There were a couple things in the comments of BenchmarkRunner.cpp (and
maybe other files, I can't really remember) that were bugging me. This
patch fixes a couple of minor issues specifically in BenchmarkRunner
like typos and makes a couple instances more clear.

The Python documentation [1] says that `PyImport_AppendInittab` should
be called before `Py_Initialize()`. Starting with Python 3.12, this is
enforced with a fatal error:

  Fatal Python error: PyImport_AppendInittab: PyImport_AppendInittab()
  may not be called after Py_Initialize()

This commit ensures we only modify the table of built-in modules if
Python hasn't been initialized. For Python embedded in LLDB, that means
this happen exactly once, before the first call to `Py_Initialize`,
which becomes a NO-OP after. However, when lldb is imported in an
existing Python interpreter, Python will have already been initialized,
but by definition, the lldb module will already have been loaded, so
it's safe to skip adding it (again).

This fixes #70453.

[1] https://docs.python.org/3.12/c-api/import.html#c.PyImport_AppendInittab

It seems the `Traits.md` file was turned into `Traits/_index.md` in
https://reviews.llvm.org/D153291, causing links to `Traits.md` to no
longer work (instead, `Traits` needs to be used).

Fixes #62128.

The getArchPointerBitWidth method provides direct access to the pointer
size for arithmetic or switch statements, instead of requiring clients
to use the isArch16Bit / isArch32Bit / isArch64Bit predicates.

Removed some redundant code. Should be NFC change.

…e-ops pass.

Reverts llvm/llvm-project#80493

This increased clang compile times by 0.5%. I'll figure out a less
expensive way to achieve this.

Currently running `statistics dump` will trigger lldb to load debug info
that's not yet loaded (eg. dwo files). Resulted in a delay in the
command return, which, can be interrupting.

This patch also added a new option `--load-all-debug-info` asking
statistics to dump all possible debug info, which will force loading all
debug info available if not yet loaded.

Use a Python Bytes and ByteArray object instead of Integers for
TestOwnedReferences and TestBorrowedReferences. These two tests were
failing when building against Python 3.12 because these Integer objects
had a refcount of 4294967296 (-1). I didn't dig into it, but I suspect
the Python runtime has adopted an optimization to decrease refcounting
traffic for these simple objects.

If the load is used by any signed setccs, we can use a sextload
instead of zextload. Then we don't have to give up on extending
the load.

Use VPBuilder to construct VPInstructions, which means there's no need
to manually inserting recipes.

Right now we've a nested ternary for the midpoint function, but this can
be simplified a bit more, using if statements. This also slightly
increases the readability of that function.

Reference: https://github.com/openbsd/src/blob/master/sys/sys/exec_elf.h

This patch adds the missing `subnormal -> normal` part for `fpext` in
`computeKnownFPClass`.
Fixes the miscompilation reported by
https://github.com/llvm/llvm-project/pull/80941#issuecomment-1947302100.

Private members of VPlanTransforms are only used inside
VPlanTransforms.cpp, just make them static.

This implements the loading of the tzdata.zi file and store its contents
in the tzdb struct.

This adds all required members except:
- the leap seconds,
- the locate_zone, and
- current_zone.

The class time_zone is incomplete and only contains the parts needed for
storing the parsed data.

The class time_zone_link is fully implemented including its non-member
functions.

Implements parts of:
- P0355 Extending <chrono> to Calendars and Time Zones
- P1614 The Mothership has Landed

Implements:
- P1982 Rename link to time_zone_link

Set DecoderNamespace and AssemblerPredicate in the base class for Real
instructions for each subtarget. This avoids some ad hoc "let" around
groups of instructions definitions, and fixes some missed cases like
BUFFER_GL0_INV_gfx10 which was missing DecoderNamespace.

This alters the lowering of G_COPYSIGN to support vector types. The
general idea is that we just lower it to vector operations using and/or
and a mask, which are now converted to a BIF/BIT/BSP.

In the process the existing AArch64LegalizerInfo::legalizeFCopySign can
be removed, replying on expanding the scalar versions to vector instead,
which just needs a small adjustment to allow widening scalars to
vectors.

This commit fixes memory leaks in sparse tensor integration tests by
adding `bufferization.dealloc_tensor` ops.

Note: Buffer deallocation will be automated in the future with the
ownership-based buffer deallocation pass, making `dealloc_tensor`
obsolete (only codegen path, not when using the runtime library).

If, like powi on windows, the libcall is unavailable we should fall back
to SDAG. Currently we try and generate a call to "".

There is no real reason to differentiate.

Extracts logic from `vector::isContiguousSlice` to check whether
the trailing dim of a memref are contiguous into a dedicated hook
in BuiitinTypes.{h|cpp}.

Follow-up for https://github.com/llvm/llvm-project/pull/76848.

Properly ignore parentheses in bitwise operators.

Closes #81515

Fixes #80579.

Fix some warnings by matching types.

This reverts commit e1164d06.

This patch introduces a new `fir.cuda_kernel_launch` operation to
represents the call to CUDA kernels with the chervon notation. The
chevrons values in the parse tree can be scalar integer expr or dim3
derived type. The operation describes the grid/block values explicitly
as i32 values.

It lowers the parse-tree call to this new operation.

assertRegexpMatches is a deprecated alias for assertRegex and has been
removed in Python 3.12. This wasn't an issue previously because we used
a vendored version of the unittest module. Now that we use the built-in
version this gets updated together with the Python version used to run
the test suite.