This reverts commit 42d669f8.

This adds 23 new helper functions to LLDB's CompilerType class, things
like IsSmartPtrType, IsPromotableIntegerType,
GetNumberofNonEmptyBaseClasses, and GetTemplateArgumentType (to name a
few).

These helper functions are needed as part of the implementation for the
Data Inspection Language, (see
https://discourse.llvm.org/t/rfc-data-inspection-language/69893).

Because `OpDot` does not require any extra capabilities or extensions,
enable it by default in the vector to spirv conversion.

Fixed #70789.

Fix for https://github.com/llvm/llvm-project/issues/68593

I don't know when, but at some point we lost test coverage to ensue that
all the headers are in the modulemap. This adds a test to make sure all
the headers (excluding a few which shouldn't be part of the modulemap)
are at least mentioned. This also fixes a few headers which bit-rotted
while we were missing the coverage.

Loop-rotate manually maintains dbg.value intrinsics -- it also needs to
manually maintain the replacement for dbg.value intrinsics, DPValue
objects. For the most part this patch adds parallel implementations
using the new type Some extra juggling is needed when loop-rotate hoists
loop-invariant instructions out of the loop: the DPValues attached to
such an instruction need to get rotated but not hoisted. Exercised by
the new test function invariant_hoist in dbgvalue.ll.

There's also a "don't insert duplicate debug intrinsics" facility in
LoopRotate. The value and correctness of this isn't clear, but to
continue preserving behaviour that's now tested in the "tak_dup"
function in dbgvalue.ll.

Other things in this patch include a helper DebugVariable constructor
for DPValues, a insertDebugValuesForPHIs handler for RemoveDIs
(exercised by the new tests), and beefing up the dbg.value checking in
dbgvalue.ll to ensure that each record is tested (and that there's an
implicit check-not).

As far as I can tell, the non-lib part never depended on MCJIT,
so I'm not sure why I added that. While here, re-sync some other
deps.

Since there's no shared library build for the gn build, no real
behavior change.

With intrinsics representing debug-info, we just clone all the
intrinsics when inlining a function and don't think about it any
further. With non-instruction debug-info however we need to be a bit
more careful and manually move the debug-info from one place to another.
For the most part, this means keeping a "cursor" during block cloning of
where we last copied debug-info from, and performing debug-info copying
whenever we successfully clone another instruction.

There are several utilities in LLVM for doing this, all of which now
need to manually call cloneDebugInfo. The testing story for this is not
well covered as we could rely on normal instruction-cloning mechanisms
to do all the hard stuff. Thus, I've added a few tests to explicitly
test dbg.value behaviours, ahead of them becoming not-instructions.

This pass isn't used anywhere upstream and thus has no test coverage.
For these reasons, remove it.

This patch adds a NewPM printer pass for the LazyValueAnalysis.

[libc++][NFC] Use __construct_at and __destroy_at instead of using preprocessor conditionals (#70866)

The tests were missing the `.pass.cpp` suffix to be recognized as tests.
As a result, they were never run and some are completely broken.

Diff Revision: https://reviews.llvm.org/D123235

Found while running libc++'s test suite with MSVC's STL.

This is structured into a series of commits for easier reviewing; I
could also split this into smaller PRs if desired.

* Add void-casts for `invoke_r` calls to fix MSVC STL `[[nodiscard]]`
warnings.
+ Our rationale is that if someone is calling `invoke_r<NonVoidType>`,
it sure looks like they care about the return value.
* Add `[[maybe_unused]]` to silence `-Wunused-parameter` warnings.
+ This happens because the parameters are used within `LIBCPP_ASSERT`,
which vanishes for MSVC's STL. This also motivates the following
changes.
* Add `[[maybe_unused]]` to fix `-Wunused-variable` warnings.
* Always void-cast `debug_comparisons` to fix `-Wunused-variable`
warnings.
+ As this was already unused with a void-cast in one
`_LIBCPP_HARDENING_MODE` branch, I'm simply lifting it next to the
variable definition.
* Add `[[maybe_unused]]` to fix `-Wunused-local-typedef` warnings.

Fixes #65890.

---------

Co-authored-by: Danny Mösch <danny.moesch@icloud.com>

ninja defaults to using ncpus + 2 build jobs, and this extra parallelism
helps the script complete faster in my testing on the GitHub runners.

Opaque ptr cleanup effort

This pass has no test coverage in upstream LLVM, is not used anywhere in
upstream LLVM, and has a NewPM equivalent. For these reasons, remove it.

12b0ab2c switched llvm-exegesis from using MCJIT to OrcJIT, which is a
different dep/library. This was ported over in the CMake build, but not
in the bazel build. This patch adjusts the llvm-exegesis deps in the
Bazel build to account for this change.

This is a follow-up to #73076. We need to reset output addresses for
deleted blocks, otherwise the address translation may mistakenly
attribute input address of a deleted block to a non-zero address.

While working on a test case, I've discovered that DWARF output ranges
were already broken for deleted basic blocks: #73428. I will provide a
test case for this PR with a DWARF address range fix PR.

Add "REQUIRES: asserts" to test as it requires the compiler to hit an assertion failure to pass and was failing in release builds.

Remove redundant check in updateOutputValues().

This will help tools like llvm-symbolizer recognizes more functions.

This pass has no test coverage upstream, is not used anywhere upstream,
and has a NewPM equivalent. For these reasons, remove it.

This pass isn't used by anywhere upstream and thus has no test coverage.
For these reasons, remove it.

This pass has a NewPM equivalent, isn't used anywhere upstream, and
doesn't have any test coverage. For these reasons, remove it.

This pass isn't used anywhere upstream, has a NewPM equivalent, and has
no test coverage. For these reasons, remove it.

This pass isn't used anywhere upstream and thus doesn't have any test
coverage. For these reasons, remove it.

This pass isn't used anywhere upstream and thus has no test coverage.
For these reasons, remove it.

This pass isn't used anywhere upstream and thus has no test coverage.
For these reasons it should be removed.

This fixes usage of PointerIntPair on 32-bit platform -
https://github.com/llvm/llvm-project/issues/73267.

This reverts commit ddc6ef46.

This relands commit 12b0ab2c.

There was an issue with certain configurations failing to build due to a
deleted Error constructor that should be fixed with this relanding.

Reland "[clang][Sema] Use original template pattern when declaring implicit deduction guides for nested template classes" (#73087)

Reland of f4183197 with proper handling
of template constructors

When a nested template is instantiated, the template pattern of the
inner class is not copied into the outer class
ClassTemplateSpecializationDecl. The specialization contains a
ClassTemplateDecl with an empty record that points to the original
template pattern instead.

As a result, when looking up the constructors of the inner class, no
results are returned. This patch finds the original template pattern and
uses that for the lookup instead.

Based on CWG2471 we must also substitute the known outer template
arguments when creating deduction guides for the inner class.

Changes from last iteration:

1. In template constructors, arguments are first rewritten to depth - 1
relative to the constructor as compared to depth 0 originally. These
arguments are needed for substitution into constraint expressions.
2. Outer arguments are then applied with the template instantiator to
produce a template argument at depth zero for use in the deduction
guide. This substitution does not evaluate constraints, which preserves
constraint arguments at the correct depth for later evaluation.
3. Tests are added that cover template constructors within nested
deduction guides for all special substitution cases.
4. Computation of the template pattern and outer instantiation arguments
are pulled into the constructor of
`ConvertConstructorToDeductionGuideTransform`.

This covers the simm12 offset case.