVERSION] in the ELF header, and should be used as the assumed
COV for the rest of the asm file.

This commit also weakens the --amdhsa-code-object-version CL flag.
Previously, the CL flag took precedence over the IR flag. Now the IR
flag/asm directive take precedence over the CL flag. This is implemented
by merging a few COV-checking functions in AMDGPUBaseInfo.h.

This patch fixes full build problems in
https://github.com/llvm/llvm-project/issues/78721 (the header problem).
The `libc.a` target can be built now.

As a separate issue, `check-libc` is failing because undefined symbols
from `libunwind`, which I do not actually know the reason yet. I will be
looking into it.

Spotted by inspection when trialling out
`_LIBCPP_REMOVE_TRANSITIVE_INCLUDES`. No other instances of `_LIPCPP_`
were found in the repo.

Add missing -fexceptions in test.

Follow-up to post-commit commens for b1bfe221.

Fix issue with constructor call being interpreted as functional cast and
considered for a replacement
with static cast or being removed as redundant.

Closes #57959

Clang-tidy 18 no longer has false positives with the spaceship operator.
Note that I'm quite sure there are more occurrences in our headers that
are not caught.

This relands https://github.com/llvm/llvm-project/pull/76659 with fixes
tested in https://github.com/llvm/llvm-project/pull/78746.

[X86][NFC] Remove unnecessary parameters for MaskedShiftAmountPats/MaskedRotateAmountPats and rename one_bit_patterns

This patch is to extract NFC in #78853 into a separate commit.

Installs the source files of the experimental libc++ modules. These
source files (.cppm) are used by the Clang to build the std and 
std.compat modules.

The design of this patch is based on a discussing in SG-15 on
12.12.2023. (SG-15 is the ISO C++ Tooling study group):

- The modules are installed at a location, that is not known to build 
  systems and compilers.
- Next to the library there will be a module manifest json file.
  This json file contains the information to build the module from the
  libraries sources. This information includes the location where the
  sources are installed. @ruoso supplied the specification of this json
  file.
- If possible, the compiler has an option to give the location of the
  module manifest file
  (https://github.com/llvm/llvm-project/pull/76451).

Currently there is no build system support, but it expected to be added
in the future.

Fixes: https://github.com/llvm/llvm-project/issues/73089

Let the std.compat module use the std module instead of duplicating the
exports.

Based on @ChuanqiXu9's suggestion in #71438.

Revert "Revert #76246 and #76083"

This reverts commit 5c150e7e.

Adds a small fix that should properly disable the tests on Windows.
Unfortunately the original poster has not provided feedback and the
original patch did not fail in the LLVM CI infrastructure.

Modules are known to fail on Windows due to non compliance of the
C library. Currently not having this patch prevents testing on other
platforms.

Implements parts of: `P2637R3` https://wg21.link/P2637R3
(https://eel.is/c++draft/variant.visit)

Implements:
`basic_format_arg.visit()`
`basic_format_arg.visit<R>()`
Deprecates:
`std::visit_format_arg()`

The tests are as close as possible to the non-member function tests.

To land after: https://github.com/llvm/llvm-project/pull/76447,
https://github.com/llvm/llvm-project/pull/76268



---------

Co-authored-by: Zingam <zingam@outlook.com>

Co-authored-by: Zingam <zingam@outlook.com>

The buffer deallocation pass checks the IR ("operation preconditions")
to make sure that there is no IR that is unsupported. In such a case,
the pass signals a failure.

The pass now rejects all ops with unknown memory effects. We do not know
whether such an op allocates memory or not. Therefore, the buffer
deallocation pass does not know whether a deallocation op should be
inserted or not.

Memory effects are queried from the `MemoryEffectOpInterface` interface.
Ops that do not implement this interface but have the
`RecursiveMemoryEffects` trait do not have any side effects (apart from
the ones that their nested ops may have).

Unregistered ops are now rejected by the pass because they do not
implement the `MemoryEffectOpInterface` and neither do we know if they
have `RecursiveMemoryEffects` or not. All test cases that currently have
unregistered ops are updated to use registered ops.

Currently, if `CLANG_HAVE_LIBXML` is defined, and the version of libxml2
is above 2.12.0, there will be two warnings when building clang.

warning: initializing 'xmlErrorPtr' (aka 'struct _xmlError *') with an
expression of type 'const xmlError *' (aka 'const struct _xmlError *')
discards qualifiers

Since this commit

https://gitlab.gnome.org/GNOME/libxml2/-/commit/45470611b047db78106dcb2fdbd4164163c15ab7,
libxml2 makes cmlGetLastError return a const error. This patch follows
libxml2. Making the result a const pointer should be compatible with
versions before 2.12.0.

Tested on ArchLinux with libxml2 2.12.3 installed.

There is already a "block inserted" notification (in
`OpBuilder::Listener`), so there should also be a "block removed"
notification.

The purpose of this change is to make the listener API more mature.
There is currently a gap between what kind of IR changes can be made and
what IR changes can be listened to. At the moment, the only way to
inform listeners about "block removal" is to send a manual
`notifyOperationModified` for the parent op (e.g., by wrapping the
`eraseBlock(b)` method call in `updateRootInPlace(b->getParentOp())`).
This tells the listener that *something* has changed, but it is somewhat
of an API abuse.

nonlazybind is for declarations, not for definitions. We could test the
behavior, but the output would be misleading.

Introduce a new `argument-within-domain` category that covers cases
where the given arguments make it impossible to produce a correct result
(or create a valid object in case of constructors). While the incorrect
result doesn't create an immediate problem within the library (like e.g.
a null pointer dereference would), it always indicates a logic error in
user code and is highly likely to lead to a bug in the program once the
value is used.

Follow up to #78862

These tests control hardening mode with `ADDITIONAL_COMPILE_FLAGS`, and
always set modes which use on TRAP. So we don't need to check
`libcpp-hardening-mode=fast`, and they must always fail with the current
HWASAN implementation.

This reverts commit aa04d2b7.

Meant to push this to my fork and did not realize I was on main and not
a separate branch.

Prepare for -fno-plt implementation.

Implements parts of: `P2637R3` https://wg21.link/P2637R3
(https://eel.is/c++draft/variant.visit)

Implements:
`variant.visit()`
`variant.visit<R>()`

The tests are as close as possible to the non-member function.

To land after: https://github.com/llvm/llvm-project/pull/76268



---------

Co-authored-by: Zingam <zingam@outlook.com>

This patch disables the mlgo-utils tests on Windows builders. MLGO is
not currently supported on Windows.

These tests were failing as some of them look for specific file paths
and the path conventions are different between Linux and Windows.

This patch updates the paths in the PR labelling config for the MLGO
label. In particular, the path for the new mlgo-utils subfolder under
llvm/utils has been added and two other files that were missed in the
original introduction.

The library files in the new mlgo-utils utilities folder only depend on
absl.logging. The builtin Python logging library is a direct drop-in
replacement here, so we can just change the include and drop the test
dependency.

This test demonstrates template instantiation via the interpreter code.
In order to do that we can allocate the object on the stack and extend
its lifetime by boxing it into a clang::Value.

That avoids the subtle problem where we call the new operator on an
object only known to the interpreter and we cannot destroy it from
compiled code since there is not suitable facility in clang::Value yet.

That should resolve the asan issues that was reported in
llvm/llvm-project#76218.

Don't insert cast ops for ops in `vector.mask` region in
`VectorDropLeadUnitDim`.

… return type.

The behavior now is consistent with the non template version.
Enabled and updated old test.