Changes

NFC Fix cost model checks in tests for debug intrinsics

These tests don't have a "Debug Info Version" module flag causing the debug
intrinsics to be dropped with a warning `warning: ignoring debug info with an
invalid version (0)`. The test passes because there are no CHECK lines for
these intrinsics. Fix the issue and auto-update the test checks.

Reviewed By: jryans

Differential Revision: https://reviews.llvm.org/D145573

[mlir] Move tosa.concat lowering from TosaToLinalg to TosaToTensor

tosa.concat is lowered to tensor.insert_slice thus it should be in
TosaToTensor rather than in TosaToLinalg.

Reviewed By: rsuderman

Differential Revision: https://reviews.llvm.org/D145952

[LLVM][OHOS] Clang toolchain and targets

Add a clang part of OpenHarmony target

Related LLVM part: D138202

~~~

Huawei RRI, OS Lab

Reviewed By: DavidSpickett

Differential Revision: https://reviews.llvm.org/D145227

[Assignment Tracking] getIntrinsicInstrCost: set dbg.assign cost to zero

Reviewed By: jmorse

Differential Revision: https://reviews.llvm.org/D144652

[Test] Add missing REQUIRES: asserts in test

[SCEV] Clarify ControlsExit comment (NFC)

ControlsExit implies a single exit branch.

[InstCombine] Regenerate test checks (NFC)

[SCEV] Fix finite loop non-strict predicate simplification (PR60944)

There are a number of issues with the current code for converting
ule -> ult (etc) predicates for comparisons controlling finite loops:

* It sets nowrap flags, which may only hold for that particular
  comparison, not globally. (PR60944)
* It doesn't check that the RHS is invariant. (I'm not sure this
  can cause practical issues independently of the previous point.)
* It runs before simplifications that may be more profitable. (PR54191)

This patch moves the handling for this into computeExitLimitFromICmp(),
because it is somewhat tightly coupled with assumptions in that code,
and addresses the aforementioned issues.

Fixes https://github.com/llvm/llvm-project/issues/60944.
Fixes https://github.com/llvm/llvm-project/issues/54191.

Differential Revision: https://reviews.llvm.org/D145510

[SCEV] Rename ControlsExit -> ControlsOnlyExit (NFC)

As suggested in https://reviews.llvm.org/D145510#4192162.

[TTI] Evaluate cost of experimental_widenable_condition as zero

This intrinsic is not supposed to live through lowering, eventually it should turn
into `true` constant and be optimized away.

Differential Revision: https://reviews.llvm.org/D146027
Reviewed By: skatkov

[AArch64] Add svboolx2_t and svboolx4_t tuple types

https://reviews.llvm.org/D145505

[DAG/AMDGPU] Use UniformityAnalysis in DAGISel

Switch DAGISel over to UniformityAnalysis, which was one of the last remaining users of the DivergenceAnalysis.
No explosions seen during internal testing so this looks like a smooth transition.

Reviewed By: sameerds

Differential Revision: https://reviews.llvm.org/D145918

[StructurizeCFG] Correctly depend on UniformityAnalysis

Small oversight in https://reviews.llvm.org/D145688 - the pass' dependency was not updated to reflect the change to UA.

Also, change DivergenceAnalysis to UniformityAnalysis in a comment. That way, StructurizeCFG only refers to UA and not DA anymore.

[IndVars] Option verify-indvars is broken (and always has been), delete it

This option is switched off by default, and it seems that it has never worked correctly.
What it basically does is: it remembers current BECount SCEV, and after all transforms
tries to validate some facts for it. However, between these two points this SCEV may
become invalid (e.g. because some SCEVUnknown it references may be deleted as dead
code). So basically it may work with broken pointers.

Besides, its implementation does strange things (e.g. forgetLoop) which are invasive and
may affect behavior in other parts of the system (specifically verification), concealing some
other problems. Another issue is that it may use SCEVCouldNotCompute object without
checking this.

The option is not used in any unit tests, and if switched on by default, the following tests
fail:
```
********************
Failed Tests (14):
  LLVM :: Transforms/IndVarSimplify/2005-06-15-InstMoveCrash.ll
  LLVM :: Transforms/IndVarSimplify/2007-06-06-DeleteDanglesPtr.ll
  LLVM :: Transforms/IndVarSimplify/2008-10-03-CouldNotCompute.ll
  LLVM :: Transforms/IndVarSimplify/2009-05-24-useafterfree.ll
  LLVM :: Transforms/IndVarSimplify/2011-10-27-lftrnull.ll
  LLVM :: Transforms/IndVarSimplify/ARM/code-size.ll
  LLVM :: Transforms/IndVarSimplify/X86/deterministic-scev-verify.ll
  LLVM :: Transforms/IndVarSimplify/X86/pr57187.ll
  LLVM :: Transforms/IndVarSimplify/X86/verify-scev.ll
  LLVM :: Transforms/IndVarSimplify/bbi-63564.ll
  LLVM :: Transforms/IndVarSimplify/invalidate-modified-lcssa-phi.ll
  LLVM :: Transforms/IndVarSimplify/loop-predication.ll
  LLVM :: Transforms/IndVarSimplify/post-inc-range.ll
  LLVM :: Transforms/IndVarSimplify/turn-to-invariant.ll

********************
Unexpectedly Passed Tests (1):
  LLVM :: Transforms/IndVarSimplify/pr55689.ll
```

None of these looks like real problems found by verification, these are
bugs in the verifying code itself (such as use of deleted SCEVs and
SCEVCouldNotCompute's).

I think it all gives enough justification for its removal.

https://github.com/llvm/llvm-project/issues/61302

Differential Revision: https://reviews.llvm.org/D145894
Reviewed By: nikic

[NFC] Drop NDEBUG around MSSA verification

Proposed on review of https://reviews.llvm.org/D145894.

[mlir][DialectUtils] Cleanup IndexingUtils and provide more affine variants while reusing implementations

Differential Revision: https://reviews.llvm.org/D145784

[clang-format] Treat &/&& as reference when followed by requires clause

Previously, the token annotator would get confused and annotate a member
function's ref qualifier as a binary operator, if said function also had
a requires clause after it.

This patch accounts for that, treating requires clauses more similarly
to `noexcept`, which also comes after the ref qualifier.

Fixes https://github.com/llvm/llvm-project/issues/61270

Reviewed By: HazardyKnusperkeks, owenpan

Differential Revision: https://reviews.llvm.org/D145656

Revert "[DAG/AMDGPU] Use UniformityAnalysis in DAGISel"

This reverts commit 0022b5803fd4f5a4e9fcf233267c0ffa1b88f763.

[RISCV][NFC] Share interleave mask checking logic

This adds two new methods to ShuffleVectorInst, isInterleave and
isInterleaveMask, so that the logic to check if a shuffle mask is an
interleave can be shared across the TTI, codegen and the interleaved
access pass.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D145971

Reapply [LowerTypeTests] Avoid creation of select constant expression

Reapply with a fix for phi handling: For phis, we need to insert
into the incoming block, not above the phi. This is especially
tricky if there are multiple incoming values from the same
predecessor, because these must all use the same value.

-----

LowerTypeTests replaces weak declarations with an icmp+select
constant expressions. As this is not a relocatable expression,
it additionally promotes initializers using it to global ctors.

As part of https://discourse.llvm.org/t/rfc-remove-most-constant-expressions/63179,
I would like to remove the select constant expression, of which LTT
is now the last user. This is a bit tricky, because we now need to
replace a constant with an instruction, which might require
converting intermediate constant expression users to instructions as
well.

We do this using the convertUsersOfConstantsToInstructions() helper.
However, it needs to be slightly extended to also support expansion
of ConstantAggregates. These are important in this context, because
the promotion of initializers to global ctors will produce stores
of such aggregates.

Differential Revision: https://reviews.llvm.org/D145247

[mlir][Affine] Add helper functions to allow reordering affine.apply operands and decompose the ops into smaller components

Care is taken to order operands from least hoistable to most hoistable and to process subexpressions in the same
order.

This allows exposing more oppportunities for licm, cse and strength reduction.

Such a step should typically be applied while we still have loops in the IR and just before lowering affine ops to arith.
This is because the affine.apply canonicalization currently tries to maximally compose chains of affine.apply operations
and could undo the effects of these decompositions.

Depends on: D145784

Differential Revision: https://reviews.llvm.org/D145685

[RISCV][test] Test case for regression when MachineOutliner and MachineCopyPropagation are both enabled

MachineCopyPropagation removes a register copy in the outlined function
as it doesn't see that it's live-out from the function.

[mlir][Transform] NFC - Various API cleanups and use RewriterBase in lieu of PatternRewriter

Depends on: D145685

Differential Revision: https://reviews.llvm.org/D145977

[X86] Add negative test for D145930

This patch adds a negative test for the issue discovered in D145930.

Differential Revision: https://reviews.llvm.org/D145933

[X86] Fix encoding for ATOMIC_LOGIC_OP

Fixes PR#61384.

Differential Revision: https://reviews.llvm.org/D145930

[Support][MemBuffer] Prevent UB on empty StringRefs

Empty StringRefs are usually identified by their length being zero, and
sometimes they'll have Data==nullptr (e.g. default constructed, or derived from
an operation like split/copy and result turned out to be empty).

If such StringRef objects are passed to llvm::MemoryBuffer::getMemBufferCopy,
it'll result in UB as neither src nor dst can be null, even if size is zero.

This patch prevents that UB by not issuing a copy whenever StringRef is empty.

Differential Revision: https://reviews.llvm.org/D144706

[mlir][GPUTransforms] NFC - Refactor GPUTransforms.cpp in preparation for improvements.

Depends on: D145977

Differential Revision: https://reviews.llvm.org/D145980

[Tooling/Inclusion] Index more sub std namespace symbols.

Fixes https://github.com/llvm/llvm-project/issues/61373

Differential Revision: https://reviews.llvm.org/D146021

Cleanup of Complex Deinterleaving pass (NFCI)

Differential Revision: https://reviews.llvm.org/D143177

[Codegen][ARM][AArch64] Support symmetric operations on complex numbers

Differential Revision: https://reviews.llvm.org/D142482

Remove an extra `//` in the IWYU pragma for gtest-matchers.h

Differential Revision: https://reviews.llvm.org/D146022

[clangd] Drop stale macro and mark ranges

I'll follow up with patching of those ranges, this is to stop bleeding
mentioned in https://github.com/clangd/clangd/issues/1537.

Differential Revision: https://reviews.llvm.org/D146024

[RISCV] Return false from shouldFormOverflowOp when type is i8 and i16

i8 and i16 are not using overflow.
Reduce the number of zero extension instructions.

To reduce the uncertainty of the unknown,
most of the checks of the virtual function are kept

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D143646

[clang][Sema] Avoid duplicate diagnostics for unreachable fallthrough attribute

This patch checks whether -Wunreachable-code-fallthrough is enabled
when clang encounters unreachable fallthrough attributes and, if so,
suppresses code will never be executed warning to avoid duplicate
warnings.

Fixes https://github.com/llvm/llvm-project/issues/60416
Differential Revision: https://reviews.llvm.org/D145842

[RISCV] Reject 'g' with explicit version in parseArchString

There is no versioning scheme for the 'g' shorthand for imafd (or in
current ISA specs, imafd_zifencei_zicsr). As such, the only sensible
behaviour to me seems to be to reject a version for it.

Differential Revision: https://reviews.llvm.org/D145954

[libc][NFC] Move memcmp implementations in subfolders

[mlir][Bazel] Adjust build after 0fa20ecafe0c3c7ffde413800eb4b1551b432273

[clang][sema][NFC] Make a few functions const

Differential Revision: https://reviews.llvm.org/D145947

[flang] Handle mismatches of procedure type args

Fortran allows type mismatch when passing actual arguments to
procedures and most cases were already being handled correctly by
Flang. However, conversion of data types to and from procedures and
conversion between procedures and char procedures were not always
handled properly. The missing cases were added and these
conversions are supported now.

Fixes #60550

Reviewed By: jeanPerier

Differential Revision: https://reviews.llvm.org/D145601

[mlir][Bazel] Adjust build after aafb52d7c9226cd9925bd5135309bd02b6e3b59d

Revert "Fix include order in CXType.cpp"

This reverts commit 4d55a0b512a17dfaa2461b8803d37b79f6c9691d.

The addition of clang_isBitFieldDecl is not needed because
clang_Cursor_isBitField already exists.

Revert "[libc][NFC] Move memcmp implementations in subfolders"

this broke arm32 build bots
This reverts commit 7f20e2a8d2ba73a8b05742903f4ae42afec82bff.

[reland][libc][NFC] Move memcmp implementations in subfolders

[AMDGPU] Don't run `llc-pipeline.ll` when expensive_checks are enabled

AMDGPU ISel can add extra passes when expensive checks are enabled. This means the pipeline can be reordered and the checks may fail.

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D146038

[NFC][OHOS] Temporarily switch off tests for windows

[libc] Correctly pass the compile options to the internal GPU compilation

Summary:
We use an internal option to create the GPU binary used for testing.
This wasn't getting the proper flags passed to it due to a missing
variable name.

[SLP]Fix PR61395: need to adjust vector factor after emitting shuffle
operation for combined entries.

The vector factor after combining of the shuffle entries is defined by
the size of the mask, not by the vector factors  of the original
entries. So, need to adjust it to emit correct code.

Reland: [DAG/AMDGPU] Use UniformityAnalysis in DAGISel

Switch DAGISel over to UniformityAnalysis, which was one of the last remaining users of the DivergenceAnalysis.
No explosions seen during internal testing so this looks like a smooth transition.

Reviewed By: sameerds

Differential Revision: https://reviews.llvm.org/D145918

[TTI] Treat AND/OR with widenable conditions as free of cost

Because widenable conditions with eventually lower into a constant, such instructions
as `and`, `or` etc. will also be optimized away. Treat them as free.

This is an important thing to have if we want that guards represented as experimental.guard
calls and in their explicit form (branch by `and` with widenable condition) have the same cost
for unroller and other passes like this.

Differential Revision: https://reviews.llvm.org/D146034
Reviewed By: nikic

[libclang] No longer attempt to get a dependent bit-width

Handle template parameter-dependent bit field widths in libclang

In a class template, a bit field's width may depend on a template
parameter. In this case the width expression cannot be evaluated.

Previously clang_getFieldDeclBitWidth() would assert, or cause memory
unsafety and return an invalid result if assertions are disabled.

This adds a check for this case which returns an error code.

This work was largely taken from an earlier patch which was reverted
due to an accidental API duplication
(https://reviews.llvm.org/D130303).

Fixes: https://github.com/llvm/llvm-project/issues/56644
Co-authored-by: Aaron Ballman <aaron@aaronballman.com>
Differential Revision: https://reviews.llvm.org/D146039

[PhaseOrdering] Add test for DAE/GlobalDCE interaction (NFC)

[lldb] Fix -Wswitch in TypeSystemClang.cpp ('SveBoolx2' and 'SveBoolx4' not handled in switch) (NFC)

/data/llvm-project/lldb/source/Plugins/TypeSystem/Clang/TypeSystemClang.cpp:4859:13: error: enumeration values 'SveBoolx2' and 'SveBoolx4' not handled in switch [-Werror,-Wswitch]
    switch (llvm::cast<clang::BuiltinType>(qual_type)->getKind()) {
            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated.

[libc] Remove dummy archive and file from integration test

This doesn't seem to be used anymore after recent changes that removed
the `--sysroot` method for the integration tests.

Reviewed By: sivachandra, lntue

Differential Revision: https://reviews.llvm.org/D146040

[X86] Fix typo in vXi64 ABDS/ABDU test cases

The select operands were commuted preventing D144789 from folding

[DAG] Fold (bitcast (logicop (bitcast x), (c))) -> (logicop x, (bitcast c)) iff the current logicop type is illegal

Try to remove extra bitcasts around logicops if we're dealing with illegal types

Fixes the regressions in D145939

Differential Revision: https://reviews.llvm.org/D146032

llvm/test/TableGen/intrinsic-pointer-to-any.td: Rework D125247.

It should've updated the number of `def i8 : ValueType<8, 3>;` to `5`,
since it renumbered ValueTypes.

FYI, the number 3 is `i2` and `IIT_I2(57)` was chosen in IntrinsicEmitter.

[mlir][vector] Add gather lowering patterns

This is for targets that do not support gather-like ops, e.g., SPIR-V.

Gather is expanded into lower-level vector ops with memory accesses
guarded with `scf.if`.

I also considered generating `vector.maskedload`s, but decided against
it to keep the `memref` and `tensor` codepath closer together. There's a
good chance that if a target doesn't support gather it does not support
masked loads either.

Issue: https://github.com/llvm/llvm-project/issues/60905

Reviewed By: ThomasRaoux

Differential Revision: https://reviews.llvm.org/D145942

[OpenMP][libomptarget][NFC] Add documentation regarding NextGen plugins

Differential Revision: https://reviews.llvm.org/D144975

[flang] Handle parent component on the LHS of intrinsic assignment

When the LHS is referring to a parent component the box need to be
reboxed to the parent component type so the runtime can handle the
assignment correctly.

Reviewed By: PeteSteinfeld

Differential Revision: https://reviews.llvm.org/D146046

[mlir][vector] Clarify OOB semantics of remaining load/store ops

This is a follow up for https://reviews.llvm.org/D145824 that clarifies
the out-of-bounds behavior for other masked load/store ops. It uses the
same wording and informal semantics syntax as `gather` and `scatter`.

Issue: https://github.com/llvm/llvm-project/issues/60905

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D145975

[CodeGen] Use *{Set,Map}::contains (NFC)

Fix false positive with unreachable C++ catch handlers

This addresses an issue found by WG21 and tracked by CWG2699 (which is
not yet publicly published). The basic problem is that Clang issues a
diagnostic about not being able to reach a handler, but that handler
*is* reached at runtime. Clang's diagnostic behavior was matching the
standard wording, and our runtime behavior was matching the standard's
intent.

This fixes the diagnostic so that it matches the runtime behavior more
closely, and reduces the number of false positives. This is the
direction of choice taken by Core for CWG2699 and it seems unlikely
that WG21 will change direction here.

Fixes https://github.com/llvm/llvm-project/issues/61177
Differential Revision: https://reviews.llvm.org/D145408

[DAG] Match select(icmp(x,y),sub(x,y),sub(y,x)) -> abd(x,y) patterns

Pulled out of PowerPC, and added ABDS support as well (hence the additional v4i32 PPC matches)

Differential Revision: https://reviews.llvm.org/D144789

[Docs] Added llvm-reduce docs in CommandGuide

[Docs] Added llvm-reduce docs in CommandGuide

Differential Revision: https://reviews.llvm.org/D146020