compiler-rt also doesn't support bits >= 64 as far as I know.

Add forward declarations for the few Instr classes we reference.

- test both 32 and 64 bit version
- probe the tail in dynamic-alloca
- generate more concise code

Differential Revision: https://reviews.llvm.org/D79482

We already have to include TargetLibraryInfo.h

This is directly defined in Instructions.h

Remove unused User.h include.
Replace SetVector.h with forward declaration.
Sort the forward declarations + remove FastMathFlags (defined in Operator.h).
Fix implicit SetVector.h dependency in LowerConstantIntrinsics.cpp.

This is directly defined in PassManager.h

This replaces the build_vector lowering code that was just added in
D80013
and matches the pattern later from the x86-specific "vzext_movl".
That seems to result in the same or better improvements and gets rid
of the 'TODO' items from that patch.

AFAICT, we always shrink wider constant vectors to 128-bit on these
patterns, so we still get the implicit zero-extension to ymm/zmm
without wasting space on larger vector constants. There's a trade-off
there because that means we miss potential load-folding.

Similarly, we could load scalar constants here with implicit
zero-extension even to 128-bit. That saves constant space, but it
means we forego load-folding, and so it increases register pressure.
This seems like a good middle-ground between those 2 options.

Differential Revision: https://reviews.llvm.org/D80131

Added pow intrinsic to LLVMIR dialect. Added a roundrip test for it.

Differential Revision: https://reviews.llvm.org/D80248

Split off PPCELFStreamer::emitPrefixedInstruction from
PPCELFStreamer::emitInstruction.

Differential Revision: https://reviews.llvm.org/D79626

```
StringMap<std::unique_ptr<MemoryBuffer>>
EmitDebugSections(llvm::DWARFYAML::Data &DI, bool ApplyFixups);
```
is unimplemented and unused.

For the supported binops (basic arithmetic, logicals + shifts), if we fail to simplify the demanded vector elts, then call SimplifyMultipleUseDemandedBits and try to peek through ops to remove unnecessary dependencies.

This helps with PR40502.

Differential Revision: https://reviews.llvm.org/D79003

Summary: 'A' constraint requires an immediate int or fp constant that can be inlined in an instruction encoding.

Reviewers: arsenm, rampitec

Differential Revision: https://reviews.llvm.org/D78494

DeclarationName is usually passed around by value as it's just a pointer.

Summary:

This is an attempt to fix https://bugs.llvm.org/show_bug.cgi?id=45988,
where SBValue::GetNumChildren returns 2, but SBValue::GetChildAtIndex(1) returns
an invalid value sentinel.

The root cause of this seems to be that GetNumChildren can return the number of
children of a wrong value. In particular, for pointers GetNumChildren just
recursively calls itself on the pointee type, so it effectively walks chains of
pointers. This is different from the logic of GetChildAtIndex, which only
recurses if pointee.IsAggregateType() returns true (IsAggregateType is false for
pointers and references), so it never follows chain of pointers.

This patch aims to make GetNumChildren (more) consistent with GetChildAtIndex by
only recursively calling GetNumChildren for aggregate types.

Ideally, GetNumChildren and GetChildAtIndex would share the code that decides
which pointers/references are followed, but that is a bit more invasive change.

Reviewers: teemperor, jingham, clayborg

Reviewed By: teemperor, clayborg

Subscribers: clayborg, labath, shafik, lldb-commits

Tags: #lldb

Differential Revision: https://reviews.llvm.org/D80254

Summary:
The cross-file rename feature is stable enough to enable it (has been
rolled out internally for a few weeks).

Reviewers: sammccall

Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D80507

Summary:
Make RAV not visit the default function decl by default.
Also update some stale comments on FunctionDecl::isDefault.

Fixes https://github.com/clangd/clangd/issues/383

Reviewers: sammccall, rsmith

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D80288

If a loop has a constant trip count known to be a multiple of MaxVF (times user
UF), LV infers that no tail will be generated for any chosen VF. This relies on
the chosen VF's being powers of 2 bound by MaxVF, and assumes MaxVF is a power
of 2. Make sure the latter holds, in particular when MaxVF is set by a memory
dependence distance which may not be a power of 2.

Differential Revision: https://reviews.llvm.org/D80491

__kmp_realloc_task_deque implicitly assumes, that the task queue is full
(ntasks == size), therefore tail = size in line 319.
An assertion is added to document this assumption.

The first check for a full queue is before the locking and might not hold
when the lock is taken. So, we need to check again for this condition when
we have the lock.

Reviewed By: AndreyChurbanov

Differential Revision: https://reviews.llvm.org/D80480

Generally:
 - found results using this method -> log
 - no results using this method -> vlog
 - method wasn't applied because ineligible -> no log

Summary: '|&' works with bash only, so it should not be used in regression
tests.

Differential Revision: https://reviews.llvm.org/D80501

We want to make sure that LINKER_IS_LLD_LINK is properly set - in
this case it shouldn't be set when building for MinGW.

Then we want to make the test for it correct and finally include
the option to build with thinlto cache since the MinGW driver now
supports that.

Differential Revision: https://reviews.llvm.org/D80493

* Enables using with more variadic sized operands;
* Generate convenience accessors for attributes;
  - The accessor are named the same as their name in ODS and returns attribute
    type (not convenience type) and no derived attributes.

This is first step to changing adapter to support verifying argument
constraints before the op is even created. This does not change the name of
adaptor nor does it require it except for ops with variadic operands to keep this change smaller.

Considered creating separate adapter but decided against that given operands also require attributes in general (and definitely for verification of operands and attributes).

Differential Revision: https://reviews.llvm.org/D80420

Fixes a build issue with libc++ configured with _LIBCPP_RAW_ITERATORS (ADL not effective)

```
llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp:1602:3: error: no matching function for call to 'transform'
  transform(HexString.begin(), HexString.end(), HexString.begin(), tolower);
  ^~~~~~~~~
```

Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D80475

Summary:
https://bugs.llvm.org/show_bug.cgi?id=46043

Git's config is generally of the format 'key=val', but a setting
'key=true' can be written as just 'key'.  The git-clang-format script
expects a value and crashes in this case; this change handles implicit
'true' values in the script.

Reviewers: MyDeveloperDay, krasimir, sammccall

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D80486

This makes many scenarios simpler by not requiring the user to write
ignoringImplicit() all the time, nor to account for non-visible
cxxConstructExpr() and cxxMemberCalExpr() nodes. This is also, in part,
inclusive of the equivalent of adding a use of ignoringParenImpCasts()
between all expr()-related matchers in an expression.

The pre-existing traverse(TK_AsIs, ...) matcher can be used to explcitly
match on implicit/invisible nodes. See

  http://lists.llvm.org/pipermail/cfe-dev/2019-December/064143.html

for more

Reviewers: aaron.ballman

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D72534

The fixed register constraint on EDX/RDX as an input
makes this not really commutable.

As is done by ignoreParenImpCasts(). We were not previously calling the
correct internal method.  Adjust tests to account for this.

If we're extracting an subvector from a broadcasted subvector of the same type then we can use the source vector directly.

If we're extracting an upper subvector from a broadcast we're better off extracting the lowest subvector instead as it avoids an actual extract instruction and might help SimplifyDemandedVectorElts further simplify the code.

EarlyCSE was added with D75145, but the motivating test is
not regressed by removing the extra pass now. That might be
because VectorCombine altered the way it processes instructions,
or it might be from (re)moving VectorCombine in the pipeline.

The extra round of EarlyCSE appears to cost approximately
0.26% in compile-time as discussed in D80236, so we need some
evidence to justify its inclusion here, but we do not have
that (yet).

I suspect that between SLP and VectorCombine, we are creating
patterns that InstCombine and/or codegen are not prepared for,
but we will need to reduce those examples and include them as
PhaseOrdering and/or test-suite benchmarks.

Currently we unconditionally get the first lane of the condition
operand, even if we later use the full vector condition. This can result
in some unnecessary instructions being generated.

Suggested as follow-up in D80219.