Major user-facing changes:
llvm/Bitcode/BitcodeReader.h no longer includes llvm/IR/ModuleSummaryIndex.h

Some statistics:
clang++ -E  -Iinclude -I../llvm/include ../llvm/lib/Bitcode/Reader/*.cpp -std=c++14 -fno-rtti -fno-exceptions | wc -l
after: 493335
before: 539640

Discourse thread on the topic: https://discourse.llvm.org/t/include-what-you-use-include-cleanup/

Differential Revision: https://reviews.llvm.org/D119091

Summary:
This patch changes the ClangLinkerWrapper to use the executable path
when searching for the lld binary. Previously we relied on the program
name. Also not finding 'llvm-strip' is not considered an error anymore
because it is an optional optimization.

As Pavel pointed out, on Apple Silicon "b main" stops at a point after
the variable has already been initialized. This patch updates the test
case to avoids that. I've also split the test into separate files so its
easier to reproduce the individual scenarios without having to build any
shared state.

This patch makes the uncontrovertial changes to the pipeline.

Reviewed By: ldionne, #libc

Spies: libcxx-commits, arichardson

Differential Revision: https://reviews.llvm.org/D119041

The vectorizer will choose at times to "vectorize" loops with a scalar
factor (VF=1) with interleaving (IC > 1). This can occasionally produce
better code than the unroller (notable for reductions where it can
produce independent reduction chains that are combined after the loop).
At times this is not very beneficial though, for example when runtime
checks are needed or when the scalar code requires predication.

This addresses the second point, preventing the vectorizer from
interleaving when the scalar loop will require predication. This
prevents it from making a bit of a mess, that is worse than the original
and better left for the unroller to unroll if beneficial. It helps
reverse some of the regressions from D118090.

Differential Revision: https://reviews.llvm.org/D118566

Add the class MultiAffineFunction which represents functions whose domain is an
IntegerPolyhedron and which produce an output given by a tuple of affine
expressions in the IntegerPolyhedron's ids.

Also add support for piece-wise MultiAffineFunctions, which are defined on a
union of IntegerPolyhedrons, and may have different output affine expressions
on each IntegerPolyhedron. Thus the function is affine on each individual
IntegerPolyhedron piece in the domain.

This is part of a series of patches leading up to parametric integer programming.

Depends on D118778.

Reviewed By: Groverkss

Differential Revision: https://reviews.llvm.org/D118779

Debug position data is cleared after ScheduleDAGMILive::schedule() due to it also calling placeDebugValues(). Make it so the data is not cleared after initial call to placeDebugValues since we will call it again after reverting a schedule.

Secondly, since we skip debug instructions when reverting the schedule on AMDGPU, all debug instructions are now moved to the end of the scheduling region. RegionEnd points to the beginning of this chunk of debug instructions since it was not incremented when a debug instruction was skipped. RegionBegin may also point to the same debug instruction if Unsched.front() is a debug instruction thus shrinking the region to 1. Fix RegionBegin and RegionEnd so that they point to the current beginning and ending before calling placeDebugValues() since both vars will be used as reference points to move debug instructions back.

Reviewed By: rampitec

Differential Revision: https://reviews.llvm.org/D119022

Separate heatmap from bolt and build it as standalone tool.

Reviewed By: maksfb

Differential Revision: https://reviews.llvm.org/D118946

Currently GCC produces lots of warnings. Most of them are `-Wattributes`, but these warnings are completly ignored by everybody. So let's disable -Wattributes and make the output cleaner.

Reviewed By: ldionne, #libc

Spies: libcxx-commits, mgorny

Differential Revision: https://reviews.llvm.org/D119140

to decrease sizeof(SymbolUnion) from 72 to 64 on ELF64 platforms.

Use a dummy `Undefined` to prevent null pointer dereference (though unused)
`*rel.sym` in InputSectionBase::relocateAlloc.

The relocation order may shuffle a bit, but otherwise there is no behavior
difference.

Add an additional access so the full runtime checks are still generated,
even after D119078.

Reviewed By: #libc, ldionne, Quuxplusone

Differential Revision: https://reviews.llvm.org/D119052

This reverts commit f9e3ca54.

The commit results in internal test failures. Test case provided offline.

By using IRBuilderBase instead of IRBuilder<> a forward declaration can
be used instead of including IRBuilder.h

This reverts commit d42765b3.

Skip the part of TestGdbRemoteMemoryAllocation.py that attempts to
allocate both writable and executable memory.

* Implement `FlatAffineConstraints::getConstantBound(EQ)`.
* Inject a simpler constraint for loops that have at most 1 iteration.
* Taking into account constant EQ bounds of FlatAffineConstraint dims/symbols during canonicalization of the resulting affine map in `canonicalizeMinMaxOp`.

Differential Revision: https://reviews.llvm.org/D119153

Add SPARC to the list of platforms for which we provide a full
unwind implementation which leads to _Unwind_Backtrace being defined within
libunwind.so.

Likewise for PPC (see D118320 for background).

Reviewed By: #libunwind, MaskRay, Arfrever

Differential Revision: https://reviews.llvm.org/D119068

This fixes TestGdbRemoteSingleStep.py and TestGdbRemote_vCont.py. This
patch updates the test to account for the possibility that the constants
are already materialized. This appears to behave differently between
embedded arm64 devices and Apple Silicon.

If we call CGOpenCLRuntime::convertOpenCLSpecificType() multiple times
we should get the same type back.

Reviewed By: svenvh

Differential Revision: https://reviews.llvm.org/D119011

lld/ELF/OutputSections.cpp includes llvm/Config/config.h for
LLVM_ENABLE_ZLIB definition, but llvm/Config/config.h doesn't exist in
standalone build.

To fix this, this patch moves LLVM_ENABLE_ZLIB from config.h to
llvm-config.h and updates OutputSections.cpp to include llvm-config.h
instead of config.h

Reviewed By: MaskRay, mgorny

Differential Revision: https://reviews.llvm.org/D119058

This is a follow-up suggested in D119060.
Instead of checking each of the bottom 2 bits individually,
we can check them together and handle the possibility that
we demand both together.

https://alive2.llvm.org/ce/z/C2ihC2

Differential Revision: https://reviews.llvm.org/D119139

Roll forward of https://reviews.llvm.org/D117056 with a fix. Proper
initialization of `IsPrivateHeader` was missing, causing failures
on Linux.

The parsing of nested names is a little lax.  This corrects that.

1) The 'L' local name prefix cannot appear before a NestedName -- only
within it.  Let's remove that check from parseName, and then adjust
parseUnscopedName to allow it with or without the 'St' prefix.

2) In a nested name, a <template-param>, <decltype> or <substitution>
can only appear as the first element.  Let's enforce that.  Note I do
not remove these from the loop, to make the change easier to follow
(such a change will come later).

3) Given that, there's no need to special case 'St' outside of the
loop, handle it with the other 'S' elements.

4) There's no need to reset 'EndsWithTemplateArgs' after each
non-template-arg component.  Rather, always clear it and then set it
in the template-args case.

5) An template-args cannot immediately follow a template-args.

6) The parsing of a CDtor name with ABITags would attach the tags to
the NestedName node, rather than the CDTor node.  This is different to
how ABITags are attached to an unscopedName.  Make it consistent.

7) We remain with only CDTor and UnscopedName requireing construction
of a NestedName, so let's drop the PushComponent lambda.

8) Add some tests to catch the new rejected manglings.

Reviewed By: ChuanqiXu

Differential Revision: https://reviews.llvm.org/D118132

After aed965d5 we no longer demangle and store the full name. The
test was updated accordingly but the comment still specified that we
should be able to find the symbol by its full demangled name.

This makes the method compatible with opaque pointers.

Turns out there's still some code referencing this. No harm in keeping it in a
bit longer.

> as it was planned for removal in clang 15 and we're now past the branch point
>
> See https://github.com/llvm/llvm-project/issues/19107
>
> Differential revision: https://reviews.llvm.org/D118762

This reverts commit 564f9be1.

I dropped this argument in cdc0573f,
but did not adjust mlir. (Or rather I did, but I modified a generated
file ... oops)

We were dropping the [gs] modifier by parsing it in parseExpr, but not
forwarding it on to parseUnresolvedName.  This is the straightforwards
fix to forward that flag -- parseExpr must see past it.

Reviewed By: ChuanqiXu

Differential Revision: https://reviews.llvm.org/D118504

The StdQualifiedName node class is used for names exactly in the std
namespace.  It is not used for nested names that descend further --
those use a NestedName with NameType("std") as the scope.
Representing the compression scheme in the node graph is layer
breaking.  We can use the same structure for those exactly in std too,
and reduce code size a bit.

Reviewed By: ChuanqiXu

Differential Revision: https://reviews.llvm.org/D118249

AArch32/Armv8A  introduced the performance deprecation of certain patterns
of IT instructions.  After some debate internal to ARM, this is now being
reverted; i.e. no IT instruction patterns are performance deprecated
anymore, as the perfomance degredation is not significant enough.

This reverts the following:

"ARMv8-A deprecates some uses of the T32 IT instruction. All uses of
IT that apply to instructions other than a single subsequent 16-bit
instruction from a restricted set are deprecated, as are explicit
references to the PC within that single 16-bit instruction. This permits
the non-deprecated forms of IT and subsequent instructions to be treated
as a single 32-bit conditional instruction."

The deprecation no longer applies, but the behaviour may be controlled
by the -arm-restrict-it and -arm-no-restrict-it command-line options,
with the latter being the default. No warnings about complex IT blocks
will be generated.

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D118044

IRBuilderBase exists specifically to avoid the need for this.

LLVM Embedded Toolchains working group regular sync up calls to start in early
March, adding details to the table of sync ups for general reference.

Differential Revision: https://reviews.llvm.org/D118884

This is effectively inverting the transform added with D116804
because the downside of the false dependency of something like
"sbb %eax, %eax" is much greater than the upside of eliminating
a zeroing instruction on (all?) Intel CPUs.

Differential Revision: https://reviews.llvm.org/D118843