Allocate storage for them, initialize the first member with the given
value and the second member to 0.

GCC complains about "type qualifiers ignored on cast result type".
Upon investigation the correct fix was to remove all `volatile` and use the `-frounding-math` option.

As of the Bazel 6.x series, there is no difference between the
`exec_tools` and `tools`. Bazel 7 removes the `exec_tools` attribute
entirely. This commit updates to use the cannonical attribute name to
allow building `clang-tidy``with bazel 7.0.0, though it does not change
the default bazel version which remains at 6.1.2.

See also https://github.com/bazelbuild/bazel/issues/19132 for more
information.

For Windows, the compiler-rt profile library contains a polyfill
reimplementation of the mmap family of functions.

Previously, the runtime library exposed those symbols like, "mmap", in
the user symbol namespace. This could cause misdetections by configure
scripts that check for the "mmap" function just by linking, without
including headers.

Add a prefix on the symbols, and make an undeclared function static.

This fixes such an issue reported at
https://github.com/mstorsjo/llvm-mingw/issues/390.

Given clang18 is going to be released, this patch collects the modules
related change to release notes to make users have a clear
vision for modules.

PlatformDarwinKernel::GetSharedModule, which can find a kernel or kext
from a local filesystem scan, needed a little cleanup. The method which
finds kernels was (1) not looking for the SymbolFileSpec when creating a
Module, and (2) adding that newly created Module to a Target, which
GetSharedModule should not be doing - after auditing many other subclass
implementations of this method, I haven't found any others doing it.
Platform::GetSharedModule didn't have a headerdoc so it took a little
work to piece together the intended behaviors.

This is addressing a bug where
PlatformDarwinKernel::GetSharedModuleKernel would find the ObjectFile
for a kernel, create a Module, and add it to the Target. Then up in
DynamicLoaderDarwinKernel, it would check if the Module had a SymbolFile
FileSpec, and because it did not, it would do its own search for a
binary & dSYM, find them, and then add that to the Target. Now we have
two copies of the Module in the Target, one with a dSYM and the other
without, and only one of them has its load addresses set.

GetSharedModule should not be adding binaries to the Target, and it
should set the SymbolFile FileSpec when it is creating the Module.

rdar://120895951

Fixes: https://github.com/llvm/llvm-project/issues/78290



See the bug for more context.
```cpp
Gen ACoroutine() {
  if constexpr (0) // remove it make clang compile.
    co_return;
  co_await Gen{};
}
```
We miss symbol of ctor of promise_type if the first coroutine statement
happens to be inside the disabled branch of `if constexpr`.

This happens because the promise object is built when we see the first
coroutine statement which is present in
`ExpressionEvaluationContext::DiscardedStatement` context due to `if
constexpr (0)`. This makes clang believe that the promise constructor is
only odr-used and not really "used".

The expr evaluation context for the coroutine body should not be related
to the context in which the first coroutine statement appears. We
override the context to `PotentiallyEvaluated`.

---------

Co-authored-by: cor3ntin <corentinjabot@gmail.com>

Drop -menable-experimenta-extensions where it isn't needed.

This file has sections for non-experimental and experimental extensions,
but we keep forgetting to move things when we change the extension
status.

[RISCV] Don't look through EXTRACT_ELEMENT in lowerScalarInsert if the element types are different. (#78668)

If the element type of the vector we're extracting from doesn't match the type we're
inserting into, we can't directly insert or extract the subvector.

Currently ODS generates `populateDefaultAttrs` or
`populateDefaultProperties` based on whether the dialect opted into
usePropertiesForAttributes. But since individual ops might get opted
into using properties (as long as it has one property), it should
actually just check whether the op itself uses properties. Otherwise
`populateDefaultAttrs` will overwrite existing attrs inside properties
when creating an op. Understandably this becomes moot once everything
switches over to using properties, but this fixes it for now.

This PR makes ODS generate `populateDefaultProperties` as long as the op
itself uses properties.

This header first appeared in 4.4BSD and is provided by a number of C
libraries including Newlib. Several of our embedded projects use this
header and so to make LLVM libc a drop-in replacement, we need to
provide it as well.

For the initial commit, we only implement singly linked variants (SLIST
and STAILQ). The doubly linked variants (LIST, TAILQ and CIRCLEQ) can be
implemented in the future as needed.

routine, data, parallel, serial, kernels and loop construct all support
the device_type clause. This clause takes a list of device_type.
Previously the lowering code was assuming that the list s a single item.
This PR updates the lowering to handle any number of device_types.

This reverts commit 99cae9a4.

Temporarily until I reproduce and fix a linker issue:
```
FAILED: tools/flang/unittests/Frontend/FlangFrontendTests
...
/usr/bin/ld: tools/flang/unittests/Frontend/CMakeFiles/FlangFrontendTests.dir/CodeGenActionTest.cpp.o: undefined reference to symbol '_ZN4llvm11LLVMContextC1Ev'
/usr/bin/ld: /work1/omp-nightly/build/git/trunk18.0/build/llvm-project/lib/libLLVMCore.so.18git: error adding symbols: DSO missing from command line
```

Fix the bug where merge-fdata unconditionally outputs boltedcollection
line, regardless of whether input files have it set.

Test Plan:
Added bolt/test/X86/merge-fdata-nobat-mode.test which fails without this
fix.

This reverts commit 82bc33ea.

Accidentally pushed unrelated changes.

Fix the bug where merge-fdata unconditionally outputs boltedcollection 
line, regardless of whether input files have it set.

Test Plan:
Added bolt/test/X86/merge-fdata-nobat-mode.test which fails without this
fix.

Builtin templates e.g. `__make_integer_seq`, `__type_pack_element` are
such that they don't have alias *Decls*.
[D133262](https://reviews.llvm.org/D133262) marked these as alias
templates, resulting in an attempt to collect their null "using" Decls
within our `TargetFinder`.

This fixes https://github.com/clangd/clangd/issues/1906.

`ResumeIndex` isn't part of the frame struct header, so it necessarily
appears after the promise.

Co-authored-by: Yoni Lavi <yoni.lavi@nextsilicon.com>

Previously, we'd keyed the table by the vector type, but we were actually assigning the same cost for all the types with a common element type.  Unless we'd missed an entry, this means that effectively we were performing an SEW lookup.

Restructure the table to make this SEW dependence more explicit, and in the process greatly reduce the size of the table.

stores/insertelement instructions, NFC.

Reverts llvm/llvm-project#75882

To recover build bots :
https://lab.llvm.org/buildbot/#/builders/239/builds/5361
https://lab.llvm.org/buildbot/#/builders/168/builds/18126

This patch make vsetvl insertion pass could track the number of
inserted/removed vsetvl instruction from `-stats` option.

This adds minimal support for 7 new unprivileged extensions that were
defined as a part of
the RISC-V Profiles specification here:

https://github.com/riscv/riscv-profiles/blob/main/profiles.adoc#7-new-isa-extensions

* Ziccif: Main memory supports instruction fetch with atomicity
requirement
* Ziccrse: Main memory supports forward progress on LR/SC sequences
* Ziccamoa: Main memory supports all atomics in A
* Zicclsm: Main memory supports misaligned loads/stores
* Za64rs: Reservation set size of 64 bytes
* Za128rs: Reservation set size of 128 bytes
* Zic64b: Cache block size isf 64 bytes

As stated in the specification, these extensions don't add any new
features but
describe existing features. So this patch only adds parsing and
subtarget
features.

Also convert from std::vector to SmallVector.

This matches the ELF linker where these were moved into the ctx object
in 9a572164 and converted to SmallVector in ba948c5a.

These cause test build failures on Windows.

This reverts the following commits:
  57ca7484
  d06ae33e

This ports msan https://reviews.llvm.org/D14795 to dfsan.
dfsan, like msan, clears shadow for globals in a newly opened DSO in
case the DSO occupies the address of a previously labeled/poisoned area.
The operation should not happen on the main executable.

In addition, for a DT_EXEC executable, l_addr is zero and will lead to a
null pointer dereference in ForEachMappedRegion.

See lld/ELF/Config.h