Without this, legalization might not recursively handle child ops properly.
Additionally, this is required for pattern rewriting to properly rollback conversions.

Differential Revision: https://reviews.llvm.org/D89122

The Callbacks.cpp test was taking a long time to compile on some build bots
causing timeouts. This patch splits up that test into five separate cpp
files and a header file.

Reviewed By: gribozavr2

Differential Revision: https://reviews.llvm.org/D88886

We cannot guarantee that the replacement expression is loop-invariant in
all AddRecs in the source expression. Use a rewriter that skips
AddRecExpr for now.

Fixes PR47776.

Reviewed By: hokein

Differential Revision: https://reviews.llvm.org/D89117

Allows us to drop the !isa<ConstantExpr> check.

The updated version of kernel outlining did not handle cases correctly
where an operand of a candidate for sinking itself was defined by an operation
that is a sinking candidate. In such cases, it could happen that sunk
operations were inserted in the wrong order, breaking ssa properties.

Differential Revision: https://reviews.llvm.org/D89112

This option just takes a single symbol name per invocation of the
option.

Differential Revision: https://reviews.llvm.org/D89007

Differential Revision: https://reviews.llvm.org/D89091

Differential Revision: https://reviews.llvm.org/D86828

Pushed by accident :(

This reverts commit 4eb627ed.

(Required for CodeActionContext.only)

Differential Revision: https://reviews.llvm.org/D88724

I messed up runlines in the original commit.

This patch extracts the ExprMutAnalyzer changes from https://reviews.llvm.org/D54943
into its own revision for simpler review and more atomic changes.

The analysis results are improved. Nested expressions (e.g. conditional
operators) are now detected properly. Some edge cases, especially
template induced imprecisions are improved upon.

Reviewed By: aaron.ballman

Differential Revision: https://reviews.llvm.org/D88088

Summary: Enabling warning after dereferece of smart pointer after branching on unknown inner pointer.

Reviewers: NoQ, Szelethus, vsavchenko, xazax.hun
Reviewed By: NoQ
Subscribers: martong, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D87043

IV widening is sometimes a strictly harmful transform (some examples
of this are shown in tests 11, 12 in widen-loop-comp.ll). One of the
reasons of this is that sometimes SCEV fails to prove some facts after
part of guards has been widened.

Though each single such case looks like a bug that can be addressed,
it seems that disabling of IV widening may be profitable in some cases.
We want to have an option to do so. By default, existing behavior is
preserved and IV widening is on.

When attempting to compute a differential orderIndex we were calculating the
bailout condition correctly, but then an errant "+ 1" meant the orderIndex we
created was invalid.

Added test.

Reviewed By: ftynse

Differential Revision: https://reviews.llvm.org/D89115

Buildbots aren't happy, need to investigate.
This reverts commit 32cc8f79.

Since AGR clobbers CC it should not be used here.

Fixes https://bugs.llvm.org/show_bug.cgi?id=47736.

Review: Ulrich Weigand
Differential Revision: https://reviews.llvm.org/D89034

Differential Revision: https://reviews.llvm.org/D87990

Reviewed By: mehdi_amini

Differential Revision: https://reviews.llvm.org/D89033

Summary: This patch is derived from D87384.
In this patch we expand the existing decomposition of mul-by-constant to be more general by implementing 2 patterns:
```
  mul x, (2^N + 2^M) --> (add (shl x, N), (shl x, M))
  mul x, (2^N - 2^M) --> (sub (shl x, N), (shl x, M))
```
The conversion will be trigged if the multiplier is a big constant that the target can't use a single multiplication instruction to handle. This is controlled by the hook `decomposeMulByConstant`.
More over, the conversion benefits from an ILP improvement since the instructions are independent. A case with the sequence like following also gets benefit since a shift instruction is saved.

```
*res1 = a * 0x8800;
*res2 = a * 0x8080;
```

Reviewed By: spatel

Differential Revision: https://reviews.llvm.org/D88201

This removes the precompiled binary and rewrites test to use YAML.

After this change we'll have no more precompiled inputs in `llvm-readobj/ELF/Inputs`.

Differential revision: https://reviews.llvm.org/D89097

This is a polymorphic class, copying it is a bad idea.

This was not a problem because most classes inheriting from it were
deleting their copy operations themselves. However, this enables us to
delete those explicit deletions, and ensure noone forgets to add them in
the future.

This adds methods to APFixedPoint for converting to and from
floating point values.

Differential Revision: https://reviews.llvm.org/D85961

Differential Revision: https://reviews.llvm.org/D88648

It is possible to get a fltSemantics of a particular Type,
but there is no way to produce a Type based on a
fltSemantics.

This adds the function Type::getFloatingPointTy, which
will return the appropriate floating point Type for a given
fltSemantics.

ConstantFP is modified to use this function instead of
implementing it itself. Also some minor refactors to use
Type::getFltSemantics instead of a hand-rolled version.

Differential Revision: https://reviews.llvm.org/D87512

This will enable queries like "clangd::" to find symbols under clangd
namespace, without requiring full "clang::clangd::" qualification.

Since Fuzzyfind performs the search under all scopes and only boosts the symbols
from relevant namespaces, we might get symbols from non-matching namespaces.
This patch chooses to drop those as they clearly do not match the query.

Fixes https://github.com/clangd/clangd/issues/550.

Differential Revision: https://reviews.llvm.org/D88814

This patch introduces hoisting detection logic into prepare state with
a partial AST traversal of the enclosing function.

We had some complaints from the users about this code action being almost always
available but failing most of the time. Hopefully this should reduce that noise.

The latency/correctness tradeoff is a bunch of hand-waving, but at least today
we don't have any other actions that are available on selection of statements,
so when we get to do the traversal it is quite likely that all the other checks
will bail out early. But this is still up for discussion, I am happy to abandon
the patch if you believe this is not practical.

Differential Revision: https://reviews.llvm.org/D85354

Differential Revision: https://reviews.llvm.org/D89036

`ninja mlir-doc` fails when Flang is configured with this.

Revert "[MLIR] Add a foldTrait() mechanism to allow traits to define folding and test it with an Involution trait"

This reverts commit 1ceaffd9.

The build is broken with  -DBUILD_SHARED_LIBS=ON ; seems like a possible
layering issue to investigate:

tools/mlir/lib/IR/CMakeFiles/obj.MLIRIR.dir/Operation.cpp.o: In function `mlir::MemoryEffectOpInterface::hasNoEffect(mlir::Operation*)':
Operation.cpp:(.text._ZN4mlir23MemoryEffectOpInterface11hasNoEffectEPNS_9OperationE[_ZN4mlir23MemoryEffectOpInterface11hasNoEffectEPNS_9OperationE]+0x9c): undefined reference to `mlir::MemoryEffectOpInterface::getEffects(llvm::SmallVectorImpl<mlir::SideEffects::EffectInstance<mlir::MemoryEffects::Effect> >&)'

Reviewed By: mehdi_amini, jpienaar

Differential Revision: https://reviews.llvm.org/D89014